git-svn.perlon commit svn: allow branches outside of refs/remotes (6f5748e)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $sha1 $sha1_short $_revision $_repository
   8                $_q $_authors $_authors_prog %users/;
   9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  10$VERSION = '@@GIT_VERSION@@';
  11
  12# From which subdir have we been invoked?
  13my $cmd_dir_prefix = eval {
  14        command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
  15} || '';
  16
  17my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
  18$ENV{GIT_DIR} ||= '.git';
  19$Git::SVN::default_repo_id = 'svn';
  20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  21$Git::SVN::Ra::_log_window_size = 100;
  22$Git::SVN::_minimize_url = 'unset';
  23
  24$Git::SVN::Log::TZ = $ENV{TZ};
  25$ENV{TZ} = 'UTC';
  26$| = 1; # unbuffer STDOUT
  27
  28sub fatal (@) { print STDERR "@_\n"; exit 1 }
  29require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  30require SVN::Ra;
  31require SVN::Delta;
  32if ($SVN::Core::VERSION lt '1.1.0') {
  33        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  34}
  35my $can_compress = eval { require Compress::Zlib; 1};
  36push @Git::SVN::Ra::ISA, 'SVN::Ra';
  37push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  38push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  39use Carp qw/croak/;
  40use Digest::MD5;
  41use IO::File qw//;
  42use File::Basename qw/dirname basename/;
  43use File::Path qw/mkpath/;
  44use File::Spec;
  45use File::Find;
  46use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  47use IPC::Open3;
  48use Git;
  49
  50BEGIN {
  51        # import functions from Git into our packages, en masse
  52        no strict 'refs';
  53        foreach (qw/command command_oneline command_noisy command_output_pipe
  54                    command_input_pipe command_close_pipe
  55                    command_bidi_pipe command_close_bidi_pipe/) {
  56                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  57                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  58                        __PACKAGE__) {
  59                        *{"${package}::$_"} = \&{"Git::$_"};
  60                }
  61        }
  62}
  63
  64my ($SVN);
  65
  66$sha1 = qr/[a-f\d]{40}/;
  67$sha1_short = qr/[a-f\d]{4,40}/;
  68my ($_stdin, $_help, $_edit,
  69        $_message, $_file, $_branch_dest,
  70        $_template, $_shared,
  71        $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
  72        $_merge, $_strategy, $_dry_run, $_local,
  73        $_prefix, $_no_checkout, $_url, $_verbose,
  74        $_git_format, $_commit_url, $_tag);
  75$Git::SVN::_follow_parent = 1;
  76$_q ||= 0;
  77my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  78                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  79                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
  80                    'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
  81my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  82                'authors-file|A=s' => \$_authors,
  83                'authors-prog=s' => \$_authors_prog,
  84                'repack:i' => \$Git::SVN::_repack,
  85                'noMetadata' => \$Git::SVN::_no_metadata,
  86                'useSvmProps' => \$Git::SVN::_use_svm_props,
  87                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  88                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  89                'no-checkout' => \$_no_checkout,
  90                'quiet|q+' => \$_q,
  91                'repack-flags|repack-args|repack-opts=s' =>
  92                   \$Git::SVN::_repack_flags,
  93                'use-log-author' => \$Git::SVN::_use_log_author,
  94                'add-author-from' => \$Git::SVN::_add_author_from,
  95                'localtime' => \$Git::SVN::_localtime,
  96                %remote_opts );
  97
  98my ($_trunk, @_tags, @_branches, $_stdlayout);
  99my %icv;
 100my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
 101                  'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
 102                  'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
 103                  'stdlayout|s' => \$_stdlayout,
 104                  'minimize-url|m!' => \$Git::SVN::_minimize_url,
 105                  'no-metadata' => sub { $icv{noMetadata} = 1 },
 106                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
 107                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
 108                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 109                  %remote_opts );
 110my %cmt_opts = ( 'edit|e' => \$_edit,
 111                'rmdir' => \$SVN::Git::Editor::_rmdir,
 112                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 113                'l=i' => \$SVN::Git::Editor::_rename_limit,
 114                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 115);
 116
 117my %cmd = (
 118        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 119                        { 'revision|r=s' => \$_revision,
 120                          'fetch-all|all' => \$_fetch_all,
 121                          'parent|p' => \$_fetch_parent,
 122                           %fc_opts } ],
 123        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 124                        { 'revision|r=s' => \$_revision,
 125                           %fc_opts, %init_opts } ],
 126        init => [ \&cmd_init, "Initialize a repo for tracking" .
 127                          " (requires URL argument)",
 128                          \%init_opts ],
 129        'multi-init' => [ \&cmd_multi_init,
 130                          "Deprecated alias for ".
 131                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 132                          \%init_opts ],
 133        dcommit => [ \&cmd_dcommit,
 134                     'Commit several diffs to merge with upstream',
 135                        { 'merge|m|M' => \$_merge,
 136                          'strategy|s=s' => \$_strategy,
 137                          'verbose|v' => \$_verbose,
 138                          'dry-run|n' => \$_dry_run,
 139                          'fetch-all|all' => \$_fetch_all,
 140                          'commit-url=s' => \$_commit_url,
 141                          'revision|r=i' => \$_revision,
 142                          'no-rebase' => \$_no_rebase,
 143                        %cmt_opts, %fc_opts } ],
 144        branch => [ \&cmd_branch,
 145                    'Create a branch in the SVN repository',
 146                    { 'message|m=s' => \$_message,
 147                      'destination|d=s' => \$_branch_dest,
 148                      'dry-run|n' => \$_dry_run,
 149                      'tag|t' => \$_tag } ],
 150        tag => [ sub { $_tag = 1; cmd_branch(@_) },
 151                 'Create a tag in the SVN repository',
 152                 { 'message|m=s' => \$_message,
 153                   'destination|d=s' => \$_branch_dest,
 154                   'dry-run|n' => \$_dry_run } ],
 155        'set-tree' => [ \&cmd_set_tree,
 156                        "Set an SVN repository to a git tree-ish",
 157                        { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
 158        'create-ignore' => [ \&cmd_create_ignore,
 159                             'Create a .gitignore per svn:ignore',
 160                             { 'revision|r=i' => \$_revision
 161                             } ],
 162        'propget' => [ \&cmd_propget,
 163                       'Print the value of a property on a file or directory',
 164                       { 'revision|r=i' => \$_revision } ],
 165        'proplist' => [ \&cmd_proplist,
 166                       'List all properties of a file or directory',
 167                       { 'revision|r=i' => \$_revision } ],
 168        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 169                        { 'revision|r=i' => \$_revision
 170                        } ],
 171        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 172                        { 'revision|r=i' => \$_revision
 173                        } ],
 174        'multi-fetch' => [ \&cmd_multi_fetch,
 175                           "Deprecated alias for $0 fetch --all",
 176                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 177        'migrate' => [ sub { },
 178                       # no-op, we automatically run this anyways,
 179                       'Migrate configuration/metadata/layout from
 180                        previous versions of git-svn',
 181                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 182                         %remote_opts } ],
 183        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 184                        { 'limit=i' => \$Git::SVN::Log::limit,
 185                          'revision|r=s' => \$_revision,
 186                          'verbose|v' => \$Git::SVN::Log::verbose,
 187                          'incremental' => \$Git::SVN::Log::incremental,
 188                          'oneline' => \$Git::SVN::Log::oneline,
 189                          'show-commit' => \$Git::SVN::Log::show_commit,
 190                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 191                          'authors-file|A=s' => \$_authors,
 192                          'color' => \$Git::SVN::Log::color,
 193                          'pager=s' => \$Git::SVN::Log::pager
 194                        } ],
 195        'find-rev' => [ \&cmd_find_rev,
 196                        "Translate between SVN revision numbers and tree-ish",
 197                        {} ],
 198        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 199                        { 'merge|m|M' => \$_merge,
 200                          'verbose|v' => \$_verbose,
 201                          'strategy|s=s' => \$_strategy,
 202                          'local|l' => \$_local,
 203                          'fetch-all|all' => \$_fetch_all,
 204                          'dry-run|n' => \$_dry_run,
 205                          %fc_opts } ],
 206        'commit-diff' => [ \&cmd_commit_diff,
 207                           'Commit a diff between two trees',
 208                        { 'message|m=s' => \$_message,
 209                          'file|F=s' => \$_file,
 210                          'revision|r=s' => \$_revision,
 211                        %cmt_opts } ],
 212        'info' => [ \&cmd_info,
 213                    "Show info about the latest SVN revision
 214                     on the current branch",
 215                    { 'url' => \$_url, } ],
 216        'blame' => [ \&Git::SVN::Log::cmd_blame,
 217                    "Show what revision and author last modified each line of a file",
 218                    { 'git-format' => \$_git_format } ],
 219        'reset' => [ \&cmd_reset,
 220                     "Undo fetches back to the specified SVN revision",
 221                     { 'revision|r=s' => \$_revision,
 222                       'parent|p' => \$_fetch_parent } ],
 223        'gc' => [ \&cmd_gc,
 224                  "Compress unhandled.log files in .git/svn and remove " .
 225                  "index files in .git/svn",
 226                {} ],
 227);
 228
 229my $cmd;
 230for (my $i = 0; $i < @ARGV; $i++) {
 231        if (defined $cmd{$ARGV[$i]}) {
 232                $cmd = $ARGV[$i];
 233                splice @ARGV, $i, 1;
 234                last;
 235        } elsif ($ARGV[$i] eq 'help') {
 236                $cmd = $ARGV[$i+1];
 237                usage(0);
 238        }
 239};
 240
 241# make sure we're always running at the top-level working directory
 242unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 243        unless (-d $ENV{GIT_DIR}) {
 244                if ($git_dir_user_set) {
 245                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 246                            "but it is not a directory\n";
 247                }
 248                my $git_dir = delete $ENV{GIT_DIR};
 249                my $cdup = undef;
 250                git_cmd_try {
 251                        $cdup = command_oneline(qw/rev-parse --show-cdup/);
 252                        $git_dir = '.' unless ($cdup);
 253                        chomp $cdup if ($cdup);
 254                        $cdup = "." unless ($cdup && length $cdup);
 255                } "Already at toplevel, but $git_dir not found\n";
 256                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 257                unless (-d $git_dir) {
 258                        die "$git_dir still not found after going to ",
 259                            "'$cdup'\n";
 260                }
 261                $ENV{GIT_DIR} = $git_dir;
 262        }
 263        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 264}
 265
 266my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 267
 268read_repo_config(\%opts);
 269if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 270        Getopt::Long::Configure('pass_through');
 271}
 272my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 273                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 274                    'id|i=s' => \$Git::SVN::default_ref_id,
 275                    'svn-remote|remote|R=s' => sub {
 276                       $Git::SVN::no_reuse_existing = 1;
 277                       $Git::SVN::default_repo_id = $_[1] });
 278exit 1 if (!$rv && $cmd && $cmd ne 'log');
 279
 280usage(0) if $_help;
 281version() if $_version;
 282usage(1) unless defined $cmd;
 283load_authors() if $_authors;
 284if (defined $_authors_prog) {
 285        $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
 286}
 287
 288unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 289        Git::SVN::Migration::migration_check();
 290}
 291Git::SVN::init_vars();
 292eval {
 293        Git::SVN::verify_remotes_sanity();
 294        $cmd{$cmd}->[0]->(@ARGV);
 295};
 296fatal $@ if $@;
 297post_fetch_checkout();
 298exit 0;
 299
 300####################### primary functions ######################
 301sub usage {
 302        my $exit = shift || 0;
 303        my $fd = $exit ? \*STDERR : \*STDOUT;
 304        print $fd <<"";
 305git-svn - bidirectional operations between a single Subversion tree and git
 306Usage: git svn <command> [options] [arguments]\n
 307
 308        print $fd "Available commands:\n" unless $cmd;
 309
 310        foreach (sort keys %cmd) {
 311                next if $cmd && $cmd ne $_;
 312                next if /^multi-/; # don't show deprecated commands
 313                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 314                foreach (sort keys %{$cmd{$_}->[2]}) {
 315                        # mixed-case options are for .git/config only
 316                        next if /[A-Z]/ && /^[a-z]+$/i;
 317                        # prints out arguments as they should be passed:
 318                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 319                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 320                                                        "--$_" : "-$_" }
 321                                                split /\|/,$_)," $x\n";
 322                }
 323        }
 324        print $fd <<"";
 325\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 326arbitrary identifier if you're tracking multiple SVN branches/repositories in
 327one git repository and want to keep them separate.  See git-svn(1) for more
 328information.
 329
 330        exit $exit;
 331}
 332
 333sub version {
 334        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 335        exit 0;
 336}
 337
 338sub do_git_init_db {
 339        unless (-d $ENV{GIT_DIR}) {
 340                my @init_db = ('init');
 341                push @init_db, "--template=$_template" if defined $_template;
 342                if (defined $_shared) {
 343                        if ($_shared =~ /[a-z]/) {
 344                                push @init_db, "--shared=$_shared";
 345                        } else {
 346                                push @init_db, "--shared";
 347                        }
 348                }
 349                command_noisy(@init_db);
 350                $_repository = Git->repository(Repository => ".git");
 351        }
 352        command_noisy('config', 'core.autocrlf', 'false');
 353        my $set;
 354        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 355        foreach my $i (keys %icv) {
 356                die "'$set' and '$i' cannot both be set\n" if $set;
 357                next unless defined $icv{$i};
 358                command_noisy('config', "$pfx.$i", $icv{$i});
 359                $set = $i;
 360        }
 361        my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
 362        command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
 363                if defined $$ignore_regex;
 364}
 365
 366sub init_subdir {
 367        my $repo_path = shift or return;
 368        mkpath([$repo_path]) unless -d $repo_path;
 369        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 370        $ENV{GIT_DIR} = '.git';
 371        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 372}
 373
 374sub cmd_clone {
 375        my ($url, $path) = @_;
 376        if (!defined $path &&
 377            (defined $_trunk || @_branches || @_tags ||
 378             defined $_stdlayout) &&
 379            $url !~ m#^[a-z\+]+://#) {
 380                $path = $url;
 381        }
 382        $path = basename($url) if !defined $path || !length $path;
 383        cmd_init($url, $path);
 384        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 385        command_oneline('config', 'svn.authorsfile', $_authors) if $_authors;
 386}
 387
 388sub cmd_init {
 389        if (defined $_stdlayout) {
 390                $_trunk = 'trunk' if (!defined $_trunk);
 391                @_tags = 'tags' if (! @_tags);
 392                @_branches = 'branches' if (! @_branches);
 393        }
 394        if (defined $_trunk || @_branches || @_tags) {
 395                return cmd_multi_init(@_);
 396        }
 397        my $url = shift or die "SVN repository location required ",
 398                               "as a command-line argument\n";
 399        $url = canonicalize_url($url);
 400        init_subdir(@_);
 401        do_git_init_db();
 402
 403        if ($Git::SVN::_minimize_url eq 'unset') {
 404                $Git::SVN::_minimize_url = 0;
 405        }
 406
 407        Git::SVN->init($url);
 408}
 409
 410sub cmd_fetch {
 411        if (grep /^\d+=./, @_) {
 412                die "'<rev>=<commit>' fetch arguments are ",
 413                    "no longer supported.\n";
 414        }
 415        my ($remote) = @_;
 416        if (@_ > 1) {
 417                die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
 418        }
 419        if ($_fetch_parent) {
 420                my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 421                unless ($gs) {
 422                        die "Unable to determine upstream SVN information from ",
 423                            "working tree history\n";
 424                }
 425                # just fetch, don't checkout.
 426                $_no_checkout = 'true';
 427                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 428        } elsif ($_fetch_all) {
 429                cmd_multi_fetch();
 430        } else {
 431                $remote ||= $Git::SVN::default_repo_id;
 432                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 433        }
 434}
 435
 436sub cmd_set_tree {
 437        my (@commits) = @_;
 438        if ($_stdin || !@commits) {
 439                print "Reading from stdin...\n";
 440                @commits = ();
 441                while (<STDIN>) {
 442                        if (/\b($sha1_short)\b/o) {
 443                                unshift @commits, $1;
 444                        }
 445                }
 446        }
 447        my @revs;
 448        foreach my $c (@commits) {
 449                my @tmp = command('rev-parse',$c);
 450                if (scalar @tmp == 1) {
 451                        push @revs, $tmp[0];
 452                } elsif (scalar @tmp > 1) {
 453                        push @revs, reverse(command('rev-list',@tmp));
 454                } else {
 455                        fatal "Failed to rev-parse $c";
 456                }
 457        }
 458        my $gs = Git::SVN->new;
 459        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 460        $gs->fetch;
 461        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 462                fatal "There are new revisions that were fetched ",
 463                      "and need to be merged (or acknowledged) ",
 464                      "before committing.\nlast rev: $r_last\n",
 465                      " current: $gs->{last_rev}";
 466        }
 467        $gs->set_tree($_) foreach @revs;
 468        print "Done committing ",scalar @revs," revisions to SVN\n";
 469        unlink $gs->{index};
 470}
 471
 472sub cmd_dcommit {
 473        my $head = shift;
 474        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 475                'Cannot dcommit with a dirty index.  Commit your changes first, '
 476                . "or stash them with `git stash'.\n";
 477        $head ||= 'HEAD';
 478
 479        my $old_head;
 480        if ($head ne 'HEAD') {
 481                $old_head = eval {
 482                        command_oneline([qw/symbolic-ref -q HEAD/])
 483                };
 484                if ($old_head) {
 485                        $old_head =~ s{^refs/heads/}{};
 486                } else {
 487                        $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
 488                }
 489                command(['checkout', $head], STDERR => 0);
 490        }
 491
 492        my @refs;
 493        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
 494        unless ($gs) {
 495                die "Unable to determine upstream SVN information from ",
 496                    "$head history.\nPerhaps the repository is empty.";
 497        }
 498
 499        if (defined $_commit_url) {
 500                $url = $_commit_url;
 501        } else {
 502                $url = eval { command_oneline('config', '--get',
 503                              "svn-remote.$gs->{repo_id}.commiturl") };
 504                if (!$url) {
 505                        $url = $gs->full_url
 506                }
 507        }
 508
 509        my $last_rev = $_revision if defined $_revision;
 510        if ($url) {
 511                print "Committing to $url ...\n";
 512        }
 513        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 514        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 515                warn "Attempting to commit more than one change while ",
 516                     "--no-rebase is enabled.\n",
 517                     "If these changes depend on each other, re-running ",
 518                     "without --no-rebase may be required."
 519        }
 520        my $expect_url = $url;
 521        Git::SVN::remove_username($expect_url);
 522        while (1) {
 523                my $d = shift @$linear_refs or last;
 524                unless (defined $last_rev) {
 525                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 526                        unless (defined $last_rev) {
 527                                fatal "Unable to extract revision information ",
 528                                      "from commit $d~1";
 529                        }
 530                }
 531                if ($_dry_run) {
 532                        print "diff-tree $d~1 $d\n";
 533                } else {
 534                        my $cmt_rev;
 535                        my %ed_opts = ( r => $last_rev,
 536                                        log => get_commit_entry($d)->{log},
 537                                        ra => Git::SVN::Ra->new($url),
 538                                        config => SVN::Core::config_get_config(
 539                                                $Git::SVN::Ra::config_dir
 540                                        ),
 541                                        tree_a => "$d~1",
 542                                        tree_b => $d,
 543                                        editor_cb => sub {
 544                                               print "Committed r$_[0]\n";
 545                                               $cmt_rev = $_[0];
 546                                        },
 547                                        svn_path => '');
 548                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 549                                print "No changes\n$d~1 == $d\n";
 550                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 551                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 552                                                               $parents->{$d};
 553                        }
 554                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 555                        $last_rev = $cmt_rev;
 556                        next if $_no_rebase;
 557
 558                        # we always want to rebase against the current HEAD,
 559                        # not any head that was passed to us
 560                        my @diff = command('diff-tree', $d,
 561                                           $gs->refname, '--');
 562                        my @finish;
 563                        if (@diff) {
 564                                @finish = rebase_cmd();
 565                                print STDERR "W: $d and ", $gs->refname,
 566                                             " differ, using @finish:\n",
 567                                             join("\n", @diff), "\n";
 568                        } else {
 569                                print "No changes between current HEAD and ",
 570                                      $gs->refname,
 571                                      "\nResetting to the latest ",
 572                                      $gs->refname, "\n";
 573                                @finish = qw/reset --mixed/;
 574                        }
 575                        command_noisy(@finish, $gs->refname);
 576                        if (@diff) {
 577                                @refs = ();
 578                                my ($url_, $rev_, $uuid_, $gs_) =
 579                                              working_head_info('HEAD', \@refs);
 580                                my ($linear_refs_, $parents_) =
 581                                              linearize_history($gs_, \@refs);
 582                                if (scalar(@$linear_refs) !=
 583                                    scalar(@$linear_refs_)) {
 584                                        fatal "# of revisions changed ",
 585                                          "\nbefore:\n",
 586                                          join("\n", @$linear_refs),
 587                                          "\n\nafter:\n",
 588                                          join("\n", @$linear_refs_), "\n",
 589                                          'If you are attempting to commit ',
 590                                          "merges, try running:\n\t",
 591                                          'git rebase --interactive',
 592                                          '--preserve-merges ',
 593                                          $gs->refname,
 594                                          "\nBefore dcommitting";
 595                                }
 596                                if ($url_ ne $expect_url) {
 597                                        fatal "URL mismatch after rebase: ",
 598                                              "$url_ != $expect_url";
 599                                }
 600                                if ($uuid_ ne $uuid) {
 601                                        fatal "uuid mismatch after rebase: ",
 602                                              "$uuid_ != $uuid";
 603                                }
 604                                # remap parents
 605                                my (%p, @l, $i);
 606                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 607                                        my $new = $linear_refs_->[$i] or next;
 608                                        $p{$new} =
 609                                                $parents->{$linear_refs->[$i]};
 610                                        push @l, $new;
 611                                }
 612                                $parents = \%p;
 613                                $linear_refs = \@l;
 614                        }
 615                }
 616        }
 617
 618        if ($old_head) {
 619                my $new_head = command_oneline(qw/rev-parse HEAD/);
 620                my $new_is_symbolic = eval {
 621                        command_oneline(qw/symbolic-ref -q HEAD/);
 622                };
 623                if ($new_is_symbolic) {
 624                        print "dcommitted the branch ", $head, "\n";
 625                } else {
 626                        print "dcommitted on a detached HEAD because you gave ",
 627                              "a revision argument.\n",
 628                              "The rewritten commit is: ", $new_head, "\n";
 629                }
 630                command(['checkout', $old_head], STDERR => 0);
 631        }
 632
 633        unlink $gs->{index};
 634}
 635
 636sub cmd_branch {
 637        my ($branch_name, $head) = @_;
 638
 639        unless (defined $branch_name && length $branch_name) {
 640                die(($_tag ? "tag" : "branch") . " name required\n");
 641        }
 642        $head ||= 'HEAD';
 643
 644        my ($src, $rev, undef, $gs) = working_head_info($head);
 645
 646        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
 647        my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
 648        my $glob;
 649        if ($#{$allglobs} == 0) {
 650                $glob = $allglobs->[0];
 651        } else {
 652                unless(defined $_branch_dest) {
 653                        die "Multiple ",
 654                            $_tag ? "tag" : "branch",
 655                            " paths defined for Subversion repository.\n",
 656                            "You must specify where you want to create the ",
 657                            $_tag ? "tag" : "branch",
 658                            " with the --destination argument.\n";
 659                }
 660                foreach my $g (@{$allglobs}) {
 661                        # SVN::Git::Editor could probably be moved to Git.pm..
