Andrew's git - gitweb.git/blob

50b7dcf255eacbeb6bdf25e7f420410a2f05c5d9
   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$VERSION = '@@GIT_VERSION@@';
  11
  12$ENV{GIT_DIR} ||= '.git';
  13$Git::SVN::default_repo_id = 'svn';
  14$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  15
  16$Git::SVN::Log::TZ = $ENV{TZ};
  17$ENV{TZ} = 'UTC';
  18$| = 1; # unbuffer STDOUT
  19
  20sub fatal (@) { print STDERR @_; exit 1 }
  21require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  22require SVN::Ra;
  23require SVN::Delta;
  24if ($SVN::Core::VERSION lt '1.1.0') {
  25        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
  26}
  27push @Git::SVN::Ra::ISA, 'SVN::Ra';
  28push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  29push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  30use Carp qw/croak/;
  31use IO::File qw//;
  32use File::Basename qw/dirname basename/;
  33use File::Path qw/mkpath/;
  34use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  35use IPC::Open3;
  36use Git;
  37
  38BEGIN {
  39        my $s;
  40        foreach (qw/command command_oneline command_noisy command_output_pipe
  41                    command_input_pipe command_close_pipe/) {
  42                $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
  43                      "*Git::SVN::Migration::$_ = ".
  44                      "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
  45        }
  46        eval $s;
  47}
  48
  49my ($SVN);
  50
  51my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
  52$sha1 = qr/[a-f\d]{40}/;
  53$sha1_short = qr/[a-f\d]{4,40}/;
  54my ($_stdin, $_help, $_edit,
  55        $_message, $_file,
  56        $_template, $_shared,
  57        $_version,
  58        $_merge, $_strategy, $_dry_run,
  59        $_prefix);
  60$Git::SVN::_follow_parent = 1;
  61my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  62                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  63                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  64my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  65                'authors-file|A=s' => \$_authors,
  66                'repack:i' => \$Git::SVN::_repack,
  67                'no-metadata' => \$Git::SVN::_no_metadata,
  68                'quiet|q' => \$_q,
  69                'repack-flags|repack-args|repack-opts=s' =>
  70                   \$Git::SVN::_repack_flags,
  71                %remote_opts );
  72
  73my ($_trunk, $_tags, $_branches);
  74my %multi_opts = ( 'trunk|T=s' => \$_trunk,
  75                'tags|t=s' => \$_tags,
  76                'branches|b=s' => \$_branches );
  77my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
  78my %cmt_opts = ( 'edit|e' => \$_edit,
  79                'rmdir' => \$SVN::Git::Editor::_rmdir,
  80                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
  81                'l=i' => \$SVN::Git::Editor::_rename_limit,
  82                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
  83);
  84
  85my %cmd = (
  86        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
  87                        { 'revision|r=s' => \$_revision, %fc_opts } ],
  88        init => [ \&cmd_init, "Initialize a repo for tracking" .
  89                          " (requires URL argument)",
  90                          \%init_opts ],
  91        dcommit => [ \&cmd_dcommit,
  92                     'Commit several diffs to merge with upstream',
  93                        { 'merge|m|M' => \$_merge,
  94                          'strategy|s=s' => \$_strategy,
  95                          'dry-run|n' => \$_dry_run,
  96                        %cmt_opts, %fc_opts } ],
  97        'set-tree' => [ \&cmd_set_tree,
  98                        "Set an SVN repository to a git tree-ish",
  99                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 100        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 101                        { 'revision|r=i' => \$_revision } ],
 102        'multi-init' => [ \&cmd_multi_init,
 103                        'Initialize multiple trees (like git-svnimport)',
 104                        { %multi_opts, %init_opts, %remote_opts,
 105                         'revision|r=i' => \$_revision,
 106                         'prefix=s' => \$_prefix,
 107                        } ],
 108        'multi-fetch' => [ \&cmd_multi_fetch,
 109                        'Fetch multiple trees (like git-svnimport)',
 110                        \%fc_opts ],
 111        'migrate' => [ sub { },
 112                       # no-op, we automatically run this anyways,
 113                       'Migrate configuration/metadata/layout from
 114                        previous versions of git-svn',
 115                        \%remote_opts ],
 116        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 117                        { 'limit=i' => \$Git::SVN::Log::limit,
 118                          'revision|r=s' => \$_revision,
 119                          'verbose|v' => \$Git::SVN::Log::verbose,
 120                          'incremental' => \$Git::SVN::Log::incremental,
 121                          'oneline' => \$Git::SVN::Log::oneline,
 122                          'show-commit' => \$Git::SVN::Log::show_commit,
 123                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 124                          'authors-file|A=s' => \$_authors,
 125                          'color' => \$Git::SVN::Log::color,
 126                          'pager=s' => \$Git::SVN::Log::pager,
 127                        } ],
 128        'commit-diff' => [ \&cmd_commit_diff,
 129                           'Commit a diff between two trees',
 130                        { 'message|m=s' => \$_message,
 131                          'file|F=s' => \$_file,
 132                          'revision|r=s' => \$_revision,
 133                        %cmt_opts } ],
 134);
 135
 136my $cmd;
 137for (my $i = 0; $i < @ARGV; $i++) {
 138        if (defined $cmd{$ARGV[$i]}) {
 139                $cmd = $ARGV[$i];
 140                splice @ARGV, $i, 1;
 141                last;
 142        }
 143};
 144
 145my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 146
 147read_repo_config(\%opts);
 148my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 149                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 150                    'id|i=s' => \$Git::SVN::default_ref_id,
 151                    'svn-remote|remote|R=s' => \$Git::SVN::default_repo_id);
 152exit 1 if (!$rv && $cmd ne 'log');
 153
 154usage(0) if $_help;
 155version() if $_version;
 156usage(1) unless defined $cmd;
 157load_authors() if $_authors;
 158unless ($cmd =~ /^(?:init|multi-init|commit-diff)$/) {
 159        Git::SVN::Migration::migration_check();
 160}
 161Git::SVN::init_vars();
 162eval {
 163        Git::SVN::verify_remotes_sanity();
 164        $cmd{$cmd}->[0]->(@ARGV);
 165};
 166fatal $@ if $@;
 167exit 0;
 168
 169####################### primary functions ######################
 170sub usage {
 171        my $exit = shift || 0;
 172        my $fd = $exit ? \*STDERR : \*STDOUT;
 173        print $fd <<"";
 174git-svn - bidirectional operations between a single Subversion tree and git
 175Usage: $0 <command> [options] [arguments]\n
 176
 177        print $fd "Available commands:\n" unless $cmd;
 178
 179        foreach (sort keys %cmd) {
 180                next if $cmd && $cmd ne $_;
 181                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 182                foreach (keys %{$cmd{$_}->[2]}) {
 183                        # prints out arguments as they should be passed:
 184                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 185                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 186                                                        "--$_" : "-$_" }
 187                                                split /\|/,$_)," $x\n";
 188                }
 189        }
 190        print $fd <<"";
 191\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 192arbitrary identifier if you're tracking multiple SVN branches/repositories in
 193one git repository and want to keep them separate.  See git-svn(1) for more
 194information.
 195
 196        exit $exit;
 197}
 198
 199sub version {
 200        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 201        exit 0;
 202}
 203
 204sub do_git_init_db {
 205        unless (-d $ENV{GIT_DIR}) {
 206                my @init_db = ('init');
 207                push @init_db, "--template=$_template" if defined $_template;
 208                push @init_db, "--shared" if defined $_shared;
 209                command_noisy(@init_db);
 210        }
 211}
 212
 213sub cmd_init {
 214        my $url = shift or die "SVN repository location required " .
 215                                "as a command-line argument\n";
 216        if (my $repo_path = shift) {
 217                unless (-d $repo_path) {
 218                        mkpath([$repo_path]);
 219                }
 220                chdir $repo_path or croak $!;
 221                $ENV{GIT_DIR} = $repo_path . "/.git";
 222        }
 223        do_git_init_db();
 224
 225        Git::SVN->init($url);
 226}
 227
 228sub cmd_fetch {
 229        if (@_) {
 230                die "Additional fetch arguments are no longer supported.\n",
 231                    "Use --follow-parent if you have moved/copied directories
 232                    instead.\n";
 233        }
 234        my $gs = Git::SVN->new;
 235        $gs->fetch(parse_revision_argument());
 236        if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
 237                command_noisy(qw(update-ref refs/heads/master),
 238                              $gs->{last_commit});
 239        }
 240}
 241
 242sub cmd_set_tree {
 243        my (@commits) = @_;
 244        if ($_stdin || !@commits) {
 245                print "Reading from stdin...\n";
 246                @commits = ();
 247                while (<STDIN>) {
 248                        if (/\b($sha1_short)\b/o) {
 249                                unshift @commits, $1;
 250                        }
 251                }
 252        }
 253        my @revs;
 254        foreach my $c (@commits) {
 255                my @tmp = command('rev-parse',$c);
 256                if (scalar @tmp == 1) {
 257                        push @revs, $tmp[0];
 258                } elsif (scalar @tmp > 1) {
 259                        push @revs, reverse(command('rev-list',@tmp));
 260                } else {
 261                        fatal "Failed to rev-parse $c\n";
 262                }
 263        }
 264        my $gs = Git::SVN->new;
 265        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 266        $gs->fetch;
 267        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 268                fatal "There are new revisions that were fetched ",
 269                      "and need to be merged (or acknowledged) ",
 270                      "before committing.\nlast rev: $r_last\n",
 271                      " current: $gs->{last_rev}\n";
 272        }
 273        $gs->set_tree($_) foreach @revs;
 274        print "Done committing ",scalar @revs," revisions to SVN\n";
 275}
 276
 277sub cmd_dcommit {
 278        my $head = shift;
 279        my $gs = Git::SVN->new;
 280        $head ||= 'HEAD';
 281        my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
 282        my $last_rev;
 283        foreach my $d (reverse @refs) {
 284                if (!verify_ref("$d~1")) {
 285                        fatal "Commit $d\n",
 286                              "has no parent commit, and therefore ",
 287                              "nothing to diff against.\n",
 288                              "You should be working from a repository ",
 289                              "originally created by git-svn\n";
 290                }
 291                unless (defined $last_rev) {
 292                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 293                        unless (defined $last_rev) {
 294                                fatal "Unable to extract revision information ",
 295                                      "from commit $d~1\n";
 296                        }
 297                }
 298                if ($_dry_run) {
 299                        print "diff-tree $d~1 $d\n";
 300                } else {
 301                        my %ed_opts = ( r => $last_rev,
 302                                        log => get_commit_entry($d)->{log},
 303                                        ra => $gs->ra,
 304                                        tree_a => "$d~1",
 305                                        tree_b => $d,
 306                                        editor_cb => sub {
 307                                               print "Committed r$_[0]\n";
 308                                               $last_rev = $_[0]; },
 309                                        svn_path => $gs->{path} );
 310                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 311                                print "No changes\n$d~1 == $d\n";
 312                        }
 313                }
 314        }
 315        return if $_dry_run;
 316        $gs->fetch;
 317        # we always want to rebase against the current HEAD, not any
 318        # head that was passed to us
 319        my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
 320        my @finish;
 321        if (@diff) {
 322                @finish = qw/rebase/;
 323                push @finish, qw/--merge/ if $_merge;
 324                push @finish, "--strategy=$_strategy" if $_strategy;
 325                print STDERR "W: HEAD and ", $gs->refname, " differ, ",
 326                             "using @finish:\n", "@diff";
 327        } else {
 328                print "No changes between current HEAD and ",
 329                      $gs->refname, "\nResetting to the latest ",
 330                      $gs->refname, "\n";
 331                @finish = qw/reset --mixed/;
 332        }
 333        command_noisy(@finish, $gs->refname);
 334}
 335
 336sub cmd_show_ignore {
 337        my $gs = Git::SVN->new;
 338        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 339        $gs->traverse_ignore(\*STDOUT, '', $r);
 340}
 341
 342sub cmd_multi_init {
 343        my $url = shift;
 344        unless (defined $_trunk || defined $_branches || defined $_tags) {
 345                usage(1);
 346        }
 347        do_git_init_db();
 348        $_prefix = '' unless defined $_prefix;
 349        $url =~ s#/+$## if defined $url;
 350        if (defined $_trunk) {
 351                my $trunk_ref = $_prefix . 'trunk';
 352                # try both old-style and new-style lookups:
 353                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 354                unless ($gs_trunk) {
 355                        my ($trunk_url, $trunk_path) =
 356                                              complete_svn_url($url, $_trunk);
 357                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 358                                                   undef, $trunk_ref);
 359                }
 360        }
 361        return unless defined $_branches || defined $_tags;
 362        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 363        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 364        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 365}
 366
 367sub cmd_multi_fetch {
 368        my $remotes = Git::SVN::read_all_remotes();
 369        foreach my $repo_id (sort keys %$remotes) {
 370                if ($remotes->{$repo_id}->{url} &&
 371                    $remotes->{$repo_id}->{fetch}) {
 372                        Git::SVN::fetch_all($repo_id, $remotes);
 373                }
 374        }
 375}
 376
 377# this command is special because it requires no metadata
 378sub cmd_commit_diff {
 379        my ($ta, $tb, $url) = @_;
 380        my $usage = "Usage: $0 commit-diff -r<revision> ".
