Andrew's git - gitweb.git/blob

a6d98f1608f7a98d3368697717df380f2250686c
   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,
  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                          'fetch-all|all' => \$_fetch_all,
 149                          %fc_opts } ],
 150        'commit-diff' => [ \&cmd_commit_diff,
 151                           'Commit a diff between two trees',
 152                        { 'message|m=s' => \$_message,
 153                          'file|F=s' => \$_file,
 154                          'revision|r=s' => \$_revision,
 155                        %cmt_opts } ],
 156);
 157
 158my $cmd;
 159for (my $i = 0; $i < @ARGV; $i++) {
 160        if (defined $cmd{$ARGV[$i]}) {
 161                $cmd = $ARGV[$i];
 162                splice @ARGV, $i, 1;
 163                last;
 164        }
 165};
 166
 167my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 168
 169read_repo_config(\%opts);
 170my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 171                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 172                    'id|i=s' => \$Git::SVN::default_ref_id,
 173                    'svn-remote|remote|R=s' => sub {
 174                       $Git::SVN::no_reuse_existing = 1;
 175                       $Git::SVN::default_repo_id = $_[1] });
 176exit 1 if (!$rv && $cmd ne 'log');
 177
 178usage(0) if $_help;
 179version() if $_version;
 180usage(1) unless defined $cmd;
 181load_authors() if $_authors;
 182
 183# make sure we're always running
 184unless ($cmd =~ /(?:clone|init|multi-init)$/) {
 185        unless (-d $ENV{GIT_DIR}) {
 186                if ($git_dir_user_set) {
 187                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 188                            "but it is not a directory\n";
 189                }
 190                my $git_dir = delete $ENV{GIT_DIR};
 191                chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
 192                unless (length $cdup) {
 193                        die "Already at toplevel, but $git_dir ",
 194                            "not found '$cdup'\n";
 195                }
 196                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 197                unless (-d $git_dir) {
 198                        die "$git_dir still not found after going to ",
 199                            "'$cdup'\n";
 200                }
 201                $ENV{GIT_DIR} = $git_dir;
 202        }
 203}
 204unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 205        Git::SVN::Migration::migration_check();
 206}
 207Git::SVN::init_vars();
 208eval {
 209        Git::SVN::verify_remotes_sanity();
 210        $cmd{$cmd}->[0]->(@ARGV);
 211};
 212fatal $@ if $@;
 213post_fetch_checkout();
 214exit 0;
 215
 216####################### primary functions ######################
 217sub usage {
 218        my $exit = shift || 0;
 219        my $fd = $exit ? \*STDERR : \*STDOUT;
 220        print $fd <<"";
 221git-svn - bidirectional operations between a single Subversion tree and git
 222Usage: $0 <command> [options] [arguments]\n
 223
 224        print $fd "Available commands:\n" unless $cmd;
 225
 226        foreach (sort keys %cmd) {
 227                next if $cmd && $cmd ne $_;
 228                next if /^multi-/; # don't show deprecated commands
 229                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 230                foreach (keys %{$cmd{$_}->[2]}) {
 231                        # prints out arguments as they should be passed:
 232                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 233                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 234                                                        "--$_" : "-$_" }
 235                                                split /\|/,$_)," $x\n";
 236                }
 237        }
 238        print $fd <<"";
 239\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 240arbitrary identifier if you're tracking multiple SVN branches/repositories in
 241one git repository and want to keep them separate.  See git-svn(1) for more
 242information.
 243
 244        exit $exit;
 245}
 246
 247sub version {
 248        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 249        exit 0;
 250}
 251
 252sub do_git_init_db {
 253        unless (-d $ENV{GIT_DIR}) {
 254                my @init_db = ('init');
 255                push @init_db, "--template=$_template" if defined $_template;
 256                if (defined $_shared) {
 257                        if ($_shared =~ /[a-z]/) {
 258                                push @init_db, "--shared=$_shared";
 259                        } else {
 260                                push @init_db, "--shared";
 261                        }
 262                }
 263                command_noisy(@init_db);
 264        }
 265        my $set;
 266        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 267        foreach my $i (keys %icv) {
 268                die "'$set' and '$i' cannot both be set\n" if $set;
 269                next unless defined $icv{$i};
 270                command_noisy('config', "$pfx.$i", $icv{$i});
 271                $set = $i;
 272        }
 273}
 274
 275sub init_subdir {
 276        my $repo_path = shift or return;
 277        mkpath([$repo_path]) unless -d $repo_path;
 278        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 279        $ENV{GIT_DIR} = $repo_path . "/.git";
 280}
 281
 282sub cmd_clone {
 283        my ($url, $path) = @_;
 284        if (!defined $path &&
 285            (defined $_trunk || defined $_branches || defined $_tags) &&
 286            $url !~ m#^[a-z\+]+://#) {
 287                $path = $url;
 288        }
 289        warn "--path: $path\n" if defined $path;
 290        $path = basename($url) if !defined $path || !length $path;
 291        warn "++path: $path\n" if defined $path;
 292        mkpath([$path]);
 293        chdir $path or die "Couldn't chdir to $path\n";
 294        cmd_init(@_);
 295        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 296}
 297
 298sub cmd_init {
 299        if (defined $_trunk || defined $_branches || defined $_tags) {
 300                return cmd_multi_init(@_);
 301        }
 302        my $url = shift or die "SVN repository location required ",
 303                               "as a command-line argument\n";
 304        init_subdir(@_);
 305        do_git_init_db();
 306
 307        Git::SVN->init($url);
 308}
 309
 310sub cmd_fetch {
 311        if (grep /^\d+=./, @_) {
 312                die "'<rev>=<commit>' fetch arguments are ",
 313                    "no longer supported.\n";
 314        }
 315        my ($remote) = @_;
 316        if (@_ > 1) {
 317                die "Usage: $0 fetch [--all] [svn-remote]\n";
 318        }
 319        $remote ||= $Git::SVN::default_repo_id;
 320        if ($_fetch_all) {
 321                cmd_multi_fetch();
 322        } else {
 323                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 324        }
 325}
 326
 327sub cmd_set_tree {
 328        my (@commits) = @_;
 329        if ($_stdin || !@commits) {
 330                print "Reading from stdin...\n";
 331                @commits = ();
 332                while (<STDIN>) {
 333                        if (/\b($sha1_short)\b/o) {
 334                                unshift @commits, $1;
 335                        }
 336                }
 337        }
 338        my @revs;
 339        foreach my $c (@commits) {
 340                my @tmp = command('rev-parse',$c);
 341                if (scalar @tmp == 1) {
 342                        push @revs, $tmp[0];
 343                } elsif (scalar @tmp > 1) {
 344                        push @revs, reverse(command('rev-list',@tmp));
 345                } else {
 346                        fatal "Failed to rev-parse $c\n";
 347                }
 348        }
 349        my $gs = Git::SVN->new;
 350        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 351        $gs->fetch;
 352        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 353                fatal "There are new revisions that were fetched ",
 354                      "and need to be merged (or acknowledged) ",
 355                      "before committing.\nlast rev: $r_last\n",
 356                      " current: $gs->{last_rev}\n";
 357        }
 358        $gs->set_tree($_) foreach @revs;
 359        print "Done committing ",scalar @revs," revisions to SVN\n";
 360}
 361
 362sub cmd_dcommit {
 363        my $head = shift;
 364        $head ||= 'HEAD';
 365        my @refs;
 366        my ($url, $rev, $uuid) = working_head_info($head, \@refs);
 367        my $c = $refs[-1];
 368        unless (defined $url && defined $rev && defined $uuid) {
 369                die "Unable to determine upstream SVN information from ",
 370                    "$head history\n";
 371        }
 372        my $gs = Git::SVN->find_by_url($url);
 373        my $last_rev;
 374        foreach my $d (@refs) {
 375                if (!verify_ref("$d~1")) {
 376                        fatal "Commit $d\n",
 377                              "has no parent commit, and therefore ",
 378                              "nothing to diff against.\n",
 379                              "You should be working from a repository ",
 380                              "originally created by git-svn\n";
 381                }
 382                unless (defined $last_rev) {
 383                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 384                        unless (defined $last_rev) {
 385                                fatal "Unable to extract revision information ",
 386                                      "from commit $d~1\n";
 387                        }
 388                }
 389                if ($_dry_run) {
 390                        print "diff-tree $d~1 $d\n";
 391                } else {
 392                        my %ed_opts = ( r => $last_rev,
 393                                        log => get_commit_entry($d)->{log},
 394                                        ra => Git::SVN::Ra->new($url),
 395                                        tree_a => "$d~1",
 396                                        tree_b => $d,
 397                                        editor_cb => sub {
 398                                               print "Committed r$_[0]\n";
 399                                               $last_rev = $_[0]; },
 400                                        svn_path => '');
 401                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 402                                print "No changes\n$d~1 == $d\n";
 403                        }
 404                }
 405        }
 406        return if $_dry_run;
 407        unless ($gs) {
 408                warn "Could not determine fetch information for $url\n",
 409                     "Will not attempt to fetch and rebase commits.\n",
 410                     "This probably means you have useSvmProps and should\n",
 411                     "now resync your SVN::Mirror repository.\n";
 412                return;
 413        }
 414        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 415        # we always want to rebase against the current HEAD, not any
 416        # head that was passed to us
 417        my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
 418        my @finish;
 419        if (@diff) {
 420                @finish = rebase_cmd();
 421                print STDERR "W: HEAD and ", $gs->refname, " differ, ",
 422                             "using @finish:\n", "@diff";
 423        } else {
 424                print "No changes between current HEAD and ",
 425                      $gs->refname, "\nResetting to the latest ",
 426                      $gs->refname, "\n";
 427                @finish = qw/reset --mixed/;
 428        }
 429        command_noisy(@finish, $gs->refname);
 430}
 431
 432sub cmd_rebase {
 433        command_noisy(qw/update-index --refresh/);
 434        my $url = (working_head_info('HEAD'))[0];
 435        if (!defined $url) {
 436                die "Unable to determine upstream SVN information from ",
 437                    "working tree history\n";
 438        }
 439
 440        my $gs = Git::SVN->find_by_url($url);
 441        if (command(qw/diff-index HEAD --/)) {
 442                print STDERR "Cannot rebase with uncommited changes:\n";
 443                command_noisy('status');
 444                exit 1;
 445        }
 446        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 447        command_noisy(rebase_cmd(), $gs->refname);
 448}
 449
 450sub cmd_show_ignore {
 451        my $url = (::working_head_info('HEAD'))[0];
 452        my $gs = Git::SVN->find_by_url($url) || Git::SVN->new;
 453        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 454        $gs->traverse_ignore(\*STDOUT, '', $r);
 455}
 456
 457sub cmd_multi_init {
 458        my $url = shift;
 459        unless (defined $_trunk || defined $_branches || defined $_tags) {
 460                usage(1);
 461        }
 462        do_git_init_db();
 463        $_prefix = '' unless defined $_prefix;
 464        if (defined $url) {
 465                $url =~ s#/+$##;
 466                init_subdir(@_);
 467        }
 468        if (defined $_trunk) {
 469                my $trunk_ref = $_prefix . 'trunk';
 470                # try both old-style and new-style lookups:
 471                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 472                unless ($gs_trunk) {
 473                        my ($trunk_url, $trunk_path) =
 474                                              complete_svn_url($url, $_trunk);
 475                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 476                                                   undef, $trunk_ref);
 477                }
 478        }
 479        return unless defined $_branches || defined $_tags;
 480        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 481        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 482        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 483}
 484
 485sub cmd_multi_fetch {
 486        my $remotes = Git::SVN::read_all_remotes();
 487        foreach my $repo_id (sort keys %$remotes) {
 488                if ($remotes->{$repo_id}->{url}) {
 489                        Git::SVN::fetch_all($repo_id, $remotes);
 490                }
 491        }
 492}
 493
 494# this command is special because it requires no metadata
 495sub cmd_commit_diff {
 496        my ($ta, $tb, $url) = @_;
 497        my $usage = "Usage: $0 commit-diff -r<revision> ".