 662                        my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
 663                        if ($_branch_dest =~ /$re/) {
 664                                $glob = $g;
 665                                last;
 666                        }
 667                }
 668                unless (defined $glob) {
 669                        my $dest_re = qr/\b\Q$_branch_dest\E\b/;
 670                        foreach my $g (@{$allglobs}) {
 671                                $g->{path}->{left} =~ /$dest_re/ or next;
 672                                if (defined $glob) {
 673                                        die "Ambiguous destination: ",
 674                                            $_branch_dest, "\nmatches both '",
 675                                            $glob->{path}->{left}, "' and '",
 676                                            $g->{path}->{left}, "'\n";
 677                                }
 678                                $glob = $g;
 679                        }
 680                        unless (defined $glob) {
 681                                die "Unknown ",
 682                                    $_tag ? "tag" : "branch",
 683                                    " destination $_branch_dest\n";
 684                        }
 685                }
 686        }
 687        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
 688        my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
 689
 690        my $ctx = SVN::Client->new(
 691                auth    => Git::SVN::Ra::_auth_providers(),
 692                log_msg => sub {
 693                        ${ $_[0] } = defined $_message
 694                                ? $_message
 695                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
 696                                . $branch_name;
 697                },
 698        );
 699
 700        eval {
 701                $ctx->ls($dst, 'HEAD', 0);
 702        } and die "branch ${branch_name} already exists\n";
 703
 704        print "Copying ${src} at r${rev} to ${dst}...\n";
 705        $ctx->copy($src, $rev, $dst)
 706                unless $_dry_run;
 707
 708        $gs->fetch_all;
 709}
 710
 711sub cmd_find_rev {
 712        my $revision_or_hash = shift or die "SVN or git revision required ",
 713                                            "as a command-line argument\n";
 714        my $result;
 715        if ($revision_or_hash =~ /^r\d+$/) {
 716                my $head = shift;
 717                $head ||= 'HEAD';
 718                my @refs;
 719                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 720                unless ($gs) {
 721                        die "Unable to determine upstream SVN information from ",
 722                            "$head history\n";
 723                }
 724                my $desired_revision = substr($revision_or_hash, 1);
 725                $result = $gs->rev_map_get($desired_revision, $uuid);
 726        } else {
 727                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 728                $result = $rev;
 729        }
 730        print "$result\n" if $result;
 731}
 732
 733sub cmd_rebase {
 734        command_noisy(qw/update-index --refresh/);
 735        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 736        unless ($gs) {
 737                die "Unable to determine upstream SVN information from ",
 738                    "working tree history\n";
 739        }
 740        if ($_dry_run) {
 741                print "Remote Branch: " . $gs->refname . "\n";
 742                print "SVN URL: " . $url . "\n";
 743                return;
 744        }
 745        if (command(qw/diff-index HEAD --/)) {
 746                print STDERR "Cannot rebase with uncommited changes:\n";
 747                command_noisy('status');
 748                exit 1;
 749        }
 750        unless ($_local) {
 751                # rebase will checkout for us, so no need to do it explicitly
 752                $_no_checkout = 'true';
 753                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 754        }
 755        command_noisy(rebase_cmd(), $gs->refname);
 756}
 757
 758sub cmd_show_ignore {
 759        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 760        $gs ||= Git::SVN->new;
 761        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 762        $gs->prop_walk($gs->{path}, $r, sub {
 763                my ($gs, $path, $props) = @_;
 764                print STDOUT "\n# $path\n";
 765                my $s = $props->{'svn:ignore'} or return;
 766                $s =~ s/[\r\n]+/\n/g;
 767                $s =~ s/^\n+//;
 768                chomp $s;
 769                $s =~ s#^#$path#gm;
 770                print STDOUT "$s\n";
 771        });
 772}
 773
 774sub cmd_show_externals {
 775        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 776        $gs ||= Git::SVN->new;
 777        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 778        $gs->prop_walk($gs->{path}, $r, sub {
 779                my ($gs, $path, $props) = @_;
 780                print STDOUT "\n# $path\n";
 781                my $s = $props->{'svn:externals'} or return;
 782                $s =~ s/[\r\n]+/\n/g;
 783                chomp $s;
 784                $s =~ s#^#$path#gm;
 785                print STDOUT "$s\n";
 786        });
 787}
 788
 789sub cmd_create_ignore {
 790        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 791        $gs ||= Git::SVN->new;
 792        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 793        $gs->prop_walk($gs->{path}, $r, sub {
 794                my ($gs, $path, $props) = @_;
 795                # $path is of the form /path/to/dir/
 796                $path = '.' . $path;
 797                # SVN can have attributes on empty directories,
 798                # which git won't track
 799                mkpath([$path]) unless -d $path;
 800                my $ignore = $path . '.gitignore';
 801                my $s = $props->{'svn:ignore'} or return;
 802                open(GITIGNORE, '>', $ignore)
 803                  or fatal("Failed to open `$ignore' for writing: $!");
 804                $s =~ s/[\r\n]+/\n/g;
 805                $s =~ s/^\n+//;
 806                chomp $s;
 807                # Prefix all patterns so that the ignore doesn't apply
 808                # to sub-directories.
 809                $s =~ s#^#/#gm;
 810                print GITIGNORE "$s\n";
 811                close(GITIGNORE)
 812                  or fatal("Failed to close `$ignore': $!");
 813                command_noisy('add', '-f', $ignore);
 814        });
 815}
 816
 817sub canonicalize_path {
 818        my ($path) = @_;
 819        my $dot_slash_added = 0;
 820        if (substr($path, 0, 1) ne "/") {
 821                $path = "./" . $path;
 822                $dot_slash_added = 1;
 823        }
 824        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 825        # good reason), so let's do this manually.
 826        $path =~ s#/+#/#g;
 827        $path =~ s#/\.(?:/|$)#/#g;
 828        $path =~ s#/[^/]+/\.\.##g;
 829        $path =~ s#/$##g;
 830        $path =~ s#^\./## if $dot_slash_added;
 831        $path =~ s#^/##;
 832        $path =~ s#^\.$##;
 833        return $path;
 834}
 835
 836sub canonicalize_url {
 837        my ($url) = @_;
 838        $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
 839        return $url;
 840}
 841
 842# get_svnprops(PATH)
 843# ------------------
 844# Helper for cmd_propget and cmd_proplist below.
 845sub get_svnprops {
 846        my $path = shift;
 847        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 848        $gs ||= Git::SVN->new;
 849
 850        # prefix THE PATH by the sub-directory from which the user
 851        # invoked us.
 852        $path = $cmd_dir_prefix . $path;
 853        fatal("No such file or directory: $path") unless -e $path;
 854        my $is_dir = -d $path ? 1 : 0;
 855        $path = $gs->{path} . '/' . $path;
 856
 857        # canonicalize the path (otherwise libsvn will abort or fail to
 858        # find the file)
 859        $path = canonicalize_path($path);
 860
 861        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 862        my $props;
 863        if ($is_dir) {
 864                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 865        }
 866        else {
 867                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 868        }
 869        return $props;
 870}
 871
 872# cmd_propget (PROP, PATH)
 873# ------------------------
 874# Print the SVN property PROP for PATH.
 875sub cmd_propget {
 876        my ($prop, $path) = @_;
 877        $path = '.' if not defined $path;
 878        usage(1) if not defined $prop;
 879        my $props = get_svnprops($path);
 880        if (not defined $props->{$prop}) {
 881                fatal("`$path' does not have a `$prop' SVN property.");
 882        }
 883        print $props->{$prop} . "\n";
 884}
 885
 886# cmd_proplist (PATH)
 887# -------------------
 888# Print the list of SVN properties for PATH.
 889sub cmd_proplist {
 890        my $path = shift;
 891        $path = '.' if not defined $path;
 892        my $props = get_svnprops($path);
 893        print "Properties on '$path':\n";
 894        foreach (sort keys %{$props}) {
 895                print "  $_\n";
 896        }
 897}
 898
 899sub cmd_multi_init {
 900        my $url = shift;
 901        unless (defined $_trunk || @_branches || @_tags) {
 902                usage(1);
 903        }
 904
 905        $_prefix = '' unless defined $_prefix;
 906        if (defined $url) {
 907                $url = canonicalize_url($url);
 908                init_subdir(@_);
 909        }
 910        do_git_init_db();
 911        if (defined $_trunk) {
 912                my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
 913                # try both old-style and new-style lookups:
 914                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 915                unless ($gs_trunk) {
 916                        my ($trunk_url, $trunk_path) =
 917                                              complete_svn_url($url, $_trunk);
 918                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 919                                                   undef, $trunk_ref);
 920                }
 921        }
 922        return unless @_branches || @_tags;
 923        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 924        foreach my $path (@_branches) {
 925                complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
 926        }
 927        foreach my $path (@_tags) {
 928                complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
 929        }
 930}
 931
 932sub cmd_multi_fetch {
 933        my $remotes = Git::SVN::read_all_remotes();
 934        foreach my $repo_id (sort keys %$remotes) {
 935                if ($remotes->{$repo_id}->{url}) {
 936                        Git::SVN::fetch_all($repo_id, $remotes);
 937                }
 938        }
 939}
 940
 941# this command is special because it requires no metadata
 942sub cmd_commit_diff {
 943        my ($ta, $tb, $url) = @_;
 944        my $usage = "Usage: $0 commit-diff -r<revision> ".
 945                    "<tree-ish> <tree-ish> [<URL>]";
 946        fatal($usage) if (!defined $ta || !defined $tb);
 947        my $svn_path = '';
 948        if (!defined $url) {
 949                my $gs = eval { Git::SVN->new };
 950                if (!$gs) {
 951                        fatal("Needed URL or usable git-svn --id in ",
 952                              "the command-line\n", $usage);
 953                }
 954                $url = $gs->{url};
 955                $svn_path = $gs->{path};
 956        }
 957        unless (defined $_revision) {
 958                fatal("-r|--revision is a required argument\n", $usage);
 959        }
 960        if (defined $_message && defined $_file) {
 961                fatal("Both --message/-m and --file/-F specified ",
 962                      "for the commit message.\n",
 963                      "I have no idea what you mean");
 964        }
 965        if (defined $_file) {
 966                $_message = file_to_s($_file);
 967        } else {
 968                $_message ||= get_commit_entry($tb)->{log};
 969        }
 970        my $ra ||= Git::SVN::Ra->new($url);
 971        my $r = $_revision;
 972        if ($r eq 'HEAD') {
 973                $r = $ra->get_latest_revnum;
 974        } elsif ($r !~ /^\d+$/) {
 975                die "revision argument: $r not understood by git-svn\n";
 976        }
 977        my %ed_opts = ( r => $r,
 978                        log => $_message,
 979                        ra => $ra,
 980                        tree_a => $ta,
 981                        tree_b => $tb,
 982                        editor_cb => sub { print "Committed r$_[0]\n" },
 983                        svn_path => $svn_path );
 984        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 985                print "No changes\n$ta == $tb\n";
 986        }
 987}
 988
 989sub escape_uri_only {
 990        my ($uri) = @_;
 991        my @tmp;
 992        foreach (split m{/}, $uri) {
 993                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
 994                push @tmp, $_;
 995        }
 996        join('/', @tmp);
 997}
 998
 999sub escape_url {
1000        my ($url) = @_;
1001        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1002                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1003                $url = "$scheme://$domain$uri";
1004        }
1005        $url;
1006}
1007
1008sub cmd_info {
1009        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1010        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1011        if (exists $_[1]) {
1012                die "Too many arguments specified\n";
1013        }
1014
1015        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1016
1017        if (!$file_type && !$diff_status) {
1018                print STDERR "svn: '$path' is not under version control\n";
1019                exit 1;
1020        }
1021
1022        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1023        unless ($gs) {
1024                die "Unable to determine upstream SVN information from ",
1025                    "working tree history\n";
1026        }
1027
1028        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1029        $path = "." if $path eq "";
1030
1031        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1032
1033        if ($_url) {
1034                print escape_url($full_url), "\n";
1035                return;
1036        }
1037
1038        my $result = "Path: $path\n";
1039        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1040        $result .= "URL: " . escape_url($full_url) . "\n";
1041
1042        eval {
1043                my $repos_root = $gs->repos_root;
1044                Git::SVN::remove_username($repos_root);
1045                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1046        };
1047        if ($@) {
1048                $result .= "Repository Root: (offline)\n";
1049        }
1050        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1051                ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1052        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1053
1054        $result .= "Node Kind: " .
1055                   ($file_type eq "dir" ? "directory" : "file") . "\n";
1056
1057        my $schedule = $diff_status eq "A"
1058                       ? "add"
1059                       : ($diff_status eq "D" ? "delete" : "normal");
1060        $result .= "Schedule: $schedule\n";
1061
1062        if ($diff_status eq "A") {
1063                print $result, "\n";
1064                return;
1065        }
1066
1067        my ($lc_author, $lc_rev, $lc_date_utc);
1068        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1069        my $log = command_output_pipe(@args);
1070        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1071        while (<$log>) {
1072                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1073                        $lc_author = $1;
1074                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1075                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
1076                        (undef, $lc_rev, undef) = ::extract_metadata($1);
1077                }
1078        }
1079        close $log;
1080
1081        Git::SVN::Log::set_local_timezone();
1082
1083        $result .= "Last Changed Author: $lc_author\n";
1084        $result .= "Last Changed Rev: $lc_rev\n";
1085        $result .= "Last Changed Date: " .
1086                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1087
1088        if ($file_type ne "dir") {
1089                my $text_last_updated_date =
1090                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1091                $result .=
1092                    "Text Last Updated: " .
1093                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
1094                    "\n";
1095                my $checksum;
1096                if ($diff_status eq "D") {
1097                        my ($fh, $ctx) =
1098                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
1099                        if ($file_type eq "link") {
1100                                my $file_name = <$fh>;
1101                                $checksum = md5sum("link $file_name");
1102                        } else {
1103                                $checksum = md5sum($fh);
1104                        }
1105                        command_close_pipe($fh, $ctx);
1106                } elsif ($file_type eq "link") {
1107                        my $file_name =
1108                            command(qw(cat-file blob), "HEAD:$path");
1109                        $checksum =
1110                            md5sum("link " . $file_name);
1111                } else {
1112                        open FILE, "<", $path or die $!;
1113                        $checksum = md5sum(\*FILE);
1114                        close FILE or die $!;
1115                }
1116                $result .= "Checksum: " . $checksum . "\n";
1117        }
1118
1119        print $result, "\n";
1120}
1121
1122sub cmd_reset {
1123        my $target = shift || $_revision or die "SVN revision required\n";
1124        $target = $1 if $target =~ /^r(\d+)$/;
1125        $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1126        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1127        unless ($gs) {
1128                die "Unable to determine upstream SVN information from ".
1129                    "history\n";
1130        }
1131        my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1132        $gs->rev_map_set($r, $c, 'reset', $uuid);
1133        print "r$r = $c ($gs->{ref_id})\n";
1134}
1135
1136sub cmd_gc {
1137        if (!$can_compress) {
1138                warn "Compress::Zlib could not be found; unhandled.log " .
1139                     "files will not be compressed.\n";
1140        }
1141        find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1142}
1143
1144########################### utility functions #########################
1145
1146sub rebase_cmd {
1147        my @cmd = qw/rebase/;
1148        push @cmd, '-v' if $_verbose;
1149        push @cmd, qw/--merge/ if $_merge;
1150        push @cmd, "--strategy=$_strategy" if $_strategy;
1151        @cmd;
1152}
1153
1154sub post_fetch_checkout {
1155        return if $_no_checkout;
1156        my $gs = $Git::SVN::_head or return;
1157        return if verify_ref('refs/heads/master^0');
1158
1159        # look for "trunk" ref if it exists
1160        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1161        my $fetch = $remote->{fetch};
1162        if ($fetch) {
1163                foreach my $p (keys %$fetch) {
1164                        basename($fetch->{$p}) eq 'trunk' or next;
1165                        $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1166                        last;
1167                }
1168        }
1169
1170        my $valid_head = verify_ref('HEAD^0');
1171        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1172        return if ($valid_head || !verify_ref('HEAD^0'));
1173
1174        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1175        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1176        return if -f $index;
1177
1178        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1179        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1180        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1181        print STDERR "Checked out HEAD:\n  ",
1182                     $gs->full_url, " r", $gs->last_rev, "\n";
1183}
1184
1185sub complete_svn_url {
1186        my ($url, $path) = @_;
1187        $path =~ s#/+$##;
1188        if ($path !~ m#^[a-z\+]+://#) {
1189                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1190                        fatal("E: '$path' is not a complete URL ",
1191                              "and a separate URL is not specified");
1192                }
1193                return ($url, $path);
1194        }
1195        return ($path, '');
1196}
1197
1198sub complete_url_ls_init {
1199        my ($ra, $repo_path, $switch, $pfx) = @_;
1200        unless ($repo_path) {
1201                print STDERR "W: $switch not specified\n";
1202                return;
1203        }
1204        $repo_path =~ s#/+$##;
1205        if ($repo_path =~ m#^[a-z\+]+://#) {
1206                $ra = Git::SVN::Ra->new($repo_path);
1207                $repo_path = '';
1208        } else {
1209                $repo_path =~ s#^/+##;
1210                unless ($ra) {
1211                        fatal("E: '$repo_path' is not a complete URL ",
1212                              "and a separate URL is not specified");
1213                }
1214        }
1215        my $url = $ra->{url};
1216        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1217        my $k = "svn-remote.$gs->{repo_id}.url";
1218        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1219        if ($orig_url && ($orig_url ne $gs->{url})) {
1220                die "$k already set: $orig_url\n",
1221                    "wanted to set to: $gs->{url}\n";
1222        }
1223        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1224        my $remote_path = "$gs->{path}/$repo_path";
1225        $remote_path =~ s#/+#/#g;
1226        $remote_path =~ s#^/##g;
1227        $remote_path .= "/*" if $remote_path !~ /\*/;
1228        my ($n) = ($switch =~ /^--(\w+)/);
1229        if (length $pfx && $pfx !~ m#/$#) {
1230                die "--prefix='$pfx' must have a trailing slash '/'\n";
1231        }
1232        command_noisy('config',
1233                      '--add',
1234                      "svn-remote.$gs->{repo_id}.$n",
1235                      "$remote_path:refs/remotes/$pfx*" .
1236                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1237}
1238
1239sub verify_ref {
1240        my ($ref) = @_;
1241        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1242                               { STDERR => 0 }); };
1243}
1244
1245sub get_tree_from_treeish {
1246        my ($treeish) = @_;
1247        # $treeish can be a symbolic ref, too:
1248        my $type = command_oneline(qw/cat-file -t/, $treeish);
1249        my $expected;
1250        while ($type eq 'tag') {
1251                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1252        }
1253        if ($type eq 'commit') {
1254                $expected = (grep /^tree /, command(qw/cat-file commit/,
1255                                                    $treeish))[0];
1256                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1257                die "Unable to get tree from $treeish\n" unless $expected;
1258        } elsif ($type eq 'tree') {
1259                $expected = $treeish;
1260        } else {
1261                die "$treeish is a $type, expected tree, tag or commit\n";
1262        }
1263        return $expected;
1264}
1265
1266sub get_commit_entry {
1267        my ($treeish) = shift;
1268        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1269        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1270        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1271        open my $log_fh, '>', $commit_editmsg or croak $!;
1272
1273        my $type = command_oneline(qw/cat-file -t/, $treeish);
1274        if ($type eq 'commit' || $type eq 'tag') {
1275                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1276                                                         $type, $treeish);
1277                my $in_msg = 0;
1278                my $author;
1279                my $saw_from = 0;
1280                my $msgbuf = "";
1281                while (<$msg_fh>) {
1282                        if (!$in_msg) {
1283                                $in_msg = 1 if (/^\s*$/);
1284                                $author = $1 if (/^author (.*>)/);
1285                        } elsif (/^git-svn-id: /) {
1286                                # skip this for now, we regenerate the
1287                                # correct one on re-fetch anyways
1288                                # TODO: set *:merge properties or like...
1289                        } else {
1290                                if (/^From:/ || /^Signed-off-by:/) {
1291                                        $saw_from = 1;
1292                                }
1293                                $msgbuf .= $_;
1294                        }
1295                }
1296                $msgbuf =~ s/\s+$//s;
1297                if ($Git::SVN::_add_author_from && defined($author)
1298                    && !$saw_from) {
1299                        $msgbuf .= "\n\nFrom: $author";
1300                }
1301                print $log_fh $msgbuf or croak $!;
1302                command_close_pipe($msg_fh, $ctx);
1303        }
1304        close $log_fh or croak $!;
1305
1306        if ($_edit || ($type eq 'tree')) {
1307                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1308                # TODO: strip out spaces, comments, like git-commit.sh
1309                system($editor, $commit_editmsg);
1310        }
1311        rename $commit_editmsg, $commit_msg or croak $!;
1312        {
1313                require Encode;
1314                # SVN requires messages to be UTF-8 when entering the repo
1315                local $/;
1316                open $log_fh, '<', $commit_msg or croak $!;
1317                binmode $log_fh;
1318                chomp($log_entry{log} = <$log_fh>);
1319
1320                my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1321                my $msg = $log_entry{log};
1322
1323                eval { $msg = Encode::decode($enc, $msg, 1) };
1324                if ($@) {
1325                        die "Could not decode as $enc:\n", $msg,
1326                            "\nPerhaps you need to set i18n.commitencoding\n";
1327                }
1328
1329                eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1330                die "Could not encode as UTF-8:\n$msg\n" if $@;
1331
1332                $log_entry{log} = $msg;
1333
1334                close $log_fh or croak $!;
1335        }
1336        unlink $commit_msg;
1337        \%log_entry;
1338}
1339
1340sub s_to_file {
1341        my ($str, $file, $mode) = @_;
1342        open my $fd,'>',$file or croak $!;
1343        print $fd $str,"\n" or croak $!;
1344        close $fd or croak $!;
1345        chmod ($mode &~ umask, $file) if (defined $mode);
1346}
1347
1348sub file_to_s {
1349        my $file = shift;
1350        open my $fd,'<',$file or croak "$!: file: $file\n";
1351        local $/;
1352        my $ret = <$fd>;
1353        close $fd or croak $!;
1354        $ret =~ s/\s*$//s;
1355        return $ret;
1356}
1357
1358# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1359sub load_authors {
1360        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1361        my $log = $cmd eq 'log';
1362        while (<$authors>) {
1363                chomp;
1364                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1365                my ($user, $name, $email) = ($1, $2, $3);
1366                if ($log) {
1367                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1368                } else {
1369                        $users{$user} = [$name, $email];
1370                }
1371        }
1372        close $authors or croak $!;
1373}
1374
1375# convert GetOpt::Long specs for use by git-config
1376sub read_repo_config {
1377        return unless -d $ENV{GIT_DIR};
1378        my $opts = shift;
1379        my @config_only;
1380        foreach my $o (keys %$opts) {
1381                # if we have mixedCase and a long option-only, then
1382                # it's a config-only variable that we don't need for
1383                # the command-line.