 381                    "<tree-ish> <tree-ish> [<URL>]\n";
 382        fatal($usage) if (!defined $ta || !defined $tb);
 383        my $svn_path;
 384        if (!defined $url) {
 385                my $gs = eval { Git::SVN->new };
 386                if (!$gs) {
 387                        fatal("Needed URL or usable git-svn --id in ",
 388                              "the command-line\n", $usage);
 389                }
 390                $url = $gs->{url};
 391                $svn_path = $gs->{path};
 392        }
 393        unless (defined $_revision) {
 394                fatal("-r|--revision is a required argument\n", $usage);
 395        }
 396        if (defined $_message && defined $_file) {
 397                fatal("Both --message/-m and --file/-F specified ",
 398                      "for the commit message.\n",
 399                      "I have no idea what you mean\n");
 400        }
 401        if (defined $_file) {
 402                $_message = file_to_s($_file);
 403        } else {
 404                $_message ||= get_commit_entry($tb)->{log};
 405        }
 406        my $ra ||= Git::SVN::Ra->new($url);
 407        $svn_path ||= $ra->{svn_path};
 408        my $r = $_revision;
 409        if ($r eq 'HEAD') {
 410                $r = $ra->get_latest_revnum;
 411        } elsif ($r !~ /^\d+$/) {
 412                die "revision argument: $r not understood by git-svn\n";
 413        }
 414        my %ed_opts = ( r => $r,
 415                        log => $_message,
 416                        ra => $ra,
 417                        tree_a => $ta,
 418                        tree_b => $tb,
 419                        editor_cb => sub { print "Committed r$_[0]\n" },
 420                        svn_path => $svn_path );
 421        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 422                print "No changes\n$ta == $tb\n";
 423        }
 424}
 425
 426########################### utility functions #########################
 427
 428sub parse_revision_argument {
 429        if (!defined $_revision || $_revision eq 'BASE:HEAD') {
 430                return (undef, undef);
 431        }
 432        return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
 433        return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
 434        return (undef, $1) if ($_revision =~ /^BASE:(\d+)$/);
 435        return ($1, undef) if ($_revision =~ /^(\d+):HEAD$/);
 436        die "revision argument: $_revision not understood by git-svn\n",
 437            "Try using the command-line svn client instead\n";
 438}
 439
 440sub complete_svn_url {
 441        my ($url, $path) = @_;
 442        $path =~ s#/+$##;
 443        if ($path !~ m#^[a-z\+]+://#) {
 444                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 445                        fatal("E: '$path' is not a complete URL ",
 446                              "and a separate URL is not specified\n");
 447                }
 448                return ($url, $path);
 449        }
 450        return ($path, '');
 451}
 452
 453sub complete_url_ls_init {
 454        my ($ra, $repo_path, $switch, $pfx) = @_;
 455        unless ($repo_path) {
 456                print STDERR "W: $switch not specified\n";
 457                return;
 458        }
 459        $repo_path =~ s#/+$##;
 460        if ($repo_path =~ m#^[a-z\+]+://#) {
 461                $ra = Git::SVN::Ra->new($repo_path);
 462                $repo_path = '';
 463        } else {
 464                $repo_path =~ s#^/+##;
 465                unless ($ra) {
 466                        fatal("E: '$repo_path' is not a complete URL ",
 467                              "and a separate URL is not specified\n");
 468                }
 469        }
 470        my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
 471        my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
 472        my $url = $ra->{url};
 473        my $remote_id;
 474        my $remote_path;
 475        foreach my $d (sort keys %$dirent) {
 476                next if ($dirent->{$d}->kind != $SVN::Node::dir);
 477                my $path =  "$repo_path/$d";
 478                my $ref = "$pfx$d";
 479                my $gs = eval { Git::SVN->new($ref) };
 480                # don't try to init already existing refs
 481                unless ($gs) {
 482                        print "init $url/$path => $ref\n";
 483                        $gs = Git::SVN->init($url, $path, undef, $ref, 1);
 484                }
 485                if ($gs) {
 486                        $remote_id = $gs->{repo_id};
 487                        last;
 488                }
 489        }
 490        if (defined $remote_id) {
 491                $remote_path = "$ra->{svn_path}/$repo_path/*";
 492                $remote_path =~ s#/+#/#g;
 493                $remote_path =~ s#^/##g;
 494                my ($n) = ($switch =~ /^--(\w+)/);
 495                if (length $pfx && $pfx !~ m#/$#) {
 496                        die "--prefix='$pfx' must have a trailing slash '/'\n";
 497                }
 498                command_noisy('config', "svn-remote.$remote_id.$n",
 499                                        "$remote_path:refs/remotes/$pfx*");
 500        }
 501}
 502
 503sub verify_ref {
 504        my ($ref) = @_;
 505        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 506                               { STDERR => 0 }); };
 507}
 508
 509sub get_tree_from_treeish {
 510        my ($treeish) = @_;
 511        # $treeish can be a symbolic ref, too:
 512        my $type = command_oneline(qw/cat-file -t/, $treeish);
 513        my $expected;
 514        while ($type eq 'tag') {
 515                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
 516        }
 517        if ($type eq 'commit') {
 518                $expected = (grep /^tree /, command(qw/cat-file commit/,
 519                                                    $treeish))[0];
 520                ($expected) = ($expected =~ /^tree ($sha1)$/o);
 521                die "Unable to get tree from $treeish\n" unless $expected;
 522        } elsif ($type eq 'tree') {
 523                $expected = $treeish;
 524        } else {
 525                die "$treeish is a $type, expected tree, tag or commit\n";
 526        }
 527        return $expected;
 528}
 529
 530sub get_commit_entry {
 531        my ($treeish) = shift;
 532        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
 533        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
 534        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
 535        open my $log_fh, '>', $commit_editmsg or croak $!;
 536
 537        my $type = command_oneline(qw/cat-file -t/, $treeish);
 538        if ($type eq 'commit' || $type eq 'tag') {
 539                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
 540                                                         $type, $treeish);
 541                my $in_msg = 0;
 542                while (<$msg_fh>) {
 543                        if (!$in_msg) {
 544                                $in_msg = 1 if (/^\s*$/);
 545                        } elsif (/^git-svn-id: /) {
 546                                # skip this for now, we regenerate the
 547                                # correct one on re-fetch anyways
 548                                # TODO: set *:merge properties or like...
 549                        } else {
 550                                print $log_fh $_ or croak $!;
 551                        }
 552                }
 553                command_close_pipe($msg_fh, $ctx);
 554        }
 555        close $log_fh or croak $!;
 556
 557        if ($_edit || ($type eq 'tree')) {
 558                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 559                # TODO: strip out spaces, comments, like git-commit.sh
 560                system($editor, $commit_editmsg);
 561        }
 562        rename $commit_editmsg, $commit_msg or croak $!;
 563        open $log_fh, '<', $commit_msg or croak $!;
 564        { local $/; chomp($log_entry{log} = <$log_fh>); }
 565        close $log_fh or croak $!;
 566        unlink $commit_msg;
 567        \%log_entry;
 568}
 569
 570sub s_to_file {
 571        my ($str, $file, $mode) = @_;
 572        open my $fd,'>',$file or croak $!;
 573        print $fd $str,"\n" or croak $!;
 574        close $fd or croak $!;
 575        chmod ($mode &~ umask, $file) if (defined $mode);
 576}
 577
 578sub file_to_s {
 579        my $file = shift;
 580        open my $fd,'<',$file or croak "$!: file: $file\n";
 581        local $/;
 582        my $ret = <$fd>;
 583        close $fd or croak $!;
 584        $ret =~ s/\s*$//s;
 585        return $ret;
 586}
 587
 588# '<svn username> = real-name <email address>' mapping based on git-svnimport:
 589sub load_authors {
 590        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
 591        my $log = $cmd eq 'log';
 592        while (<$authors>) {
 593                chomp;
 594                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
 595                my ($user, $name, $email) = ($1, $2, $3);
 596                if ($log) {
 597                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
 598                } else {
 599                        $users{$user} = [$name, $email];
 600                }
 601        }
 602        close $authors or croak $!;
 603}
 604
 605# convert GetOpt::Long specs for use by git-config
 606sub read_repo_config {
 607        return unless -d $ENV{GIT_DIR};
 608        my $opts = shift;
 609        foreach my $o (keys %$opts) {
 610                my $v = $opts->{$o};
 611                my ($key) = ($o =~ /^([a-z\-]+)/);
 612                $key =~ s/-//g;
 613                my $arg = 'git-config';
 614                $arg .= ' --int' if ($o =~ /[:=]i$/);
 615                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
 616                if (ref $v eq 'ARRAY') {
 617                        chomp(my @tmp = `$arg --get-all svn.$key`);
 618                        @$v = @tmp if @tmp;
 619                } else {
 620                        chomp(my $tmp = `$arg --get svn.$key`);
 621                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
 622                                $$v = $tmp;
 623                        }
 624                }
 625        }
 626}
 627
 628sub extract_metadata {
 629        my $id = shift or return (undef, undef, undef);
 630        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 631                                                        \s([a-f\d\-]+)$/x);
 632        if (!defined $rev || !$uuid || !$url) {
 633                # some of the original repositories I made had
 634                # identifiers like this:
 635                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
 636        }
 637        return ($url, $rev, $uuid);
 638}
 639
 640sub cmt_metadata {
 641        return extract_metadata((grep(/^git-svn-id: /,
 642                command(qw/cat-file commit/, shift)))[-1]);
 643}
 644
 645package Git::SVN;
 646use strict;
 647use warnings;
 648use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
 649            $_repack $_repack_flags/;
 650use Carp qw/croak/;
 651use File::Path qw/mkpath/;
 652use File::Copy qw/copy/;
 653use IPC::Open3;
 654
 655my $_repack_nr;
 656# properties that we do not log:
 657my %SKIP_PROP;
 658BEGIN {
 659        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
 660                                        svn:special svn:executable
 661                                        svn:entry:committed-rev
 662                                        svn:entry:last-author
 663                                        svn:entry:uuid
 664                                        svn:entry:committed-date/;
 665}
 666
 667my %LOCKFILES;
 668END { unlink keys %LOCKFILES if %LOCKFILES }
 669
 670sub resolve_local_globs {
 671        my ($url, $fetch, $glob_spec) = @_;
 672        return unless defined $glob_spec;
 673        my $ref = $glob_spec->{ref};
 674        my $path = $glob_spec->{path};
 675        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
 676                next unless m#^refs/remotes/$ref->{regex}$#;
 677                my $p = $1;
 678                my $pathname = $path->full_path($p);
 679                my $refname = $ref->full_path($p);
 680                if (my $existing = $fetch->{$pathname}) {
 681                        if ($existing ne $refname) {
 682                                die "Refspec conflict:\n",
 683                                    "existing: refs/remotes/$existing\n",
 684                                    " globbed: refs/remotes/$refname\n";
 685                        }
 686                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
 687                        $u =~ s!^\Q$url\E(/|$)!! or die
 688                          "refs/remotes/$refname: '$url' not found in '$u'\n";
 689                        if ($pathname ne $u) {
 690                                warn "W: Refspec glob conflict ",
 691                                     "(ref: refs/remotes/$refname):\n",
 692                                     "expected path: $pathname\n",
 693                                     "    real path: $u\n",
 694                                     "Continuing ahead with $u\n";
 695                                next;
 696                        }
 697                } else {
 698                        $fetch->{$pathname} = $refname;
 699                }
 700        }
 701}
 702
 703sub fetch_all {
 704        my ($repo_id, $remotes) = @_;
 705        my $remote = $remotes->{$repo_id};
 706        my $fetch = $remote->{fetch};
 707        my $url = $remote->{url};
 708        my (@gs, @globs);
 709        my $ra = Git::SVN::Ra->new($url);
 710        my $uuid = $ra->uuid;
 711        my $head = $ra->get_latest_revnum;
 712        my $base = $head;
 713
 714        # read the max revs for wildcard expansion (branches/*, tags/*)
 715        foreach my $t (qw/branches tags/) {
 716                defined $remote->{$t} or next;
 717                push @globs, $remote->{$t};
 718                my $f = "$ENV{GIT_DIR}/svn/.$uuid.$t";
 719                if (open my $fh, '<', $f) {
 720                        chomp(my $max_rev = <$fh>);
 721                        close $fh or die "Error closing $f: $!\n";
 722
 723                        if ($max_rev !~ /^\d+$/) {
 724                                die "$max_rev (in $f) is not an integer!\n";
 725                        }
 726                        $remote->{$t}->{max_rev} = $max_rev;
 727                        $base = $max_rev if ($max_rev < $base);
 728                }
 729        }
 730
 731        foreach my $p (sort keys %$fetch) {
 732                my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
 733                my $lr = $gs->rev_db_max;
 734                if (defined $lr) {
 735                        $base = $lr if ($lr < $base);
 736                }
 737                push @gs, $gs;
 738        }
 739        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
 740}
 741
 742sub read_all_remotes {
 743        my $r = {};
 744        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
 745                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
 746                        $r->{$1}->{fetch}->{$2} = $3;
 747                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
 748                        $r->{$1}->{url} = $2;
 749                } elsif (m!^(.+)\.(branches|tags)=
 750                           (.*):refs/remotes/(.+)\s*$/!x) {
 751                        my ($p, $g) = ($3, $4);
 752                        my $rs = $r->{$1}->{$2} = {
 753                                          t => $2,
 754                                          path => Git::SVN::GlobSpec->new($p),
 755                                          ref => Git::SVN::GlobSpec->new($g) };
 756                        if (length($rs->{ref}->{right}) != 0) {
 757                                die "The '*' glob character must be the last ",
 758                                    "character of '$g'\n";
 759                        }
 760                }
 761        }
 762        $r;
 763}
 764
 765sub init_vars {
 766        if (defined $_repack) {
 767                $_repack = 1000 if ($_repack <= 0);
 768                $_repack_nr = $_repack;
 769                $_repack_flags ||= '-d';
 770        }
 771}
 772
 773sub verify_remotes_sanity {
 774        return unless -d $ENV{GIT_DIR};
 775        my %seen;
 776        foreach (command(qw/config -l/)) {
 777                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
 778                        if ($seen{$1}) {
 779                                die "Remote ref refs/remote/$1 is tracked by",
 780                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
 781                                    "Please resolve this ambiguity in ",
 782                                    "your git configuration file before ",
 783                                    "continuing\n";
 784                        }
 785                        $seen{$1} = $_;
 786                }
 787        }
 788}
 789
 790# we allow more chars than remotes2config.sh...
