Andrew's git - gitweb.git/blob

   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $sha1 $sha1_short $_revision
   8                $_q $_authors %users/;
   9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  10$VERSION = '@@GIT_VERSION@@';
  11
  12my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
  13$ENV{GIT_DIR} ||= '.git';
  14$Git::SVN::default_repo_id = 'svn';
  15$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  16$Git::SVN::Ra::_log_window_size = 100;
  17
  18$Git::SVN::Log::TZ = $ENV{TZ};
  19$ENV{TZ} = 'UTC';
  20$| = 1; # unbuffer STDOUT
  21
  22sub fatal (@) { print STDERR @_; exit 1 }
  23require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  24require SVN::Ra;
  25require SVN::Delta;
  26if ($SVN::Core::VERSION lt '1.1.0') {
  27        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
  28}
  29push @Git::SVN::Ra::ISA, 'SVN::Ra';
  30push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  31push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  32use Carp qw/croak/;
  33use IO::File qw//;
  34use File::Basename qw/dirname basename/;
  35use File::Path qw/mkpath/;
  36use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  37use IPC::Open3;
  38use Git;
  39
  40BEGIN {
  41        my $s;
  42        foreach (qw/command command_oneline command_noisy command_output_pipe
  43                    command_input_pipe command_close_pipe/) {
  44                $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
  45                      "*Git::SVN::Migration::$_ = ".
  46                      "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
  47        }
  48        eval $s;
  49}
  50
  51my ($SVN);
  52
  53$sha1 = qr/[a-f\d]{40}/;
  54$sha1_short = qr/[a-f\d]{4,40}/;
  55my ($_stdin, $_help, $_edit,
  56        $_message, $_file,
  57        $_template, $_shared,
  58        $_version, $_fetch_all, $_no_rebase,
  59        $_merge, $_strategy, $_dry_run, $_local,
  60        $_prefix, $_no_checkout, $_verbose);
  61$Git::SVN::_follow_parent = 1;
  62my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  63                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  64                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  65my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  66                'authors-file|A=s' => \$_authors,
  67                'repack:i' => \$Git::SVN::_repack,
  68                'noMetadata' => \$Git::SVN::_no_metadata,
  69                'useSvmProps' => \$Git::SVN::_use_svm_props,
  70                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  71                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  72                'no-checkout' => \$_no_checkout,
  73                'quiet|q' => \$_q,
  74                'repack-flags|repack-args|repack-opts=s' =>
  75                   \$Git::SVN::_repack_flags,
  76                %remote_opts );
  77
  78my ($_trunk, $_tags, $_branches);
  79my %icv;
  80my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
  81                  'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
  82                  'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
  83                  'minimize-url|m' => \$Git::SVN::_minimize_url,
  84                  'no-metadata' => sub { $icv{noMetadata} = 1 },
  85                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
  86                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
  87                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
  88                  %remote_opts );
  89my %cmt_opts = ( 'edit|e' => \$_edit,
  90                'rmdir' => \$SVN::Git::Editor::_rmdir,
  91                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
  92                'l=i' => \$SVN::Git::Editor::_rename_limit,
  93                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
  94);
  95
  96my %cmd = (
  97        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
  98                        { 'revision|r=s' => \$_revision,
  99                          'fetch-all|all' => \$_fetch_all,
 100                           %fc_opts } ],
 101        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 102                        { 'revision|r=s' => \$_revision,
 103                           %fc_opts, %init_opts } ],
 104        init => [ \&cmd_init, "Initialize a repo for tracking" .
 105                          " (requires URL argument)",
 106                          \%init_opts ],
 107        'multi-init' => [ \&cmd_multi_init,
 108                          "Deprecated alias for ".
 109                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 110                          \%init_opts ],
 111        dcommit => [ \&cmd_dcommit,
 112                     'Commit several diffs to merge with upstream',
 113                        { 'merge|m|M' => \$_merge,
 114                          'strategy|s=s' => \$_strategy,
 115                          'verbose|v' => \$_verbose,
 116                          'dry-run|n' => \$_dry_run,
 117                          'fetch-all|all' => \$_fetch_all,
 118                          'no-rebase' => \$_no_rebase,
 119                        %cmt_opts, %fc_opts } ],
 120        'set-tree' => [ \&cmd_set_tree,
 121                        "Set an SVN repository to a git tree-ish",
 122                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 123        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 124                        { 'revision|r=i' => \$_revision } ],
 125        'multi-fetch' => [ \&cmd_multi_fetch,
 126                           "Deprecated alias for $0 fetch --all",
 127                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 128        'migrate' => [ sub { },
 129                       # no-op, we automatically run this anyways,
 130                       'Migrate configuration/metadata/layout from
 131                        previous versions of git-svn',
 132                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 133                         %remote_opts } ],
 134        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 135                        { 'limit=i' => \$Git::SVN::Log::limit,
 136                          'revision|r=s' => \$_revision,
 137                          'verbose|v' => \$Git::SVN::Log::verbose,
 138                          'incremental' => \$Git::SVN::Log::incremental,
 139                          'oneline' => \$Git::SVN::Log::oneline,
 140                          'show-commit' => \$Git::SVN::Log::show_commit,
 141                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 142                          'authors-file|A=s' => \$_authors,
 143                          'color' => \$Git::SVN::Log::color,
 144                          'pager=s' => \$Git::SVN::Log::pager,
 145                        } ],
 146        'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
 147                        { } ],
 148        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 149                        { 'merge|m|M' => \$_merge,
 150                          'verbose|v' => \$_verbose,
 151                          'strategy|s=s' => \$_strategy,
 152                          'local|l' => \$_local,
 153                          'fetch-all|all' => \$_fetch_all,
 154                          %fc_opts } ],
 155        'commit-diff' => [ \&cmd_commit_diff,
 156                           'Commit a diff between two trees',
 157                        { 'message|m=s' => \$_message,
 158                          'file|F=s' => \$_file,
 159                          'revision|r=s' => \$_revision,
 160                        %cmt_opts } ],
 161);
 162
 163my $cmd;
 164for (my $i = 0; $i < @ARGV; $i++) {
 165        if (defined $cmd{$ARGV[$i]}) {
 166                $cmd = $ARGV[$i];
 167                splice @ARGV, $i, 1;
 168                last;
 169        }
 170};
 171
 172my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 173
 174read_repo_config(\%opts);
 175Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
 176my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 177                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 178                    'id|i=s' => \$Git::SVN::default_ref_id,
 179                    'svn-remote|remote|R=s' => sub {
 180                       $Git::SVN::no_reuse_existing = 1;
 181                       $Git::SVN::default_repo_id = $_[1] });
 182exit 1 if (!$rv && $cmd && $cmd ne 'log');
 183
 184usage(0) if $_help;
 185version() if $_version;
 186usage(1) unless defined $cmd;
 187load_authors() if $_authors;
 188
 189# make sure we're always running
 190unless ($cmd =~ /(?:clone|init|multi-init)$/) {
 191        unless (-d $ENV{GIT_DIR}) {
 192                if ($git_dir_user_set) {
 193                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 194                            "but it is not a directory\n";
 195                }
 196                my $git_dir = delete $ENV{GIT_DIR};
 197                chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
 198                unless (length $cdup) {
 199                        die "Already at toplevel, but $git_dir ",
 200                            "not found '$cdup'\n";
 201                }
 202                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 203                unless (-d $git_dir) {
 204                        die "$git_dir still not found after going to ",
 205                            "'$cdup'\n";
 206                }
 207                $ENV{GIT_DIR} = $git_dir;
 208        }
 209}
 210unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 211        Git::SVN::Migration::migration_check();
 212}
 213Git::SVN::init_vars();
 214eval {
 215        Git::SVN::verify_remotes_sanity();
 216        $cmd{$cmd}->[0]->(@ARGV);
 217};
 218fatal $@ if $@;
 219post_fetch_checkout();
 220exit 0;
 221
 222####################### primary functions ######################
 223sub usage {
 224        my $exit = shift || 0;
 225        my $fd = $exit ? \*STDERR : \*STDOUT;
 226        print $fd <<"";
 227git-svn - bidirectional operations between a single Subversion tree and git
 228Usage: $0 <command> [options] [arguments]\n
 229
 230        print $fd "Available commands:\n" unless $cmd;
 231
 232        foreach (sort keys %cmd) {
 233                next if $cmd && $cmd ne $_;
 234                next if /^multi-/; # don't show deprecated commands
 235                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 236                foreach (keys %{$cmd{$_}->[2]}) {
 237                        # mixed-case options are for .git/config only
 238                        next if /[A-Z]/ && /^[a-z]+$/i;
 239                        # prints out arguments as they should be passed:
 240                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 241                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 242                                                        "--$_" : "-$_" }
 243                                                split /\|/,$_)," $x\n";
 244                }
 245        }
 246        print $fd <<"";
 247\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 248arbitrary identifier if you're tracking multiple SVN branches/repositories in
 249one git repository and want to keep them separate.  See git-svn(1) for more
 250information.
 251
 252        exit $exit;
 253}
 254
 255sub version {
 256        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 257        exit 0;
 258}
 259
 260sub do_git_init_db {
 261        unless (-d $ENV{GIT_DIR}) {
 262                my @init_db = ('init');
 263                push @init_db, "--template=$_template" if defined $_template;
 264                if (defined $_shared) {
 265                        if ($_shared =~ /[a-z]/) {
 266                                push @init_db, "--shared=$_shared";
 267                        } else {
 268                                push @init_db, "--shared";
 269                        }
 270                }
 271                command_noisy(@init_db);
 272        }
 273        my $set;
 274        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 275        foreach my $i (keys %icv) {
 276                die "'$set' and '$i' cannot both be set\n" if $set;
 277                next unless defined $icv{$i};
 278                command_noisy('config', "$pfx.$i", $icv{$i});
 279                $set = $i;
 280        }
 281}
 282
 283sub init_subdir {
 284        my $repo_path = shift or return;
 285        mkpath([$repo_path]) unless -d $repo_path;
 286        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 287        $ENV{GIT_DIR} = '.git';
 288}
 289
 290sub cmd_clone {
 291        my ($url, $path) = @_;
 292        if (!defined $path &&
 293            (defined $_trunk || defined $_branches || defined $_tags) &&
 294            $url !~ m#^[a-z\+]+://#) {
 295                $path = $url;
 296        }
 297        $path = basename($url) if !defined $path || !length $path;
 298        cmd_init($url, $path);
 299        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 300}
 301
 302sub cmd_init {
 303        if (defined $_trunk || defined $_branches || defined $_tags) {
 304                return cmd_multi_init(@_);
 305        }
 306        my $url = shift or die "SVN repository location required ",
 307                               "as a command-line argument\n";
 308        init_subdir(@_);
 309        do_git_init_db();
 310
 311        Git::SVN->init($url);
 312}
 313
 314sub cmd_fetch {
 315        if (grep /^\d+=./, @_) {
 316                die "'<rev>=<commit>' fetch arguments are ",
 317                    "no longer supported.\n";
 318        }
 319        my ($remote) = @_;
 320        if (@_ > 1) {
 321                die "Usage: $0 fetch [--all] [svn-remote]\n";
 322        }
 323        $remote ||= $Git::SVN::default_repo_id;
 324        if ($_fetch_all) {
 325                cmd_multi_fetch();
 326        } else {
 327                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 328        }
 329}
 330
 331sub cmd_set_tree {
 332        my (@commits) = @_;
 333        if ($_stdin || !@commits) {
 334                print "Reading from stdin...\n";
 335                @commits = ();
 336                while (<STDIN>) {
 337                        if (/\b($sha1_short)\b/o) {
 338                                unshift @commits, $1;
 339                        }
 340                }
 341        }
 342        my @revs;
 343        foreach my $c (@commits) {
 344                my @tmp = command('rev-parse',$c);
 345                if (scalar @tmp == 1) {
 346                        push @revs, $tmp[0];
 347                } elsif (scalar @tmp > 1) {
 348                        push @revs, reverse(command('rev-list',@tmp));
 349                } else {
 350                        fatal "Failed to rev-parse $c\n";
 351                }
 352        }
 353        my $gs = Git::SVN->new;
 354        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 355        $gs->fetch;
 356        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 357                fatal "There are new revisions that were fetched ",
 358                      "and need to be merged (or acknowledged) ",
 359                      "before committing.\nlast rev: $r_last\n",
 360                      " current: $gs->{last_rev}\n";
 361        }
 362        $gs->set_tree($_) foreach @revs;
 363        print "Done committing ",scalar @revs," revisions to SVN\n";
 364}
 365
 366sub cmd_dcommit {
 367        my $head = shift;
 368        $head ||= 'HEAD';
 369        my @refs;
 370        my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
 371        unless ($gs) {
 372                die "Unable to determine upstream SVN information from ",
 373                    "$head history\n";
 374        }
 375        my $last_rev;
 376        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 377        foreach my $d (@$linear_refs) {
 378                unless (defined $last_rev) {
 379                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 380                        unless (defined $last_rev) {
 381                                fatal "Unable to extract revision information ",
 382                                      "from commit $d~1\n";
 383                        }
 384                }
 385                if ($_dry_run) {
 386                        print "diff-tree $d~1 $d\n";
 387                } else {
 388                        my %ed_opts = ( r => $last_rev,
 389                                        log => get_commit_entry($d)->{log},
 390                                        ra => Git::SVN::Ra->new($gs->full_url),
 391                                        tree_a => "$d~1",
 392                                        tree_b => $d,
 393                                        editor_cb => sub {
 394                                               print "Committed r$_[0]\n";
 395                                               $last_rev = $_[0]; },
 396                                        svn_path => '');
 397                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 398                                print "No changes\n$d~1 == $d\n";
 399                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 400                                $gs->{inject_parents_dcommit}->{$last_rev} =
 401                                                               $parents->{$d};
 402                        }
 403                }
 404        }
 405        return if $_dry_run;
 406        unless ($gs) {
 407                warn "Could not determine fetch information for $url\n",
 408                     "Will not attempt to fetch and rebase commits.\n",
 409                     "This probably means you have useSvmProps and should\n",
 410                     "now resync your SVN::Mirror repository.\n";
 411                return;
 412        }
 413        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 414        unless ($_no_rebase) {
 415                # we always want to rebase against the current HEAD, not any
 416                # head that was passed to us
 417                my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
 418                my @finish;
 419                if (@diff) {
 420                        @finish = rebase_cmd();
 421                        print STDERR "W: HEAD and ", $gs->refname, " differ, ",
 422                                     "using @finish:\n", "@diff";
 423                } else {
 424                        print "No changes between current HEAD and ",
 425                              $gs->refname, "\nResetting to the latest ",
 426                              $gs->refname, "\n";
 427                        @finish = qw/reset --mixed/;
 428                }
 429                command_noisy(@finish, $gs->refname);
 430        }
 431}
 432
 433sub cmd_find_rev {
 434        my $revision_or_hash = shift;
 435        my $result;
 436        if ($revision_or_hash =~ /^r\d+$/) {
 437                my $head = shift;
 438                $head ||= 'HEAD';
 439                my @refs;
 440                my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
 441                unless ($gs) {
 442                        die "Unable to determine upstream SVN information from ",
 443                            "$head history\n";
 444                }
 445                my $desired_revision = substr($revision_or_hash, 1);
 446                $result = $gs->rev_db_get($desired_revision);
 447        } else {
 448                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 449                $result = $rev;
 450        }
 451        print "$result\n" if $result;
 452}
 453
 454sub cmd_rebase {
 455        command_noisy(qw/update-index --refresh/);
 456        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 457        unless ($gs) {
 458                die "Unable to determine upstream SVN information from ",
 459                    "working tree history\n";
 460        }
 461        if (command(qw/diff-index HEAD --/)) {
 462                print STDERR "Cannot rebase with uncommited changes:\n";
 463                command_noisy('status');
 464                exit 1;
 465        }
 466        unless ($_local) {
 467                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 468        }
 469        command_noisy(rebase_cmd(), $gs->refname);
 470}
 471
 472sub cmd_show_ignore {
 473        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 474        $gs ||= Git::SVN->new;
 475        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 476        $gs->traverse_ignore(\*STDOUT, $gs->{path}, $r);
 477}
 478
 479sub cmd_multi_init {
 480        my $url = shift;
 481        unless (defined $_trunk || defined $_branches || defined $_tags) {
 482                usage(1);
 483        }
 484
 485        # there are currently some bugs that prevent multi-init/multi-fetch
 486        # setups from working well without this.
 487        $Git::SVN::_minimize_url = 1;
 488
 489        $_prefix = '' unless defined $_prefix;
 490        if (defined $url) {
 491                $url =~ s#/+$##;
 492                init_subdir(@_);
 493        }
 494        do_git_init_db();
 495        if (defined $_trunk) {
 496                my $trunk_ref = $_prefix . 'trunk';
 497                # try both old-style and new-style lookups:
 498                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 499                unless ($gs_trunk) {
 500                        my ($trunk_url, $trunk_path) =
 501                                              complete_svn_url($url, $_trunk);
 502                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 503                                                   undef, $trunk_ref);
 504                }
 505        }
 506        return unless defined $_branches || defined $_tags;
 507        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 508        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 509        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 510}
 511
 512sub cmd_multi_fetch {
 513        my $remotes = Git::SVN::read_all_remotes();
 514        foreach my $repo_id (sort keys %$remotes) {
 515                if ($remotes->{$repo_id}->{url}) {
 516                        Git::SVN::fetch_all($repo_id, $remotes);
 517                }
 518        }
 519}
 520
 521# this command is special because it requires no metadata
 522sub cmd_commit_diff {
 523        my ($ta, $tb, $url) = @_;
 524        my $usage = "Usage: $0 commit-diff -r<revision> ".