1384                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1385                my $v = $opts->{$o};
1386                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1387                $key =~ s/-//g;
1388                my $arg = 'git config';
1389                $arg .= ' --int' if ($o =~ /[:=]i$/);
1390                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1391                if (ref $v eq 'ARRAY') {
1392                        chomp(my @tmp = `$arg --get-all svn.$key`);
1393                        @$v = @tmp if @tmp;
1394                } else {
1395                        chomp(my $tmp = `$arg --get svn.$key`);
1396                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1397                                $$v = $tmp;
1398                        }
1399                }
1400        }
1401        delete @$opts{@config_only} if @config_only;
1402}
1403
1404sub extract_metadata {
1405        my $id = shift or return (undef, undef, undef);
1406        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1407                                                        \s([a-f\d\-]+)$/ix);
1408        if (!defined $rev || !$uuid || !$url) {
1409                # some of the original repositories I made had
1410                # identifiers like this:
1411                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1412        }
1413        return ($url, $rev, $uuid);
1414}
1415
1416sub cmt_metadata {
1417        return extract_metadata((grep(/^git-svn-id: /,
1418                command(qw/cat-file commit/, shift)))[-1]);
1419}
1420
1421sub cmt_sha2rev_batch {
1422        my %s2r;
1423        my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1424        my $list = shift;
1425
1426        foreach my $sha (@{$list}) {
1427                my $first = 1;
1428                my $size = 0;
1429                print $out $sha, "\n";
1430
1431                while (my $line = <$in>) {
1432                        if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1433                                last;
1434                        } elsif ($first &&
1435                               $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1436                                $first = 0;
1437                                $size = $1;
1438                                next;
1439                        } elsif ($line =~ /^(git-svn-id: )/) {
1440                                my (undef, $rev, undef) =
1441                                                      extract_metadata($line);
1442                                $s2r{$sha} = $rev;
1443                        }
1444
1445                        $size -= length($line);
1446                        last if ($size == 0);
1447                }
1448        }
1449
1450        command_close_bidi_pipe($pid, $in, $out, $ctx);
1451
1452        return \%s2r;
1453}
1454
1455sub working_head_info {
1456        my ($head, $refs) = @_;
1457        my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1458        my ($fh, $ctx) = command_output_pipe(@args, $head);
1459        my $hash;
1460        my %max;
1461        while (<$fh>) {
1462                if ( m{^commit ($::sha1)$} ) {
1463                        unshift @$refs, $hash if $hash and $refs;
1464                        $hash = $1;
1465                        next;
1466                }
1467                next unless s{^\s*(git-svn-id:)}{$1};
1468                my ($url, $rev, $uuid) = extract_metadata($_);
1469                if (defined $url && defined $rev) {
1470                        next if $max{$url} and $max{$url} < $rev;
1471                        if (my $gs = Git::SVN->find_by_url($url)) {
1472                                my $c = $gs->rev_map_get($rev, $uuid);
1473                                if ($c && $c eq $hash) {
1474                                        close $fh; # break the pipe
1475                                        return ($url, $rev, $uuid, $gs);
1476                                } else {
1477                                        $max{$url} ||= $gs->rev_map_max;
1478                                }
1479                        }
1480                }
1481        }
1482        command_close_pipe($fh, $ctx);
1483        (undef, undef, undef, undef);
1484}
1485
1486sub read_commit_parents {
1487        my ($parents, $c) = @_;
1488        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1489        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1490        @{$parents->{$c}} = split(/ /, $p);
1491}
1492
1493sub linearize_history {
1494        my ($gs, $refs) = @_;
1495        my %parents;
1496        foreach my $c (@$refs) {
1497                read_commit_parents(\%parents, $c);
1498        }
1499
1500        my @linear_refs;
1501        my %skip = ();
1502        my $last_svn_commit = $gs->last_commit;
1503        foreach my $c (reverse @$refs) {
1504                next if $c eq $last_svn_commit;
1505                last if $skip{$c};
1506
1507                unshift @linear_refs, $c;
1508                $skip{$c} = 1;
1509
1510                # we only want the first parent to diff against for linear
1511                # history, we save the rest to inject when we finalize the
1512                # svn commit
1513                my $fp_a = verify_ref("$c~1");
1514                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1515                if (!$fp_a || !$fp_b) {
1516                        die "Commit $c\n",
1517                            "has no parent commit, and therefore ",
1518                            "nothing to diff against.\n",
1519                            "You should be working from a repository ",
1520                            "originally created by git-svn\n";
1521                }
1522                if ($fp_a ne $fp_b) {
1523                        die "$c~1 = $fp_a, however parsing commit $c ",
1524                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1525                }
1526
1527                foreach my $p (@{$parents{$c}}) {
1528                        $skip{$p} = 1;
1529                }
1530        }
1531        (\@linear_refs, \%parents);
1532}
1533
1534sub find_file_type_and_diff_status {
1535        my ($path) = @_;
1536        return ('dir', '') if $path eq '';
1537
1538        my $diff_output =
1539            command_oneline(qw(diff --cached --name-status --), $path) || "";
1540        my $diff_status = (split(' ', $diff_output))[0] || "";
1541
1542        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1543
1544        return (undef, undef) if !$diff_status && !$ls_tree;
1545
1546        if ($diff_status eq "A") {
1547                return ("link", $diff_status) if -l $path;
1548                return ("dir", $diff_status) if -d $path;
1549                return ("file", $diff_status);
1550        }
1551
1552        my $mode = (split(' ', $ls_tree))[0] || "";
1553
1554        return ("link", $diff_status) if $mode eq "120000";
1555        return ("dir", $diff_status) if $mode eq "040000";
1556        return ("file", $diff_status);
1557}
1558
1559sub md5sum {
1560        my $arg = shift;
1561        my $ref = ref $arg;
1562        my $md5 = Digest::MD5->new();
1563        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1564                $md5->addfile($arg) or croak $!;
1565        } elsif ($ref eq 'SCALAR') {
1566                $md5->add($$arg) or croak $!;
1567        } elsif (!$ref) {
1568                $md5->add($arg) or croak $!;
1569        } else {
1570                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1571        }
1572        return $md5->hexdigest();
1573}
1574
1575sub gc_directory {
1576        if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1577                my $out_filename = $_ . ".gz";
1578                open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1579                binmode $in_fh;
1580                my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1581                                die "Unable to open $out_filename: $!\n";
1582
1583                my $res;
1584                while ($res = sysread($in_fh, my $str, 1024)) {
1585                        $gz->gzwrite($str) or
1586                                die "Unable to write: ".$gz->gzerror()."!\n";
1587                }
1588                unlink $_ or die "unlink $File::Find::name: $!\n";
1589        } elsif (-f $_ && basename($_) eq "index") {
1590                unlink $_ or die "unlink $_: $!\n";
1591        }
1592}
1593
1594package Git::SVN;
1595use strict;
1596use warnings;
1597use Fcntl qw/:DEFAULT :seek/;
1598use constant rev_map_fmt => 'NH40';
1599use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1600            $_repack $_repack_flags $_use_svm_props $_head
1601            $_use_svnsync_props $no_reuse_existing $_minimize_url
1602            $_use_log_author $_add_author_from $_localtime/;
1603use Carp qw/croak/;
1604use File::Path qw/mkpath/;
1605use File::Copy qw/copy/;
1606use IPC::Open3;
1607
1608my ($_gc_nr, $_gc_period);
1609
1610# properties that we do not log:
1611my %SKIP_PROP;
1612BEGIN {
1613        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1614                                        svn:special svn:executable
1615                                        svn:entry:committed-rev
1616                                        svn:entry:last-author
1617                                        svn:entry:uuid
1618                                        svn:entry:committed-date/;
1619
1620        # some options are read globally, but can be overridden locally
1621        # per [svn-remote "..."] section.  Command-line options will *NOT*
1622        # override options set in an [svn-remote "..."] section
1623        no strict 'refs';
1624        for my $option (qw/follow_parent no_metadata use_svm_props
1625                           use_svnsync_props/) {
1626                my $key = $option;
1627                $key =~ tr/_//d;
1628                my $prop = "-$option";
1629                *$option = sub {
1630                        my ($self) = @_;
1631                        return $self->{$prop} if exists $self->{$prop};
1632                        my $k = "svn-remote.$self->{repo_id}.$key";
1633                        eval { command_oneline(qw/config --get/, $k) };
1634                        if ($@) {
1635                                $self->{$prop} = ${"Git::SVN::_$option"};
1636                        } else {
1637                                my $v = command_oneline(qw/config --bool/,$k);
1638                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1639                        }
1640                        return $self->{$prop};
1641                }
1642        }
1643}
1644
1645
1646my (%LOCKFILES, %INDEX_FILES);
1647END {
1648        unlink keys %LOCKFILES if %LOCKFILES;
1649        unlink keys %INDEX_FILES if %INDEX_FILES;
1650}
1651
1652sub resolve_local_globs {
1653        my ($url, $fetch, $glob_spec) = @_;
1654        return unless defined $glob_spec;
1655        my $ref = $glob_spec->{ref};
1656        my $path = $glob_spec->{path};
1657        foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1658                next unless m#^$ref->{regex}$#;
1659                my $p = $1;
1660                my $pathname = desanitize_refname($path->full_path($p));
1661                my $refname = desanitize_refname($ref->full_path($p));
1662                if (my $existing = $fetch->{$pathname}) {
1663                        if ($existing ne $refname) {
1664                                die "Refspec conflict:\n",
1665                                    "existing: $existing\n",
1666                                    " globbed: $refname\n";
1667                        }
1668                        my $u = (::cmt_metadata("$refname"))[0];
1669                        $u =~ s!^\Q$url\E(/|$)!! or die
1670                          "$refname: '$url' not found in '$u'\n";
1671                        if ($pathname ne $u) {
1672                                warn "W: Refspec glob conflict ",
1673                                     "(ref: $refname):\n",
1674                                     "expected path: $pathname\n",
1675                                     "    real path: $u\n",
1676                                     "Continuing ahead with $u\n";
1677                                next;
1678                        }
1679                } else {
1680                        $fetch->{$pathname} = $refname;
1681                }
1682        }
1683}
1684
1685sub parse_revision_argument {
1686        my ($base, $head) = @_;
1687        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1688                return ($base, $head);
1689        }
1690        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1691        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1692        return ($head, $head) if ($::_revision eq 'HEAD');
1693        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1694        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1695        die "revision argument: $::_revision not understood by git-svn\n";
1696}
1697
1698sub fetch_all {
1699        my ($repo_id, $remotes) = @_;
1700        if (ref $repo_id) {
1701                my $gs = $repo_id;
1702                $repo_id = undef;
1703                $repo_id = $gs->{repo_id};
1704        }
1705        $remotes ||= read_all_remotes();
1706        my $remote = $remotes->{$repo_id} or
1707                     die "[svn-remote \"$repo_id\"] unknown\n";
1708        my $fetch = $remote->{fetch};
1709        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1710        my (@gs, @globs);
1711        my $ra = Git::SVN::Ra->new($url);
1712        my $uuid = $ra->get_uuid;
1713        my $head = $ra->get_latest_revnum;
1714        $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1715        my $base = defined $fetch ? $head : 0;
1716
1717        # read the max revs for wildcard expansion (branches/*, tags/*)
1718        foreach my $t (qw/branches tags/) {
1719                defined $remote->{$t} or next;
1720                push @globs, @{$remote->{$t}};
1721
1722                my $max_rev = eval { tmp_config(qw/--int --get/,
1723                                         "svn-remote.$repo_id.${t}-maxRev") };
1724                if (defined $max_rev && ($max_rev < $base)) {
1725                        $base = $max_rev;
1726                } elsif (!defined $max_rev) {
1727                        $base = 0;
1728                }
1729        }
1730
1731        if ($fetch) {
1732                foreach my $p (sort keys %$fetch) {
1733                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1734                        my $lr = $gs->rev_map_max;
1735                        if (defined $lr) {
1736                                $base = $lr if ($lr < $base);
1737                        }
1738                        push @gs, $gs;
1739                }
1740        }
1741
1742        ($base, $head) = parse_revision_argument($base, $head);
1743        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1744}
1745
1746sub read_all_remotes {
1747        my $r = {};
1748        my $use_svm_props = eval { command_oneline(qw/config --bool
1749            svn.useSvmProps/) };
1750        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1751        my $svn_refspec = qr{\s*/?(.*?)\s*:\s*(.+?)\s*};
1752        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1753                if (m!^(.+)\.fetch=$svn_refspec$!) {
1754                        my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1755                        die("svn-remote.$remote: remote ref '$remote_ref' "
1756                            . "must start with 'refs/'\n")
1757                                unless $remote_ref =~ m{^refs/};
1758                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1759                        $r->{$remote}->{svm} = {} if $use_svm_props;
1760                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1761                        $r->{$1}->{svm} = {};
1762                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1763                        $r->{$1}->{url} = $2;
1764                } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1765                        my ($remote, $t, $local_ref, $remote_ref) =
1766                                                             ($1, $2, $3, $4);
1767                        die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1768                            . "must start with 'refs/'\n")
1769                                unless $remote_ref =~ m{^refs/};
1770                        my $rs = {
1771                            t => $t,
1772                            remote => $remote,
1773                            path => Git::SVN::GlobSpec->new($local_ref),
1774                            ref => Git::SVN::GlobSpec->new($remote_ref) };
1775                        if (length($rs->{ref}->{right}) != 0) {
1776                                die "The '*' glob character must be the last ",
1777                                    "character of '$remote_ref'\n";
1778                        }
1779                        push @{ $r->{$remote}->{$t} }, $rs;
1780                }
1781        }
1782
1783        map {
1784                if (defined $r->{$_}->{svm}) {
1785                        my $svm;
1786                        eval {
1787                                my $section = "svn-remote.$_";
1788                                $svm = {
1789                                        source => tmp_config('--get',
1790                                            "$section.svm-source"),
1791                                        replace => tmp_config('--get',
1792                                            "$section.svm-replace"),
1793                                }
1794                        };
1795                        $r->{$_}->{svm} = $svm;
1796                }
1797        } keys %$r;
1798
1799        $r;
1800}
1801
1802sub init_vars {
1803        $_gc_nr = $_gc_period = 1000;
1804        if (defined $_repack || defined $_repack_flags) {
1805               warn "Repack options are obsolete; they have no effect.\n";
1806        }
1807}
1808
1809sub verify_remotes_sanity {
1810        return unless -d $ENV{GIT_DIR};
1811        my %seen;
1812        foreach (command(qw/config -l/)) {
1813                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1814                        if ($seen{$1}) {
1815                                die "Remote ref refs/remote/$1 is tracked by",
1816                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1817                                    "Please resolve this ambiguity in ",
1818                                    "your git configuration file before ",
1819                                    "continuing\n";
1820                        }
1821                        $seen{$1} = $_;
1822                }
1823        }
1824}
1825
1826sub find_existing_remote {
1827        my ($url, $remotes) = @_;
1828        return undef if $no_reuse_existing;
1829        my $existing;
1830        foreach my $repo_id (keys %$remotes) {
1831                my $u = $remotes->{$repo_id}->{url} or next;
1832                next if $u ne $url;
1833                $existing = $repo_id;
1834                last;
1835        }
1836        $existing;
1837}
1838
1839sub init_remote_config {
1840        my ($self, $url, $no_write) = @_;
1841        $url =~ s!/+$!!; # strip trailing slash
1842        my $r = read_all_remotes();
1843        my $existing = find_existing_remote($url, $r);
1844        if ($existing) {
1845                unless ($no_write) {
1846                        print STDERR "Using existing ",
1847                                     "[svn-remote \"$existing\"]\n";
1848                }
1849                $self->{repo_id} = $existing;
1850        } elsif ($_minimize_url) {
1851                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1852                $existing = find_existing_remote($min_url, $r);
1853                if ($existing) {
1854                        unless ($no_write) {
1855                                print STDERR "Using existing ",
1856                                             "[svn-remote \"$existing\"]\n";
1857                        }
1858                        $self->{repo_id} = $existing;
1859                }
1860                if ($min_url ne $url) {
1861                        unless ($no_write) {
1862                                print STDERR "Using higher level of URL: ",
1863                                             "$url => $min_url\n";
1864                        }
1865                        my $old_path = $self->{path};
1866                        $self->{path} = $url;
1867                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1868                        if (length $old_path) {
1869                                $self->{path} .= "/$old_path";
1870                        }
1871                        $url = $min_url;
1872                }
1873        }
1874        my $orig_url;
1875        if (!$existing) {
1876                # verify that we aren't overwriting anything:
1877                $orig_url = eval {
1878                        command_oneline('config', '--get',
1879                                        "svn-remote.$self->{repo_id}.url")
1880                };
1881                if ($orig_url && ($orig_url ne $url)) {
1882                        die "svn-remote.$self->{repo_id}.url already set: ",
1883                            "$orig_url\nwanted to set to: $url\n";
1884                }
1885        }
1886        my ($xrepo_id, $xpath) = find_ref($self->refname);
1887        if (!$no_write && defined $xpath) {
1888                die "svn-remote.$xrepo_id.fetch already set to track ",
1889                    "$xpath:", $self->refname, "\n";
1890        }
1891        unless ($no_write) {
1892                command_noisy('config',
1893                              "svn-remote.$self->{repo_id}.url", $url);
1894                $self->{path} =~ s{^/}{};
1895                command_noisy('config', '--add',
1896                              "svn-remote.$self->{repo_id}.fetch",
1897                              "$self->{path}:".$self->refname);
1898        }
1899        $self->{url} = $url;
1900}
1901
1902sub find_by_url { # repos_root and, path are optional
1903        my ($class, $full_url, $repos_root, $path) = @_;
1904
1905        return undef unless defined $full_url;
1906        remove_username($full_url);
1907        remove_username($repos_root) if defined $repos_root;
1908        my $remotes = read_all_remotes();
1909        if (defined $full_url && defined $repos_root && !defined $path) {
1910                $path = $full_url;
1911                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1912        }
1913        foreach my $repo_id (keys %$remotes) {
1914                my $u = $remotes->{$repo_id}->{url} or next;
1915                remove_username($u);
1916                next if defined $repos_root && $repos_root ne $u;
1917
1918                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1919                foreach my $t (qw/branches tags/) {
1920                        foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1921                                resolve_local_globs($u, $fetch, $globspec);
1922                        }
1923                }
1924                my $p = $path;
1925                my $rwr = rewrite_root({repo_id => $repo_id});
1926                my $svm = $remotes->{$repo_id}->{svm}
1927                        if defined $remotes->{$repo_id}->{svm};
1928                unless (defined $p) {
1929                        $p = $full_url;
1930                        my $z = $u;
1931                        my $prefix = '';
1932                        if ($rwr) {
1933                                $z = $rwr;
1934                                remove_username($z);
1935                        } elsif (defined $svm) {
1936                                $z = $svm->{source};
1937                                $prefix = $svm->{replace};
1938                                $prefix =~ s#^\Q$u\E(?:/|$)##;
1939                                $prefix =~ s#/$##;
1940                        }
1941                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1942                }
1943                foreach my $f (keys %$fetch) {
1944                        next if $f ne $p;
1945                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1946                }
1947        }
1948        undef;
1949}
1950
1951sub init {
1952        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1953        my $self = _new($class, $repo_id, $ref_id, $path);
1954        if (defined $url) {
1955                $self->init_remote_config($url, $no_write);
1956        }
1957        $self;
1958}
1959
1960sub find_ref {
1961        my ($ref_id) = @_;
1962        foreach (command(qw/config -l/)) {
1963                next unless m!^svn-remote\.(.+)\.fetch=
1964                              \s*/?(.*?)\s*:\s*(.+?)\s*$!x;
1965                my ($repo_id, $path, $ref) = ($1, $2, $3);
1966                if ($ref eq $ref_id) {
1967                        $path = '' if ($path =~ m#^\./?#);
1968                        return ($repo_id, $path);
1969                }
1970        }
1971        (undef, undef, undef);
1972}
1973
1974sub new {
1975        my ($class, $ref_id, $repo_id, $path) = @_;
1976        if (defined $ref_id && !defined $repo_id && !defined $path) {
1977                ($repo_id, $path) = find_ref($ref_id);
1978                if (!defined $repo_id) {
1979                        die "Could not find a \"svn-remote.*.fetch\" key ",
1980                            "in the repository configuration matching: ",
1981                            "$ref_id\n";
1982                }
1983        }
1984        my $self = _new($class, $repo_id, $ref_id, $path);
1985        if (!defined $self->{path} || !length $self->{path}) {
1986                my $fetch = command_oneline('config', '--get',
1987                                            "svn-remote.$repo_id.fetch",
1988                                            ":$ref_id\$") or
1989                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1990                         "\":$ref_id\$\" in config\n";
1991                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1992        }
1993        $self->{url} = command_oneline('config', '--get',
1994                                       "svn-remote.$repo_id.url") or
1995                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1996        $self->rebuild;
1997        $self;
1998}
1999
2000sub refname {
2001        my ($refname) = $_[0]->{ref_id} ;
2002
2003        # It cannot end with a slash /, we'll throw up on this because
2004        # SVN can't have directories with a slash in their name, either:
2005        if ($refname =~ m{/$}) {
2006                die "ref: '$refname' ends with a trailing slash, this is ",
2007                    "not permitted by git nor Subversion\n";
2008        }
2009
2010        # It cannot have ASCII control character space, tilde ~, caret ^,
2011        # colon :, question-mark ?, asterisk *, space, or open bracket [
2012        # anywhere.
2013        #
2014        # Additionally, % must be escaped because it is used for escaping
2015        # and we want our escaped refname to be reversible
2016        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2017
2018        # no slash-separated component can begin with a dot .
2019        # /.* becomes /%2E*
2020        $refname =~ s{/\.}{/%2E}g;
2021
2022        # It cannot have two consecutive dots .. anywhere
2023        # .. becomes %2E%2E
2024        $refname =~ s{\.\.}{%2E%2E}g;
2025
2026        return $refname;
2027}
2028
2029sub desanitize_refname {
2030        my ($refname) = @_;
2031        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2032        return $refname;
2033}
2034
2035sub svm_uuid {
2036        my ($self) = @_;
2037        return $self->{svm}->{uuid} if $self->svm;
2038        $self->ra;
2039        unless ($self->{svm}) {
2040                die "SVM UUID not cached, and reading remotely failed\n";
2041        }
2042        $self->{svm}->{uuid};
2043}
2044
2045sub svm {
2046        my ($self) = @_;
2047        return $self->{svm} if $self->{svm};
2048        my $svm;
2049        # see if we have it in our config, first:
2050        eval {
2051                my $section = "svn-remote.$self->{repo_id}";
2052                $svm = {
2053                  source => tmp_config('--get', "$section.svm-source"),
2054                  uuid => tmp_config('--get', "$section.svm-uuid"),
2055                  replace => tmp_config('--get', "$section.svm-replace"),
2056                }
2057        };
2058        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2059                $self->{svm} = $svm;
2060        }
2061        $self->{svm};
2062}
2063
2064sub _set_svm_vars {
2065        my ($self, $ra) = @_;
2066        return $ra if $self->svm;
2067
2068        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2069                    "(svm:source, svm:uuid) ",
2070                    "from the following URLs:\n" );
2071        sub read_svm_props {
2072                my ($self, $ra, $path, $r) = @_;
2073                my $props = ($ra->get_dir($path, $r))[2];
2074                my $src = $props->{'svm:source'};
2075                my $uuid = $props->{'svm:uuid'};
2076                return undef if (!$src || !$uuid);
2077
2078                chomp($src, $uuid);
2079
2080                $uuid =~ m{^[0-9a-f\-]{30,}$}i
2081                    or die "doesn't look right - svm:uuid is '$uuid'\n";
2082
2083                # the '!' is used to mark the repos_root!/relative/path
2084                $src =~ s{/?!/?}{/};
2085                $src =~ s{/+$}{}; # no trailing slashes please
2086                # username is of no interest
2087                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2088
2089                my $replace = $ra->{url};
2090                $replace .= "/$path" if length $path;
2091
2092                my $section = "svn-remote.$self->{repo_id}";
2093                tmp_config("$section.svm-source", $src);
2094                tmp_config("$section.svm-replace", $replace);
2095                tmp_config("$section.svm-uuid", $uuid);
2096                $self->{svm} = {
2097                        source => $src,
2098                        uuid => $uuid,
2099                        replace => $replace
2100                };
2101        }
2102
2103        my $r = $ra->get_latest_revnum;
2104        my $path = $self->{path};
2105        my %tried;
2106        while (length $path) {
2107                unless ($tried{"$self->{url}/$path"}) {
2108                        return $ra if $self->read_svm_props($ra, $path, $r);
2109                        $tried{"$self->{url}/$path"} = 1;
2110                }
2111                $path =~ s#/?[^/]+$##;
2112        }
2113        die "Path: '$path' should be ''\n" if $path ne '';
2114        return $ra if $self->read_svm_props($ra, $path, $r);
2115        $tried{"$self->{url}/$path"} = 1;
2116
2117        if ($ra->{repos_root} eq $self->{url}) {
2118                die @err, (map { "  $_\n" } keys %tried), "\n";
2119        }
2120
2121        # nope, make sure we're connected to the repository root:
2122        my $ok;
2123        my @tried_b;
2124        $path = $ra->{svn_path};
2125        $ra = Git::SVN::Ra->new($ra->{repos_root});
2126        while (length $path) {
2127                unless ($tried{"$ra->{url}/$path"}) {
2128                        $ok = $self->read_svm_props($ra, $path, $r);
2129                        last if $ok;
2130                        $tried{"$ra->{url}/$path"} = 1;
2131                }
2132                $path =~ s#/?[^/]+$##;
2133        }
2134        die "Path: '$path' should be ''\n" if $path ne '';
2135        $ok ||= $self->read_svm_props($ra, $path, $r);
2136        $tried{"$ra->{url}/$path"} = 1;
2137        if (!$ok) {
2138                die @err, (map { "  $_\n" } keys %tried), "\n";
2139        }
2140        Git::SVN::Ra->new($self->{url});
2141}
2142
2143sub svnsync {
2144        my ($self) = @_;
2145        return $self->{svnsync} if $self->{svnsync};
2146
2147        if ($self->no_metadata) {
2148                die "Can't have both 'noMetadata' and ",
2149                    "'useSvnsyncProps' options set!\n";
2150        }
2151        if ($self->rewrite_root) {
2152                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2153                    "options set!\n";
2154        }
2155
2156        my $svnsync;
2157        # see if we have it in our config, first:
2158        eval {
2159                my $section = "svn-remote.$self->{repo_id}";
2160
2161                my $url = tmp_config('--get', "$section.svnsync-url");
2162                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2163                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2164
2165                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2166                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2167                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2168
2169                $svnsync = { url => $url, uuid => $uuid }
2170        };
2171        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2172                return $self->{svnsync} = $svnsync;
2173        }
2174
2175        my $err = "useSvnsyncProps set, but failed to read " .