 791sub sanitize_remote_name {
 792        my ($name) = @_;
 793        $name =~ tr{A-Za-z0-9:,/+-}{.}c;
 794        $name;
 795}
 796
 797sub find_existing_remote {
 798        my ($url, $remotes) = @_;
 799        my $existing;
 800        foreach my $repo_id (keys %$remotes) {
 801                my $u = $remotes->{$repo_id}->{url} or next;
 802                next if $u ne $url;
 803                $existing = $repo_id;
 804                last;
 805        }
 806        $existing;
 807}
 808
 809sub init_remote_config {
 810        my ($self, $url, $no_write) = @_;
 811        $url =~ s!/+$!!; # strip trailing slash
 812        my $r = read_all_remotes();
 813        my $existing = find_existing_remote($url, $r);
 814        if ($existing) {
 815                unless ($no_write) {
 816                        print STDERR "Using existing ",
 817                                     "[svn-remote \"$existing\"]\n";
 818                }
 819                $self->{repo_id} = $existing;
 820        } else {
 821                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
 822                $existing = find_existing_remote($min_url, $r);
 823                if ($existing) {
 824                        unless ($no_write) {
 825                                print STDERR "Using existing ",
 826                                             "[svn-remote \"$existing\"]\n";
 827                        }
 828                        $self->{repo_id} = $existing;
 829                }
 830                if ($min_url ne $url) {
 831                        unless ($no_write) {
 832                                print STDERR "Using higher level of URL: ",
 833                                             "$url => $min_url\n";
 834                        }
 835                        my $old_path = $self->{path};
 836                        $self->{path} = $url;
 837                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
 838                        if (length $old_path) {
 839                                $self->{path} .= "/$old_path";
 840                        }
 841                        $url = $min_url;
 842                }
 843        }
 844        my $orig_url;
 845        if (!$existing) {
 846                # verify that we aren't overwriting anything:
 847                $orig_url = eval {
 848                        command_oneline('config', '--get',
 849                                        "svn-remote.$self->{repo_id}.url")
 850                };
 851                if ($orig_url && ($orig_url ne $url)) {
 852                        die "svn-remote.$self->{repo_id}.url already set: ",
 853                            "$orig_url\nwanted to set to: $url\n";
 854                }
 855        }
 856        my ($xrepo_id, $xpath) = find_ref($self->refname);
 857        if (defined $xpath) {
 858                die "svn-remote.$xrepo_id.fetch already set to track ",
 859                    "$xpath:refs/remotes/", $self->refname, "\n";
 860        }
 861        unless ($no_write) {
 862                command_noisy('config',
 863                              "svn-remote.$self->{repo_id}.url", $url);
 864                command_noisy('config', '--add',
 865                              "svn-remote.$self->{repo_id}.fetch",
 866                              "$self->{path}:".$self->refname);
 867        }
 868        $self->{url} = $url;
 869}
 870
 871sub init {
 872        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
 873        my $self = _new($class, $repo_id, $ref_id, $path);
 874        if (defined $url) {
 875                $self->init_remote_config($url, $no_write);
 876        }
 877        $self;
 878}
 879
 880sub find_ref {
 881        my ($ref_id) = @_;
 882        foreach (command(qw/config -l/)) {
 883                next unless m!^svn-remote\.(.+)\.fetch=
 884                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
 885                my ($repo_id, $path, $ref) = ($1, $2, $3);
 886                if ($ref eq $ref_id) {
 887                        $path = '' if ($path =~ m#^\./?#);
 888                        return ($repo_id, $path);
 889                }
 890        }
 891        (undef, undef, undef);
 892}
 893
 894sub new {
 895        my ($class, $ref_id, $repo_id, $path) = @_;
 896        if (defined $ref_id && !defined $repo_id && !defined $path) {
 897                ($repo_id, $path) = find_ref($ref_id);
 898                if (!defined $repo_id) {
 899                        die "Could not find a \"svn-remote.*.fetch\" key ",
 900                            "in the repository configuration matching: ",
 901                            "refs/remotes/$ref_id\n";
 902                }
 903        }
 904        my $self = _new($class, $repo_id, $ref_id, $path);
 905        if (!defined $self->{path} || !length $self->{path}) {
 906                my $fetch = command_oneline('config', '--get',
 907                                            "svn-remote.$repo_id.fetch",
 908                                            ":refs/remotes/$ref_id\$") or
 909                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
 910                         "\":refs/remotes/$ref_id\$\" in config\n";
 911                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
 912        }
 913        $self->{url} = command_oneline('config', '--get',
 914                                       "svn-remote.$repo_id.url") or
 915                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
 916        if (-z $self->{db_path} && ::verify_ref($self->refname.'^0')) {
 917                $self->rebuild;
 918        }
 919        $self;
 920}
 921
 922sub refname { "refs/remotes/$_[0]->{ref_id}" }
 923
 924sub ra {
 925        my ($self) = shift;
 926        Git::SVN::Ra->new($self->{url});
 927}
 928
 929sub rel_path {
 930        my ($self) = @_;
 931        my $repos_root = $self->ra->{repos_root};
 932        return $self->{path} if ($self->{url} eq $repos_root);
 933        die "BUG: rel_path failed! repos_root: $repos_root, Ra URL: ",
 934            $self->ra->{url}, " path: $self->{path},  URL: $self->{url}\n";
 935}
 936
 937sub traverse_ignore {
 938        my ($self, $fh, $path, $r) = @_;
 939        $path =~ s#^/+##g;
 940        my $ra = $self->ra;
 941        my ($dirent, undef, $props) = $ra->get_dir($path, $r);
 942        my $p = $path;
 943        $p =~ s#^\Q$ra->{svn_path}\E/##;
 944        print $fh length $p ? "\n# $p\n" : "\n# /\n";
 945        if (my $s = $props->{'svn:ignore'}) {
 946                $s =~ s/[\r\n]+/\n/g;
 947                chomp $s;
 948                if (length $p == 0) {
 949                        $s =~ s#\n#\n/$p#g;
 950                        print $fh "/$s\n";
 951                } else {
 952                        $s =~ s#\n#\n/$p/#g;
 953                        print $fh "/$p/$s\n";
 954                }
 955        }
 956        foreach (sort keys %$dirent) {
 957                next if $dirent->{$_}->kind != $SVN::Node::dir;
 958                $self->traverse_ignore($fh, "$path/$_", $r);
 959        }
 960}
 961
 962sub last_rev { ($_[0]->last_rev_commit)[0] }
 963sub last_commit { ($_[0]->last_rev_commit)[1] }
 964
 965# returns the newest SVN revision number and newest commit SHA1
 966sub last_rev_commit {
 967        my ($self) = @_;
 968        if (defined $self->{last_rev} && defined $self->{last_commit}) {
 969                return ($self->{last_rev}, $self->{last_commit});
 970        }
 971        my $c = ::verify_ref($self->refname.'^0');
 972        if ($c) {
 973                my $rev = (::cmt_metadata($c))[1];
 974                if (defined $rev) {
 975                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
 976                        return ($rev, $c);
 977                }
 978        }
 979        my $offset = -41; # from tail
 980        my $rl;
 981        open my $fh, '<', $self->{db_path} or
 982                                 croak "$self->{db_path} not readable: $!\n";
 983        sysseek($fh, $offset, 2); # don't care for errors
 984        sysread($fh, $rl, 41) == 41 or return (undef, undef);
 985        chomp $rl;
 986        while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
 987                $offset -= 41;
 988                sysseek($fh, $offset, 2); # don't care for errors
 989                sysread($fh, $rl, 41) == 41 or return (undef, undef);
 990                chomp $rl;
 991        }
 992        if ($c) {
 993                die "$self->{db_path} and ", $self->refname,
 994                    " inconsistent!:\n$c != $rl\n";
 995        }
 996        my $rev = sysseek($fh, 0, 1) or croak $!;
 997        $rev =  ($rev - 41) / 41;
 998        close $fh or croak $!;
 999        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1000        return ($rev, $c);
1001}
1002
1003sub get_fetch_range {
1004        my ($self, $min, $max) = @_;
1005        $max ||= $self->ra->get_latest_revnum;
1006        $min ||= $self->rev_db_max;
1007        (++$min, $max);
1008}
1009
1010sub tmp_index_do {
1011        my ($self, $sub) = @_;
1012        my $old_index = $ENV{GIT_INDEX_FILE};
1013        $ENV{GIT_INDEX_FILE} = $self->{index};
1014        my @ret = &$sub;
1015        if ($old_index) {
1016                $ENV{GIT_INDEX_FILE} = $old_index;
1017        } else {
1018                delete $ENV{GIT_INDEX_FILE};
1019        }
1020        wantarray ? @ret : $ret[0];
1021}
1022
1023sub assert_index_clean {
1024        my ($self, $treeish) = @_;
1025
1026        $self->tmp_index_do(sub {
1027                command_noisy('read-tree', $treeish) unless -e $self->{index};
1028                my $x = command_oneline('write-tree');
1029                my ($y) = (command(qw/cat-file commit/, $treeish) =~
1030                           /^tree ($::sha1)/mo);
1031                if ($y ne $x) {
1032                        unlink $self->{index} or croak $!;
1033                        command_noisy('read-tree', $treeish);
1034                }
1035                $x = command_oneline('write-tree');
1036                if ($y ne $x) {
1037                        ::fatal "trees ($treeish) $y != $x\n",
1038                                "Something is seriously wrong...\n";
1039                }
1040        });
1041}
1042
1043sub get_commit_parents {
1044        my ($self, $log_entry) = @_;
1045        my (%seen, @ret, @tmp);
1046        # legacy support for 'set-tree'; this is only used by set_tree_cb:
1047        if (my $ip = $self->{inject_parents}) {
1048                if (my $commit = delete $ip->{$log_entry->{revision}}) {
1049                        push @tmp, $commit;
1050                }
1051        }
1052        if (my $cur = ::verify_ref($self->refname.'^0')) {
1053                push @tmp, $cur;
1054        }
1055        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1056        while (my $p = shift @tmp) {
1057                next if $seen{$p};
1058                $seen{$p} = 1;
1059                push @ret, $p;
1060                # MAXPARENT is defined to 16 in commit-tree.c:
1061                last if @ret >= 16;
1062        }
1063        if (@tmp) {
1064                die "r$log_entry->{revision}: No room for parents:\n\t",
1065                    join("\n\t", @tmp), "\n";
1066        }
1067        @ret;
1068}
1069
1070sub full_url {
1071        my ($self) = @_;
1072        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1073}
1074
1075sub do_git_commit {
1076        my ($self, $log_entry) = @_;
1077        my $lr = $self->last_rev;
1078        if (defined $lr && $lr >= $log_entry->{revision}) {
1079                die "Last fetched revision of ", $self->refname,
1080                    " was r$lr, but we are about to fetch: ",
1081                    "r$log_entry->{revision}!\n";
1082        }
1083        if (my $c = $self->rev_db_get($log_entry->{revision})) {
1084                croak "$log_entry->{revision} = $c already exists! ",
1085                      "Why are we refetching it?\n";
1086        }
1087        my $author = $log_entry->{author};
1088        my ($name, $email) = (defined $::users{$author} ? @{$::users{$author}}
1089                           : ($author, "$author\@".$self->ra->uuid));
1090        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1091        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1092        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1093
1094        my $tree = $log_entry->{tree};
1095        if (!defined $tree) {
1096                $tree = $self->tmp_index_do(sub {
1097                                            command_oneline('write-tree') });
1098        }
1099        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1100
1101        my @exec = ('git-commit-tree', $tree);
1102        foreach ($self->get_commit_parents($log_entry)) {
1103                push @exec, '-p', $_;
1104        }
1105        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1106                                                                   or croak $!;
1107        print $msg_fh $log_entry->{log} or croak $!;
1108        unless ($_no_metadata) {
1109                print $msg_fh "\ngit-svn-id: ", $self->full_url, '@',
1110                              $log_entry->{revision}, ' ',
1111                              $self->ra->uuid, "\n" or croak $!;
1112        }
1113        $msg_fh->flush == 0 or croak $!;
1114        close $msg_fh or croak $!;
1115        chomp(my $commit = do { local $/; <$out_fh> });
1116        close $out_fh or croak $!;
1117        waitpid $pid, 0;
1118        croak $? if $?;
1119        if ($commit !~ /^$::sha1$/o) {
1120                die "Failed to commit, invalid sha1: $commit\n";
1121        }
1122
1123        $self->rev_db_set($log_entry->{revision}, $commit, 1);
1124
1125        $self->{last_rev} = $log_entry->{revision};
1126        $self->{last_commit} = $commit;
1127        print "r$log_entry->{revision} = $commit ($self->{ref_id})\n";
1128        if (defined $_repack && (--$_repack_nr == 0)) {
1129                $_repack_nr = $_repack;
1130                # repack doesn't use any arguments with spaces in them, does it?
1131                print "Running git repack $_repack_flags ...\n";
1132                command_noisy('repack', split(/\s+/, $_repack_flags));
1133                print "Done repacking\n";
1134        }
1135        return $commit;
1136}
1137
1138sub match_paths {
1139        my ($self, $paths, $r) = @_;
1140        return 1 if $self->{path} eq '';
1141        if (my $path = $paths->{"/$self->{path}"}) {
1142                return ($path->{action} eq 'D') ? 0 : 1;
1143        }
1144        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1145        if (grep /$self->{path_regex}/, keys %$paths) {
1146                return 1;
1147        }
1148        my $c = '';
1149        foreach (split m#/#, $self->{path}) {
1150                $c .= "/$_";
1151                next unless ($paths->{$c} &&
1152                             ($paths->{$c}->{action} =~ /^[AR]$/));
1153                if ($self->ra->check_path($self->{path}, $r) ==
1154                    $SVN::Node::dir) {
1155                        return 1;
1156                }
1157        }
1158        return 0;
1159}
1160
1161sub find_parent_branch {
1162        my ($self, $paths, $rev) = @_;
1163        return undef unless $_follow_parent;
1164        unless (defined $paths) {
1165                my $err_handler = $SVN::Error::handler;
1166                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1167                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1168                                   $paths =
1169                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
1170                $SVN::Error::handler = $err_handler;
1171        }
1172        return undef unless defined $paths;
1173
1174        # look for a parent from another branch:
1175        my @b_path_components = split m#/#, $self->rel_path;
1176        my @a_path_components;
1177        my $i;
1178        while (@b_path_components) {
1179                $i = $paths->{'/'.join('/', @b_path_components)};
1180                last if $i && defined $i->{copyfrom_path};
1181                unshift(@a_path_components, pop(@b_path_components));
1182        }
1183        return undef unless defined $i && defined $i->{copyfrom_path};
1184        my $branch_from = $i->{copyfrom_path};
1185        if (@a_path_components) {
1186                print STDERR "branch_from: $branch_from => ";
1187                $branch_from .= '/'.join('/', @a_path_components);
1188                print STDERR $branch_from, "\n";
1189        }
1190        my $r = $i->{copyfrom_rev};
1191        my $repos_root = $self->ra->{repos_root};
1192        my $url = $self->ra->{url};
1193        my $new_url = $repos_root . $branch_from;
1194        print STDERR  "Found possible branch point: ",
1195                      "$new_url => ", $self->full_url, ", $r\n";
1196        $branch_from =~ s#^/##;
1197        my $remotes = read_all_remotes();
1198        my $gs;
1199        foreach my $repo_id (keys %$remotes) {
1200                my $u = $remotes->{$repo_id}->{url} or next;
1201                next if $url ne $u;
1202                my $fetch = $remotes->{$repo_id}->{fetch};
1203                foreach (qw/branches tags/) {
1204                        resolve_local_globs($url, $fetch,
1205                                            $remotes->{$repo_id}->{$_});
1206                }
1207                foreach my $f (keys %$fetch) {
1208                        next if $f ne $branch_from;
1209                        $gs = Git::SVN->new($fetch->{$f}, $repo_id, $f);
1210                        last;
1211                }
1212                last if $gs;
1213        }
1214        unless ($gs) {
1215                my $ref_id = $self->{ref_id};
1216                $ref_id =~ s/\@\d+$//;
1217                $ref_id .= "\@$r";
1218                # just grow a tail if we're not unique enough :x
1219                $ref_id .= '-' while find_ref($ref_id);
1220                print STDERR "Initializing parent: $ref_id\n";
1221                $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1222        }
1223        my ($r0, $parent) = $gs->find_rev_before($r, 1);
1224        if ($_follow_parent && (!defined $r0 || !defined $parent)) {
1225                $gs->fetch(0, $r);
1226                ($r0, $parent) = $gs->last_rev_commit;
1227        }
1228        if (defined $r0 && defined $parent) {
1229                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1230                $self->assert_index_clean($parent);
1231                my $ed;
1232                if ($self->ra->can_do_switch) {
1233                        print STDERR "Following parent with do_switch\n";
1234                        # do_switch works with svn/trunk >= r22312, but that
1235                        # is not included with SVN 1.4.3 (the latest version
1236                        # at the moment), so we can't rely on it
1237                        $self->{last_commit} = $parent;
1238                        $ed = SVN::Git::Fetcher->new($self);
1239                        $gs->ra->gs_do_switch($r0, $rev, $gs,
1240                                              $self->full_url, $ed)
1241                          or die "SVN connection failed somewhere...\n";
1242                } else {
1243                        print STDERR "Following parent with do_update\n";
1244                        $ed = SVN::Git::Fetcher->new($self);
1245                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
1246                          or die "SVN connection failed somewhere...\n";
1247                }
1248                print STDERR "Successfully followed parent\n";
1249                return $self->make_log_entry($rev, [$parent], $ed);
1250        }
1251        return undef;
1252}
1253
1254sub do_fetch {
1255        my ($self, $paths, $rev) = @_;
1256        my $ed;
1257        my ($last_rev, @parents);
1258        if (my $lc = $self->last_commit) {
1259                # we can have a branch that was deleted, then re-added
1260                # under the same name but copied from another path, in
1261                # which case we'll have multiple parents (we don't
1262                # want to break the original ref, nor lose copypath info):
1263                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1264                        push @{$log_entry->{parents}}, $lc;
1265                        return $log_entry;
1266                }
1267                $ed = SVN::Git::Fetcher->new($self);
1268                $last_rev = $self->{last_rev};
1269                $ed->{c} = $lc;
1270                @parents = ($lc);
1271        } else {
1272                $last_rev = $rev;
1273                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1274                        return $log_entry;
1275                }
1276                $ed = SVN::Git::Fetcher->new($self);
1277        }
1278        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1279                die "SVN connection failed somewhere...\n";
1280        }
1281        $self->make_log_entry($rev, \@parents, $ed);
1282}
1283
1284sub get_untracked {
1285        my ($self, $ed) = @_;
1286        my @out;
1287        my $h = $ed->{empty};
1288        foreach (sort keys %$h) {
1289                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1290                push @out, "  $act: " . uri_encode($_);
1291                warn "W: $act: $_\n";
1292        }
1293        foreach my $t (qw/dir_prop file_prop/) {
1294                $h = $ed->{$t} or next;
1295                foreach my $path (sort keys %$h) {
1296                        my $ppath = $path eq '' ? '.' : $path;
1297                        foreach my $prop (sort keys %{$h->{$path}}) {
1298                                next if $SKIP_PROP{$prop};
1299                                my $v = $h->{$path}->{$prop};
1300                                my $t_ppath_prop = "$t: " .