 525                    "<tree-ish> <tree-ish> [<URL>]\n";
 526        fatal($usage) if (!defined $ta || !defined $tb);
 527        my $svn_path;
 528        if (!defined $url) {
 529                my $gs = eval { Git::SVN->new };
 530                if (!$gs) {
 531                        fatal("Needed URL or usable git-svn --id in ",
 532                              "the command-line\n", $usage);
 533                }
 534                $url = $gs->{url};
 535                $svn_path = $gs->{path};
 536        }
 537        unless (defined $_revision) {
 538                fatal("-r|--revision is a required argument\n", $usage);
 539        }
 540        if (defined $_message && defined $_file) {
 541                fatal("Both --message/-m and --file/-F specified ",
 542                      "for the commit message.\n",
 543                      "I have no idea what you mean\n");
 544        }
 545        if (defined $_file) {
 546                $_message = file_to_s($_file);
 547        } else {
 548                $_message ||= get_commit_entry($tb)->{log};
 549        }
 550        my $ra ||= Git::SVN::Ra->new($url);
 551        $svn_path ||= $ra->{svn_path};
 552        my $r = $_revision;
 553        if ($r eq 'HEAD') {
 554                $r = $ra->get_latest_revnum;
 555        } elsif ($r !~ /^\d+$/) {
 556                die "revision argument: $r not understood by git-svn\n";
 557        }
 558        my %ed_opts = ( r => $r,
 559                        log => $_message,
 560                        ra => $ra,
 561                        tree_a => $ta,
 562                        tree_b => $tb,
 563                        editor_cb => sub { print "Committed r$_[0]\n" },
 564                        svn_path => $svn_path );
 565        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 566                print "No changes\n$ta == $tb\n";
 567        }
 568}
 569
 570########################### utility functions #########################
 571
 572sub rebase_cmd {
 573        my @cmd = qw/rebase/;
 574        push @cmd, '-v' if $_verbose;
 575        push @cmd, qw/--merge/ if $_merge;
 576        push @cmd, "--strategy=$_strategy" if $_strategy;
 577        @cmd;
 578}
 579
 580sub post_fetch_checkout {
 581        return if $_no_checkout;
 582        my $gs = $Git::SVN::_head or return;
 583        return if verify_ref('refs/heads/master^0');
 584
 585        my $valid_head = verify_ref('HEAD^0');
 586        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
 587        return if ($valid_head || !verify_ref('HEAD^0'));
 588
 589        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
 590        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
 591        return if -f $index;
 592
 593        chomp(my $bare = `git config --bool --get core.bare`);
 594        return if $bare eq 'true';
 595        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
 596        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
 597        print STDERR "Checked out HEAD:\n  ",
 598                     $gs->full_url, " r", $gs->last_rev, "\n";
 599}
 600
 601sub complete_svn_url {
 602        my ($url, $path) = @_;
 603        $path =~ s#/+$##;
 604        if ($path !~ m#^[a-z\+]+://#) {
 605                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 606                        fatal("E: '$path' is not a complete URL ",
 607                              "and a separate URL is not specified\n");
 608                }
 609                return ($url, $path);
 610        }
 611        return ($path, '');
 612}
 613
 614sub complete_url_ls_init {
 615        my ($ra, $repo_path, $switch, $pfx) = @_;
 616        unless ($repo_path) {
 617                print STDERR "W: $switch not specified\n";
 618                return;
 619        }
 620        $repo_path =~ s#/+$##;
 621        if ($repo_path =~ m#^[a-z\+]+://#) {
 622                $ra = Git::SVN::Ra->new($repo_path);
 623                $repo_path = '';
 624        } else {
 625                $repo_path =~ s#^/+##;
 626                unless ($ra) {
 627                        fatal("E: '$repo_path' is not a complete URL ",
 628                              "and a separate URL is not specified\n");
 629                }
 630        }
 631        my $url = $ra->{url};
 632        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
 633        my $k = "svn-remote.$gs->{repo_id}.url";
 634        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
 635        if ($orig_url && ($orig_url ne $gs->{url})) {
 636                die "$k already set: $orig_url\n",
 637                    "wanted to set to: $gs->{url}\n";
 638        }
 639        command_oneline('config', $k, $gs->{url}) unless $orig_url;
 640        my $remote_path = "$ra->{svn_path}/$repo_path/*";
 641        $remote_path =~ s#/+#/#g;
 642        $remote_path =~ s#^/##g;
 643        my ($n) = ($switch =~ /^--(\w+)/);
 644        if (length $pfx && $pfx !~ m#/$#) {
 645                die "--prefix='$pfx' must have a trailing slash '/'\n";
 646        }
 647        command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
 648                                "$remote_path:refs/remotes/$pfx*");
 649}
 650
 651sub verify_ref {
 652        my ($ref) = @_;
 653        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 654                               { STDERR => 0 }); };
 655}
 656
 657sub get_tree_from_treeish {
 658        my ($treeish) = @_;
 659        # $treeish can be a symbolic ref, too:
 660        my $type = command_oneline(qw/cat-file -t/, $treeish);
 661        my $expected;
 662        while ($type eq 'tag') {
 663                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
 664        }
 665        if ($type eq 'commit') {
 666                $expected = (grep /^tree /, command(qw/cat-file commit/,
 667                                                    $treeish))[0];
 668                ($expected) = ($expected =~ /^tree ($sha1)$/o);
 669                die "Unable to get tree from $treeish\n" unless $expected;
 670        } elsif ($type eq 'tree') {
 671                $expected = $treeish;
 672        } else {
 673                die "$treeish is a $type, expected tree, tag or commit\n";
 674        }
 675        return $expected;
 676}
 677
 678sub get_commit_entry {
 679        my ($treeish) = shift;
 680        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
 681        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
 682        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
 683        open my $log_fh, '>', $commit_editmsg or croak $!;
 684
 685        my $type = command_oneline(qw/cat-file -t/, $treeish);
 686        if ($type eq 'commit' || $type eq 'tag') {
 687                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
 688                                                         $type, $treeish);
 689                my $in_msg = 0;
 690                while (<$msg_fh>) {
 691                        if (!$in_msg) {
 692                                $in_msg = 1 if (/^\s*$/);
 693                        } elsif (/^git-svn-id: /) {
 694                                # skip this for now, we regenerate the
 695                                # correct one on re-fetch anyways
 696                                # TODO: set *:merge properties or like...
 697                        } else {
 698                                print $log_fh $_ or croak $!;
 699                        }
 700                }
 701                command_close_pipe($msg_fh, $ctx);
 702        }
 703        close $log_fh or croak $!;
 704
 705        if ($_edit || ($type eq 'tree')) {
 706                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 707                # TODO: strip out spaces, comments, like git-commit.sh
 708                system($editor, $commit_editmsg);
 709        }
 710        rename $commit_editmsg, $commit_msg or croak $!;
 711        open $log_fh, '<', $commit_msg or croak $!;
 712        { local $/; chomp($log_entry{log} = <$log_fh>); }
 713        close $log_fh or croak $!;
 714        unlink $commit_msg;
 715        \%log_entry;
 716}
 717
 718sub s_to_file {
 719        my ($str, $file, $mode) = @_;
 720        open my $fd,'>',$file or croak $!;
 721        print $fd $str,"\n" or croak $!;
 722        close $fd or croak $!;
 723        chmod ($mode &~ umask, $file) if (defined $mode);
 724}
 725
 726sub file_to_s {
 727        my $file = shift;
 728        open my $fd,'<',$file or croak "$!: file: $file\n";
 729        local $/;
 730        my $ret = <$fd>;
 731        close $fd or croak $!;
 732        $ret =~ s/\s*$//s;
 733        return $ret;
 734}
 735
 736# '<svn username> = real-name <email address>' mapping based on git-svnimport:
 737sub load_authors {
 738        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
 739        my $log = $cmd eq 'log';
 740        while (<$authors>) {
 741                chomp;
 742                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
 743                my ($user, $name, $email) = ($1, $2, $3);
 744                if ($log) {
 745                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
 746                } else {
 747                        $users{$user} = [$name, $email];
 748                }
 749        }
 750        close $authors or croak $!;
 751}
 752
 753# convert GetOpt::Long specs for use by git-config
 754sub read_repo_config {
 755        return unless -d $ENV{GIT_DIR};
 756        my $opts = shift;
 757        my @config_only;
 758        foreach my $o (keys %$opts) {
 759                # if we have mixedCase and a long option-only, then
 760                # it's a config-only variable that we don't need for
 761                # the command-line.
 762                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
 763                my $v = $opts->{$o};
 764                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
 765                $key =~ s/-//g;
 766                my $arg = 'git-config';
 767                $arg .= ' --int' if ($o =~ /[:=]i$/);
 768                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
 769                if (ref $v eq 'ARRAY') {
 770                        chomp(my @tmp = `$arg --get-all svn.$key`);
 771                        @$v = @tmp if @tmp;
 772                } else {
 773                        chomp(my $tmp = `$arg --get svn.$key`);
 774                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
 775                                $$v = $tmp;
 776                        }
 777                }
 778        }
 779        delete @$opts{@config_only} if @config_only;
 780}
 781
 782sub extract_metadata {
 783        my $id = shift or return (undef, undef, undef);
 784        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 785                                                        \s([a-f\d\-]+)$/x);
 786        if (!defined $rev || !$uuid || !$url) {
 787                # some of the original repositories I made had
 788                # identifiers like this:
 789                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
 790        }
 791        return ($url, $rev, $uuid);
 792}
 793
 794sub cmt_metadata {
 795        return extract_metadata((grep(/^git-svn-id: /,
 796                command(qw/cat-file commit/, shift)))[-1]);
 797}
 798
 799sub working_head_info {
 800        my ($head, $refs) = @_;
 801        my ($fh, $ctx) = command_output_pipe('rev-list', $head);
 802        while (my $hash = <$fh>) {
 803                chomp($hash);
 804                my ($url, $rev, $uuid) = cmt_metadata($hash);
 805                if (defined $url && defined $rev) {
 806                        if (my $gs = Git::SVN->find_by_url($url)) {
 807                                my $c = $gs->rev_db_get($rev);
 808                                if ($c && $c eq $hash) {
 809                                        close $fh; # break the pipe
 810                                        return ($url, $rev, $uuid, $gs);
 811                                }
 812                        }
 813                }
 814                unshift @$refs, $hash if $refs;
 815        }
 816        command_close_pipe($fh, $ctx);
 817        (undef, undef, undef, undef);
 818}
 819
 820sub read_commit_parents {
 821        my ($parents, $c) = @_;
 822        my ($fh, $ctx) = command_output_pipe(qw/cat-file commit/, $c);
 823        while (<$fh>) {
 824                chomp;
 825                last if '';
 826                /^parent ($sha1)/ or next;
 827                push @{$parents->{$c}}, $1;
 828        }
 829        close $fh; # break the pipe
 830}
 831
 832sub linearize_history {
 833        my ($gs, $refs) = @_;
 834        my %parents;
 835        foreach my $c (@$refs) {
 836                read_commit_parents(\%parents, $c);
 837        }
 838
 839        my @linear_refs;
 840        my %skip = ();
 841        my $last_svn_commit = $gs->last_commit;
 842        foreach my $c (reverse @$refs) {
 843                next if $c eq $last_svn_commit;
 844                last if $skip{$c};
 845
 846                unshift @linear_refs, $c;
 847                $skip{$c} = 1;
 848
 849                # we only want the first parent to diff against for linear
 850                # history, we save the rest to inject when we finalize the
 851                # svn commit
 852                my $fp_a = verify_ref("$c~1");
 853                my $fp_b = shift @{$parents{$c}} if $parents{$c};
 854                if (!$fp_a || !$fp_b) {
 855                        die "Commit $c\n",
 856                            "has no parent commit, and therefore ",
 857                            "nothing to diff against.\n",
 858                            "You should be working from a repository ",
 859                            "originally created by git-svn\n";
 860                }
 861                if ($fp_a ne $fp_b) {
 862                        die "$c~1 = $fp_a, however parsing commit $c ",
 863                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
 864                }
 865
 866                foreach my $p (@{$parents{$c}}) {
 867                        $skip{$p} = 1;
 868                }
 869        }
 870        (\@linear_refs, \%parents);
 871}
 872
 873package Git::SVN;
 874use strict;
 875use warnings;
 876use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
 877            $_repack $_repack_flags $_use_svm_props $_head
 878            $_use_svnsync_props $no_reuse_existing $_minimize_url/;
 879use Carp qw/croak/;
 880use File::Path qw/mkpath/;
 881use File::Copy qw/copy/;
 882use IPC::Open3;
 883
 884my $_repack_nr;
 885# properties that we do not log:
 886my %SKIP_PROP;
 887BEGIN {
 888        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
 889                                        svn:special svn:executable
 890                                        svn:entry:committed-rev
 891                                        svn:entry:last-author
 892                                        svn:entry:uuid
 893                                        svn:entry:committed-date/;
 894
 895        # some options are read globally, but can be overridden locally
 896        # per [svn-remote "..."] section.  Command-line options will *NOT*
 897        # override options set in an [svn-remote "..."] section
 898        my $e;
 899        foreach (qw/follow_parent no_metadata use_svm_props
 900                    use_svnsync_props/) {
 901                my $key = $_;
 902                $key =~ tr/_//d;
 903                $e .= "sub $_ {
 904                        my (\$self) = \@_;
 905                        return \$self->{-$_} if exists \$self->{-$_};
 906                        my \$k = \"svn-remote.\$self->{repo_id}\.$key\";
 907                        eval { command_oneline(qw/config --get/, \$k) };
 908                        if (\$@) {
 909                                \$self->{-$_} = \$Git::SVN::_$_;
 910                        } else {
 911                                my \$v = command_oneline(qw/config --bool/,\$k);
 912                                \$self->{-$_} = \$v eq 'false' ? 0 : 1;
 913                        }
 914                        return \$self->{-$_} }\n";
 915        }
 916        $e .= "1;\n";
 917        eval $e or die $@;
 918}
 919
 920my %LOCKFILES;
 921END { unlink keys %LOCKFILES if %LOCKFILES }
 922
 923sub resolve_local_globs {
 924        my ($url, $fetch, $glob_spec) = @_;
 925        return unless defined $glob_spec;
 926        my $ref = $glob_spec->{ref};
 927        my $path = $glob_spec->{path};
 928        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
 929                next unless m#^refs/remotes/$ref->{regex}$#;
 930                my $p = $1;
 931                my $pathname = $path->full_path($p);
 932                my $refname = $ref->full_path($p);
 933                if (my $existing = $fetch->{$pathname}) {
 934                        if ($existing ne $refname) {
 935                                die "Refspec conflict:\n",
 936                                    "existing: refs/remotes/$existing\n",
 937                                    " globbed: refs/remotes/$refname\n";
 938                        }
 939                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
 940                        $u =~ s!^\Q$url\E(/|$)!! or die
 941                          "refs/remotes/$refname: '$url' not found in '$u'\n";
 942                        if ($pathname ne $u) {
 943                                warn "W: Refspec glob conflict ",
 944                                     "(ref: refs/remotes/$refname):\n",
 945                                     "expected path: $pathname\n",
 946                                     "    real path: $u\n",
 947                                     "Continuing ahead with $u\n";
 948                                next;
 949                        }
 950                } else {
 951                        $fetch->{$pathname} = $refname;
 952                }
 953        }
 954}
 955
 956sub parse_revision_argument {
 957        my ($base, $head) = @_;
 958        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
 959                return ($base, $head);
 960        }
 961        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
 962        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
 963        return ($head, $head) if ($::_revision eq 'HEAD');
 964        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
 965        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
 966        die "revision argument: $::_revision not understood by git-svn\n";
 967}
 968
 969sub fetch_all {
 970        my ($repo_id, $remotes) = @_;
 971        if (ref $repo_id) {
 972                my $gs = $repo_id;
 973                $repo_id = undef;
 974                $repo_id = $gs->{repo_id};
 975        }
 976        $remotes ||= read_all_remotes();
 977        my $remote = $remotes->{$repo_id} or
 978                     die "[svn-remote \"$repo_id\"] unknown\n";
 979        my $fetch = $remote->{fetch};
 980        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
 981        my (@gs, @globs);
 982        my $ra = Git::SVN::Ra->new($url);
 983        my $uuid = $ra->get_uuid;
 984        my $head = $ra->get_latest_revnum;
 985        my $base = defined $fetch ? $head : 0;
 986
 987        # read the max revs for wildcard expansion (branches/*, tags/*)
 988        foreach my $t (qw/branches tags/) {
 989                defined $remote->{$t} or next;
 990                push @globs, $remote->{$t};
 991                my $max_rev = eval { tmp_config(qw/--int --get/,
 992                                         "svn-remote.$repo_id.${t}-maxRev") };
 993                if (defined $max_rev && ($max_rev < $base)) {
 994                        $base = $max_rev;
 995                } elsif (!defined $max_rev) {
 996                        $base = 0;
 997                }
 998        }
 999
1000        if ($fetch) {
1001                foreach my $p (sort keys %$fetch) {
1002                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1003                        my $lr = $gs->rev_db_max;
1004                        if (defined $lr) {
1005                                $base = $lr if ($lr < $base);
1006                        }
1007                        push @gs, $gs;
1008                }
1009        }
1010
1011        ($base, $head) = parse_revision_argument($base, $head);
1012        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1013}
1014
1015sub read_all_remotes {
1016        my $r = {};
1017        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1018                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
1019                        $r->{$1}->{fetch}->{$2} = $3;
1020                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1021                        $r->{$1}->{url} = $2;
1022                } elsif (m!^(.+)\.(branches|tags)=
1023                           (.*):refs/remotes/(.+)\s*$/!x) {
1024                        my ($p, $g) = ($3, $4);
1025                        my $rs = $r->{$1}->{$2} = {
1026                                          t => $2,
1027                                          remote => $1,
1028                                          path => Git::SVN::GlobSpec->new($p),
1029                                          ref => Git::SVN::GlobSpec->new($g) };
1030                        if (length($rs->{ref}->{right}) != 0) {
1031                                die "The '*' glob character must be the last ",
1032                                    "character of '$g'\n";
1033                        }
1034                }
1035        }
1036        $r;
1037}
1038
1039sub init_vars {
1040        if (defined $_repack) {
1041                $_repack = 1000 if ($_repack <= 0);
1042                $_repack_nr = $_repack;
1043                $_repack_flags ||= '-d';
1044        }
1045}
1046
1047sub verify_remotes_sanity {
1048        return unless -d $ENV{GIT_DIR};
1049        my %seen;
1050        foreach (command(qw/config -l/)) {
1051                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1052                        if ($seen{$1}) {
1053                                die "Remote ref refs/remote/$1 is tracked by",
1054                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1055                                    "Please resolve this ambiguity in ",
1056                                    "your git configuration file before ",
1057                                    "continuing\n";
1058                        }
1059                        $seen{$1} = $_;
1060                }
1061        }
1062}
1063
1064# we allow more chars than remotes2config.sh...