 498                    "<tree-ish> <tree-ish> [<URL>]\n";
 499        fatal($usage) if (!defined $ta || !defined $tb);
 500        my $svn_path;
 501        if (!defined $url) {
 502                my $gs = eval { Git::SVN->new };
 503                if (!$gs) {
 504                        fatal("Needed URL or usable git-svn --id in ",
 505                              "the command-line\n", $usage);
 506                }
 507                $url = $gs->{url};
 508                $svn_path = $gs->{path};
 509        }
 510        unless (defined $_revision) {
 511                fatal("-r|--revision is a required argument\n", $usage);
 512        }
 513        if (defined $_message && defined $_file) {
 514                fatal("Both --message/-m and --file/-F specified ",
 515                      "for the commit message.\n",
 516                      "I have no idea what you mean\n");
 517        }
 518        if (defined $_file) {
 519                $_message = file_to_s($_file);
 520        } else {
 521                $_message ||= get_commit_entry($tb)->{log};
 522        }
 523        my $ra ||= Git::SVN::Ra->new($url);
 524        $svn_path ||= $ra->{svn_path};
 525        my $r = $_revision;
 526        if ($r eq 'HEAD') {
 527                $r = $ra->get_latest_revnum;
 528        } elsif ($r !~ /^\d+$/) {
 529                die "revision argument: $r not understood by git-svn\n";
 530        }
 531        my %ed_opts = ( r => $r,
 532                        log => $_message,
 533                        ra => $ra,
 534                        tree_a => $ta,
 535                        tree_b => $tb,
 536                        editor_cb => sub { print "Committed r$_[0]\n" },
 537                        svn_path => $svn_path );
 538        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 539                print "No changes\n$ta == $tb\n";
 540        }
 541}
 542
 543########################### utility functions #########################
 544
 545sub rebase_cmd {
 546        my @cmd = qw/rebase/;
 547        push @cmd, '-v' if $_verbose;
 548        push @cmd, qw/--merge/ if $_merge;
 549        push @cmd, "--strategy=$_strategy" if $_strategy;
 550        @cmd;
 551}
 552
 553sub post_fetch_checkout {
 554        return if $_no_checkout;
 555        my $gs = $Git::SVN::_head or return;
 556        return if verify_ref('refs/heads/master^0');
 557
 558        my $valid_head = verify_ref('HEAD^0');
 559        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
 560        return if ($valid_head || !verify_ref('HEAD^0'));
 561
 562        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
 563        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
 564        return if -f $index;
 565
 566        chomp(my $bare = `git config --bool --get core.bare`);
 567        return if $bare eq 'true';
 568        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
 569        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
 570        print STDERR "Checked out HEAD:\n  ",
 571                     $gs->full_url, " r", $gs->last_rev, "\n";
 572}
 573
 574sub complete_svn_url {
 575        my ($url, $path) = @_;
 576        $path =~ s#/+$##;
 577        if ($path !~ m#^[a-z\+]+://#) {
 578                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 579                        fatal("E: '$path' is not a complete URL ",
 580                              "and a separate URL is not specified\n");
 581                }
 582                return ($url, $path);
 583        }
 584        return ($path, '');
 585}
 586
 587sub complete_url_ls_init {
 588        my ($ra, $repo_path, $switch, $pfx) = @_;
 589        unless ($repo_path) {
 590                print STDERR "W: $switch not specified\n";
 591                return;
 592        }
 593        $repo_path =~ s#/+$##;
 594        if ($repo_path =~ m#^[a-z\+]+://#) {
 595                $ra = Git::SVN::Ra->new($repo_path);
 596                $repo_path = '';
 597        } else {
 598                $repo_path =~ s#^/+##;
 599                unless ($ra) {
 600                        fatal("E: '$repo_path' is not a complete URL ",
 601                              "and a separate URL is not specified\n");
 602                }
 603        }
 604        my $url = $ra->{url};
 605        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
 606        my $k = "svn-remote.$gs->{repo_id}.url";
 607        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
 608        if ($orig_url && ($orig_url ne $gs->{url})) {
 609                die "$k already set: $orig_url\n",
 610                    "wanted to set to: $gs->{url}\n";
 611        }
 612        command_oneline('config', $k, $gs->{url}) unless $orig_url;
 613        my $remote_path = "$ra->{svn_path}/$repo_path/*";
 614        $remote_path =~ s#/+#/#g;
 615        $remote_path =~ s#^/##g;
 616        my ($n) = ($switch =~ /^--(\w+)/);
 617        if (length $pfx && $pfx !~ m#/$#) {
 618                die "--prefix='$pfx' must have a trailing slash '/'\n";
 619        }
 620        command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
 621                                "$remote_path:refs/remotes/$pfx*");
 622}
 623
 624sub verify_ref {
 625        my ($ref) = @_;
 626        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 627                               { STDERR => 0 }); };
 628}
 629
 630sub get_tree_from_treeish {
 631        my ($treeish) = @_;
 632        # $treeish can be a symbolic ref, too:
 633        my $type = command_oneline(qw/cat-file -t/, $treeish);
 634        my $expected;
 635        while ($type eq 'tag') {
 636                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
 637        }
 638        if ($type eq 'commit') {
 639                $expected = (grep /^tree /, command(qw/cat-file commit/,
 640                                                    $treeish))[0];
 641                ($expected) = ($expected =~ /^tree ($sha1)$/o);
 642                die "Unable to get tree from $treeish\n" unless $expected;
 643        } elsif ($type eq 'tree') {
 644                $expected = $treeish;
 645        } else {
 646                die "$treeish is a $type, expected tree, tag or commit\n";
 647        }
 648        return $expected;
 649}
 650
 651sub get_commit_entry {
 652        my ($treeish) = shift;
 653        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
 654        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
 655        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
 656        open my $log_fh, '>', $commit_editmsg or croak $!;
 657
 658        my $type = command_oneline(qw/cat-file -t/, $treeish);
 659        if ($type eq 'commit' || $type eq 'tag') {
 660                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
 661                                                         $type, $treeish);
 662                my $in_msg = 0;
 663                while (<$msg_fh>) {
 664                        if (!$in_msg) {
 665                                $in_msg = 1 if (/^\s*$/);
 666                        } elsif (/^git-svn-id: /) {
 667                                # skip this for now, we regenerate the
 668                                # correct one on re-fetch anyways
 669                                # TODO: set *:merge properties or like...
 670                        } else {
 671                                print $log_fh $_ or croak $!;
 672                        }
 673                }
 674                command_close_pipe($msg_fh, $ctx);
 675        }
 676        close $log_fh or croak $!;
 677
 678        if ($_edit || ($type eq 'tree')) {
 679                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 680                # TODO: strip out spaces, comments, like git-commit.sh
 681                system($editor, $commit_editmsg);
 682        }
 683        rename $commit_editmsg, $commit_msg or croak $!;
 684        open $log_fh, '<', $commit_msg or croak $!;
 685        { local $/; chomp($log_entry{log} = <$log_fh>); }
 686        close $log_fh or croak $!;
 687        unlink $commit_msg;
 688        \%log_entry;
 689}
 690
 691sub s_to_file {
 692        my ($str, $file, $mode) = @_;
 693        open my $fd,'>',$file or croak $!;
 694        print $fd $str,"\n" or croak $!;
 695        close $fd or croak $!;
 696        chmod ($mode &~ umask, $file) if (defined $mode);
 697}
 698
 699sub file_to_s {
 700        my $file = shift;
 701        open my $fd,'<',$file or croak "$!: file: $file\n";
 702        local $/;
 703        my $ret = <$fd>;
 704        close $fd or croak $!;
 705        $ret =~ s/\s*$//s;
 706        return $ret;
 707}
 708
 709# '<svn username> = real-name <email address>' mapping based on git-svnimport:
 710sub load_authors {
 711        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
 712        my $log = $cmd eq 'log';
 713        while (<$authors>) {
 714                chomp;
 715                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
 716                my ($user, $name, $email) = ($1, $2, $3);
 717                if ($log) {
 718                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
 719                } else {
 720                        $users{$user} = [$name, $email];
 721                }
 722        }
 723        close $authors or croak $!;
 724}
 725
 726# convert GetOpt::Long specs for use by git-config
 727sub read_repo_config {
 728        return unless -d $ENV{GIT_DIR};
 729        my $opts = shift;
 730        my @config_only;
 731        foreach my $o (keys %$opts) {
 732                # if we have mixedCase and a long option-only, then
 733                # it's a config-only variable that we don't need for
 734                # the command-line.