1301                                                    uri_encode($ppath) . ' ' .
1302                                                    uri_encode($prop);
1303                                if (defined $v) {
1304                                        push @out, "  +$t_ppath_prop " .
1305                                                   uri_encode($v);
1306                                } else {
1307                                        push @out, "  -$t_ppath_prop";
1308                                }
1309                        }
1310                }
1311        }
1312        foreach my $t (qw/absent_file absent_directory/) {
1313                $h = $ed->{$t} or next;
1314                foreach my $parent (sort keys %$h) {
1315                        foreach my $path (sort @{$h->{$parent}}) {
1316                                push @out, "  $t: " .
1317                                           uri_encode("$parent/$path");
1318                                warn "W: $t: $parent/$path ",
1319                                     "Insufficient permissions?\n";
1320                        }
1321                }
1322        }
1323        \@out;
1324}
1325
1326sub parse_svn_date {
1327        my $date = shift || return '+0000 1970-01-01 00:00:00';
1328        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1329                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1330                                         croak "Unable to parse date: $date\n";
1331        "+0000 $Y-$m-$d $H:$M:$S";
1332}
1333
1334sub check_author {
1335        my ($author) = @_;
1336        if (!defined $author || length $author == 0) {
1337                $author = '(no author)';
1338        }
1339        if (defined $::_authors && ! defined $::users{$author}) {
1340                die "Author: $author not defined in $::_authors file\n";
1341        }
1342        $author;
1343}
1344
1345sub make_log_entry {
1346        my ($self, $rev, $parents, $ed) = @_;
1347        my $untracked = $self->get_untracked($ed);
1348
1349        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1350        print $un "r$rev\n" or croak $!;
1351        print $un $_, "\n" foreach @$untracked;
1352        my %log_entry = ( parents => $parents || [], revision => $rev,
1353                          log => '');
1354
1355        my $logged = delete $self->{logged_rev_props};
1356        if (!$logged || $self->{-want_extra_revprops}) {
1357                my $rp = $self->ra->rev_proplist($rev);
1358                foreach (sort keys %$rp) {
1359                        my $v = $rp->{$_};
1360                        if (/^svn:(author|date|log)$/) {
1361                                $log_entry{$1} = $v;
1362                        } else {
1363                                print $un "  rev_prop: ", uri_encode($_), ' ',
1364                                          uri_encode($v), "\n";
1365                        }
1366                }
1367        } else {
1368                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1369        }
1370        close $un or croak $!;
1371
1372        $log_entry{date} = parse_svn_date($log_entry{date});
1373        $log_entry{author} = check_author($log_entry{author});
1374        $log_entry{log} .= "\n";
1375        \%log_entry;
1376}
1377
1378sub fetch {
1379        my ($self, $min_rev, $max_rev, @parents) = @_;
1380        my ($last_rev, $last_commit) = $self->last_rev_commit;
1381        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1382        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1383}
1384
1385sub set_tree_cb {
1386        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1387        # TODO: enable and test optimized commits:
1388        if (0 && $rev == ($self->{last_rev} + 1)) {
1389                $log_entry->{revision} = $rev;
1390                $log_entry->{author} = $author;
1391                $self->do_git_commit($log_entry, "$rev=$tree");
1392        } else {
1393                $self->{inject_parents} = { $rev => $tree };
1394                $self->fetch(undef, undef);
1395        }
1396}
1397
1398sub set_tree {
1399        my ($self, $tree) = (shift, shift);
1400        my $log_entry = ::get_commit_entry($tree);
1401        unless ($self->{last_rev}) {
1402                fatal("Must have an existing revision to commit\n");
1403        }
1404        my %ed_opts = ( r => $self->{last_rev},
1405                        log => $log_entry->{log},
1406                        ra => $self->ra,
1407                        tree_a => $self->{last_commit},
1408                        tree_b => $tree,
1409                        editor_cb => sub {
1410                               $self->set_tree_cb($log_entry, $tree, @_) },
1411                        svn_path => $self->{path} );
1412        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1413                print "No changes\nr$self->{last_rev} = $tree\n";
1414        }
1415}
1416
1417sub rebuild {
1418        my ($self) = @_;
1419        print "Rebuilding $self->{db_path} ...\n";
1420        my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1421        my $latest;
1422        my $full_url = $self->full_url;
1423        my $svn_uuid;
1424        while (<$rev_list>) {
1425                chomp;
1426                my $c = $_;
1427                die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1428                my ($url, $rev, $uuid) = ::cmt_metadata($c);
1429
1430                # ignore merges (from set-tree)
1431                next if (!defined $rev || !$uuid);
1432
1433                # if we merged or otherwise started elsewhere, this is
1434                # how we break out of it
1435                if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1436                    ($full_url && $url && ($url ne $full_url))) {
1437                        next;
1438                }
1439                $latest ||= $rev;
1440                $svn_uuid ||= $uuid;
1441
1442                $self->rev_db_set($rev, $c);
1443                print "r$rev = $c\n";
1444        }
1445        command_close_pipe($rev_list, $ctx);
1446        print "Done rebuilding $self->{db_path}\n";
1447}
1448
1449# rev_db:
1450# Tie::File seems to be prone to offset errors if revisions get sparse,
1451# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
1452# one of my favorite modules is out :<  Next up would be one of the DBM
1453# modules, but I'm not sure which is most portable...  So I'll just
1454# go with something that's plain-text, but still capable of
1455# being randomly accessed.  So here's my ultra-simple fixed-width
1456# database.  All records are 40 characters + "\n", so it's easy to seek
1457# to a revision: (41 * rev) is the byte offset.
1458# A record of 40 0s denotes an empty revision.
1459# And yes, it's still pretty fast (faster than Tie::File).
1460# These files are disposable unless --no-metadata is set
1461
1462sub rev_db_set {
1463        my ($self, $rev, $commit, $update_ref) = @_;
1464        length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
1465        my ($db, $db_lock) = ($self->{db_path}, "$self->{db_path}.lock");
1466        my $sig;
1467        if ($update_ref) {
1468                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1469                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
1470        }
1471        $LOCKFILES{$db_lock} = 1;
1472        if ($_no_metadata) {
1473                copy($db, $db_lock) or die "rev_db_set(@_): ",
1474                                           "Failed to copy: ",
1475                                           "$db => $db_lock ($!)\n";
1476        } else {
1477                rename $db, $db_lock or die "rev_db_set(@_): ",
1478                                            "Failed to rename: ",
1479                                            "$db => $db_lock ($!)\n";
1480        }
1481        open my $fh, '+<', $db_lock or croak $!;
1482        my $offset = $rev * 41;
1483        # assume that append is the common case:
1484        seek $fh, 0, 2 or croak $!;
1485        my $pos = tell $fh;
1486        if ($pos < $offset) {
1487                for (1 .. (($offset - $pos) / 41)) {
1488                        print $fh (('0' x 40),"\n") or croak $!;
1489                }
1490        }
1491        seek $fh, $offset, 0 or croak $!;
1492        print $fh $commit,"\n" or croak $!;
1493        close $fh or croak $!;
1494        if ($update_ref) {
1495                command_noisy('update-ref', '-m', "r$rev",
1496                              $self->refname, $commit);
1497        }
1498        rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
1499                                    "$db_lock => $db ($!)\n";
1500        delete $LOCKFILES{$db_lock};
1501        if ($update_ref) {
1502                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1503                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
1504                kill $sig, $$ if defined $sig;
1505        }
1506}
1507
1508sub rev_db_max {
1509        my ($self) = @_;
1510        my @stat = stat $self->{db_path} or
1511                        die "Couldn't stat $self->{db_path}: $!\n";
1512        ($stat[7] % 41) == 0 or
1513                        die "$self->{db_path} inconsistent size:$stat[7]\n";
1514        my $max = $stat[7] / 41;
1515        (($max > 0) ? $max - 1 : 0);
1516}
1517
1518sub rev_db_get {
1519        my ($self, $rev) = @_;
1520        my $ret;
1521        my $offset = $rev * 41;
1522        open my $fh, '<', $self->{db_path} or croak $!;
1523        if (sysseek($fh, $offset, 0) == $offset) {
1524                my $read = sysread($fh, $ret, 40);
1525                $ret = undef if ($read != 40 || $ret eq ('0'x40));
1526        }
1527        close $fh or croak $!;
1528        $ret;
1529}
1530
1531sub find_rev_before {
1532        my ($self, $rev, $eq_ok) = @_;
1533        --$rev unless $eq_ok;
1534        while ($rev > 0) {
1535                if (my $c = $self->rev_db_get($rev)) {
1536                        return ($rev, $c);
1537                }
1538                --$rev;
1539        }
1540        return (undef, undef);
1541}
1542
1543sub _new {
1544        my ($class, $repo_id, $ref_id, $path) = @_;
1545        unless (defined $repo_id && length $repo_id) {
1546                $repo_id = $Git::SVN::default_repo_id;
1547        }
1548        unless (defined $ref_id && length $ref_id) {
1549                $_[2] = $ref_id = $Git::SVN::default_ref_id;
1550        }
1551        $_[1] = $repo_id = sanitize_remote_name($repo_id);
1552        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1553        $_[3] = $path = '' unless (defined $path);
1554        mkpath([$dir]);
1555        unless (-f "$dir/.rev_db") {
1556                open my $fh, '>>', "$dir/.rev_db" or croak $!;
1557                close $fh or croak $!;
1558        }
1559        bless { ref_id => $ref_id, dir => $dir, index => "$dir/index",
1560                path => $path,
1561                db_path => "$dir/.rev_db", repo_id => $repo_id }, $class;
1562}
1563
1564sub uri_encode {
1565        my ($f) = @_;
1566        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1567        $f
1568}
1569
1570package Git::SVN::Prompt;
1571use strict;
1572use warnings;
1573require SVN::Core;
1574use vars qw/$_no_auth_cache $_username/;
1575
1576sub simple {
1577        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1578        $may_save = undef if $_no_auth_cache;
1579        $default_username = $_username if defined $_username;
1580        if (defined $default_username && length $default_username) {
1581                if (defined $realm && length $realm) {
1582                        print STDERR "Authentication realm: $realm\n";
1583                        STDERR->flush;
1584                }
1585                $cred->username($default_username);
1586        } else {
1587                username($cred, $realm, $may_save, $pool);
1588        }
1589        $cred->password(_read_password("Password for '" .
1590                                       $cred->username . "': ", $realm));
1591        $cred->may_save($may_save);
1592        $SVN::_Core::SVN_NO_ERROR;
1593}
1594
1595sub ssl_server_trust {
1596        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1597        $may_save = undef if $_no_auth_cache;
1598        print STDERR "Error validating server certificate for '$realm':\n";
1599        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1600                print STDERR " - The certificate is not issued by a trusted ",
1601                      "authority. Use the\n",
1602                      "   fingerprint to validate the certificate manually!\n";
1603        }
1604        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1605                print STDERR " - The certificate hostname does not match.\n";
1606        }
1607        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1608                print STDERR " - The certificate is not yet valid.\n";
1609        }
1610        if ($failures & $SVN::Auth::SSL::EXPIRED) {
1611                print STDERR " - The certificate has expired.\n";
1612        }
1613        if ($failures & $SVN::Auth::SSL::OTHER) {
1614                print STDERR " - The certificate has an unknown error.\n";
1615        }
1616        printf STDERR
1617                "Certificate information:\n".
1618                " - Hostname: %s\n".
1619                " - Valid: from %s until %s\n".
1620                " - Issuer: %s\n".
1621                " - Fingerprint: %s\n",
1622                map $cert_info->$_, qw(hostname valid_from valid_until
1623                                       issuer_dname fingerprint);
1624        my $choice;
1625prompt:
1626        print STDERR $may_save ?