2176                  "svnsync property: svn:sync-from-";
2177        my $rp = $self->ra->rev_proplist(0);
2178
2179        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2180        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2181                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2182
2183        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2184        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2185                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2186
2187        my $section = "svn-remote.$self->{repo_id}";
2188        tmp_config('--add', "$section.svnsync-uuid", $uuid);
2189        tmp_config('--add', "$section.svnsync-url", $url);
2190        return $self->{svnsync} = { url => $url, uuid => $uuid };
2191}
2192
2193# this allows us to memoize our SVN::Ra UUID locally and avoid a
2194# remote lookup (useful for 'git svn log').
2195sub ra_uuid {
2196        my ($self) = @_;
2197        unless ($self->{ra_uuid}) {
2198                my $key = "svn-remote.$self->{repo_id}.uuid";
2199                my $uuid = eval { tmp_config('--get', $key) };
2200                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2201                        $self->{ra_uuid} = $uuid;
2202                } else {
2203                        die "ra_uuid called without URL\n" unless $self->{url};
2204                        $self->{ra_uuid} = $self->ra->get_uuid;
2205                        tmp_config('--add', $key, $self->{ra_uuid});
2206                }
2207        }
2208        $self->{ra_uuid};
2209}
2210
2211sub _set_repos_root {
2212        my ($self, $repos_root) = @_;
2213        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2214        $repos_root ||= $self->ra->{repos_root};
2215        tmp_config($k, $repos_root);
2216        $repos_root;
2217}
2218
2219sub repos_root {
2220        my ($self) = @_;
2221        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2222        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2223}
2224
2225sub ra {
2226        my ($self) = shift;
2227        my $ra = Git::SVN::Ra->new($self->{url});
2228        $self->_set_repos_root($ra->{repos_root});
2229        if ($self->use_svm_props && !$self->{svm}) {
2230                if ($self->no_metadata) {
2231                        die "Can't have both 'noMetadata' and ",
2232                            "'useSvmProps' options set!\n";
2233                } elsif ($self->use_svnsync_props) {
2234                        die "Can't have both 'useSvnsyncProps' and ",
2235                            "'useSvmProps' options set!\n";
2236                }
2237                $ra = $self->_set_svm_vars($ra);
2238                $self->{-want_revprops} = 1;
2239        }
2240        $ra;
2241}
2242
2243# prop_walk(PATH, REV, SUB)
2244# -------------------------
2245# Recursively traverse PATH at revision REV and invoke SUB for each
2246# directory that contains a SVN property.  SUB will be invoked as
2247# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2248# Git::SVN, `path' the path to the directory where the properties
2249# `props' were found.  The `path' will be relative to point of checkout,
2250# that is, if url://repo/trunk is the current Git branch, and that
2251# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2252# as `path' (note the trailing `/').
2253sub prop_walk {
2254        my ($self, $path, $rev, $sub) = @_;
2255
2256        $path =~ s#^/##;
2257        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2258        $path =~ s#^/*#/#g;
2259        my $p = $path;
2260        # Strip the irrelevant part of the path.
2261        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2262        # Ensure the path is terminated by a `/'.
2263        $p =~ s#/*$#/#;
2264
2265        # The properties contain all the internal SVN stuff nobody
2266        # (usually) cares about.
2267        my $interesting_props = 0;
2268        foreach (keys %{$props}) {
2269                # If it doesn't start with `svn:', it must be a
2270                # user-defined property.
2271                ++$interesting_props and next if $_ !~ /^svn:/;
2272                # FIXME: Fragile, if SVN adds new public properties,
2273                # this needs to be updated.
2274                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2275                                                 |eol-style|mime-type
2276                                                 |externals|needs-lock)$/x;
2277        }
2278        &$sub($self, $p, $props) if $interesting_props;
2279
2280        foreach (sort keys %$dirent) {
2281                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2282                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2283        }
2284}
2285
2286sub last_rev { ($_[0]->last_rev_commit)[0] }
2287sub last_commit { ($_[0]->last_rev_commit)[1] }
2288
2289# returns the newest SVN revision number and newest commit SHA1
2290sub last_rev_commit {
2291        my ($self) = @_;
2292        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2293                return ($self->{last_rev}, $self->{last_commit});
2294        }
2295        my $c = ::verify_ref($self->refname.'^0');
2296        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2297                my $rev = (::cmt_metadata($c))[1];
2298                if (defined $rev) {
2299                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2300                        return ($rev, $c);
2301                }
2302        }
2303        my $map_path = $self->map_path;
2304        unless (-e $map_path) {
2305                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2306                return (undef, undef);
2307        }
2308        my ($rev, $commit) = $self->rev_map_max(1);
2309        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2310        return ($rev, $commit);
2311}
2312
2313sub get_fetch_range {
2314        my ($self, $min, $max) = @_;
2315        $max ||= $self->ra->get_latest_revnum;
2316        $min ||= $self->rev_map_max;
2317        (++$min, $max);
2318}
2319
2320sub tmp_config {
2321        my (@args) = @_;
2322        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2323        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2324        if (! -f $config && -f $old_def_config) {
2325                rename $old_def_config, $config or
2326                       die "Failed rename $old_def_config => $config: $!\n";
2327        }
2328        my $old_config = $ENV{GIT_CONFIG};
2329        $ENV{GIT_CONFIG} = $config;
2330        $@ = undef;
2331        my @ret = eval {
2332                unless (-f $config) {
2333                        mkfile($config);
2334                        open my $fh, '>', $config or
2335                            die "Can't open $config: $!\n";
2336                        print $fh "; This file is used internally by ",
2337                                  "git-svn\n" or die
2338                                  "Couldn't write to $config: $!\n";
2339                        print $fh "; You should not have to edit it\n" or
2340                              die "Couldn't write to $config: $!\n";
2341                        close $fh or die "Couldn't close $config: $!\n";
2342                }
2343                command('config', @args);
2344        };
2345        my $err = $@;
2346        if (defined $old_config) {
2347                $ENV{GIT_CONFIG} = $old_config;
2348        } else {
2349                delete $ENV{GIT_CONFIG};
2350        }
2351        die $err if $err;
2352        wantarray ? @ret : $ret[0];
2353}
2354
2355sub tmp_index_do {
2356        my ($self, $sub) = @_;
2357        my $old_index = $ENV{GIT_INDEX_FILE};
2358        $ENV{GIT_INDEX_FILE} = $self->{index};
2359        $@ = undef;
2360        my @ret = eval {
2361                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2362                mkpath([$dir]) unless -d $dir;
2363                &$sub;
2364        };
2365        my $err = $@;
2366        if (defined $old_index) {
2367                $ENV{GIT_INDEX_FILE} = $old_index;
2368        } else {
2369                delete $ENV{GIT_INDEX_FILE};
2370        }
2371        die $err if $err;
2372        wantarray ? @ret : $ret[0];
2373}
2374
2375sub assert_index_clean {
2376        my ($self, $treeish) = @_;
2377
2378        $self->tmp_index_do(sub {
2379                command_noisy('read-tree', $treeish) unless -e $self->{index};
2380                my $x = command_oneline('write-tree');
2381                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2382                           /^tree ($::sha1)/mo);
2383                return if $y eq $x;
2384
2385                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2386                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2387                command_noisy('read-tree', $treeish);
2388                $x = command_oneline('write-tree');
2389                if ($y ne $x) {
2390                        ::fatal "trees ($treeish) $y != $x\n",
2391                                "Something is seriously wrong...";
2392                }
2393        });
2394}
2395
2396sub get_commit_parents {
2397        my ($self, $log_entry) = @_;
2398        my (%seen, @ret, @tmp);
2399        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2400        if (my $ip = $self->{inject_parents}) {
2401                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2402                        push @tmp, $commit;
2403                }
2404        }
2405        if (my $cur = ::verify_ref($self->refname.'^0')) {
2406                push @tmp, $cur;
2407        }
2408        if (my $ipd = $self->{inject_parents_dcommit}) {
2409                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2410                        push @tmp, @$commit;
2411                }
2412        }
2413        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2414        while (my $p = shift @tmp) {
2415                next if $seen{$p};
2416                $seen{$p} = 1;
2417                push @ret, $p;
2418                # MAXPARENT is defined to 16 in commit-tree.c:
2419                last if @ret >= 16;
2420        }
2421        if (@tmp) {
2422                die "r$log_entry->{revision}: No room for parents:\n\t",
2423                    join("\n\t", @tmp), "\n";
2424        }
2425        @ret;
2426}
2427
2428sub rewrite_root {
2429        my ($self) = @_;
2430        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2431        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2432        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2433        if ($rwr) {
2434                $rwr =~ s#/+$##;
2435                if ($rwr !~ m#^[a-z\+]+://#) {
2436                        die "$rwr is not a valid URL (key: $k)\n";
2437                }
2438        }
2439        $self->{-rewrite_root} = $rwr;
2440}
2441
2442sub metadata_url {
2443        my ($self) = @_;
2444        ($self->rewrite_root || $self->{url}) .
2445           (length $self->{path} ? '/' . $self->{path} : '');
2446}
2447
2448sub full_url {
2449        my ($self) = @_;
2450        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2451}
2452
2453
2454sub set_commit_header_env {
2455        my ($log_entry) = @_;
2456        my %env;
2457        foreach my $ned (qw/NAME EMAIL DATE/) {
2458                foreach my $ac (qw/AUTHOR COMMITTER/) {
2459                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2460                }
2461        }
2462
2463        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2464        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2465        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2466
2467        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2468                                                ? $log_entry->{commit_name}
2469                                                : $log_entry->{name};
2470        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2471                                                ? $log_entry->{commit_email}
2472                                                : $log_entry->{email};
2473        \%env;
2474}
2475
2476sub restore_commit_header_env {
2477        my ($env) = @_;
2478        foreach my $ned (qw/NAME EMAIL DATE/) {
2479                foreach my $ac (qw/AUTHOR COMMITTER/) {
2480                        my $k = "GIT_${ac}_${ned}";
2481                        if (defined $env->{$k}) {
2482                                $ENV{$k} = $env->{$k};
2483                        } else {
2484                                delete $ENV{$k};
2485                        }
2486                }
2487        }
2488}
2489
2490sub gc {
2491        command_noisy('gc', '--auto');
2492};
2493
2494sub do_git_commit {
2495        my ($self, $log_entry) = @_;
2496        my $lr = $self->last_rev;
2497        if (defined $lr && $lr >= $log_entry->{revision}) {
2498                die "Last fetched revision of ", $self->refname,
2499                    " was r$lr, but we are about to fetch: ",
2500                    "r$log_entry->{revision}!\n";
2501        }
2502        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2503                croak "$log_entry->{revision} = $c already exists! ",
2504                      "Why are we refetching it?\n";
2505        }
2506        my $old_env = set_commit_header_env($log_entry);
2507        my $tree = $log_entry->{tree};
2508        if (!defined $tree) {
2509                $tree = $self->tmp_index_do(sub {
2510                                            command_oneline('write-tree') });
2511        }
2512        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2513
2514        my @exec = ('git', 'commit-tree', $tree);
2515        foreach ($self->get_commit_parents($log_entry)) {
2516                push @exec, '-p', $_;
2517        }
2518        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2519                                                                   or croak $!;
2520        binmode $msg_fh;
2521
2522        # we always get UTF-8 from SVN, but we may want our commits in
2523        # a different encoding.
2524        if (my $enc = Git::config('i18n.commitencoding')) {
2525                require Encode;
2526                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2527        }
2528        print $msg_fh $log_entry->{log} or croak $!;
2529        restore_commit_header_env($old_env);
2530        unless ($self->no_metadata) {
2531                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2532                              or croak $!;
2533        }
2534        $msg_fh->flush == 0 or croak $!;
2535        close $msg_fh or croak $!;
2536        chomp(my $commit = do { local $/; <$out_fh> });
2537        close $out_fh or croak $!;
2538        waitpid $pid, 0;
2539        croak $? if $?;
2540        if ($commit !~ /^$::sha1$/o) {
2541                die "Failed to commit, invalid sha1: $commit\n";
2542        }
2543
2544        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2545
2546        $self->{last_rev} = $log_entry->{revision};
2547        $self->{last_commit} = $commit;
2548        print "r$log_entry->{revision}" unless $::_q > 1;
2549        if (defined $log_entry->{svm_revision}) {
2550                 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2551                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2552                                   0, $self->svm_uuid);
2553        }
2554        print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2555        if (--$_gc_nr == 0) {
2556                $_gc_nr = $_gc_period;
2557                gc();
2558        }
2559        return $commit;
2560}
2561
2562sub match_paths {
2563        my ($self, $paths, $r) = @_;
2564        return 1 if $self->{path} eq '';
2565        if (my $path = $paths->{"/$self->{path}"}) {
2566                return ($path->{action} eq 'D') ? 0 : 1;
2567        }
2568        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2569        if (grep /$self->{path_regex}/, keys %$paths) {
2570                return 1;
2571        }
2572        my $c = '';
2573        foreach (split m#/#, $self->{path}) {
2574                $c .= "/$_";
2575                next unless ($paths->{$c} &&
2576                             ($paths->{$c}->{action} =~ /^[AR]$/));
2577                if ($self->ra->check_path($self->{path}, $r) ==
2578                    $SVN::Node::dir) {
2579                        return 1;
2580                }
2581        }
2582        return 0;
2583}
2584
2585sub find_parent_branch {
2586        my ($self, $paths, $rev) = @_;
2587        return undef unless $self->follow_parent;
2588        unless (defined $paths) {
2589                my $err_handler = $SVN::Error::handler;
2590                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2591                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2592                                   sub { $paths = $_[0] });
2593                $SVN::Error::handler = $err_handler;
2594        }
2595        return undef unless defined $paths;
2596
2597        # look for a parent from another branch:
2598        my @b_path_components = split m#/#, $self->{path};
2599        my @a_path_components;
2600        my $i;
2601        while (@b_path_components) {
2602                $i = $paths->{'/'.join('/', @b_path_components)};
2603                last if $i && defined $i->{copyfrom_path};
2604                unshift(@a_path_components, pop(@b_path_components));
2605        }
2606        return undef unless defined $i && defined $i->{copyfrom_path};
2607        my $branch_from = $i->{copyfrom_path};
2608        if (@a_path_components) {
2609                print STDERR "branch_from: $branch_from => ";
2610                $branch_from .= '/'.join('/', @a_path_components);
2611                print STDERR $branch_from, "\n";
2612        }
2613        my $r = $i->{copyfrom_rev};
2614        my $repos_root = $self->ra->{repos_root};
2615        my $url = $self->ra->{url};
2616        my $new_url = $url . $branch_from;
2617        print STDERR  "Found possible branch point: ",
2618                      "$new_url => ", $self->full_url, ", $r\n";
2619        $branch_from =~ s#^/##;
2620        my $gs = $self->other_gs($new_url, $url,
2621                                 $branch_from, $r, $self->{ref_id});
2622        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2623        {
2624                my ($base, $head);
2625                if (!defined $r0 || !defined $parent) {
2626                        ($base, $head) = parse_revision_argument(0, $r);
2627                } else {
2628                        if ($r0 < $r) {
2629                                $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2630                                        0, 1, sub { $base = $_[1] - 1 });
2631                        }
2632                }
2633                if (defined $base && $base <= $r) {
2634                        $gs->fetch($base, $r);
2635                }
2636                ($r0, $parent) = $gs->find_rev_before($r, 1);
2637        }
2638        if (defined $r0 && defined $parent) {
2639                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2640                my $ed;
2641                if ($self->ra->can_do_switch) {
2642                        $self->assert_index_clean($parent);
2643                        print STDERR "Following parent with do_switch\n";
2644                        # do_switch works with svn/trunk >= r22312, but that
2645                        # is not included with SVN 1.4.3 (the latest version
2646                        # at the moment), so we can't rely on it
2647                        $self->{last_rev} = $r0;
2648                        $self->{last_commit} = $parent;
2649                        $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2650                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2651                                              $self->full_url, $ed)
2652                          or die "SVN connection failed somewhere...\n";
2653                } elsif ($self->ra->trees_match($new_url, $r0,
2654                                                $self->full_url, $rev)) {
2655                        print STDERR "Trees match:\n",
2656                                     "  $new_url\@$r0\n",
2657                                     "  ${\$self->full_url}\@$rev\n",
2658                                     "Following parent with no changes\n";
2659                        $self->tmp_index_do(sub {
2660                            command_noisy('read-tree', $parent);
2661                        });
2662                        $self->{last_commit} = $parent;
2663                } else {
2664                        print STDERR "Following parent with do_update\n";
2665                        $ed = SVN::Git::Fetcher->new($self);
2666                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2667                          or die "SVN connection failed somewhere...\n";
2668                }
2669                print STDERR "Successfully followed parent\n";
2670                return $self->make_log_entry($rev, [$parent], $ed);
2671        }
2672        return undef;
2673}
2674
2675sub do_fetch {
2676        my ($self, $paths, $rev) = @_;
2677        my $ed;
2678        my ($last_rev, @parents);
2679        if (my $lc = $self->last_commit) {
2680                # we can have a branch that was deleted, then re-added
2681                # under the same name but copied from another path, in
2682                # which case we'll have multiple parents (we don't
2683                # want to break the original ref, nor lose copypath info):
2684                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2685                        push @{$log_entry->{parents}}, $lc;
2686                        return $log_entry;
2687                }
2688                $ed = SVN::Git::Fetcher->new($self);
2689                $last_rev = $self->{last_rev};
2690                $ed->{c} = $lc;
2691                @parents = ($lc);
2692        } else {
2693                $last_rev = $rev;
2694                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2695                        return $log_entry;
2696                }
2697                $ed = SVN::Git::Fetcher->new($self);
2698        }
2699        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2700                die "SVN connection failed somewhere...\n";
2701        }
2702        $self->make_log_entry($rev, \@parents, $ed);
2703}
2704
2705sub get_untracked {
2706        my ($self, $ed) = @_;
2707        my @out;
2708        my $h = $ed->{empty};
2709        foreach (sort keys %$h) {
2710                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2711                push @out, "  $act: " . uri_encode($_);
2712                warn "W: $act: $_\n";
2713        }
2714        foreach my $t (qw/dir_prop file_prop/) {
2715                $h = $ed->{$t} or next;
2716                foreach my $path (sort keys %$h) {
2717                        my $ppath = $path eq '' ? '.' : $path;
2718                        foreach my $prop (sort keys %{$h->{$path}}) {
2719                                next if $SKIP_PROP{$prop};
2720                                my $v = $h->{$path}->{$prop};
2721                                my $t_ppath_prop = "$t: " .
2722                                                    uri_encode($ppath) . ' ' .
2723                                                    uri_encode($prop);
2724                                if (defined $v) {
2725                                        push @out, "  +$t_ppath_prop " .
2726                                                   uri_encode($v);
2727                                } else {
2728                                        push @out, "  -$t_ppath_prop";
2729                                }
2730                        }
2731                }
2732        }
2733        foreach my $t (qw/absent_file absent_directory/) {
2734                $h = $ed->{$t} or next;
2735                foreach my $parent (sort keys %$h) {
2736                        foreach my $path (sort @{$h->{$parent}}) {
2737                                push @out, "  $t: " .
2738                                           uri_encode("$parent/$path");
2739                                warn "W: $t: $parent/$path ",
2740                                     "Insufficient permissions?\n";
2741                        }
2742                }
2743        }
2744        \@out;
2745}
2746
2747# parse_svn_date(DATE)
2748# --------------------
2749# Given a date (in UTC) from Subversion, return a string in the format
2750# "<TZ Offset> <local date/time>" that Git will use.
2751#
2752# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2753# is true we'll convert it to the local timezone instead.
2754sub parse_svn_date {
2755        my $date = shift || return '+0000 1970-01-01 00:00:00';
2756        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2757                                            (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2758                                         croak "Unable to parse date: $date\n";
2759        my $parsed_date;    # Set next.
2760
2761        if ($Git::SVN::_localtime) {
2762                # Translate the Subversion datetime to an epoch time.
2763                # Begin by switching ourselves to $date's timezone, UTC.
2764                my $old_env_TZ = $ENV{TZ};
2765                $ENV{TZ} = 'UTC';
2766
2767                my $epoch_in_UTC =
2768                    POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2769
2770                # Determine our local timezone (including DST) at the
2771                # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
2772                # value of TZ, if any, at the time we were run.
2773                if (defined $Git::SVN::Log::TZ) {
2774                        $ENV{TZ} = $Git::SVN::Log::TZ;
2775                } else {
2776                        delete $ENV{TZ};
2777                }
2778
2779                my $our_TZ =
2780                    POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2781
2782                # This converts $epoch_in_UTC into our local timezone.
2783                my ($sec, $min, $hour, $mday, $mon, $year,
2784                    $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2785
2786                $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2787                                       $our_TZ, $year + 1900, $mon + 1,
2788                                       $mday, $hour, $min, $sec);
2789
2790                # Reset us to the timezone in effect when we entered
2791                # this routine.
