Andrew's git - gitweb.git/blob

   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $sha1 $sha1_short $_revision
   8                $_q $_authors %users/;
   9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  10$VERSION = '@@GIT_VERSION@@';
  11
  12my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
  13$ENV{GIT_DIR} ||= '.git';
  14$Git::SVN::default_repo_id = 'svn';
  15$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  16$Git::SVN::Ra::_log_window_size = 100;
  17
  18$Git::SVN::Log::TZ = $ENV{TZ};
  19$ENV{TZ} = 'UTC';
  20$| = 1; # unbuffer STDOUT
  21
  22sub fatal (@) { print STDERR @_; exit 1 }
  23require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  24require SVN::Ra;
  25require SVN::Delta;
  26if ($SVN::Core::VERSION lt '1.1.0') {
  27        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
  28}
  29push @Git::SVN::Ra::ISA, 'SVN::Ra';
  30push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  31push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  32use Carp qw/croak/;
  33use IO::File qw//;
  34use File::Basename qw/dirname basename/;
  35use File::Path qw/mkpath/;
  36use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  37use IPC::Open3;
  38use Git;
  39
  40BEGIN {
  41        my $s;
  42        foreach (qw/command command_oneline command_noisy command_output_pipe
  43                    command_input_pipe command_close_pipe/) {
  44                $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
  45                      "*Git::SVN::Migration::$_ = ".
  46                      "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
  47        }
  48        eval $s;
  49}
  50
  51my ($SVN);
  52
  53$sha1 = qr/[a-f\d]{40}/;
  54$sha1_short = qr/[a-f\d]{4,40}/;
  55my ($_stdin, $_help, $_edit,
  56        $_message, $_file,
  57        $_template, $_shared,
  58        $_version, $_fetch_all,
  59        $_merge, $_strategy, $_dry_run, $_local,
  60        $_prefix, $_no_checkout, $_verbose);
  61$Git::SVN::_follow_parent = 1;
  62my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  63                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  64                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  65my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  66                'authors-file|A=s' => \$_authors,
  67                'repack:i' => \$Git::SVN::_repack,
  68                'noMetadata' => \$Git::SVN::_no_metadata,
  69                'useSvmProps' => \$Git::SVN::_use_svm_props,
  70                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  71                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  72                'no-checkout' => \$_no_checkout,
  73                'quiet|q' => \$_q,
  74                'repack-flags|repack-args|repack-opts=s' =>
  75                   \$Git::SVN::_repack_flags,
  76                %remote_opts );
  77
  78my ($_trunk, $_tags, $_branches);
  79my %icv;
  80my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
  81                  'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
  82                  'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
  83                  'no-metadata' => sub { $icv{noMetadata} = 1 },
  84                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
  85                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
  86                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
  87                  %remote_opts );
  88my %cmt_opts = ( 'edit|e' => \$_edit,
  89                'rmdir' => \$SVN::Git::Editor::_rmdir,
  90                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
  91                'l=i' => \$SVN::Git::Editor::_rename_limit,
  92                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
  93);
  94
  95my %cmd = (
  96        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
  97                        { 'revision|r=s' => \$_revision,
  98                          'fetch-all|all' => \$_fetch_all,
  99                           %fc_opts } ],
 100        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 101                        { 'revision|r=s' => \$_revision,
 102                           %fc_opts, %init_opts } ],
 103        init => [ \&cmd_init, "Initialize a repo for tracking" .
 104                          " (requires URL argument)",
 105                          \%init_opts ],
 106        'multi-init' => [ \&cmd_multi_init,
 107                          "Deprecated alias for ".
 108                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 109                          \%init_opts ],
 110        dcommit => [ \&cmd_dcommit,
 111                     'Commit several diffs to merge with upstream',
 112                        { 'merge|m|M' => \$_merge,
 113                          'strategy|s=s' => \$_strategy,
 114                          'verbose|v' => \$_verbose,
 115                          'dry-run|n' => \$_dry_run,
 116                          'fetch-all|all' => \$_fetch_all,
 117                        %cmt_opts, %fc_opts } ],
 118        'set-tree' => [ \&cmd_set_tree,
 119                        "Set an SVN repository to a git tree-ish",
 120                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 121        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 122                        { 'revision|r=i' => \$_revision } ],
 123        'multi-fetch' => [ \&cmd_multi_fetch,
 124                           "Deprecated alias for $0 fetch --all",
 125                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 126        'migrate' => [ sub { },
 127                       # no-op, we automatically run this anyways,
 128                       'Migrate configuration/metadata/layout from
 129                        previous versions of git-svn',
 130                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 131                         %remote_opts } ],
 132        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 133                        { 'limit=i' => \$Git::SVN::Log::limit,
 134                          'revision|r=s' => \$_revision,
 135                          'verbose|v' => \$Git::SVN::Log::verbose,
 136                          'incremental' => \$Git::SVN::Log::incremental,
 137                          'oneline' => \$Git::SVN::Log::oneline,
 138                          'show-commit' => \$Git::SVN::Log::show_commit,
 139                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 140                          'authors-file|A=s' => \$_authors,
 141                          'color' => \$Git::SVN::Log::color,
 142                          'pager=s' => \$Git::SVN::Log::pager,
 143                        } ],
 144        'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
 145                        { } ],
 146        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 147                        { 'merge|m|M' => \$_merge,
 148                          'verbose|v' => \$_verbose,
 149                          'strategy|s=s' => \$_strategy,
 150                          'local|l' => \$_local,
 151                          'fetch-all|all' => \$_fetch_all,
 152                          %fc_opts } ],
 153        'commit-diff' => [ \&cmd_commit_diff,
 154                           'Commit a diff between two trees',
 155                        { 'message|m=s' => \$_message,
 156                          'file|F=s' => \$_file,
 157                          'revision|r=s' => \$_revision,
 158                        %cmt_opts } ],
 159);
 160
 161my $cmd;
 162for (my $i = 0; $i < @ARGV; $i++) {
 163        if (defined $cmd{$ARGV[$i]}) {
 164                $cmd = $ARGV[$i];
 165                splice @ARGV, $i, 1;
 166                last;
 167        }
 168};
 169
 170my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 171
 172read_repo_config(\%opts);
 173Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
 174my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 175                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 176                    'id|i=s' => \$Git::SVN::default_ref_id,
 177                    'svn-remote|remote|R=s' => sub {
 178                       $Git::SVN::no_reuse_existing = 1;
 179                       $Git::SVN::default_repo_id = $_[1] });
 180exit 1 if (!$rv && $cmd && $cmd ne 'log');
 181
 182usage(0) if $_help;
 183version() if $_version;
 184usage(1) unless defined $cmd;
 185load_authors() if $_authors;
 186
 187# make sure we're always running
 188unless ($cmd =~ /(?:clone|init|multi-init)$/) {
 189        unless (-d $ENV{GIT_DIR}) {
 190                if ($git_dir_user_set) {
 191                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 192                            "but it is not a directory\n";
 193                }
 194                my $git_dir = delete $ENV{GIT_DIR};
 195                chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
 196                unless (length $cdup) {
 197                        die "Already at toplevel, but $git_dir ",
 198                            "not found '$cdup'\n";
 199                }
 200                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 201                unless (-d $git_dir) {
 202                        die "$git_dir still not found after going to ",
 203                            "'$cdup'\n";
 204                }
 205                $ENV{GIT_DIR} = $git_dir;
 206        }
 207}
 208unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 209        Git::SVN::Migration::migration_check();
 210}
 211Git::SVN::init_vars();
 212eval {
 213        Git::SVN::verify_remotes_sanity();
 214        $cmd{$cmd}->[0]->(@ARGV);
 215};
 216fatal $@ if $@;
 217post_fetch_checkout();
 218exit 0;
 219
 220####################### primary functions ######################
 221sub usage {
 222        my $exit = shift || 0;
 223        my $fd = $exit ? \*STDERR : \*STDOUT;
 224        print $fd <<"";
 225git-svn - bidirectional operations between a single Subversion tree and git
 226Usage: $0 <command> [options] [arguments]\n
 227
 228        print $fd "Available commands:\n" unless $cmd;
 229
 230        foreach (sort keys %cmd) {
 231                next if $cmd && $cmd ne $_;
 232                next if /^multi-/; # don't show deprecated commands
 233                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 234                foreach (keys %{$cmd{$_}->[2]}) {
 235                        # mixed-case options are for .git/config only
 236                        next if /[A-Z]/ && /^[a-z]+$/i;
 237                        # prints out arguments as they should be passed:
 238                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 239                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 240                                                        "--$_" : "-$_" }
 241                                                split /\|/,$_)," $x\n";
 242                }
 243        }
 244        print $fd <<"";
 245\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 246arbitrary identifier if you're tracking multiple SVN branches/repositories in
 247one git repository and want to keep them separate.  See git-svn(1) for more
 248information.
 249
 250        exit $exit;
 251}
 252
 253sub version {
 254        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 255        exit 0;
 256}
 257
 258sub do_git_init_db {
 259        unless (-d $ENV{GIT_DIR}) {
 260                my @init_db = ('init');
 261                push @init_db, "--template=$_template" if defined $_template;
 262                if (defined $_shared) {
 263                        if ($_shared =~ /[a-z]/) {
 264                                push @init_db, "--shared=$_shared";
 265                        } else {
 266                                push @init_db, "--shared";
 267                        }
 268                }
 269                command_noisy(@init_db);
 270        }
 271        my $set;
 272        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 273        foreach my $i (keys %icv) {
 274                die "'$set' and '$i' cannot both be set\n" if $set;
 275                next unless defined $icv{$i};
 276                command_noisy('config', "$pfx.$i", $icv{$i});
 277                $set = $i;
 278        }
 279}
 280
 281sub init_subdir {
 282        my $repo_path = shift or return;
 283        mkpath([$repo_path]) unless -d $repo_path;
 284        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 285        $ENV{GIT_DIR} = '.git';
 286}
 287
 288sub cmd_clone {
 289        my ($url, $path) = @_;
 290        if (!defined $path &&
 291            (defined $_trunk || defined $_branches || defined $_tags) &&
 292            $url !~ m#^[a-z\+]+://#) {
 293                $path = $url;
 294        }
 295        $path = basename($url) if !defined $path || !length $path;
 296        cmd_init($url, $path);
 297        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 298}
 299
 300sub cmd_init {
 301        if (defined $_trunk || defined $_branches || defined $_tags) {
 302                return cmd_multi_init(@_);
 303        }
 304        my $url = shift or die "SVN repository location required ",
 305                               "as a command-line argument\n";
 306        init_subdir(@_);
 307        do_git_init_db();
 308
 309        Git::SVN->init($url);
 310}
 311
 312sub cmd_fetch {
 313        if (grep /^\d+=./, @_) {
 314                die "'<rev>=<commit>' fetch arguments are ",
 315                    "no longer supported.\n";
 316        }
 317        my ($remote) = @_;
 318        if (@_ > 1) {
 319                die "Usage: $0 fetch [--all] [svn-remote]\n";
 320        }
 321        $remote ||= $Git::SVN::default_repo_id;
 322        if ($_fetch_all) {
 323                cmd_multi_fetch();
 324        } else {
 325                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 326        }
 327}
 328
 329sub cmd_set_tree {
 330        my (@commits) = @_;
 331        if ($_stdin || !@commits) {
 332                print "Reading from stdin...\n";
 333                @commits = ();
 334                while (<STDIN>) {
 335                        if (/\b($sha1_short)\b/o) {
 336                                unshift @commits, $1;
 337                        }
 338                }
 339        }
 340        my @revs;
 341        foreach my $c (@commits) {
 342                my @tmp = command('rev-parse',$c);
 343                if (scalar @tmp == 1) {
 344                        push @revs, $tmp[0];
 345                } elsif (scalar @tmp > 1) {
 346                        push @revs, reverse(command('rev-list',@tmp));
 347                } else {
 348                        fatal "Failed to rev-parse $c\n";
 349                }
 350        }
 351        my $gs = Git::SVN->new;
 352        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 353        $gs->fetch;
 354        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 355                fatal "There are new revisions that were fetched ",
 356                      "and need to be merged (or acknowledged) ",
 357                      "before committing.\nlast rev: $r_last\n",
 358                      " current: $gs->{last_rev}\n";
 359        }
 360        $gs->set_tree($_) foreach @revs;
 361        print "Done committing ",scalar @revs," revisions to SVN\n";
 362}
 363
 364sub cmd_dcommit {
 365        my $head = shift;
 366        $head ||= 'HEAD';
 367        my @refs;
 368        my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
 369        unless ($gs) {
 370                die "Unable to determine upstream SVN information from ",
 371                    "$head history\n";
 372        }
 373        my $c = $refs[-1];
 374        my $last_rev;
 375        foreach my $d (@refs) {
 376                if (!verify_ref("$d~1")) {
 377                        fatal "Commit $d\n",
 378                              "has no parent commit, and therefore ",
 379                              "nothing to diff against.\n",
 380                              "You should be working from a repository ",
 381                              "originally created by git-svn\n";
 382                }
 383                unless (defined $last_rev) {
 384                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 385                        unless (defined $last_rev) {
 386                                fatal "Unable to extract revision information ",
 387                                      "from commit $d~1\n";
 388                        }
 389                }
 390                if ($_dry_run) {
 391                        print "diff-tree $d~1 $d\n";
 392                } else {
 393                        my %ed_opts = ( r => $last_rev,
 394                                        log => get_commit_entry($d)->{log},
 395                                        ra => Git::SVN::Ra->new($url),
 396                                        tree_a => "$d~1",
 397                                        tree_b => $d,
 398                                        editor_cb => sub {
 399                                               print "Committed r$_[0]\n";
 400                                               $last_rev = $_[0]; },
 401                                        svn_path => '');
 402                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 403                                print "No changes\n$d~1 == $d\n";
 404                        }
 405                }
 406        }
 407        return if $_dry_run;
 408        unless ($gs) {
 409                warn "Could not determine fetch information for $url\n",
 410                     "Will not attempt to fetch and rebase commits.\n",
 411                     "This probably means you have useSvmProps and should\n",
 412                     "now resync your SVN::Mirror repository.\n";
 413                return;
 414        }
 415        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 416        # we always want to rebase against the current HEAD, not any
 417        # head that was passed to us
 418        my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
 419        my @finish;
 420        if (@diff) {
 421                @finish = rebase_cmd();
 422                print STDERR "W: HEAD and ", $gs->refname, " differ, ",
 423                             "using @finish:\n", "@diff";
 424        } else {
 425                print "No changes between current HEAD and ",
 426                      $gs->refname, "\nResetting to the latest ",
 427                      $gs->refname, "\n";
 428                @finish = qw/reset --mixed/;
 429        }
 430        command_noisy(@finish, $gs->refname);
 431}
 432
 433sub cmd_find_rev {
 434        my $revision_or_hash = shift;
 435        my $result;
 436        if ($revision_or_hash =~ /^r\d+$/) {
 437                my $head = shift;
 438                $head ||= 'HEAD';
 439                my @refs;
 440                my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
 441                unless ($gs) {
 442                        die "Unable to determine upstream SVN information from ",
 443                            "$head history\n";
 444                }
 445                my $desired_revision = substr($revision_or_hash, 1);
 446                $result = $gs->rev_db_get($desired_revision);
 447        } else {
 448                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 449                $result = $rev;
 450        }
 451        print "$result\n" if $result;
 452}
 453
 454sub cmd_rebase {
 455        command_noisy(qw/update-index --refresh/);
 456        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 457        unless ($gs) {
 458                die "Unable to determine upstream SVN information from ",
 459                    "working tree history\n";
 460        }
 461        if (command(qw/diff-index HEAD --/)) {
 462                print STDERR "Cannot rebase with uncommited changes:\n";
 463                command_noisy('status');
 464                exit 1;
 465        }
 466        unless ($_local) {
 467                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 468        }
 469        command_noisy(rebase_cmd(), $gs->refname);
 470}
 471
 472sub cmd_show_ignore {
 473        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 474        $gs ||= Git::SVN->new;
 475        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 476        $gs->traverse_ignore(\*STDOUT, $gs->{path}, $r);
 477}
 478
 479sub cmd_multi_init {
 480        my $url = shift;
 481        unless (defined $_trunk || defined $_branches || defined $_tags) {
 482                usage(1);
 483        }
 484        $_prefix = '' unless defined $_prefix;
 485        if (defined $url) {
 486                $url =~ s#/+$##;
 487                init_subdir(@_);
 488        }
 489        do_git_init_db();
 490        if (defined $_trunk) {
 491                my $trunk_ref = $_prefix . 'trunk';
 492                # try both old-style and new-style lookups:
 493                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 494                unless ($gs_trunk) {
 495                        my ($trunk_url, $trunk_path) =
 496                                              complete_svn_url($url, $_trunk);
 497                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 498                                                   undef, $trunk_ref);
 499                }
 500        }
 501        return unless defined $_branches || defined $_tags;
 502        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 503        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 504        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 505}
 506
 507sub cmd_multi_fetch {
 508        my $remotes = Git::SVN::read_all_remotes();
 509        foreach my $repo_id (sort keys %$remotes) {
 510                if ($remotes->{$repo_id}->{url}) {
 511                        Git::SVN::fetch_all($repo_id, $remotes);
 512                }
 513        }
 514}
 515
 516# this command is special because it requires no metadata
 517sub cmd_commit_diff {
 518        my ($ta, $tb, $url) = @_;
 519        my $usage = "Usage: $0 commit-diff -r<revision> ".