 735                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
 736                my $v = $opts->{$o};
 737                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
 738                $key =~ s/-//g;
 739                my $arg = 'git-config';
 740                $arg .= ' --int' if ($o =~ /[:=]i$/);
 741                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
 742                if (ref $v eq 'ARRAY') {
 743                        chomp(my @tmp = `$arg --get-all svn.$key`);
 744                        @$v = @tmp if @tmp;
 745                } else {
 746                        chomp(my $tmp = `$arg --get svn.$key`);
 747                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
 748                                $$v = $tmp;
 749                        }
 750                }
 751        }
 752        delete @$opts{@config_only} if @config_only;
 753}
 754
 755sub extract_metadata {
 756        my $id = shift or return (undef, undef, undef);
 757        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 758                                                        \s([a-f\d\-]+)$/x);
 759        if (!defined $rev || !$uuid || !$url) {
 760                # some of the original repositories I made had
 761                # identifiers like this:
 762                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
 763        }
 764        return ($url, $rev, $uuid);
 765}
 766
 767sub cmt_metadata {
 768        return extract_metadata((grep(/^git-svn-id: /,
 769                command(qw/cat-file commit/, shift)))[-1]);
 770}
 771
 772sub working_head_info {
 773        my ($head, $refs) = @_;
 774        my ($url, $rev, $uuid);
 775        my ($fh, $ctx) = command_output_pipe('rev-list', $head);
 776        while (<$fh>) {
 777                chomp;
 778                ($url, $rev, $uuid) = cmt_metadata($_);
 779                last if (defined $url && defined $rev && defined $uuid);
 780                unshift @$refs, $_ if $refs;
 781        }
 782        close $fh; # break the pipe
 783        ($url, $rev, $uuid);
 784}
 785
 786package Git::SVN;
 787use strict;
 788use warnings;
 789use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
 790            $_repack $_repack_flags $_use_svm_props $_head
 791            $_use_svnsync_props $no_reuse_existing/;
 792use Carp qw/croak/;
 793use File::Path qw/mkpath/;
 794use File::Copy qw/copy/;
 795use IPC::Open3;
 796
 797my $_repack_nr;
 798# properties that we do not log:
 799my %SKIP_PROP;
 800BEGIN {
 801        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
 802                                        svn:special svn:executable
 803                                        svn:entry:committed-rev
 804                                        svn:entry:last-author
 805                                        svn:entry:uuid
 806                                        svn:entry:committed-date/;
 807
 808        # some options are read globally, but can be overridden locally
 809        # per [svn-remote "..."] section.  Command-line options will *NOT*
 810        # override options set in an [svn-remote "..."] section
 811        my $e;
 812        foreach (qw/follow_parent no_metadata use_svm_props
 813                    use_svnsync_props/) {
 814                my $key = $_;
 815                $key =~ tr/_//d;
 816                $e .= "sub $_ {
 817                        my (\$self) = \@_;
 818                        return \$self->{-$_} if exists \$self->{-$_};
 819                        my \$k = \"svn-remote.\$self->{repo_id}\.$key\";
 820                        eval { command_oneline(qw/config --get/, \$k) };
 821                        if (\$@) {
 822                                \$self->{-$_} = \$Git::SVN::_$_;
 823                        } else {
 824                                my \$v = command_oneline(qw/config --bool/,\$k);
 825                                \$self->{-$_} = \$v eq 'false' ? 0 : 1;
 826                        }
 827                        return \$self->{-$_} }\n";
 828        }
 829        $e .= "1;\n";
 830        eval $e or die $@;
 831}
 832
 833my %LOCKFILES;
 834END { unlink keys %LOCKFILES if %LOCKFILES }
 835
 836sub resolve_local_globs {
 837        my ($url, $fetch, $glob_spec) = @_;
 838        return unless defined $glob_spec;
 839        my $ref = $glob_spec->{ref};
 840        my $path = $glob_spec->{path};
 841        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
 842                next unless m#^refs/remotes/$ref->{regex}$#;
 843                my $p = $1;
 844                my $pathname = $path->full_path($p);
 845                my $refname = $ref->full_path($p);
 846                if (my $existing = $fetch->{$pathname}) {
 847                        if ($existing ne $refname) {
 848                                die "Refspec conflict:\n",
 849                                    "existing: refs/remotes/$existing\n",
 850                                    " globbed: refs/remotes/$refname\n";
 851                        }
 852                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
 853                        $u =~ s!^\Q$url\E(/|$)!! or die
 854                          "refs/remotes/$refname: '$url' not found in '$u'\n";
 855                        if ($pathname ne $u) {
 856                                warn "W: Refspec glob conflict ",
 857                                     "(ref: refs/remotes/$refname):\n",
 858                                     "expected path: $pathname\n",
 859                                     "    real path: $u\n",
 860                                     "Continuing ahead with $u\n";
 861                                next;
 862                        }
 863                } else {
 864                        $fetch->{$pathname} = $refname;
 865                }
 866        }
 867}
 868
 869sub parse_revision_argument {
 870        my ($base, $head) = @_;
 871        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
 872                return ($base, $head);
 873        }
 874        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
 875        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
 876        return ($head, $head) if ($::_revision eq 'HEAD');
 877        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
 878        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
 879        die "revision argument: $::_revision not understood by git-svn\n";
 880}
 881
 882sub fetch_all {
 883        my ($repo_id, $remotes) = @_;
 884        if (ref $repo_id) {
 885                my $gs = $repo_id;
 886                $repo_id = undef;
 887                $repo_id = $gs->{repo_id};
 888        }
 889        $remotes ||= read_all_remotes();
 890        my $remote = $remotes->{$repo_id} or
 891                     die "[svn-remote \"$repo_id\"] unknown\n";
 892        my $fetch = $remote->{fetch};
 893        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
 894        my (@gs, @globs);
 895        my $ra = Git::SVN::Ra->new($url);
 896        my $uuid = $ra->get_uuid;
 897        my $head = $ra->get_latest_revnum;
 898        my $base = defined $fetch ? $head : 0;
 899
 900        # read the max revs for wildcard expansion (branches/*, tags/*)
 901        foreach my $t (qw/branches tags/) {
 902                defined $remote->{$t} or next;
 903                push @globs, $remote->{$t};
 904                my $max_rev = eval { tmp_config(qw/--int --get/,
 905                                         "svn-remote.$repo_id.${t}-maxRev") };
 906                if (defined $max_rev && ($max_rev < $base)) {
 907                        $base = $max_rev;
 908                } elsif (!defined $max_rev) {
 909                        $base = 0;
 910                }
 911        }
 912
 913        if ($fetch) {
 914                foreach my $p (sort keys %$fetch) {
 915                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
 916                        my $lr = $gs->rev_db_max;
 917                        if (defined $lr) {
 918                                $base = $lr if ($lr < $base);
 919                        }
 920                        push @gs, $gs;
 921                }
 922        }
 923
 924        ($base, $head) = parse_revision_argument($base, $head);
 925        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
 926}
 927
 928sub read_all_remotes {
 929        my $r = {};
 930        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
 931                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
 932                        $r->{$1}->{fetch}->{$2} = $3;
 933                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
 934                        $r->{$1}->{url} = $2;
 935                } elsif (m!^(.+)\.(branches|tags)=
 936                           (.*):refs/remotes/(.+)\s*$/!x) {
 937                        my ($p, $g) = ($3, $4);
 938                        my $rs = $r->{$1}->{$2} = {
 939                                          t => $2,
 940                                          remote => $1,
 941                                          path => Git::SVN::GlobSpec->new($p),
 942                                          ref => Git::SVN::GlobSpec->new($g) };
 943                        if (length($rs->{ref}->{right}) != 0) {
 944                                die "The '*' glob character must be the last ",
 945                                    "character of '$g'\n";
 946                        }
 947                }
 948        }
 949        $r;
 950}
 951
 952sub init_vars {
 953        if (defined $_repack) {
 954                $_repack = 1000 if ($_repack <= 0);
 955                $_repack_nr = $_repack;
 956                $_repack_flags ||= '-d';
 957        }
 958}
 959
 960sub verify_remotes_sanity {
 961        return unless -d $ENV{GIT_DIR};
 962        my %seen;
 963        foreach (command(qw/config -l/)) {
 964                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
 965                        if ($seen{$1}) {
 966                                die "Remote ref refs/remote/$1 is tracked by",
 967                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
 968                                    "Please resolve this ambiguity in ",
 969                                    "your git configuration file before ",
 970                                    "continuing\n";
 971                        }
 972                        $seen{$1} = $_;
 973                }
 974        }
 975}
 976
 977# we allow more chars than remotes2config.sh...
 978sub sanitize_remote_name {
 979        my ($name) = @_;
 980        $name =~ tr{A-Za-z0-9:,/+-}{.}c;
 981        $name;
 982}
 983
 984sub find_existing_remote {
 985        my ($url, $remotes) = @_;
 986        return undef if $no_reuse_existing;
 987        my $existing;
 988        foreach my $repo_id (keys %$remotes) {
 989                my $u = $remotes->{$repo_id}->{url} or next;
 990                next if $u ne $url;
 991                $existing = $repo_id;
 992                last;
 993        }
 994        $existing;
 995}
 996
 997sub init_remote_config {
 998        my ($self, $url, $no_write) = @_;
 999        $url =~ s!/+$!!; # strip trailing slash
1000        my $r = read_all_remotes();
1001        my $existing = find_existing_remote($url, $r);
1002        if ($existing) {
1003                unless ($no_write) {
1004                        print STDERR "Using existing ",
1005                                     "[svn-remote \"$existing\"]\n";
1006                }
1007                $self->{repo_id} = $existing;
1008        } else {
1009                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1010                $existing = find_existing_remote($min_url, $r);
1011                if ($existing) {
1012                        unless ($no_write) {
1013                                print STDERR "Using existing ",
1014                                             "[svn-remote \"$existing\"]\n";
1015                        }
1016                        $self->{repo_id} = $existing;
1017                }
1018                if ($min_url ne $url) {
1019                        unless ($no_write) {
1020                                print STDERR "Using higher level of URL: ",
1021                                             "$url => $min_url\n";
1022                        }
1023                        my $old_path = $self->{path};
1024                        $self->{path} = $url;
1025                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1026                        if (length $old_path) {
1027                                $self->{path} .= "/$old_path";
1028                        }
1029                        $url = $min_url;
1030                }
1031        }
1032        my $orig_url;
1033        if (!$existing) {
1034                # verify that we aren't overwriting anything:
1035                $orig_url = eval {
1036                        command_oneline('config', '--get',
1037                                        "svn-remote.$self->{repo_id}.url")
1038                };
1039                if ($orig_url && ($orig_url ne $url)) {
1040                        die "svn-remote.$self->{repo_id}.url already set: ",
1041                            "$orig_url\nwanted to set to: $url\n";
1042                }
1043        }
1044        my ($xrepo_id, $xpath) = find_ref($self->refname);
1045        if (defined $xpath) {
1046                die "svn-remote.$xrepo_id.fetch already set to track ",
1047                    "$xpath:refs/remotes/", $self->refname, "\n";
1048        }
1049        unless ($no_write) {
1050                command_noisy('config',
1051                              "svn-remote.$self->{repo_id}.url", $url);
1052                command_noisy('config', '--add',
1053                              "svn-remote.$self->{repo_id}.fetch",
1054                              "$self->{path}:".$self->refname);
1055        }
1056        $self->{url} = $url;
1057}
1058
1059sub find_by_url { # repos_root and, path are optional
1060        my ($class, $full_url, $repos_root, $path) = @_;
1061        return undef unless defined $full_url;
1062        my $remotes = read_all_remotes();
1063        if (defined $full_url && defined $repos_root && !defined $path) {
1064                $path = $full_url;
1065                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1066        }
1067        foreach my $repo_id (keys %$remotes) {
1068                my $u = $remotes->{$repo_id}->{url} or next;
1069                next if defined $repos_root && $repos_root ne $u;
1070
1071                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1072                foreach (qw/branches tags/) {
1073                        resolve_local_globs($u, $fetch,
1074                                            $remotes->{$repo_id}->{$_});
1075                }
1076                my $p = $path;
1077                unless (defined $p) {
1078                        $p = $full_url;
1079                        $p =~ s#^\Q$u\E(?:/|$)## or next;
1080                }
1081                foreach my $f (keys %$fetch) {
1082                        next if $f ne $p;
1083                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1084                }
1085        }
1086        undef;
1087}
1088
1089sub init {
1090        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1091        my $self = _new($class, $repo_id, $ref_id, $path);
1092        if (defined $url) {
1093                $self->init_remote_config($url, $no_write);
1094        }
1095        $self;
1096}
1097
1098sub find_ref {
1099        my ($ref_id) = @_;
1100        foreach (command(qw/config -l/)) {
1101                next unless m!^svn-remote\.(.+)\.fetch=
1102                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1103                my ($repo_id, $path, $ref) = ($1, $2, $3);
1104                if ($ref eq $ref_id) {
1105                        $path = '' if ($path =~ m#^\./?#);
1106                        return ($repo_id, $path);
1107                }
1108        }
1109        (undef, undef, undef);
1110}
1111
1112sub new {
1113        my ($class, $ref_id, $repo_id, $path) = @_;
1114        if (defined $ref_id && !defined $repo_id && !defined $path) {
1115                ($repo_id, $path) = find_ref($ref_id);
1116                if (!defined $repo_id) {
1117                        die "Could not find a \"svn-remote.*.fetch\" key ",
1118                            "in the repository configuration matching: ",