2792                if (defined $old_env_TZ) {
2793                        $ENV{TZ} = $old_env_TZ;
2794                } else {
2795                        delete $ENV{TZ};
2796                }
2797        } else {
2798                $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2799        }
2800
2801        return $parsed_date;
2802}
2803
2804sub other_gs {
2805        my ($self, $new_url, $url,
2806            $branch_from, $r, $old_ref_id) = @_;
2807        my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2808        unless ($gs) {
2809                my $ref_id = $old_ref_id;
2810                $ref_id =~ s/\@\d+$//;
2811                $ref_id .= "\@$r";
2812                # just grow a tail if we're not unique enough :x
2813                $ref_id .= '-' while find_ref($ref_id);
2814                print STDERR "Initializing parent: $ref_id\n";
2815                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2816                if ($u =~ s#^\Q$url\E(/|$)##) {
2817                        $p = $u;
2818                        $u = $url;
2819                        $repo_id = $self->{repo_id};
2820                }
2821                $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2822        }
2823        $gs
2824}
2825
2826sub call_authors_prog {
2827        my ($orig_author) = @_;
2828        my $author = `$::_authors_prog $orig_author`;
2829        if ($? != 0) {
2830                die "$::_authors_prog failed with exit code $?\n"
2831        }
2832        if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2833                my ($name, $email) = ($1, $2);
2834                $email = undef if length $2 == 0;
2835                return [$name, $email];
2836        } else {
2837                die "Author: $orig_author: $::_authors_prog returned "
2838                        . "invalid author format: $author\n";
2839        }
2840}
2841
2842sub check_author {
2843        my ($author) = @_;
2844        if (!defined $author || length $author == 0) {
2845                $author = '(no author)';
2846        }
2847        if (!defined $::users{$author}) {
2848                if (defined $::_authors_prog) {
2849                        $::users{$author} = call_authors_prog($author);
2850                } elsif (defined $::_authors) {
2851                        die "Author: $author not defined in $::_authors file\n";
2852                }
2853        }
2854        $author;
2855}
2856
2857sub make_log_entry {
2858        my ($self, $rev, $parents, $ed) = @_;
2859        my $untracked = $self->get_untracked($ed);
2860
2861        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2862        print $un "r$rev\n" or croak $!;
2863        print $un $_, "\n" foreach @$untracked;
2864        my %log_entry = ( parents => $parents || [], revision => $rev,
2865                          log => '');
2866
2867        my $headrev;
2868        my $logged = delete $self->{logged_rev_props};
2869        if (!$logged || $self->{-want_revprops}) {
2870                my $rp = $self->ra->rev_proplist($rev);
2871                foreach (sort keys %$rp) {
2872                        my $v = $rp->{$_};
2873                        if (/^svn:(author|date|log)$/) {
2874                                $log_entry{$1} = $v;
2875                        } elsif ($_ eq 'svm:headrev') {
2876                                $headrev = $v;
2877                        } else {
2878                                print $un "  rev_prop: ", uri_encode($_), ' ',
2879                                          uri_encode($v), "\n";
2880                        }
2881                }
2882        } else {
2883                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2884        }
2885        close $un or croak $!;
2886
2887        $log_entry{date} = parse_svn_date($log_entry{date});
2888        $log_entry{log} .= "\n";
2889        my $author = $log_entry{author} = check_author($log_entry{author});
2890        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2891                                                       : ($author, undef);
2892
2893        my ($commit_name, $commit_email) = ($name, $email);
2894        if ($_use_log_author) {
2895                my $name_field;
2896                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2897                        $name_field = $1;
2898                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2899                        $name_field = $1;
2900                }
2901                if (!defined $name_field) {
2902                        if (!defined $email) {
2903                                $email = $name;
2904                        }
2905                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2906                        ($name, $email) = ($1, $2);
2907                } elsif ($name_field =~ /(.*)@/) {
2908                        ($name, $email) = ($1, $name_field);
2909                } else {
2910                        ($name, $email) = ($name_field, $name_field);
2911                }
2912        }
2913        if (defined $headrev && $self->use_svm_props) {
2914                if ($self->rewrite_root) {
2915                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2916                            "options set!\n";
2917                }
2918                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
2919                # we don't want "SVM: initializing mirror for junk" ...
2920                return undef if $r == 0;
2921                my $svm = $self->svm;
2922                if ($uuid ne $svm->{uuid}) {
2923                        die "UUID mismatch on SVM path:\n",
2924                            "expected: $svm->{uuid}\n",
2925                            "     got: $uuid\n";
2926                }
2927                my $full_url = $self->full_url;
2928                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2929                             die "Failed to replace '$svm->{replace}' with ",
2930                                 "'$svm->{source}' in $full_url\n";
2931                # throw away username for storing in records
2932                remove_username($full_url);
2933                $log_entry{metadata} = "$full_url\@$r $uuid";
2934                $log_entry{svm_revision} = $r;
2935                $email ||= "$author\@$uuid";
2936                $commit_email ||= "$author\@$uuid";
2937        } elsif ($self->use_svnsync_props) {
2938                my $full_url = $self->svnsync->{url};
2939                $full_url .= "/$self->{path}" if length $self->{path};
2940                remove_username($full_url);
2941                my $uuid = $self->svnsync->{uuid};
2942                $log_entry{metadata} = "$full_url\@$rev $uuid";
2943                $email ||= "$author\@$uuid";
2944                $commit_email ||= "$author\@$uuid";
2945        } else {
2946                my $url = $self->metadata_url;
2947                remove_username($url);
2948                $log_entry{metadata} = "$url\@$rev " .
2949                                       $self->ra->get_uuid;
2950                $email ||= "$author\@" . $self->ra->get_uuid;
2951                $commit_email ||= "$author\@" . $self->ra->get_uuid;
2952        }
2953        $log_entry{name} = $name;
2954        $log_entry{email} = $email;
2955        $log_entry{commit_name} = $commit_name;
2956        $log_entry{commit_email} = $commit_email;
2957        \%log_entry;
2958}
2959
2960sub fetch {
2961        my ($self, $min_rev, $max_rev, @parents) = @_;
2962        my ($last_rev, $last_commit) = $self->last_rev_commit;
2963        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2964        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2965}
2966
2967sub set_tree_cb {
2968        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2969        $self->{inject_parents} = { $rev => $tree };
2970        $self->fetch(undef, undef);
2971}
2972
2973sub set_tree {
2974        my ($self, $tree) = (shift, shift);
2975        my $log_entry = ::get_commit_entry($tree);
2976        unless ($self->{last_rev}) {
2977                ::fatal("Must have an existing revision to commit");
2978        }
2979        my %ed_opts = ( r => $self->{last_rev},
2980                        log => $log_entry->{log},
2981                        ra => $self->ra,
2982                        tree_a => $self->{last_commit},
2983                        tree_b => $tree,
2984                        editor_cb => sub {
2985                               $self->set_tree_cb($log_entry, $tree, @_) },
2986                        svn_path => $self->{path} );
2987        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2988                print "No changes\nr$self->{last_rev} = $tree\n";
2989        }
2990}
2991
2992sub rebuild_from_rev_db {
2993        my ($self, $path) = @_;
2994        my $r = -1;
2995        open my $fh, '<', $path or croak "open: $!";
2996        binmode $fh or croak "binmode: $!";
2997        while (<$fh>) {
2998                length($_) == 41 or croak "inconsistent size in ($_) != 41";
2999                chomp($_);
3000                ++$r;
3001                next if $_ eq ('0' x 40);
3002                $self->rev_map_set($r, $_);
3003                print "r$r = $_\n";
3004        }
3005        close $fh or croak "close: $!";
3006        unlink $path or croak "unlink: $!";
3007}
3008
3009sub rebuild {
3010        my ($self) = @_;
3011        my $map_path = $self->map_path;
3012        my $partial = (-e $map_path && ! -z $map_path);
3013        return unless ::verify_ref($self->refname.'^0');
3014        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3015                my $rev_db = $self->rev_db_path;
3016                $self->rebuild_from_rev_db($rev_db);
3017                if ($self->use_svm_props) {
3018                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3019                        $self->rebuild_from_rev_db($svm_rev_db);
3020                }
3021                $self->unlink_rev_db_symlink;
3022                return;
3023        }
3024        print "Rebuilding $map_path ...\n" if (!$partial);
3025        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3026                (undef, undef));
3027        my ($log, $ctx) =
3028            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3029                                ($head ? "$head.." : "") . $self->refname,
3030                                '--');
3031        my $metadata_url = $self->metadata_url;
3032        remove_username($metadata_url);
3033        my $svn_uuid = $self->ra_uuid;
3034        my $c;
3035        while (<$log>) {
3036                if ( m{^commit ($::sha1)$} ) {
3037                        $c = $1;
3038                        next;
3039                }
3040                next unless s{^\s*(git-svn-id:)}{$1};
3041                my ($url, $rev, $uuid) = ::extract_metadata($_);
3042                remove_username($url);
3043
3044                # ignore merges (from set-tree)
3045                next if (!defined $rev || !$uuid);
3046
3047                # if we merged or otherwise started elsewhere, this is
3048                # how we break out of it
3049                if (($uuid ne $svn_uuid) ||
3050                    ($metadata_url && $url && ($url ne $metadata_url))) {
3051                        next;
3052                }
3053                if ($partial && $head) {
3054                        print "Partial-rebuilding $map_path ...\n";
3055                        print "Currently at $base_rev = $head\n";
3056                        $head = undef;
3057                }
3058
3059                $self->rev_map_set($rev, $c);
3060                print "r$rev = $c\n";
3061        }
3062        command_close_pipe($log, $ctx);
3063        print "Done rebuilding $map_path\n" if (!$partial || !$head);
3064        my $rev_db_path = $self->rev_db_path;
3065        if (-f $self->rev_db_path) {
3066                unlink $self->rev_db_path or croak "unlink: $!";
3067        }
3068        $self->unlink_rev_db_symlink;
3069}
3070
3071# rev_map:
3072# Tie::File seems to be prone to offset errors if revisions get sparse,
3073# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3074# one of my favorite modules is out :<  Next up would be one of the DBM
3075# modules, but I'm not sure which is most portable...
3076#
3077# This is the replacement for the rev_db format, which was too big
3078# and inefficient for large repositories with a lot of sparse history
3079# (mainly tags)
3080#
3081# The format is this:
3082#   - 24 bytes for every record,
3083#     * 4 bytes for the integer representing an SVN revision number
3084#     * 20 bytes representing the sha1 of a git commit
3085#   - No empty padding records like the old format
3086#     (except the last record, which can be overwritten)
3087#   - new records are written append-only since SVN revision numbers
3088#     increase monotonically
3089#   - lookups on SVN revision number are done via a binary search
3090#   - Piping the file to xxd -c24 is a good way of dumping it for
3091#     viewing or editing (piped back through xxd -r), should the need
3092#     ever arise.
3093#   - The last record can be padding revision with an all-zero sha1
3094#     This is used to optimize fetch performance when using multiple
3095#     "fetch" directives in .git/config
3096#
3097# These files are disposable unless noMetadata or useSvmProps is set
3098
3099sub _rev_map_set {
3100        my ($fh, $rev, $commit) = @_;
3101
3102        binmode $fh or croak "binmode: $!";
3103        my $size = (stat($fh))[7];
3104        ($size % 24) == 0 or croak "inconsistent size: $size";
3105
3106        my $wr_offset = 0;
3107        if ($size > 0) {
3108                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3109                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3110                $read == 24 or croak "read only $read bytes (!= 24)";
3111                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3112                if ($last_commit eq ('0' x40)) {
3113                        if ($size >= 48) {
3114                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3115                                $read = sysread($fh, $buf, 24) or
3116                                    croak "read: $!";
3117                                $read == 24 or
3118                                    croak "read only $read bytes (!= 24)";
3119                                ($last_rev, $last_commit) =
3120                                    unpack(rev_map_fmt, $buf);
3121                                if ($last_commit eq ('0' x40)) {
3122                                        croak "inconsistent .rev_map\n";
3123                                }
3124                        }
3125                        if ($last_rev >= $rev) {
3126                                croak "last_rev is higher!: $last_rev >= $rev";
3127                        }
3128                        $wr_offset = -24;
3129                }
3130        }
3131        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3132        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3133          croak "write: $!";
3134}
3135
3136sub _rev_map_reset {
3137        my ($fh, $rev, $commit) = @_;
3138        my $c = _rev_map_get($fh, $rev);
3139        $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3140        my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3141        truncate $fh, $offset or croak "truncate: $!";
3142}
3143
3144sub mkfile {
3145        my ($path) = @_;
3146        unless (-e $path) {
3147                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3148                mkpath([$dir]) unless -d $dir;
3149                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3150                close $fh or die "Couldn't close (create) $path: $!\n";
3151        }
3152}
3153
3154sub rev_map_set {
3155        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3156        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3157        my $db = $self->map_path($uuid);
3158        my $db_lock = "$db.lock";
3159        my $sig;
3160        $update_ref ||= 0;
3161        if ($update_ref) {
3162                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3163                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3164        }
3165        mkfile($db);
3166
3167        $LOCKFILES{$db_lock} = 1;
3168        my $sync;
3169        # both of these options make our .rev_db file very, very important
3170        # and we can't afford to lose it because rebuild() won't work
3171        if ($self->use_svm_props || $self->no_metadata) {
3172                $sync = 1;
3173                copy($db, $db_lock) or die "rev_map_set(@_): ",
3174                                           "Failed to copy: ",
3175                                           "$db => $db_lock ($!)\n";
3176        } else {
3177                rename $db, $db_lock or die "rev_map_set(@_): ",
3178                                            "Failed to rename: ",
3179                                            "$db => $db_lock ($!)\n";
3180        }
3181
3182        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3183             or croak "Couldn't open $db_lock: $!\n";
3184        $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3185                                 _rev_map_set($fh, $rev, $commit);
3186        if ($sync) {
3187                $fh->flush or die "Couldn't flush $db_lock: $!\n";
3188                $fh->sync or die "Couldn't sync $db_lock: $!\n";
3189        }
3190        close $fh or croak $!;
3191        if ($update_ref) {
3192                $_head = $self;
3193                my $note = "";
3194                $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3195                command_noisy('update-ref', '-m', "r$rev$note",
3196                              $self->refname, $commit);
3197        }
3198        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3199                                    "$db_lock => $db ($!)\n";
3200        delete $LOCKFILES{$db_lock};
3201        if ($update_ref) {
3202                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3203                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3204                kill $sig, $$ if defined $sig;
3205        }
3206}
3207
3208# If want_commit, this will return an array of (rev, commit) where
3209# commit _must_ be a valid commit in the archive.
3210# Otherwise, it'll return the max revision (whether or not the
3211# commit is valid or just a 0x40 placeholder).
3212sub rev_map_max {
3213        my ($self, $want_commit) = @_;
3214        $self->rebuild;
3215        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3216        $want_commit ? ($r, $c) : $r;
3217}
3218
3219sub rev_map_max_norebuild {
3220        my ($self, $want_commit) = @_;
3221        my $map_path = $self->map_path;
3222        stat $map_path or return $want_commit ? (0, undef) : 0;
3223        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3224        binmode $fh or croak "binmode: $!";
3225        my $size = (stat($fh))[7];
3226        ($size % 24) == 0 or croak "inconsistent size: $size";
3227
3228        if ($size == 0) {
3229                close $fh or croak "close: $!";
3230                return $want_commit ? (0, undef) : 0;
3231        }
3232
3233        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3234        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3235        my ($r, $c) = unpack(rev_map_fmt, $buf);
3236        if ($want_commit && $c eq ('0' x40)) {
3237                if ($size < 48) {
3238                        return $want_commit ? (0, undef) : 0;
3239                }
3240                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3241                sysread($fh, $buf, 24) == 24 or croak "read: $!";
3242                ($r, $c) = unpack(rev_map_fmt, $buf);
3243                if ($c eq ('0'x40)) {
3244                        croak "Penultimate record is all-zeroes in $map_path";
3245                }
3246        }
3247        close $fh or croak "close: $!";
3248        $want_commit ? ($r, $c) : $r;
3249}
3250
3251sub rev_map_get {
3252        my ($self, $rev, $uuid) = @_;
3253        my $map_path = $self->map_path($uuid);
3254        return undef unless -e $map_path;
3255
3256        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3257        my $c = _rev_map_get($fh, $rev);
3258        close($fh) or croak "close: $!";
3259        $c
3260}
3261
3262sub _rev_map_get {
3263        my ($fh, $rev) = @_;
3264
3265        binmode $fh or croak "binmode: $!";
3266        my $size = (stat($fh))[7];
3267        ($size % 24) == 0 or croak "inconsistent size: $size";
3268
3269        if ($size == 0) {
3270                return undef;
3271        }
3272
3273        my ($l, $u) = (0, $size - 24);
3274        my ($r, $c, $buf);
3275
3276        while ($l <= $u) {
3277                my $i = int(($l/24 + $u/24) / 2) * 24;
3278                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3279                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3280                my ($r, $c) = unpack('NH40', $buf);
3281
3282                if ($r < $rev) {
3283                        $l = $i + 24;
3284                } elsif ($r > $rev) {
3285                        $u = $i - 24;
3286                } else { # $r == $rev
3287                        return $c eq ('0' x 40) ? undef : $c;
3288                }
3289        }
3290        undef;
3291}
3292
3293# Finds the first svn revision that exists on (if $eq_ok is true) or
3294# before $rev for the current branch.  It will not search any lower
3295# than $min_rev.  Returns the git commit hash and svn revision number
3296# if found, else (undef, undef).
3297sub find_rev_before {
3298        my ($self, $rev, $eq_ok, $min_rev) = @_;
3299        --$rev unless $eq_ok;
3300        $min_rev ||= 1;
3301        my $max_rev = $self->rev_map_max;
3302        $rev = $max_rev if ($rev > $max_rev);
3303        while ($rev >= $min_rev) {
3304                if (my $c = $self->rev_map_get($rev)) {
3305                        return ($rev, $c);
3306                }
3307                --$rev;
3308        }
3309        return (undef, undef);
3310}
3311
3312# Finds the first svn revision that exists on (if $eq_ok is true) or
3313# after $rev for the current branch.  It will not search any higher
3314# than $max_rev.  Returns the git commit hash and svn revision number
3315# if found, else (undef, undef).
3316sub find_rev_after {
3317        my ($self, $rev, $eq_ok, $max_rev) = @_;
3318        ++$rev unless $eq_ok;
3319        $max_rev ||= $self->rev_map_max;
3320        while ($rev <= $max_rev) {
3321                if (my $c = $self->rev_map_get($rev)) {
3322                        return ($rev, $c);
3323                }
3324                ++$rev;
3325        }
3326        return (undef, undef);
3327}
3328
3329sub _new {
3330        my ($class, $repo_id, $ref_id, $path) = @_;
3331        unless (defined $repo_id && length $repo_id) {
3332                $repo_id = $Git::SVN::default_repo_id;
3333        }
3334        unless (defined $ref_id && length $ref_id) {
3335                $_prefix = '' unless defined($_prefix);
3336                $_[2] = $ref_id =
3337                             "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3338        }
3339        $_[1] = $repo_id;
3340        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3341
3342        # Older repos imported by us used $GIT_DIR/svn/foo instead of
3343        # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3344        if ($ref_id =~ m{^refs/remotes/(.*)}) {
3345                my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3346                if (-d $old_dir && ! -d $dir) {
3347                        $dir = $old_dir;
3348                }
3349        }
3350
3351        $_[3] = $path = '' unless (defined $path);
3352        mkpath([$dir]);
3353        bless {
3354                ref_id => $ref_id, dir => $dir, index => "$dir/index",
3355                path => $path, config => "$ENV{GIT_DIR}/svn/config",
3356                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3357}
3358
3359# for read-only access of old .rev_db formats
3360sub unlink_rev_db_symlink {
3361        my ($self) = @_;
3362        my $link = $self->rev_db_path;
3363        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3364        if (-l $link) {
3365                unlink $link or croak "unlink: $link failed!";
3366        }
3367}
3368
3369sub rev_db_path {
3370        my ($self, $uuid) = @_;
3371        my $db_path = $self->map_path($uuid);
3372        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3373            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3374        $db_path;
3375}
3376
3377# the new replacement for .rev_db
3378sub map_path {
3379        my ($self, $uuid) = @_;
3380        $uuid ||= $self->ra_uuid;
3381        "$self->{map_root}.$uuid";
3382}
3383
3384sub uri_encode {
3385        my ($f) = @_;
3386        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3387        $f
3388}
3389
3390sub remove_username {
3391        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3392}
3393
3394package Git::SVN::Prompt;
3395use strict;
3396use warnings;
3397require SVN::Core;
3398use vars qw/$_no_auth_cache $_username/;
3399
3400sub simple {
3401        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3402        $may_save = undef if $_no_auth_cache;
3403        $default_username = $_username if defined $_username;
3404        if (defined $default_username && length $default_username) {
3405                if (defined $realm && length $realm) {
3406                        print STDERR "Authentication realm: $realm\n";
3407                        STDERR->flush;
3408                }
3409                $cred->username($default_username);
3410        } else {
3411                username($cred, $realm, $may_save, $pool);
3412        }
3413        $cred->password(_read_password("Password for '" .
3414                                       $cred->username . "': ", $realm));
3415        $cred->may_save($may_save);
3416        $SVN::_Core::SVN_NO_ERROR;
3417}
3418
3419sub ssl_server_trust {
3420        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3421        $may_save = undef if $_no_auth_cache;
3422        print STDERR "Error validating server certificate for '$realm':\n";
3423        {
3424                no warnings 'once';
3425                # All variables SVN::Auth::SSL::* are used only once,
3426                # so we're shutting up Perl warnings about this.
3427                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3428                        print STDERR " - The certificate is not issued ",
3429                            "by a trusted authority. Use the\n",
3430                            "   fingerprint to validate ",
3431                            "the certificate manually!\n";
3432                }
3433                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3434                        print STDERR " - The certificate hostname ",
3435                            "does not match.\n";
3436                }
3437                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3438                        print STDERR " - The certificate is not yet valid.\n";
3439                }
3440                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3441                        print STDERR " - The certificate has expired.\n";
3442                }
3443                if ($failures & $SVN::Auth::SSL::OTHER) {
3444                        print STDERR " - The certificate has ",
3445                            "an unknown error.\n";
3446                }
3447        } # no warnings 'once'
3448        printf STDERR
3449                "Certificate information:\n".
3450                " - Hostname: %s\n".
3451                " - Valid: from %s until %s\n".
3452                " - Issuer: %s\n".
3453                " - Fingerprint: %s\n",
3454                map $cert_info->$_, qw(hostname valid_from valid_until
3455                                       issuer_dname fingerprint);
3456        my $choice;
3457prompt:
3458        print STDERR $may_save ?
3459              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3460              "(R)eject or accept (t)emporarily? ";
3461        STDERR->flush;
3462        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3463        if ($choice =~ /^t$/i) {
3464                $cred->may_save(undef);
3465        } elsif ($choice =~ /^r$/i) {
3466                return -1;
3467        } elsif ($may_save && $choice =~ /^p$/i) {
3468                $cred->may_save($may_save);
3469        } else {
3470                goto prompt;
3471        }
3472        $cred->accepted_failures($failures);
3473        $SVN::_Core::SVN_NO_ERROR;
3474}
3475
3476sub ssl_client_cert {
3477        my ($cred, $realm, $may_save, $pool) = @_;
3478        $may_save = undef if $_no_auth_cache;
3479        print STDERR "Client certificate filename: ";
3480        STDERR->flush;
3481        chomp(my $filename = <STDIN>);
3482        $cred->cert_file($filename);
3483        $cred->may_save($may_save);
3484        $SVN::_Core::SVN_NO_ERROR;
3485}
3486
3487sub ssl_client_cert_pw {
3488        my ($cred, $realm, $may_save, $pool) = @_;
3489        $may_save = undef if $_no_auth_cache;
3490        $cred->password(_read_password("Password: ", $realm));
3491        $cred->may_save($may_save);
3492        $SVN::_Core::SVN_NO_ERROR;
3493}
3494
3495sub username {
3496        my ($cred, $realm, $may_save, $pool) = @_;
3497        $may_save = undef if $_no_auth_cache;
3498        if (defined $realm && length $realm) {
3499                print STDERR "Authentication realm: $realm\n";
3500        }
3501        my $username;
3502        if (defined $_username) {
3503                $username = $_username;
3504        } else {
3505                print STDERR "Username: ";
3506                STDERR->flush;
3507                chomp($username = <STDIN>);
3508        }
3509        $cred->username($username);
3510        $cred->may_save($may_save);
3511        $SVN::_Core::SVN_NO_ERROR;
3512}
3513
3514sub _read_password {
3515        my ($prompt, $realm) = @_;
3516        print STDERR $prompt;
3517        STDERR->flush;
3518        require Term::ReadKey;
3519        Term::ReadKey::ReadMode('noecho');
3520        my $password = '';
3521        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3522                last if $key =~ /[\012\015]/; # \n\r
3523                $password .= $key;
3524        }
3525        Term::ReadKey::ReadMode('restore');
3526        print STDERR "\n";
3527        STDERR->flush;
3528        $password;
3529}
3530
3531package SVN::Git::Fetcher;
3532use vars qw/@ISA/;
3533use strict;
3534use warnings;
3535use Carp qw/croak/;
3536use File::Temp qw/tempfile/;
3537use IO::File qw//;
3538use vars qw/$_ignore_regex/;
3539
3540# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3541sub new {
3542        my ($class, $git_svn, $switch_path) = @_;
3543        my $self = SVN::Delta::Editor->new;
3544        bless $self, $class;
3545        if (exists $git_svn->{last_commit}) {
3546                $self->{c} = $git_svn->{last_commit};
3547                $self->{empty_symlinks} =
3548                                  _mark_empty_symlinks($git_svn, $switch_path);
3549        }
3550        $self->{ignore_regex} = eval { command_oneline('config', '--get',
3551                             "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3552        $self->{empty} = {};
3553        $self->{dir_prop} = {};
3554        $self->{file_prop} = {};
3555        $self->{absent_dir} = {};
3556        $self->{absent_file} = {};
3557        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3558        $self;
3559}
3560
3561# this uses the Ra object, so it must be called before do_{switch,update},
3562# not inside them (when the Git::SVN::Fetcher object is passed) to
3563# do_{switch,update}
3564sub _mark_empty_symlinks {
3565        my ($git_svn, $switch_path) = @_;
3566        my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3567        return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3568
3569        my %ret;
3570        my ($rev, $cmt) = $git_svn->last_rev_commit;
3571        return {} unless ($rev && $cmt);
3572
3573        # allow the warning to be printed for each revision we fetch to
3574        # ensure the user sees it.  The user can also disable the workaround
3575        # on the repository even while git svn is running and the next
3576        # revision fetched will skip this expensive function.