1065sub sanitize_remote_name {
1066        my ($name) = @_;
1067        $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1068        $name;
1069}
1070
1071sub find_existing_remote {
1072        my ($url, $remotes) = @_;
1073        return undef if $no_reuse_existing;
1074        my $existing;
1075        foreach my $repo_id (keys %$remotes) {
1076                my $u = $remotes->{$repo_id}->{url} or next;
1077                next if $u ne $url;
1078                $existing = $repo_id;
1079                last;
1080        }
1081        $existing;
1082}
1083
1084sub init_remote_config {
1085        my ($self, $url, $no_write) = @_;
1086        $url =~ s!/+$!!; # strip trailing slash
1087        my $r = read_all_remotes();
1088        my $existing = find_existing_remote($url, $r);
1089        if ($existing) {
1090                unless ($no_write) {
1091                        print STDERR "Using existing ",
1092                                     "[svn-remote \"$existing\"]\n";
1093                }
1094                $self->{repo_id} = $existing;
1095        } elsif ($_minimize_url) {
1096                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1097                $existing = find_existing_remote($min_url, $r);
1098                if ($existing) {
1099                        unless ($no_write) {
1100                                print STDERR "Using existing ",
1101                                             "[svn-remote \"$existing\"]\n";
1102                        }
1103                        $self->{repo_id} = $existing;
1104                }
1105                if ($min_url ne $url) {
1106                        unless ($no_write) {
1107                                print STDERR "Using higher level of URL: ",
1108                                             "$url => $min_url\n";
1109                        }
1110                        my $old_path = $self->{path};
1111                        $self->{path} = $url;
1112                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1113                        if (length $old_path) {
1114                                $self->{path} .= "/$old_path";
1115                        }
1116                        $url = $min_url;
1117                }
1118        }
1119        my $orig_url;
1120        if (!$existing) {
1121                # verify that we aren't overwriting anything:
1122                $orig_url = eval {
1123                        command_oneline('config', '--get',
1124                                        "svn-remote.$self->{repo_id}.url")
1125                };
1126                if ($orig_url && ($orig_url ne $url)) {
1127                        die "svn-remote.$self->{repo_id}.url already set: ",
1128                            "$orig_url\nwanted to set to: $url\n";
1129                }
1130        }
1131        my ($xrepo_id, $xpath) = find_ref($self->refname);
1132        if (defined $xpath) {
1133                die "svn-remote.$xrepo_id.fetch already set to track ",
1134                    "$xpath:refs/remotes/", $self->refname, "\n";
1135        }
1136        unless ($no_write) {
1137                command_noisy('config',
1138                              "svn-remote.$self->{repo_id}.url", $url);
1139                command_noisy('config', '--add',
1140                              "svn-remote.$self->{repo_id}.fetch",
1141                              "$self->{path}:".$self->refname);
1142        }
1143        $self->{url} = $url;
1144}
1145
1146sub find_by_url { # repos_root and, path are optional
1147        my ($class, $full_url, $repos_root, $path) = @_;
1148
1149        return undef unless defined $full_url;
1150        remove_username($full_url);
1151        remove_username($repos_root) if defined $repos_root;
1152        my $remotes = read_all_remotes();
1153        if (defined $full_url && defined $repos_root && !defined $path) {
1154                $path = $full_url;
1155                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1156        }
1157        foreach my $repo_id (keys %$remotes) {
1158                my $u = $remotes->{$repo_id}->{url} or next;
1159                remove_username($u);
1160                next if defined $repos_root && $repos_root ne $u;
1161
1162                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1163                foreach (qw/branches tags/) {
1164                        resolve_local_globs($u, $fetch,
1165                                            $remotes->{$repo_id}->{$_});
1166                }
1167                my $p = $path;
1168                unless (defined $p) {
1169                        $p = $full_url;
1170                        $p =~ s#^\Q$u\E(?:/|$)## or next;
1171                }
1172                foreach my $f (keys %$fetch) {
1173                        next if $f ne $p;
1174                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1175                }
1176        }
1177        undef;
1178}
1179
1180sub init {
1181        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1182        my $self = _new($class, $repo_id, $ref_id, $path);
1183        if (defined $url) {
1184                $self->init_remote_config($url, $no_write);
1185        }
1186        $self;
1187}
1188
1189sub find_ref {
1190        my ($ref_id) = @_;
1191        foreach (command(qw/config -l/)) {
1192                next unless m!^svn-remote\.(.+)\.fetch=
1193                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1194                my ($repo_id, $path, $ref) = ($1, $2, $3);
1195                if ($ref eq $ref_id) {
1196                        $path = '' if ($path =~ m#^\./?#);
1197                        return ($repo_id, $path);
1198                }
1199        }
1200        (undef, undef, undef);
1201}
1202
1203sub new {
1204        my ($class, $ref_id, $repo_id, $path) = @_;
1205        if (defined $ref_id && !defined $repo_id && !defined $path) {
1206                ($repo_id, $path) = find_ref($ref_id);
1207                if (!defined $repo_id) {
1208                        die "Could not find a \"svn-remote.*.fetch\" key ",
1209                            "in the repository configuration matching: ",
1210                            "refs/remotes/$ref_id\n";
1211                }
1212        }
1213        my $self = _new($class, $repo_id, $ref_id, $path);
1214        if (!defined $self->{path} || !length $self->{path}) {
1215                my $fetch = command_oneline('config', '--get',
1216                                            "svn-remote.$repo_id.fetch",
1217                                            ":refs/remotes/$ref_id\$") or
1218                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1219                         "\":refs/remotes/$ref_id\$\" in config\n";
1220                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1221        }
1222        $self->{url} = command_oneline('config', '--get',
1223                                       "svn-remote.$repo_id.url") or
1224                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1225        $self->rebuild;
1226        $self;
1227}
1228
1229sub refname { "refs/remotes/$_[0]->{ref_id}" }
1230
1231sub svm_uuid {
1232        my ($self) = @_;
1233        return $self->{svm}->{uuid} if $self->svm;
1234        $self->ra;
1235        unless ($self->{svm}) {
1236                die "SVM UUID not cached, and reading remotely failed\n";
1237        }
1238        $self->{svm}->{uuid};
1239}
1240
1241sub svm {
1242        my ($self) = @_;
1243        return $self->{svm} if $self->{svm};
1244        my $svm;
1245        # see if we have it in our config, first:
1246        eval {
1247                my $section = "svn-remote.$self->{repo_id}";
1248                $svm = {
1249                  source => tmp_config('--get', "$section.svm-source"),
1250                  uuid => tmp_config('--get', "$section.svm-uuid"),
1251                  replace => tmp_config('--get', "$section.svm-replace"),
1252                }
1253        };
1254        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1255                $self->{svm} = $svm;
1256        }
1257        $self->{svm};
1258}
1259
1260sub _set_svm_vars {
1261        my ($self, $ra) = @_;
1262        return $ra if $self->svm;
1263
1264        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1265                    "(svm:source, svm:uuid) ",
1266                    "from the following URLs:\n" );
1267        sub read_svm_props {
1268                my ($self, $ra, $path, $r) = @_;
1269                my $props = ($ra->get_dir($path, $r))[2];
1270                my $src = $props->{'svm:source'};
1271                my $uuid = $props->{'svm:uuid'};
1272                return undef if (!$src || !$uuid);
1273
1274                chomp($src, $uuid);
1275
1276                $uuid =~ m{^[0-9a-f\-]{30,}$}
1277                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1278
1279                # the '!' is used to mark the repos_root!/relative/path
1280                $src =~ s{/?!/?}{/};
1281                $src =~ s{/+$}{}; # no trailing slashes please
1282                # username is of no interest
1283                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1284
1285                my $replace = $ra->{url};
1286                $replace .= "/$path" if length $path;
1287
1288                my $section = "svn-remote.$self->{repo_id}";
1289                tmp_config("$section.svm-source", $src);
1290                tmp_config("$section.svm-replace", $replace);
1291                tmp_config("$section.svm-uuid", $uuid);
1292                $self->{svm} = {
1293                        source => $src,
1294                        uuid => $uuid,
1295                        replace => $replace
1296                };
1297        }
1298
1299        my $r = $ra->get_latest_revnum;
1300        my $path = $self->{path};
1301        my %tried;
1302        while (length $path) {
1303                unless ($tried{"$self->{url}/$path"}) {
1304                        return $ra if $self->read_svm_props($ra, $path, $r);
1305                        $tried{"$self->{url}/$path"} = 1;
1306                }
1307                $path =~ s#/?[^/]+$##;
1308        }
1309        die "Path: '$path' should be ''\n" if $path ne '';
1310        return $ra if $self->read_svm_props($ra, $path, $r);
1311        $tried{"$self->{url}/$path"} = 1;
1312
1313        if ($ra->{repos_root} eq $self->{url}) {
1314                die @err, (map { "  $_\n" } keys %tried), "\n";
1315        }
1316
1317        # nope, make sure we're connected to the repository root:
1318        my $ok;
1319        my @tried_b;
1320        $path = $ra->{svn_path};
1321        $ra = Git::SVN::Ra->new($ra->{repos_root});
1322        while (length $path) {
1323                unless ($tried{"$ra->{url}/$path"}) {
1324                        $ok = $self->read_svm_props($ra, $path, $r);
1325                        last if $ok;
1326                        $tried{"$ra->{url}/$path"} = 1;
1327                }
1328                $path =~ s#/?[^/]+$##;
1329        }
1330        die "Path: '$path' should be ''\n" if $path ne '';
1331        $ok ||= $self->read_svm_props($ra, $path, $r);
1332        $tried{"$ra->{url}/$path"} = 1;
1333        if (!$ok) {
1334                die @err, (map { "  $_\n" } keys %tried), "\n";
1335        }
1336        Git::SVN::Ra->new($self->{url});
1337}
1338
1339sub svnsync {
1340        my ($self) = @_;
1341        return $self->{svnsync} if $self->{svnsync};
1342
1343        if ($self->no_metadata) {
1344                die "Can't have both 'noMetadata' and ",
1345                    "'useSvnsyncProps' options set!\n";
1346        }
1347        if ($self->rewrite_root) {
1348                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1349                    "options set!\n";
1350        }
1351
1352        my $svnsync;
1353        # see if we have it in our config, first:
1354        eval {
1355                my $section = "svn-remote.$self->{repo_id}";
1356                $svnsync = {
1357                  url => tmp_config('--get', "$section.svnsync-url"),
1358                  uuid => tmp_config('--get', "$section.svnsync-uuid"),
1359                }
1360        };
1361        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1362                return $self->{svnsync} = $svnsync;
1363        }
1364
1365        my $err = "useSvnsyncProps set, but failed to read " .
1366                  "svnsync property: svn:sync-from-";
1367        my $rp = $self->ra->rev_proplist(0);
1368
1369        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1370        $url =~ m{^[a-z\+]+://} or
1371                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1372
1373        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1374        $uuid =~ m{^[0-9a-f\-]{30,}$} or
1375                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1376
1377        my $section = "svn-remote.$self->{repo_id}";
1378        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1379        tmp_config('--add', "$section.svnsync-url", $url);
1380        return $self->{svnsync} = { url => $url, uuid => $uuid };
1381}
1382
1383# this allows us to memoize our SVN::Ra UUID locally and avoid a
1384# remote lookup (useful for 'git svn log').
1385sub ra_uuid {
1386        my ($self) = @_;
1387        unless ($self->{ra_uuid}) {
1388                my $key = "svn-remote.$self->{repo_id}.uuid";
1389                my $uuid = eval { tmp_config('--get', $key) };
1390                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1391                        $self->{ra_uuid} = $uuid;
1392                } else {
1393                        die "ra_uuid called without URL\n" unless $self->{url};
1394                        $self->{ra_uuid} = $self->ra->get_uuid;
1395                        tmp_config('--add', $key, $self->{ra_uuid});
1396                }
1397        }
1398        $self->{ra_uuid};
1399}
1400
1401sub ra {
1402        my ($self) = shift;
1403        my $ra = Git::SVN::Ra->new($self->{url});
1404        if ($self->use_svm_props && !$self->{svm}) {
1405                if ($self->no_metadata) {
1406                        die "Can't have both 'noMetadata' and ",
1407                            "'useSvmProps' options set!\n";
1408                } elsif ($self->use_svnsync_props) {
1409                        die "Can't have both 'useSvnsyncProps' and ",
1410                            "'useSvmProps' options set!\n";
1411                }
1412                $ra = $self->_set_svm_vars($ra);
1413                $self->{-want_revprops} = 1;
1414        }
1415        $ra;
1416}
1417
1418sub rel_path {
1419        my ($self) = @_;
1420        my $repos_root = $self->ra->{repos_root};
1421        return $self->{path} if ($self->{url} eq $repos_root);
1422        my $url = $self->{url} .
