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