1119                            "refs/remotes/$ref_id\n";
1120                }
1121        }
1122        my $self = _new($class, $repo_id, $ref_id, $path);
1123        if (!defined $self->{path} || !length $self->{path}) {
1124                my $fetch = command_oneline('config', '--get',
1125                                            "svn-remote.$repo_id.fetch",
1126                                            ":refs/remotes/$ref_id\$") or
1127                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1128                         "\":refs/remotes/$ref_id\$\" in config\n";
1129                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1130        }
1131        $self->{url} = command_oneline('config', '--get',
1132                                       "svn-remote.$repo_id.url") or
1133                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1134        $self->rebuild;
1135        $self;
1136}
1137
1138sub refname { "refs/remotes/$_[0]->{ref_id}" }
1139
1140sub svm_uuid {
1141        my ($self) = @_;
1142        return $self->{svm}->{uuid} if $self->svm;
1143        $self->ra;
1144        unless ($self->{svm}) {
1145                die "SVM UUID not cached, and reading remotely failed\n";
1146        }
1147        $self->{svm}->{uuid};
1148}
1149
1150sub svm {
1151        my ($self) = @_;
1152        return $self->{svm} if $self->{svm};
1153        my $svm;
1154        # see if we have it in our config, first:
1155        eval {
1156                my $section = "svn-remote.$self->{repo_id}";
1157                $svm = {
1158                  source => tmp_config('--get', "$section.svm-source"),
1159                  uuid => tmp_config('--get', "$section.svm-uuid"),
1160                  replace => tmp_config('--get', "$section.svm-replace"),
1161                }
1162        };
1163        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1164                $self->{svm} = $svm;
1165        }
1166        $self->{svm};
1167}
1168
1169sub _set_svm_vars {
1170        my ($self, $ra) = @_;
1171        return $ra if $self->svm;
1172
1173        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1174                    "(svm:source, svm:uuid) ",
1175                    "from the following URLs:\n" );
1176        sub read_svm_props {
1177                my ($self, $ra, $path, $r) = @_;
1178                my $props = ($ra->get_dir($path, $r))[2];
1179                my $src = $props->{'svm:source'};
1180                my $uuid = $props->{'svm:uuid'};
1181                return undef if (!$src || !$uuid);
1182
1183                chomp($src, $uuid);
1184
1185                $uuid =~ m{^[0-9a-f\-]{30,}$}
1186                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1187
1188                # the '!' is used to mark the repos_root!/relative/path
1189                $src =~ s{/?!/?}{/};
1190                $src =~ s{/+$}{}; # no trailing slashes please
1191                # username is of no interest
1192                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1193
1194                my $replace = $ra->{url};
1195                $replace .= "/$path" if length $path;
1196
1197                my $section = "svn-remote.$self->{repo_id}";
1198                tmp_config("$section.svm-source", $src);
1199                tmp_config("$section.svm-replace", $replace);
1200                tmp_config("$section.svm-uuid", $uuid);
1201                $self->{svm} = {
1202                        source => $src,
1203                        uuid => $uuid,
1204                        replace => $replace
1205                };
1206        }
1207
1208        my $r = $ra->get_latest_revnum;
1209        my $path = $self->{path};
1210        my %tried;
1211        while (length $path) {
1212                unless ($tried{"$self->{url}/$path"}) {
1213                        return $ra if $self->read_svm_props($ra, $path, $r);
1214                        $tried{"$self->{url}/$path"} = 1;
1215                }
1216                $path =~ s#/?[^/]+$##;
1217        }
1218        die "Path: '$path' should be ''\n" if $path ne '';
1219        return $ra if $self->read_svm_props($ra, $path, $r);
1220        $tried{"$self->{url}/$path"} = 1;
1221
1222        if ($ra->{repos_root} eq $self->{url}) {
1223                die @err, (map { "  $_\n" } keys %tried), "\n";
1224        }
1225
1226        # nope, make sure we're connected to the repository root:
1227        my $ok;
1228        my @tried_b;
1229        $path = $ra->{svn_path};
1230        $ra = Git::SVN::Ra->new($ra->{repos_root});
1231        while (length $path) {
1232                unless ($tried{"$ra->{url}/$path"}) {
1233                        $ok = $self->read_svm_props($ra, $path, $r);
1234                        last if $ok;
1235                        $tried{"$ra->{url}/$path"} = 1;
1236                }
1237                $path =~ s#/?[^/]+$##;
1238        }
1239        die "Path: '$path' should be ''\n" if $path ne '';
1240        $ok ||= $self->read_svm_props($ra, $path, $r);
1241        $tried{"$ra->{url}/$path"} = 1;
1242        if (!$ok) {
1243                die @err, (map { "  $_\n" } keys %tried), "\n";
1244        }
1245        Git::SVN::Ra->new($self->{url});
1246}
1247
1248sub svnsync {
1249        my ($self) = @_;
1250        return $self->{svnsync} if $self->{svnsync};
1251
1252        if ($self->no_metadata) {
1253                die "Can't have both 'noMetadata' and ",
1254                    "'useSvnsyncProps' options set!\n";
1255        }
1256        if ($self->rewrite_root) {
1257                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1258                    "options set!\n";
1259        }
1260
1261        my $svnsync;
1262        # see if we have it in our config, first:
1263        eval {
1264                my $section = "svn-remote.$self->{repo_id}";
1265                $svnsync = {
1266                  url => tmp_config('--get', "$section.svnsync-url"),
1267                  uuid => tmp_config('--get', "$section.svnsync-uuid"),
1268                }
1269        };
1270        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1271                return $self->{svnsync} = $svnsync;
1272        }
1273
1274        my $err = "useSvnsyncProps set, but failed to read " .
1275                  "svnsync property: svn:sync-from-";
1276        my $rp = $self->ra->rev_proplist(0);
1277
1278        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1279        $url =~ m{^[a-z\+]+://} or
1280                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1281
1282        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1283        $uuid =~ m{^[0-9a-f\-]{30,}$} or
1284                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1285
1286        my $section = "svn-remote.$self->{repo_id}";
1287        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1288        tmp_config('--add', "$section.svnsync-url", $url);
1289        return $self->{svnsync} = { url => $url, uuid => $uuid };
1290}
1291
1292# this allows us to memoize our SVN::Ra UUID locally and avoid a
1293# remote lookup (useful for 'git svn log').
1294sub ra_uuid {
1295        my ($self) = @_;
1296        unless ($self->{ra_uuid}) {
1297                my $key = "svn-remote.$self->{repo_id}.uuid";
1298                my $uuid = eval { tmp_config('--get', $key) };
1299                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1300                        $self->{ra_uuid} = $uuid;
1301                } else {
1302                        die "ra_uuid called without URL\n" unless $self->{url};
1303                        $self->{ra_uuid} = $self->ra->get_uuid;
1304                        tmp_config('--add', $key, $self->{ra_uuid});
1305                }
1306        }
1307        $self->{ra_uuid};
1308}
1309
1310sub ra {
1311        my ($self) = shift;
1312        my $ra = Git::SVN::Ra->new($self->{url});
1313        if ($self->use_svm_props && !$self->{svm}) {
1314                if ($self->no_metadata) {
1315                        die "Can't have both 'noMetadata' and ",
1316                            "'useSvmProps' options set!\n";
1317                } elsif ($self->use_svnsync_props) {
1318                        die "Can't have both 'useSvnsyncProps' and ",
1319                            "'useSvmProps' options set!\n";
1320                }
1321                $ra = $self->_set_svm_vars($ra);
1322                $self->{-want_revprops} = 1;
1323        }
1324        $ra;
1325}
1326
1327sub rel_path {
1328        my ($self) = @_;
1329        my $repos_root = $self->ra->{repos_root};
1330        return $self->{path} if ($self->{url} eq $repos_root);
1331        die "BUG: rel_path failed! repos_root: $repos_root, Ra URL: ",
1332            $self->ra->{url}, " path: $self->{path},  URL: $self->{url}\n";
1333}
1334
1335sub traverse_ignore {
1336        my ($self, $fh, $path, $r) = @_;
1337        $path =~ s#^/+##g;
1338        my $ra = $self->ra;
1339        my ($dirent, undef, $props) = $ra->get_dir($path, $r);
1340        my $p = $path;
1341        $p =~ s#^\Q$ra->{svn_path}\E/##;
1342        print $fh length $p ? "\n# $p\n" : "\n# /\n";
1343        if (my $s = $props->{'svn:ignore'}) {
1344                $s =~ s/[\r\n]+/\n/g;
1345                chomp $s;
1346                if (length $p == 0) {
1347                        $s =~ s#\n#\n/$p#g;
1348                        print $fh "/$s\n";
1349                } else {
1350                        $s =~ s#\n#\n/$p/#g;
1351                        print $fh "/$p/$s\n";
1352                }
1353        }
1354        foreach (sort keys %$dirent) {
1355                next if $dirent->{$_}->kind != $SVN::Node::dir;
1356                $self->traverse_ignore($fh, "$path/$_", $r);
1357        }
1358}
1359
1360sub last_rev { ($_[0]->last_rev_commit)[0] }
1361sub last_commit { ($_[0]->last_rev_commit)[1] }
1362
1363# returns the newest SVN revision number and newest commit SHA1
1364sub last_rev_commit {
1365        my ($self) = @_;
1366        if (defined $self->{last_rev} && defined $self->{last_commit}) {
1367                return ($self->{last_rev}, $self->{last_commit});
1368        }
1369        my $c = ::verify_ref($self->refname.'^0');
1370        if ($c && !$self->use_svm_props && !$self->no_metadata) {
1371                my $rev = (::cmt_metadata($c))[1];
1372                if (defined $rev) {
1373                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1374                        return ($rev, $c);
1375                }
1376        }
1377        my $db_path = $self->db_path;
1378        unless (-e $db_path) {
1379                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1380                return (undef, undef);
1381        }
1382        my $offset = -41; # from tail
1383        my $rl;
1384        open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1385        sysseek($fh, $offset, 2); # don't care for errors
1386        sysread($fh, $rl, 41) == 41 or return (undef, undef);
1387        chomp $rl;
1388        while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1389                $offset -= 41;
1390                sysseek($fh, $offset, 2); # don't care for errors
1391                sysread($fh, $rl, 41) == 41 or return (undef, undef);
1392                chomp $rl;
1393        }
1394        if ($c && $c ne $rl) {
1395                die "$db_path and ", $self->refname,
1396                    " inconsistent!:\n$c != $rl\n";
1397        }
1398        my $rev = sysseek($fh, 0, 1) or croak $!;
1399        $rev =  ($rev - 41) / 41;
1400        close $fh or croak $!;
1401        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1402        return ($rev, $c);
1403}
1404
1405sub get_fetch_range {
1406        my ($self, $min, $max) = @_;
1407        $max ||= $self->ra->get_latest_revnum;
1408        $min ||= $self->rev_db_max;
1409        (++$min, $max);
1410}
1411
1412sub tmp_config {
1413        my (@args) = @_;
1414        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1415        my $config = "$ENV{GIT_DIR}/svn/.metadata";
1416        if (-e $old_def_config && ! -e $config) {
1417                rename $old_def_config, $config or
1418                       die "Failed rename $old_def_config => $config: $!\n";
1419        }
1420        my $old_config = $ENV{GIT_CONFIG};
1421        $ENV{GIT_CONFIG} = $config;
1422        $@ = undef;
1423        my @ret = eval {
1424                unless (-f $config) {
1425                        mkfile($config);
1426                        open my $fh, '>', $config or
1427                            die "Can't open $config: $!\n";
1428                        print $fh "; This file is used internally by ",
1429                                  "git-svn\n" or die
1430                                  "Couldn't write to $config: $!\n";
1431                        print $fh "; You should not have to edit it\n" or
1432                              die "Couldn't write to $config: $!\n";
1433                        close $fh or die "Couldn't close $config: $!\n";
1434                }
1435                command('config', @args);
1436        };
1437        my $err = $@;
1438        if (defined $old_config) {
1439                $ENV{GIT_CONFIG} = $old_config;
1440        } else {
1441                delete $ENV{GIT_CONFIG};
1442        }
1443        die $err if $err;
1444        wantarray ? @ret : $ret[0];
1445}
1446
1447sub tmp_index_do {
1448        my ($self, $sub) = @_;
1449        my $old_index = $ENV{GIT_INDEX_FILE};
1450        $ENV{GIT_INDEX_FILE} = $self->{index};
1451        $@ = undef;
1452        my @ret = eval {
1453                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1454                mkpath([$dir]) unless -d $dir;
1455                &$sub;
1456        };
1457        my $err = $@;
1458        if (defined $old_index) {
1459                $ENV{GIT_INDEX_FILE} = $old_index;
1460        } else {
1461                delete $ENV{GIT_INDEX_FILE};
1462        }
1463        die $err if $err;
1464        wantarray ? @ret : $ret[0];
1465}
1466
1467sub assert_index_clean {
1468        my ($self, $treeish) = @_;
1469
1470        $self->tmp_index_do(sub {
1471                command_noisy('read-tree', $treeish) unless -e $self->{index};
1472                my $x = command_oneline('write-tree');
1473                my ($y) = (command(qw/cat-file commit/, $treeish) =~
1474                           /^tree ($::sha1)/mo);
1475                return if $y eq $x;
1476
1477                warn "Index mismatch: $y != $x\nrereading $treeish\n";
1478                unlink $self->{index} or die "unlink $self->{index}: $!\n";
1479                command_noisy('read-tree', $treeish);
1480                $x = command_oneline('write-tree');
1481                if ($y ne $x) {
1482                        ::fatal "trees ($treeish) $y != $x\n",
1483                                "Something is seriously wrong...\n";
1484                }
1485        });
1486}
1487
1488sub get_commit_parents {
1489        my ($self, $log_entry) = @_;
1490        my (%seen, @ret, @tmp);
1491        # legacy support for 'set-tree'; this is only used by set_tree_cb:
1492        if (my $ip = $self->{inject_parents}) {
1493                if (my $commit = delete $ip->{$log_entry->{revision}}) {
1494                        push @tmp, $commit;
1495                }
1496        }
1497        if (my $cur = ::verify_ref($self->refname.'^0')) {
1498                push @tmp, $cur;
1499        }
1500        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1501        while (my $p = shift @tmp) {
1502                next if $seen{$p};
1503                $seen{$p} = 1;
1504                push @ret, $p;
1505                # MAXPARENT is defined to 16 in commit-tree.c:
1506                last if @ret >= 16;
1507        }
1508        if (@tmp) {
1509                die "r$log_entry->{revision}: No room for parents:\n\t",
1510                    join("\n\t", @tmp), "\n";
1511        }
1512        @ret;
1513}
1514
1515sub rewrite_root {
1516        my ($self) = @_;
1517        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
1518        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
1519        my $rwr = eval { command_oneline(qw/config --get/, $k) };
1520        if ($rwr) {
1521                $rwr =~ s#/+$##;
1522                if ($rwr !~ m#^[a-z\+]+://#) {
1523                        die "$rwr is not a valid URL (key: $k)\n";
1524                }
1525        }
1526        $self->{-rewrite_root} = $rwr;
1527}
1528
1529sub metadata_url {
1530        my ($self) = @_;
1531        ($self->rewrite_root || $self->{url}) .