1423                  (length $self->{path} ? "/$self->{path}" : $self->{path});
1424        $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1425        $url;
1426}
1427
1428sub traverse_ignore {
1429        my ($self, $fh, $path, $r) = @_;
1430        $path =~ s#^/+##g;
1431        my $ra = $self->ra;
1432        my ($dirent, undef, $props) = $ra->get_dir($path, $r);
1433        my $p = $path;
1434        $p =~ s#^\Q$self->{path}\E(/|$)##;
1435        print $fh length $p ? "\n# $p\n" : "\n# /\n";
1436        if (my $s = $props->{'svn:ignore'}) {
1437                $s =~ s/[\r\n]+/\n/g;
1438                chomp $s;
1439                if (length $p == 0) {
1440                        $s =~ s#\n#\n/$p#g;
1441                        print $fh "/$s\n";
1442                } else {
1443                        $s =~ s#\n#\n/$p/#g;
1444                        print $fh "/$p/$s\n";
1445                }
1446        }
1447        foreach (sort keys %$dirent) {
1448                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1449                $self->traverse_ignore($fh, "$path/$_", $r);
1450        }
1451}
1452
1453sub last_rev { ($_[0]->last_rev_commit)[0] }
1454sub last_commit { ($_[0]->last_rev_commit)[1] }
1455
1456# returns the newest SVN revision number and newest commit SHA1
1457sub last_rev_commit {
1458        my ($self) = @_;
1459        if (defined $self->{last_rev} && defined $self->{last_commit}) {
1460                return ($self->{last_rev}, $self->{last_commit});
1461        }
1462        my $c = ::verify_ref($self->refname.'^0');
1463        if ($c && !$self->use_svm_props && !$self->no_metadata) {
1464                my $rev = (::cmt_metadata($c))[1];
1465                if (defined $rev) {
1466                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1467                        return ($rev, $c);
1468                }
1469        }
1470        my $db_path = $self->db_path;
1471        unless (-e $db_path) {
1472                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1473                return (undef, undef);
1474        }
1475        my $offset = -41; # from tail
1476        my $rl;
1477        open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1478        sysseek($fh, $offset, 2); # don't care for errors
1479        sysread($fh, $rl, 41) == 41 or return (undef, undef);
1480        chomp $rl;
1481        while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1482                $offset -= 41;
1483                sysseek($fh, $offset, 2); # don't care for errors
1484                sysread($fh, $rl, 41) == 41 or return (undef, undef);
1485                chomp $rl;
1486        }
1487        if ($c && $c ne $rl) {
1488                die "$db_path and ", $self->refname,
1489                    " inconsistent!:\n$c != $rl\n";
1490        }
1491        my $rev = sysseek($fh, 0, 1) or croak $!;
1492        $rev =  ($rev - 41) / 41;
1493        close $fh or croak $!;
1494        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1495        return ($rev, $c);
1496}
1497
1498sub get_fetch_range {
1499        my ($self, $min, $max) = @_;
1500        $max ||= $self->ra->get_latest_revnum;
1501        $min ||= $self->rev_db_max;
1502        (++$min, $max);
1503}
1504
1505sub tmp_config {
1506        my (@args) = @_;
1507        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1508        my $config = "$ENV{GIT_DIR}/svn/.metadata";
1509        if (! -f $config && -f $old_def_config) {
1510                rename $old_def_config, $config or
1511                       die "Failed rename $old_def_config => $config: $!\n";
1512        }
1513        my $old_config = $ENV{GIT_CONFIG};
1514        $ENV{GIT_CONFIG} = $config;
1515        $@ = undef;
1516        my @ret = eval {
1517                unless (-f $config) {
1518                        mkfile($config);
1519                        open my $fh, '>', $config or
1520                            die "Can't open $config: $!\n";
1521                        print $fh "; This file is used internally by ",
1522                                  "git-svn\n" or die
1523                                  "Couldn't write to $config: $!\n";
1524                        print $fh "; You should not have to edit it\n" or
1525                              die "Couldn't write to $config: $!\n";
1526                        close $fh or die "Couldn't close $config: $!\n";
1527                }
1528                command('config', @args);
1529        };
1530        my $err = $@;
1531        if (defined $old_config) {
1532                $ENV{GIT_CONFIG} = $old_config;
1533        } else {
1534                delete $ENV{GIT_CONFIG};
1535        }
1536        die $err if $err;
1537        wantarray ? @ret : $ret[0];
1538}
1539
1540sub tmp_index_do {
1541        my ($self, $sub) = @_;
1542        my $old_index = $ENV{GIT_INDEX_FILE};
1543        $ENV{GIT_INDEX_FILE} = $self->{index};
1544        $@ = undef;
1545        my @ret = eval {
1546                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1547                mkpath([$dir]) unless -d $dir;
1548                &$sub;
1549        };
1550        my $err = $@;
1551        if (defined $old_index) {
1552                $ENV{GIT_INDEX_FILE} = $old_index;
1553        } else {
1554                delete $ENV{GIT_INDEX_FILE};
1555        }
1556        die $err if $err;
1557        wantarray ? @ret : $ret[0];
1558}
1559
1560sub assert_index_clean {
1561        my ($self, $treeish) = @_;
1562
1563        $self->tmp_index_do(sub {
1564                command_noisy('read-tree', $treeish) unless -e $self->{index};
1565                my $x = command_oneline('write-tree');
1566                my ($y) = (command(qw/cat-file commit/, $treeish) =~
1567                           /^tree ($::sha1)/mo);
1568                return if $y eq $x;
1569
1570                warn "Index mismatch: $y != $x\nrereading $treeish\n";
1571                unlink $self->{index} or die "unlink $self->{index}: $!\n";
1572                command_noisy('read-tree', $treeish);
1573                $x = command_oneline('write-tree');
1574                if ($y ne $x) {
1575                        ::fatal "trees ($treeish) $y != $x\n",
1576                                "Something is seriously wrong...\n";
1577                }
1578        });
1579}
1580
1581sub get_commit_parents {
1582        my ($self, $log_entry) = @_;
1583        my (%seen, @ret, @tmp);
1584        # legacy support for 'set-tree'; this is only used by set_tree_cb:
1585        if (my $ip = $self->{inject_parents}) {
1586                if (my $commit = delete $ip->{$log_entry->{revision}}) {
1587                        push @tmp, $commit;
1588                }
1589        }
1590        if (my $cur = ::verify_ref($self->refname.'^0')) {
1591                push @tmp, $cur;
1592        }
1593        if (my $ipd = $self->{inject_parents_dcommit}) {
1594                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
1595                        push @tmp, @$commit;
1596                }
1597        }
1598        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1599        while (my $p = shift @tmp) {
1600                next if $seen{$p};
1601                $seen{$p} = 1;
1602                push @ret, $p;
1603                # MAXPARENT is defined to 16 in commit-tree.c:
1604                last if @ret >= 16;
1605        }
1606        if (@tmp) {
1607                die "r$log_entry->{revision}: No room for parents:\n\t",
1608                    join("\n\t", @tmp), "\n";
1609        }
1610        @ret;
1611}
1612
1613sub rewrite_root {
1614        my ($self) = @_;
1615        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
1616        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
1617        my $rwr = eval { command_oneline(qw/config --get/, $k) };
1618        if ($rwr) {
1619                $rwr =~ s#/+$##;
1620                if ($rwr !~ m#^[a-z\+]+://#) {
1621                        die "$rwr is not a valid URL (key: $k)\n";
1622                }
1623        }
1624        $self->{-rewrite_root} = $rwr;
1625}
1626
1627sub metadata_url {
1628        my ($self) = @_;
1629        ($self->rewrite_root || $self->{url}) .
1630           (length $self->{path} ? '/' . $self->{path} : '');
1631}
1632
1633sub full_url {
1634        my ($self) = @_;
1635        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1636}
1637
1638sub do_git_commit {
1639        my ($self, $log_entry) = @_;
1640        my $lr = $self->last_rev;
1641        if (defined $lr && $lr >= $log_entry->{revision}) {
1642                die "Last fetched revision of ", $self->refname,
1643                    " was r$lr, but we are about to fetch: ",
1644                    "r$log_entry->{revision}!\n";
1645        }
1646        if (my $c = $self->rev_db_get($log_entry->{revision})) {
1647                croak "$log_entry->{revision} = $c already exists! ",
1648                      "Why are we refetching it?\n";
1649        }
1650        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1651        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1652                                                          $log_entry->{email};
1653        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1654
1655        my $tree = $log_entry->{tree};
1656        if (!defined $tree) {
1657                $tree = $self->tmp_index_do(sub {
1658                                            command_oneline('write-tree') });
1659        }
1660        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1661
1662        my @exec = ('git-commit-tree', $tree);
1663        foreach ($self->get_commit_parents($log_entry)) {
1664                push @exec, '-p', $_;
1665        }
1666        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1667                                                                   or croak $!;
1668        print $msg_fh $log_entry->{log} or croak $!;
1669        unless ($self->no_metadata) {
1670                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1671                              or croak $!;
1672        }
1673        $msg_fh->flush == 0 or croak $!;
1674        close $msg_fh or croak $!;
1675        chomp(my $commit = do { local $/; <$out_fh> });
1676        close $out_fh or croak $!;
1677        waitpid $pid, 0;
1678        croak $? if $?;
1679        if ($commit !~ /^$::sha1$/o) {
1680                die "Failed to commit, invalid sha1: $commit\n";
1681        }
1682
1683        $self->rev_db_set($log_entry->{revision}, $commit, 1);
1684
1685        $self->{last_rev} = $log_entry->{revision};
1686        $self->{last_commit} = $commit;
1687        print "r$log_entry->{revision}";
1688        if (defined $log_entry->{svm_revision}) {
1689                 print " (\@$log_entry->{svm_revision})";
1690                 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1691                                   0, $self->svm_uuid);
1692        }
1693        print " = $commit ($self->{ref_id})\n";
1694        if (defined $_repack && (--$_repack_nr == 0)) {
1695                $_repack_nr = $_repack;
1696                # repack doesn't use any arguments with spaces in them, does it?
1697                print "Running git repack $_repack_flags ...\n";
1698                command_noisy('repack', split(/\s+/, $_repack_flags));
1699                print "Done repacking\n";
1700        }
1701        return $commit;
1702}
1703
1704sub match_paths {
1705        my ($self, $paths, $r) = @_;
1706        return 1 if $self->{path} eq '';
1707        if (my $path = $paths->{"/$self->{path}"}) {
1708                return ($path->{action} eq 'D') ? 0 : 1;
1709        }
1710        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1711        if (grep /$self->{path_regex}/, keys %$paths) {
1712                return 1;
1713        }
1714        my $c = '';
1715        foreach (split m#/#, $self->{path}) {
1716                $c .= "/$_";
1717                next unless ($paths->{$c} &&
1718                             ($paths->{$c}->{action} =~ /^[AR]$/));
1719                if ($self->ra->check_path($self->{path}, $r) ==
1720                    $SVN::Node::dir) {
1721                        return 1;
1722                }
1723        }
1724        return 0;
1725}
1726
1727sub find_parent_branch {
1728        my ($self, $paths, $rev) = @_;
1729        return undef unless $self->follow_parent;
1730        unless (defined $paths) {
1731                my $err_handler = $SVN::Error::handler;
1732                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1733                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1734                                   $paths =
1735                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
1736                $SVN::Error::handler = $err_handler;
1737        }
1738        return undef unless defined $paths;
1739
1740        # look for a parent from another branch:
1741        my @b_path_components = split m#/#, $self->rel_path;
1742        my @a_path_components;
1743        my $i;
1744        while (@b_path_components) {
1745                $i = $paths->{'/'.join('/', @b_path_components)};
1746                last if $i && defined $i->{copyfrom_path};
1747                unshift(@a_path_components, pop(@b_path_components));
1748        }
1749        return undef unless defined $i && defined $i->{copyfrom_path};
1750        my $branch_from = $i->{copyfrom_path};
1751        if (@a_path_components) {
1752                print STDERR "branch_from: $branch_from => ";
1753                $branch_from .= '/'.join('/', @a_path_components);
1754                print STDERR $branch_from, "\n";
1755        }
1756        my $r = $i->{copyfrom_rev};
1757        my $repos_root = $self->ra->{repos_root};
1758        my $url = $self->ra->{url};
1759        my $new_url = $repos_root . $branch_from;
1760        print STDERR  "Found possible branch point: ",
1761                      "$new_url => ", $self->full_url, ", $r\n";
1762        $branch_from =~ s#^/##;
1763        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
1764        unless ($gs) {
1765                my $ref_id = $self->{ref_id};
1766                $ref_id =~ s/\@\d+$//;
1767                $ref_id .= "\@$r";
1768                # just grow a tail if we're not unique enough :x
1769                $ref_id .= '-' while find_ref($ref_id);
1770                print STDERR "Initializing parent: $ref_id\n";
1771                $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1772        }
1773        my ($r0, $parent) = $gs->find_rev_before($r, 1);
1774        if (!defined $r0 || !defined $parent) {
1775                my ($base, $head) = parse_revision_argument(0, $r);
1776                if ($base <= $r) {
1777                        $gs->fetch($base, $r);
1778                }
1779                ($r0, $parent) = $gs->last_rev_commit;
1780        }
1781        if (defined $r0 && defined $parent) {
1782                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1783                my $ed;
1784                if ($self->ra->can_do_switch) {
1785                        $self->assert_index_clean($parent);
1786                        print STDERR "Following parent with do_switch\n";
1787                        # do_switch works with svn/trunk >= r22312, but that
1788                        # is not included with SVN 1.4.3 (the latest version
1789                        # at the moment), so we can't rely on it
1790                        $self->{last_commit} = $parent;
1791                        $ed = SVN::Git::Fetcher->new($self);
1792                        $gs->ra->gs_do_switch($r0, $rev, $gs,
1793                                              $self->full_url, $ed)
1794                          or die "SVN connection failed somewhere...\n";
1795                } else {
1796                        print STDERR "Following parent with do_update\n";
1797                        $ed = SVN::Git::Fetcher->new($self);
1798                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
1799                          or die "SVN connection failed somewhere...\n";
1800                }
1801                print STDERR "Successfully followed parent\n";
1802                return $self->make_log_entry($rev, [$parent], $ed);
1803        }
1804        return undef;
1805}
1806
1807sub do_fetch {
1808        my ($self, $paths, $rev) = @_;
1809        my $ed;
1810        my ($last_rev, @parents);
1811        if (my $lc = $self->last_commit) {
1812                # we can have a branch that was deleted, then re-added
1813                # under the same name but copied from another path, in
1814                # which case we'll have multiple parents (we don't
1815                # want to break the original ref, nor lose copypath info):
1816                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1817                        push @{$log_entry->{parents}}, $lc;
1818                        return $log_entry;
1819                }
1820                $ed = SVN::Git::Fetcher->new($self);
1821                $last_rev = $self->{last_rev};
1822                $ed->{c} = $lc;
1823                @parents = ($lc);
1824        } else {
1825                $last_rev = $rev;
1826                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1827                        return $log_entry;
1828                }
1829                $ed = SVN::Git::Fetcher->new($self);
1830        }
1831        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1832                die "SVN connection failed somewhere...\n";
1833        }
1834        $self->make_log_entry($rev, \@parents, $ed);
1835}
1836
1837sub get_untracked {
1838        my ($self, $ed) = @_;
1839        my @out;
1840        my $h = $ed->{empty};
1841        foreach (sort keys %$h) {
1842                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1843                push @out, "  $act: " . uri_encode($_);
1844                warn "W: $act: $_\n";
1845        }
1846        foreach my $t (qw/dir_prop file_prop/) {
1847                $h = $ed->{$t} or next;
1848                foreach my $path (sort keys %$h) {
1849                        my $ppath = $path eq '' ? '.' : $path;
1850                        foreach my $prop (sort keys %{$h->{$path}}) {
1851                                next if $SKIP_PROP{$prop};
1852                                my $v = $h->{$path}->{$prop};
1853                                my $t_ppath_prop = "$t: " .
1854                                                    uri_encode($ppath) . ' ' .
1855                                                    uri_encode($prop);
1856                                if (defined $v) {
1857                                        push @out, "  +$t_ppath_prop " .
1858                                                   uri_encode($v);
1859                                } else {
1860                                        push @out, "  -$t_ppath_prop";
1861                                }
1862                        }
1863                }
1864        }
1865        foreach my $t (qw/absent_file absent_directory/) {
1866                $h = $ed->{$t} or next;
1867                foreach my $parent (sort keys %$h) {
1868                        foreach my $path (sort @{$h->{$parent}}) {
1869                                push @out, "  $t: " .
1870                                           uri_encode("$parent/$path");
1871                                warn "W: $t: $parent/$path ",
1872                                     "Insufficient permissions?\n";
1873                        }
1874                }
1875        }
1876        \@out;
1877}
1878
1879sub parse_svn_date {
1880        my $date = shift || return '+0000 1970-01-01 00:00:00';
1881        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1882                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1883                                         croak "Unable to parse date: $date\n";
1884        "+0000 $Y-$m-$d $H:$M:$S";
1885}
1886
1887sub check_author {
1888        my ($author) = @_;
1889        if (!defined $author || length $author == 0) {
1890                $author = '(no author)';
1891        }
1892        if (defined $::_authors && ! defined $::users{$author}) {
1893                die "Author: $author not defined in $::_authors file\n";
1894        }
1895        $author;
1896}
1897
1898sub make_log_entry {
1899        my ($self, $rev, $parents, $ed) = @_;
1900        my $untracked = $self->get_untracked($ed);
1901
1902        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1903        print $un "r$rev\n" or croak $!;
1904        print $un $_, "\n" foreach @$untracked;
1905        my %log_entry = ( parents => $parents || [], revision => $rev,
1906                          log => '');
1907
1908        my $headrev;
1909        my $logged = delete $self->{logged_rev_props};
1910        if (!$logged || $self->{-want_revprops}) {
1911                my $rp = $self->ra->rev_proplist($rev);
1912                foreach (sort keys %$rp) {
1913                        my $v = $rp->{$_};
1914                        if (/^svn:(author|date|log)$/) {
1915                                $log_entry{$1} = $v;
1916                        } elsif ($_ eq 'svm:headrev') {
1917                                $headrev = $v;
1918                        } else {
1919                                print $un "  rev_prop: ", uri_encode($_), ' ',
1920                                          uri_encode($v), "\n";
1921                        }
1922                }
1923        } else {
1924                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1925        }
1926        close $un or croak $!;
1927
1928        $log_entry{date} = parse_svn_date($log_entry{date});
1929        $log_entry{log} .= "\n";
1930        my $author = $log_entry{author} = check_author($log_entry{author});
1931        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1932                                                       : ($author, undef);
1933        if (defined $headrev && $self->use_svm_props) {
1934                if ($self->rewrite_root) {
1935                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
1936                            "options set!\n";
1937                }
1938                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1939                # we don't want "SVM: initializing mirror for junk" ...
1940                return undef if $r == 0;
1941                my $svm = $self->svm;
1942                if ($uuid ne $svm->{uuid}) {
1943                        die "UUID mismatch on SVM path:\n",
1944                            "expected: $svm->{uuid}\n",
1945                            "     got: $uuid\n";
1946                }
1947                my $full_url = $self->full_url;
1948                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
1949                             die "Failed to replace '$svm->{replace}' with ",
1950                                 "'$svm->{source}' in $full_url\n";
1951                # throw away username for storing in records
1952                remove_username($full_url);
1953                $log_entry{metadata} = "$full_url\@$r $uuid";
1954                $log_entry{svm_revision} = $r;
1955                $email ||= "$author\@$uuid"
1956        } elsif ($self->use_svnsync_props) {
1957                my $full_url = $self->svnsync->{url};
1958                $full_url .= "/$self->{path}" if length $self->{path};
1959                remove_username($full_url);
1960                my $uuid = $self->svnsync->{uuid};
1961                $log_entry{metadata} = "$full_url\@$rev $uuid";
1962                $email ||= "$author\@$uuid"
1963        } else {
1964                my $url = $self->metadata_url;
1965                remove_username($url);
1966                $log_entry{metadata} = "$url\@$rev " .