3577        my $printed_warning;
3578        chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3579        my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3580        local $/ = "\0";
3581        my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3582        $pfx .= '/' if length($pfx);
3583        while (<$ls>) {
3584                chomp;
3585                s/\A100644 blob $empty_blob\t//o or next;
3586                unless ($printed_warning) {
3587                        print STDERR "Scanning for empty symlinks, ",
3588                                     "this may take a while if you have ",
3589                                     "many empty files\n",
3590                                     "You may disable this with `",
3591                                     "git config svn.brokenSymlinkWorkaround ",
3592                                     "false'.\n",
3593                                     "This may be done in a different ",
3594                                     "terminal without restarting ",
3595                                     "git svn\n";
3596                        $printed_warning = 1;
3597                }
3598                my $path = $_;
3599                my (undef, $props) =
3600                               $git_svn->ra->get_file($pfx.$path, $rev, undef);
3601                if ($props->{'svn:special'}) {
3602                        $ret{$path} = 1;
3603                }
3604        }
3605        command_close_pipe($ls, $ctx);
3606        \%ret;
3607}
3608
3609# returns true if a given path is inside a ".git" directory
3610sub in_dot_git {
3611        $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3612}
3613
3614# return value: 0 -- don't ignore, 1 -- ignore
3615sub is_path_ignored {
3616        my ($self, $path) = @_;
3617        return 1 if in_dot_git($path);
3618        return 1 if defined($self->{ignore_regex}) &&
3619                    $path =~ m!$self->{ignore_regex}!;
3620        return 0 unless defined($_ignore_regex);
3621        return 1 if $path =~ m!$_ignore_regex!o;
3622        return 0;
3623}
3624
3625sub set_path_strip {
3626        my ($self, $path) = @_;
3627        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3628}
3629
3630sub open_root {
3631        { path => '' };
3632}
3633
3634sub open_directory {
3635        my ($self, $path, $pb, $rev) = @_;
3636        { path => $path };
3637}
3638
3639sub git_path {
3640        my ($self, $path) = @_;
3641        if ($self->{path_strip}) {
3642                $path =~ s!$self->{path_strip}!! or
3643                  die "Failed to strip path '$path' ($self->{path_strip})\n";
3644        }
3645        $path;
3646}
3647
3648sub delete_entry {
3649        my ($self, $path, $rev, $pb) = @_;
3650        return undef if $self->is_path_ignored($path);
3651
3652        my $gpath = $self->git_path($path);
3653        return undef if ($gpath eq '');
3654
3655        # remove entire directories.
3656        my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3657                         =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3658        if ($tree) {
3659                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3660                                                     -r --name-only -z/,
3661                                                     $tree);
3662                local $/ = "\0";
3663                while (<$ls>) {
3664                        chomp;
3665                        my $rmpath = "$gpath/$_";
3666                        $self->{gii}->remove($rmpath);
3667                        print "\tD\t$rmpath\n" unless $::_q;
3668                }
3669                print "\tD\t$gpath/\n" unless $::_q;
3670                command_close_pipe($ls, $ctx);
3671                $self->{empty}->{$path} = 0
3672        } else {
3673                $self->{gii}->remove($gpath);
3674                print "\tD\t$gpath\n" unless $::_q;
3675        }
3676        undef;
3677}
3678
3679sub open_file {
3680        my ($self, $path, $pb, $rev) = @_;
3681        my ($mode, $blob);
3682
3683        goto out if $self->is_path_ignored($path);
3684
3685        my $gpath = $self->git_path($path);
3686        ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3687                             =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3688        unless (defined $mode && defined $blob) {
3689                die "$path was not found in commit $self->{c} (r$rev)\n";
3690        }
3691        if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3692                $mode = '120000';
3693        }
3694out:
3695        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3696          pool => SVN::Pool->new, action => 'M' };
3697}
3698
3699sub add_file {
3700        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3701        my $mode;
3702
3703        if (!$self->is_path_ignored($path)) {
3704                my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3705                delete $self->{empty}->{$dir};
3706                $mode = '100644';
3707        }
3708        { path => $path, mode_a => $mode, mode_b => $mode,
3709          pool => SVN::Pool->new, action => 'A' };
3710}
3711
3712sub add_directory {
3713        my ($self, $path, $cp_path, $cp_rev) = @_;
3714        goto out if $self->is_path_ignored($path);
3715        my $gpath = $self->git_path($path);
3716        if ($gpath eq '') {
3717                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3718                                                     -r --name-only -z/,
3719                                                     $self->{c});
3720                local $/ = "\0";
3721                while (<$ls>) {
3722                        chomp;
3723                        $self->{gii}->remove($_);
3724                        print "\tD\t$_\n" unless $::_q;
3725                }
3726                command_close_pipe($ls, $ctx);
3727                $self->{empty}->{$path} = 0;
3728        }
3729        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3730        delete $self->{empty}->{$dir};
3731        $self->{empty}->{$path} = 1;
3732out:
3733        { path => $path };
3734}
3735
3736sub change_dir_prop {
3737        my ($self, $db, $prop, $value) = @_;
3738        return undef if $self->is_path_ignored($db->{path});
3739        $self->{dir_prop}->{$db->{path}} ||= {};
3740        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3741        undef;
3742}
3743
3744sub absent_directory {
3745        my ($self, $path, $pb) = @_;
3746        return undef if $self->is_path_ignored($path);
3747        $self->{absent_dir}->{$pb->{path}} ||= [];
3748        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3749        undef;
3750}
3751
3752sub absent_file {
3753        my ($self, $path, $pb) = @_;
3754        return undef if $self->is_path_ignored($path);
3755        $self->{absent_file}->{$pb->{path}} ||= [];
3756        push @{$self->{absent_file}->{$pb->{path}}}, $path;
3757        undef;
3758}
3759
3760sub change_file_prop {
3761        my ($self, $fb, $prop, $value) = @_;
3762        return undef if $self->is_path_ignored($fb->{path});
3763        if ($prop eq 'svn:executable') {
3764                if ($fb->{mode_b} != 120000) {
3765                        $fb->{mode_b} = defined $value ? 100755 : 100644;
3766                }
3767        } elsif ($prop eq 'svn:special') {
3768                $fb->{mode_b} = defined $value ? 120000 : 100644;
3769        } else {
3770                $self->{file_prop}->{$fb->{path}} ||= {};
3771                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3772        }
3773        undef;
3774}
3775
3776sub apply_textdelta {
3777        my ($self, $fb, $exp) = @_;
3778        return undef if $self->is_path_ignored($fb->{path});
3779        my $fh = $::_repository->temp_acquire('svn_delta');
3780        # $fh gets auto-closed() by SVN::TxDelta::apply(),
3781        # (but $base does not,) so dup() it for reading in close_file
3782        open my $dup, '<&', $fh or croak $!;
3783        my $base = $::_repository->temp_acquire('git_blob');
3784
3785        if ($fb->{blob}) {
3786                my ($base_is_link, $size);
3787
3788                if ($fb->{mode_a} eq '120000' &&
3789                    ! $self->{empty_symlinks}->{$fb->{path}}) {
3790                        print $base 'link ' or die "print $!\n";
3791                        $base_is_link = 1;
3792                }
3793        retry:
3794                $size = $::_repository->cat_blob($fb->{blob}, $base);
3795                die "Failed to read object $fb->{blob}" if ($size < 0);
3796
3797                if (defined $exp) {
3798                        seek $base, 0, 0 or croak $!;
3799                        my $got = ::md5sum($base);
3800                        if ($got ne $exp) {
3801                                my $err = "Checksum mismatch: ".
3802                                       "$fb->{path} $fb->{blob}\n" .
3803                                       "expected: $exp\n" .
3804                                       "     got: $got\n";
3805                                if ($base_is_link) {
3806                                        warn $err,
3807                                             "Retrying... (possibly ",
3808                                             "a bad symlink from SVN)\n";
3809                                        $::_repository->temp_reset($base);
3810                                        $base_is_link = 0;
3811                                        goto retry;
3812                                }
3813                                die $err;
3814                        }
3815                }
3816        }
3817        seek $base, 0, 0 or croak $!;
3818        $fb->{fh} = $fh;
3819        $fb->{base} = $base;
3820        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3821}
3822
3823sub close_file {
3824        my ($self, $fb, $exp) = @_;
3825        return undef if $self->is_path_ignored($fb->{path});
3826
3827        my $hash;
3828        my $path = $self->git_path($fb->{path});
3829        if (my $fh = $fb->{fh}) {
3830                if (defined $exp) {
3831                        seek($fh, 0, 0) or croak $!;
3832                        my $got = ::md5sum($fh);
3833                        if ($got ne $exp) {
3834                                die "Checksum mismatch: $path\n",
3835                                    "expected: $exp\n    got: $got\n";
3836                        }
3837                }
3838                if ($fb->{mode_b} == 120000) {
3839                        sysseek($fh, 0, 0) or croak $!;
3840                        my $rd = sysread($fh, my $buf, 5);
3841
3842                        if (!defined $rd) {
3843                                croak "sysread: $!\n";
3844                        } elsif ($rd == 0) {
3845                                warn "$path has mode 120000",
3846                                     " but it points to nothing\n",
3847                                     "converting to an empty file with mode",
3848                                     " 100644\n";
3849                                $fb->{mode_b} = '100644';
3850                        } elsif ($buf ne 'link ') {
3851                                warn "$path has mode 120000",
3852                                     " but is not a link\n";
3853                        } else {
3854                                my $tmp_fh = $::_repository->temp_acquire(
3855                                        'svn_hash');
3856                                my $res;
3857                                while ($res = sysread($fh, my $str, 1024)) {
3858                                        my $out = syswrite($tmp_fh, $str, $res);
3859                                        defined($out) && $out == $res
3860                                                or croak("write ",
3861                                                        Git::temp_path($tmp_fh),
3862                                                        ": $!\n");
3863                                }
3864                                defined $res or croak $!;
3865
3866                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
3867                                Git::temp_release($tmp_fh, 1);
3868                        }
3869                }
3870
3871                $hash = $::_repository->hash_and_insert_object(
3872                                Git::temp_path($fh));
3873                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3874
3875                Git::temp_release($fb->{base}, 1);
3876                Git::temp_release($fh, 1);
3877        } else {
3878                $hash = $fb->{blob} or die "no blob information\n";
3879        }
3880        $fb->{pool}->clear;
3881        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3882        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3883        undef;
3884}
3885
3886sub abort_edit {
3887        my $self = shift;
3888        $self->{nr} = $self->{gii}->{nr};
3889        delete $self->{gii};
3890        $self->SUPER::abort_edit(@_);
3891}
3892
3893sub close_edit {
3894        my $self = shift;
3895        $self->{git_commit_ok} = 1;
3896        $self->{nr} = $self->{gii}->{nr};
3897        delete $self->{gii};
3898        $self->SUPER::close_edit(@_);
3899}
3900
3901package SVN::Git::Editor;
3902use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3903use strict;
3904use warnings;
3905use Carp qw/croak/;
3906use IO::File;
3907
3908sub new {
3909        my ($class, $opts) = @_;
3910        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3911                die "$_ required!\n" unless (defined $opts->{$_});
3912        }
3913
3914        my $pool = SVN::Pool->new;
3915        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3916        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3917                                     $opts->{r}, $mods);
3918
3919        # $opts->{ra} functions should not be used after this:
3920        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
3921                                                $opts->{editor_cb}, $pool);
3922        my $self = SVN::Delta::Editor->new(@ce, $pool);
3923        bless $self, $class;
3924        foreach (qw/svn_path r tree_a tree_b/) {
3925                $self->{$_} = $opts->{$_};
3926        }
3927        $self->{url} = $opts->{ra}->{url};
3928        $self->{mods} = $mods;
3929        $self->{types} = $types;
3930        $self->{pool} = $pool;
3931        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3932        $self->{rm} = { };
3933        $self->{path_prefix} = length $self->{svn_path} ?
3934                               "$self->{svn_path}/" : '';
3935        $self->{config} = $opts->{config};
3936        return $self;
3937}
3938
3939sub generate_diff {
3940        my ($tree_a, $tree_b) = @_;
3941        my @diff_tree = qw(diff-tree -z -r);
3942        if ($_cp_similarity) {
3943                push @diff_tree, "-C$_cp_similarity";
3944        } else {
3945                push @diff_tree, '-C';
3946        }
3947        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3948        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3949        push @diff_tree, $tree_a, $tree_b;
3950        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3951        local $/ = "\0";
3952        my $state = 'meta';
3953        my @mods;
3954        while (<$diff_fh>) {
3955                chomp $_; # this gets rid of the trailing "\0"
3956                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3957                                        ($::sha1)\s($::sha1)\s
3958                                        ([MTCRAD])\d*$/xo) {
3959                        push @mods, {   mode_a => $1, mode_b => $2,
3960                                        sha1_a => $3, sha1_b => $4,
3961                                        chg => $5 };
3962                        if ($5 =~ /^(?:C|R)$/) {
3963                                $state = 'file_a';
3964                        } else {
3965                                $state = 'file_b';
3966                        }
3967                } elsif ($state eq 'file_a') {
3968                        my $x = $mods[$#mods] or croak "Empty array\n";
3969                        if ($x->{chg} !~ /^(?:C|R)$/) {
3970                                croak "Error parsing $_, $x->{chg}\n";
3971                        }
3972                        $x->{file_a} = $_;
3973                        $state = 'file_b';
3974                } elsif ($state eq 'file_b') {
3975                        my $x = $mods[$#mods] or croak "Empty array\n";
3976                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3977                                croak "Error parsing $_, $x->{chg}\n";
3978                        }
3979                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3980                                croak "Error parsing $_, $x->{chg}\n";
3981                        }
3982                        $x->{file_b} = $_;
3983                        $state = 'meta';
3984                } else {
3985                        croak "Error parsing $_\n";
3986                }
3987        }
3988        command_close_pipe($diff_fh, $ctx);
3989        \@mods;
3990}
3991
3992sub check_diff_paths {
3993        my ($ra, $pfx, $rev, $mods) = @_;
3994        my %types;
3995        $pfx .= '/' if length $pfx;
3996
3997        sub type_diff_paths {
3998                my ($ra, $types, $path, $rev) = @_;
3999                my @p = split m#/+#, $path;
4000                my $c = shift @p;
4001                unless (defined $types->{$c}) {
4002                        $types->{$c} = $ra->check_path($c, $rev);
4003                }
4004                while (@p) {
4005                        $c .= '/' . shift @p;
4006                        next if defined $types->{$c};
4007                        $types->{$c} = $ra->check_path($c, $rev);
4008                }
4009        }
4010
4011        foreach my $m (@$mods) {
4012                foreach my $f (qw/file_a file_b/) {
4013                        next unless defined $m->{$f};
4014                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4015                        if (length $pfx.$dir && ! defined $types{$dir}) {
4016                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4017                        }
4018                }
4019        }
4020        \%types;
4021}
4022
4023sub split_path {
4024        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4025}
4026
4027sub repo_path {
4028        my ($self, $path) = @_;
4029        $self->{path_prefix}.(defined $path ? $path : '');
4030}
4031
4032sub url_path {
4033        my ($self, $path) = @_;
4034        if ($self->{url} =~ m#^https?://#) {
4035                $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4036        }
4037        $self->{url} . '/' . $self->repo_path($path);
4038}
4039
4040sub rmdirs {
4041        my ($self) = @_;
4042        my $rm = $self->{rm};
4043        delete $rm->{''}; # we never delete the url we're tracking
4044        return unless %$rm;
4045
4046        foreach (keys %$rm) {
4047                my @d = split m#/#, $_;
4048                my $c = shift @d;
4049                $rm->{$c} = 1;
4050                while (@d) {
4051                        $c .= '/' . shift @d;
4052                        $rm->{$c} = 1;
4053                }
4054        }
4055        delete $rm->{$self->{svn_path}};
4056        delete $rm->{''}; # we never delete the url we're tracking
4057        return unless %$rm;
4058
4059        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4060                                             $self->{tree_b});
4061        local $/ = "\0";
4062        while (<$fh>) {
4063                chomp;
4064                my @dn = split m#/#, $_;
4065                while (pop @dn) {
4066                        delete $rm->{join '/', @dn};
4067                }
4068                unless (%$rm) {
4069                        close $fh;
4070                        return;
4071                }
4072        }
4073        command_close_pipe($fh, $ctx);
4074
4075        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4076        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4077                $self->close_directory($bat->{$d}, $p);
4078                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4079                print "\tD+\t$d/\n" unless $::_q;
4080                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4081                delete $bat->{$d};
4082        }
4083}
4084
4085sub open_or_add_dir {
4086        my ($self, $full_path, $baton) = @_;
4087        my $t = $self->{types}->{$full_path};
4088        if (!defined $t) {
4089                die "$full_path not known in r$self->{r} or we have a bug!\n";
4090        }
4091        {
4092                no warnings 'once';
4093                # SVN::Node::none and SVN::Node::file are used only once,
4094                # so we're shutting up Perl's warnings about them.
4095                if ($t == $SVN::Node::none) {
4096                        return $self->add_directory($full_path, $baton,
4097                            undef, -1, $self->{pool});
4098                } elsif ($t == $SVN::Node::dir) {
4099                        return $self->open_directory($full_path, $baton,
4100                            $self->{r}, $self->{pool});
4101                } # no warnings 'once'
4102                print STDERR "$full_path already exists in repository at ",
4103                    "r$self->{r} and it is not a directory (",
4104                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4105        } # no warnings 'once'
4106        exit 1;
4107}
4108
4109sub ensure_path {
4110        my ($self, $path) = @_;
4111        my $bat = $self->{bat};
4112        my $repo_path = $self->repo_path($path);
4113        return $bat->{''} unless (length $repo_path);
4114        my @p = split m#/+#, $repo_path;
4115        my $c = shift @p;
4116        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4117        while (@p) {
4118                my $c0 = $c;
4119                $c .= '/' . shift @p;
4120                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4121        }
4122        return $bat->{$c};
4123}
4124
4125# Subroutine to convert a globbing pattern to a regular expression.
4126# From perl cookbook.
4127sub glob2pat {
4128        my $globstr = shift;
4129        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4130        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4131        return '^' . $globstr . '$';
4132}
4133
4134sub check_autoprop {
4135        my ($self, $pattern, $properties, $file, $fbat) = @_;
4136        # Convert the globbing pattern to a regular expression.
4137        my $regex = glob2pat($pattern);
4138        # Check if the pattern matches the file name.
4139        if($file =~ m/($regex)/) {
4140                # Parse the list of properties to set.
4141                my @props = split(/;/, $properties);
4142                foreach my $prop (@props) {
4143                        # Parse 'name=value' syntax and set the property.
4144                        if ($prop =~ /([^=]+)=(.*)/) {
4145                                my ($n,$v) = ($1,$2);
4146                                for ($n, $v) {
4147                                        s/^\s+//; s/\s+$//;
4148                                }
4149                                $self->change_file_prop($fbat, $n, $v);
4150                        }
4151                }
4152        }
4153}
4154
4155sub apply_autoprops {
4156        my ($self, $file, $fbat) = @_;
4157        my $conf_t = ${$self->{config}}{'config'};
4158        no warnings 'once';
4159        # Check [miscellany]/enable-auto-props in svn configuration.
4160        if (SVN::_Core::svn_config_get_bool(
4161                $conf_t,
4162                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4163                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4164                0)) {
4165                # Auto-props are enabled.  Enumerate them to look for matches.