1532           (length $self->{path} ? '/' . $self->{path} : '');
1533}
1534
1535sub full_url {
1536        my ($self) = @_;
1537        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1538}
1539
1540sub do_git_commit {
1541        my ($self, $log_entry) = @_;
1542        my $lr = $self->last_rev;
1543        if (defined $lr && $lr >= $log_entry->{revision}) {
1544                die "Last fetched revision of ", $self->refname,
1545                    " was r$lr, but we are about to fetch: ",
1546                    "r$log_entry->{revision}!\n";
1547        }
1548        if (my $c = $self->rev_db_get($log_entry->{revision})) {
1549                croak "$log_entry->{revision} = $c already exists! ",
1550                      "Why are we refetching it?\n";
1551        }
1552        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1553        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1554                                                          $log_entry->{email};
1555        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1556
1557        my $tree = $log_entry->{tree};
1558        if (!defined $tree) {
1559                $tree = $self->tmp_index_do(sub {
1560                                            command_oneline('write-tree') });
1561        }
1562        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1563
1564        my @exec = ('git-commit-tree', $tree);
1565        foreach ($self->get_commit_parents($log_entry)) {
1566                push @exec, '-p', $_;
1567        }
1568        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1569                                                                   or croak $!;
1570        print $msg_fh $log_entry->{log} or croak $!;
1571        unless ($self->no_metadata) {
1572                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1573                              or croak $!;
1574        }
1575        $msg_fh->flush == 0 or croak $!;
1576        close $msg_fh or croak $!;
1577        chomp(my $commit = do { local $/; <$out_fh> });
1578        close $out_fh or croak $!;
1579        waitpid $pid, 0;
1580        croak $? if $?;
1581        if ($commit !~ /^$::sha1$/o) {
1582                die "Failed to commit, invalid sha1: $commit\n";
1583        }
1584
1585        $self->rev_db_set($log_entry->{revision}, $commit, 1);
1586
1587        $self->{last_rev} = $log_entry->{revision};
1588        $self->{last_commit} = $commit;
1589        print "r$log_entry->{revision}";
1590        if (defined $log_entry->{svm_revision}) {
1591                 print " (\@$log_entry->{svm_revision})";
1592                 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1593                                   0, $self->svm_uuid);
1594        }
1595        print " = $commit ($self->{ref_id})\n";
1596        if (defined $_repack && (--$_repack_nr == 0)) {
1597                $_repack_nr = $_repack;
1598                # repack doesn't use any arguments with spaces in them, does it?
1599                print "Running git repack $_repack_flags ...\n";
1600                command_noisy('repack', split(/\s+/, $_repack_flags));
1601                print "Done repacking\n";
1602        }
1603        return $commit;
1604}
1605
1606sub match_paths {
1607        my ($self, $paths, $r) = @_;
1608        return 1 if $self->{path} eq '';
1609        if (my $path = $paths->{"/$self->{path}"}) {
1610                return ($path->{action} eq 'D') ? 0 : 1;
1611        }
1612        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1613        if (grep /$self->{path_regex}/, keys %$paths) {
1614                return 1;
1615        }
1616        my $c = '';
1617        foreach (split m#/#, $self->{path}) {
1618                $c .= "/$_";
1619                next unless ($paths->{$c} &&
1620                             ($paths->{$c}->{action} =~ /^[AR]$/));
1621                if ($self->ra->check_path($self->{path}, $r) ==
1622                    $SVN::Node::dir) {
1623                        return 1;
1624                }
1625        }
1626        return 0;
1627}
1628
1629sub find_parent_branch {
1630        my ($self, $paths, $rev) = @_;
1631        return undef unless $self->follow_parent;
1632        unless (defined $paths) {
1633                my $err_handler = $SVN::Error::handler;
1634                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1635                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1636                                   $paths =
1637                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
1638                $SVN::Error::handler = $err_handler;
1639        }
1640        return undef unless defined $paths;
1641
1642        # look for a parent from another branch:
1643        my @b_path_components = split m#/#, $self->rel_path;
1644        my @a_path_components;
1645        my $i;
1646        while (@b_path_components) {
1647                $i = $paths->{'/'.join('/', @b_path_components)};
1648                last if $i && defined $i->{copyfrom_path};
1649                unshift(@a_path_components, pop(@b_path_components));
1650        }
1651        return undef unless defined $i && defined $i->{copyfrom_path};
1652        my $branch_from = $i->{copyfrom_path};
1653        if (@a_path_components) {
1654                print STDERR "branch_from: $branch_from => ";
1655                $branch_from .= '/'.join('/', @a_path_components);
1656                print STDERR $branch_from, "\n";
1657        }
1658        my $r = $i->{copyfrom_rev};
1659        my $repos_root = $self->ra->{repos_root};
1660        my $url = $self->ra->{url};
1661        my $new_url = $repos_root . $branch_from;
1662        print STDERR  "Found possible branch point: ",
1663                      "$new_url => ", $self->full_url, ", $r\n";
1664        $branch_from =~ s#^/##;
1665        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
1666        unless ($gs) {
1667                my $ref_id = $self->{ref_id};
1668                $ref_id =~ s/\@\d+$//;
1669                $ref_id .= "\@$r";
1670                # just grow a tail if we're not unique enough :x
1671                $ref_id .= '-' while find_ref($ref_id);
1672                print STDERR "Initializing parent: $ref_id\n";
1673                $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1674        }
1675        my ($r0, $parent) = $gs->find_rev_before($r, 1);
1676        if (!defined $r0 || !defined $parent) {
1677                $gs->fetch(0, $r);
1678                ($r0, $parent) = $gs->last_rev_commit;
1679        }
1680        if (defined $r0 && defined $parent) {
1681                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1682                $self->assert_index_clean($parent);
1683                my $ed;
1684                if ($self->ra->can_do_switch) {
1685                        print STDERR "Following parent with do_switch\n";
1686                        # do_switch works with svn/trunk >= r22312, but that
1687                        # is not included with SVN 1.4.3 (the latest version
1688                        # at the moment), so we can't rely on it
1689                        $self->{last_commit} = $parent;
1690                        $ed = SVN::Git::Fetcher->new($self);
1691                        $gs->ra->gs_do_switch($r0, $rev, $gs,
1692                                              $self->full_url, $ed)
1693                          or die "SVN connection failed somewhere...\n";
1694                } else {
1695                        print STDERR "Following parent with do_update\n";
1696                        $ed = SVN::Git::Fetcher->new($self);
1697                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
1698                          or die "SVN connection failed somewhere...\n";
1699                }
1700                print STDERR "Successfully followed parent\n";
1701                return $self->make_log_entry($rev, [$parent], $ed);
1702        }
1703        return undef;
1704}
1705
1706sub do_fetch {
1707        my ($self, $paths, $rev) = @_;
1708        my $ed;
1709        my ($last_rev, @parents);
1710        if (my $lc = $self->last_commit) {
1711                # we can have a branch that was deleted, then re-added
1712                # under the same name but copied from another path, in
1713                # which case we'll have multiple parents (we don't
1714                # want to break the original ref, nor lose copypath info):
1715                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1716                        push @{$log_entry->{parents}}, $lc;
1717                        return $log_entry;
1718                }
1719                $ed = SVN::Git::Fetcher->new($self);
1720                $last_rev = $self->{last_rev};
1721                $ed->{c} = $lc;
1722                @parents = ($lc);
1723        } else {
1724                $last_rev = $rev;
1725                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1726                        return $log_entry;
1727                }
1728                $ed = SVN::Git::Fetcher->new($self);
1729        }
1730        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1731                die "SVN connection failed somewhere...\n";
1732        }
1733        $self->make_log_entry($rev, \@parents, $ed);
1734}
1735
1736sub get_untracked {
1737        my ($self, $ed) = @_;
1738        my @out;
1739        my $h = $ed->{empty};
1740        foreach (sort keys %$h) {
1741                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1742                push @out, "  $act: " . uri_encode($_);
1743                warn "W: $act: $_\n";
1744        }
1745        foreach my $t (qw/dir_prop file_prop/) {
1746                $h = $ed->{$t} or next;
1747                foreach my $path (sort keys %$h) {
1748                        my $ppath = $path eq '' ? '.' : $path;
1749                        foreach my $prop (sort keys %{$h->{$path}}) {
1750                                next if $SKIP_PROP{$prop};
1751                                my $v = $h->{$path}->{$prop};
1752                                my $t_ppath_prop = "$t: " .
1753                                                    uri_encode($ppath) . ' ' .
1754                                                    uri_encode($prop);
1755                                if (defined $v) {
1756                                        push @out, "  +$t_ppath_prop " .
1757                                                   uri_encode($v);
1758                                } else {
1759                                        push @out, "  -$t_ppath_prop";
1760                                }
1761                        }
1762                }
1763        }
1764        foreach my $t (qw/absent_file absent_directory/) {
1765                $h = $ed->{$t} or next;
1766                foreach my $parent (sort keys %$h) {
1767                        foreach my $path (sort @{$h->{$parent}}) {
1768                                push @out, "  $t: " .