 520                    "<tree-ish> <tree-ish> [<URL>]\n";
 521        fatal($usage) if (!defined $ta || !defined $tb);
 522        my $svn_path;
 523        if (!defined $url) {
 524                my $gs = eval { Git::SVN->new };
 525                if (!$gs) {
 526                        fatal("Needed URL or usable git-svn --id in ",
 527                              "the command-line\n", $usage);
 528                }
 529                $url = $gs->{url};
 530                $svn_path = $gs->{path};
 531        }
 532        unless (defined $_revision) {
 533                fatal("-r|--revision is a required argument\n", $usage);
 534        }
 535        if (defined $_message && defined $_file) {
 536                fatal("Both --message/-m and --file/-F specified ",
 537                      "for the commit message.\n",
 538                      "I have no idea what you mean\n");
 539        }
 540        if (defined $_file) {
 541                $_message = file_to_s($_file);
 542        } else {
 543                $_message ||= get_commit_entry($tb)->{log};
 544        }
 545        my $ra ||= Git::SVN::Ra->new($url);
 546        $svn_path ||= $ra->{svn_path};
 547        my $r = $_revision;
 548        if ($r eq 'HEAD') {
 549                $r = $ra->get_latest_revnum;
 550        } elsif ($r !~ /^\d+$/) {
 551                die "revision argument: $r not understood by git-svn\n";
 552        }
 553        my %ed_opts = ( r => $r,
 554                        log => $_message,
 555                        ra => $ra,
 556                        tree_a => $ta,
 557                        tree_b => $tb,
 558                        editor_cb => sub { print "Committed r$_[0]\n" },
 559                        svn_path => $svn_path );
 560        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 561                print "No changes\n$ta == $tb\n";
 562        }
 563}
 564
 565########################### utility functions #########################
 566
 567sub rebase_cmd {
 568        my @cmd = qw/rebase/;
 569        push @cmd, '-v' if $_verbose;
 570        push @cmd, qw/--merge/ if $_merge;
 571        push @cmd, "--strategy=$_strategy" if $_strategy;
 572        @cmd;
 573}
 574
 575sub post_fetch_checkout {
 576        return if $_no_checkout;
 577        my $gs = $Git::SVN::_head or return;
 578        return if verify_ref('refs/heads/master^0');
 579
 580        my $valid_head = verify_ref('HEAD^0');
 581        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
 582        return if ($valid_head || !verify_ref('HEAD^0'));
 583
 584        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
 585        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
 586        return if -f $index;
 587
 588        chomp(my $bare = `git config --bool --get core.bare`);
 589        return if $bare eq 'true';
 590        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
 591        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
 592        print STDERR "Checked out HEAD:\n  ",
 593                     $gs->full_url, " r", $gs->last_rev, "\n";
 594}
 595
 596sub complete_svn_url {
 597        my ($url, $path) = @_;
 598        $path =~ s#/+$##;
 599        if ($path !~ m#^[a-z\+]+://#) {
 600                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 601                        fatal("E: '$path' is not a complete URL ",
 602                              "and a separate URL is not specified\n");
 603                }
 604                return ($url, $path);
 605        }
 606        return ($path, '');
 607}
 608
 609sub complete_url_ls_init {
 610        my ($ra, $repo_path, $switch, $pfx) = @_;
 611        unless ($repo_path) {
 612                print STDERR "W: $switch not specified\n";
 613                return;
 614        }
 615        $repo_path =~ s#/+$##;
 616        if ($repo_path =~ m#^[a-z\+]+://#) {
 617                $ra = Git::SVN::Ra->new($repo_path);
 618                $repo_path = '';
 619        } else {
 620                $repo_path =~ s#^/+##;
 621                unless ($ra) {
 622                        fatal("E: '$repo_path' is not a complete URL ",
 623                              "and a separate URL is not specified\n");
 624                }
 625        }
 626        my $url = $ra->{url};
 627        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
 628        my $k = "svn-remote.$gs->{repo_id}.url";
 629        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
 630        if ($orig_url && ($orig_url ne $gs->{url})) {
 631                die "$k already set: $orig_url\n",
 632                    "wanted to set to: $gs->{url}\n";
 633        }
 634        command_oneline('config', $k, $gs->{url}) unless $orig_url;
 635        my $remote_path = "$ra->{svn_path}/$repo_path/*";
 636        $remote_path =~ s#/+#/#g;
 637        $remote_path =~ s#^/##g;
 638        my ($n) = ($switch =~ /^--(\w+)/);
 639        if (length $pfx && $pfx !~ m#/$#) {
 640                die "--prefix='$pfx' must have a trailing slash '/'\n";
 641        }
 642        command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
 643                                "$remote_path:refs/remotes/$pfx*");
 644}
 645
 646sub verify_ref {
 647        my ($ref) = @_;
 648        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 649                               { STDERR => 0 }); };
 650}
 651
 652sub get_tree_from_treeish {
 653        my ($treeish) = @_;
 654        # $treeish can be a symbolic ref, too:
 655        my $type = command_oneline(qw/cat-file -t/, $treeish);
 656        my $expected;
 657        while ($type eq 'tag') {
 658                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
 659        }
 660        if ($type eq 'commit') {
 661                $expected = (grep /^tree /, command(qw/cat-file commit/,
 662                                                    $treeish))[0];
 663                ($expected) = ($expected =~ /^tree ($sha1)$/o);
 664                die "Unable to get tree from $treeish\n" unless $expected;
 665        } elsif ($type eq 'tree') {
 666                $expected = $treeish;
 667        } else {
 668                die "$treeish is a $type, expected tree, tag or commit\n";
 669        }
 670        return $expected;
 671}
 672
 673sub get_commit_entry {
 674        my ($treeish) = shift;
 675        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
 676        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
 677        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
 678        open my $log_fh, '>', $commit_editmsg or croak $!;
 679
 680        my $type = command_oneline(qw/cat-file -t/, $treeish);
 681        if ($type eq 'commit' || $type eq 'tag') {
 682                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
 683                                                         $type, $treeish);
 684                my $in_msg = 0;
 685                while (<$msg_fh>) {
 686                        if (!$in_msg) {
 687                                $in_msg = 1 if (/^\s*$/);
 688                        } elsif (/^git-svn-id: /) {
 689                                # skip this for now, we regenerate the
 690                                # correct one on re-fetch anyways
 691                                # TODO: set *:merge properties or like...
 692                        } else {
 693                                print $log_fh $_ or croak $!;
 694                        }
 695                }
 696                command_close_pipe($msg_fh, $ctx);
 697        }
 698        close $log_fh or croak $!;
 699
 700        if ($_edit || ($type eq 'tree')) {
 701                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 702                # TODO: strip out spaces, comments, like git-commit.sh
 703                system($editor, $commit_editmsg);
 704        }
 705        rename $commit_editmsg, $commit_msg or croak $!;
 706        open $log_fh, '<', $commit_msg or croak $!;
 707        { local $/; chomp($log_entry{log} = <$log_fh>); }
 708        close $log_fh or croak $!;
 709        unlink $commit_msg;
 710        \%log_entry;
 711}
 712
 713sub s_to_file {
 714        my ($str, $file, $mode) = @_;
 715        open my $fd,'>',$file or croak $!;
 716        print $fd $str,"\n" or croak $!;
 717        close $fd or croak $!;
 718        chmod ($mode &~ umask, $file) if (defined $mode);
 719}
 720
 721sub file_to_s {
 722        my $file = shift;
 723        open my $fd,'<',$file or croak "$!: file: $file\n";
 724        local $/;
 725        my $ret = <$fd>;
 726        close $fd or croak $!;
 727        $ret =~ s/\s*$//s;
 728        return $ret;
 729}
 730
 731# '<svn username> = real-name <email address>' mapping based on git-svnimport:
 732sub load_authors {
 733        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
 734        my $log = $cmd eq 'log';
 735        while (<$authors>) {
 736                chomp;
 737                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
 738                my ($user, $name, $email) = ($1, $2, $3);
 739                if ($log) {
 740                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
 741                } else {
 742                        $users{$user} = [$name, $email];
 743                }
 744        }
 745        close $authors or croak $!;
 746}
 747
 748# convert GetOpt::Long specs for use by git-config
 749sub read_repo_config {
 750        return unless -d $ENV{GIT_DIR};
 751        my $opts = shift;
 752        my @config_only;
 753        foreach my $o (keys %$opts) {
 754                # if we have mixedCase and a long option-only, then
 755                # it's a config-only variable that we don't need for
 756                # the command-line.
 757                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
 758                my $v = $opts->{$o};
 759                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
 760                $key =~ s/-//g;
 761                my $arg = 'git-config';
 762                $arg .= ' --int' if ($o =~ /[:=]i$/);
 763                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
 764                if (ref $v eq 'ARRAY') {
 765                        chomp(my @tmp = `$arg --get-all svn.$key`);
 766                        @$v = @tmp if @tmp;
 767                } else {
 768                        chomp(my $tmp = `$arg --get svn.$key`);
 769                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
 770                                $$v = $tmp;
 771                        }
 772                }
 773        }
 774        delete @$opts{@config_only} if @config_only;
 775}
 776
 777sub extract_metadata {
 778        my $id = shift or return (undef, undef, undef);
 779        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 780                                                        \s([a-f\d\-]+)$/x);
 781        if (!defined $rev || !$uuid || !$url) {
 782                # some of the original repositories I made had
 783                # identifiers like this:
 784                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
 785        }
 786        return ($url, $rev, $uuid);
 787}
 788
 789sub cmt_metadata {
 790        return extract_metadata((grep(/^git-svn-id: /,
 791                command(qw/cat-file commit/, shift)))[-1]);
 792}
 793
 794sub working_head_info {
 795        my ($head, $refs) = @_;
 796        my ($fh, $ctx) = command_output_pipe('rev-list', $head);
 797        while (my $hash = <$fh>) {
 798                chomp($hash);
 799                my ($url, $rev, $uuid) = cmt_metadata($hash);
 800                if (defined $url && defined $rev) {
 801                        if (my $gs = Git::SVN->find_by_url($url)) {
 802                                my $c = $gs->rev_db_get($rev);
 803                                if ($c && $c eq $hash) {
 804                                        close $fh; # break the pipe
 805                                        return ($url, $rev, $uuid, $gs);
 806                                }
 807                        }
 808                }
 809                unshift @$refs, $hash if $refs;
 810        }
 811        command_close_pipe($fh, $ctx);
 812        (undef, undef, undef, undef);
 813}
 814
 815package Git::SVN;
 816use strict;
 817use warnings;
 818use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
 819            $_repack $_repack_flags $_use_svm_props $_head
 820            $_use_svnsync_props $no_reuse_existing/;
 821use Carp qw/croak/;
 822use File::Path qw/mkpath/;
 823use File::Copy qw/copy/;
 824use IPC::Open3;
 825
 826my $_repack_nr;
 827# properties that we do not log:
 828my %SKIP_PROP;
 829BEGIN {
 830        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
 831                                        svn:special svn:executable
 832                                        svn:entry:committed-rev
 833                                        svn:entry:last-author
 834                                        svn:entry:uuid
 835                                        svn:entry:committed-date/;
 836
 837        # some options are read globally, but can be overridden locally
 838        # per [svn-remote "..."] section.  Command-line options will *NOT*
 839        # override options set in an [svn-remote "..."] section
 840        my $e;
 841        foreach (qw/follow_parent no_metadata use_svm_props
 842                    use_svnsync_props/) {
 843                my $key = $_;
 844                $key =~ tr/_//d;
 845                $e .= "sub $_ {
 846                        my (\$self) = \@_;
 847                        return \$self->{-$_} if exists \$self->{-$_};
 848                        my \$k = \"svn-remote.\$self->{repo_id}\.$key\";
 849                        eval { command_oneline(qw/config --get/, \$k) };
 850                        if (\$@) {
 851                                \$self->{-$_} = \$Git::SVN::_$_;
 852                        } else {
 853                                my \$v = command_oneline(qw/config --bool/,\$k);
 854                                \$self->{-$_} = \$v eq 'false' ? 0 : 1;
 855                        }
 856                        return \$self->{-$_} }\n";
 857        }
 858        $e .= "1;\n";
 859        eval $e or die $@;
 860}
 861
 862my %LOCKFILES;
 863END { unlink keys %LOCKFILES if %LOCKFILES }
 864
 865sub resolve_local_globs {
 866        my ($url, $fetch, $glob_spec) = @_;
 867        return unless defined $glob_spec;
 868        my $ref = $glob_spec->{ref};
 869        my $path = $glob_spec->{path};
 870        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
 871                next unless m#^refs/remotes/$ref->{regex}$#;
 872                my $p = $1;
 873                my $pathname = $path->full_path($p);
 874                my $refname = $ref->full_path($p);
 875                if (my $existing = $fetch->{$pathname}) {
 876                        if ($existing ne $refname) {
 877                                die "Refspec conflict:\n",
 878                                    "existing: refs/remotes/$existing\n",
 879                                    " globbed: refs/remotes/$refname\n";
 880                        }
 881                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
 882                        $u =~ s!^\Q$url\E(/|$)!! or die
 883                          "refs/remotes/$refname: '$url' not found in '$u'\n";
 884                        if ($pathname ne $u) {
 885                                warn "W: Refspec glob conflict ",
 886                                     "(ref: refs/remotes/$refname):\n",
 887                                     "expected path: $pathname\n",
 888                                     "    real path: $u\n",
 889                                     "Continuing ahead with $u\n";
 890                                next;
 891                        }
 892                } else {
 893                        $fetch->{$pathname} = $refname;
 894                }
 895        }
 896}
 897
 898sub parse_revision_argument {
 899        my ($base, $head) = @_;
 900        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
 901                return ($base, $head);
 902        }
 903        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
 904        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
 905        return ($head, $head) if ($::_revision eq 'HEAD');
 906        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
 907        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
 908        die "revision argument: $::_revision not understood by git-svn\n";
 909}
 910
 911sub fetch_all {
 912        my ($repo_id, $remotes) = @_;
 913        if (ref $repo_id) {
 914                my $gs = $repo_id;
 915                $repo_id = undef;
 916                $repo_id = $gs->{repo_id};
 917        }
 918        $remotes ||= read_all_remotes();
 919        my $remote = $remotes->{$repo_id} or
 920                     die "[svn-remote \"$repo_id\"] unknown\n";
 921        my $fetch = $remote->{fetch};
 922        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
 923        my (@gs, @globs);
 924        my $ra = Git::SVN::Ra->new($url);
 925        my $uuid = $ra->get_uuid;
 926        my $head = $ra->get_latest_revnum;
 927        my $base = defined $fetch ? $head : 0;
 928
 929        # read the max revs for wildcard expansion (branches/*, tags/*)
 930        foreach my $t (qw/branches tags/) {
 931                defined $remote->{$t} or next;
 932                push @globs, $remote->{$t};
 933                my $max_rev = eval { tmp_config(qw/--int --get/,
 934                                         "svn-remote.$repo_id.${t}-maxRev") };
 935                if (defined $max_rev && ($max_rev < $base)) {
 936                        $base = $max_rev;
 937                } elsif (!defined $max_rev) {
 938                        $base = 0;
 939                }
 940        }
 941
 942        if ($fetch) {
 943                foreach my $p (sort keys %$fetch) {
 944                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
 945                        my $lr = $gs->rev_db_max;
 946                        if (defined $lr) {
 947                                $base = $lr if ($lr < $base);
 948                        }
 949                        push @gs, $gs;
 950                }
 951        }
 952
 953        ($base, $head) = parse_revision_argument($base, $head);
 954        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
 955}
 956
 957sub read_all_remotes {
 958        my $r = {};
 959        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
 960                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
 961                        $r->{$1}->{fetch}->{$2} = $3;
 962                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
 963                        $r->{$1}->{url} = $2;
 964                } elsif (m!^(.+)\.(branches|tags)=
 965                           (.*):refs/remotes/(.+)\s*$/!x) {
 966                        my ($p, $g) = ($3, $4);
 967                        my $rs = $r->{$1}->{$2} = {
 968                                          t => $2,
 969                                          remote => $1,
 970                                          path => Git::SVN::GlobSpec->new($p),
 971                                          ref => Git::SVN::GlobSpec->new($g) };
 972                        if (length($rs->{ref}->{right}) != 0) {
 973                                die "The '*' glob character must be the last ",
 974                                    "character of '$g'\n";
 975                        }
 976                }
 977        }
 978        $r;
 979}
 980
 981sub init_vars {
 982        if (defined $_repack) {
 983                $_repack = 1000 if ($_repack <= 0);
 984                $_repack_nr = $_repack;
 985                $_repack_flags ||= '-d';
 986        }
 987}
 988
 989sub verify_remotes_sanity {
 990        return unless -d $ENV{GIT_DIR};
 991        my %seen;
 992        foreach (command(qw/config -l/)) {
 993                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
 994                        if ($seen{$1}) {
 995                                die "Remote ref refs/remote/$1 is tracked by",
 996                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
 997                                    "Please resolve this ambiguity in ",
 998                                    "your git configuration file before ",
 999                                    "continuing\n";
1000                        }
1001                        $seen{$1} = $_;
1002                }
1003        }
1004}
1005
1006# we allow more chars than remotes2config.sh...