1627              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1628              "(R)eject or accept (t)emporarily? ";
1629        STDERR->flush;
1630        $choice = lc(substr(<STDIN> || 'R', 0, 1));
1631        if ($choice =~ /^t$/i) {
1632                $cred->may_save(undef);
1633        } elsif ($choice =~ /^r$/i) {
1634                return -1;
1635        } elsif ($may_save && $choice =~ /^p$/i) {
1636                $cred->may_save($may_save);
1637        } else {
1638                goto prompt;
1639        }
1640        $cred->accepted_failures($failures);
1641        $SVN::_Core::SVN_NO_ERROR;
1642}
1643
1644sub ssl_client_cert {
1645        my ($cred, $realm, $may_save, $pool) = @_;
1646        $may_save = undef if $_no_auth_cache;
1647        print STDERR "Client certificate filename: ";
1648        STDERR->flush;
1649        chomp(my $filename = <STDIN>);
1650        $cred->cert_file($filename);
1651        $cred->may_save($may_save);
1652        $SVN::_Core::SVN_NO_ERROR;
1653}
1654
1655sub ssl_client_cert_pw {
1656        my ($cred, $realm, $may_save, $pool) = @_;
1657        $may_save = undef if $_no_auth_cache;
1658        $cred->password(_read_password("Password: ", $realm));
1659        $cred->may_save($may_save);
1660        $SVN::_Core::SVN_NO_ERROR;
1661}
1662
1663sub username {
1664        my ($cred, $realm, $may_save, $pool) = @_;
1665        $may_save = undef if $_no_auth_cache;
1666        if (defined $realm && length $realm) {
1667                print STDERR "Authentication realm: $realm\n";
1668        }
1669        my $username;
1670        if (defined $_username) {
1671                $username = $_username;
1672        } else {
1673                print STDERR "Username: ";
1674                STDERR->flush;
1675                chomp($username = <STDIN>);
1676        }
1677        $cred->username($username);
1678        $cred->may_save($may_save);
1679        $SVN::_Core::SVN_NO_ERROR;
1680}
1681
1682sub _read_password {
1683        my ($prompt, $realm) = @_;
1684        print STDERR $prompt;
1685        STDERR->flush;
1686        require Term::ReadKey;
1687        Term::ReadKey::ReadMode('noecho');
1688        my $password = '';
1689        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1690                last if $key =~ /[\012\015]/; # \n\r
1691                $password .= $key;
1692        }
1693        Term::ReadKey::ReadMode('restore');
1694        print STDERR "\n";
1695        STDERR->flush;
1696        $password;
1697}
1698
1699package main;
1700
1701{
1702        my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1703                                $SVN::Node::dir.$SVN::Node::unknown.
1704                                $SVN::Node::none.$SVN::Node::file.
1705                                $SVN::Node::dir.$SVN::Node::unknown.
1706                                $SVN::Auth::SSL::CNMISMATCH.
1707                                $SVN::Auth::SSL::NOTYETVALID.
1708                                $SVN::Auth::SSL::EXPIRED.
1709                                $SVN::Auth::SSL::UNKNOWNCA.
1710                                $SVN::Auth::SSL::OTHER;
1711}
1712
1713package SVN::Git::Fetcher;
1714use vars qw/@ISA/;
1715use strict;
1716use warnings;
1717use Carp qw/croak/;
1718use IO::File qw//;
1719use Digest::MD5;
1720
1721# file baton members: path, mode_a, mode_b, pool, fh, blob, base
1722sub new {
1723        my ($class, $git_svn) = @_;
1724        my $self = SVN::Delta::Editor->new;
1725        bless $self, $class;
1726        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
1727        $self->{empty} = {};
1728        $self->{dir_prop} = {};
1729        $self->{file_prop} = {};
1730        $self->{absent_dir} = {};
1731        $self->{absent_file} = {};
1732        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
1733        $self;
1734}
1735
1736sub set_path_strip {
1737        my ($self, $path) = @_;
1738        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
1739}
1740
1741sub open_root {
1742        { path => '' };
1743}
1744
1745sub open_directory {
1746        my ($self, $path, $pb, $rev) = @_;
1747        { path => $path };
1748}
1749
1750sub git_path {
1751        my ($self, $path) = @_;
1752        if ($self->{path_strip}) {
1753                $path =~ s!$self->{path_strip}!! or
1754                  die "Failed to strip path '$path' ($self->{path_strip})\n";
1755        }
1756        $path;
1757}
1758
1759sub delete_entry {
1760        my ($self, $path, $rev, $pb) = @_;
1761
1762        my $gpath = $self->git_path($path);
1763        return undef if ($gpath eq '');
1764
1765        # remove entire directories.
1766        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
1767                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
1768                                                     -r --name-only -z/,
1769                                                     $self->{c}, '--', $gpath);
1770                local $/ = "\0";
1771                while (<$ls>) {
1772                        chomp;
1773                        $self->{gii}->remove($_);
1774                        print "\tD\t$_\n" unless $::_q;
1775                }
1776                print "\tD\t$gpath/\n" unless $::_q;
1777                command_close_pipe($ls, $ctx);
1778                $self->{empty}->{$path} = 0
1779        } else {
1780                $self->{gii}->remove($gpath);
1781                print "\tD\t$gpath\n" unless $::_q;
1782        }
1783        undef;
1784}
1785
1786sub open_file {
1787        my ($self, $path, $pb, $rev) = @_;
1788        my $gpath = $self->git_path($path);
1789        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
1790                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
1791        unless (defined $mode && defined $blob) {
1792                die "$path was not found in commit $self->{c} (r$rev)\n";
1793        }
1794        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
1795          pool => SVN::Pool->new, action => 'M' };
1796}
1797
1798sub add_file {
1799        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
1800        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1801        delete $self->{empty}->{$dir};
1802        { path => $path, mode_a => 100644, mode_b => 100644,
1803          pool => SVN::Pool->new, action => 'A' };
1804}
1805
1806sub add_directory {
1807        my ($self, $path, $cp_path, $cp_rev) = @_;
1808        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1809        delete $self->{empty}->{$dir};
1810        $self->{empty}->{$path} = 1;
1811        { path => $path };
1812}
1813
1814sub change_dir_prop {
1815        my ($self, $db, $prop, $value) = @_;
1816        $self->{dir_prop}->{$db->{path}} ||= {};
1817        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
1818        undef;
1819}
1820
1821sub absent_directory {
1822        my ($self, $path, $pb) = @_;
1823        $self->{absent_dir}->{$pb->{path}} ||= [];
1824        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
1825        undef;
1826}
1827
1828sub absent_file {
1829        my ($self, $path, $pb) = @_;
1830        $self->{absent_file}->{$pb->{path}} ||= [];
1831        push @{$self->{absent_file}->{$pb->{path}}}, $path;
1832        undef;
1833}
1834
1835sub change_file_prop {
1836        my ($self, $fb, $prop, $value) = @_;
1837        if ($prop eq 'svn:executable') {
1838                if ($fb->{mode_b} != 120000) {
1839                        $fb->{mode_b} = defined $value ? 100755 : 100644;
1840                }
1841        } elsif ($prop eq 'svn:special') {
1842                $fb->{mode_b} = defined $value ? 120000 : 100644;
1843        } else {
1844                $self->{file_prop}->{$fb->{path}} ||= {};
1845                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
1846        }
1847        undef;
1848}
1849
1850sub apply_textdelta {
1851        my ($self, $fb, $exp) = @_;
1852        my $fh = IO::File->new_tmpfile;
1853        $fh->autoflush(1);
1854        # $fh gets auto-closed() by SVN::TxDelta::apply(),
1855        # (but $base does not,) so dup() it for reading in close_file
1856        open my $dup, '<&', $fh or croak $!;
1857        my $base = IO::File->new_tmpfile;
1858        $base->autoflush(1);
1859        if ($fb->{blob}) {
1860                defined (my $pid = fork) or croak $!;
1861                if (!$pid) {
1862                        open STDOUT, '>&', $base or croak $!;
1863                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
1864                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
1865                }
1866                waitpid $pid, 0;
1867                croak $? if $?;
1868
1869                if (defined $exp) {
1870                        seek $base, 0, 0 or croak $!;
1871                        my $md5 = Digest::MD5->new;
1872                        $md5->addfile($base);
1873                        my $got = $md5->hexdigest;
1874                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
1875                            "expected: $exp\n",
1876                            "     got: $got\n" if ($got ne $exp);
1877                }
1878        }
1879        seek $base, 0, 0 or croak $!;
1880        $fb->{fh} = $dup;
1881        $fb->{base} = $base;
1882        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
1883}
1884
1885sub close_file {
1886        my ($self, $fb, $exp) = @_;
1887        my $hash;
1888        my $path = $self->git_path($fb->{path});
1889        if (my $fh = $fb->{fh}) {
1890                seek($fh, 0, 0) or croak $!;
1891                my $md5 = Digest::MD5->new;
1892                $md5->addfile($fh);
1893                my $got = $md5->hexdigest;
1894                die "Checksum mismatch: $path\n",
1895                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
1896                seek($fh, 0, 0) or croak $!;
1897                if ($fb->{mode_b} == 120000) {
1898                        read($fh, my $buf, 5) == 5 or croak $!;
1899                        $buf eq 'link ' or die "$path has mode 120000",
1900                                               "but is not a link\n";
1901                }
1902                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
1903                if (!$pid) {
1904                        open STDIN, '<&', $fh or croak $!;
1905                        exec qw/git-hash-object -w --stdin/ or croak $!;
1906                }
1907                chomp($hash = do { local $/; <$out> });
1908                close $out or croak $!;
1909                close $fh or croak $!;
1910                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
1911                close $fb->{base} or croak $!;
1912        } else {
1913                $hash = $fb->{blob} or die "no blob information\n";
1914        }
1915        $fb->{pool}->clear;
1916        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
1917        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
1918        undef;
1919}
1920
1921sub abort_edit {
1922        my $self = shift;
1923        $self->{nr} = $self->{gii}->{nr};
1924        delete $self->{gii};
1925        $self->SUPER::abort_edit(@_);
1926}
1927
1928sub close_edit {
1929        my $self = shift;
1930        $self->{git_commit_ok} = 1;
1931        $self->{nr} = $self->{gii}->{nr};
1932        delete $self->{gii};
1933        $self->SUPER::close_edit(@_);
1934}
1935
1936package SVN::Git::Editor;
1937use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
1938use strict;
1939use warnings;
1940use Carp qw/croak/;
1941use IO::File;
1942use Digest::MD5;
1943
1944sub new {
1945        my ($class, $opts) = @_;
1946        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
1947                die "$_ required!\n" unless (defined $opts->{$_});
1948        }
1949
1950        my $pool = SVN::Pool->new;
1951        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
1952        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
1953                                     $opts->{r}, $mods);
1954
1955        # $opts->{ra} functions should not be used after this:
1956        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
1957                                                $opts->{editor_cb}, $pool);
1958        my $self = SVN::Delta::Editor->new(@ce, $pool);
1959        bless $self, $class;
1960        foreach (qw/svn_path r tree_a tree_b/) {
1961                $self->{$_} = $opts->{$_};
1962        }
1963        $self->{url} = $opts->{ra}->{url};
1964        $self->{mods} = $mods;
1965        $self->{types} = $types;
1966        $self->{pool} = $pool;
1967        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
1968        $self->{rm} = { };
1969        $self->{path_prefix} = length $self->{svn_path} ?