1769                                           uri_encode("$parent/$path");
1770                                warn "W: $t: $parent/$path ",
1771                                     "Insufficient permissions?\n";
1772                        }
1773                }
1774        }
1775        \@out;
1776}
1777
1778sub parse_svn_date {
1779        my $date = shift || return '+0000 1970-01-01 00:00:00';
1780        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1781                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1782                                         croak "Unable to parse date: $date\n";
1783        "+0000 $Y-$m-$d $H:$M:$S";
1784}
1785
1786sub check_author {
1787        my ($author) = @_;
1788        if (!defined $author || length $author == 0) {
1789                $author = '(no author)';
1790        }
1791        if (defined $::_authors && ! defined $::users{$author}) {
1792                die "Author: $author not defined in $::_authors file\n";
1793        }
1794        $author;
1795}
1796
1797sub make_log_entry {
1798        my ($self, $rev, $parents, $ed) = @_;
1799        my $untracked = $self->get_untracked($ed);
1800
1801        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1802        print $un "r$rev\n" or croak $!;
1803        print $un $_, "\n" foreach @$untracked;
1804        my %log_entry = ( parents => $parents || [], revision => $rev,
1805                          log => '');
1806
1807        my $headrev;
1808        my $logged = delete $self->{logged_rev_props};
1809        if (!$logged || $self->{-want_revprops}) {
1810                my $rp = $self->ra->rev_proplist($rev);
1811                foreach (sort keys %$rp) {
1812                        my $v = $rp->{$_};
1813                        if (/^svn:(author|date|log)$/) {
1814                                $log_entry{$1} = $v;
1815                        } elsif ($_ eq 'svm:headrev') {
1816                                $headrev = $v;
1817                        } else {
1818                                print $un "  rev_prop: ", uri_encode($_), ' ',
1819                                          uri_encode($v), "\n";
1820                        }
1821                }
1822        } else {
1823                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1824        }
1825        close $un or croak $!;
1826
1827        $log_entry{date} = parse_svn_date($log_entry{date});
1828        $log_entry{log} .= "\n";
1829        my $author = $log_entry{author} = check_author($log_entry{author});
1830        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1831                                                       : ($author, undef);
1832        if (defined $headrev && $self->use_svm_props) {
1833                if ($self->rewrite_root) {
1834                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
1835                            "options set!\n";
1836                }
1837                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1838                # we don't want "SVM: initializing mirror for junk" ...
1839                return undef if $r == 0;
1840                my $svm = $self->svm;
1841                if ($uuid ne $svm->{uuid}) {
1842                        die "UUID mismatch on SVM path:\n",
1843                            "expected: $svm->{uuid}\n",
1844                            "     got: $uuid\n";
1845                }
1846                my $full_url = $self->full_url;
1847                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
1848                             die "Failed to replace '$svm->{replace}' with ",
1849                                 "'$svm->{source}' in $full_url\n";
1850                $log_entry{metadata} = "$full_url\@$r $uuid";
1851                $log_entry{svm_revision} = $r;
1852                $email ||= "$author\@$uuid"
1853        } elsif ($self->use_svnsync_props) {
1854                my $full_url = $self->svnsync->{url};
1855                $full_url .= "/$self->{path}" if length $self->{path};
1856                my $uuid = $self->svnsync->{uuid};
1857                $log_entry{metadata} = "$full_url\@$rev $uuid";
1858                $email ||= "$author\@$uuid"
1859        } else {
1860                $log_entry{metadata} = $self->metadata_url. "\@$rev " .
1861                                       $self->ra->get_uuid;
1862                $email ||= "$author\@" . $self->ra->get_uuid;
1863        }
1864        $log_entry{name} = $name;
1865        $log_entry{email} = $email;
1866        \%log_entry;
1867}
1868
1869sub fetch {
1870        my ($self, $min_rev, $max_rev, @parents) = @_;
1871        my ($last_rev, $last_commit) = $self->last_rev_commit;
1872        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1873        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1874}
1875
1876sub set_tree_cb {
1877        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1878        $self->{inject_parents} = { $rev => $tree };
1879        $self->fetch(undef, undef);
1880}
1881
1882sub set_tree {
1883        my ($self, $tree) = (shift, shift);
1884        my $log_entry = ::get_commit_entry($tree);
1885        unless ($self->{last_rev}) {
1886                fatal("Must have an existing revision to commit\n");
1887        }
1888        my %ed_opts = ( r => $self->{last_rev},
1889                        log => $log_entry->{log},
1890                        ra => $self->ra,
1891                        tree_a => $self->{last_commit},
1892                        tree_b => $tree,
1893                        editor_cb => sub {
1894                               $self->set_tree_cb($log_entry, $tree, @_) },
1895                        svn_path => $self->{path} );
1896        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1897                print "No changes\nr$self->{last_rev} = $tree\n";
1898        }
1899}
1900
1901sub rebuild {
1902        my ($self) = @_;
1903        my $db_path = $self->db_path;
1904        return if (-e $db_path && ! -z $db_path);
1905        return unless ::verify_ref($self->refname.'^0');
1906        if (-f $self->{db_root}) {
1907                rename $self->{db_root}, $db_path or die
1908                     "rename $self->{db_root} => $db_path failed: $!\n";
1909                my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
1910                symlink $base, $self->{db_root} or die
1911                     "symlink $base => $self->{db_root} failed: $!\n";
1912                return;
1913        }
1914        print "Rebuilding $db_path ...\n";
1915        my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1916        my $latest;
1917        my $full_url = $self->full_url;
1918        my $svn_uuid;
1919        while (<$rev_list>) {
1920                chomp;
1921                my $c = $_;
1922                die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1923                my ($url, $rev, $uuid) = ::cmt_metadata($c);
1924
1925                # ignore merges (from set-tree)
1926                next if (!defined $rev || !$uuid);
1927
1928                # if we merged or otherwise started elsewhere, this is
1929                # how we break out of it
1930                if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1931                    ($full_url && $url && ($url ne $full_url))) {
1932                        next;
1933                }
1934                $latest ||= $rev;
1935                $svn_uuid ||= $uuid;
1936
1937                $self->rev_db_set($rev, $c);
1938                print "r$rev = $c\n";
1939        }
1940        command_close_pipe($rev_list, $ctx);
1941        print "Done rebuilding $db_path\n";
1942}
1943
1944# rev_db:
1945# Tie::File seems to be prone to offset errors if revisions get sparse,
1946# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
1947# one of my favorite modules is out :<  Next up would be one of the DBM
1948# modules, but I'm not sure which is most portable...  So I'll just
1949# go with something that's plain-text, but still capable of
1950# being randomly accessed.  So here's my ultra-simple fixed-width
1951# database.  All records are 40 characters + "\n", so it's easy to seek
1952# to a revision: (41 * rev) is the byte offset.
1953# A record of 40 0s denotes an empty revision.
1954# And yes, it's still pretty fast (faster than Tie::File).
1955# These files are disposable unless noMetadata or useSvmProps is set
1956
1957sub _rev_db_set {
1958        my ($fh, $rev, $commit) = @_;
1959        my $offset = $rev * 41;
1960        # assume that append is the common case:
1961        seek $fh, 0, 2 or croak $!;
1962        my $pos = tell $fh;
1963        if ($pos < $offset) {
1964                for (1 .. (($offset - $pos) / 41)) {
1965                        print $fh (('0' x 40),"\n") or croak $!;
1966                }
1967        }
1968        seek $fh, $offset, 0 or croak $!;
1969        print $fh $commit,"\n" or croak $!;
1970}
1971
1972sub mkfile {
1973        my ($path) = @_;
1974        unless (-e $path) {
1975                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
1976                mkpath([$dir]) unless -d $dir;
1977                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
1978                close $fh or die "Couldn't close (create) $path: $!\n";
1979        }
1980}
1981
1982sub rev_db_set {
1983        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
1984        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
1985        my $db = $self->db_path($uuid);
1986        my $db_lock = "$db.lock";
1987        my $sig;
1988        if ($update_ref) {
1989                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1990                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
1991        }
1992        mkfile($db);
1993
1994        $LOCKFILES{$db_lock} = 1;
1995        my $sync;
1996        # both of these options make our .rev_db file very, very important
1997        # and we can't afford to lose it because rebuild() won't work
1998        if ($self->use_svm_props || $self->no_metadata) {
1999                $sync = 1;
2000                copy($db, $db_lock) or die "rev_db_set(@_): ",
2001                                           "Failed to copy: ",
2002                                           "$db => $db_lock ($!)\n";
2003        } else {
2004                rename $db, $db_lock or die "rev_db_set(@_): ",
2005                                            "Failed to rename: ",
2006                                            "$db => $db_lock ($!)\n";
2007        }
2008        open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
2009        _rev_db_set($fh, $rev, $commit);
2010        if ($sync) {
2011                $fh->flush or die "Couldn't flush $db_lock: $!\n";
2012                $fh->sync or die "Couldn't sync $db_lock: $!\n";
2013        }
2014        close $fh or croak $!;
2015        if ($update_ref) {
2016                $_head = $self;
2017                command_noisy('update-ref', '-m', "r$rev",
2018                              $self->refname, $commit);
2019        }
2020        rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
2021                                    "$db_lock => $db ($!)\n";
2022        delete $LOCKFILES{$db_lock};
2023        if ($update_ref) {
2024                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2025                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2026                kill $sig, $$ if defined $sig;
2027        }
2028}
2029
2030sub rev_db_max {
2031        my ($self) = @_;
2032        $self->rebuild;
2033        my $db_path = $self->db_path;
2034        my @stat = stat $db_path or return 0;
2035        ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
2036        my $max = $stat[7] / 41;
2037        (($max > 0) ? $max - 1 : 0);
2038}
2039
2040sub rev_db_get {
2041        my ($self, $rev, $uuid) = @_;
2042        my $ret;
2043        my $offset = $rev * 41;
2044        my $db_path = $self->db_path($uuid);
2045        return undef unless -e $db_path;
2046        open my $fh, '<', $db_path or croak $!;
2047        if (sysseek($fh, $offset, 0) == $offset) {
2048                my $read = sysread($fh, $ret, 40);
2049                $ret = undef if ($read != 40 || $ret eq ('0'x40));
2050        }
2051        close $fh or croak $!;
2052        $ret;
2053}
2054
2055sub find_rev_before {
2056        my ($self, $rev, $eq_ok) = @_;
2057        --$rev unless $eq_ok;
2058        while ($rev > 0) {
2059                if (my $c = $self->rev_db_get($rev)) {
2060                        return ($rev, $c);
2061                }
2062                --$rev;
2063        }
2064        return (undef, undef);
2065}
2066
2067sub _new {
2068        my ($class, $repo_id, $ref_id, $path) = @_;
2069        unless (defined $repo_id && length $repo_id) {
2070                $repo_id = $Git::SVN::default_repo_id;
2071        }
2072        unless (defined $ref_id && length $ref_id) {
2073                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2074        }
2075        $_[1] = $repo_id = sanitize_remote_name($repo_id);
2076        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2077        $_[3] = $path = '' unless (defined $path);
2078        mkpath(["$ENV{GIT_DIR}/svn"]);
2079        bless {
2080                ref_id => $ref_id, dir => $dir, index => "$dir/index",
2081                path => $path, config => "$ENV{GIT_DIR}/svn/config",
2082                db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2083}
2084
2085sub db_path {
2086        my ($self, $uuid) = @_;
2087        $uuid ||= $self->ra_uuid;
2088        "$self->{db_root}.$uuid";
2089}
2090
2091sub uri_encode {
2092        my ($f) = @_;
2093        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2094        $f
2095}
2096
2097package Git::SVN::Prompt;
2098use strict;
2099use warnings;
2100require SVN::Core;
2101use vars qw/$_no_auth_cache $_username/;
2102
2103sub simple {
2104        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2105        $may_save = undef if $_no_auth_cache;
2106        $default_username = $_username if defined $_username;
2107        if (defined $default_username && length $default_username) {
2108                if (defined $realm && length $realm) {
2109                        print STDERR "Authentication realm: $realm\n";
2110                        STDERR->flush;
2111                }
2112                $cred->username($default_username);
2113        } else {
2114                username($cred, $realm, $may_save, $pool);
2115        }
2116        $cred->password(_read_password("Password for '" .