1007sub sanitize_remote_name {
1008        my ($name) = @_;
1009        $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1010        $name;
1011}
1012
1013sub find_existing_remote {
1014        my ($url, $remotes) = @_;
1015        return undef if $no_reuse_existing;
1016        my $existing;
1017        foreach my $repo_id (keys %$remotes) {
1018                my $u = $remotes->{$repo_id}->{url} or next;
1019                next if $u ne $url;
1020                $existing = $repo_id;
1021                last;
1022        }
1023        $existing;
1024}
1025
1026sub init_remote_config {
1027        my ($self, $url, $no_write) = @_;
1028        $url =~ s!/+$!!; # strip trailing slash
1029        my $r = read_all_remotes();
1030        my $existing = find_existing_remote($url, $r);
1031        if ($existing) {
1032                unless ($no_write) {
1033                        print STDERR "Using existing ",
1034                                     "[svn-remote \"$existing\"]\n";
1035                }
1036                $self->{repo_id} = $existing;
1037        } else {
1038                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1039                $existing = find_existing_remote($min_url, $r);
1040                if ($existing) {
1041                        unless ($no_write) {
1042                                print STDERR "Using existing ",
1043                                             "[svn-remote \"$existing\"]\n";
1044                        }
1045                        $self->{repo_id} = $existing;
1046                }
1047                if ($min_url ne $url) {
1048                        unless ($no_write) {
1049                                print STDERR "Using higher level of URL: ",
1050                                             "$url => $min_url\n";
1051                        }
1052                        my $old_path = $self->{path};
1053                        $self->{path} = $url;
1054                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1055                        if (length $old_path) {
1056                                $self->{path} .= "/$old_path";
1057                        }
1058                        $url = $min_url;
1059                }
1060        }
1061        my $orig_url;
1062        if (!$existing) {
1063                # verify that we aren't overwriting anything:
1064                $orig_url = eval {
1065                        command_oneline('config', '--get',
1066                                        "svn-remote.$self->{repo_id}.url")
1067                };
1068                if ($orig_url && ($orig_url ne $url)) {
1069                        die "svn-remote.$self->{repo_id}.url already set: ",
1070                            "$orig_url\nwanted to set to: $url\n";
1071                }
1072        }
1073        my ($xrepo_id, $xpath) = find_ref($self->refname);
1074        if (defined $xpath) {
1075                die "svn-remote.$xrepo_id.fetch already set to track ",
1076                    "$xpath:refs/remotes/", $self->refname, "\n";
1077        }
1078        unless ($no_write) {
1079                command_noisy('config',
1080                              "svn-remote.$self->{repo_id}.url", $url);
1081                command_noisy('config', '--add',
1082                              "svn-remote.$self->{repo_id}.fetch",
1083                              "$self->{path}:".$self->refname);
1084        }
1085        $self->{url} = $url;
1086}
1087
1088sub find_by_url { # repos_root and, path are optional
1089        my ($class, $full_url, $repos_root, $path) = @_;
1090
1091        return undef unless defined $full_url;
1092        remove_username($full_url);
1093        remove_username($repos_root) if defined $repos_root;
1094        my $remotes = read_all_remotes();
1095        if (defined $full_url && defined $repos_root && !defined $path) {
1096                $path = $full_url;
1097                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1098        }
1099        foreach my $repo_id (keys %$remotes) {
1100                my $u = $remotes->{$repo_id}->{url} or next;
1101                remove_username($u);
1102                next if defined $repos_root && $repos_root ne $u;
1103
1104                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1105                foreach (qw/branches tags/) {
1106                        resolve_local_globs($u, $fetch,
1107                                            $remotes->{$repo_id}->{$_});
1108                }
1109                my $p = $path;
1110                unless (defined $p) {
1111                        $p = $full_url;
1112                        $p =~ s#^\Q$u\E(?:/|$)## or next;
1113                }
1114                foreach my $f (keys %$fetch) {
1115                        next if $f ne $p;
1116                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1117                }
1118        }
1119        undef;
1120}
1121
1122sub init {
1123        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1124        my $self = _new($class, $repo_id, $ref_id, $path);
1125        if (defined $url) {
1126                $self->init_remote_config($url, $no_write);
1127        }
1128        $self;
1129}
1130
1131sub find_ref {
1132        my ($ref_id) = @_;
1133        foreach (command(qw/config -l/)) {
1134                next unless m!^svn-remote\.(.+)\.fetch=
1135                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1136                my ($repo_id, $path, $ref) = ($1, $2, $3);
1137                if ($ref eq $ref_id) {
1138                        $path = '' if ($path =~ m#^\./?#);
1139                        return ($repo_id, $path);
1140                }
1141        }
1142        (undef, undef, undef);
1143}
1144
1145sub new {
1146        my ($class, $ref_id, $repo_id, $path) = @_;
1147        if (defined $ref_id && !defined $repo_id && !defined $path) {
1148                ($repo_id, $path) = find_ref($ref_id);
1149                if (!defined $repo_id) {
1150                        die "Could not find a \"svn-remote.*.fetch\" key ",
1151                            "in the repository configuration matching: ",
1152                            "refs/remotes/$ref_id\n";
1153                }
1154        }
1155        my $self = _new($class, $repo_id, $ref_id, $path);
1156        if (!defined $self->{path} || !length $self->{path}) {
1157                my $fetch = command_oneline('config', '--get',
1158                                            "svn-remote.$repo_id.fetch",
1159                                            ":refs/remotes/$ref_id\$") or
1160                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1161                         "\":refs/remotes/$ref_id\$\" in config\n";
1162                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1163        }
1164        $self->{url} = command_oneline('config', '--get',
1165                                       "svn-remote.$repo_id.url") or
1166                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1167        $self->rebuild;
1168        $self;
1169}
1170
1171sub refname { "refs/remotes/$_[0]->{ref_id}" }
1172
1173sub svm_uuid {
1174        my ($self) = @_;
1175        return $self->{svm}->{uuid} if $self->svm;
1176        $self->ra;
1177        unless ($self->{svm}) {
1178                die "SVM UUID not cached, and reading remotely failed\n";
1179        }
1180        $self->{svm}->{uuid};
1181}
1182
1183sub svm {
1184        my ($self) = @_;
1185        return $self->{svm} if $self->{svm};
1186        my $svm;
1187        # see if we have it in our config, first:
1188        eval {
1189                my $section = "svn-remote.$self->{repo_id}";
1190                $svm = {
1191                  source => tmp_config('--get', "$section.svm-source"),
1192                  uuid => tmp_config('--get', "$section.svm-uuid"),
1193                  replace => tmp_config('--get', "$section.svm-replace"),
1194                }
1195        };
1196        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1197                $self->{svm} = $svm;
1198        }
1199        $self->{svm};
1200}
1201
1202sub _set_svm_vars {
1203        my ($self, $ra) = @_;
1204        return $ra if $self->svm;
1205
1206        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1207                    "(svm:source, svm:uuid) ",
1208                    "from the following URLs:\n" );
1209        sub read_svm_props {
1210                my ($self, $ra, $path, $r) = @_;
1211                my $props = ($ra->get_dir($path, $r))[2];
1212                my $src = $props->{'svm:source'};
1213                my $uuid = $props->{'svm:uuid'};
1214                return undef if (!$src || !$uuid);
1215
1216                chomp($src, $uuid);
1217
1218                $uuid =~ m{^[0-9a-f\-]{30,}$}
1219                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1220
1221                # the '!' is used to mark the repos_root!/relative/path
1222                $src =~ s{/?!/?}{/};
1223                $src =~ s{/+$}{}; # no trailing slashes please
1224                # username is of no interest
1225                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1226
1227                my $replace = $ra->{url};
1228                $replace .= "/$path" if length $path;
1229
1230                my $section = "svn-remote.$self->{repo_id}";
1231                tmp_config("$section.svm-source", $src);
1232                tmp_config("$section.svm-replace", $replace);
1233                tmp_config("$section.svm-uuid", $uuid);
1234                $self->{svm} = {
1235                        source => $src,
1236                        uuid => $uuid,
1237                        replace => $replace
1238                };
1239        }
1240
1241        my $r = $ra->get_latest_revnum;
1242        my $path = $self->{path};
1243        my %tried;
1244        while (length $path) {
1245                unless ($tried{"$self->{url}/$path"}) {
1246                        return $ra if $self->read_svm_props($ra, $path, $r);
1247                        $tried{"$self->{url}/$path"} = 1;
1248                }
1249                $path =~ s#/?[^/]+$##;
1250        }
1251        die "Path: '$path' should be ''\n" if $path ne '';
1252        return $ra if $self->read_svm_props($ra, $path, $r);
1253        $tried{"$self->{url}/$path"} = 1;
1254
1255        if ($ra->{repos_root} eq $self->{url}) {
1256                die @err, (map { "  $_\n" } keys %tried), "\n";
1257        }
1258
1259        # nope, make sure we're connected to the repository root:
1260        my $ok;
1261        my @tried_b;
1262        $path = $ra->{svn_path};
1263        $ra = Git::SVN::Ra->new($ra->{repos_root});
1264        while (length $path) {
1265                unless ($tried{"$ra->{url}/$path"}) {
1266                        $ok = $self->read_svm_props($ra, $path, $r);
1267                        last if $ok;
1268                        $tried{"$ra->{url}/$path"} = 1;
1269                }
1270                $path =~ s#/?[^/]+$##;
1271        }
1272        die "Path: '$path' should be ''\n" if $path ne '';
1273        $ok ||= $self->read_svm_props($ra, $path, $r);
1274        $tried{"$ra->{url}/$path"} = 1;
1275        if (!$ok) {
1276                die @err, (map { "  $_\n" } keys %tried), "\n";
1277        }
1278        Git::SVN::Ra->new($self->{url});
1279}
1280
1281sub svnsync {
1282        my ($self) = @_;
1283        return $self->{svnsync} if $self->{svnsync};
1284
1285        if ($self->no_metadata) {
1286                die "Can't have both 'noMetadata' and ",
1287                    "'useSvnsyncProps' options set!\n";
1288        }
1289        if ($self->rewrite_root) {
1290                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1291                    "options set!\n";
1292        }
1293
1294        my $svnsync;
1295        # see if we have it in our config, first:
1296        eval {
1297                my $section = "svn-remote.$self->{repo_id}";
1298                $svnsync = {
1299                  url => tmp_config('--get', "$section.svnsync-url"),
1300                  uuid => tmp_config('--get', "$section.svnsync-uuid"),
1301                }
1302        };
1303        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1304                return $self->{svnsync} = $svnsync;
1305        }
1306
1307        my $err = "useSvnsyncProps set, but failed to read " .
1308                  "svnsync property: svn:sync-from-";
1309        my $rp = $self->ra->rev_proplist(0);
1310
1311        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1312        $url =~ m{^[a-z\+]+://} or
1313                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1314
1315        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1316        $uuid =~ m{^[0-9a-f\-]{30,}$} or
1317                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1318
1319        my $section = "svn-remote.$self->{repo_id}";
1320        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1321        tmp_config('--add', "$section.svnsync-url", $url);
1322        return $self->{svnsync} = { url => $url, uuid => $uuid };
1323}
1324
1325# this allows us to memoize our SVN::Ra UUID locally and avoid a
1326# remote lookup (useful for 'git svn log').
1327sub ra_uuid {
1328        my ($self) = @_;
1329        unless ($self->{ra_uuid}) {
1330                my $key = "svn-remote.$self->{repo_id}.uuid";
1331                my $uuid = eval { tmp_config('--get', $key) };
1332                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1333                        $self->{ra_uuid} = $uuid;
1334                } else {
1335                        die "ra_uuid called without URL\n" unless $self->{url};
1336                        $self->{ra_uuid} = $self->ra->get_uuid;
1337                        tmp_config('--add', $key, $self->{ra_uuid});
1338                }
1339        }
1340        $self->{ra_uuid};
1341}
1342
1343sub ra {
1344        my ($self) = shift;
1345        my $ra = Git::SVN::Ra->new($self->{url});
1346        if ($self->use_svm_props && !$self->{svm}) {
1347                if ($self->no_metadata) {
1348                        die "Can't have both 'noMetadata' and ",
1349                            "'useSvmProps' options set!\n";
1350                } elsif ($self->use_svnsync_props) {
1351                        die "Can't have both 'useSvnsyncProps' and ",
1352                            "'useSvmProps' options set!\n";
1353                }
1354                $ra = $self->_set_svm_vars($ra);
1355                $self->{-want_revprops} = 1;
1356        }
1357        $ra;
1358}
1359
1360sub rel_path {
1361        my ($self) = @_;
1362        my $repos_root = $self->ra->{repos_root};
1363        return $self->{path} if ($self->{url} eq $repos_root);
1364        my $url = $self->{url} .
1365                  (length $self->{path} ? "/$self->{path}" : $self->{path});
1366        $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1367        $url;
1368}
1369
1370sub traverse_ignore {
1371        my ($self, $fh, $path, $r) = @_;
1372        $path =~ s#^/+##g;
1373        my $ra = $self->ra;
1374        my ($dirent, undef, $props) = $ra->get_dir($path, $r);
1375        my $p = $path;
1376        $p =~ s#^\Q$self->{path}\E(/|$)##;
1377        print $fh length $p ? "\n# $p\n" : "\n# /\n";
1378        if (my $s = $props->{'svn:ignore'}) {
1379                $s =~ s/[\r\n]+/\n/g;
1380                chomp $s;
1381                if (length $p == 0) {
1382                        $s =~ s#\n#\n/$p#g;
1383                        print $fh "/$s\n";
1384                } else {
1385                        $s =~ s#\n#\n/$p/#g;
1386                        print $fh "/$p/$s\n";
1387                }
1388        }
1389        foreach (sort keys %$dirent) {
1390                next if $dirent->{$_}->kind != $SVN::Node::dir;
1391                $self->traverse_ignore($fh, "$path/$_", $r);
1392        }
1393}
1394
1395sub last_rev { ($_[0]->last_rev_commit)[0] }
1396sub last_commit { ($_[0]->last_rev_commit)[1] }
1397
1398# returns the newest SVN revision number and newest commit SHA1
1399sub last_rev_commit {
1400        my ($self) = @_;
1401        if (defined $self->{last_rev} && defined $self->{last_commit}) {
1402                return ($self->{last_rev}, $self->{last_commit});
1403        }
1404        my $c = ::verify_ref($self->refname.'^0');
1405        if ($c && !$self->use_svm_props && !$self->no_metadata) {
1406                my $rev = (::cmt_metadata($c))[1];
1407                if (defined $rev) {
1408                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1409                        return ($rev, $c);
1410                }
1411        }
1412        my $db_path = $self->db_path;
1413        unless (-e $db_path) {
1414                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1415                return (undef, undef);
1416        }
1417        my $offset = -41; # from tail
1418        my $rl;
1419        open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1420        sysseek($fh, $offset, 2); # don't care for errors
1421        sysread($fh, $rl, 41) == 41 or return (undef, undef);
1422        chomp $rl;
1423        while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1424                $offset -= 41;
1425                sysseek($fh, $offset, 2); # don't care for errors
1426                sysread($fh, $rl, 41) == 41 or return (undef, undef);
1427                chomp $rl;
1428        }
1429        if ($c && $c ne $rl) {
1430                die "$db_path and ", $self->refname,
1431                    " inconsistent!:\n$c != $rl\n";
1432        }
1433        my $rev = sysseek($fh, 0, 1) or croak $!;
1434        $rev =  ($rev - 41) / 41;
1435        close $fh or croak $!;
1436        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1437        return ($rev, $c);
1438}
1439
1440sub get_fetch_range {
1441        my ($self, $min, $max) = @_;
1442        $max ||= $self->ra->get_latest_revnum;
1443        $min ||= $self->rev_db_max;
1444        (++$min, $max);
1445}
1446
1447sub tmp_config {
1448        my (@args) = @_;
1449        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1450        my $config = "$ENV{GIT_DIR}/svn/.metadata";
1451        if (-e $old_def_config && ! -e $config) {
1452                rename $old_def_config, $config or
1453                       die "Failed rename $old_def_config => $config: $!\n";
1454        }
1455        my $old_config = $ENV{GIT_CONFIG};
1456        $ENV{GIT_CONFIG} = $config;
1457        $@ = undef;
1458        my @ret = eval {
1459                unless (-f $config) {
1460                        mkfile($config);
1461                        open my $fh, '>', $config or
1462                            die "Can't open $config: $!\n";
1463                        print $fh "; This file is used internally by ",
1464                                  "git-svn\n" or die
1465                                  "Couldn't write to $config: $!\n";
1466                        print $fh "; You should not have to edit it\n" or
1467                              die "Couldn't write to $config: $!\n";
1468                        close $fh or die "Couldn't close $config: $!\n";
1469                }
1470                command('config', @args);
1471        };
1472        my $err = $@;
1473        if (defined $old_config) {
1474                $ENV{GIT_CONFIG} = $old_config;
1475        } else {
1476                delete $ENV{GIT_CONFIG};
1477        }
1478        die $err if $err;
1479        wantarray ? @ret : $ret[0];
1480}
1481
1482sub tmp_index_do {
1483        my ($self, $sub) = @_;
1484        my $old_index = $ENV{GIT_INDEX_FILE};
1485        $ENV{GIT_INDEX_FILE} = $self->{index};
1486        $@ = undef;
1487        my @ret = eval {
1488                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1489                mkpath([$dir]) unless -d $dir;
1490                &$sub;
1491        };
1492        my $err = $@;
1493        if (defined $old_index) {
1494                $ENV{GIT_INDEX_FILE} = $old_index;
1495        } else {
1496                delete $ENV{GIT_INDEX_FILE};
1497        }
1498        die $err if $err;
1499        wantarray ? @ret : $ret[0];
1500}
1501
1502sub assert_index_clean {
1503        my ($self, $treeish) = @_;
1504
1505        $self->tmp_index_do(sub {
1506                command_noisy('read-tree', $treeish) unless -e $self->{index};
1507                my $x = command_oneline('write-tree');
1508                my ($y) = (command(qw/cat-file commit/, $treeish) =~
1509                           /^tree ($::sha1)/mo);
1510                return if $y eq $x;
1511
1512                warn "Index mismatch: $y != $x\nrereading $treeish\n";
1513                unlink $self->{index} or die "unlink $self->{index}: $!\n";
1514                command_noisy('read-tree', $treeish);
1515                $x = command_oneline('write-tree');
1516                if ($y ne $x) {
1517                        ::fatal "trees ($treeish) $y != $x\n",
1518                                "Something is seriously wrong...\n";
1519                }
1520        });
1521}
1522
1523sub get_commit_parents {
1524        my ($self, $log_entry) = @_;
1525        my (%seen, @ret, @tmp);
1526        # legacy support for 'set-tree'; this is only used by set_tree_cb:
1527        if (my $ip = $self->{inject_parents}) {
1528                if (my $commit = delete $ip->{$log_entry->{revision}}) {
1529                        push @tmp, $commit;
1530                }
1531        }
1532        if (my $cur = ::verify_ref($self->refname.'^0')) {
1533                push @tmp, $cur;
1534        }
1535        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1536        while (my $p = shift @tmp) {
1537                next if $seen{$p};
1538                $seen{$p} = 1;
1539                push @ret, $p;
1540                # MAXPARENT is defined to 16 in commit-tree.c:
1541                last if @ret >= 16;
1542        }
1543        if (@tmp) {
1544                die "r$log_entry->{revision}: No room for parents:\n\t",
1545                    join("\n\t", @tmp), "\n";
1546        }
1547        @ret;
1548}
1549
1550sub rewrite_root {
1551        my ($self) = @_;
1552        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
1553        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
1554        my $rwr = eval { command_oneline(qw/config --get/, $k) };
1555        if ($rwr) {
1556                $rwr =~ s#/+$##;
1557                if ($rwr !~ m#^[a-z\+]+://#) {
1558                        die "$rwr is not a valid URL (key: $k)\n";
1559                }
1560        }
1561        $self->{-rewrite_root} = $rwr;
1562}
1563
1564sub metadata_url {
1565        my ($self) = @_;
1566        ($self->rewrite_root || $self->{url}) .