1967                                       $self->ra->get_uuid;
1968                $email ||= "$author\@" . $self->ra->get_uuid;
1969        }
1970        $log_entry{name} = $name;
1971        $log_entry{email} = $email;
1972        \%log_entry;
1973}
1974
1975sub fetch {
1976        my ($self, $min_rev, $max_rev, @parents) = @_;
1977        my ($last_rev, $last_commit) = $self->last_rev_commit;
1978        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1979        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1980}
1981
1982sub set_tree_cb {
1983        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1984        $self->{inject_parents} = { $rev => $tree };
1985        $self->fetch(undef, undef);
1986}
1987
1988sub set_tree {
1989        my ($self, $tree) = (shift, shift);
1990        my $log_entry = ::get_commit_entry($tree);
1991        unless ($self->{last_rev}) {
1992                fatal("Must have an existing revision to commit\n");
1993        }
1994        my %ed_opts = ( r => $self->{last_rev},
1995                        log => $log_entry->{log},
1996                        ra => $self->ra,
1997                        tree_a => $self->{last_commit},
1998                        tree_b => $tree,
1999                        editor_cb => sub {
2000                               $self->set_tree_cb($log_entry, $tree, @_) },
2001                        svn_path => $self->{path} );
2002        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2003                print "No changes\nr$self->{last_rev} = $tree\n";
2004        }
2005}
2006
2007sub rebuild {
2008        my ($self) = @_;
2009        my $db_path = $self->db_path;
2010        return if (-e $db_path && ! -z $db_path);
2011        return unless ::verify_ref($self->refname.'^0');
2012        if (-f $self->{db_root}) {
2013                rename $self->{db_root}, $db_path or die
2014                     "rename $self->{db_root} => $db_path failed: $!\n";
2015                my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
2016                symlink $base, $self->{db_root} or die
2017                     "symlink $base => $self->{db_root} failed: $!\n";
2018                return;
2019        }
2020        print "Rebuilding $db_path ...\n";
2021        my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
2022        my $latest;
2023        my $full_url = $self->full_url;
2024        remove_username($full_url);
2025        my $svn_uuid;
2026        while (<$rev_list>) {
2027                chomp;
2028                my $c = $_;
2029                die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
2030                my ($url, $rev, $uuid) = ::cmt_metadata($c);
2031                remove_username($url);
2032
2033                # ignore merges (from set-tree)
2034                next if (!defined $rev || !$uuid);
2035
2036                # if we merged or otherwise started elsewhere, this is
2037                # how we break out of it
2038                if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
2039                    ($full_url && $url && ($url ne $full_url))) {
2040                        next;
2041                }
2042                $latest ||= $rev;
2043                $svn_uuid ||= $uuid;
2044
2045                $self->rev_db_set($rev, $c);
2046                print "r$rev = $c\n";
2047        }
2048        command_close_pipe($rev_list, $ctx);
2049        print "Done rebuilding $db_path\n";
2050}
2051
2052# rev_db:
2053# Tie::File seems to be prone to offset errors if revisions get sparse,
2054# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
2055# one of my favorite modules is out :<  Next up would be one of the DBM
2056# modules, but I'm not sure which is most portable...  So I'll just
2057# go with something that's plain-text, but still capable of
2058# being randomly accessed.  So here's my ultra-simple fixed-width
2059# database.  All records are 40 characters + "\n", so it's easy to seek
2060# to a revision: (41 * rev) is the byte offset.
2061# A record of 40 0s denotes an empty revision.
2062# And yes, it's still pretty fast (faster than Tie::File).
2063# These files are disposable unless noMetadata or useSvmProps is set
2064
2065sub _rev_db_set {
2066        my ($fh, $rev, $commit) = @_;
2067        my $offset = $rev * 41;
2068        # assume that append is the common case:
2069        seek $fh, 0, 2 or croak $!;
2070        my $pos = tell $fh;
2071        if ($pos < $offset) {
2072                for (1 .. (($offset - $pos) / 41)) {
2073                        print $fh (('0' x 40),"\n") or croak $!;
2074                }
2075        }
2076        seek $fh, $offset, 0 or croak $!;
2077        print $fh $commit,"\n" or croak $!;
2078}
2079
2080sub mkfile {
2081        my ($path) = @_;
2082        unless (-e $path) {
2083                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2084                mkpath([$dir]) unless -d $dir;
2085                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2086                close $fh or die "Couldn't close (create) $path: $!\n";
2087        }
2088}
2089
2090sub rev_db_set {
2091        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2092        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2093        my $db = $self->db_path($uuid);
2094        my $db_lock = "$db.lock";
2095        my $sig;
2096        if ($update_ref) {
2097                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2098                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2099        }
2100        mkfile($db);
2101
2102        $LOCKFILES{$db_lock} = 1;
2103        my $sync;
2104        # both of these options make our .rev_db file very, very important
2105        # and we can't afford to lose it because rebuild() won't work
2106        if ($self->use_svm_props || $self->no_metadata) {
2107                $sync = 1;
2108                copy($db, $db_lock) or die "rev_db_set(@_): ",
2109                                           "Failed to copy: ",
2110                                           "$db => $db_lock ($!)\n";
2111        } else {
2112                rename $db, $db_lock or die "rev_db_set(@_): ",
2113                                            "Failed to rename: ",
2114                                            "$db => $db_lock ($!)\n";
2115        }
2116        open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
2117        _rev_db_set($fh, $rev, $commit);
2118        if ($sync) {
2119                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2120                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2121        }
2122        close $fh or croak $!;
2123        if ($update_ref) {
2124                $_head = $self;
2125                command_noisy('update-ref', '-m', "r$rev",
2126                              $self->refname, $commit);
2127        }
2128        rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
2129                                    "$db_lock => $db ($!)\n";
2130        delete $LOCKFILES{$db_lock};
2131        if ($update_ref) {
2132                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2133                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2134                kill $sig, $$ if defined $sig;
2135        }
2136}
2137
2138sub rev_db_max {
2139        my ($self) = @_;
2140        $self->rebuild;
2141        my $db_path = $self->db_path;
2142        my @stat = stat $db_path or return 0;
2143        ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
2144        my $max = $stat[7] / 41;
2145        (($max > 0) ? $max - 1 : 0);
2146}
2147
2148sub rev_db_get {
2149        my ($self, $rev, $uuid) = @_;
2150        my $ret;
2151        my $offset = $rev * 41;
2152        my $db_path = $self->db_path($uuid);
2153        return undef unless -e $db_path;
2154        open my $fh, '<', $db_path or croak $!;
2155        if (sysseek($fh, $offset, 0) == $offset) {
2156                my $read = sysread($fh, $ret, 40);
2157                $ret = undef if ($read != 40 || $ret eq ('0'x40));
2158        }
2159        close $fh or croak $!;
2160        $ret;
2161}
2162
2163sub find_rev_before {
2164        my ($self, $rev, $eq_ok) = @_;
2165        --$rev unless $eq_ok;
2166        while ($rev > 0) {
2167                if (my $c = $self->rev_db_get($rev)) {
2168                        return ($rev, $c);
2169                }
2170                --$rev;
2171        }
2172        return (undef, undef);
2173}
2174
2175sub _new {
2176        my ($class, $repo_id, $ref_id, $path) = @_;
2177        unless (defined $repo_id && length $repo_id) {
2178                $repo_id = $Git::SVN::default_repo_id;
2179        }
2180        unless (defined $ref_id && length $ref_id) {
2181                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2182        }
2183        $_[1] = $repo_id = sanitize_remote_name($repo_id);
2184        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2185        $_[3] = $path = '' unless (defined $path);
2186        mkpath(["$ENV{GIT_DIR}/svn"]);
2187        bless {
2188                ref_id => $ref_id, dir => $dir, index => "$dir/index",
2189                path => $path, config => "$ENV{GIT_DIR}/svn/config",
2190                db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2191}
2192
2193sub db_path {
2194        my ($self, $uuid) = @_;
2195        $uuid ||= $self->ra_uuid;
2196        "$self->{db_root}.$uuid";
2197}
2198
2199sub uri_encode {
2200        my ($f) = @_;
2201        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2202        $f
2203}
2204
2205sub remove_username {
2206        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2207}
2208
2209package Git::SVN::Prompt;
2210use strict;
2211use warnings;
2212require SVN::Core;
2213use vars qw/$_no_auth_cache $_username/;
2214
2215sub simple {
2216        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2217        $may_save = undef if $_no_auth_cache;
2218        $default_username = $_username if defined $_username;
2219        if (defined $default_username && length $default_username) {
2220                if (defined $realm && length $realm) {
2221                        print STDERR "Authentication realm: $realm\n";
2222                        STDERR->flush;
2223                }
2224                $cred->username($default_username);
2225        } else {
2226                username($cred, $realm, $may_save, $pool);
2227        }
2228        $cred->password(_read_password("Password for '" .
2229                                       $cred->username . "': ", $realm));
2230        $cred->may_save($may_save);
2231        $SVN::_Core::SVN_NO_ERROR;
2232}
2233
2234sub ssl_server_trust {
2235        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2236        $may_save = undef if $_no_auth_cache;
2237        print STDERR "Error validating server certificate for '$realm':\n";
2238        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2239                print STDERR " - The certificate is not issued by a trusted ",
2240                      "authority. Use the\n",
2241                      "   fingerprint to validate the certificate manually!\n";
2242        }
2243        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2244                print STDERR " - The certificate hostname does not match.\n";
2245        }
2246        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2247                print STDERR " - The certificate is not yet valid.\n";
2248        }
2249        if ($failures & $SVN::Auth::SSL::EXPIRED) {
2250                print STDERR " - The certificate has expired.\n";
2251        }
2252        if ($failures & $SVN::Auth::SSL::OTHER) {
2253                print STDERR " - The certificate has an unknown error.\n";
2254        }
2255        printf STDERR
2256                "Certificate information:\n".
2257                " - Hostname: %s\n".
2258                " - Valid: from %s until %s\n".
2259                " - Issuer: %s\n".
2260                " - Fingerprint: %s\n",
2261                map $cert_info->$_, qw(hostname valid_from valid_until
2262                                       issuer_dname fingerprint);
2263        my $choice;
2264prompt:
2265        print STDERR $may_save ?
2266              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2267              "(R)eject or accept (t)emporarily? ";
2268        STDERR->flush;
2269        $choice = lc(substr(<STDIN> || 'R', 0, 1));
2270        if ($choice =~ /^t$/i) {
2271                $cred->may_save(undef);
2272        } elsif ($choice =~ /^r$/i) {
2273                return -1;
2274        } elsif ($may_save && $choice =~ /^p$/i) {
2275                $cred->may_save($may_save);
2276        } else {
2277                goto prompt;
2278        }
2279        $cred->accepted_failures($failures);
2280        $SVN::_Core::SVN_NO_ERROR;
2281}
2282
2283sub ssl_client_cert {
2284        my ($cred, $realm, $may_save, $pool) = @_;
2285        $may_save = undef if $_no_auth_cache;
2286        print STDERR "Client certificate filename: ";
2287        STDERR->flush;
2288        chomp(my $filename = <STDIN>);
2289        $cred->cert_file($filename);
2290        $cred->may_save($may_save);
2291        $SVN::_Core::SVN_NO_ERROR;
2292}
2293
2294sub ssl_client_cert_pw {
2295        my ($cred, $realm, $may_save, $pool) = @_;
2296        $may_save = undef if $_no_auth_cache;
2297        $cred->password(_read_password("Password: ", $realm));
2298        $cred->may_save($may_save);
2299        $SVN::_Core::SVN_NO_ERROR;
2300}
2301
2302sub username {
2303        my ($cred, $realm, $may_save, $pool) = @_;
2304        $may_save = undef if $_no_auth_cache;
2305        if (defined $realm && length $realm) {
2306                print STDERR "Authentication realm: $realm\n";
2307        }
2308        my $username;
2309        if (defined $_username) {
2310                $username = $_username;
2311        } else {
2312                print STDERR "Username: ";
2313                STDERR->flush;
2314                chomp($username = <STDIN>);
2315        }
2316        $cred->username($username);
2317        $cred->may_save($may_save);
2318        $SVN::_Core::SVN_NO_ERROR;
2319}
2320
2321sub _read_password {
2322        my ($prompt, $realm) = @_;
2323        print STDERR $prompt;
2324        STDERR->flush;
2325        require Term::ReadKey;
2326        Term::ReadKey::ReadMode('noecho');
2327        my $password = '';
2328        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2329                last if $key =~ /[\012\015]/; # \n\r
2330                $password .= $key;
2331        }
2332        Term::ReadKey::ReadMode('restore');
2333        print STDERR "\n";
2334        STDERR->flush;
2335        $password;
2336}
2337
2338package main;
2339
2340{
2341        my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2342                                $SVN::Node::dir.$SVN::Node::unknown.
2343                                $SVN::Node::none.$SVN::Node::file.
2344                                $SVN::Node::dir.$SVN::Node::unknown.
2345                                $SVN::Auth::SSL::CNMISMATCH.
2346                                $SVN::Auth::SSL::NOTYETVALID.
2347                                $SVN::Auth::SSL::EXPIRED.
2348                                $SVN::Auth::SSL::UNKNOWNCA.
2349                                $SVN::Auth::SSL::OTHER;
2350}
2351
2352package SVN::Git::Fetcher;
2353use vars qw/@ISA/;
2354use strict;
2355use warnings;
2356use Carp qw/croak/;
2357use IO::File qw//;
2358use Digest::MD5;
2359
2360# file baton members: path, mode_a, mode_b, pool, fh, blob, base
2361sub new {
2362        my ($class, $git_svn) = @_;
2363        my $self = SVN::Delta::Editor->new;
2364        bless $self, $class;
2365        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2366        $self->{empty} = {};
2367        $self->{dir_prop} = {};
2368        $self->{file_prop} = {};
2369        $self->{absent_dir} = {};
2370        $self->{absent_file} = {};
2371        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2372        $self;
2373}
2374
2375sub set_path_strip {
2376        my ($self, $path) = @_;
2377        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2378}
2379
2380sub open_root {
2381        { path => '' };
2382}
2383
2384sub open_directory {
2385        my ($self, $path, $pb, $rev) = @_;
2386        { path => $path };
2387}
2388
2389sub git_path {
2390        my ($self, $path) = @_;
2391        if ($self->{path_strip}) {
2392                $path =~ s!$self->{path_strip}!! or
2393                  die "Failed to strip path '$path' ($self->{path_strip})\n";
2394        }
2395        $path;
2396}
2397
2398sub delete_entry {
2399        my ($self, $path, $rev, $pb) = @_;
2400
2401        my $gpath = $self->git_path($path);
2402        return undef if ($gpath eq '');
2403
2404        # remove entire directories.
2405        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2406                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2407                                                     -r --name-only -z/,
2408                                                     $self->{c}, '--', $gpath);
2409                local $/ = "\0";
2410                while (<$ls>) {
2411                        chomp;
2412                        $self->{gii}->remove($_);
2413                        print "\tD\t$_\n" unless $::_q;
2414                }
2415                print "\tD\t$gpath/\n" unless $::_q;
2416                command_close_pipe($ls, $ctx);
2417                $self->{empty}->{$path} = 0
2418        } else {
2419                $self->{gii}->remove($gpath);
2420                print "\tD\t$gpath\n" unless $::_q;
2421        }
2422        undef;
2423}
2424
2425sub open_file {
2426        my ($self, $path, $pb, $rev) = @_;
2427        my $gpath = $self->git_path($path);
2428        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2429                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2430        unless (defined $mode && defined $blob) {
2431                die "$path was not found in commit $self->{c} (r$rev)\n";
2432        }
2433        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2434          pool => SVN::Pool->new, action => 'M' };
2435}
2436
2437sub add_file {
2438        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2439        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2440        delete $self->{empty}->{$dir};
2441        { path => $path, mode_a => 100644, mode_b => 100644,
2442          pool => SVN::Pool->new, action => 'A' };
2443}
2444
2445sub add_directory {
2446        my ($self, $path, $cp_path, $cp_rev) = @_;
2447        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2448        delete $self->{empty}->{$dir};
2449        $self->{empty}->{$path} = 1;
2450        { path => $path };
2451}
2452
2453sub change_dir_prop {
2454        my ($self, $db, $prop, $value) = @_;
2455        $self->{dir_prop}->{$db->{path}} ||= {};
2456        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2457        undef;
2458}
2459
2460sub absent_directory {
2461        my ($self, $path, $pb) = @_;
2462        $self->{absent_dir}->{$pb->{path}} ||= [];
2463        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2464        undef;
2465}
2466
2467sub absent_file {
2468        my ($self, $path, $pb) = @_;
2469        $self->{absent_file}->{$pb->{path}} ||= [];
2470        push @{$self->{absent_file}->{$pb->{path}}}, $path;
2471        undef;
2472}
2473
2474sub change_file_prop {
2475        my ($self, $fb, $prop, $value) = @_;
2476        if ($prop eq 'svn:executable') {
2477                if ($fb->{mode_b} != 120000) {
2478                        $fb->{mode_b} = defined $value ? 100755 : 100644;
2479                }
2480        } elsif ($prop eq 'svn:special') {
2481                $fb->{mode_b} = defined $value ? 120000 : 100644;
2482        } else {
2483                $self->{file_prop}->{$fb->{path}} ||= {};
2484                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2485        }
2486        undef;
2487}
2488
2489sub apply_textdelta {
2490        my ($self, $fb, $exp) = @_;
2491        my $fh = IO::File->new_tmpfile;
2492        $fh->autoflush(1);
2493        # $fh gets auto-closed() by SVN::TxDelta::apply(),
2494        # (but $base does not,) so dup() it for reading in close_file
2495        open my $dup, '<&', $fh or croak $!;
2496        my $base = IO::File->new_tmpfile;
2497        $base->autoflush(1);
2498        if ($fb->{blob}) {
2499                defined (my $pid = fork) or croak $!;
2500                if (!$pid) {
2501                        open STDOUT, '>&', $base or croak $!;
2502                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2503                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2504                }
2505                waitpid $pid, 0;
2506                croak $? if $?;
2507
2508                if (defined $exp) {
2509                        seek $base, 0, 0 or croak $!;
2510                        my $md5 = Digest::MD5->new;
2511                        $md5->addfile($base);
2512                        my $got = $md5->hexdigest;
2513                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2514                            "expected: $exp\n",
2515                            "     got: $got\n" if ($got ne $exp);
2516                }
2517        }
2518        seek $base, 0, 0 or croak $!;
2519        $fb->{fh} = $dup;
2520        $fb->{base} = $base;
2521        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2522}
2523
2524sub close_file {
2525        my ($self, $fb, $exp) = @_;
2526        my $hash;
2527        my $path = $self->git_path($fb->{path});
2528        if (my $fh = $fb->{fh}) {
2529                if (defined $exp) {
2530                        seek($fh, 0, 0) or croak $!;
2531                        my $md5 = Digest::MD5->new;
2532                        $md5->addfile($fh);
2533                        my $got = $md5->hexdigest;
2534                        if ($got ne $exp) {
2535                                die "Checksum mismatch: $path\n",
2536                                    "expected: $exp\n    got: $got\n";
2537                        }
2538                }
2539                sysseek($fh, 0, 0) or croak $!;
2540                if ($fb->{mode_b} == 120000) {
2541                        sysread($fh, my $buf, 5) == 5 or croak $!;
2542                        $buf eq 'link ' or die "$path has mode 120000",
2543                                               "but is not a link\n";
2544                }
2545                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2546                if (!$pid) {
2547                        open STDIN, '<&', $fh or croak $!;
2548                        exec qw/git-hash-object -w --stdin/ or croak $!;
2549                }
2550                chomp($hash = do { local $/; <$out> });
2551                close $out or croak $!;
2552                close $fh or croak $!;
2553                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2554                close $fb->{base} or croak $!;
2555        } else {
2556                $hash = $fb->{blob} or die "no blob information\n";
2557        }
2558        $fb->{pool}->clear;
2559        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2560        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2561        undef;
2562}
2563
2564sub abort_edit {
2565        my $self = shift;
2566        $self->{nr} = $self->{gii}->{nr};
2567        delete $self->{gii};
2568        $self->SUPER::abort_edit(@_);
2569}
2570
2571sub close_edit {
2572        my $self = shift;
2573        $self->{git_commit_ok} = 1;
2574        $self->{nr} = $self->{gii}->{nr};
2575        delete $self->{gii};
2576        $self->SUPER::close_edit(@_);
2577}
2578
2579package SVN::Git::Editor;
2580use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2581use strict;
2582use warnings;
2583use Carp qw/croak/;
2584use IO::File;
2585use Digest::MD5;
2586
2587sub new {
2588        my ($class, $opts) = @_;
2589        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2590                die "$_ required!\n" unless (defined $opts->{$_});
2591        }
2592
2593        my $pool = SVN::Pool->new;
2594        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2595        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2596                                     $opts->{r}, $mods);
2597
2598        # $opts->{ra} functions should not be used after this:
2599        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
2600                                                $opts->{editor_cb}, $pool);
2601        my $self = SVN::Delta::Editor->new(@ce, $pool);
2602        bless $self, $class;
2603        foreach (qw/svn_path r tree_a tree_b/) {
2604                $self->{$_} = $opts->{$_};
2605        }
2606        $self->{url} = $opts->{ra}->{url};
2607        $self->{mods} = $mods;
2608        $self->{types} = $types;
2609        $self->{pool} = $pool;
2610        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2611        $self->{rm} = { };
2612        $self->{path_prefix} = length $self->{svn_path} ?