2117                                       $cred->username . "': ", $realm));
2118        $cred->may_save($may_save);
2119        $SVN::_Core::SVN_NO_ERROR;
2120}
2121
2122sub ssl_server_trust {
2123        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2124        $may_save = undef if $_no_auth_cache;
2125        print STDERR "Error validating server certificate for '$realm':\n";
2126        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2127                print STDERR " - The certificate is not issued by a trusted ",
2128                      "authority. Use the\n",
2129                      "   fingerprint to validate the certificate manually!\n";
2130        }
2131        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2132                print STDERR " - The certificate hostname does not match.\n";
2133        }
2134        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2135                print STDERR " - The certificate is not yet valid.\n";
2136        }
2137        if ($failures & $SVN::Auth::SSL::EXPIRED) {
2138                print STDERR " - The certificate has expired.\n";
2139        }
2140        if ($failures & $SVN::Auth::SSL::OTHER) {
2141                print STDERR " - The certificate has an unknown error.\n";
2142        }
2143        printf STDERR
2144                "Certificate information:\n".
2145                " - Hostname: %s\n".
2146                " - Valid: from %s until %s\n".
2147                " - Issuer: %s\n".
2148                " - Fingerprint: %s\n",
2149                map $cert_info->$_, qw(hostname valid_from valid_until
2150                                       issuer_dname fingerprint);
2151        my $choice;
2152prompt:
2153        print STDERR $may_save ?
2154              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2155              "(R)eject or accept (t)emporarily? ";
2156        STDERR->flush;
2157        $choice = lc(substr(<STDIN> || 'R', 0, 1));
2158        if ($choice =~ /^t$/i) {
2159                $cred->may_save(undef);
2160        } elsif ($choice =~ /^r$/i) {
2161                return -1;
2162        } elsif ($may_save && $choice =~ /^p$/i) {
2163                $cred->may_save($may_save);
2164        } else {
2165                goto prompt;
2166        }
2167        $cred->accepted_failures($failures);
2168        $SVN::_Core::SVN_NO_ERROR;
2169}
2170
2171sub ssl_client_cert {
2172        my ($cred, $realm, $may_save, $pool) = @_;
2173        $may_save = undef if $_no_auth_cache;
2174        print STDERR "Client certificate filename: ";
2175        STDERR->flush;
2176        chomp(my $filename = <STDIN>);
2177        $cred->cert_file($filename);
2178        $cred->may_save($may_save);
2179        $SVN::_Core::SVN_NO_ERROR;
2180}
2181
2182sub ssl_client_cert_pw {
2183        my ($cred, $realm, $may_save, $pool) = @_;
2184        $may_save = undef if $_no_auth_cache;
2185        $cred->password(_read_password("Password: ", $realm));
2186        $cred->may_save($may_save);
2187        $SVN::_Core::SVN_NO_ERROR;
2188}
2189
2190sub username {
2191        my ($cred, $realm, $may_save, $pool) = @_;
2192        $may_save = undef if $_no_auth_cache;
2193        if (defined $realm && length $realm) {
2194                print STDERR "Authentication realm: $realm\n";
2195        }
2196        my $username;
2197        if (defined $_username) {
2198                $username = $_username;
2199        } else {
2200                print STDERR "Username: ";
2201                STDERR->flush;
2202                chomp($username = <STDIN>);
2203        }
2204        $cred->username($username);
2205        $cred->may_save($may_save);
2206        $SVN::_Core::SVN_NO_ERROR;
2207}
2208
2209sub _read_password {
2210        my ($prompt, $realm) = @_;
2211        print STDERR $prompt;
2212        STDERR->flush;
2213        require Term::ReadKey;
2214        Term::ReadKey::ReadMode('noecho');
2215        my $password = '';
2216        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2217                last if $key =~ /[\012\015]/; # \n\r
2218                $password .= $key;
2219        }
2220        Term::ReadKey::ReadMode('restore');
2221        print STDERR "\n";
2222        STDERR->flush;
2223        $password;
2224}
2225
2226package main;
2227
2228{
2229        my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2230                                $SVN::Node::dir.$SVN::Node::unknown.
2231                                $SVN::Node::none.$SVN::Node::file.
2232                                $SVN::Node::dir.$SVN::Node::unknown.
2233                                $SVN::Auth::SSL::CNMISMATCH.
2234                                $SVN::Auth::SSL::NOTYETVALID.
2235                                $SVN::Auth::SSL::EXPIRED.
2236                                $SVN::Auth::SSL::UNKNOWNCA.
2237                                $SVN::Auth::SSL::OTHER;
2238}
2239
2240package SVN::Git::Fetcher;
2241use vars qw/@ISA/;
2242use strict;
2243use warnings;
2244use Carp qw/croak/;
2245use IO::File qw//;
2246use Digest::MD5;
2247
2248# file baton members: path, mode_a, mode_b, pool, fh, blob, base
2249sub new {
2250        my ($class, $git_svn) = @_;
2251        my $self = SVN::Delta::Editor->new;
2252        bless $self, $class;
2253        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2254        $self->{empty} = {};
2255        $self->{dir_prop} = {};
2256        $self->{file_prop} = {};
2257        $self->{absent_dir} = {};
2258        $self->{absent_file} = {};
2259        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2260        $self;
2261}
2262
2263sub set_path_strip {
2264        my ($self, $path) = @_;
2265        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2266}
2267
2268sub open_root {
2269        { path => '' };
2270}
2271
2272sub open_directory {
2273        my ($self, $path, $pb, $rev) = @_;
2274        { path => $path };
2275}
2276
2277sub git_path {
2278        my ($self, $path) = @_;
2279        if ($self->{path_strip}) {
2280                $path =~ s!$self->{path_strip}!! or
2281                  die "Failed to strip path '$path' ($self->{path_strip})\n";
2282        }
2283        $path;
2284}
2285
2286sub delete_entry {
2287        my ($self, $path, $rev, $pb) = @_;
2288
2289        my $gpath = $self->git_path($path);
2290        return undef if ($gpath eq '');
2291
2292        # remove entire directories.
2293        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2294                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2295                                                     -r --name-only -z/,
2296                                                     $self->{c}, '--', $gpath);
2297                local $/ = "\0";
2298                while (<$ls>) {
2299                        chomp;
2300                        $self->{gii}->remove($_);
2301                        print "\tD\t$_\n" unless $::_q;
2302                }
2303                print "\tD\t$gpath/\n" unless $::_q;
2304                command_close_pipe($ls, $ctx);
2305                $self->{empty}->{$path} = 0
2306        } else {
2307                $self->{gii}->remove($gpath);
2308                print "\tD\t$gpath\n" unless $::_q;
2309        }
2310        undef;
2311}
2312
2313sub open_file {
2314        my ($self, $path, $pb, $rev) = @_;
2315        my $gpath = $self->git_path($path);
2316        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2317                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2318        unless (defined $mode && defined $blob) {
2319                die "$path was not found in commit $self->{c} (r$rev)\n";
2320        }
2321        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2322          pool => SVN::Pool->new, action => 'M' };
2323}
2324
2325sub add_file {
2326        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2327        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2328        delete $self->{empty}->{$dir};
2329        { path => $path, mode_a => 100644, mode_b => 100644,
2330          pool => SVN::Pool->new, action => 'A' };
2331}
2332
2333sub add_directory {
2334        my ($self, $path, $cp_path, $cp_rev) = @_;
2335        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2336        delete $self->{empty}->{$dir};
2337        $self->{empty}->{$path} = 1;
2338        { path => $path };
2339}
2340
2341sub change_dir_prop {
2342        my ($self, $db, $prop, $value) = @_;
2343        $self->{dir_prop}->{$db->{path}} ||= {};
2344        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2345        undef;
2346}
2347
2348sub absent_directory {
2349        my ($self, $path, $pb) = @_;
2350        $self->{absent_dir}->{$pb->{path}} ||= [];
2351        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2352        undef;
2353}
2354
2355sub absent_file {
2356        my ($self, $path, $pb) = @_;
2357        $self->{absent_file}->{$pb->{path}} ||= [];
2358        push @{$self->{absent_file}->{$pb->{path}}}, $path;
2359        undef;
2360}
2361
2362sub change_file_prop {
2363        my ($self, $fb, $prop, $value) = @_;
2364        if ($prop eq 'svn:executable') {
2365                if ($fb->{mode_b} != 120000) {
2366                        $fb->{mode_b} = defined $value ? 100755 : 100644;
2367                }
2368        } elsif ($prop eq 'svn:special') {
2369                $fb->{mode_b} = defined $value ? 120000 : 100644;
2370        } else {
2371                $self->{file_prop}->{$fb->{path}} ||= {};
2372                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2373        }
2374        undef;
2375}
2376
2377sub apply_textdelta {
2378        my ($self, $fb, $exp) = @_;
2379        my $fh = IO::File->new_tmpfile;
2380        $fh->autoflush(1);
2381        # $fh gets auto-closed() by SVN::TxDelta::apply(),
2382        # (but $base does not,) so dup() it for reading in close_file
2383        open my $dup, '<&', $fh or croak $!;
2384        my $base = IO::File->new_tmpfile;
2385        $base->autoflush(1);
2386        if ($fb->{blob}) {
2387                defined (my $pid = fork) or croak $!;
2388                if (!$pid) {
2389                        open STDOUT, '>&', $base or croak $!;
2390                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2391                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2392                }
2393                waitpid $pid, 0;
2394                croak $? if $?;
2395
2396                if (defined $exp) {
2397                        seek $base, 0, 0 or croak $!;
2398                        my $md5 = Digest::MD5->new;
2399                        $md5->addfile($base);
2400                        my $got = $md5->hexdigest;
2401                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2402                            "expected: $exp\n",
2403                            "     got: $got\n" if ($got ne $exp);
2404                }
2405        }
2406        seek $base, 0, 0 or croak $!;
2407        $fb->{fh} = $dup;
2408        $fb->{base} = $base;
2409        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2410}
2411
2412sub close_file {
2413        my ($self, $fb, $exp) = @_;
2414        my $hash;
2415        my $path = $self->git_path($fb->{path});
2416        if (my $fh = $fb->{fh}) {
2417                seek($fh, 0, 0) or croak $!;
2418                my $md5 = Digest::MD5->new;
2419                $md5->addfile($fh);
2420                my $got = $md5->hexdigest;
2421                die "Checksum mismatch: $path\n",
2422                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
2423                seek($fh, 0, 0) or croak $!;
2424                if ($fb->{mode_b} == 120000) {
2425                        read($fh, my $buf, 5) == 5 or croak $!;
2426                        $buf eq 'link ' or die "$path has mode 120000",
2427                                               "but is not a link\n";
2428                }
2429                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2430                if (!$pid) {
2431                        open STDIN, '<&', $fh or croak $!;
2432                        exec qw/git-hash-object -w --stdin/ or croak $!;
2433                }
2434                chomp($hash = do { local $/; <$out> });
2435                close $out or croak $!;
2436                close $fh or croak $!;
2437                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2438                close $fb->{base} or croak $!;
2439        } else {
2440                $hash = $fb->{blob} or die "no blob information\n";
2441        }
2442        $fb->{pool}->clear;
2443        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2444        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2445        undef;
2446}
2447
2448sub abort_edit {
2449        my $self = shift;
2450        $self->{nr} = $self->{gii}->{nr};
2451        delete $self->{gii};
2452        $self->SUPER::abort_edit(@_);
2453}
2454
2455sub close_edit {
2456        my $self = shift;
2457        $self->{git_commit_ok} = 1;
2458        $self->{nr} = $self->{gii}->{nr};
2459        delete $self->{gii};
2460        $self->SUPER::close_edit(@_);
2461}
2462
2463package SVN::Git::Editor;
2464use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2465use strict;
2466use warnings;
2467use Carp qw/croak/;
2468use IO::File;
2469use Digest::MD5;
2470
2471sub new {
2472        my ($class, $opts) = @_;
2473        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2474                die "$_ required!\n" unless (defined $opts->{$_});
2475        }
2476
2477        my $pool = SVN::Pool->new;
2478        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2479        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2480                                     $opts->{r}, $mods);
2481
2482        # $opts->{ra} functions should not be used after this:
2483        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
2484                                                $opts->{editor_cb}, $pool);
2485        my $self = SVN::Delta::Editor->new(@ce, $pool);
2486        bless $self, $class;
2487        foreach (qw/svn_path r tree_a tree_b/) {
2488                $self->{$_} = $opts->{$_};
2489        }
2490        $self->{url} = $opts->{ra}->{url};
2491        $self->{mods} = $mods;
2492        $self->{types} = $types;
2493        $self->{pool} = $pool;
2494        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2495        $self->{rm} = { };
2496        $self->{path_prefix} = length $self->{svn_path} ?