1567           (length $self->{path} ? '/' . $self->{path} : '');
1568}
1569
1570sub full_url {
1571        my ($self) = @_;
1572        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1573}
1574
1575sub do_git_commit {
1576        my ($self, $log_entry) = @_;
1577        my $lr = $self->last_rev;
1578        if (defined $lr && $lr >= $log_entry->{revision}) {
1579                die "Last fetched revision of ", $self->refname,
1580                    " was r$lr, but we are about to fetch: ",
1581                    "r$log_entry->{revision}!\n";
1582        }
1583        if (my $c = $self->rev_db_get($log_entry->{revision})) {
1584                croak "$log_entry->{revision} = $c already exists! ",
1585                      "Why are we refetching it?\n";
1586        }
1587        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1588        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1589                                                          $log_entry->{email};
1590        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1591
1592        my $tree = $log_entry->{tree};
1593        if (!defined $tree) {
1594                $tree = $self->tmp_index_do(sub {
1595                                            command_oneline('write-tree') });
1596        }
1597        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1598
1599        my @exec = ('git-commit-tree', $tree);
1600        foreach ($self->get_commit_parents($log_entry)) {
1601                push @exec, '-p', $_;
1602        }
1603        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1604                                                                   or croak $!;
1605        print $msg_fh $log_entry->{log} or croak $!;
1606        unless ($self->no_metadata) {
1607                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1608                              or croak $!;
1609        }
1610        $msg_fh->flush == 0 or croak $!;
1611        close $msg_fh or croak $!;
1612        chomp(my $commit = do { local $/; <$out_fh> });
1613        close $out_fh or croak $!;
1614        waitpid $pid, 0;
1615        croak $? if $?;
1616        if ($commit !~ /^$::sha1$/o) {
1617                die "Failed to commit, invalid sha1: $commit\n";
1618        }
1619
1620        $self->rev_db_set($log_entry->{revision}, $commit, 1);
1621
1622        $self->{last_rev} = $log_entry->{revision};
1623        $self->{last_commit} = $commit;
1624        print "r$log_entry->{revision}";
1625        if (defined $log_entry->{svm_revision}) {
1626                 print " (\@$log_entry->{svm_revision})";
1627                 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1628                                   0, $self->svm_uuid);
1629        }
1630        print " = $commit ($self->{ref_id})\n";
1631        if (defined $_repack && (--$_repack_nr == 0)) {
1632                $_repack_nr = $_repack;
1633                # repack doesn't use any arguments with spaces in them, does it?
1634                print "Running git repack $_repack_flags ...\n";
1635                command_noisy('repack', split(/\s+/, $_repack_flags));
1636                print "Done repacking\n";
1637        }
1638        return $commit;
1639}
1640
1641sub match_paths {
1642        my ($self, $paths, $r) = @_;
1643        return 1 if $self->{path} eq '';
1644        if (my $path = $paths->{"/$self->{path}"}) {
1645                return ($path->{action} eq 'D') ? 0 : 1;
1646        }
1647        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1648        if (grep /$self->{path_regex}/, keys %$paths) {
1649                return 1;
1650        }
1651        my $c = '';
1652        foreach (split m#/#, $self->{path}) {
1653                $c .= "/$_";
1654                next unless ($paths->{$c} &&
1655                             ($paths->{$c}->{action} =~ /^[AR]$/));
1656                if ($self->ra->check_path($self->{path}, $r) ==
1657                    $SVN::Node::dir) {
1658                        return 1;
1659                }
1660        }
1661        return 0;
1662}
1663
1664sub find_parent_branch {
1665        my ($self, $paths, $rev) = @_;
1666        return undef unless $self->follow_parent;
1667        unless (defined $paths) {
1668                my $err_handler = $SVN::Error::handler;
1669                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1670                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1671                                   $paths =
1672                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
1673                $SVN::Error::handler = $err_handler;
1674        }
1675        return undef unless defined $paths;
1676
1677        # look for a parent from another branch:
1678        my @b_path_components = split m#/#, $self->rel_path;
1679        my @a_path_components;
1680        my $i;
1681        while (@b_path_components) {
1682                $i = $paths->{'/'.join('/', @b_path_components)};
1683                last if $i && defined $i->{copyfrom_path};
1684                unshift(@a_path_components, pop(@b_path_components));
1685        }
1686        return undef unless defined $i && defined $i->{copyfrom_path};
1687        my $branch_from = $i->{copyfrom_path};
1688        if (@a_path_components) {
1689                print STDERR "branch_from: $branch_from => ";
1690                $branch_from .= '/'.join('/', @a_path_components);
1691                print STDERR $branch_from, "\n";
1692        }
1693        my $r = $i->{copyfrom_rev};
1694        my $repos_root = $self->ra->{repos_root};
1695        my $url = $self->ra->{url};
1696        my $new_url = $repos_root . $branch_from;
1697        print STDERR  "Found possible branch point: ",
1698                      "$new_url => ", $self->full_url, ", $r\n";
1699        $branch_from =~ s#^/##;
1700        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
1701        unless ($gs) {
1702                my $ref_id = $self->{ref_id};
1703                $ref_id =~ s/\@\d+$//;
1704                $ref_id .= "\@$r";
1705                # just grow a tail if we're not unique enough :x
1706                $ref_id .= '-' while find_ref($ref_id);
1707                print STDERR "Initializing parent: $ref_id\n";
1708                $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1709        }
1710        my ($r0, $parent) = $gs->find_rev_before($r, 1);
1711        if (!defined $r0 || !defined $parent) {
1712                my ($base, $head) = parse_revision_argument(0, $r);
1713                if ($base <= $r) {
1714                        $gs->fetch($base, $r);
1715                }
1716                ($r0, $parent) = $gs->last_rev_commit;
1717        }
1718        if (defined $r0 && defined $parent) {
1719                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1720                my $ed;
1721                if ($self->ra->can_do_switch) {
1722                        $self->assert_index_clean($parent);
1723                        print STDERR "Following parent with do_switch\n";
1724                        # do_switch works with svn/trunk >= r22312, but that
1725                        # is not included with SVN 1.4.3 (the latest version
1726                        # at the moment), so we can't rely on it
1727                        $self->{last_commit} = $parent;
1728                        $ed = SVN::Git::Fetcher->new($self);
1729                        $gs->ra->gs_do_switch($r0, $rev, $gs,
1730                                              $self->full_url, $ed)
1731                          or die "SVN connection failed somewhere...\n";
1732                } else {
1733                        print STDERR "Following parent with do_update\n";
1734                        $ed = SVN::Git::Fetcher->new($self);
1735                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
1736                          or die "SVN connection failed somewhere...\n";
1737                }
1738                print STDERR "Successfully followed parent\n";
1739                return $self->make_log_entry($rev, [$parent], $ed);
1740        }
1741        return undef;
1742}
1743
1744sub do_fetch {
1745        my ($self, $paths, $rev) = @_;
1746        my $ed;
1747        my ($last_rev, @parents);
1748        if (my $lc = $self->last_commit) {
1749                # we can have a branch that was deleted, then re-added
1750                # under the same name but copied from another path, in
1751                # which case we'll have multiple parents (we don't
1752                # want to break the original ref, nor lose copypath info):
1753                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1754                        push @{$log_entry->{parents}}, $lc;
1755                        return $log_entry;
1756                }
1757                $ed = SVN::Git::Fetcher->new($self);
1758                $last_rev = $self->{last_rev};
1759                $ed->{c} = $lc;
1760                @parents = ($lc);
1761        } else {
1762                $last_rev = $rev;
1763                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1764                        return $log_entry;
1765                }
1766                $ed = SVN::Git::Fetcher->new($self);
1767        }
1768        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1769                die "SVN connection failed somewhere...\n";
1770        }
1771        $self->make_log_entry($rev, \@parents, $ed);
1772}
1773
1774sub get_untracked {
1775        my ($self, $ed) = @_;
1776        my @out;
1777        my $h = $ed->{empty};
1778        foreach (sort keys %$h) {
1779                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1780                push @out, "  $act: " . uri_encode($_);
1781                warn "W: $act: $_\n";
1782        }
1783        foreach my $t (qw/dir_prop file_prop/) {
1784                $h = $ed->{$t} or next;
1785                foreach my $path (sort keys %$h) {
1786                        my $ppath = $path eq '' ? '.' : $path;
1787                        foreach my $prop (sort keys %{$h->{$path}}) {
1788                                next if $SKIP_PROP{$prop};
1789                                my $v = $h->{$path}->{$prop};
1790                                my $t_ppath_prop = "$t: " .
1791                                                    uri_encode($ppath) . ' ' .
1792                                                    uri_encode($prop);
1793                                if (defined $v) {
1794                                        push @out, "  +$t_ppath_prop " .
1795                                                   uri_encode($v);
1796                                } else {
1797                                        push @out, "  -$t_ppath_prop";
1798                                }
1799                        }
1800                }
1801        }
1802        foreach my $t (qw/absent_file absent_directory/) {
1803                $h = $ed->{$t} or next;
1804                foreach my $parent (sort keys %$h) {
1805                        foreach my $path (sort @{$h->{$parent}}) {
1806                                push @out, "  $t: " .
1807                                           uri_encode("$parent/$path");
1808                                warn "W: $t: $parent/$path ",
1809                                     "Insufficient permissions?\n";
1810                        }
1811                }
1812        }
1813        \@out;
1814}
1815
1816sub parse_svn_date {
1817        my $date = shift || return '+0000 1970-01-01 00:00:00';
1818        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1819                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1820                                         croak "Unable to parse date: $date\n";
1821        "+0000 $Y-$m-$d $H:$M:$S";
1822}
1823
1824sub check_author {
1825        my ($author) = @_;
1826        if (!defined $author || length $author == 0) {
1827                $author = '(no author)';
1828        }
1829        if (defined $::_authors && ! defined $::users{$author}) {
1830                die "Author: $author not defined in $::_authors file\n";
1831        }
1832        $author;
1833}
1834
1835sub make_log_entry {
1836        my ($self, $rev, $parents, $ed) = @_;
1837        my $untracked = $self->get_untracked($ed);
1838
1839        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1840        print $un "r$rev\n" or croak $!;
1841        print $un $_, "\n" foreach @$untracked;
1842        my %log_entry = ( parents => $parents || [], revision => $rev,
1843                          log => '');
1844
1845        my $headrev;
1846        my $logged = delete $self->{logged_rev_props};
1847        if (!$logged || $self->{-want_revprops}) {
1848                my $rp = $self->ra->rev_proplist($rev);
1849                foreach (sort keys %$rp) {
1850                        my $v = $rp->{$_};
1851                        if (/^svn:(author|date|log)$/) {
1852                                $log_entry{$1} = $v;
1853                        } elsif ($_ eq 'svm:headrev') {
1854                                $headrev = $v;
1855                        } else {
1856                                print $un "  rev_prop: ", uri_encode($_), ' ',
1857                                          uri_encode($v), "\n";
1858                        }
1859                }
1860        } else {
1861                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1862        }
1863        close $un or croak $!;
1864
1865        $log_entry{date} = parse_svn_date($log_entry{date});
1866        $log_entry{log} .= "\n";
1867        my $author = $log_entry{author} = check_author($log_entry{author});
1868        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1869                                                       : ($author, undef);
1870        if (defined $headrev && $self->use_svm_props) {
1871                if ($self->rewrite_root) {
1872                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
1873                            "options set!\n";
1874                }
1875                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1876                # we don't want "SVM: initializing mirror for junk" ...
1877                return undef if $r == 0;
1878                my $svm = $self->svm;
1879                if ($uuid ne $svm->{uuid}) {
1880                        die "UUID mismatch on SVM path:\n",
1881                            "expected: $svm->{uuid}\n",
1882                            "     got: $uuid\n";
1883                }
1884                my $full_url = $self->full_url;
1885                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
1886                             die "Failed to replace '$svm->{replace}' with ",
1887                                 "'$svm->{source}' in $full_url\n";
1888                # throw away username for storing in records
1889                remove_username($full_url);
1890                $log_entry{metadata} = "$full_url\@$r $uuid";
1891                $log_entry{svm_revision} = $r;
1892                $email ||= "$author\@$uuid"
1893        } elsif ($self->use_svnsync_props) {
1894                my $full_url = $self->svnsync->{url};
1895                $full_url .= "/$self->{path}" if length $self->{path};
1896                remove_username($full_url);
1897                my $uuid = $self->svnsync->{uuid};
1898                $log_entry{metadata} = "$full_url\@$rev $uuid";
1899                $email ||= "$author\@$uuid"
1900        } else {
1901                my $url = $self->metadata_url;
1902                remove_username($url);
1903                $log_entry{metadata} = "$url\@$rev " .