2613                               "$self->{svn_path}/" : '';
2614        return $self;
2615}
2616
2617sub generate_diff {
2618        my ($tree_a, $tree_b) = @_;
2619        my @diff_tree = qw(diff-tree -z -r);
2620        if ($_cp_similarity) {
2621                push @diff_tree, "-C$_cp_similarity";
2622        } else {
2623                push @diff_tree, '-C';
2624        }
2625        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2626        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2627        push @diff_tree, $tree_a, $tree_b;
2628        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2629        local $/ = "\0";
2630        my $state = 'meta';
2631        my @mods;
2632        while (<$diff_fh>) {
2633                chomp $_; # this gets rid of the trailing "\0"
2634                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2635                                        $::sha1\s($::sha1)\s
2636                                        ([MTCRAD])\d*$/xo) {
2637                        push @mods, {   mode_a => $1, mode_b => $2,
2638                                        sha1_b => $3, chg => $4 };
2639                        if ($4 =~ /^(?:C|R)$/) {
2640                                $state = 'file_a';
2641                        } else {
2642                                $state = 'file_b';
2643                        }
2644                } elsif ($state eq 'file_a') {
2645                        my $x = $mods[$#mods] or croak "Empty array\n";
2646                        if ($x->{chg} !~ /^(?:C|R)$/) {
2647                                croak "Error parsing $_, $x->{chg}\n";
2648                        }
2649                        $x->{file_a} = $_;
2650                        $state = 'file_b';
2651                } elsif ($state eq 'file_b') {
2652                        my $x = $mods[$#mods] or croak "Empty array\n";
2653                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2654                                croak "Error parsing $_, $x->{chg}\n";
2655                        }
2656                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2657                                croak "Error parsing $_, $x->{chg}\n";
2658                        }
2659                        $x->{file_b} = $_;
2660                        $state = 'meta';
2661                } else {
2662                        croak "Error parsing $_\n";
2663                }
2664        }
2665        command_close_pipe($diff_fh, $ctx);
2666        \@mods;
2667}
2668
2669sub check_diff_paths {
2670        my ($ra, $pfx, $rev, $mods) = @_;
2671        my %types;
2672        $pfx .= '/' if length $pfx;
2673
2674        sub type_diff_paths {
2675                my ($ra, $types, $path, $rev) = @_;
2676                my @p = split m#/+#, $path;
2677                my $c = shift @p;
2678                unless (defined $types->{$c}) {
2679                        $types->{$c} = $ra->check_path($c, $rev);
2680                }
2681                while (@p) {
2682                        $c .= '/' . shift @p;
2683                        next if defined $types->{$c};
2684                        $types->{$c} = $ra->check_path($c, $rev);
2685                }
2686        }
2687
2688        foreach my $m (@$mods) {
2689                foreach my $f (qw/file_a file_b/) {
2690                        next unless defined $m->{$f};
2691                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2692                        if (length $pfx.$dir && ! defined $types{$dir}) {
2693                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2694                        }
2695                }
2696        }
2697        \%types;
2698}
2699
2700sub split_path {
2701        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2702}
2703
2704sub repo_path {
2705        my ($self, $path) = @_;
2706        $self->{path_prefix}.(defined $path ? $path : '');
2707}
2708
2709sub url_path {
2710        my ($self, $path) = @_;
2711        $self->{url} . '/' . $self->repo_path($path);
2712}
2713
2714sub rmdirs {
2715        my ($self) = @_;
2716        my $rm = $self->{rm};
2717        delete $rm->{''}; # we never delete the url we're tracking
2718        return unless %$rm;
2719
2720        foreach (keys %$rm) {
2721                my @d = split m#/#, $_;
2722                my $c = shift @d;
2723                $rm->{$c} = 1;
2724                while (@d) {
2725                        $c .= '/' . shift @d;
2726                        $rm->{$c} = 1;
2727                }
2728        }
2729        delete $rm->{$self->{svn_path}};
2730        delete $rm->{''}; # we never delete the url we're tracking
2731        return unless %$rm;
2732
2733        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2734                                             $self->{tree_b});
2735        local $/ = "\0";
2736        while (<$fh>) {
2737                chomp;
2738                my @dn = split m#/#, $_;
2739                while (pop @dn) {
2740                        delete $rm->{join '/', @dn};
2741                }
2742                unless (%$rm) {
2743                        close $fh;
2744                        return;
2745                }
2746        }
2747        command_close_pipe($fh, $ctx);
2748
2749        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2750        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2751                $self->close_directory($bat->{$d}, $p);
2752                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2753                print "\tD+\t$d/\n" unless $::_q;
2754                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2755                delete $bat->{$d};
2756        }
2757}
2758
2759sub open_or_add_dir {
2760        my ($self, $full_path, $baton) = @_;
2761        my $t = $self->{types}->{$full_path};
2762        if (!defined $t) {
2763                die "$full_path not known in r$self->{r} or we have a bug!\n";
2764        }
2765        if ($t == $SVN::Node::none) {
2766                return $self->add_directory($full_path, $baton,
2767                                                undef, -1, $self->{pool});
2768        } elsif ($t == $SVN::Node::dir) {
2769                return $self->open_directory($full_path, $baton,
2770                                                $self->{r}, $self->{pool});
2771        }
2772        print STDERR "$full_path already exists in repository at ",
2773                "r$self->{r} and it is not a directory (",
2774                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2775        exit 1;
2776}
2777
2778sub ensure_path {
2779        my ($self, $path) = @_;
2780        my $bat = $self->{bat};
2781        my $repo_path = $self->repo_path($path);
2782        return $bat->{''} unless (length $repo_path);
2783        my @p = split m#/+#, $repo_path;
2784        my $c = shift @p;
2785        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2786        while (@p) {
2787                my $c0 = $c;
2788                $c .= '/' . shift @p;
2789                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2790        }
2791        return $bat->{$c};
2792}
2793
2794sub A {
2795        my ($self, $m) = @_;
2796        my ($dir, $file) = split_path($m->{file_b});
2797        my $pbat = $self->ensure_path($dir);
2798        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2799                                        undef, -1);
2800        print "\tA\t$m->{file_b}\n" unless $::_q;
2801        $self->chg_file($fbat, $m);
2802        $self->close_file($fbat,undef,$self->{pool});
2803}
2804
2805sub C {
2806        my ($self, $m) = @_;
2807        my ($dir, $file) = split_path($m->{file_b});
2808        my $pbat = $self->ensure_path($dir);
2809        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2810                                $self->url_path($m->{file_a}), $self->{r});
2811        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2812        $self->chg_file($fbat, $m);
2813        $self->close_file($fbat,undef,$self->{pool});
2814}
2815
2816sub delete_entry {
2817        my ($self, $path, $pbat) = @_;
2818        my $rpath = $self->repo_path($path);
2819        my ($dir, $file) = split_path($rpath);
2820        $self->{rm}->{$dir} = 1;
2821        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2822}
2823
2824sub R {
2825        my ($self, $m) = @_;
2826        my ($dir, $file) = split_path($m->{file_b});
2827        my $pbat = $self->ensure_path($dir);
2828        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2829                                $self->url_path($m->{file_a}), $self->{r});
2830        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2831        $self->chg_file($fbat, $m);
2832        $self->close_file($fbat,undef,$self->{pool});
2833
2834        ($dir, $file) = split_path($m->{file_a});
2835        $pbat = $self->ensure_path($dir);
2836        $self->delete_entry($m->{file_a}, $pbat);
2837}
2838
2839sub M {
2840        my ($self, $m) = @_;
2841        my ($dir, $file) = split_path($m->{file_b});
2842        my $pbat = $self->ensure_path($dir);
2843        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2844                                $pbat,$self->{r},$self->{pool});
2845        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2846        $self->chg_file($fbat, $m);
2847        $self->close_file($fbat,undef,$self->{pool});
2848}
2849
2850sub T { shift->M(@_) }
2851
2852sub change_file_prop {
2853        my ($self, $fbat, $pname, $pval) = @_;
2854        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2855}
2856
2857sub chg_file {
2858        my ($self, $fbat, $m) = @_;
2859        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2860                $self->change_file_prop($fbat,'svn:executable','*');
2861        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2862                $self->change_file_prop($fbat,'svn:executable',undef);
2863        }
2864        my $fh = IO::File->new_tmpfile or croak $!;
2865        if ($m->{mode_b} =~ /^120/) {
2866                print $fh 'link ' or croak $!;
2867                $self->change_file_prop($fbat,'svn:special','*');
2868        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2869                $self->change_file_prop($fbat,'svn:special',undef);
2870        }
2871        defined(my $pid = fork) or croak $!;
2872        if (!$pid) {
2873                open STDOUT, '>&', $fh or croak $!;
2874                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2875        }
2876        waitpid $pid, 0;
2877        croak $? if $?;
2878        $fh->flush == 0 or croak $!;
2879        seek $fh, 0, 0 or croak $!;
2880
2881        my $md5 = Digest::MD5->new;
2882        $md5->addfile($fh) or croak $!;
2883        seek $fh, 0, 0 or croak $!;
2884
2885        my $exp = $md5->hexdigest;
2886        my $pool = SVN::Pool->new;
2887        my $atd = $self->apply_textdelta($fbat, undef, $pool);
2888        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2889        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2890        $pool->clear;
2891
2892        close $fh or croak $!;
2893}
2894
2895sub D {
2896        my ($self, $m) = @_;
2897        my ($dir, $file) = split_path($m->{file_b});
2898        my $pbat = $self->ensure_path($dir);
2899        print "\tD\t$m->{file_b}\n" unless $::_q;
2900        $self->delete_entry($m->{file_b}, $pbat);
2901}
2902
2903sub close_edit {
2904        my ($self) = @_;
2905        my ($p,$bat) = ($self->{pool}, $self->{bat});
2906        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2907                next if $_ eq '';
2908                $self->close_directory($bat->{$_}, $p);
2909        }
2910        $self->close_directory($bat->{''}, $p);
2911        $self->SUPER::close_edit($p);
2912        $p->clear;
2913}
2914
2915sub abort_edit {
2916        my ($self) = @_;
2917        $self->SUPER::abort_edit($self->{pool});
2918}
2919
2920sub DESTROY {
2921        my $self = shift;
2922        $self->SUPER::DESTROY(@_);
2923        $self->{pool}->clear;
2924}
2925
2926# this drives the editor
2927sub apply_diff {
2928        my ($self) = @_;
2929        my $mods = $self->{mods};
2930        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2931        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2932                my $f = $m->{chg};
2933                if (defined $o{$f}) {
2934                        $self->$f($m);
2935                } else {
2936                        fatal("Invalid change type: $f\n");
2937                }
2938        }
2939        $self->rmdirs if $_rmdir;
2940        if (@$mods == 0) {
2941                $self->abort_edit;
2942        } else {
2943                $self->close_edit;
2944        }
2945        return scalar @$mods;
2946}
2947
2948package Git::SVN::Ra;
2949use vars qw/@ISA $config_dir $_log_window_size/;
2950use strict;
2951use warnings;
2952my ($can_do_switch, %ignored_err, $RA);
2953
2954BEGIN {
2955        # enforce temporary pool usage for some simple functions
2956        my $e;
2957        foreach (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
2958                $e .= "sub $_ {
2959                        my \$self = shift;
2960                        my \$pool = SVN::Pool->new;
2961                        my \@ret = \$self->SUPER::$_(\@_,\$pool);
2962                        \$pool->clear;
2963                        wantarray ? \@ret : \$ret[0]; }\n";
2964        }
2965
2966        eval "$e; 1;" or die $@;
2967}
2968
2969sub new {
2970        my ($class, $url) = @_;
2971        $url =~ s!/+$!!;
2972        return $RA if ($RA && $RA->{url} eq $url);
2973
2974        SVN::_Core::svn_config_ensure($config_dir, undef);
2975        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2976            SVN::Client::get_simple_provider(),
2977            SVN::Client::get_ssl_server_trust_file_provider(),
2978            SVN::Client::get_simple_prompt_provider(
2979              \&Git::SVN::Prompt::simple, 2),
2980            SVN::Client::get_ssl_client_cert_prompt_provider(
2981              \&Git::SVN::Prompt::ssl_client_cert, 2),
2982            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2983              \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2984            SVN::Client::get_username_provider(),
2985            SVN::Client::get_ssl_server_trust_prompt_provider(
2986              \&Git::SVN::Prompt::ssl_server_trust),
2987            SVN::Client::get_username_prompt_provider(
2988              \&Git::SVN::Prompt::username, 2),
2989          ]);
2990        my $config = SVN::Core::config_get_config($config_dir);
2991        my $self = SVN::Ra->new(url => $url, auth => $baton,
2992                              config => $config,
2993                              pool => SVN::Pool->new,
2994                              auth_provider_callbacks => $callbacks);
2995        $self->{svn_path} = $url;
2996        $self->{repos_root} = $self->get_repos_root;
2997        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
2998        $self->{cache} = { check_path => { r => 0, data => {} },
2999                           get_dir => { r => 0, data => {} } };
3000        $RA = bless $self, $class;
3001}
3002
3003sub check_path {
3004        my ($self, $path, $r) = @_;
3005        my $cache = $self->{cache}->{check_path};
3006        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3007                return $cache->{data}->{$path};
3008        }
3009        my $pool = SVN::Pool->new;
3010        my $t = $self->SUPER::check_path($path, $r, $pool);
3011        $pool->clear;
3012        if ($r != $cache->{r}) {
3013                %{$cache->{data}} = ();
3014                $cache->{r} = $r;
3015        }
3016        $cache->{data}->{$path} = $t;
3017}
3018
3019sub get_dir {
3020        my ($self, $dir, $r) = @_;
3021        my $cache = $self->{cache}->{get_dir};
3022        if ($r == $cache->{r}) {
3023                if (my $x = $cache->{data}->{$dir}) {
3024                        return wantarray ? @$x : $x->[0];
3025                }
3026        }
3027        my $pool = SVN::Pool->new;
3028        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3029        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3030        $pool->clear;
3031        if ($r != $cache->{r}) {
3032                %{$cache->{data}} = ();
3033                $cache->{r} = $r;
3034        }
3035        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3036        wantarray ? (\%dirents, $r, $props) : \%dirents;
3037}
3038
3039sub DESTROY {
3040        # do not call the real DESTROY since we store ourselves in $RA
3041}
3042
3043sub get_log {
3044        my ($self, @args) = @_;
3045        my $pool = SVN::Pool->new;
3046        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3047        my $ret = $self->SUPER::get_log(@args, $pool);
3048        $pool->clear;
3049        $ret;
3050}
3051
3052sub get_commit_editor {
3053        my ($self, $log, $cb, $pool) = @_;
3054        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3055        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3056}
3057
3058sub gs_do_update {
3059        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3060        my $new = ($rev_a == $rev_b);
3061        my $path = $gs->{path};
3062
3063        if ($new && -e $gs->{index}) {
3064                unlink $gs->{index} or die
3065                  "Couldn't unlink index: $gs->{index}: $!\n";
3066        }
3067        my $pool = SVN::Pool->new;
3068        $editor->set_path_strip($path);
3069        my (@pc) = split m#/#, $path;
3070        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3071                                        1, $editor, $pool);
3072        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3073
3074        # Since we can't rely on svn_ra_reparent being available, we'll
3075        # just have to do some magic with set_path to make it so
3076        # we only want a partial path.