4166                my $callback = sub {
4167                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
4168                };
4169                SVN::_Core::svn_config_enumerate(
4170                        $conf_t,
4171                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4172                        $callback);
4173        }
4174}
4175
4176sub A {
4177        my ($self, $m) = @_;
4178        my ($dir, $file) = split_path($m->{file_b});
4179        my $pbat = $self->ensure_path($dir);
4180        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4181                                        undef, -1);
4182        print "\tA\t$m->{file_b}\n" unless $::_q;
4183        $self->apply_autoprops($file, $fbat);
4184        $self->chg_file($fbat, $m);
4185        $self->close_file($fbat,undef,$self->{pool});
4186}
4187
4188sub C {
4189        my ($self, $m) = @_;
4190        my ($dir, $file) = split_path($m->{file_b});
4191        my $pbat = $self->ensure_path($dir);
4192        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4193                                $self->url_path($m->{file_a}), $self->{r});
4194        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4195        $self->chg_file($fbat, $m);
4196        $self->close_file($fbat,undef,$self->{pool});
4197}
4198
4199sub delete_entry {
4200        my ($self, $path, $pbat) = @_;
4201        my $rpath = $self->repo_path($path);
4202        my ($dir, $file) = split_path($rpath);
4203        $self->{rm}->{$dir} = 1;
4204        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4205}
4206
4207sub R {
4208        my ($self, $m) = @_;
4209        my ($dir, $file) = split_path($m->{file_b});
4210        my $pbat = $self->ensure_path($dir);
4211        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4212                                $self->url_path($m->{file_a}), $self->{r});
4213        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4214        $self->apply_autoprops($file, $fbat);
4215        $self->chg_file($fbat, $m);
4216        $self->close_file($fbat,undef,$self->{pool});
4217
4218        ($dir, $file) = split_path($m->{file_a});
4219        $pbat = $self->ensure_path($dir);
4220        $self->delete_entry($m->{file_a}, $pbat);
4221}
4222
4223sub M {
4224        my ($self, $m) = @_;
4225        my ($dir, $file) = split_path($m->{file_b});
4226        my $pbat = $self->ensure_path($dir);
4227        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4228                                $pbat,$self->{r},$self->{pool});
4229        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4230        $self->chg_file($fbat, $m);
4231        $self->close_file($fbat,undef,$self->{pool});
4232}
4233
4234sub T { shift->M(@_) }
4235
4236sub change_file_prop {
4237        my ($self, $fbat, $pname, $pval) = @_;
4238        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4239}
4240
4241sub _chg_file_get_blob ($$$$) {
4242        my ($self, $fbat, $m, $which) = @_;
4243        my $fh = $::_repository->temp_acquire("git_blob_$which");
4244        if ($m->{"mode_$which"} =~ /^120/) {
4245                print $fh 'link ' or croak $!;
4246                $self->change_file_prop($fbat,'svn:special','*');
4247        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4248                $self->change_file_prop($fbat,'svn:special',undef);
4249        }
4250        my $blob = $m->{"sha1_$which"};
4251        return ($fh,) if ($blob =~ /^0{40}$/);
4252        my $size = $::_repository->cat_blob($blob, $fh);
4253        croak "Failed to read object $blob" if ($size < 0);
4254        $fh->flush == 0 or croak $!;
4255        seek $fh, 0, 0 or croak $!;
4256
4257        my $exp = ::md5sum($fh);
4258        seek $fh, 0, 0 or croak $!;
4259        return ($fh, $exp);
4260}
4261
4262sub chg_file {
4263        my ($self, $fbat, $m) = @_;
4264        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4265                $self->change_file_prop($fbat,'svn:executable','*');
4266        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4267                $self->change_file_prop($fbat,'svn:executable',undef);
4268        }
4269        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4270        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4271        my $pool = SVN::Pool->new;
4272        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4273        if (-s $fh_a) {
4274                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4275                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4276                if (defined $res) {
4277                        die "Unexpected result from send_txstream: $res\n",
4278                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4279                }
4280        } else {
4281                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4282                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4283                    if ($got ne $exp_b);
4284        }
4285        Git::temp_release($fh_b, 1);
4286        Git::temp_release($fh_a, 1);
4287        $pool->clear;
4288}
4289
4290sub D {
4291        my ($self, $m) = @_;
4292        my ($dir, $file) = split_path($m->{file_b});
4293        my $pbat = $self->ensure_path($dir);
4294        print "\tD\t$m->{file_b}\n" unless $::_q;
4295        $self->delete_entry($m->{file_b}, $pbat);
4296}
4297
4298sub close_edit {
4299        my ($self) = @_;
4300        my ($p,$bat) = ($self->{pool}, $self->{bat});
4301        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4302                next if $_ eq '';
4303                $self->close_directory($bat->{$_}, $p);
4304        }
4305        $self->close_directory($bat->{''}, $p);
4306        $self->SUPER::close_edit($p);
4307        $p->clear;
4308}
4309
4310sub abort_edit {
4311        my ($self) = @_;
4312        $self->SUPER::abort_edit($self->{pool});
4313}
4314
4315sub DESTROY {
4316        my $self = shift;
4317        $self->SUPER::DESTROY(@_);
4318        $self->{pool}->clear;
4319}
4320
4321# this drives the editor
4322sub apply_diff {
4323        my ($self) = @_;
4324        my $mods = $self->{mods};
4325        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4326        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4327                my $f = $m->{chg};
4328                if (defined $o{$f}) {
4329                        $self->$f($m);
4330                } else {
4331                        fatal("Invalid change type: $f");
4332                }
4333        }
4334        $self->rmdirs if $_rmdir;
4335        if (@$mods == 0) {
4336                $self->abort_edit;
4337        } else {
4338                $self->close_edit;
4339        }
4340        return scalar @$mods;
4341}
4342
4343package Git::SVN::Ra;
4344use vars qw/@ISA $config_dir $_log_window_size/;
4345use strict;
4346use warnings;
4347my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4348
4349BEGIN {
4350        # enforce temporary pool usage for some simple functions
4351        no strict 'refs';
4352        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4353                      get_file/) {
4354                my $SUPER = "SUPER::$f";
4355                *$f = sub {
4356                        my $self = shift;
4357                        my $pool = SVN::Pool->new;
4358                        my @ret = $self->$SUPER(@_,$pool);
4359                        $pool->clear;
4360                        wantarray ? @ret : $ret[0];
4361                };
4362        }
4363}
4364
4365sub _auth_providers () {
4366        [
4367          SVN::Client::get_simple_provider(),
4368          SVN::Client::get_ssl_server_trust_file_provider(),
4369          SVN::Client::get_simple_prompt_provider(
4370            \&Git::SVN::Prompt::simple, 2),
4371          SVN::Client::get_ssl_client_cert_file_provider(),
4372          SVN::Client::get_ssl_client_cert_prompt_provider(
4373            \&Git::SVN::Prompt::ssl_client_cert, 2),
4374          SVN::Client::get_ssl_client_cert_pw_file_provider(),
4375          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4376            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4377          SVN::Client::get_username_provider(),
4378          SVN::Client::get_ssl_server_trust_prompt_provider(
4379            \&Git::SVN::Prompt::ssl_server_trust),
4380          SVN::Client::get_username_prompt_provider(
4381            \&Git::SVN::Prompt::username, 2)
4382        ]
4383}
4384
4385sub escape_uri_only {
4386        my ($uri) = @_;
4387        my @tmp;
4388        foreach (split m{/}, $uri) {
4389                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4390                push @tmp, $_;
4391        }
4392        join('/', @tmp);
4393}
4394
4395sub escape_url {
4396        my ($url) = @_;
4397        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4398                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4399                $url = "$scheme://$domain$uri";
4400        }
4401        $url;
4402}
4403
4404sub new {
4405        my ($class, $url) = @_;
4406        $url =~ s!/+$!!;
4407        return $RA if ($RA && $RA->{url} eq $url);
4408
4409        SVN::_Core::svn_config_ensure($config_dir, undef);
4410        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4411        my $config = SVN::Core::config_get_config($config_dir);
4412        $RA = undef;
4413        my $dont_store_passwords = 1;
4414        my $conf_t = ${$config}{'config'};
4415        {
4416                no warnings 'once';
4417                # The usage of $SVN::_Core::SVN_CONFIG_* variables
4418                # produces warnings that variables are used only once.
4419                # I had not found the better way to shut them up, so
4420                # the warnings of type 'once' are disabled in this block.
4421                if (SVN::_Core::svn_config_get_bool($conf_t,
4422                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4423                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4424                    1) == 0) {
4425                        SVN::_Core::svn_auth_set_parameter($baton,
4426                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4427                            bless (\$dont_store_passwords, "_p_void"));
4428                }
4429                if (SVN::_Core::svn_config_get_bool($conf_t,
4430                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4431                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4432                    1) == 0) {
4433                        $Git::SVN::Prompt::_no_auth_cache = 1;
4434                }
4435        } # no warnings 'once'
4436        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4437                              config => $config,
4438                              pool => SVN::Pool->new,
4439                              auth_provider_callbacks => $callbacks);
4440        $self->{url} = $url;
4441        $self->{svn_path} = $url;
4442        $self->{repos_root} = $self->get_repos_root;
4443        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4444        $self->{cache} = { check_path => { r => 0, data => {} },
4445                           get_dir => { r => 0, data => {} } };
4446        $RA = bless $self, $class;
4447}
4448
4449sub check_path {
4450        my ($self, $path, $r) = @_;
4451        my $cache = $self->{cache}->{check_path};
4452        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4453                return $cache->{data}->{$path};
4454        }
4455        my $pool = SVN::Pool->new;
4456        my $t = $self->SUPER::check_path($path, $r, $pool);
4457        $pool->clear;
4458        if ($r != $cache->{r}) {
4459                %{$cache->{data}} = ();
4460                $cache->{r} = $r;
4461        }
4462        $cache->{data}->{$path} = $t;
4463}
4464
4465sub get_dir {
4466        my ($self, $dir, $r) = @_;
4467        my $cache = $self->{cache}->{get_dir};
4468        if ($r == $cache->{r}) {
4469                if (my $x = $cache->{data}->{$dir}) {
4470                        return wantarray ? @$x : $x->[0];
4471                }
4472        }
4473        my $pool = SVN::Pool->new;
4474        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4475        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4476        $pool->clear;
4477        if ($r != $cache->{r}) {
4478                %{$cache->{data}} = ();
4479                $cache->{r} = $r;
4480        }
4481        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4482        wantarray ? (\%dirents, $r, $props) : \%dirents;
4483}
4484
4485sub DESTROY {
4486        # do not call the real DESTROY since we store ourselves in $RA
4487}
4488
4489# get_log(paths, start, end, limit,
4490#         discover_changed_paths, strict_node_history, receiver)
4491sub get_log {
4492        my ($self, @args) = @_;
4493        my $pool = SVN::Pool->new;
4494
4495        # svn_log_changed_path_t objects passed to get_log are likely to be
4496        # overwritten even if only the refs are copied to an external variable,
4497        # so we should dup the structures in their entirety.  Using an
4498        # externally passed pool (instead of our temporary and quickly cleared
4499        # pool in Git::SVN::Ra) does not help matters at all...
4500        my $receiver = pop @args;
4501        my $prefix = "/".$self->{svn_path};
4502        $prefix =~ s#/+($)##;
4503        my $prefix_regex = qr#^\Q$prefix\E#;
4504        push(@args, sub {
4505                my ($paths) = $_[0];
4506                return &$receiver(@_) unless $paths;
4507                $_[0] = ();
4508                foreach my $p (keys %$paths) {
4509                        my $i = $paths->{$p};
4510                        # Make path relative to our url, not repos_root
4511                        $p =~ s/$prefix_regex//;
4512                        my %s = map { $_ => $i->$_; }
4513                                qw/copyfrom_path copyfrom_rev action/;
4514                        if ($s{'copyfrom_path'}) {
4515                                $s{'copyfrom_path'} =~ s/$prefix_regex//;
4516                        }
4517                        $_[0]{$p} = \%s;
4518                }
4519                &$receiver(@_);
4520        });
4521
4522
4523        # the limit parameter was not supported in SVN 1.1.x, so we
4524        # drop it.  Therefore, the receiver callback passed to it
4525        # is made aware of this limitation by being wrapped if
4526        # the limit passed to is being wrapped.
4527        if ($SVN::Core::VERSION le '1.2.0') {
4528                my $limit = splice(@args, 3, 1);
4529                if ($limit > 0) {
4530                        my $receiver = pop @args;
4531                        push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4532                }
4533        }
4534        my $ret = $self->SUPER::get_log(@args, $pool);
4535        $pool->clear;
4536        $ret;
4537}
4538
4539sub trees_match {
4540        my ($self, $url1, $rev1, $url2, $rev2) = @_;
4541        my $ctx = SVN::Client->new(auth => _auth_providers);
4542        my $out = IO::File->new_tmpfile;
4543
4544        # older SVN (1.1.x) doesn't take $pool as the last parameter for
4545        # $ctx->diff(), so we'll create a default one
4546        my $pool = SVN::Pool->new_default_sub;
4547
4548        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4549        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4550        $out->flush;
4551        my $ret = (($out->stat)[7] == 0);
4552        close $out or croak $!;
4553
4554        $ret;
4555}
4556
4557sub get_commit_editor {
4558        my ($self, $log, $cb, $pool) = @_;
4559        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4560        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4561}
4562
4563sub gs_do_update {
4564        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4565        my $new = ($rev_a == $rev_b);
4566        my $path = $gs->{path};
4567
4568        if ($new && -e $gs->{index}) {
4569                unlink $gs->{index} or die
4570                  "Couldn't unlink index: $gs->{index}: $!\n";
4571        }
4572        my $pool = SVN::Pool->new;
4573        $editor->set_path_strip($path);
4574        my (@pc) = split m#/#, $path;
4575        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4576                                        1, $editor, $pool);
4577        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4578
4579        # Since we can't rely on svn_ra_reparent being available, we'll
4580        # just have to do some magic with set_path to make it so
4581        # we only want a partial path.
4582        my $sp = '';
4583        my $final = join('/', @pc);
4584        while (@pc) {
4585                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4586                $sp .= '/' if length $sp;
4587                $sp .= shift @pc;
4588        }
4589        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4590
4591        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4592
4593        $reporter->finish_report($pool);
4594        $pool->clear;
4595        $editor->{git_commit_ok};
4596}
4597
4598# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4599# svn_ra_reparent didn't work before 1.4)
4600sub gs_do_switch {
4601        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4602        my $path = $gs->{path};
4603        my $pool = SVN::Pool->new;
4604
4605        my $full_url = $self->{url};
4606        my $old_url = $full_url;
4607        $full_url .= '/' . $path if length $path;
4608        my ($ra, $reparented);
4609
4610        if ($old_url =~ m#^svn(\+ssh)?://# ||
4611            ($full_url =~ m#^https?://# &&
4612             escape_url($full_url) ne $full_url)) {
4613                $_[0] = undef;
4614                $self = undef;
4615                $RA = undef;
4616                $ra = Git::SVN::Ra->new($full_url);
4617                $ra_invalid = 1;
4618        } elsif ($old_url ne $full_url) {
4619                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4620                $self->{url} = $full_url;
4621                $reparented = 1;
4622        }
4623
4624        $ra ||= $self;
4625        $url_b = escape_url($url_b);
4626        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4627        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4628        $reporter->set_path('', $rev_a, 0, @lock, $pool);
4629        $reporter->finish_report($pool);
4630
4631        if ($reparented) {
4632                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4633                $self->{url} = $old_url;
4634        }
4635
4636        $pool->clear;
4637        $editor->{git_commit_ok};
4638}
4639
4640sub longest_common_path {
4641        my ($gsv, $globs) = @_;
4642        my %common;
4643        my $common_max = scalar @$gsv;
4644
4645        foreach my $gs (@$gsv) {
4646                my @tmp = split m#/#, $gs->{path};
4647                my $p = '';
4648                foreach (@tmp) {
4649                        $p .= length($p) ? "/$_" : $_;
4650                        $common{$p} ||= 0;
4651                        $common{$p}++;
4652                }
4653        }
4654        $globs ||= [];
4655        $common_max += scalar @$globs;
4656        foreach my $glob (@$globs) {
4657                my @tmp = split m#/#, $glob->{path}->{left};
4658                my $p = '';
4659                foreach (@tmp) {
4660                        $p .= length($p) ? "/$_" : $_;
4661                        $common{$p} ||= 0;
4662                        $common{$p}++;
4663                }
4664        }
4665
4666        my $longest_path = '';
4667        foreach (sort {length $b <=> length $a} keys %common) {
4668                if ($common{$_} == $common_max) {
4669                        $longest_path = $_;
4670                        last;
4671                }
4672        }
4673        $longest_path;
4674}
4675
4676sub gs_fetch_loop_common {
4677        my ($self, $base, $head, $gsv, $globs) = @_;
4678        return if ($base > $head);
4679        my $inc = $_log_window_size;
4680        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4681        my $longest_path = longest_common_path($gsv, $globs);
4682        my $ra_url = $self->{url};
4683        my $find_trailing_edge;
4684        while (1) {
4685                my %revs;
4686                my $err;
4687                my $err_handler = $SVN::Error::handler;
4688                $SVN::Error::handler = sub {
4689                        ($err) = @_;
4690                        skip_unknown_revs($err);
4691                };
4692                sub _cb {
4693                        my ($paths, $r, $author, $date, $log) = @_;
4694                        [ $paths,
4695                          { author => $author, date => $date, log => $log } ];
4696                }
4697                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4698                               sub { $revs{$_[1]} = _cb(@_) });
4699                if ($err) {
4700                        print "Checked through r$max\r";
4701                } else {
4702                        $find_trailing_edge = 1;
4703                }
4704                if ($err and $find_trailing_edge) {
4705                        print STDERR "Path '$longest_path' ",
4706                                     "was probably deleted:\n",
4707                                     $err->expanded_message,
4708                                     "\nWill attempt to follow ",
4709                                     "revisions r$min .. r$max ",
4710                                     "committed before the deletion\n";
4711                        my $hi = $max;
4712                        while (--$hi >= $min) {
4713                                my $ok;
4714                                $self->get_log([$longest_path], $min, $hi,
4715                                               0, 1, 1, sub {
4716                                               $ok = $_[1];
4717                                               $revs{$_[1]} = _cb(@_) });
4718                                if ($ok) {
4719                                        print STDERR "r$min .. r$ok OK\n";
4720                                        last;
4721                                }
4722                        }
4723                        $find_trailing_edge = 0;
4724                }
4725                $SVN::Error::handler = $err_handler;
4726
4727                my %exists = map { $_->{path} => $_ } @$gsv;
4728                foreach my $r (sort {$a <=> $b} keys %revs) {
4729                        my ($paths, $logged) = @{$revs{$r}};
4730
4731                        foreach my $gs ($self->match_globs(\%exists, $paths,
4732                                                           $globs, $r)) {
4733                                if ($gs->rev_map_max >= $r) {
4734                                        next;
4735                                }
4736                                next unless $gs->match_paths($paths, $r);
4737                                $gs->{logged_rev_props} = $logged;
4738                                if (my $last_commit = $gs->last_commit) {
4739                                        $gs->assert_index_clean($last_commit);
4740                                }
4741                                my $log_entry = $gs->do_fetch($paths, $r);
4742                                if ($log_entry) {
4743                                        $gs->do_git_commit($log_entry);
4744                                }
4745                                $INDEX_FILES{$gs->{index}} = 1;
4746                        }
4747                        foreach my $g (@$globs) {
4748                                my $k = "svn-remote.$g->{remote}." .
4749                                        "$g->{t}-maxRev";
4750                                Git::SVN::tmp_config($k, $r);
4751                        }
4752                        if ($ra_invalid) {
4753                                $_[0] = undef;
4754                                $self = undef;
4755                                $RA = undef;
4756                                $self = Git::SVN::Ra->new($ra_url);
4757                                $ra_invalid = undef;
4758                        }
4759                }
4760                # pre-fill the .rev_db since it'll eventually get filled in
4761                # with '0' x40 if something new gets committed
4762                foreach my $gs (@$gsv) {
4763                        next if $gs->rev_map_max >= $max;
4764                        next if defined $gs->rev_map_get($max);
4765                        $gs->rev_map_set($max, 0 x40);
4766                }
4767                foreach my $g (@$globs) {
4768                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4769                        Git::SVN::tmp_config($k, $max);
4770                }
4771                last if $max >= $head;
4772                $min = $max + 1;
4773                $max += $inc;
4774                $max = $head if ($max > $head);
4775        }
4776        Git::SVN::gc();
4777}
4778
4779sub get_dir_globbed {
4780        my ($self, $left, $depth, $r) = @_;
4781
4782        my @x = eval { $self->get_dir($left, $r) };
4783        return unless scalar @x == 3;
4784        my $dirents = $x[0];
4785        my @finalents;
4786        foreach my $de (keys %$dirents) {
4787                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4788                if ($depth > 1) {
4789                        my @args = ("$left/$de", $depth - 1, $r);
4790                        foreach my $dir ($self->get_dir_globbed(@args)) {
4791                                push @finalents, "$de/$dir";
4792                        }
4793                } else {
4794                        push @finalents, $de;
4795                }
4796        }
4797        @finalents;
4798}
4799
4800sub match_globs {
4801        my ($self, $exists, $paths, $globs, $r) = @_;
4802
4803        sub get_dir_check {
4804                my ($self, $exists, $g, $r) = @_;
4805
4806                my @dirs = $self->get_dir_globbed($g->{path}->{left},
4807                                                  $g->{path}->{depth},
4808                                                  $r);
4809
4810                foreach my $de (@dirs) {
4811                        my $p = $g->{path}->full_path($de);
4812                        next if $exists->{$p};
4813                        next if (length $g->{path}->{right} &&
4814                                 ($self->check_path($p, $r) !=
4815                                  $SVN::Node::dir));
4816                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4817                                         $g->{ref}->full_path($de), 1);
4818                }
4819        }
4820        foreach my $g (@$globs) {
4821                if (my $path = $paths->{"/$g->{path}->{left}"}) {
4822                        if ($path->{action} =~ /^[AR]$/) {
4823                                get_dir_check($self, $exists, $g, $r);
4824                        }
4825                }
4826                foreach (keys %$paths) {
4827                        if (/$g->{path}->{left_regex}/ &&
4828                            !/$g->{path}->{regex}/) {
4829                                next if $paths->{$_}->{action} !~ /^[AR]$/;
4830                                get_dir_check($self, $exists, $g, $r);
4831                        }
4832                        next unless /$g->{path}->{regex}/;
4833                        my $p = $1;
4834                        my $pathname = $g->{path}->full_path($p);
4835                        next if $exists->{$pathname};
4836                        next if ($self->check_path($pathname, $r) !=
4837                                 $SVN::Node::dir);
4838                        $exists->{$pathname} = Git::SVN->init(
4839                                              $self->{url}, $pathname, undef,
4840                                              $g->{ref}->full_path($p), 1);
4841                }
4842                my $c = '';
4843                foreach (split m#/#, $g->{path}->{left}) {
4844                        $c .= "/$_";
4845                        next unless ($paths->{$c} &&
4846                                     ($paths->{$c}->{action} =~ /^[AR]$/));
4847                        get_dir_check($self, $exists, $g, $r);
4848                }
4849        }
4850        values %$exists;
4851}
4852
4853sub minimize_url {
4854        my ($self) = @_;
4855        return $self->{url} if ($self->{url} eq $self->{repos_root});
4856        my $url = $self->{repos_root};
4857        my @components = split(m!/!, $self->{svn_path});
4858        my $c = '';
4859        do {
4860                $url .= "/$c" if length $c;
4861                eval {
4862                        my $ra = (ref $self)->new($url);
4863                        my $latest = $ra->get_latest_revnum;
4864                        $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
4865                };
4866        } while ($@ && ($c = shift @components));
4867        $url;
4868}
4869
4870sub can_do_switch {
4871        my $self = shift;
4872        unless (defined $can_do_switch) {
4873                my $pool = SVN::Pool->new;
4874                my $rep = eval {
4875                        $self->do_switch(1, '', 0, $self->{url},
4876                                         SVN::Delta::Editor->new, $pool);
4877                };
4878                if ($@) {
4879                        $can_do_switch = 0;
4880                } else {
4881                        $rep->abort_report($pool);
4882                        $can_do_switch = 1;
4883                }
4884                $pool->clear;
4885        }
4886        $can_do_switch;
4887}
4888
4889sub skip_unknown_revs {
4890        my ($err) = @_;
4891        my $errno = $err->apr_err();
4892        # Maybe the branch we're tracking didn't
4893        # exist when the repo started, so it's
4894        # not an error if it doesn't, just continue
4895        #
4896        # Wonderfully consistent library, eh?
4897        # 160013 - svn:// and file://
4898        # 175002 - http(s)://
4899        # 175007 - http(s):// (this repo required authorization, too...)
4900        #   More codes may be discovered later...
4901        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4902                my $err_key = $err->expanded_message;
4903                # revision numbers change every time, filter them out
4904                $err_key =~ s/\d+/\0/g;
4905                $err_key = "$errno\0$err_key";
4906                unless ($ignored_err{$err_key}) {
4907                        warn "W: Ignoring error from SVN, path probably ",
4908                             "does not exist: ($errno): ",
4909                             $err->expanded_message,"\n";
4910                        warn "W: Do not be alarmed at the above message ",
4911                             "git-svn is just searching aggressively for ",
4912                             "old history.\n",
4913                             "This may take a while on large repositories\n";
4914                        $ignored_err{$err_key} = 1;
4915                }
4916                return;
4917        }
4918        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4919}
4920
4921package Git::SVN::Log;
4922use strict;
4923use warnings;
4924use POSIX qw/strftime/;
4925use Time::Local;
4926use constant commit_log_separator => ('-' x 72) . "\n";
4927use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4928            %rusers $show_commit $incremental/;
4929my $l_fmt;
4930
4931sub cmt_showable {
4932        my ($c) = @_;
4933        return 1 if defined $c->{r};
4934
4935        # big commit message got truncated by the 16k pretty buffer in rev-list
4936        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4937                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4938                @{$c->{l}} = ();
4939                my @log = command(qw/cat-file commit/, $c->{c});
4940
4941                # shift off the headers
4942                shift @log while ($log[0] ne '');
4943                shift @log;
4944
4945                # TODO: make $c->{l} not have a trailing newline in the future
4946                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4947
4948                (undef, $c->{r}, undef) = ::extract_metadata(
4949                                (grep(/^git-svn-id: /, @log))[-1]);
4950        }
4951        return defined $c->{r};
4952}
4953
4954sub log_use_color {
4955        return $color || Git->repository->get_colorbool('color.diff');
4956}
4957
4958sub git_svn_log_cmd {
4959        my ($r_min, $r_max, @args) = @_;
4960        my $head = 'HEAD';
4961        my (@files, @log_opts);
4962        foreach my $x (@args) {
4963                if ($x eq '--' || @files) {
4964                        push @files, $x;
4965                } else {
4966                        if (::verify_ref("$x^0")) {
4967                                $head = $x;
4968                        } else {
4969                                push @log_opts, $x;
4970                        }
4971                }
4972        }
4973
4974        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4975        $gs ||= Git::SVN->_new;
4976        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4977                   $gs->refname);
4978        push @cmd, '-r' unless $non_recursive;
4979        push @cmd, qw/--raw --name-status/ if $verbose;
4980        push @cmd, '--color' if log_use_color();
4981        push @cmd, @log_opts;
4982        if (defined $r_max && $r_max == $r_min) {
4983                push @cmd, '--max-count=1';
4984                if (my $c = $gs->rev_map_get($r_max)) {
4985                        push @cmd, $c;
4986                }
4987        } elsif (defined $r_max) {
4988                if ($r_max < $r_min) {
4989                        ($r_min, $r_max) = ($r_max, $r_min);
4990                }
4991                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4992                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4993                # If there are no commits in the range, both $c_max and $c_min
4994                # will be undefined.  If there is at least 1 commit in the
4995                # range, both will be defined.