1904                                       $self->ra->get_uuid;
1905                $email ||= "$author\@" . $self->ra->get_uuid;
1906        }
1907        $log_entry{name} = $name;
1908        $log_entry{email} = $email;
1909        \%log_entry;
1910}
1911
1912sub fetch {
1913        my ($self, $min_rev, $max_rev, @parents) = @_;
1914        my ($last_rev, $last_commit) = $self->last_rev_commit;
1915        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1916        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1917}
1918
1919sub set_tree_cb {
1920        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1921        $self->{inject_parents} = { $rev => $tree };
1922        $self->fetch(undef, undef);
1923}
1924
1925sub set_tree {
1926        my ($self, $tree) = (shift, shift);
1927        my $log_entry = ::get_commit_entry($tree);
1928        unless ($self->{last_rev}) {
1929                fatal("Must have an existing revision to commit\n");
1930        }
1931        my %ed_opts = ( r => $self->{last_rev},
1932                        log => $log_entry->{log},
1933                        ra => $self->ra,
1934                        tree_a => $self->{last_commit},
1935                        tree_b => $tree,
1936                        editor_cb => sub {
1937                               $self->set_tree_cb($log_entry, $tree, @_) },
1938                        svn_path => $self->{path} );
1939        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1940                print "No changes\nr$self->{last_rev} = $tree\n";
1941        }
1942}
1943
1944sub rebuild {
1945        my ($self) = @_;
1946        my $db_path = $self->db_path;
1947        return if (-e $db_path && ! -z $db_path);
1948        return unless ::verify_ref($self->refname.'^0');
1949        if (-f $self->{db_root}) {
1950                rename $self->{db_root}, $db_path or die
1951                     "rename $self->{db_root} => $db_path failed: $!\n";
1952                my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
1953                symlink $base, $self->{db_root} or die
1954                     "symlink $base => $self->{db_root} failed: $!\n";
1955                return;
1956        }
1957        print "Rebuilding $db_path ...\n";
1958        my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1959        my $latest;
1960        my $full_url = $self->full_url;
1961        remove_username($full_url);
1962        my $svn_uuid;
1963        while (<$rev_list>) {
1964                chomp;
1965                my $c = $_;
1966                die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1967                my ($url, $rev, $uuid) = ::cmt_metadata($c);
1968                remove_username($url);
1969
1970                # ignore merges (from set-tree)
1971                next if (!defined $rev || !$uuid);
1972
1973                # if we merged or otherwise started elsewhere, this is
1974                # how we break out of it
1975                if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1976                    ($full_url && $url && ($url ne $full_url))) {
1977                        next;
1978                }
1979                $latest ||= $rev;
1980                $svn_uuid ||= $uuid;
1981
1982                $self->rev_db_set($rev, $c);
1983                print "r$rev = $c\n";
1984        }
1985        command_close_pipe($rev_list, $ctx);
1986        print "Done rebuilding $db_path\n";
1987}
1988
1989# rev_db:
1990# Tie::File seems to be prone to offset errors if revisions get sparse,
1991# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
1992# one of my favorite modules is out :<  Next up would be one of the DBM
1993# modules, but I'm not sure which is most portable...  So I'll just
1994# go with something that's plain-text, but still capable of
1995# being randomly accessed.  So here's my ultra-simple fixed-width
1996# database.  All records are 40 characters + "\n", so it's easy to seek
1997# to a revision: (41 * rev) is the byte offset.
1998# A record of 40 0s denotes an empty revision.
1999# And yes, it's still pretty fast (faster than Tie::File).
2000# These files are disposable unless noMetadata or useSvmProps is set
2001
2002sub _rev_db_set {
2003        my ($fh, $rev, $commit) = @_;
2004        my $offset = $rev * 41;
2005        # assume that append is the common case:
2006        seek $fh, 0, 2 or croak $!;
2007        my $pos = tell $fh;
2008        if ($pos < $offset) {
2009                for (1 .. (($offset - $pos) / 41)) {
2010                        print $fh (('0' x 40),"\n") or croak $!;
2011                }
2012        }
2013        seek $fh, $offset, 0 or croak $!;
2014        print $fh $commit,"\n" or croak $!;
2015}
2016
2017sub mkfile {
2018        my ($path) = @_;
2019        unless (-e $path) {
2020                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2021                mkpath([$dir]) unless -d $dir;
2022                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2023                close $fh or die "Couldn't close (create) $path: $!\n";
2024        }
2025}
2026
2027sub rev_db_set {
2028        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2029        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2030        my $db = $self->db_path($uuid);
2031        my $db_lock = "$db.lock";
2032        my $sig;
2033        if ($update_ref) {
2034                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2035                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2036        }
2037        mkfile($db);
2038
2039        $LOCKFILES{$db_lock} = 1;
2040        my $sync;
2041        # both of these options make our .rev_db file very, very important
2042        # and we can't afford to lose it because rebuild() won't work
2043        if ($self->use_svm_props || $self->no_metadata) {
2044                $sync = 1;
2045                copy($db, $db_lock) or die "rev_db_set(@_): ",
2046                                           "Failed to copy: ",
2047                                           "$db => $db_lock ($!)\n";
2048        } else {
2049                rename $db, $db_lock or die "rev_db_set(@_): ",
2050                                            "Failed to rename: ",
2051                                            "$db => $db_lock ($!)\n";
2052        }
2053        open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
2054        _rev_db_set($fh, $rev, $commit);
2055        if ($sync) {
2056                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2057                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2058        }
2059        close $fh or croak $!;
2060        if ($update_ref) {
2061                $_head = $self;
2062                command_noisy('update-ref', '-m', "r$rev",
2063                              $self->refname, $commit);
2064        }
2065        rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
2066                                    "$db_lock => $db ($!)\n";
2067        delete $LOCKFILES{$db_lock};
2068        if ($update_ref) {
2069                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2070                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2071                kill $sig, $$ if defined $sig;
2072        }
2073}
2074
2075sub rev_db_max {
2076        my ($self) = @_;
2077        $self->rebuild;
2078        my $db_path = $self->db_path;
2079        my @stat = stat $db_path or return 0;
2080        ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
2081        my $max = $stat[7] / 41;
2082        (($max > 0) ? $max - 1 : 0);
2083}
2084
2085sub rev_db_get {
2086        my ($self, $rev, $uuid) = @_;
2087        my $ret;
2088        my $offset = $rev * 41;
2089        my $db_path = $self->db_path($uuid);
2090        return undef unless -e $db_path;
2091        open my $fh, '<', $db_path or croak $!;
2092        if (sysseek($fh, $offset, 0) == $offset) {
2093                my $read = sysread($fh, $ret, 40);
2094                $ret = undef if ($read != 40 || $ret eq ('0'x40));
2095        }
2096        close $fh or croak $!;
2097        $ret;
2098}
2099
2100sub find_rev_before {
2101        my ($self, $rev, $eq_ok) = @_;
2102        --$rev unless $eq_ok;
2103        while ($rev > 0) {
2104                if (my $c = $self->rev_db_get($rev)) {
2105                        return ($rev, $c);
2106                }
2107                --$rev;
2108        }
2109        return (undef, undef);
2110}
2111
2112sub _new {
2113        my ($class, $repo_id, $ref_id, $path) = @_;
2114        unless (defined $repo_id && length $repo_id) {
2115                $repo_id = $Git::SVN::default_repo_id;
2116        }
2117        unless (defined $ref_id && length $ref_id) {
2118                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2119        }
2120        $_[1] = $repo_id = sanitize_remote_name($repo_id);
2121        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2122        $_[3] = $path = '' unless (defined $path);
2123        mkpath(["$ENV{GIT_DIR}/svn"]);
2124        bless {
2125                ref_id => $ref_id, dir => $dir, index => "$dir/index",
2126                path => $path, config => "$ENV{GIT_DIR}/svn/config",
2127                db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2128}
2129
2130sub db_path {
2131        my ($self, $uuid) = @_;
2132        $uuid ||= $self->ra_uuid;
2133        "$self->{db_root}.$uuid";
2134}
2135
2136sub uri_encode {
2137        my ($f) = @_;
2138        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2139        $f
2140}
2141
2142sub remove_username {
2143        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2144}
2145
2146package Git::SVN::Prompt;
2147use strict;
2148use warnings;
2149require SVN::Core;
2150use vars qw/$_no_auth_cache $_username/;
2151
2152sub simple {
2153        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2154        $may_save = undef if $_no_auth_cache;
2155        $default_username = $_username if defined $_username;
2156        if (defined $default_username && length $default_username) {
2157                if (defined $realm && length $realm) {
2158                        print STDERR "Authentication realm: $realm\n";
2159                        STDERR->flush;
2160                }
2161                $cred->username($default_username);
2162        } else {
2163                username($cred, $realm, $may_save, $pool);
2164        }
2165        $cred->password(_read_password("Password for '" .
2166                                       $cred->username . "': ", $realm));
2167        $cred->may_save($may_save);
2168        $SVN::_Core::SVN_NO_ERROR;
2169}
2170
2171sub ssl_server_trust {
2172        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2173        $may_save = undef if $_no_auth_cache;
2174        print STDERR "Error validating server certificate for '$realm':\n";
2175        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2176                print STDERR " - The certificate is not issued by a trusted ",
2177                      "authority. Use the\n",
2178                      "   fingerprint to validate the certificate manually!\n";
2179        }
2180        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2181                print STDERR " - The certificate hostname does not match.\n";
2182        }
2183        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2184                print STDERR " - The certificate is not yet valid.\n";
2185        }
2186        if ($failures & $SVN::Auth::SSL::EXPIRED) {
2187                print STDERR " - The certificate has expired.\n";
2188        }
2189        if ($failures & $SVN::Auth::SSL::OTHER) {
2190                print STDERR " - The certificate has an unknown error.\n";
2191        }
2192        printf STDERR
2193                "Certificate information:\n".
2194                " - Hostname: %s\n".
2195                " - Valid: from %s until %s\n".
2196                " - Issuer: %s\n".
2197                " - Fingerprint: %s\n",
2198                map $cert_info->$_, qw(hostname valid_from valid_until
2199                                       issuer_dname fingerprint);
2200        my $choice;
2201prompt:
2202        print STDERR $may_save ?
2203              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2204              "(R)eject or accept (t)emporarily? ";
2205        STDERR->flush;
2206        $choice = lc(substr(<STDIN> || 'R', 0, 1));
2207        if ($choice =~ /^t$/i) {
2208                $cred->may_save(undef);
2209        } elsif ($choice =~ /^r$/i) {
2210                return -1;
2211        } elsif ($may_save && $choice =~ /^p$/i) {
2212                $cred->may_save($may_save);
2213        } else {
2214                goto prompt;
2215        }
2216        $cred->accepted_failures($failures);
2217        $SVN::_Core::SVN_NO_ERROR;
2218}
2219
2220sub ssl_client_cert {
2221        my ($cred, $realm, $may_save, $pool) = @_;
2222        $may_save = undef if $_no_auth_cache;
2223        print STDERR "Client certificate filename: ";
2224        STDERR->flush;
2225        chomp(my $filename = <STDIN>);
2226        $cred->cert_file($filename);
2227        $cred->may_save($may_save);
2228        $SVN::_Core::SVN_NO_ERROR;
2229}
2230
2231sub ssl_client_cert_pw {
2232        my ($cred, $realm, $may_save, $pool) = @_;
2233        $may_save = undef if $_no_auth_cache;
2234        $cred->password(_read_password("Password: ", $realm));
2235        $cred->may_save($may_save);
2236        $SVN::_Core::SVN_NO_ERROR;
2237}
2238
2239sub username {
2240        my ($cred, $realm, $may_save, $pool) = @_;
2241        $may_save = undef if $_no_auth_cache;
2242        if (defined $realm && length $realm) {
2243                print STDERR "Authentication realm: $realm\n";
2244        }
2245        my $username;
2246        if (defined $_username) {
2247                $username = $_username;
2248        } else {
2249                print STDERR "Username: ";
2250                STDERR->flush;
2251                chomp($username = <STDIN>);
2252        }
2253        $cred->username($username);
2254        $cred->may_save($may_save);
2255        $SVN::_Core::SVN_NO_ERROR;
2256}
2257
2258sub _read_password {
2259        my ($prompt, $realm) = @_;
2260        print STDERR $prompt;
2261        STDERR->flush;
2262        require Term::ReadKey;
2263        Term::ReadKey::ReadMode('noecho');
2264        my $password = '';
2265        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2266                last if $key =~ /[\012\015]/; # \n\r
2267                $password .= $key;
2268        }
2269        Term::ReadKey::ReadMode('restore');
2270        print STDERR "\n";
2271        STDERR->flush;
2272        $password;
2273}
2274
2275package main;
2276
2277{
2278        my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2279                                $SVN::Node::dir.$SVN::Node::unknown.
2280                                $SVN::Node::none.$SVN::Node::file.
2281                                $SVN::Node::dir.$SVN::Node::unknown.
2282                                $SVN::Auth::SSL::CNMISMATCH.
2283                                $SVN::Auth::SSL::NOTYETVALID.
2284                                $SVN::Auth::SSL::EXPIRED.
2285                                $SVN::Auth::SSL::UNKNOWNCA.
2286                                $SVN::Auth::SSL::OTHER;
2287}
2288
2289package SVN::Git::Fetcher;
2290use vars qw/@ISA/;
2291use strict;
2292use warnings;
2293use Carp qw/croak/;
2294use IO::File qw//;
2295use Digest::MD5;
2296
2297# file baton members: path, mode_a, mode_b, pool, fh, blob, base
2298sub new {
2299        my ($class, $git_svn) = @_;
2300        my $self = SVN::Delta::Editor->new;
2301        bless $self, $class;
2302        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2303        $self->{empty} = {};
2304        $self->{dir_prop} = {};
2305        $self->{file_prop} = {};
2306        $self->{absent_dir} = {};
2307        $self->{absent_file} = {};
2308        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2309        $self;
2310}
2311
2312sub set_path_strip {
2313        my ($self, $path) = @_;
2314        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2315}
2316
2317sub open_root {
2318        { path => '' };
2319}
2320
2321sub open_directory {
2322        my ($self, $path, $pb, $rev) = @_;
2323        { path => $path };
2324}
2325
2326sub git_path {
2327        my ($self, $path) = @_;
2328        if ($self->{path_strip}) {
2329                $path =~ s!$self->{path_strip}!! or
2330                  die "Failed to strip path '$path' ($self->{path_strip})\n";
2331        }
2332        $path;
2333}
2334
2335sub delete_entry {
2336        my ($self, $path, $rev, $pb) = @_;
2337
2338        my $gpath = $self->git_path($path);
2339        return undef if ($gpath eq '');
2340
2341        # remove entire directories.
2342        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2343                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2344                                                     -r --name-only -z/,
2345                                                     $self->{c}, '--', $gpath);
2346                local $/ = "\0";
2347                while (<$ls>) {
2348                        chomp;
2349                        $self->{gii}->remove($_);
2350                        print "\tD\t$_\n" unless $::_q;
2351                }
2352                print "\tD\t$gpath/\n" unless $::_q;
2353                command_close_pipe($ls, $ctx);
2354                $self->{empty}->{$path} = 0
2355        } else {
2356                $self->{gii}->remove($gpath);
2357                print "\tD\t$gpath\n" unless $::_q;
2358        }
2359        undef;
2360}
2361
2362sub open_file {
2363        my ($self, $path, $pb, $rev) = @_;
2364        my $gpath = $self->git_path($path);
2365        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2366                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2367        unless (defined $mode && defined $blob) {
2368                die "$path was not found in commit $self->{c} (r$rev)\n";
2369        }
2370        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2371          pool => SVN::Pool->new, action => 'M' };
2372}
2373
2374sub add_file {
2375        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2376        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2377        delete $self->{empty}->{$dir};
2378        { path => $path, mode_a => 100644, mode_b => 100644,
2379          pool => SVN::Pool->new, action => 'A' };
2380}
2381
2382sub add_directory {
2383        my ($self, $path, $cp_path, $cp_rev) = @_;
2384        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2385        delete $self->{empty}->{$dir};
2386        $self->{empty}->{$path} = 1;
2387        { path => $path };
2388}
2389
2390sub change_dir_prop {
2391        my ($self, $db, $prop, $value) = @_;
2392        $self->{dir_prop}->{$db->{path}} ||= {};
2393        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2394        undef;
2395}
2396
2397sub absent_directory {
2398        my ($self, $path, $pb) = @_;
2399        $self->{absent_dir}->{$pb->{path}} ||= [];
2400        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2401        undef;
2402}
2403
2404sub absent_file {
2405        my ($self, $path, $pb) = @_;
2406        $self->{absent_file}->{$pb->{path}} ||= [];
2407        push @{$self->{absent_file}->{$pb->{path}}}, $path;
2408        undef;
2409}
2410
2411sub change_file_prop {
2412        my ($self, $fb, $prop, $value) = @_;
2413        if ($prop eq 'svn:executable') {
2414                if ($fb->{mode_b} != 120000) {
2415                        $fb->{mode_b} = defined $value ? 100755 : 100644;
2416                }
2417        } elsif ($prop eq 'svn:special') {
2418                $fb->{mode_b} = defined $value ? 120000 : 100644;
2419        } else {
2420                $self->{file_prop}->{$fb->{path}} ||= {};
2421                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2422        }
2423        undef;
2424}
2425
2426sub apply_textdelta {
2427        my ($self, $fb, $exp) = @_;
2428        my $fh = IO::File->new_tmpfile;
2429        $fh->autoflush(1);
2430        # $fh gets auto-closed() by SVN::TxDelta::apply(),
2431        # (but $base does not,) so dup() it for reading in close_file
2432        open my $dup, '<&', $fh or croak $!;
2433        my $base = IO::File->new_tmpfile;
2434        $base->autoflush(1);
2435        if ($fb->{blob}) {
2436                defined (my $pid = fork) or croak $!;
2437                if (!$pid) {
2438                        open STDOUT, '>&', $base or croak $!;
2439                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2440                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2441                }
2442                waitpid $pid, 0;
2443                croak $? if $?;
2444
2445                if (defined $exp) {
2446                        seek $base, 0, 0 or croak $!;
2447                        my $md5 = Digest::MD5->new;
2448                        $md5->addfile($base);
2449                        my $got = $md5->hexdigest;
2450                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2451                            "expected: $exp\n",
2452                            "     got: $got\n" if ($got ne $exp);
2453                }
2454        }
2455        seek $base, 0, 0 or croak $!;
2456        $fb->{fh} = $dup;
2457        $fb->{base} = $base;
2458        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2459}
2460
2461sub close_file {
2462        my ($self, $fb, $exp) = @_;
2463        my $hash;
2464        my $path = $self->git_path($fb->{path});
2465        if (my $fh = $fb->{fh}) {
2466                seek($fh, 0, 0) or croak $!;
2467                my $md5 = Digest::MD5->new;
2468                $md5->addfile($fh);
2469                my $got = $md5->hexdigest;
2470                die "Checksum mismatch: $path\n",
2471                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
2472                sysseek($fh, 0, 0) or croak $!;
2473                if ($fb->{mode_b} == 120000) {
2474                        sysread($fh, my $buf, 5) == 5 or croak $!;
2475                        $buf eq 'link ' or die "$path has mode 120000",
2476                                               "but is not a link\n";
2477                }
2478                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2479                if (!$pid) {
2480                        open STDIN, '<&', $fh or croak $!;
2481                        exec qw/git-hash-object -w --stdin/ or croak $!;
2482                }
2483                chomp($hash = do { local $/; <$out> });
2484                close $out or croak $!;
2485                close $fh or croak $!;
2486                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2487                close $fb->{base} or croak $!;
2488        } else {
2489                $hash = $fb->{blob} or die "no blob information\n";
2490        }
2491        $fb->{pool}->clear;
2492        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2493        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2494        undef;
2495}
2496
2497sub abort_edit {
2498        my $self = shift;
2499        $self->{nr} = $self->{gii}->{nr};
2500        delete $self->{gii};
2501        $self->SUPER::abort_edit(@_);
2502}
2503
2504sub close_edit {
2505        my $self = shift;
2506        $self->{git_commit_ok} = 1;
2507        $self->{nr} = $self->{gii}->{nr};
2508        delete $self->{gii};
2509        $self->SUPER::close_edit(@_);
2510}
2511
2512package SVN::Git::Editor;
2513use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2514use strict;
2515use warnings;
2516use Carp qw/croak/;
2517use IO::File;
2518use Digest::MD5;
2519
2520sub new {
2521        my ($class, $opts) = @_;
2522        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2523                die "$_ required!\n" unless (defined $opts->{$_});
2524        }
2525
2526        my $pool = SVN::Pool->new;
2527        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2528        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2529                                     $opts->{r}, $mods);
2530
2531        # $opts->{ra} functions should not be used after this:
2532        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
2533                                                $opts->{editor_cb}, $pool);
2534        my $self = SVN::Delta::Editor->new(@ce, $pool);
2535        bless $self, $class;
2536        foreach (qw/svn_path r tree_a tree_b/) {
2537                $self->{$_} = $opts->{$_};
2538        }
2539        $self->{url} = $opts->{ra}->{url};
2540        $self->{mods} = $mods;
2541        $self->{types} = $types;
2542        $self->{pool} = $pool;
2543        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2544        $self->{rm} = { };
2545        $self->{path_prefix} = length $self->{svn_path} ?