3077        my $sp = '';
3078        my $final = join('/', @pc);
3079        while (@pc) {
3080                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3081                $sp .= '/' if length $sp;
3082                $sp .= shift @pc;
3083        }
3084        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3085
3086        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3087
3088        $reporter->finish_report($pool);
3089        $pool->clear;
3090        $editor->{git_commit_ok};
3091}
3092
3093# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3094# svn_ra_reparent didn't work before 1.4)
3095sub gs_do_switch {
3096        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3097        my $path = $gs->{path};
3098        my $pool = SVN::Pool->new;
3099
3100        my $full_url = $self->{url};
3101        my $old_url = $full_url;
3102        $full_url .= "/$path" if length $path;
3103        my ($ra, $reparented);
3104        if ($old_url ne $full_url) {
3105                if ($old_url !~ m#^svn(\+ssh)?://#) {
3106                        SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3107                                                  $pool);
3108                        $self->{url} = $full_url;
3109                        $reparented = 1;
3110                } else {
3111                        $ra = Git::SVN::Ra->new($full_url);
3112                }
3113        }
3114        $ra ||= $self;
3115        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3116        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3117        $reporter->set_path('', $rev_a, 0, @lock, $pool);
3118        $reporter->finish_report($pool);
3119
3120        if ($reparented) {
3121                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3122                $self->{url} = $old_url;
3123        }
3124
3125        $pool->clear;
3126        $editor->{git_commit_ok};
3127}
3128
3129sub longest_common_path {
3130        my ($gsv, $globs) = @_;
3131        my %common;
3132        my $common_max = scalar @$gsv;
3133
3134        foreach my $gs (@$gsv) {
3135                my @tmp = split m#/#, $gs->{path};
3136                my $p = '';
3137                foreach (@tmp) {
3138                        $p .= length($p) ? "/$_" : $_;
3139                        $common{$p} ||= 0;
3140                        $common{$p}++;
3141                }
3142        }
3143        $globs ||= [];
3144        $common_max += scalar @$globs;
3145        foreach my $glob (@$globs) {
3146                my @tmp = split m#/#, $glob->{path}->{left};
3147                my $p = '';
3148                foreach (@tmp) {
3149                        $p .= length($p) ? "/$_" : $_;
3150                        $common{$p} ||= 0;
3151                        $common{$p}++;
3152                }
3153        }
3154
3155        my $longest_path = '';
3156        foreach (sort {length $b <=> length $a} keys %common) {
3157                if ($common{$_} == $common_max) {
3158                        $longest_path = $_;
3159                        last;
3160                }
3161        }
3162        $longest_path;
3163}
3164
3165sub gs_fetch_loop_common {
3166        my ($self, $base, $head, $gsv, $globs) = @_;
3167        return if ($base > $head);
3168        my $inc = $_log_window_size;
3169        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3170        my $longest_path = longest_common_path($gsv, $globs);
3171        while (1) {
3172                my %revs;
3173                my $err;
3174                my $err_handler = $SVN::Error::handler;
3175                $SVN::Error::handler = sub {
3176                        ($err) = @_;
3177                        skip_unknown_revs($err);
3178                };
3179                sub _cb {
3180                        my ($paths, $r, $author, $date, $log) = @_;
3181                        [ dup_changed_paths($paths),
3182                          { author => $author, date => $date, log => $log } ];
3183                }
3184                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3185                               sub { $revs{$_[1]} = _cb(@_) });
3186                if ($err && $max >= $head) {
3187                        print STDERR "Path '$longest_path' ",
3188                                     "was probably deleted:\n",
3189                                     $err->expanded_message,
3190                                     "\nWill attempt to follow ",
3191                                     "revisions r$min .. r$max ",
3192                                     "committed before the deletion\n";
3193                        my $hi = $max;
3194                        while (--$hi >= $min) {
3195                                my $ok;
3196                                $self->get_log([$longest_path], $min, $hi,
3197                                               0, 1, 1, sub {
3198                                               $ok ||= $_[1];
3199                                               $revs{$_[1]} = _cb(@_) });
3200                                if ($ok) {
3201                                        print STDERR "r$min .. r$ok OK\n";
3202                                        last;
3203                                }
3204                        }
3205                }
3206                $SVN::Error::handler = $err_handler;
3207
3208                my %exists = map { $_->{path} => $_ } @$gsv;
3209                foreach my $r (sort {$a <=> $b} keys %revs) {
3210                        my ($paths, $logged) = @{$revs{$r}};
3211
3212                        foreach my $gs ($self->match_globs(\%exists, $paths,
3213                                                           $globs, $r)) {
3214                                if ($gs->rev_db_max >= $r) {
3215                                        next;
3216                                }
3217                                next unless $gs->match_paths($paths, $r);
3218                                $gs->{logged_rev_props} = $logged;
3219                                if (my $last_commit = $gs->last_commit) {
3220                                        $gs->assert_index_clean($last_commit);
3221                                }
3222                                my $log_entry = $gs->do_fetch($paths, $r);
3223                                if ($log_entry) {
3224                                        $gs->do_git_commit($log_entry);
3225                                }
3226                        }
3227                        foreach my $g (@$globs) {
3228                                my $k = "svn-remote.$g->{remote}." .
3229                                        "$g->{t}-maxRev";
3230                                Git::SVN::tmp_config($k, $r);
3231                        }
3232                }
3233                # pre-fill the .rev_db since it'll eventually get filled in
3234                # with '0' x40 if something new gets committed
3235                foreach my $gs (@$gsv) {
3236                        next if defined $gs->rev_db_get($max);
3237                        $gs->rev_db_set($max, 0 x40);
3238                }
3239                foreach my $g (@$globs) {
3240                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3241                        Git::SVN::tmp_config($k, $max);
3242                }
3243                last if $max >= $head;
3244                $min = $max + 1;
3245                $max += $inc;
3246                $max = $head if ($max > $head);
3247        }
3248}
3249
3250sub match_globs {
3251        my ($self, $exists, $paths, $globs, $r) = @_;
3252
3253        sub get_dir_check {
3254                my ($self, $exists, $g, $r) = @_;
3255                my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3256                return unless scalar @x == 3;
3257                my $dirents = $x[0];
3258                foreach my $de (keys %$dirents) {
3259                        next if $dirents->{$de}->{kind} != $SVN::Node::dir;
3260                        my $p = $g->{path}->full_path($de);
3261                        next if $exists->{$p};
3262                        next if (length $g->{path}->{right} &&
3263                                 ($self->check_path($p, $r) !=
3264                                  $SVN::Node::dir));
3265                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3266                                         $g->{ref}->full_path($de), 1);
3267                }
3268        }
3269        foreach my $g (@$globs) {
3270                if (my $path = $paths->{"/$g->{path}->{left}"}) {
3271                        if ($path->{action} =~ /^[AR]$/) {
3272                                get_dir_check($self, $exists, $g, $r);
3273                        }
3274                }
3275                foreach (keys %$paths) {
3276                        if (/$g->{path}->{left_regex}/ &&
3277                            !/$g->{path}->{regex}/) {
3278                                next if $paths->{$_}->{action} !~ /^[AR]$/;
3279                                get_dir_check($self, $exists, $g, $r);
3280                        }
3281                        next unless /$g->{path}->{regex}/;
3282                        my $p = $1;
3283                        my $pathname = $g->{path}->full_path($p);
3284                        next if $exists->{$pathname};
3285                        next if ($self->check_path($pathname, $r) !=
3286                                 $SVN::Node::dir);
3287                        $exists->{$pathname} = Git::SVN->init(
3288                                              $self->{url}, $pathname, undef,
3289                                              $g->{ref}->full_path($p), 1);
3290                }
3291                my $c = '';
3292                foreach (split m#/#, $g->{path}->{left}) {
3293                        $c .= "/$_";
3294                        next unless ($paths->{$c} &&
3295                                     ($paths->{$c}->{action} =~ /^[AR]$/));
3296                        get_dir_check($self, $exists, $g, $r);
3297                }
3298        }
3299        values %$exists;
3300}
3301
3302sub minimize_url {
3303        my ($self) = @_;
3304        return $self->{url} if ($self->{url} eq $self->{repos_root});
3305        my $url = $self->{repos_root};
3306        my @components = split(m!/!, $self->{svn_path});
3307        my $c = '';
3308        do {
3309                $url .= "/$c" if length $c;
3310                eval { (ref $self)->new($url)->get_latest_revnum };
3311        } while ($@ && ($c = shift @components));
3312        $url;
3313}
3314
3315sub can_do_switch {
3316        my $self = shift;
3317        unless (defined $can_do_switch) {
3318                my $pool = SVN::Pool->new;
3319                my $rep = eval {
3320                        $self->do_switch(1, '', 0, $self->{url},
3321                                         SVN::Delta::Editor->new, $pool);
3322                };
3323                if ($@) {
3324                        $can_do_switch = 0;
3325                } else {
3326                        $rep->abort_report($pool);
3327                        $can_do_switch = 1;
3328                }
3329                $pool->clear;
3330        }
3331        $can_do_switch;
3332}
3333
3334sub skip_unknown_revs {
3335        my ($err) = @_;
3336        my $errno = $err->apr_err();
3337        # Maybe the branch we're tracking didn't
3338        # exist when the repo started, so it's
3339        # not an error if it doesn't, just continue
3340        #
3341        # Wonderfully consistent library, eh?
3342        # 160013 - svn:// and file://
3343        # 175002 - http(s)://
3344        # 175007 - http(s):// (this repo required authorization, too...)
3345        #   More codes may be discovered later...
3346        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3347                my $err_key = $err->expanded_message;
3348                # revision numbers change every time, filter them out
3349                $err_key =~ s/\d+/\0/g;
3350                $err_key = "$errno\0$err_key";
3351                unless ($ignored_err{$err_key}) {
3352                        warn "W: Ignoring error from SVN, path probably ",
3353                             "does not exist: ($errno): ",
3354                             $err->expanded_message,"\n";
3355                        $ignored_err{$err_key} = 1;
3356                }
3357                return;
3358        }
3359        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3360}
3361
3362# svn_log_changed_path_t objects passed to get_log are likely to be
3363# overwritten even if only the refs are copied to an external variable,
3364# so we should dup the structures in their entirety.  Using an externally
3365# passed pool (instead of our temporary and quickly cleared pool in
3366# Git::SVN::Ra) does not help matters at all...
3367sub dup_changed_paths {
3368        my ($paths) = @_;
3369        return undef unless $paths;
3370        my %ret;
3371        foreach my $p (keys %$paths) {
3372                my $i = $paths->{$p};
3373                my %s = map { $_ => $i->$_ }
3374                              qw/copyfrom_path copyfrom_rev action/;
3375                $ret{$p} = \%s;
3376        }
3377        \%ret;
3378}
3379
3380package Git::SVN::Log;
3381use strict;
3382use warnings;
3383use POSIX qw/strftime/;
3384use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3385            %rusers $show_commit $incremental/;
3386my $l_fmt;
3387
3388sub cmt_showable {
3389        my ($c) = @_;
3390        return 1 if defined $c->{r};
3391
3392        # big commit message got truncated by the 16k pretty buffer in rev-list
3393        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3394                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3395                @{$c->{l}} = ();
3396                my @log = command(qw/cat-file commit/, $c->{c});
3397
3398                # shift off the headers
3399                shift @log while ($log[0] ne '');
3400                shift @log;
3401
3402                # TODO: make $c->{l} not have a trailing newline in the future
3403                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
3404
3405                (undef, $c->{r}, undef) = ::extract_metadata(
3406                                (grep(/^git-svn-id: /, @log))[-1]);
3407        }
3408        return defined $c->{r};
3409}
3410
3411sub log_use_color {
3412        return 1 if $color;
3413        my ($dc, $dcvar);
3414        $dcvar = 'color.diff';
3415        $dc = `git-config --get $dcvar`;
3416        if ($dc eq '') {
3417                # nothing at all; fallback to "diff.color"
3418                $dcvar = 'diff.color';
3419                $dc = `git-config --get $dcvar`;
3420        }
3421        chomp($dc);
3422        if ($dc eq 'auto') {
3423                my $pc;
3424                $pc = `git-config --get color.pager`;
3425                if ($pc eq '') {
3426                        # does not have it -- fallback to pager.color
3427                        $pc = `git-config --bool --get pager.color`;
3428                }
3429                else {
3430                        $pc = `git-config --bool --get color.pager`;
3431                        if ($?) {
3432                                $pc = 'false';
3433                        }
3434                }
3435                chomp($pc);
3436                if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3437                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3438                }
3439                return 0;
3440        }
3441        return 0 if $dc eq 'never';
3442        return 1 if $dc eq 'always';
3443        chomp($dc = `git-config --bool --get $dcvar`);
3444        return ($dc eq 'true');
3445}
3446
3447sub git_svn_log_cmd {
3448        my ($r_min, $r_max, @args) = @_;
3449        my $head = 'HEAD';
3450        foreach my $x (@args) {
3451                last if $x eq '--';
3452                next unless ::verify_ref("$x^0");
3453                $head = $x;
3454                last;
3455        }
3456
3457        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
3458        $gs ||= Git::SVN->_new;
3459        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3460                   $gs->refname);
3461        push @cmd, '-r' unless $non_recursive;
3462        push @cmd, qw/--raw --name-status/ if $verbose;
3463        push @cmd, '--color' if log_use_color();
3464        return @cmd unless defined $r_max;
3465        if ($r_max == $r_min) {
3466                push @cmd, '--max-count=1';
3467                if (my $c = $gs->rev_db_get($r_max)) {
3468                        push @cmd, $c;
3469                }
3470        } else {
3471                my ($c_min, $c_max);
3472                $c_max = $gs->rev_db_get($r_max);
3473                $c_min = $gs->rev_db_get($r_min);
3474                if (defined $c_min && defined $c_max) {
3475                        if ($r_max > $r_max) {
3476                                push @cmd, "$c_min..$c_max";
3477                        } else {
3478                                push @cmd, "$c_max..$c_min";
3479                        }
3480                } elsif ($r_max > $r_min) {
3481                        push @cmd, $c_max;
3482                } else {
3483                        push @cmd, $c_min;
3484                }
3485        }
3486        return @cmd;
3487}
3488
3489# adapted from pager.c
3490sub config_pager {
3491        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3492        if (!defined $pager) {
3493                $pager = 'less';
3494        } elsif (length $pager == 0 || $pager eq 'cat') {
3495                $pager = undef;
3496        }
3497}
3498
3499sub run_pager {
3500        return unless -t *STDOUT;
3501        pipe my $rfd, my $wfd or return;
3502        defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
3503        if (!$pid) {
3504                open STDOUT, '>&', $wfd or
3505                                     ::fatal "Can't redirect to stdout: $!\n";
3506                return;
3507        }
3508        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
3509        $ENV{LESS} ||= 'FRSX';
3510        exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
3511}
3512
3513sub tz_to_s_offset {
3514        my ($tz) = @_;
3515        $tz =~ s/(\d\d)$//;
3516        return ($1 * 60) + ($tz * 3600);
3517}
3518
3519sub get_author_info {
3520        my ($dest, $author, $t, $tz) = @_;
3521        $author =~ s/(?:^\s*|\s*$)//g;
3522        $dest->{a_raw} = $author;
3523        my $au;
3524        if ($::_authors) {
3525                $au = $rusers{$author} || undef;
3526        }
3527        if (!$au) {
3528                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3529        }
3530        $dest->{t} = $t;
3531        $dest->{tz} = $tz;
3532        $dest->{a} = $au;
3533        # Date::Parse isn't in the standard Perl distro :(
3534        if ($tz =~ s/^\+//) {
3535                $t += tz_to_s_offset($tz);
3536        } elsif ($tz =~ s/^\-//) {
3537                $t -= tz_to_s_offset($tz);
3538        }
3539        $dest->{t_utc} = $t;
3540}
3541
3542sub process_commit {
3543        my ($c, $r_min, $r_max, $defer) = @_;
3544        if (defined $r_min && defined $r_max) {
3545                if ($r_min == $c->{r} && $r_min == $r_max) {
3546                        show_commit($c);
3547                        return 0;
3548                }
3549                return 1 if $r_min == $r_max;
3550                if ($r_min < $r_max) {
3551                        # we need to reverse the print order
3552                        return 0 if (defined $limit && --$limit < 0);
3553                        push @$defer, $c;
3554                        return 1;
3555                }
3556                if ($r_min != $r_max) {
3557                        return 1 if ($r_min < $c->{r});
3558                        return 1 if ($r_max > $c->{r});
3559                }
3560        }
3561        return 0 if (defined $limit && --$limit < 0);
3562        show_commit($c);
3563        return 1;
3564}
3565
3566sub show_commit {
3567        my $c = shift;
3568        if ($oneline) {
3569                my $x = "\n";
3570                if (my $l = $c->{l}) {
3571                        while ($l->[0] =~ /^\s*$/) { shift @$l }
3572                        $x = $l->[0];
3573                }
3574                $l_fmt ||= 'A' . length($c->{r});
3575                print 'r',pack($l_fmt, $c->{r}),' | ';
3576                print "$c->{c} | " if $show_commit;
3577                print $x;
3578        } else {
3579                show_commit_normal($c);
3580        }
3581}
3582
3583sub show_commit_changed_paths {
3584        my ($c) = @_;
3585        return unless $c->{changed};
3586        print "Changed paths:\n", @{$c->{changed}};
3587}
3588
3589sub show_commit_normal {
3590        my ($c) = @_;
3591        print '-' x72, "\nr$c->{r} | ";
3592        print "$c->{c} | " if $show_commit;
3593        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3594                                 localtime($c->{t_utc})), ' | ';
3595        my $nr_line = 0;
3596
3597        if (my $l = $c->{l}) {
3598                while ($l->[$#$l] eq "\n" && $#$l > 0
3599                                          && $l->[($#$l - 1)] eq "\n") {
3600                        pop @$l;
3601                }
3602                $nr_line = scalar @$l;
3603                if (!$nr_line) {
3604                        print "1 line\n\n\n";
3605                } else {
3606                        if ($nr_line == 1) {
3607                                $nr_line = '1 line';
3608                        } else {
3609                                $nr_line .= ' lines';
3610                        }
3611                        print $nr_line, "\n";
3612                        show_commit_changed_paths($c);
3613                        print "\n";
3614                        print $_ foreach @$l;
3615                }
3616        } else {
3617                print "1 line\n";
3618                show_commit_changed_paths($c);
3619                print "\n";
3620
3621        }
3622        foreach my $x (qw/raw stat diff/) {
3623                if ($c->{$x}) {
3624                        print "\n";
3625                        print $_ foreach @{$c->{$x}}
3626                }
3627        }
3628}
3629
3630sub cmd_show_log {
3631        my (@args) = @_;
3632        my ($r_min, $r_max);
3633        my $r_last = -1; # prevent dupes
3634        if (defined $TZ) {
3635                $ENV{TZ} = $TZ;
3636        } else {
3637                delete $ENV{TZ};
3638        }
3639        if (defined $::_revision) {
3640                if ($::_revision =~ /^(\d+):(\d+)$/) {
3641                        ($r_min, $r_max) = ($1, $2);
3642                } elsif ($::_revision =~ /^\d+$/) {
3643                        $r_min = $r_max = $::_revision;
3644                } else {
3645                        ::fatal "-r$::_revision is not supported, use ",
3646                                "standard \'git log\' arguments instead\n";
3647                }
3648        }
3649
3650        config_pager();
3651        @args = (git_svn_log_cmd($r_min, $r_max, @args), @args);
3652        my $log = command_output_pipe(@args);
3653        run_pager();
3654        my (@k, $c, $d, $stat);
3655        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3656        while (<$log>) {
3657                if (/^${esc_color}commit ($::sha1_short)/o) {
3658                        my $cmt = $1;
3659                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3660                                $r_last = $c->{r};
3661                                process_commit($c, $r_min, $r_max, \@k) or
3662                                                                goto out;
3663                        }
3664                        $d = undef;
3665                        $c = { c => $cmt };
3666                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3667                        get_author_info($c, $1, $2, $3);
3668                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3669                        # ignore
3670                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3671                        push @{$c->{raw}}, $_;
3672                } elsif (/^${esc_color}[ACRMDT]\t/) {
3673                        # we could add $SVN->{svn_path} here, but that requires
3674                        # remote access at the moment (repo_path_split)...