4996                return () if !defined $c_min || !defined $c_max;
4997                if ($c_min eq $c_max) {
4998                        push @cmd, '--max-count=1', $c_min;
4999                } else {
5000                        push @cmd, '--boundary', "$c_min..$c_max";
5001                }
5002        }
5003        return (@cmd, @files);
5004}
5005
5006# adapted from pager.c
5007sub config_pager {
5008        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
5009        if (!defined $pager) {
5010                $pager = 'less';
5011        } elsif (length $pager == 0 || $pager eq 'cat') {
5012                $pager = undef;
5013        }
5014        $ENV{GIT_PAGER_IN_USE} = defined($pager);
5015}
5016
5017sub run_pager {
5018        return unless -t *STDOUT && defined $pager;
5019        pipe my ($rfd, $wfd) or return;
5020        defined(my $pid = fork) or ::fatal "Can't fork: $!";
5021        if (!$pid) {
5022                open STDOUT, '>&', $wfd or
5023                                     ::fatal "Can't redirect to stdout: $!";
5024                return;
5025        }
5026        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5027        $ENV{LESS} ||= 'FRSX';
5028        exec $pager or ::fatal "Can't run pager: $! ($pager)";
5029}
5030
5031sub format_svn_date {
5032        # some systmes don't handle or mishandle %z, so be creative.
5033        my $t = shift || time;
5034        my $gm = timelocal(gmtime($t));
5035        my $sign = qw( + + - )[ $t <=> $gm ];
5036        my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5037        return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5038}
5039
5040sub parse_git_date {
5041        my ($t, $tz) = @_;
5042        # Date::Parse isn't in the standard Perl distro :(
5043        if ($tz =~ s/^\+//) {
5044                $t += tz_to_s_offset($tz);
5045        } elsif ($tz =~ s/^\-//) {
5046                $t -= tz_to_s_offset($tz);
5047        }
5048        return $t;
5049}
5050
5051sub set_local_timezone {
5052        if (defined $TZ) {
5053                $ENV{TZ} = $TZ;
5054        } else {
5055                delete $ENV{TZ};
5056        }
5057}
5058
5059sub tz_to_s_offset {
5060        my ($tz) = @_;
5061        $tz =~ s/(\d\d)$//;
5062        return ($1 * 60) + ($tz * 3600);
5063}
5064
5065sub get_author_info {
5066        my ($dest, $author, $t, $tz) = @_;
5067        $author =~ s/(?:^\s*|\s*$)//g;
5068        $dest->{a_raw} = $author;
5069        my $au;
5070        if ($::_authors) {
5071                $au = $rusers{$author} || undef;
5072        }
5073        if (!$au) {
5074                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5075        }
5076        $dest->{t} = $t;
5077        $dest->{tz} = $tz;
5078        $dest->{a} = $au;
5079        $dest->{t_utc} = parse_git_date($t, $tz);
5080}
5081
5082sub process_commit {
5083        my ($c, $r_min, $r_max, $defer) = @_;
5084        if (defined $r_min && defined $r_max) {
5085                if ($r_min == $c->{r} && $r_min == $r_max) {
5086                        show_commit($c);
5087                        return 0;
5088                }
5089                return 1 if $r_min == $r_max;
5090                if ($r_min < $r_max) {
5091                        # we need to reverse the print order
5092                        return 0 if (defined $limit && --$limit < 0);
5093                        push @$defer, $c;
5094                        return 1;
5095                }
5096                if ($r_min != $r_max) {
5097                        return 1 if ($r_min < $c->{r});
5098                        return 1 if ($r_max > $c->{r});
5099                }
5100        }
5101        return 0 if (defined $limit && --$limit < 0);
5102        show_commit($c);
5103        return 1;
5104}
5105
5106sub show_commit {
5107        my $c = shift;
5108        if ($oneline) {
5109                my $x = "\n";
5110                if (my $l = $c->{l}) {
5111                        while ($l->[0] =~ /^\s*$/) { shift @$l }
5112                        $x = $l->[0];
5113                }
5114                $l_fmt ||= 'A' . length($c->{r});
5115                print 'r',pack($l_fmt, $c->{r}),' | ';
5116                print "$c->{c} | " if $show_commit;
5117                print $x;
5118        } else {
5119                show_commit_normal($c);
5120        }
5121}
5122
5123sub show_commit_changed_paths {
5124        my ($c) = @_;
5125        return unless $c->{changed};
5126        print "Changed paths:\n", @{$c->{changed}};
5127}
5128
5129sub show_commit_normal {
5130        my ($c) = @_;
5131        print commit_log_separator, "r$c->{r} | ";
5132        print "$c->{c} | " if $show_commit;
5133        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5134        my $nr_line = 0;
5135
5136        if (my $l = $c->{l}) {
5137                while ($l->[$#$l] eq "\n" && $#$l > 0
5138                                          && $l->[($#$l - 1)] eq "\n") {
5139                        pop @$l;
5140                }
5141                $nr_line = scalar @$l;
5142                if (!$nr_line) {
5143                        print "1 line\n\n\n";
5144                } else {
5145                        if ($nr_line == 1) {
5146                                $nr_line = '1 line';
5147                        } else {
5148                                $nr_line .= ' lines';
5149                        }
5150                        print $nr_line, "\n";
5151                        show_commit_changed_paths($c);
5152                        print "\n";
5153                        print $_ foreach @$l;
5154                }
5155        } else {
5156                print "1 line\n";
5157                show_commit_changed_paths($c);
5158                print "\n";
5159
5160        }
5161        foreach my $x (qw/raw stat diff/) {
5162                if ($c->{$x}) {
5163                        print "\n";
5164                        print $_ foreach @{$c->{$x}}
5165                }
5166        }
5167}
5168
5169sub cmd_show_log {
5170        my (@args) = @_;
5171        my ($r_min, $r_max);
5172        my $r_last = -1; # prevent dupes
5173        set_local_timezone();
5174        if (defined $::_revision) {
5175                if ($::_revision =~ /^(\d+):(\d+)$/) {
5176                        ($r_min, $r_max) = ($1, $2);
5177                } elsif ($::_revision =~ /^\d+$/) {
5178                        $r_min = $r_max = $::_revision;
5179                } else {
5180                        ::fatal "-r$::_revision is not supported, use ",
5181                                "standard 'git log' arguments instead";
5182                }
5183        }
5184
5185        config_pager();
5186        @args = git_svn_log_cmd($r_min, $r_max, @args);
5187        if (!@args) {
5188                print commit_log_separator unless $incremental || $oneline;
5189                return;
5190        }
5191        my $log = command_output_pipe(@args);
5192        run_pager();
5193        my (@k, $c, $d, $stat);
5194        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5195        while (<$log>) {
5196                if (/^${esc_color}commit -?($::sha1_short)/o) {
5197                        my $cmt = $1;
5198                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5199                                $r_last = $c->{r};
5200                                process_commit($c, $r_min, $r_max, \@k) or
5201                                                                goto out;
5202                        }
5203                        $d = undef;
5204                        $c = { c => $cmt };
5205                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5206                        get_author_info($c, $1, $2, $3);
5207                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5208                        # ignore
5209                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5210                        push @{$c->{raw}}, $_;
5211                } elsif (/^${esc_color}[ACRMDT]\t/) {
5212                        # we could add $SVN->{svn_path} here, but that requires
5213                        # remote access at the moment (repo_path_split)...
5214                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
5215                        push @{$c->{changed}}, $_;
5216                } elsif (/^${esc_color}diff /o) {
5217                        $d = 1;
5218                        push @{$c->{diff}}, $_;
5219                } elsif ($d) {
5220                        push @{$c->{diff}}, $_;
5221                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5222                          $esc_color*[\+\-]*$esc_color$/x) {
5223                        $stat = 1;
5224                        push @{$c->{stat}}, $_;
5225                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5226                        push @{$c->{stat}}, $_;
5227                        $stat = undef;
5228                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
5229                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5230                } elsif (s/^${esc_color}    //o) {
5231                        push @{$c->{l}}, $_;
5232                }
5233        }
5234        if ($c && defined $c->{r} && $c->{r} != $r_last) {
5235                $r_last = $c->{r};
5236                process_commit($c, $r_min, $r_max, \@k);
5237        }
5238        if (@k) {
5239                ($r_min, $r_max) = ($r_max, $r_min);
5240                process_commit($_, $r_min, $r_max) foreach reverse @k;
5241        }
5242out:
5243        close $log;
5244        print commit_log_separator unless $incremental || $oneline;
5245}
5246
5247sub cmd_blame {
5248        my $path = pop;
5249
5250        config_pager();
5251        run_pager();
5252
5253        my ($fh, $ctx, $rev);
5254
5255        if ($_git_format) {
5256                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5257                while (my $line = <$fh>) {
5258                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5259                                # Uncommitted edits show up as a rev ID of
5260                                # all zeros, which we can't look up with
5261                                # cmt_metadata
5262                                if ($1 !~ /^0+$/) {
5263                                        (undef, $rev, undef) =
5264                                                ::cmt_metadata($1);
5265                                        $rev = '0' if (!$rev);
5266                                } else {
5267                                        $rev = '0';
5268                                }
5269                                $rev = sprintf('%-10s', $rev);
5270                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5271                        }
5272                        print $line;
5273                }
5274        } else {
5275                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5276                                                  '--', $path);
5277                my ($sha1);
5278                my %authors;
5279                my @buffer;
5280                my %dsha; #distinct sha keys
5281
5282                while (my $line = <$fh>) {
5283                        push @buffer, $line;
5284                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5285                                $dsha{$1} = 1;
5286                        }
5287                }
5288
5289                my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5290
5291                foreach my $line (@buffer) {
5292                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5293                                $rev = $s2r->{$1};
5294                                $rev = '0' if (!$rev)
5295                        }
5296                        elsif ($line =~ /^author (.*)/) {
5297                                $authors{$rev} = $1;
5298                                $authors{$rev} =~ s/\s/_/g;
5299                        }
5300                        elsif ($line =~ /^\t(.*)$/) {
5301                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5302                        }
5303                }
5304        }
5305        command_close_pipe($fh, $ctx);
5306}
5307
5308package Git::SVN::Migration;
5309# these version numbers do NOT correspond to actual version numbers
5310# of git nor git-svn.  They are just relative.
5311#
5312# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5313#
5314# v1 layout: .git/$id/info/url, refs/remotes/$id
5315#
5316# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5317#
5318# v3 layout: .git/svn/$id, refs/remotes/$id
5319#            - info/url may remain for backwards compatibility
5320#            - this is what we migrate up to this layout automatically,
5321#            - this will be used by git svn init on single branches
5322# v3.1 layout (auto migrated):
5323#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5324#              for backwards compatibility
5325#
5326# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5327#            - this is only created for newly multi-init-ed
5328#              repositories.  Similar in spirit to the
5329#              --use-separate-remotes option in git-clone (now default)
5330#            - we do not automatically migrate to this (following
5331#              the example set by core git)
5332#
5333# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5334#            - newer, more-efficient format that uses 24-bytes per record
5335#              with no filler space.
5336#            - use xxd -c24 < .rev_map.$UUID to view and debug
5337#            - This is a one-way migration, repositories updated to the
5338#              new format will not be able to use old git-svn without
5339#              rebuilding the .rev_db.  Rebuilding the rev_db is not
5340#              possible if noMetadata or useSvmProps are set; but should
5341#              be no problem for users that use the (sensible) defaults.
5342use strict;
5343use warnings;
5344use Carp qw/croak/;
5345use File::Path qw/mkpath/;
5346use File::Basename qw/dirname basename/;
5347use vars qw/$_minimize/;
5348
5349sub migrate_from_v0 {
5350        my $git_dir = $ENV{GIT_DIR};
5351        return undef unless -d $git_dir;
5352        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5353        my $migrated = 0;
5354        while (<$fh>) {
5355                chomp;
5356                my ($id, $orig_ref) = ($_, $_);
5357                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5358                next unless -f "$git_dir/$id/info/url";
5359                my $new_ref = "refs/remotes/$id";
5360                if (::verify_ref("$new_ref^0")) {
5361                        print STDERR "W: $orig_ref is probably an old ",
5362                                     "branch used by an ancient version of ",
5363                                     "git-svn.\n",
5364                                     "However, $new_ref also exists.\n",
5365                                     "We will not be able ",
5366                                     "to use this branch until this ",
5367                                     "ambiguity is resolved.\n";
5368                        next;
5369                }
5370                print STDERR "Migrating from v0 layout...\n" if !$migrated;
5371                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5372                command_noisy('update-ref', $new_ref, $orig_ref);
5373                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5374                $migrated++;
5375        }
5376        command_close_pipe($fh, $ctx);
5377        print STDERR "Done migrating from v0 layout...\n" if $migrated;
5378        $migrated;
5379}
5380
5381sub migrate_from_v1 {
5382        my $git_dir = $ENV{GIT_DIR};
5383        my $migrated = 0;
5384        return $migrated unless -d $git_dir;
5385        my $svn_dir = "$git_dir/svn";
5386
5387        # just in case somebody used 'svn' as their $id at some point...
5388        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5389
5390        print STDERR "Migrating from a git-svn v1 layout...\n";
5391        mkpath([$svn_dir]);
5392        print STDERR "Data from a previous version of git-svn exists, but\n\t",
5393                     "$svn_dir\n\t(required for this version ",
5394                     "($::VERSION) of git-svn) does not exist.\n";
5395        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5396        while (<$fh>) {
5397                my $x = $_;
5398                next unless $x =~ s#^refs/remotes/##;
5399                chomp $x;
5400                next unless -f "$git_dir/$x/info/url";
5401                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5402                next unless $u;
5403                my $dn = dirname("$git_dir/svn/$x");
5404                mkpath([$dn]) unless -d $dn;
5405                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5406                        mkpath(["$git_dir/svn/svn"]);
5407                        print STDERR " - $git_dir/$x/info => ",
5408                                        "$git_dir/svn/$x/info\n";
5409                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5410                               croak "$!: $x";
5411                        # don't worry too much about these, they probably
5412                        # don't exist with repos this old (save for index,
5413                        # and we can easily regenerate that)
5414                        foreach my $f (qw/unhandled.log index .rev_db/) {
5415                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5416                        }
5417                } else {
5418                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5419                        rename "$git_dir/$x", "$git_dir/svn/$x" or
5420                               croak "$!: $x";
5421                }
5422                $migrated++;
5423        }
5424        command_close_pipe($fh, $ctx);
5425        print STDERR "Done migrating from a git-svn v1 layout\n";
5426        $migrated;
5427}
5428
5429sub read_old_urls {
5430        my ($l_map, $pfx, $path) = @_;
5431        my @dir;
5432        foreach (<$path/*>) {
5433                if (-r "$_/info/url") {
5434                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5435                        my $ref_id = $pfx . basename $_;
5436                        my $url = ::file_to_s("$_/info/url");
5437                        $l_map->{$ref_id} = $url;
5438                } elsif (-d $_) {
5439                        push @dir, $_;
5440                }
5441        }
5442        foreach (@dir) {
5443                my $x = $_;
5444                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5445                read_old_urls($l_map, $x, $_);
5446        }
5447}
5448
5449sub migrate_from_v2 {
5450        my @cfg = command(qw/config -l/);
5451        return if grep /^svn-remote\..+\.url=/, @cfg;
5452        my %l_map;
5453        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5454        my $migrated = 0;
5455
5456        foreach my $ref_id (sort keys %l_map) {
5457                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5458                if ($@) {
5459                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5460                }
5461                $migrated++;
5462        }
5463        $migrated;
5464}
5465
5466sub minimize_connections {
5467        my $r = Git::SVN::read_all_remotes();
5468        my $new_urls = {};
5469        my $root_repos = {};
5470        foreach my $repo_id (keys %$r) {
5471                my $url = $r->{$repo_id}->{url} or next;
5472                my $fetch = $r->{$repo_id}->{fetch} or next;
5473                my $ra = Git::SVN::Ra->new($url);
5474
5475                # skip existing cases where we already connect to the root
5476                if (($ra->{url} eq $ra->{repos_root}) ||
5477                    ($ra->{repos_root} eq $repo_id)) {
5478                        $root_repos->{$ra->{url}} = $repo_id;
5479                        next;
5480                }
5481
5482                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5483                my $root_path = $ra->{url};
5484                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5485                foreach my $path (keys %$fetch) {
5486                        my $ref_id = $fetch->{$path};
5487                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5488
5489                        # make sure we can read when connecting to
5490                        # a higher level of a repository
5491                        my ($last_rev, undef) = $gs->last_rev_commit;
5492                        if (!defined $last_rev) {
5493                                $last_rev = eval {
5494                                        $root_ra->get_latest_revnum;
5495                                };
5496                                next if $@;
5497                        }
5498                        my $new = $root_path;
5499                        $new .= length $path ? "/$path" : '';
5500                        eval {
5501                                $root_ra->get_log([$new], $last_rev, $last_rev,
5502                                                  0, 0, 1, sub { });
5503                        };
5504                        next if $@;
5505                        $new_urls->{$ra->{repos_root}}->{$new} =
5506                                { ref_id => $ref_id,
5507                                  old_repo_id => $repo_id,
5508                                  old_path => $path };
5509                }
5510        }
5511
5512        my @emptied;
5513        foreach my $url (keys %$new_urls) {
5514                # see if we can re-use an existing [svn-remote "repo_id"]
5515                # instead of creating a(n ugly) new section:
5516                my $repo_id = $root_repos->{$url} || $url;
5517
5518                my $fetch = $new_urls->{$url};
5519                foreach my $path (keys %$fetch) {
5520                        my $x = $fetch->{$path};
5521                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5522                        my $pfx = "svn-remote.$x->{old_repo_id}";
5523
5524                        my $old_fetch = quotemeta("$x->{old_path}:".
5525                                                  "$x->{ref_id}");
5526                        command_noisy(qw/config --unset/,
5527                                      "$pfx.fetch", '^'. $old_fetch . '$');
5528                        delete $r->{$x->{old_repo_id}}->
5529                               {fetch}->{$x->{old_path}};
5530                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5531                                command_noisy(qw/config --unset/,
5532                                              "$pfx.url");
5533                                push @emptied, $x->{old_repo_id}
5534                        }
5535                }
5536        }
5537        if (@emptied) {
5538                my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5539                print STDERR <<EOF;
5540The following [svn-remote] sections in your config file ($file) are empty
5541and can be safely removed:
5542EOF
5543                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5544        }
5545}
5546
5547sub migration_check {
5548        migrate_from_v0();
5549        migrate_from_v1();
5550        migrate_from_v2();
5551        minimize_connections() if $_minimize;
5552}
5553
5554package Git::IndexInfo;
5555use strict;
5556use warnings;
5557use Git qw/command_input_pipe command_close_pipe/;
5558
5559sub new {
5560        my ($class) = @_;
5561        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5562        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5563}
5564
5565sub remove {
5566        my ($self, $path) = @_;
5567        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5568                return ++$self->{nr};
5569        }
5570        undef;
5571}
5572
5573sub update {
5574        my ($self, $mode, $hash, $path) = @_;
5575        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5576                return ++$self->{nr};
5577        }
5578        undef;
5579}
5580
5581sub DESTROY {
5582        my ($self) = @_;
5583        command_close_pipe($self->{gui}, $self->{ctx});
5584}
5585
5586package Git::SVN::GlobSpec;
5587use strict;
5588use warnings;
5589
5590sub new {
5591        my ($class, $glob) = @_;
5592        my $re = $glob;
5593        $re =~ s!/+$!!g; # no need for trailing slashes
5594        $re =~ m!^([^*]*)(\*(?:/\*)*)(.*)$!;
5595        my $temp = $re;
5596        my ($left, $right) = ($1, $3);
5597        $re = $2;
5598        my $depth = $re =~ tr/*/*/;
5599        if ($depth != $temp =~ tr/*/*/) {
5600                die "Only one set of wildcard directories " .
5601                        "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5602        }
5603        if ($depth == 0) {
5604                die "One '*' is needed for glob: '$glob'\n";
5605        }
5606        $re =~ s!\*!\[^/\]*!g;
5607        $re = quotemeta($left) . "($re)" . quotemeta($right);
5608        if (length $left && !($left =~ s!/+$!!g)) {
5609                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5610        }
5611        if (length $right && !($right =~ s!^/+!!g)) {
5612                die "Missing leading '/' on right side of: '$glob' ($right)\n";
5613        }
5614        my $left_re = qr/^\/\Q$left\E(\/|$)/;
5615        bless { left => $left, right => $right, left_regex => $left_re,
5616                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5617}
5618
5619sub full_path {
5620        my ($self, $path) = @_;
5621        return (length $self->{left} ? "$self->{left}/" : '') .
5622               $path . (length $self->{right} ? "/$self->{right}" : '');
5623}
5624
5625__END__
5626
5627Data structures:
5628
5629
5630$remotes = { # returned by read_all_remotes()
5631        'svn' => {
5632                # svn-remote.svn.url=https://svn.musicpd.org
5633                url => 'https://svn.musicpd.org',
5634                # svn-remote.svn.fetch=mpd/trunk:trunk
5635                fetch => {
5636                        'mpd/trunk' => 'trunk',
5637                },
5638                # svn-remote.svn.tags=mpd/tags/*:tags/*
5639                tags => {
5640                        path => {
5641                                left => 'mpd/tags',
5642                                right => '',
5643                                regex => qr!mpd/tags/([^/]+)$!,
5644                                glob => 'tags/*',
5645                        },
5646                        ref => {
5647                                left => 'tags',
5648                                right => '',
5649                                regex => qr!tags/([^/]+)$!,
5650                                glob => 'tags/*',
5651                        },
5652                }
5653        }
5654};
5655
5656$log_entry hashref as returned by libsvn_log_entry()
5657{
5658        log => 'whitespace-formatted log entry
5659',                                              # trailing newline is preserved
5660        revision => '8',                        # integer
5661        date => '2004-02-24T17:01:44.108345Z',  # commit date
5662        author => 'committer name'
5663};
5664
5665
5666# this is generated by generate_diff();
5667@mods = array of diff-index line hashes, each element represents one line
5668        of diff-index output
5669
5670diff-index line ($m hash)
5671{
5672        mode_a => first column of diff-index output, no leading ':',
5673        mode_b => second column of diff-index output,
5674        sha1_b => sha1sum of the final blob,
5675        chg => change type [MCRADT],
5676        file_a => original file name of a file (iff chg is 'C' or 'R')
5677        file_b => new/current file name of a file (any chg)
5678}
5679;
5680
5681# retval of read_url_paths{,_all}();
5682$l_map = {
5683        # repository root url
5684        'https://svn.musicpd.org' => {
5685                # repository path               # GIT_SVN_ID
5686                'mpd/trunk'             =>      'trunk',
5687                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
5688        },
5689}
5690
5691Notes:
5692        I don't trust the each() function on unless I created %hash myself
5693        because the internal iterator may not have started at base.