2546                               "$self->{svn_path}/" : '';
2547        return $self;
2548}
2549
2550sub generate_diff {
2551        my ($tree_a, $tree_b) = @_;
2552        my @diff_tree = qw(diff-tree -z -r);
2553        if ($_cp_similarity) {
2554                push @diff_tree, "-C$_cp_similarity";
2555        } else {
2556                push @diff_tree, '-C';
2557        }
2558        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2559        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2560        push @diff_tree, $tree_a, $tree_b;
2561        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2562        local $/ = "\0";
2563        my $state = 'meta';
2564        my @mods;
2565        while (<$diff_fh>) {
2566                chomp $_; # this gets rid of the trailing "\0"
2567                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2568                                        $::sha1\s($::sha1)\s
2569                                        ([MTCRAD])\d*$/xo) {
2570                        push @mods, {   mode_a => $1, mode_b => $2,
2571                                        sha1_b => $3, chg => $4 };
2572                        if ($4 =~ /^(?:C|R)$/) {
2573                                $state = 'file_a';
2574                        } else {
2575                                $state = 'file_b';
2576                        }
2577                } elsif ($state eq 'file_a') {
2578                        my $x = $mods[$#mods] or croak "Empty array\n";
2579                        if ($x->{chg} !~ /^(?:C|R)$/) {
2580                                croak "Error parsing $_, $x->{chg}\n";
2581                        }
2582                        $x->{file_a} = $_;
2583                        $state = 'file_b';
2584                } elsif ($state eq 'file_b') {
2585                        my $x = $mods[$#mods] or croak "Empty array\n";
2586                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2587                                croak "Error parsing $_, $x->{chg}\n";
2588                        }
2589                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2590                                croak "Error parsing $_, $x->{chg}\n";
2591                        }
2592                        $x->{file_b} = $_;
2593                        $state = 'meta';
2594                } else {
2595                        croak "Error parsing $_\n";
2596                }
2597        }
2598        command_close_pipe($diff_fh, $ctx);
2599        \@mods;
2600}
2601
2602sub check_diff_paths {
2603        my ($ra, $pfx, $rev, $mods) = @_;
2604        my %types;
2605        $pfx .= '/' if length $pfx;
2606
2607        sub type_diff_paths {
2608                my ($ra, $types, $path, $rev) = @_;
2609                my @p = split m#/+#, $path;
2610                my $c = shift @p;
2611                unless (defined $types->{$c}) {
2612                        $types->{$c} = $ra->check_path($c, $rev);
2613                }
2614                while (@p) {
2615                        $c .= '/' . shift @p;
2616                        next if defined $types->{$c};
2617                        $types->{$c} = $ra->check_path($c, $rev);
2618                }
2619        }
2620
2621        foreach my $m (@$mods) {
2622                foreach my $f (qw/file_a file_b/) {
2623                        next unless defined $m->{$f};
2624                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2625                        if (length $pfx.$dir && ! defined $types{$dir}) {
2626                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2627                        }
2628                }
2629        }
2630        \%types;
2631}
2632
2633sub split_path {
2634        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2635}
2636
2637sub repo_path {
2638        my ($self, $path) = @_;
2639        $self->{path_prefix}.(defined $path ? $path : '');
2640}
2641
2642sub url_path {
2643        my ($self, $path) = @_;
2644        $self->{url} . '/' . $self->repo_path($path);
2645}
2646
2647sub rmdirs {
2648        my ($self) = @_;
2649        my $rm = $self->{rm};
2650        delete $rm->{''}; # we never delete the url we're tracking
2651        return unless %$rm;
2652
2653        foreach (keys %$rm) {
2654                my @d = split m#/#, $_;
2655                my $c = shift @d;
2656                $rm->{$c} = 1;
2657                while (@d) {
2658                        $c .= '/' . shift @d;
2659                        $rm->{$c} = 1;
2660                }
2661        }
2662        delete $rm->{$self->{svn_path}};
2663        delete $rm->{''}; # we never delete the url we're tracking
2664        return unless %$rm;
2665
2666        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2667                                             $self->{tree_b});
2668        local $/ = "\0";
2669        while (<$fh>) {
2670                chomp;
2671                my @dn = split m#/#, $_;
2672                while (pop @dn) {
2673                        delete $rm->{join '/', @dn};
2674                }
2675                unless (%$rm) {
2676                        close $fh;
2677                        return;
2678                }
2679        }
2680        command_close_pipe($fh, $ctx);
2681
2682        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2683        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2684                $self->close_directory($bat->{$d}, $p);
2685                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2686                print "\tD+\t$d/\n" unless $::_q;
2687                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2688                delete $bat->{$d};
2689        }
2690}
2691
2692sub open_or_add_dir {
2693        my ($self, $full_path, $baton) = @_;
2694        my $t = $self->{types}->{$full_path};
2695        if (!defined $t) {
2696                die "$full_path not known in r$self->{r} or we have a bug!\n";
2697        }
2698        if ($t == $SVN::Node::none) {
2699                return $self->add_directory($full_path, $baton,
2700                                                undef, -1, $self->{pool});
2701        } elsif ($t == $SVN::Node::dir) {
2702                return $self->open_directory($full_path, $baton,
2703                                                $self->{r}, $self->{pool});
2704        }
2705        print STDERR "$full_path already exists in repository at ",
2706                "r$self->{r} and it is not a directory (",
2707                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2708        exit 1;
2709}
2710
2711sub ensure_path {
2712        my ($self, $path) = @_;
2713        my $bat = $self->{bat};
2714        my $repo_path = $self->repo_path($path);
2715        return $bat->{''} unless (length $repo_path);
2716        my @p = split m#/+#, $repo_path;
2717        my $c = shift @p;
2718        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2719        while (@p) {
2720                my $c0 = $c;
2721                $c .= '/' . shift @p;
2722                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2723        }
2724        return $bat->{$c};
2725}
2726
2727sub A {
2728        my ($self, $m) = @_;
2729        my ($dir, $file) = split_path($m->{file_b});
2730        my $pbat = $self->ensure_path($dir);
2731        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2732                                        undef, -1);
2733        print "\tA\t$m->{file_b}\n" unless $::_q;
2734        $self->chg_file($fbat, $m);
2735        $self->close_file($fbat,undef,$self->{pool});
2736}
2737
2738sub C {
2739        my ($self, $m) = @_;
2740        my ($dir, $file) = split_path($m->{file_b});
2741        my $pbat = $self->ensure_path($dir);
2742        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2743                                $self->url_path($m->{file_a}), $self->{r});
2744        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2745        $self->chg_file($fbat, $m);
2746        $self->close_file($fbat,undef,$self->{pool});
2747}
2748
2749sub delete_entry {
2750        my ($self, $path, $pbat) = @_;
2751        my $rpath = $self->repo_path($path);
2752        my ($dir, $file) = split_path($rpath);
2753        $self->{rm}->{$dir} = 1;
2754        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2755}
2756
2757sub R {
2758        my ($self, $m) = @_;
2759        my ($dir, $file) = split_path($m->{file_b});
2760        my $pbat = $self->ensure_path($dir);
2761        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2762                                $self->url_path($m->{file_a}), $self->{r});
2763        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2764        $self->chg_file($fbat, $m);
2765        $self->close_file($fbat,undef,$self->{pool});
2766
2767        ($dir, $file) = split_path($m->{file_a});
2768        $pbat = $self->ensure_path($dir);
2769        $self->delete_entry($m->{file_a}, $pbat);
2770}
2771
2772sub M {
2773        my ($self, $m) = @_;
2774        my ($dir, $file) = split_path($m->{file_b});
2775        my $pbat = $self->ensure_path($dir);
2776        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2777                                $pbat,$self->{r},$self->{pool});
2778        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2779        $self->chg_file($fbat, $m);
2780        $self->close_file($fbat,undef,$self->{pool});
2781}
2782
2783sub T { shift->M(@_) }
2784
2785sub change_file_prop {
2786        my ($self, $fbat, $pname, $pval) = @_;
2787        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2788}
2789
2790sub chg_file {
2791        my ($self, $fbat, $m) = @_;
2792        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2793                $self->change_file_prop($fbat,'svn:executable','*');
2794        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2795                $self->change_file_prop($fbat,'svn:executable',undef);
2796        }
2797        my $fh = IO::File->new_tmpfile or croak $!;
2798        if ($m->{mode_b} =~ /^120/) {
2799                print $fh 'link ' or croak $!;
2800                $self->change_file_prop($fbat,'svn:special','*');
2801        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2802                $self->change_file_prop($fbat,'svn:special',undef);
2803        }
2804        defined(my $pid = fork) or croak $!;
2805        if (!$pid) {
2806                open STDOUT, '>&', $fh or croak $!;
2807                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2808        }
2809        waitpid $pid, 0;
2810        croak $? if $?;
2811        $fh->flush == 0 or croak $!;
2812        seek $fh, 0, 0 or croak $!;
2813
2814        my $md5 = Digest::MD5->new;
2815        $md5->addfile($fh) or croak $!;
2816        seek $fh, 0, 0 or croak $!;
2817
2818        my $exp = $md5->hexdigest;
2819        my $pool = SVN::Pool->new;
2820        my $atd = $self->apply_textdelta($fbat, undef, $pool);
2821        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2822        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2823        $pool->clear;
2824
2825        close $fh or croak $!;
2826}
2827
2828sub D {
2829        my ($self, $m) = @_;
2830        my ($dir, $file) = split_path($m->{file_b});
2831        my $pbat = $self->ensure_path($dir);
2832        print "\tD\t$m->{file_b}\n" unless $::_q;
2833        $self->delete_entry($m->{file_b}, $pbat);
2834}
2835
2836sub close_edit {
2837        my ($self) = @_;
2838        my ($p,$bat) = ($self->{pool}, $self->{bat});
2839        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2840                $self->close_directory($bat->{$_}, $p);
2841        }
2842        $self->SUPER::close_edit($p);
2843        $p->clear;
2844}
2845
2846sub abort_edit {
2847        my ($self) = @_;
2848        $self->SUPER::abort_edit($self->{pool});
2849}
2850
2851sub DESTROY {
2852        my $self = shift;
2853        $self->SUPER::DESTROY(@_);
2854        $self->{pool}->clear;
2855}
2856
2857# this drives the editor
2858sub apply_diff {
2859        my ($self) = @_;
2860        my $mods = $self->{mods};
2861        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2862        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2863                my $f = $m->{chg};
2864                if (defined $o{$f}) {
2865                        $self->$f($m);
2866                } else {
2867                        fatal("Invalid change type: $f\n");
2868                }
2869        }
2870        $self->rmdirs if $_rmdir;
2871        if (@$mods == 0) {
2872                $self->abort_edit;
2873        } else {
2874                $self->close_edit;
2875        }
2876        return scalar @$mods;
2877}
2878
2879package Git::SVN::Ra;
2880use vars qw/@ISA $config_dir $_log_window_size/;
2881use strict;
2882use warnings;
2883my ($can_do_switch, %ignored_err, $RA);
2884
2885BEGIN {
2886        # enforce temporary pool usage for some simple functions
2887        my $e;
2888        foreach (qw/get_latest_revnum get_uuid get_repos_root/) {
2889                $e .= "sub $_ {
2890                        my \$self = shift;
2891                        my \$pool = SVN::Pool->new;
2892                        my \@ret = \$self->SUPER::$_(\@_,\$pool);
2893                        \$pool->clear;
2894                        wantarray ? \@ret : \$ret[0]; }\n";
2895        }
2896
2897        # get_dir needs $pool held in cache for dirents to work,
2898        # check_path is cacheable and rev_proplist is close enough
2899        # for our purposes.
2900        foreach (qw/check_path get_dir rev_proplist/) {
2901                $e .= "my \%${_}_cache; my \$${_}_rev = 0; sub $_ {
2902                        my \$self = shift;
2903                        my \$r = pop;
2904                        my \$k = join(\"\\0\", \@_);
2905                        if (my \$x = \$${_}_cache{\$r}->{\$k}) {
2906                                return wantarray ? \@\$x : \$x->[0];
2907                        }
2908                        my \$pool = SVN::Pool->new;
2909                        my \@ret = \$self->SUPER::$_(\@_, \$r, \$pool);
2910                        if (\$r != \$${_}_rev) {
2911                                \%${_}_cache = ( pool => [] );
2912                                \$${_}_rev = \$r;
2913                        }
2914                        \$${_}_cache{\$r}->{\$k} = \\\@ret;
2915                        push \@{\$${_}_cache{pool}}, \$pool;
2916                        wantarray ? \@ret : \$ret[0]; }\n";
2917        }
2918        $e .= "\n1;";
2919        eval $e or die $@;
2920}
2921
2922sub new {
2923        my ($class, $url) = @_;
2924        $url =~ s!/+$!!;
2925        return $RA if ($RA && $RA->{url} eq $url);
2926        $RA->{pool}->clear if $RA;
2927
2928        SVN::_Core::svn_config_ensure($config_dir, undef);
2929        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2930            SVN::Client::get_simple_provider(),
2931            SVN::Client::get_ssl_server_trust_file_provider(),
2932            SVN::Client::get_simple_prompt_provider(
2933              \&Git::SVN::Prompt::simple, 2),
2934            SVN::Client::get_ssl_client_cert_prompt_provider(
2935              \&Git::SVN::Prompt::ssl_client_cert, 2),
2936            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2937              \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2938            SVN::Client::get_username_provider(),
2939            SVN::Client::get_ssl_server_trust_prompt_provider(
2940              \&Git::SVN::Prompt::ssl_server_trust),
2941            SVN::Client::get_username_prompt_provider(
2942              \&Git::SVN::Prompt::username, 2),
2943          ]);
2944        my $config = SVN::Core::config_get_config($config_dir);
2945        my $self = SVN::Ra->new(url => $url, auth => $baton,
2946                              config => $config,
2947                              pool => SVN::Pool->new,
2948                              auth_provider_callbacks => $callbacks);
2949        $self->{svn_path} = $url;
2950        $self->{repos_root} = $self->get_repos_root;
2951        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
2952        $RA = bless $self, $class;
2953}
2954
2955sub DESTROY {
2956        # do not call the real DESTROY since we store ourselves in $RA
2957}
2958
2959sub get_log {
2960        my ($self, @args) = @_;
2961        my $pool = SVN::Pool->new;
2962        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2963        my $ret = $self->SUPER::get_log(@args, $pool);
2964        $pool->clear;
2965        $ret;
2966}
2967
2968sub get_commit_editor {
2969        my ($self, $log, $cb, $pool) = @_;
2970        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2971        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2972}
2973
2974sub gs_do_update {
2975        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2976        my $new = ($rev_a == $rev_b);
2977        my $path = $gs->{path};
2978
2979        if ($new && -e $gs->{index}) {
2980                unlink $gs->{index} or die
2981                  "Couldn't unlink index: $gs->{index}: $!\n";
2982        }
2983        my $pool = SVN::Pool->new;
2984        $editor->set_path_strip($path);
2985        my (@pc) = split m#/#, $path;
2986        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2987                                        1, $editor, $pool);
2988        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2989
2990        # Since we can't rely on svn_ra_reparent being available, we'll
2991        # just have to do some magic with set_path to make it so
2992        # we only want a partial path.