1970                               "$self->{svn_path}/" : '';
1971        return $self;
1972}
1973
1974sub generate_diff {
1975        my ($tree_a, $tree_b) = @_;
1976        my @diff_tree = qw(diff-tree -z -r);
1977        if ($_cp_similarity) {
1978                push @diff_tree, "-C$_cp_similarity";
1979        } else {
1980                push @diff_tree, '-C';
1981        }
1982        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1983        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
1984        push @diff_tree, $tree_a, $tree_b;
1985        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
1986        local $/ = "\0";
1987        my $state = 'meta';
1988        my @mods;
1989        while (<$diff_fh>) {
1990                chomp $_; # this gets rid of the trailing "\0"
1991                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1992                                        $::sha1\s($::sha1)\s
1993                                        ([MTCRAD])\d*$/xo) {
1994                        push @mods, {   mode_a => $1, mode_b => $2,
1995                                        sha1_b => $3, chg => $4 };
1996                        if ($4 =~ /^(?:C|R)$/) {
1997                                $state = 'file_a';
1998                        } else {
1999                                $state = 'file_b';
2000                        }
2001                } elsif ($state eq 'file_a') {
2002                        my $x = $mods[$#mods] or croak "Empty array\n";
2003                        if ($x->{chg} !~ /^(?:C|R)$/) {
2004                                croak "Error parsing $_, $x->{chg}\n";
2005                        }
2006                        $x->{file_a} = $_;
2007                        $state = 'file_b';
2008                } elsif ($state eq 'file_b') {
2009                        my $x = $mods[$#mods] or croak "Empty array\n";
2010                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2011                                croak "Error parsing $_, $x->{chg}\n";
2012                        }
2013                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2014                                croak "Error parsing $_, $x->{chg}\n";
2015                        }
2016                        $x->{file_b} = $_;
2017                        $state = 'meta';
2018                } else {
2019                        croak "Error parsing $_\n";
2020                }
2021        }
2022        command_close_pipe($diff_fh, $ctx);
2023        \@mods;
2024}
2025
2026sub check_diff_paths {
2027        my ($ra, $pfx, $rev, $mods) = @_;
2028        my %types;
2029        $pfx .= '/' if length $pfx;
2030
2031        sub type_diff_paths {
2032                my ($ra, $types, $path, $rev) = @_;
2033                my @p = split m#/+#, $path;
2034                my $c = shift @p;
2035                unless (defined $types->{$c}) {
2036                        $types->{$c} = $ra->check_path($c, $rev);
2037                }
2038                while (@p) {
2039                        $c .= '/' . shift @p;
2040                        next if defined $types->{$c};
2041                        $types->{$c} = $ra->check_path($c, $rev);
2042                }
2043        }
2044
2045        foreach my $m (@$mods) {
2046                foreach my $f (qw/file_a file_b/) {
2047                        next unless defined $m->{$f};
2048                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2049                        if (length $pfx.$dir && ! defined $types{$dir}) {
2050                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2051                        }
2052                }
2053        }
2054        \%types;
2055}
2056
2057sub split_path {
2058        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2059}
2060
2061sub repo_path {
2062        my ($self, $path) = @_;
2063        $self->{path_prefix}.(defined $path ? $path : '');
2064}
2065
2066sub url_path {
2067        my ($self, $path) = @_;
2068        $self->{url} . '/' . $self->repo_path($path);
2069}
2070
2071sub rmdirs {
2072        my ($self) = @_;
2073        my $rm = $self->{rm};
2074        delete $rm->{''}; # we never delete the url we're tracking
2075        return unless %$rm;
2076
2077        foreach (keys %$rm) {
2078                my @d = split m#/#, $_;
2079                my $c = shift @d;
2080                $rm->{$c} = 1;
2081                while (@d) {
2082                        $c .= '/' . shift @d;
2083                        $rm->{$c} = 1;
2084                }
2085        }
2086        delete $rm->{$self->{svn_path}};
2087        delete $rm->{''}; # we never delete the url we're tracking
2088        return unless %$rm;
2089
2090        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2091                                             $self->{tree_b});
2092        local $/ = "\0";
2093        while (<$fh>) {
2094                chomp;
2095                my @dn = split m#/#, $_;
2096                while (pop @dn) {
2097                        delete $rm->{join '/', @dn};
2098                }
2099                unless (%$rm) {
2100                        close $fh;
2101                        return;
2102                }
2103        }
2104        command_close_pipe($fh, $ctx);
2105
2106        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2107        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2108                $self->close_directory($bat->{$d}, $p);
2109                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2110                print "\tD+\t$d/\n" unless $::_q;
2111                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2112                delete $bat->{$d};
2113        }
2114}
2115
2116sub open_or_add_dir {
2117        my ($self, $full_path, $baton) = @_;
2118        my $t = $self->{types}->{$full_path};
2119        if (!defined $t) {
2120                die "$full_path not known in r$self->{r} or we have a bug!\n";
2121        }
2122        if ($t == $SVN::Node::none) {
2123                return $self->add_directory($full_path, $baton,
2124                                                undef, -1, $self->{pool});
2125        } elsif ($t == $SVN::Node::dir) {
2126                return $self->open_directory($full_path, $baton,
2127                                                $self->{r}, $self->{pool});
2128        }
2129        print STDERR "$full_path already exists in repository at ",
2130                "r$self->{r} and it is not a directory (",
2131                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2132        exit 1;
2133}
2134
2135sub ensure_path {
2136        my ($self, $path) = @_;
2137        my $bat = $self->{bat};
2138        my $repo_path = $self->repo_path($path);
2139        return $bat->{''} unless (length $repo_path);
2140        my @p = split m#/+#, $repo_path;
2141        my $c = shift @p;
2142        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2143        while (@p) {
2144                my $c0 = $c;
2145                $c .= '/' . shift @p;
2146                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2147        }
2148        return $bat->{$c};
2149}
2150
2151sub A {
2152        my ($self, $m) = @_;
2153        my ($dir, $file) = split_path($m->{file_b});
2154        my $pbat = $self->ensure_path($dir);
2155        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2156                                        undef, -1);
2157        print "\tA\t$m->{file_b}\n" unless $::_q;
2158        $self->chg_file($fbat, $m);
2159        $self->close_file($fbat,undef,$self->{pool});
2160}
2161
2162sub C {
2163        my ($self, $m) = @_;
2164        my ($dir, $file) = split_path($m->{file_b});
2165        my $pbat = $self->ensure_path($dir);
2166        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2167                                $self->url_path($m->{file_a}), $self->{r});
2168        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2169        $self->chg_file($fbat, $m);
2170        $self->close_file($fbat,undef,$self->{pool});
2171}
2172
2173sub delete_entry {
2174        my ($self, $path, $pbat) = @_;
2175        my $rpath = $self->repo_path($path);
2176        my ($dir, $file) = split_path($rpath);
2177        $self->{rm}->{$dir} = 1;
2178        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2179}
2180
2181sub R {
2182        my ($self, $m) = @_;
2183        my ($dir, $file) = split_path($m->{file_b});
2184        my $pbat = $self->ensure_path($dir);
2185        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2186                                $self->url_path($m->{file_a}), $self->{r});
2187        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2188        $self->chg_file($fbat, $m);
2189        $self->close_file($fbat,undef,$self->{pool});
2190
2191        ($dir, $file) = split_path($m->{file_a});
2192        $pbat = $self->ensure_path($dir);
2193        $self->delete_entry($m->{file_a}, $pbat);
2194}
2195
2196sub M {
2197        my ($self, $m) = @_;
2198        my ($dir, $file) = split_path($m->{file_b});
2199        my $pbat = $self->ensure_path($dir);
2200        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2201                                $pbat,$self->{r},$self->{pool});
2202        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2203        $self->chg_file($fbat, $m);
2204        $self->close_file($fbat,undef,$self->{pool});
2205}
2206
2207sub T { shift->M(@_) }
2208
2209sub change_file_prop {
2210        my ($self, $fbat, $pname, $pval) = @_;
2211        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2212}
2213
2214sub chg_file {
2215        my ($self, $fbat, $m) = @_;
2216        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2217                $self->change_file_prop($fbat,'svn:executable','*');
2218        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2219                $self->change_file_prop($fbat,'svn:executable',undef);
2220        }
2221        my $fh = IO::File->new_tmpfile or croak $!;
2222        if ($m->{mode_b} =~ /^120/) {
2223                print $fh 'link ' or croak $!;
2224                $self->change_file_prop($fbat,'svn:special','*');
2225        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2226                $self->change_file_prop($fbat,'svn:special',undef);
2227        }
2228        defined(my $pid = fork) or croak $!;
2229        if (!$pid) {
2230                open STDOUT, '>&', $fh or croak $!;
2231                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2232        }
2233        waitpid $pid, 0;
2234        croak $? if $?;
2235        $fh->flush == 0 or croak $!;
2236        seek $fh, 0, 0 or croak $!;
2237
2238        my $md5 = Digest::MD5->new;
2239        $md5->addfile($fh) or croak $!;
2240        seek $fh, 0, 0 or croak $!;
2241
2242        my $exp = $md5->hexdigest;
2243        my $pool = SVN::Pool->new;
2244        my $atd = $self->apply_textdelta($fbat, undef, $pool);
2245        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2246        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2247        $pool->clear;
2248
2249        close $fh or croak $!;
2250}
2251
2252sub D {
2253        my ($self, $m) = @_;
2254        my ($dir, $file) = split_path($m->{file_b});
2255        my $pbat = $self->ensure_path($dir);
2256        print "\tD\t$m->{file_b}\n" unless $::_q;
2257        $self->delete_entry($m->{file_b}, $pbat);
2258}
2259
2260sub close_edit {
2261        my ($self) = @_;
2262        my ($p,$bat) = ($self->{pool}, $self->{bat});
2263        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2264                $self->close_directory($bat->{$_}, $p);
2265        }
2266        $self->SUPER::close_edit($p);
2267        $p->clear;
2268}
2269
2270sub abort_edit {
2271        my ($self) = @_;
2272        $self->SUPER::abort_edit($self->{pool});
2273}
2274
2275sub DESTROY {
2276        my $self = shift;
2277        $self->SUPER::DESTROY(@_);
2278        $self->{pool}->clear;
2279}
2280
2281# this drives the editor
2282sub apply_diff {
2283        my ($self) = @_;
2284        my $mods = $self->{mods};
2285        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2286        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2287                my $f = $m->{chg};
2288                if (defined $o{$f}) {
2289                        $self->$f($m);
2290                } else {
2291                        fatal("Invalid change type: $f\n");
2292                }
2293        }
2294        $self->rmdirs if $_rmdir;
2295        if (@$mods == 0) {
2296                $self->abort_edit;
2297        } else {
2298                $self->close_edit;
2299        }
2300        return scalar @$mods;
2301}
2302
2303package Git::SVN::Ra;
2304use vars qw/@ISA $config_dir/;
2305use strict;
2306use warnings;
2307my ($can_do_switch);
2308my $RA;
2309
2310BEGIN {
2311        # enforce temporary pool usage for some simple functions
2312        my $e;
2313        foreach (qw/get_latest_revnum get_uuid get_repos_root/) {
2314                $e .= "sub $_ {
2315                        my \$self = shift;
2316                        my \$pool = SVN::Pool->new;
2317                        my \@ret = \$self->SUPER::$_(\@_,\$pool);
2318                        \$pool->clear;
2319                        wantarray ? \@ret : \$ret[0]; }\n";
2320        }
2321
2322        # get_dir needs $pool held in cache for dirents to work,
2323        # check_path is cacheable and rev_proplist is close enough
2324        # for our purposes.
2325        foreach (qw/check_path get_dir rev_proplist/) {
2326                $e .= "my \%${_}_cache; my \$${_}_rev = 0; sub $_ {
2327                        my \$self = shift;
2328                        my \$r = pop;
2329                        my \$k = join(\"\\0\", \@_);
2330                        if (my \$x = \$${_}_cache{\$r}->{\$k}) {
2331                                return wantarray ? \@\$x : \$x->[0];
2332                        }
2333                        my \$pool = SVN::Pool->new;
2334                        my \@ret = \$self->SUPER::$_(\@_, \$r, \$pool);
2335                        if (\$r != \$${_}_rev) {
2336                                \%${_}_cache = ( pool => [] );
2337                                \$${_}_rev = \$r;
2338                        }
2339                        \$${_}_cache{\$r}->{\$k} = \\\@ret;
2340                        push \@{\$${_}_cache{pool}}, \$pool;
2341                        wantarray ? \@ret : \$ret[0]; }\n";
2342        }
2343        $e .= "\n1;";
2344        eval $e or die $@;
2345}
2346
2347sub new {
2348        my ($class, $url) = @_;
2349        $url =~ s!/+$!!;
2350        return $RA if ($RA && $RA->{url} eq $url);
2351
2352        SVN::_Core::svn_config_ensure($config_dir, undef);
2353        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2354            SVN::Client::get_simple_provider(),
2355            SVN::Client::get_ssl_server_trust_file_provider(),
2356            SVN::Client::get_simple_prompt_provider(
2357              \&Git::SVN::Prompt::simple, 2),
2358            SVN::Client::get_ssl_client_cert_prompt_provider(
2359              \&Git::SVN::Prompt::ssl_client_cert, 2),
2360            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2361              \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2362            SVN::Client::get_username_provider(),
2363            SVN::Client::get_ssl_server_trust_prompt_provider(
2364              \&Git::SVN::Prompt::ssl_server_trust),
2365            SVN::Client::get_username_prompt_provider(
2366              \&Git::SVN::Prompt::username, 2),
2367          ]);
2368        my $config = SVN::Core::config_get_config($config_dir);
2369        my $self = SVN::Ra->new(url => $url, auth => $baton,
2370                              config => $config,
2371                              pool => SVN::Pool->new,
2372                              auth_provider_callbacks => $callbacks);
2373        $self->{svn_path} = $url;
2374        $self->{repos_root} = $self->get_repos_root;
2375        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
2376        $RA = bless $self, $class;
2377}
2378
2379sub DESTROY {
2380        # do not call the real DESTROY since we store ourselves in $RA
2381}
2382
2383sub get_log {
2384        my ($self, @args) = @_;
2385        my $pool = SVN::Pool->new;
2386        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2387        my $ret = $self->SUPER::get_log(@args, $pool);
2388        $pool->clear;
2389        $ret;
2390}
2391
2392sub get_commit_editor {
2393        my ($self, $log, $cb, $pool) = @_;
2394        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2395        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2396}
2397
2398sub uuid {
2399        my ($self) = @_;
2400        $self->{uuid} ||= $self->get_uuid;
2401}
2402
2403sub gs_do_update {
2404        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2405        my $new = ($rev_a == $rev_b);
2406        my $path = $gs->{path};
2407
2408        my $pool = SVN::Pool->new;
2409        $editor->set_path_strip($path);
2410        my (@pc) = split m#/#, $path;
2411        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2412                                        1, $editor, $pool);
2413        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2414
2415        # Since we can't rely on svn_ra_reparent being available, we'll
2416        # just have to do some magic with set_path to make it so
2417        # we only want a partial path.
2418        my $sp = '';
2419        my $final = join('/', @pc);
2420        while (@pc) {
2421                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2422                $sp .= '/' if length $sp;
2423                $sp .= shift @pc;
2424        }
2425        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2426
2427        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2428
2429        $reporter->finish_report($pool);
2430        $pool->clear;
2431        $editor->{git_commit_ok};
2432}
2433
2434# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2435# svn_ra_reparent didn't work before 1.4)
2436sub gs_do_switch {
2437        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
2438        my $path = $gs->{path};
2439        my $pool = SVN::Pool->new;
2440
2441        my $full_url = $self->{url};
2442        my $old_url = $full_url;
2443        $full_url .= "/$path" if length $path;
2444        my ($ra, $reparented);
2445        if ($old_url ne $full_url) {
2446                if ($old_url !~ m#^svn(\+ssh)?://#) {
2447                        SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2448                                                  $pool);
2449                        $self->{url} = $full_url;
2450                        $reparented = 1;
2451                } else {
2452                        $ra = Git::SVN::Ra->new($full_url);
2453                }
2454        }
2455        $ra ||= $self;
2456        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
2457        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2458        $reporter->set_path('', $rev_a, 0, @lock, $pool);
2459        $reporter->finish_report($pool);
2460
2461        if ($reparented) {
2462                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2463                $self->{url} = $old_url;
2464        }
2465
2466        $pool->clear;
2467        $editor->{git_commit_ok};
2468}
2469
2470sub gs_fetch_loop_common {
2471        my ($self, $base, $head, $gsv, $globs) = @_;
2472        return if ($base > $head);
2473        my $inc = 1000;
2474        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2475        my %common;
2476        my $common_max = scalar @$gsv;
2477
2478        foreach my $gs (@$gsv) {
2479                if (my $last_commit = $gs->last_commit) {
2480                        $gs->assert_index_clean($last_commit);
2481                }
2482                my @tmp = split m#/#, $gs->{path};
2483                my $p = '';
2484                foreach (@tmp) {
2485                        $p .= length($p) ? "/$_" : $_;
2486                        $common{$p} ||= 0;
2487                        $common{$p}++;
2488                }
2489        }
2490        $globs ||= [];
2491        $common_max += scalar @$globs;
2492        foreach my $glob (@$globs) {
2493                my @tmp = split m#/#, $glob->{path}->{left};
2494                my $p = '';
2495                foreach (@tmp) {
2496                        $p .= length($p) ? "/$_" : $_;
2497                        $common{$p} ||= 0;
2498                        $common{$p}++;
2499                }
2500        }
2501
2502        my $longest_path = '';
2503        foreach (sort {length $b <=> length $a} keys %common) {
2504                if ($common{$_} == $common_max) {
2505                        $longest_path = $_;
2506                        last;
2507                }
2508        }
2509        while (1) {
2510                my %revs;
2511                my $err;
2512                my $err_handler = $SVN::Error::handler;
2513                $SVN::Error::handler = sub {
2514                        ($err) = @_;
2515                        skip_unknown_revs($err);
2516                };
2517                sub _cb {
2518                        my ($paths, $r, $author, $date, $log) = @_;
2519                        [ dup_changed_paths($paths),
2520                          { author => $author, date => $date, log => $log } ];
2521                }
2522                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
2523                               sub { $revs{$_[1]} = _cb(@_) });
2524                if ($err && $max >= $head) {
2525                        print STDERR "Path '$longest_path' ",
2526                                     "was probably deleted:\n",
2527                                     $err->expanded_message,
2528                                     "\nWill attempt to follow ",
2529                                     "revisions r$min .. r$max ",
2530                                     "committed before the deletion\n";
2531                        my $hi = $max;
2532                        while (--$hi >= $min) {
2533                                my $ok;
2534                                $self->get_log([$longest_path], $min, $hi,
2535                                               0, 1, 1, sub {
2536                                               $ok ||= $_[1];
2537                                               $revs{$_[1]} = _cb(@_) });
2538                                if ($ok) {
2539                                        print STDERR "r$min .. r$ok OK\n";
2540                                        last;
2541                                }
2542                        }
2543                }
2544                $SVN::Error::handler = $err_handler;
2545
2546                my %exists = map { $_->{path} => $_ } @$gsv;
2547                foreach my $r (sort {$a <=> $b} keys %revs) {
2548                        my ($paths, $logged) = @{$revs{$r}};
2549
2550                        foreach my $gs ($self->match_globs(\%exists, $paths,
2551                                                           $globs, $r)) {
2552                                if ($gs->rev_db_max >= $r) {
2553                                        next;
2554                                }
2555                                next unless $gs->match_paths($paths, $r);
2556                                $gs->{logged_rev_props} = $logged;
2557                                my $log_entry = $gs->do_fetch($paths, $r);
2558                                if ($log_entry) {
2559                                        $gs->do_git_commit($log_entry);
2560                                }
2561                        }
2562                        foreach my $g (@$globs) {
2563                                my $f = "$ENV{GIT_DIR}/svn/." .