3675                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
3676                        push @{$c->{changed}}, $_;
3677                } elsif (/^${esc_color}diff /o) {
3678                        $d = 1;
3679                        push @{$c->{diff}}, $_;
3680                } elsif ($d) {
3681                        push @{$c->{diff}}, $_;
3682                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
3683                          $esc_color*[\+\-]*$esc_color$/x) {
3684                        $stat = 1;
3685                        push @{$c->{stat}}, $_;
3686                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
3687                        push @{$c->{stat}}, $_;
3688                        $stat = undef;
3689                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
3690                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3691                } elsif (s/^${esc_color}    //o) {
3692                        push @{$c->{l}}, $_;
3693                }
3694        }
3695        if ($c && defined $c->{r} && $c->{r} != $r_last) {
3696                $r_last = $c->{r};
3697                process_commit($c, $r_min, $r_max, \@k);
3698        }
3699        if (@k) {
3700                my $swap = $r_max;
3701                $r_max = $r_min;
3702                $r_min = $swap;
3703                process_commit($_, $r_min, $r_max) foreach reverse @k;
3704        }
3705out:
3706        close $log;
3707        print '-' x72,"\n" unless $incremental || $oneline;
3708}
3709
3710package Git::SVN::Migration;
3711# these version numbers do NOT correspond to actual version numbers
3712# of git nor git-svn.  They are just relative.
3713#
3714# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3715#
3716# v1 layout: .git/$id/info/url, refs/remotes/$id
3717#
3718# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3719#
3720# v3 layout: .git/svn/$id, refs/remotes/$id
3721#            - info/url may remain for backwards compatibility
3722#            - this is what we migrate up to this layout automatically,
3723#            - this will be used by git svn init on single branches
3724# v3.1 layout (auto migrated):
3725#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
3726#              for backwards compatibility
3727#
3728# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3729#            - this is only created for newly multi-init-ed
3730#              repositories.  Similar in spirit to the
3731#              --use-separate-remotes option in git-clone (now default)
3732#            - we do not automatically migrate to this (following
3733#              the example set by core git)
3734use strict;
3735use warnings;
3736use Carp qw/croak/;
3737use File::Path qw/mkpath/;
3738use File::Basename qw/dirname basename/;
3739use vars qw/$_minimize/;
3740
3741sub migrate_from_v0 {
3742        my $git_dir = $ENV{GIT_DIR};
3743        return undef unless -d $git_dir;
3744        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3745        my $migrated = 0;
3746        while (<$fh>) {
3747                chomp;
3748                my ($id, $orig_ref) = ($_, $_);
3749                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3750                next unless -f "$git_dir/$id/info/url";
3751                my $new_ref = "refs/remotes/$id";
3752                if (::verify_ref("$new_ref^0")) {
3753                        print STDERR "W: $orig_ref is probably an old ",
3754                                     "branch used by an ancient version of ",
3755                                     "git-svn.\n",
3756                                     "However, $new_ref also exists.\n",
3757                                     "We will not be able ",
3758                                     "to use this branch until this ",
3759                                     "ambiguity is resolved.\n";
3760                        next;
3761                }
3762                print STDERR "Migrating from v0 layout...\n" if !$migrated;
3763                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3764                command_noisy('update-ref', $new_ref, $orig_ref);
3765                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3766                $migrated++;
3767        }
3768        command_close_pipe($fh, $ctx);
3769        print STDERR "Done migrating from v0 layout...\n" if $migrated;
3770        $migrated;
3771}
3772
3773sub migrate_from_v1 {
3774        my $git_dir = $ENV{GIT_DIR};
3775        my $migrated = 0;
3776        return $migrated unless -d $git_dir;
3777        my $svn_dir = "$git_dir/svn";
3778
3779        # just in case somebody used 'svn' as their $id at some point...
3780        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3781
3782        print STDERR "Migrating from a git-svn v1 layout...\n";
3783        mkpath([$svn_dir]);
3784        print STDERR "Data from a previous version of git-svn exists, but\n\t",
3785                     "$svn_dir\n\t(required for this version ",
3786                     "($::VERSION) of git-svn) does not. exist\n";
3787        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3788        while (<$fh>) {
3789                my $x = $_;
3790                next unless $x =~ s#^refs/remotes/##;
3791                chomp $x;
3792                next unless -f "$git_dir/$x/info/url";
3793                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3794                next unless $u;
3795                my $dn = dirname("$git_dir/svn/$x");
3796                mkpath([$dn]) unless -d $dn;
3797                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3798                        mkpath(["$git_dir/svn/svn"]);
3799                        print STDERR " - $git_dir/$x/info => ",
3800                                        "$git_dir/svn/$x/info\n";
3801                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3802                               croak "$!: $x";
3803                        # don't worry too much about these, they probably
3804                        # don't exist with repos this old (save for index,
3805                        # and we can easily regenerate that)
3806                        foreach my $f (qw/unhandled.log index .rev_db/) {
3807                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3808                        }
3809                } else {
3810                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3811                        rename "$git_dir/$x", "$git_dir/svn/$x" or
3812                               croak "$!: $x";
3813                }
3814                $migrated++;
3815        }
3816        command_close_pipe($fh, $ctx);
3817        print STDERR "Done migrating from a git-svn v1 layout\n";
3818        $migrated;
3819}
3820
3821sub read_old_urls {
3822        my ($l_map, $pfx, $path) = @_;
3823        my @dir;
3824        foreach (<$path/*>) {
3825                if (-r "$_/info/url") {
3826                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3827                        my $ref_id = $pfx . basename $_;
3828                        my $url = ::file_to_s("$_/info/url");
3829                        $l_map->{$ref_id} = $url;
3830                } elsif (-d $_) {
3831                        push @dir, $_;
3832                }
3833        }
3834        foreach (@dir) {
3835                my $x = $_;
3836                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3837                read_old_urls($l_map, $x, $_);
3838        }
3839}
3840
3841sub migrate_from_v2 {
3842        my @cfg = command(qw/config -l/);
3843        return if grep /^svn-remote\..+\.url=/, @cfg;
3844        my %l_map;
3845        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3846        my $migrated = 0;
3847
3848        foreach my $ref_id (sort keys %l_map) {
3849                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3850                if ($@) {
3851                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3852                }
3853                $migrated++;
3854        }
3855        $migrated;
3856}
3857
3858sub minimize_connections {
3859        my $r = Git::SVN::read_all_remotes();
3860        my $new_urls = {};
3861        my $root_repos = {};
3862        foreach my $repo_id (keys %$r) {
3863                my $url = $r->{$repo_id}->{url} or next;
3864                my $fetch = $r->{$repo_id}->{fetch} or next;
3865                my $ra = Git::SVN::Ra->new($url);
3866
3867                # skip existing cases where we already connect to the root
3868                if (($ra->{url} eq $ra->{repos_root}) ||
3869                    (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3870                     $repo_id)) {
3871                        $root_repos->{$ra->{url}} = $repo_id;
3872                        next;
3873                }
3874
3875                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3876                my $root_path = $ra->{url};
3877                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3878                foreach my $path (keys %$fetch) {
3879                        my $ref_id = $fetch->{$path};
3880                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3881
3882                        # make sure we can read when connecting to
3883                        # a higher level of a repository
3884                        my ($last_rev, undef) = $gs->last_rev_commit;
3885                        if (!defined $last_rev) {
3886                                $last_rev = eval {
3887                                        $root_ra->get_latest_revnum;
3888                                };
3889                                next if $@;
3890                        }
3891                        my $new = $root_path;
3892                        $new .= length $path ? "/$path" : '';
3893                        eval {
3894                                $root_ra->get_log([$new], $last_rev, $last_rev,
3895                                                  0, 0, 1, sub { });
3896                        };
3897                        next if $@;
3898                        $new_urls->{$ra->{repos_root}}->{$new} =
3899                                { ref_id => $ref_id,
3900                                  old_repo_id => $repo_id,
3901                                  old_path => $path };
3902                }
3903        }
3904
3905        my @emptied;
3906        foreach my $url (keys %$new_urls) {
3907                # see if we can re-use an existing [svn-remote "repo_id"]
3908                # instead of creating a(n ugly) new section:
3909                my $repo_id = $root_repos->{$url} ||
3910                              Git::SVN::sanitize_remote_name($url);
3911
3912                my $fetch = $new_urls->{$url};
3913                foreach my $path (keys %$fetch) {
3914                        my $x = $fetch->{$path};
3915                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3916                        my $pfx = "svn-remote.$x->{old_repo_id}";
3917
3918                        my $old_fetch = quotemeta("$x->{old_path}:".
3919                                                  "refs/remotes/$x->{ref_id}");
3920                        command_noisy(qw/config --unset/,
3921                                      "$pfx.fetch", '^'. $old_fetch . '$');
3922                        delete $r->{$x->{old_repo_id}}->
3923                               {fetch}->{$x->{old_path}};
3924                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3925                                command_noisy(qw/config --unset/,
3926                                              "$pfx.url");
3927                                push @emptied, $x->{old_repo_id}
3928                        }
3929                }
3930        }
3931        if (@emptied) {
3932                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3933                           "$ENV{GIT_DIR}/config";
3934                print STDERR <<EOF;
3935The following [svn-remote] sections in your config file ($file) are empty
3936and can be safely removed:
3937EOF
3938                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3939        }
3940}
3941
3942sub migration_check {
3943        migrate_from_v0();
3944        migrate_from_v1();
3945        migrate_from_v2();
3946        minimize_connections() if $_minimize;
3947}
3948
3949package Git::IndexInfo;
3950use strict;
3951use warnings;
3952use Git qw/command_input_pipe command_close_pipe/;
3953
3954sub new {
3955        my ($class) = @_;
3956        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3957        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3958}
3959
3960sub remove {
3961        my ($self, $path) = @_;
3962        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3963                return ++$self->{nr};
3964        }
3965        undef;
3966}
3967
3968sub update {
3969        my ($self, $mode, $hash, $path) = @_;
3970        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3971                return ++$self->{nr};
3972        }
3973        undef;
3974}
3975
3976sub DESTROY {
3977        my ($self) = @_;
3978        command_close_pipe($self->{gui}, $self->{ctx});
3979}
3980
3981package Git::SVN::GlobSpec;
3982use strict;
3983use warnings;
3984
3985sub new {
3986        my ($class, $glob) = @_;
3987        my $re = $glob;
3988        $re =~ s!/+$!!g; # no need for trailing slashes
3989        my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
3990        my ($left, $right) = ($1, $2);
3991        if ($nr > 1) {
3992                die "Only one '*' wildcard expansion ",
3993                    "is supported (got $nr): '$glob'\n";
3994        } elsif ($nr == 0) {
3995                die "One '*' is needed for glob: '$glob'\n";
3996        }
3997        $re = quotemeta($left) . $re . quotemeta($right);
3998        if (length $left && !($left =~ s!/+$!!g)) {
3999                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4000        }
4001        if (length $right && !($right =~ s!^/+!!g)) {
4002                die "Missing leading '/' on right side of: '$glob' ($right)\n";
4003        }
4004        my $left_re = qr/^\/\Q$left\E(\/|$)/;
4005        bless { left => $left, right => $right, left_regex => $left_re,
4006                regex => qr/$re/, glob => $glob }, $class;
4007}
4008
4009sub full_path {
4010        my ($self, $path) = @_;
4011        return (length $self->{left} ? "$self->{left}/" : '') .
4012               $path . (length $self->{right} ? "/$self->{right}" : '');
4013}
4014
4015__END__
4016
4017Data structures:
4018
4019
4020$remotes = { # returned by read_all_remotes()
4021        'svn' => {
4022                # svn-remote.svn.url=https://svn.musicpd.org
4023                url => 'https://svn.musicpd.org',
4024                # svn-remote.svn.fetch=mpd/trunk:trunk
4025                fetch => {
4026                        'mpd/trunk' => 'trunk',
4027                },
4028                # svn-remote.svn.tags=mpd/tags/*:tags/*
4029                tags => {
4030                        path => {
4031                                left => 'mpd/tags',
4032                                right => '',
4033                                regex => qr!mpd/tags/([^/]+)$!,
4034                                glob => 'tags/*',
4035                        },
4036                        ref => {
4037                                left => 'tags',
4038                                right => '',
4039                                regex => qr!tags/([^/]+)$!,
4040                                glob => 'tags/*',
4041                        },
4042                }
4043        }
4044};
4045
4046$log_entry hashref as returned by libsvn_log_entry()
4047{
4048        log => 'whitespace-formatted log entry
4049',                                              # trailing newline is preserved
4050        revision => '8',                        # integer
4051        date => '2004-02-24T17:01:44.108345Z',  # commit date
4052        author => 'committer name'
4053};
4054
4055
4056# this is generated by generate_diff();
4057@mods = array of diff-index line hashes, each element represents one line
4058        of diff-index output
4059
4060diff-index line ($m hash)
4061{
4062        mode_a => first column of diff-index output, no leading ':',
4063        mode_b => second column of diff-index output,
4064        sha1_b => sha1sum of the final blob,
4065        chg => change type [MCRADT],
4066        file_a => original file name of a file (iff chg is 'C' or 'R')
4067        file_b => new/current file name of a file (any chg)
4068}
4069;
4070
4071# retval of read_url_paths{,_all}();
4072$l_map = {
4073        # repository root url
4074        'https://svn.musicpd.org' => {
4075                # repository path               # GIT_SVN_ID
4076                'mpd/trunk'             =>      'trunk',
4077                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
4078        },
4079}
4080
4081Notes:
4082        I don't trust the each() function on unless I created %hash myself
4083        because the internal iterator may not have started at base.