2993        my $sp = '';
2994        my $final = join('/', @pc);
2995        while (@pc) {
2996                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2997                $sp .= '/' if length $sp;
2998                $sp .= shift @pc;
2999        }
3000        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3001
3002        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3003
3004        $reporter->finish_report($pool);
3005        $pool->clear;
3006        $editor->{git_commit_ok};
3007}
3008
3009# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3010# svn_ra_reparent didn't work before 1.4)
3011sub gs_do_switch {
3012        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3013        my $path = $gs->{path};
3014        my $pool = SVN::Pool->new;
3015
3016        my $full_url = $self->{url};
3017        my $old_url = $full_url;
3018        $full_url .= "/$path" if length $path;
3019        my ($ra, $reparented);
3020        if ($old_url ne $full_url) {
3021                if ($old_url !~ m#^svn(\+ssh)?://#) {
3022                        SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3023                                                  $pool);
3024                        $self->{url} = $full_url;
3025                        $reparented = 1;
3026                } else {
3027                        $ra = Git::SVN::Ra->new($full_url);
3028                }
3029        }
3030        $ra ||= $self;
3031        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3032        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3033        $reporter->set_path('', $rev_a, 0, @lock, $pool);
3034        $reporter->finish_report($pool);
3035
3036        if ($reparented) {
3037                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3038                $self->{url} = $old_url;
3039        }
3040
3041        $pool->clear;
3042        $editor->{git_commit_ok};
3043}
3044
3045sub gs_fetch_loop_common {
3046        my ($self, $base, $head, $gsv, $globs) = @_;
3047        return if ($base > $head);
3048        my $inc = $_log_window_size;
3049        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3050        my %common;
3051        my $common_max = scalar @$gsv;
3052
3053        foreach my $gs (@$gsv) {
3054                my @tmp = split m#/#, $gs->{path};
3055                my $p = '';
3056                foreach (@tmp) {
3057                        $p .= length($p) ? "/$_" : $_;
3058                        $common{$p} ||= 0;
3059                        $common{$p}++;
3060                }
3061        }
3062        $globs ||= [];
3063        $common_max += scalar @$globs;
3064        foreach my $glob (@$globs) {
3065                my @tmp = split m#/#, $glob->{path}->{left};
3066                my $p = '';
3067                foreach (@tmp) {
3068                        $p .= length($p) ? "/$_" : $_;
3069                        $common{$p} ||= 0;
3070                        $common{$p}++;
3071                }
3072        }
3073
3074        my $longest_path = '';
3075        foreach (sort {length $b <=> length $a} keys %common) {
3076                if ($common{$_} == $common_max) {
3077                        $longest_path = $_;
3078                        last;
3079                }
3080        }
3081        while (1) {
3082                my %revs;
3083                my $err;
3084                my $err_handler = $SVN::Error::handler;
3085                $SVN::Error::handler = sub {
3086                        ($err) = @_;
3087                        skip_unknown_revs($err);
3088                };
3089                sub _cb {
3090                        my ($paths, $r, $author, $date, $log) = @_;
3091                        [ dup_changed_paths($paths),
3092                          { author => $author, date => $date, log => $log } ];
3093                }
3094                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3095                               sub { $revs{$_[1]} = _cb(@_) });
3096                if ($err && $max >= $head) {
3097                        print STDERR "Path '$longest_path' ",
3098                                     "was probably deleted:\n",
3099                                     $err->expanded_message,
3100                                     "\nWill attempt to follow ",
3101                                     "revisions r$min .. r$max ",
3102                                     "committed before the deletion\n";
3103                        my $hi = $max;
3104                        while (--$hi >= $min) {
3105                                my $ok;
3106                                $self->get_log([$longest_path], $min, $hi,
3107                                               0, 1, 1, sub {
3108                                               $ok ||= $_[1];
3109                                               $revs{$_[1]} = _cb(@_) });
3110                                if ($ok) {
3111                                        print STDERR "r$min .. r$ok OK\n";
3112                                        last;
3113                                }
3114                        }
3115                }
3116                $SVN::Error::handler = $err_handler;
3117
3118                my %exists = map { $_->{path} => $_ } @$gsv;
3119                foreach my $r (sort {$a <=> $b} keys %revs) {
3120                        my ($paths, $logged) = @{$revs{$r}};
3121
3122                        foreach my $gs ($self->match_globs(\%exists, $paths,
3123                                                           $globs, $r)) {
3124                                if ($gs->rev_db_max >= $r) {
3125                                        next;
3126                                }
3127                                next unless $gs->match_paths($paths, $r);
3128                                $gs->{logged_rev_props} = $logged;
3129                                if (my $last_commit = $gs->last_commit) {
3130                                        $gs->assert_index_clean($last_commit);
3131                                }
3132                                my $log_entry = $gs->do_fetch($paths, $r);
3133                                if ($log_entry) {
3134                                        $gs->do_git_commit($log_entry);
3135                                }
3136                        }
3137                        foreach my $g (@$globs) {
3138                                my $k = "svn-remote.$g->{remote}." .
3139                                        "$g->{t}-maxRev";
3140                                Git::SVN::tmp_config($k, $r);
3141                        }
3142                }
3143                # pre-fill the .rev_db since it'll eventually get filled in
3144                # with '0' x40 if something new gets committed
3145                foreach my $gs (@$gsv) {
3146                        next if defined $gs->rev_db_get($max);
3147                        $gs->rev_db_set($max, 0 x40);
3148                }
3149                foreach my $g (@$globs) {
3150                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3151                        Git::SVN::tmp_config($k, $max);
3152                }
3153                last if $max >= $head;
3154                $min = $max + 1;
3155                $max += $inc;
3156                $max = $head if ($max > $head);
3157        }
3158}
3159
3160sub match_globs {
3161        my ($self, $exists, $paths, $globs, $r) = @_;
3162
3163        sub get_dir_check {
3164                my ($self, $exists, $g, $r) = @_;
3165                my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3166                return unless scalar @x == 3;
3167                my $dirents = $x[0];
3168                foreach my $de (keys %$dirents) {
3169                        next if $dirents->{$de}->kind != $SVN::Node::dir;
3170                        my $p = $g->{path}->full_path($de);
3171                        next if $exists->{$p};
3172                        next if (length $g->{path}->{right} &&
3173                                 ($self->check_path($p, $r) !=
3174                                  $SVN::Node::dir));
3175                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3176                                         $g->{ref}->full_path($de), 1);
3177                }
3178        }
3179        foreach my $g (@$globs) {
3180                if (my $path = $paths->{"/$g->{path}->{left}"}) {
3181                        if ($path->{action} =~ /^[AR]$/) {
3182                                get_dir_check($self, $exists, $g, $r);
3183                        }
3184                }
3185                foreach (keys %$paths) {
3186                        if (/$g->{path}->{left_regex}/ &&
3187                            !/$g->{path}->{regex}/) {
3188                                next if $paths->{$_}->{action} !~ /^[AR]$/;
3189                                get_dir_check($self, $exists, $g, $r);
3190                        }
3191                        next unless /$g->{path}->{regex}/;
3192                        my $p = $1;
3193                        my $pathname = $g->{path}->full_path($p);
3194                        next if $exists->{$pathname};
3195                        next if ($self->check_path($pathname, $r) !=
3196                                 $SVN::Node::dir);
3197                        $exists->{$pathname} = Git::SVN->init(
3198                                              $self->{url}, $pathname, undef,
3199                                              $g->{ref}->full_path($p), 1);
3200                }
3201                my $c = '';
3202                foreach (split m#/#, $g->{path}->{left}) {
3203                        $c .= "/$_";
3204                        next unless ($paths->{$c} &&
3205                                     ($paths->{$c}->{action} =~ /^[AR]$/));
3206                        get_dir_check($self, $exists, $g, $r);
3207                }
3208        }
3209        values %$exists;
3210}
3211
3212sub minimize_url {
3213        my ($self) = @_;
3214        return $self->{url} if ($self->{url} eq $self->{repos_root});
3215        my $url = $self->{repos_root};
3216        my @components = split(m!/!, $self->{svn_path});
3217        my $c = '';
3218        do {
3219                $url .= "/$c" if length $c;
3220                eval { (ref $self)->new($url)->get_latest_revnum };
3221        } while ($@ && ($c = shift @components));
3222        $url;
3223}
3224
3225sub can_do_switch {
3226        my $self = shift;
3227        unless (defined $can_do_switch) {
3228                my $pool = SVN::Pool->new;
3229                my $rep = eval {
3230                        $self->do_switch(1, '', 0, $self->{url},
3231                                         SVN::Delta::Editor->new, $pool);
3232                };
3233                if ($@) {
3234                        $can_do_switch = 0;
3235                } else {
3236                        $rep->abort_report($pool);
3237                        $can_do_switch = 1;
3238                }
3239                $pool->clear;
3240        }
3241        $can_do_switch;
3242}
3243
3244sub skip_unknown_revs {
3245        my ($err) = @_;
3246        my $errno = $err->apr_err();
3247        # Maybe the branch we're tracking didn't
3248        # exist when the repo started, so it's
3249        # not an error if it doesn't, just continue
3250        #
3251        # Wonderfully consistent library, eh?
3252        # 160013 - svn:// and file://
3253        # 175002 - http(s)://
3254        # 175007 - http(s):// (this repo required authorization, too...)
3255        #   More codes may be discovered later...
3256        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3257                my $err_key = $err->expanded_message;
3258                # revision numbers change every time, filter them out
3259                $err_key =~ s/\d+/\0/g;
3260                $err_key = "$errno\0$err_key";
3261                unless ($ignored_err{$err_key}) {
3262                        warn "W: Ignoring error from SVN, path probably ",
3263                             "does not exist: ($errno): ",
3264                             $err->expanded_message,"\n";
3265                        $ignored_err{$err_key} = 1;
3266                }
3267                return;
3268        }
3269        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3270}
3271
3272# svn_log_changed_path_t objects passed to get_log are likely to be
3273# overwritten even if only the refs are copied to an external variable,
3274# so we should dup the structures in their entirety.  Using an externally
3275# passed pool (instead of our temporary and quickly cleared pool in
3276# Git::SVN::Ra) does not help matters at all...
3277sub dup_changed_paths {
3278        my ($paths) = @_;
3279        return undef unless $paths;
3280        my %ret;
3281        foreach my $p (keys %$paths) {
3282                my $i = $paths->{$p};
3283                my %s = map { $_ => $i->$_ }
3284                              qw/copyfrom_path copyfrom_rev action/;
3285                $ret{$p} = \%s;
3286        }
3287        \%ret;
3288}
3289
3290package Git::SVN::Log;
3291use strict;
3292use warnings;
3293use POSIX qw/strftime/;
3294use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3295            %rusers $show_commit $incremental/;
3296my $l_fmt;
3297
3298sub cmt_showable {
3299        my ($c) = @_;
3300        return 1 if defined $c->{r};
3301
3302        # big commit message got truncated by the 16k pretty buffer in rev-list
3303        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3304                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3305                @{$c->{l}} = ();
3306                my @log = command(qw/cat-file commit/, $c->{c});
3307
3308                # shift off the headers
3309                shift @log while ($log[0] ne '');
3310                shift @log;
3311
3312                # TODO: make $c->{l} not have a trailing newline in the future
3313                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
3314
3315                (undef, $c->{r}, undef) = ::extract_metadata(
3316                                (grep(/^git-svn-id: /, @log))[-1]);
3317        }
3318        return defined $c->{r};
3319}
3320
3321sub log_use_color {
3322        return 1 if $color;
3323        my ($dc, $dcvar);
3324        $dcvar = 'color.diff';
3325        $dc = `git-config --get $dcvar`;
3326        if ($dc eq '') {
3327                # nothing at all; fallback to "diff.color"
3328                $dcvar = 'diff.color';
3329                $dc = `git-config --get $dcvar`;
3330        }
3331        chomp($dc);
3332        if ($dc eq 'auto') {
3333                my $pc;
3334                $pc = `git-config --get color.pager`;
3335                if ($pc eq '') {
3336                        # does not have it -- fallback to pager.color
3337                        $pc = `git-config --bool --get pager.color`;
3338                }
3339                else {
3340                        $pc = `git-config --bool --get color.pager`;
3341                        if ($?) {
3342                                $pc = 'false';
3343                        }
3344                }
3345                chomp($pc);
3346                if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3347                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3348                }
3349                return 0;
3350        }
3351        return 0 if $dc eq 'never';
3352        return 1 if $dc eq 'always';
3353        chomp($dc = `git-config --bool --get $dcvar`);
3354        return ($dc eq 'true');
3355}
3356
3357sub git_svn_log_cmd {
3358        my ($r_min, $r_max, @args) = @_;
3359        my $head = 'HEAD';
3360        foreach my $x (@args) {
3361                last if $x eq '--';
3362                next unless ::verify_ref("$x^0");
3363                $head = $x;
3364                last;
3365        }
3366
3367        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
3368        $gs ||= Git::SVN->_new;
3369        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3370                   $gs->refname);
3371        push @cmd, '-r' unless $non_recursive;
3372        push @cmd, qw/--raw --name-status/ if $verbose;
3373        push @cmd, '--color' if log_use_color();
3374        return @cmd unless defined $r_max;
3375        if ($r_max == $r_min) {
3376                push @cmd, '--max-count=1';
3377                if (my $c = $gs->rev_db_get($r_max)) {
3378                        push @cmd, $c;
3379                }
3380        } else {
3381                my ($c_min, $c_max);
3382                $c_max = $gs->rev_db_get($r_max);
3383                $c_min = $gs->rev_db_get($r_min);
3384                if (defined $c_min && defined $c_max) {
3385                        if ($r_max > $r_max) {
3386                                push @cmd, "$c_min..$c_max";
3387                        } else {
3388                                push @cmd, "$c_max..$c_min";
3389                        }
3390                } elsif ($r_max > $r_min) {
3391                        push @cmd, $c_max;
3392                } else {
3393                        push @cmd, $c_min;
3394                }
3395        }
3396        return @cmd;
3397}
3398
3399# adapted from pager.c
3400sub config_pager {
3401        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3402        if (!defined $pager) {
3403                $pager = 'less';
3404        } elsif (length $pager == 0 || $pager eq 'cat') {
3405                $pager = undef;
3406        }
3407}
3408
3409sub run_pager {
3410        return unless -t *STDOUT;
3411        pipe my $rfd, my $wfd or return;
3412        defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
3413        if (!$pid) {
3414                open STDOUT, '>&', $wfd or
3415                                     ::fatal "Can't redirect to stdout: $!\n";
3416                return;
3417        }
3418        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
3419        $ENV{LESS} ||= 'FRSX';
3420        exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
3421}
3422
3423sub tz_to_s_offset {
3424        my ($tz) = @_;
3425        $tz =~ s/(\d\d)$//;
3426        return ($1 * 60) + ($tz * 3600);
3427}
3428
3429sub get_author_info {
3430        my ($dest, $author, $t, $tz) = @_;
3431        $author =~ s/(?:^\s*|\s*$)//g;
3432        $dest->{a_raw} = $author;
3433        my $au;
3434        if ($::_authors) {
3435                $au = $rusers{$author} || undef;
3436        }
3437        if (!$au) {
3438                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3439        }
3440        $dest->{t} = $t;
3441        $dest->{tz} = $tz;
3442        $dest->{a} = $au;
3443        # Date::Parse isn't in the standard Perl distro :(
3444        if ($tz =~ s/^\+//) {
3445                $t += tz_to_s_offset($tz);
3446        } elsif ($tz =~ s/^\-//) {
3447                $t -= tz_to_s_offset($tz);
3448        }
3449        $dest->{t_utc} = $t;
3450}
3451
3452sub process_commit {
3453        my ($c, $r_min, $r_max, $defer) = @_;
3454        if (defined $r_min && defined $r_max) {
3455                if ($r_min == $c->{r} && $r_min == $r_max) {
3456                        show_commit($c);
3457                        return 0;
3458                }
3459                return 1 if $r_min == $r_max;
3460                if ($r_min < $r_max) {
3461                        # we need to reverse the print order
3462                        return 0 if (defined $limit && --$limit < 0);
3463                        push @$defer, $c;
3464                        return 1;
3465                }
3466                if ($r_min != $r_max) {
3467                        return 1 if ($r_min < $c->{r});
3468                        return 1 if ($r_max > $c->{r});
3469                }
3470        }
3471        return 0 if (defined $limit && --$limit < 0);
3472        show_commit($c);
3473        return 1;
3474}
3475
3476sub show_commit {
3477        my $c = shift;
3478        if ($oneline) {
3479                my $x = "\n";
3480                if (my $l = $c->{l}) {
3481                        while ($l->[0] =~ /^\s*$/) { shift @$l }
3482                        $x = $l->[0];
3483                }
3484                $l_fmt ||= 'A' . length($c->{r});
3485                print 'r',pack($l_fmt, $c->{r}),' | ';
3486                print "$c->{c} | " if $show_commit;
3487                print $x;
3488        } else {
3489                show_commit_normal($c);
3490        }
3491}
3492
3493sub show_commit_changed_paths {
3494        my ($c) = @_;
3495        return unless $c->{changed};
3496        print "Changed paths:\n", @{$c->{changed}};
3497}
3498
3499sub show_commit_normal {
3500        my ($c) = @_;
3501        print '-' x72, "\nr$c->{r} | ";
3502        print "$c->{c} | " if $show_commit;
3503        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3504                                 localtime($c->{t_utc})), ' | ';
3505        my $nr_line = 0;
3506
3507        if (my $l = $c->{l}) {
3508                while ($l->[$#$l] eq "\n" && $#$l > 0
3509                                          && $l->[($#$l - 1)] eq "\n") {
3510                        pop @$l;
3511                }
3512                $nr_line = scalar @$l;
3513                if (!$nr_line) {
3514                        print "1 line\n\n\n";
3515                } else {
3516                        if ($nr_line == 1) {
3517                                $nr_line = '1 line';
3518                        } else {
3519                                $nr_line .= ' lines';
3520                        }
3521                        print $nr_line, "\n";
3522                        show_commit_changed_paths($c);
3523                        print "\n";
3524                        print $_ foreach @$l;
3525                }
3526        } else {
3527                print "1 line\n";
3528                show_commit_changed_paths($c);
3529                print "\n";
3530
3531        }
3532        foreach my $x (qw/raw stat diff/) {
3533                if ($c->{$x}) {
3534                        print "\n";
3535                        print $_ foreach @{$c->{$x}}
3536                }
3537        }
3538}
3539
3540sub cmd_show_log {
3541        my (@args) = @_;
3542        my ($r_min, $r_max);
3543        my $r_last = -1; # prevent dupes
3544        if (defined $TZ) {
3545                $ENV{TZ} = $TZ;
3546        } else {
3547                delete $ENV{TZ};
3548        }
3549        if (defined $::_revision) {
3550                if ($::_revision =~ /^(\d+):(\d+)$/) {
3551                        ($r_min, $r_max) = ($1, $2);
3552                } elsif ($::_revision =~ /^\d+$/) {
3553                        $r_min = $r_max = $::_revision;
3554                } else {
3555                        ::fatal "-r$::_revision is not supported, use ",
3556                                "standard \'git log\' arguments instead\n";
3557                }
3558        }
3559
3560        config_pager();
3561        @args = (git_svn_log_cmd($r_min, $r_max, @args), @args);
3562        my $log = command_output_pipe(@args);
3563        run_pager();
3564        my (@k, $c, $d, $stat);
3565        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3566        while (<$log>) {
3567                if (/^${esc_color}commit ($::sha1_short)/o) {
3568                        my $cmt = $1;
3569                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3570                                $r_last = $c->{r};
3571                                process_commit($c, $r_min, $r_max, \@k) or
3572                                                                goto out;
3573                        }
3574                        $d = undef;
3575                        $c = { c => $cmt };
3576                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3577                        get_author_info($c, $1, $2, $3);
3578                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3579                        # ignore
3580                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3581                        push @{$c->{raw}}, $_;
3582                } elsif (/^${esc_color}[ACRMDT]\t/) {
3583                        # we could add $SVN->{svn_path} here, but that requires
3584                        # remote access at the moment (repo_path_split)...