2564                                        $self->uuid . ".$g->{t}";
2565                                open my $fh, '>', "$f.tmp" or
2566                                      die "Can't open $f.tmp for writing: $!";
2567                                print $fh "$r\n" or
2568                                      die "Couldn't write to $f: $!\n";
2569                                close $fh or die "Error closing $f: $!\n";
2570                                rename "$f.tmp", $f or
2571                                       die "Couldn't rename ",
2572                                           "$f.tmp => $f: $!\n";
2573                        }
2574                }
2575                # pre-fill the .rev_db since it'll eventually get filled in
2576                # with '0' x40 if something new gets committed
2577                foreach my $gs (@$gsv) {
2578                        next if defined $gs->rev_db_get($max);
2579                        $gs->rev_db_set($max, 0 x40);
2580                }
2581                last if $max >= $head;
2582                $min = $max + 1;
2583                $max += $inc;
2584                $max = $head if ($max > $head);
2585        }
2586}
2587
2588sub match_globs {
2589        my ($self, $exists, $paths, $globs, $r) = @_;
2590
2591        sub get_dir_check {
2592                my ($self, $exists, $g, $r) = @_;
2593                my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
2594                return unless scalar @x == 3;
2595                my $dirents = $x[0];
2596                foreach my $de (keys %$dirents) {
2597                        next if $dirents->{$de}->kind != $SVN::Node::dir;
2598                        my $p = $g->{path}->full_path($de);
2599                        next if $exists->{$p};
2600                        next if (length $g->{path}->{right} &&
2601                                 ($self->check_path($p, $r) !=
2602                                  $SVN::Node::dir));
2603                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
2604                                         $g->{ref}->full_path($de), 1);
2605                }
2606        }
2607        foreach my $g (@$globs) {
2608                if (my $path = $paths->{"/$g->{path}->{left}"}) {
2609                        if ($path->{action} =~ /^[AR]$/) {
2610                                get_dir_check($self, $exists, $g, $r);
2611                        }
2612                }
2613                foreach (keys %$paths) {
2614                        if (/$g->{path}->{left_regex}/) {
2615                                next if $paths->{$_}->{action} !~ /^[AR]$/;
2616                                get_dir_check($self, $exists, $g, $r);
2617                        }
2618                        next unless /$g->{path}->{regex}/;
2619                        my $p = $1;
2620                        my $pathname = $g->{path}->full_path($p);
2621                        next if $exists->{$pathname};
2622                        $exists->{$pathname} = Git::SVN->init(
2623                                              $self->{url}, $pathname, undef,
2624                                              $g->{ref}->full_path($p), 1);
2625                }
2626                my $c = '';
2627                foreach (split m#/#, $g->{path}->{left}) {
2628                        $c .= "/$_";
2629                        next unless ($paths->{$c} &&
2630                                     ($paths->{$c}->{action} =~ /^[AR]$/));
2631                        get_dir_check($self, $exists, $g, $r);
2632                }
2633        }
2634        values %$exists;
2635}
2636
2637sub minimize_url {
2638        my ($self) = @_;
2639        return $self->{url} if ($self->{url} eq $self->{repos_root});
2640        my $url = $self->{repos_root};
2641        my @components = split(m!/!, $self->{svn_path});
2642        my $c = '';
2643        do {
2644                $url .= "/$c" if length $c;
2645                eval { (ref $self)->new($url)->get_latest_revnum };
2646        } while ($@ && ($c = shift @components));
2647        $url;
2648}
2649
2650sub can_do_switch {
2651        my $self = shift;
2652        unless (defined $can_do_switch) {
2653                my $pool = SVN::Pool->new;
2654                my $rep = eval {
2655                        $self->do_switch(1, '', 0, $self->{url},
2656                                         SVN::Delta::Editor->new, $pool);
2657                };
2658                if ($@) {
2659                        $can_do_switch = 0;
2660                } else {
2661                        $rep->abort_report($pool);
2662                        $can_do_switch = 1;
2663                }
2664                $pool->clear;
2665        }
2666        $can_do_switch;
2667}
2668
2669sub skip_unknown_revs {
2670        my ($err) = @_;
2671        my $errno = $err->apr_err();
2672        # Maybe the branch we're tracking didn't
2673        # exist when the repo started, so it's
2674        # not an error if it doesn't, just continue
2675        #
2676        # Wonderfully consistent library, eh?
2677        # 160013 - svn:// and file://
2678        # 175002 - http(s)://
2679        # 175007 - http(s):// (this repo required authorization, too...)
2680        #   More codes may be discovered later...
2681        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2682                return;
2683        }
2684        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2685}
2686
2687# svn_log_changed_path_t objects passed to get_log are likely to be
2688# overwritten even if only the refs are copied to an external variable,
2689# so we should dup the structures in their entirety.  Using an externally
2690# passed pool (instead of our temporary and quickly cleared pool in
2691# Git::SVN::Ra) does not help matters at all...
2692sub dup_changed_paths {
2693        my ($paths) = @_;
2694        return undef unless $paths;
2695        my %ret;
2696        foreach my $p (keys %$paths) {
2697                my $i = $paths->{$p};
2698                my %s = map { $_ => $i->$_ }
2699                              qw/copyfrom_path copyfrom_rev action/;
2700                $ret{$p} = \%s;
2701        }
2702        \%ret;
2703}
2704
2705package Git::SVN::Log;
2706use strict;
2707use warnings;
2708use POSIX qw/strftime/;
2709use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2710            %rusers $show_commit $incremental/;
2711my $l_fmt;
2712
2713sub cmt_showable {
2714        my ($c) = @_;
2715        return 1 if defined $c->{r};
2716        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2717                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2718                my @log = command(qw/cat-file commit/, $c->{c});
2719                shift @log while ($log[0] ne "\n");
2720                shift @log;
2721                @{$c->{l}} = grep !/^git-svn-id: /, @log;
2722
2723                (undef, $c->{r}, undef) = ::extract_metadata(
2724                                (grep(/^git-svn-id: /, @log))[-1]);
2725        }
2726        return defined $c->{r};
2727}
2728
2729sub log_use_color {
2730        return 1 if $color;
2731        my ($dc, $dcvar);
2732        $dcvar = 'color.diff';
2733        $dc = `git-config --get $dcvar`;
2734        if ($dc eq '') {
2735                # nothing at all; fallback to "diff.color"
2736                $dcvar = 'diff.color';
2737                $dc = `git-config --get $dcvar`;
2738        }
2739        chomp($dc);
2740        if ($dc eq 'auto') {
2741                my $pc;
2742                $pc = `git-config --get color.pager`;
2743                if ($pc eq '') {
2744                        # does not have it -- fallback to pager.color
2745                        $pc = `git-config --bool --get pager.color`;
2746                }
2747                else {
2748                        $pc = `git-config --bool --get color.pager`;
2749                        if ($?) {
2750                                $pc = 'false';
2751                        }
2752                }
2753                chomp($pc);
2754                if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2755                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2756                }
2757                return 0;
2758        }
2759        return 0 if $dc eq 'never';
2760        return 1 if $dc eq 'always';
2761        chomp($dc = `git-config --bool --get $dcvar`);
2762        return ($dc eq 'true');
2763}
2764
2765sub git_svn_log_cmd {
2766        my ($r_min, $r_max) = @_;
2767        my $gs = Git::SVN->_new;
2768        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2769                   $gs->refname);
2770        push @cmd, '-r' unless $non_recursive;
2771        push @cmd, qw/--raw --name-status/ if $verbose;
2772        push @cmd, '--color' if log_use_color();
2773        return @cmd unless defined $r_max;
2774        if ($r_max == $r_min) {
2775                push @cmd, '--max-count=1';
2776                if (my $c = $gs->rev_db_get($r_max)) {
2777                        push @cmd, $c;
2778                }
2779        } else {
2780                my ($c_min, $c_max);
2781                $c_max = $gs->rev_db_get($r_max);
2782                $c_min = $gs->rev_db_get($r_min);
2783                if (defined $c_min && defined $c_max) {
2784                        if ($r_max > $r_max) {
2785                                push @cmd, "$c_min..$c_max";
2786                        } else {
2787                                push @cmd, "$c_max..$c_min";
2788                        }
2789                } elsif ($r_max > $r_min) {
2790                        push @cmd, $c_max;
2791                } else {
2792                        push @cmd, $c_min;
2793                }
2794        }
2795        return @cmd;
2796}
2797
2798# adapted from pager.c
2799sub config_pager {
2800        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2801        if (!defined $pager) {
2802                $pager = 'less';
2803        } elsif (length $pager == 0 || $pager eq 'cat') {
2804                $pager = undef;
2805        }
2806}
2807
2808sub run_pager {
2809        return unless -t *STDOUT;
2810        pipe my $rfd, my $wfd or return;
2811        defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2812        if (!$pid) {
2813                open STDOUT, '>&', $wfd or
2814                                     ::fatal "Can't redirect to stdout: $!\n";
2815                return;
2816        }
2817        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2818        $ENV{LESS} ||= 'FRSX';
2819        exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2820}
2821
2822sub tz_to_s_offset {
2823        my ($tz) = @_;
2824        $tz =~ s/(\d\d)$//;
2825        return ($1 * 60) + ($tz * 3600);
2826}
2827
2828sub get_author_info {
2829        my ($dest, $author, $t, $tz) = @_;
2830        $author =~ s/(?:^\s*|\s*$)//g;
2831        $dest->{a_raw} = $author;
2832        my $au;
2833        if ($::_authors) {
2834                $au = $rusers{$author} || undef;
2835        }
2836        if (!$au) {
2837                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2838        }
2839        $dest->{t} = $t;
2840        $dest->{tz} = $tz;
2841        $dest->{a} = $au;
2842        # Date::Parse isn't in the standard Perl distro :(
2843        if ($tz =~ s/^\+//) {
2844                $t += tz_to_s_offset($tz);
2845        } elsif ($tz =~ s/^\-//) {
2846                $t -= tz_to_s_offset($tz);
2847        }
2848        $dest->{t_utc} = $t;
2849}
2850
2851sub process_commit {
2852        my ($c, $r_min, $r_max, $defer) = @_;
2853        if (defined $r_min && defined $r_max) {
2854                if ($r_min == $c->{r} && $r_min == $r_max) {
2855                        show_commit($c);
2856                        return 0;
2857                }
2858                return 1 if $r_min == $r_max;
2859                if ($r_min < $r_max) {
2860                        # we need to reverse the print order
2861                        return 0 if (defined $limit && --$limit < 0);
2862                        push @$defer, $c;
2863                        return 1;
2864                }
2865                if ($r_min != $r_max) {
2866                        return 1 if ($r_min < $c->{r});
2867                        return 1 if ($r_max > $c->{r});
2868                }
2869        }
2870        return 0 if (defined $limit && --$limit < 0);
2871        show_commit($c);
2872        return 1;
2873}
2874
2875sub show_commit {
2876        my $c = shift;
2877        if ($oneline) {
2878                my $x = "\n";
2879                if (my $l = $c->{l}) {
2880                        while ($l->[0] =~ /^\s*$/) { shift @$l }
2881                        $x = $l->[0];
2882                }
2883                $l_fmt ||= 'A' . length($c->{r});
2884                print 'r',pack($l_fmt, $c->{r}),' | ';
2885                print "$c->{c} | " if $show_commit;
2886                print $x;
2887        } else {
2888                show_commit_normal($c);
2889        }
2890}
2891
2892sub show_commit_changed_paths {
2893        my ($c) = @_;
2894        return unless $c->{changed};
2895        print "Changed paths:\n", @{$c->{changed}};
2896}
2897
2898sub show_commit_normal {
2899        my ($c) = @_;
2900        print '-' x72, "\nr$c->{r} | ";
2901        print "$c->{c} | " if $show_commit;
2902        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2903                                 localtime($c->{t_utc})), ' | ';
2904        my $nr_line = 0;
2905
2906        if (my $l = $c->{l}) {
2907                while ($l->[$#$l] eq "\n" && $#$l > 0
2908                                          && $l->[($#$l - 1)] eq "\n") {
2909                        pop @$l;
2910                }
2911                $nr_line = scalar @$l;
2912                if (!$nr_line) {
2913                        print "1 line\n\n\n";
2914                } else {
2915                        if ($nr_line == 1) {
2916                                $nr_line = '1 line';
2917                        } else {
2918                                $nr_line .= ' lines';
2919                        }
2920                        print $nr_line, "\n";
2921                        show_commit_changed_paths($c);
2922                        print "\n";
2923                        print $_ foreach @$l;
2924                }
2925        } else {
2926                print "1 line\n";
2927                show_commit_changed_paths($c);
2928                print "\n";
2929
2930        }
2931        foreach my $x (qw/raw diff/) {
2932                if ($c->{$x}) {
2933                        print "\n";
2934                        print $_ foreach @{$c->{$x}}
2935                }
2936        }
2937}
2938
2939sub cmd_show_log {
2940        my (@args) = @_;
2941        my ($r_min, $r_max);
2942        my $r_last = -1; # prevent dupes
2943        if (defined $TZ) {
2944                $ENV{TZ} = $TZ;
2945        } else {
2946                delete $ENV{TZ};
2947        }
2948        if (defined $::_revision) {
2949                if ($::_revision =~ /^(\d+):(\d+)$/) {
2950                        ($r_min, $r_max) = ($1, $2);
2951                } elsif ($::_revision =~ /^\d+$/) {
2952                        $r_min = $r_max = $::_revision;
2953                } else {
2954                        ::fatal "-r$::_revision is not supported, use ",
2955                                "standard \'git log\' arguments instead\n";
2956                }
2957        }
2958
2959        config_pager();
2960        @args = (git_svn_log_cmd($r_min, $r_max), @args);
2961        my $log = command_output_pipe(@args);
2962        run_pager();
2963        my (@k, $c, $d);
2964        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
2965        while (<$log>) {
2966                if (/^${esc_color}commit ($::sha1_short)/o) {
2967                        my $cmt = $1;
2968                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
2969                                $r_last = $c->{r};
2970                                process_commit($c, $r_min, $r_max, \@k) or
2971                                                                goto out;
2972                        }
2973                        $d = undef;
2974                        $c = { c => $cmt };
2975                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
2976                        get_author_info($c, $1, $2, $3);
2977                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
2978                        # ignore
2979                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
2980                        push @{$c->{raw}}, $_;
2981                } elsif (/^${esc_color}[ACRMDT]\t/) {
2982                        # we could add $SVN->{svn_path} here, but that requires
2983                        # remote access at the moment (repo_path_split)...
2984                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
2985                        push @{$c->{changed}}, $_;
2986                } elsif (/^${esc_color}diff /o) {
2987                        $d = 1;
2988                        push @{$c->{diff}}, $_;
2989                } elsif ($d) {
2990                        push @{$c->{diff}}, $_;
2991                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
2992                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
2993                } elsif (s/^${esc_color}    //o) {
2994                        push @{$c->{l}}, $_;
2995                }
2996        }
2997        if ($c && defined $c->{r} && $c->{r} != $r_last) {
2998                $r_last = $c->{r};
2999                process_commit($c, $r_min, $r_max, \@k);
3000        }
3001        if (@k) {
3002                my $swap = $r_max;
3003                $r_max = $r_min;
3004                $r_min = $swap;
3005                process_commit($_, $r_min, $r_max) foreach reverse @k;
3006        }
3007out:
3008        close $log;
3009        print '-' x72,"\n" unless $incremental || $oneline;
3010}
3011
3012package Git::SVN::Migration;
3013# these version numbers do NOT correspond to actual version numbers
3014# of git nor git-svn.  They are just relative.
3015#
3016# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3017#
3018# v1 layout: .git/$id/info/url, refs/remotes/$id
3019#
3020# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3021#
3022# v3 layout: .git/svn/$id, refs/remotes/$id
3023#            - info/url may remain for backwards compatibility
3024#            - this is what we migrate up to this layout automatically,
3025#            - this will be used by git svn init on single branches
3026#
3027# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3028#            - this is only created for newly multi-init-ed
3029#              repositories.  Similar in spirit to the
3030#              --use-separate-remotes option in git-clone (now default)
3031#            - we do not automatically migrate to this (following
3032#              the example set by core git)
3033use strict;
3034use warnings;
3035use Carp qw/croak/;
3036use File::Path qw/mkpath/;
3037use File::Basename qw/dirname basename/;
3038use vars qw/$_minimize/;
3039
3040sub migrate_from_v0 {
3041        my $git_dir = $ENV{GIT_DIR};
3042        return undef unless -d $git_dir;
3043        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3044        my $migrated = 0;
3045        while (<$fh>) {
3046                chomp;
3047                my ($id, $orig_ref) = ($_, $_);
3048                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3049                next unless -f "$git_dir/$id/info/url";
3050                my $new_ref = "refs/remotes/$id";
3051                if (::verify_ref("$new_ref^0")) {
3052                        print STDERR "W: $orig_ref is probably an old ",
3053                                     "branch used by an ancient version of ",
3054                                     "git-svn.\n",
3055                                     "However, $new_ref also exists.\n",
3056                                     "We will not be able ",
3057                                     "to use this branch until this ",
3058                                     "ambiguity is resolved.\n";
3059                        next;
3060                }
3061                print STDERR "Migrating from v0 layout...\n" if !$migrated;
3062                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3063                command_noisy('update-ref', $new_ref, $orig_ref);
3064                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3065                $migrated++;
3066        }
3067        command_close_pipe($fh, $ctx);
3068        print STDERR "Done migrating from v0 layout...\n" if $migrated;
3069        $migrated;
3070}
3071
3072sub migrate_from_v1 {
3073        my $git_dir = $ENV{GIT_DIR};
3074        my $migrated = 0;
3075        return $migrated unless -d $git_dir;
3076        my $svn_dir = "$git_dir/svn";
3077
3078        # just in case somebody used 'svn' as their $id at some point...
3079        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3080
3081        print STDERR "Migrating from a git-svn v1 layout...\n";
3082        mkpath([$svn_dir]);
3083        print STDERR "Data from a previous version of git-svn exists, but\n\t",
3084                     "$svn_dir\n\t(required for this version ",
3085                     "($::VERSION) of git-svn) does not. exist\n";
3086        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3087        while (<$fh>) {
3088                my $x = $_;
3089                next unless $x =~ s#^refs/remotes/##;
3090                chomp $x;
3091                next unless -f "$git_dir/$x/info/url";
3092                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3093                next unless $u;
3094                my $dn = dirname("$git_dir/svn/$x");
3095                mkpath([$dn]) unless -d $dn;
3096                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3097                        mkpath(["$git_dir/svn/svn"]);
3098                        print STDERR " - $git_dir/$x/info => ",
3099                                        "$git_dir/svn/$x/info\n";
3100                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3101                               croak "$!: $x";
3102                        # don't worry too much about these, they probably
3103                        # don't exist with repos this old (save for index,
3104                        # and we can easily regenerate that)
3105                        foreach my $f (qw/unhandled.log index .rev_db/) {
3106                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3107                        }
3108                } else {
3109                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3110                        rename "$git_dir/$x", "$git_dir/svn/$x" or
3111                               croak "$!: $x";
3112                }
3113                $migrated++;
3114        }
3115        command_close_pipe($fh, $ctx);
3116        print STDERR "Done migrating from a git-svn v1 layout\n";
3117        $migrated;
3118}
3119
3120sub read_old_urls {
3121        my ($l_map, $pfx, $path) = @_;
3122        my @dir;
3123        foreach (<$path/*>) {
3124                if (-r "$_/info/url") {
3125                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3126                        my $ref_id = $pfx . basename $_;
3127                        my $url = ::file_to_s("$_/info/url");
3128                        $l_map->{$ref_id} = $url;
3129                } elsif (-d $_) {
3130                        push @dir, $_;
3131                }
3132        }
3133        foreach (@dir) {
3134                my $x = $_;
3135                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3136                read_old_urls($l_map, $x, $_);
3137        }
3138}
3139
3140sub migrate_from_v2 {
3141        my @cfg = command(qw/config -l/);
3142        return if grep /^svn-remote\..+\.url=/, @cfg;
3143        my %l_map;
3144        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3145        my $migrated = 0;
3146
3147        foreach my $ref_id (sort keys %l_map) {
3148                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3149                if ($@) {
3150                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3151                }
3152                $migrated++;
3153        }
3154        $migrated;
3155}
3156
3157sub minimize_connections {
3158        my $r = Git::SVN::read_all_remotes();
3159        my $new_urls = {};
3160        my $root_repos = {};
3161        foreach my $repo_id (keys %$r) {
3162                my $url = $r->{$repo_id}->{url} or next;
3163                my $fetch = $r->{$repo_id}->{fetch} or next;
3164                my $ra = Git::SVN::Ra->new($url);
3165
3166                # skip existing cases where we already connect to the root
3167                if (($ra->{url} eq $ra->{repos_root}) ||
3168                    (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3169                     $repo_id)) {
3170                        $root_repos->{$ra->{url}} = $repo_id;
3171                        next;
3172                }
3173
3174                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3175                my $root_path = $ra->{url};
3176                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3177                foreach my $path (keys %$fetch) {
3178                        my $ref_id = $fetch->{$path};
3179                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3180
3181                        # make sure we can read when connecting to
3182                        # a higher level of a repository
3183                        my ($last_rev, undef) = $gs->last_rev_commit;
3184                        if (!defined $last_rev) {
3185                                $last_rev = eval {
3186                                        $root_ra->get_latest_revnum;
3187                                };
3188                                next if $@;
3189                        }
3190                        my $new = $root_path;
3191                        $new .= length $path ? "/$path" : '';
3192                        eval {
3193                                $root_ra->get_log([$new], $last_rev, $last_rev,
3194                                                  0, 0, 1, sub { });
3195                        };
3196                        next if $@;
3197                        $new_urls->{$ra->{repos_root}}->{$new} =
3198                                { ref_id => $ref_id,
3199                                  old_repo_id => $repo_id,
3200                                  old_path => $path };
3201                }
3202        }
3203
3204        my @emptied;
3205        foreach my $url (keys %$new_urls) {
3206                # see if we can re-use an existing [svn-remote "repo_id"]
3207                # instead of creating a(n ugly) new section:
3208                my $repo_id = $root_repos->{$url} ||
3209                              Git::SVN::sanitize_remote_name($url);
3210
3211                my $fetch = $new_urls->{$url};
3212                foreach my $path (keys %$fetch) {
3213                        my $x = $fetch->{$path};
3214                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3215                        my $pfx = "svn-remote.$x->{old_repo_id}";
3216
3217                        my $old_fetch = quotemeta("$x->{old_path}:".
3218                                                  "refs/remotes/$x->{ref_id}");
3219                        command_noisy(qw/config --unset/,
3220                                      "$pfx.fetch", '^'. $old_fetch . '$');
3221                        delete $r->{$x->{old_repo_id}}->
3222                               {fetch}->{$x->{old_path}};
3223                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3224                                command_noisy(qw/config --unset/,
3225                                              "$pfx.url");
3226                                push @emptied, $x->{old_repo_id}
3227                        }
3228                }
3229        }
3230        if (@emptied) {
3231                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3232                           "$ENV{GIT_DIR}/config";
3233                print STDERR <<EOF;
3234The following [svn-remote] sections in your config file ($file) are empty
3235and can be safely removed:
3236EOF
3237                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3238        }
3239}
3240
3241sub migration_check {
3242        migrate_from_v0();
3243        migrate_from_v1();
3244        migrate_from_v2();
3245        minimize_connections() if $_minimize;
3246}
3247
3248package Git::IndexInfo;
3249use strict;
3250use warnings;
3251use Git qw/command_input_pipe command_close_pipe/;
3252
3253sub new {
3254        my ($class) = @_;
3255        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3256        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3257}
3258
3259sub remove {
3260        my ($self, $path) = @_;
3261        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3262                return ++$self->{nr};
3263        }
3264        undef;
3265}
3266
3267sub update {
3268        my ($self, $mode, $hash, $path) = @_;
3269        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3270                return ++$self->{nr};
3271        }
3272        undef;
3273}
3274
3275sub DESTROY {
3276        my ($self) = @_;
3277        command_close_pipe($self->{gui}, $self->{ctx});
3278}
3279
3280package Git::SVN::GlobSpec;
3281use strict;
3282use warnings;
3283
3284sub new {
3285        my ($class, $glob) = @_;
3286        my $re = $glob;
3287        $re =~ s!/+$!!g; # no need for trailing slashes
3288        my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
3289        my ($left, $right) = ($1, $2);
3290        if ($nr > 1) {
3291                die "Only one '*' wildcard expansion ",
3292                    "is supported (got $nr): '$glob'\n";
3293        } elsif ($nr == 0) {
3294                die "One '*' is needed for glob: '$glob'\n";
3295        }
3296        $re = quotemeta($left) . $re . quotemeta($right);
3297        if (length $left && !($left =~ s!/+$!!g)) {
3298                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
3299        }
3300        if (length $right && !($right =~ s!^/+!!g)) {
3301                die "Missing leading '/' on right side of: '$glob' ($right)\n";
3302        }
3303        my $left_re = qr/^\/\Q$left\E(\/|$)/;
3304        bless { left => $left, right => $right, left_regex => $left_re,
3305                regex => qr/$re/, glob => $glob }, $class;
3306}
3307
3308sub full_path {
3309        my ($self, $path) = @_;
3310        return (length $self->{left} ? "$self->{left}/" : '') .
3311               $path . (length $self->{right} ? "/$self->{right}" : '');
3312}
3313
3314__END__
3315
3316Data structures:
3317
3318
3319$remotes = { # returned by read_all_remotes()
3320        'svn' => {
3321                # svn-remote.svn.url=https://svn.musicpd.org
3322                url => 'https://svn.musicpd.org',
3323                # svn-remote.svn.fetch=mpd/trunk:trunk
3324                fetch => {
3325                        'mpd/trunk' => 'trunk',
3326                },
3327                # svn-remote.svn.tags=mpd/tags/*:tags/*
3328                tags => {
3329                        path => {
3330                                left => 'mpd/tags',
3331                                right => '',
3332                                regex => qr!mpd/tags/([^/]+)$!,
3333                                glob => 'tags/*',
3334                        },
3335                        ref => {
3336                                left => 'tags',
3337                                right => '',
3338                                regex => qr!tags/([^/]+)$!,
3339                                glob => 'tags/*',
3340                        },
3341                }
3342        }
3343};
3344
3345$log_entry hashref as returned by libsvn_log_entry()
3346{
3347        log => 'whitespace-formatted log entry
3348',                                              # trailing newline is preserved
3349        revision => '8',                        # integer
3350        date => '2004-02-24T17:01:44.108345Z',  # commit date
3351        author => 'committer name'
3352};
3353
3354
3355# this is generated by generate_diff();
3356@mods = array of diff-index line hashes, each element represents one line
3357        of diff-index output
3358
3359diff-index line ($m hash)
3360{
3361        mode_a => first column of diff-index output, no leading ':',
3362        mode_b => second column of diff-index output,
3363        sha1_b => sha1sum of the final blob,
3364        chg => change type [MCRADT],
3365        file_a => original file name of a file (iff chg is 'C' or 'R')
3366        file_b => new/current file name of a file (any chg)
3367}
3368;
3369
3370# retval of read_url_paths{,_all}();
3371$l_map = {
3372        # repository root url
3373        'https://svn.musicpd.org' => {
3374                # repository path               # GIT_SVN_ID
3375                'mpd/trunk'             =>      'trunk',
3376                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3377        },
3378}
3379
3380Notes:
3381        I don't trust the each() function on unless I created %hash myself
3382        because the internal iterator may not have started at base.