3585                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
3586                        push @{$c->{changed}}, $_;
3587                } elsif (/^${esc_color}diff /o) {
3588                        $d = 1;
3589                        push @{$c->{diff}}, $_;
3590                } elsif ($d) {
3591                        push @{$c->{diff}}, $_;
3592                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
3593                          $esc_color*[\+\-]*$esc_color$/x) {
3594                        $stat = 1;
3595                        push @{$c->{stat}}, $_;
3596                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
3597                        push @{$c->{stat}}, $_;
3598                        $stat = undef;
3599                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
3600                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3601                } elsif (s/^${esc_color}    //o) {
3602                        push @{$c->{l}}, $_;
3603                }
3604        }
3605        if ($c && defined $c->{r} && $c->{r} != $r_last) {
3606                $r_last = $c->{r};
3607                process_commit($c, $r_min, $r_max, \@k);
3608        }
3609        if (@k) {
3610                my $swap = $r_max;
3611                $r_max = $r_min;
3612                $r_min = $swap;
3613                process_commit($_, $r_min, $r_max) foreach reverse @k;
3614        }
3615out:
3616        close $log;
3617        print '-' x72,"\n" unless $incremental || $oneline;
3618}
3619
3620package Git::SVN::Migration;
3621# these version numbers do NOT correspond to actual version numbers
3622# of git nor git-svn.  They are just relative.
3623#
3624# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3625#
3626# v1 layout: .git/$id/info/url, refs/remotes/$id
3627#
3628# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3629#
3630# v3 layout: .git/svn/$id, refs/remotes/$id
3631#            - info/url may remain for backwards compatibility
3632#            - this is what we migrate up to this layout automatically,
3633#            - this will be used by git svn init on single branches
3634# v3.1 layout (auto migrated):
3635#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
3636#              for backwards compatibility
3637#
3638# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3639#            - this is only created for newly multi-init-ed
3640#              repositories.  Similar in spirit to the
3641#              --use-separate-remotes option in git-clone (now default)
3642#            - we do not automatically migrate to this (following
3643#              the example set by core git)
3644use strict;
3645use warnings;
3646use Carp qw/croak/;
3647use File::Path qw/mkpath/;
3648use File::Basename qw/dirname basename/;
3649use vars qw/$_minimize/;
3650
3651sub migrate_from_v0 {
3652        my $git_dir = $ENV{GIT_DIR};
3653        return undef unless -d $git_dir;
3654        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3655        my $migrated = 0;
3656        while (<$fh>) {
3657                chomp;
3658                my ($id, $orig_ref) = ($_, $_);
3659                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3660                next unless -f "$git_dir/$id/info/url";
3661                my $new_ref = "refs/remotes/$id";
3662                if (::verify_ref("$new_ref^0")) {
3663                        print STDERR "W: $orig_ref is probably an old ",
3664                                     "branch used by an ancient version of ",
3665                                     "git-svn.\n",
3666                                     "However, $new_ref also exists.\n",
3667                                     "We will not be able ",
3668                                     "to use this branch until this ",
3669                                     "ambiguity is resolved.\n";
3670                        next;
3671                }
3672                print STDERR "Migrating from v0 layout...\n" if !$migrated;
3673                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3674                command_noisy('update-ref', $new_ref, $orig_ref);
3675                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3676                $migrated++;
3677        }
3678        command_close_pipe($fh, $ctx);
3679        print STDERR "Done migrating from v0 layout...\n" if $migrated;
3680        $migrated;
3681}
3682
3683sub migrate_from_v1 {
3684        my $git_dir = $ENV{GIT_DIR};
3685        my $migrated = 0;
3686        return $migrated unless -d $git_dir;
3687        my $svn_dir = "$git_dir/svn";
3688
3689        # just in case somebody used 'svn' as their $id at some point...
3690        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3691
3692        print STDERR "Migrating from a git-svn v1 layout...\n";
3693        mkpath([$svn_dir]);
3694        print STDERR "Data from a previous version of git-svn exists, but\n\t",
3695                     "$svn_dir\n\t(required for this version ",
3696                     "($::VERSION) of git-svn) does not. exist\n";
3697        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3698        while (<$fh>) {
3699                my $x = $_;
3700                next unless $x =~ s#^refs/remotes/##;
3701                chomp $x;
3702                next unless -f "$git_dir/$x/info/url";
3703                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3704                next unless $u;
3705                my $dn = dirname("$git_dir/svn/$x");
3706                mkpath([$dn]) unless -d $dn;
3707                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3708                        mkpath(["$git_dir/svn/svn"]);
3709                        print STDERR " - $git_dir/$x/info => ",
3710                                        "$git_dir/svn/$x/info\n";
3711                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3712                               croak "$!: $x";
3713                        # don't worry too much about these, they probably
3714                        # don't exist with repos this old (save for index,
3715                        # and we can easily regenerate that)
3716                        foreach my $f (qw/unhandled.log index .rev_db/) {
3717                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3718                        }
3719                } else {
3720                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3721                        rename "$git_dir/$x", "$git_dir/svn/$x" or
3722                               croak "$!: $x";
3723                }
3724                $migrated++;
3725        }
3726        command_close_pipe($fh, $ctx);
3727        print STDERR "Done migrating from a git-svn v1 layout\n";
3728        $migrated;
3729}
3730
3731sub read_old_urls {
3732        my ($l_map, $pfx, $path) = @_;
3733        my @dir;
3734        foreach (<$path/*>) {
3735                if (-r "$_/info/url") {
3736                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3737                        my $ref_id = $pfx . basename $_;
3738                        my $url = ::file_to_s("$_/info/url");
3739                        $l_map->{$ref_id} = $url;
3740                } elsif (-d $_) {
3741                        push @dir, $_;
3742                }
3743        }
3744        foreach (@dir) {
3745                my $x = $_;
3746                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3747                read_old_urls($l_map, $x, $_);
3748        }
3749}
3750
3751sub migrate_from_v2 {
3752        my @cfg = command(qw/config -l/);
3753        return if grep /^svn-remote\..+\.url=/, @cfg;
3754        my %l_map;
3755        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3756        my $migrated = 0;
3757
3758        foreach my $ref_id (sort keys %l_map) {
3759                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3760                if ($@) {
3761                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3762                }
3763                $migrated++;
3764        }
3765        $migrated;
3766}
3767
3768sub minimize_connections {
3769        my $r = Git::SVN::read_all_remotes();
3770        my $new_urls = {};
3771        my $root_repos = {};
3772        foreach my $repo_id (keys %$r) {
3773                my $url = $r->{$repo_id}->{url} or next;
3774                my $fetch = $r->{$repo_id}->{fetch} or next;
3775                my $ra = Git::SVN::Ra->new($url);
3776
3777                # skip existing cases where we already connect to the root
3778                if (($ra->{url} eq $ra->{repos_root}) ||
3779                    (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3780                     $repo_id)) {
3781                        $root_repos->{$ra->{url}} = $repo_id;
3782                        next;
3783                }
3784
3785                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3786                my $root_path = $ra->{url};
3787                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3788                foreach my $path (keys %$fetch) {
3789                        my $ref_id = $fetch->{$path};
3790                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3791
3792                        # make sure we can read when connecting to
3793                        # a higher level of a repository
3794                        my ($last_rev, undef) = $gs->last_rev_commit;
3795                        if (!defined $last_rev) {
3796                                $last_rev = eval {
3797                                        $root_ra->get_latest_revnum;
3798                                };
3799                                next if $@;
3800                        }
3801                        my $new = $root_path;
3802                        $new .= length $path ? "/$path" : '';
3803                        eval {
3804                                $root_ra->get_log([$new], $last_rev, $last_rev,
3805                                                  0, 0, 1, sub { });
3806                        };
3807                        next if $@;
3808                        $new_urls->{$ra->{repos_root}}->{$new} =
3809                                { ref_id => $ref_id,
3810                                  old_repo_id => $repo_id,
3811                                  old_path => $path };
3812                }
3813        }
3814
3815        my @emptied;
3816        foreach my $url (keys %$new_urls) {
3817                # see if we can re-use an existing [svn-remote "repo_id"]
3818                # instead of creating a(n ugly) new section:
3819                my $repo_id = $root_repos->{$url} ||
3820                              Git::SVN::sanitize_remote_name($url);
3821
3822                my $fetch = $new_urls->{$url};
3823                foreach my $path (keys %$fetch) {
3824                        my $x = $fetch->{$path};
3825                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3826                        my $pfx = "svn-remote.$x->{old_repo_id}";
3827
3828                        my $old_fetch = quotemeta("$x->{old_path}:".
3829                                                  "refs/remotes/$x->{ref_id}");
3830                        command_noisy(qw/config --unset/,
3831                                      "$pfx.fetch", '^'. $old_fetch . '$');
3832                        delete $r->{$x->{old_repo_id}}->
3833                               {fetch}->{$x->{old_path}};
3834                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3835                                command_noisy(qw/config --unset/,
3836                                              "$pfx.url");
3837                                push @emptied, $x->{old_repo_id}
3838                        }
3839                }
3840        }
3841        if (@emptied) {
3842                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3843                           "$ENV{GIT_DIR}/config";
3844                print STDERR <<EOF;
3845The following [svn-remote] sections in your config file ($file) are empty
3846and can be safely removed:
3847EOF
3848                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3849        }
3850}
3851
3852sub migration_check {
3853        migrate_from_v0();
3854        migrate_from_v1();
3855        migrate_from_v2();
3856        minimize_connections() if $_minimize;
3857}
3858
3859package Git::IndexInfo;
3860use strict;
3861use warnings;
3862use Git qw/command_input_pipe command_close_pipe/;
3863
3864sub new {
3865        my ($class) = @_;
3866        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3867        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3868}
3869
3870sub remove {
3871        my ($self, $path) = @_;
3872        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3873                return ++$self->{nr};
3874        }
3875        undef;
3876}
3877
3878sub update {
3879        my ($self, $mode, $hash, $path) = @_;
3880        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3881                return ++$self->{nr};
3882        }
3883        undef;
3884}
3885
3886sub DESTROY {
3887        my ($self) = @_;
3888        command_close_pipe($self->{gui}, $self->{ctx});
3889}
3890
3891package Git::SVN::GlobSpec;
3892use strict;
3893use warnings;
3894
3895sub new {
3896        my ($class, $glob) = @_;
3897        my $re = $glob;
3898        $re =~ s!/+$!!g; # no need for trailing slashes
3899        my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
3900        my ($left, $right) = ($1, $2);
3901        if ($nr > 1) {
3902                die "Only one '*' wildcard expansion ",
3903                    "is supported (got $nr): '$glob'\n";
3904        } elsif ($nr == 0) {
3905                die "One '*' is needed for glob: '$glob'\n";
3906        }
3907        $re = quotemeta($left) . $re . quotemeta($right);
3908        if (length $left && !($left =~ s!/+$!!g)) {
3909                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
3910        }
3911        if (length $right && !($right =~ s!^/+!!g)) {
3912                die "Missing leading '/' on right side of: '$glob' ($right)\n";
3913        }
3914        my $left_re = qr/^\/\Q$left\E(\/|$)/;
3915        bless { left => $left, right => $right, left_regex => $left_re,
3916                regex => qr/$re/, glob => $glob }, $class;
3917}
3918
3919sub full_path {
3920        my ($self, $path) = @_;
3921        return (length $self->{left} ? "$self->{left}/" : '') .
3922               $path . (length $self->{right} ? "/$self->{right}" : '');
3923}
3924
3925__END__
3926
3927Data structures:
3928
3929
3930$remotes = { # returned by read_all_remotes()
3931        'svn' => {
3932                # svn-remote.svn.url=https://svn.musicpd.org
3933                url => 'https://svn.musicpd.org',
3934                # svn-remote.svn.fetch=mpd/trunk:trunk
3935                fetch => {
3936                        'mpd/trunk' => 'trunk',
3937                },
3938                # svn-remote.svn.tags=mpd/tags/*:tags/*
3939                tags => {
3940                        path => {
3941                                left => 'mpd/tags',
3942                                right => '',
3943                                regex => qr!mpd/tags/([^/]+)$!,
3944                                glob => 'tags/*',
3945                        },
3946                        ref => {
3947                                left => 'tags',
3948                                right => '',
3949                                regex => qr!tags/([^/]+)$!,
3950                                glob => 'tags/*',
3951                        },
3952                }
3953        }
3954};
3955
3956$log_entry hashref as returned by libsvn_log_entry()
3957{
3958        log => 'whitespace-formatted log entry
3959',                                              # trailing newline is preserved
3960        revision => '8',                        # integer
3961        date => '2004-02-24T17:01:44.108345Z',  # commit date
3962        author => 'committer name'
3963};
3964
3965
3966# this is generated by generate_diff();
3967@mods = array of diff-index line hashes, each element represents one line
3968        of diff-index output
3969
3970diff-index line ($m hash)
3971{
3972        mode_a => first column of diff-index output, no leading ':',
3973        mode_b => second column of diff-index output,
3974        sha1_b => sha1sum of the final blob,
3975        chg => change type [MCRADT],
3976        file_a => original file name of a file (iff chg is 'C' or 'R')
3977        file_b => new/current file name of a file (any chg)
3978}
3979;
3980
3981# retval of read_url_paths{,_all}();
3982$l_map = {
3983        # repository root url
3984        'https://svn.musicpd.org' => {
3985                # repository path               # GIT_SVN_ID
3986                'mpd/trunk'             =>      'trunk',
3987                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3988        },
3989}
3990
3991Notes:
3992        I don't trust the each() function on unless I created %hash myself
3993        because the internal iterator may not have started at base.