2497                               "$self->{svn_path}/" : '';
2498        return $self;
2499}
2500
2501sub generate_diff {
2502        my ($tree_a, $tree_b) = @_;
2503        my @diff_tree = qw(diff-tree -z -r);
2504        if ($_cp_similarity) {
2505                push @diff_tree, "-C$_cp_similarity";
2506        } else {
2507                push @diff_tree, '-C';
2508        }
2509        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2510        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2511        push @diff_tree, $tree_a, $tree_b;
2512        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2513        local $/ = "\0";
2514        my $state = 'meta';
2515        my @mods;
2516        while (<$diff_fh>) {
2517                chomp $_; # this gets rid of the trailing "\0"
2518                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2519                                        $::sha1\s($::sha1)\s
2520                                        ([MTCRAD])\d*$/xo) {
2521                        push @mods, {   mode_a => $1, mode_b => $2,
2522                                        sha1_b => $3, chg => $4 };
2523                        if ($4 =~ /^(?:C|R)$/) {
2524                                $state = 'file_a';
2525                        } else {
2526                                $state = 'file_b';
2527                        }
2528                } elsif ($state eq 'file_a') {
2529                        my $x = $mods[$#mods] or croak "Empty array\n";
2530                        if ($x->{chg} !~ /^(?:C|R)$/) {
2531                                croak "Error parsing $_, $x->{chg}\n";
2532                        }
2533                        $x->{file_a} = $_;
2534                        $state = 'file_b';
2535                } elsif ($state eq 'file_b') {
2536                        my $x = $mods[$#mods] or croak "Empty array\n";
2537                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2538                                croak "Error parsing $_, $x->{chg}\n";
2539                        }
2540                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2541                                croak "Error parsing $_, $x->{chg}\n";
2542                        }
2543                        $x->{file_b} = $_;
2544                        $state = 'meta';
2545                } else {
2546                        croak "Error parsing $_\n";
2547                }
2548        }
2549        command_close_pipe($diff_fh, $ctx);
2550        \@mods;
2551}
2552
2553sub check_diff_paths {
2554        my ($ra, $pfx, $rev, $mods) = @_;
2555        my %types;
2556        $pfx .= '/' if length $pfx;
2557
2558        sub type_diff_paths {
2559                my ($ra, $types, $path, $rev) = @_;
2560                my @p = split m#/+#, $path;
2561                my $c = shift @p;
2562                unless (defined $types->{$c}) {
2563                        $types->{$c} = $ra->check_path($c, $rev);
2564                }
2565                while (@p) {
2566                        $c .= '/' . shift @p;
2567                        next if defined $types->{$c};
2568                        $types->{$c} = $ra->check_path($c, $rev);
2569                }
2570        }
2571
2572        foreach my $m (@$mods) {
2573                foreach my $f (qw/file_a file_b/) {
2574                        next unless defined $m->{$f};
2575                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2576                        if (length $pfx.$dir && ! defined $types{$dir}) {
2577                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2578                        }
2579                }
2580        }
2581        \%types;
2582}
2583
2584sub split_path {
2585        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2586}
2587
2588sub repo_path {
2589        my ($self, $path) = @_;
2590        $self->{path_prefix}.(defined $path ? $path : '');
2591}
2592
2593sub url_path {
2594        my ($self, $path) = @_;
2595        $self->{url} . '/' . $self->repo_path($path);
2596}
2597
2598sub rmdirs {
2599        my ($self) = @_;
2600        my $rm = $self->{rm};
2601        delete $rm->{''}; # we never delete the url we're tracking
2602        return unless %$rm;
2603
2604        foreach (keys %$rm) {
2605                my @d = split m#/#, $_;
2606                my $c = shift @d;
2607                $rm->{$c} = 1;
2608                while (@d) {
2609                        $c .= '/' . shift @d;
2610                        $rm->{$c} = 1;
2611                }
2612        }
2613        delete $rm->{$self->{svn_path}};
2614        delete $rm->{''}; # we never delete the url we're tracking
2615        return unless %$rm;
2616
2617        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2618                                             $self->{tree_b});
2619        local $/ = "\0";
2620        while (<$fh>) {
2621                chomp;
2622                my @dn = split m#/#, $_;
2623                while (pop @dn) {
2624                        delete $rm->{join '/', @dn};
2625                }
2626                unless (%$rm) {
2627                        close $fh;
2628                        return;
2629                }
2630        }
2631        command_close_pipe($fh, $ctx);
2632
2633        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2634        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2635                $self->close_directory($bat->{$d}, $p);
2636                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2637                print "\tD+\t$d/\n" unless $::_q;
2638                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2639                delete $bat->{$d};
2640        }
2641}
2642
2643sub open_or_add_dir {
2644        my ($self, $full_path, $baton) = @_;
2645        my $t = $self->{types}->{$full_path};
2646        if (!defined $t) {
2647                die "$full_path not known in r$self->{r} or we have a bug!\n";
2648        }
2649        if ($t == $SVN::Node::none) {
2650                return $self->add_directory($full_path, $baton,
2651                                                undef, -1, $self->{pool});
2652        } elsif ($t == $SVN::Node::dir) {
2653                return $self->open_directory($full_path, $baton,
2654                                                $self->{r}, $self->{pool});
2655        }
2656        print STDERR "$full_path already exists in repository at ",
2657                "r$self->{r} and it is not a directory (",
2658                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2659        exit 1;
2660}
2661
2662sub ensure_path {
2663        my ($self, $path) = @_;
2664        my $bat = $self->{bat};
2665        my $repo_path = $self->repo_path($path);
2666        return $bat->{''} unless (length $repo_path);
2667        my @p = split m#/+#, $repo_path;
2668        my $c = shift @p;
2669        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2670        while (@p) {
2671                my $c0 = $c;
2672                $c .= '/' . shift @p;
2673                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2674        }
2675        return $bat->{$c};
2676}
2677
2678sub A {
2679        my ($self, $m) = @_;
2680        my ($dir, $file) = split_path($m->{file_b});
2681        my $pbat = $self->ensure_path($dir);
2682        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2683                                        undef, -1);
2684        print "\tA\t$m->{file_b}\n" unless $::_q;
2685        $self->chg_file($fbat, $m);
2686        $self->close_file($fbat,undef,$self->{pool});
2687}
2688
2689sub C {
2690        my ($self, $m) = @_;
2691        my ($dir, $file) = split_path($m->{file_b});
2692        my $pbat = $self->ensure_path($dir);
2693        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2694                                $self->url_path($m->{file_a}), $self->{r});
2695        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2696        $self->chg_file($fbat, $m);
2697        $self->close_file($fbat,undef,$self->{pool});
2698}
2699
2700sub delete_entry {
2701        my ($self, $path, $pbat) = @_;
2702        my $rpath = $self->repo_path($path);
2703        my ($dir, $file) = split_path($rpath);
2704        $self->{rm}->{$dir} = 1;
2705        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2706}
2707
2708sub R {
2709        my ($self, $m) = @_;
2710        my ($dir, $file) = split_path($m->{file_b});
2711        my $pbat = $self->ensure_path($dir);
2712        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2713                                $self->url_path($m->{file_a}), $self->{r});
2714        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2715        $self->chg_file($fbat, $m);
2716        $self->close_file($fbat,undef,$self->{pool});
2717
2718        ($dir, $file) = split_path($m->{file_a});
2719        $pbat = $self->ensure_path($dir);
2720        $self->delete_entry($m->{file_a}, $pbat);
2721}
2722
2723sub M {
2724        my ($self, $m) = @_;
2725        my ($dir, $file) = split_path($m->{file_b});
2726        my $pbat = $self->ensure_path($dir);
2727        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2728                                $pbat,$self->{r},$self->{pool});
2729        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2730        $self->chg_file($fbat, $m);
2731        $self->close_file($fbat,undef,$self->{pool});
2732}
2733
2734sub T { shift->M(@_) }
2735
2736sub change_file_prop {
2737        my ($self, $fbat, $pname, $pval) = @_;
2738        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2739}
2740
2741sub chg_file {
2742        my ($self, $fbat, $m) = @_;
2743        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2744                $self->change_file_prop($fbat,'svn:executable','*');
2745        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2746                $self->change_file_prop($fbat,'svn:executable',undef);
2747        }
2748        my $fh = IO::File->new_tmpfile or croak $!;
2749        if ($m->{mode_b} =~ /^120/) {
2750                print $fh 'link ' or croak $!;
2751                $self->change_file_prop($fbat,'svn:special','*');
2752        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2753                $self->change_file_prop($fbat,'svn:special',undef);
2754        }
2755        defined(my $pid = fork) or croak $!;
2756        if (!$pid) {
2757                open STDOUT, '>&', $fh or croak $!;
2758                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2759        }
2760        waitpid $pid, 0;
2761        croak $? if $?;
2762        $fh->flush == 0 or croak $!;
2763        seek $fh, 0, 0 or croak $!;
2764
2765        my $md5 = Digest::MD5->new;
2766        $md5->addfile($fh) or croak $!;
2767        seek $fh, 0, 0 or croak $!;
2768
2769        my $exp = $md5->hexdigest;
2770        my $pool = SVN::Pool->new;
2771        my $atd = $self->apply_textdelta($fbat, undef, $pool);
2772        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2773        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2774        $pool->clear;
2775
2776        close $fh or croak $!;
2777}
2778
2779sub D {
2780        my ($self, $m) = @_;
2781        my ($dir, $file) = split_path($m->{file_b});
2782        my $pbat = $self->ensure_path($dir);
2783        print "\tD\t$m->{file_b}\n" unless $::_q;
2784        $self->delete_entry($m->{file_b}, $pbat);
2785}
2786
2787sub close_edit {
2788        my ($self) = @_;
2789        my ($p,$bat) = ($self->{pool}, $self->{bat});
2790        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2791                $self->close_directory($bat->{$_}, $p);
2792        }
2793        $self->SUPER::close_edit($p);
2794        $p->clear;
2795}
2796
2797sub abort_edit {
2798        my ($self) = @_;
2799        $self->SUPER::abort_edit($self->{pool});
2800}
2801
2802sub DESTROY {
2803        my $self = shift;
2804        $self->SUPER::DESTROY(@_);
2805        $self->{pool}->clear;
2806}
2807
2808# this drives the editor
2809sub apply_diff {
2810        my ($self) = @_;
2811        my $mods = $self->{mods};
2812        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2813        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2814                my $f = $m->{chg};
2815                if (defined $o{$f}) {
2816                        $self->$f($m);
2817                } else {
2818                        fatal("Invalid change type: $f\n");
2819                }
2820        }
2821        $self->rmdirs if $_rmdir;
2822        if (@$mods == 0) {
2823                $self->abort_edit;
2824        } else {
2825                $self->close_edit;
2826        }
2827        return scalar @$mods;
2828}
2829
2830package Git::SVN::Ra;
2831use vars qw/@ISA $config_dir $_log_window_size/;
2832use strict;
2833use warnings;
2834my ($can_do_switch);
2835my $RA;
2836
2837BEGIN {
2838        # enforce temporary pool usage for some simple functions
2839        my $e;
2840        foreach (qw/get_latest_revnum get_uuid get_repos_root/) {
2841                $e .= "sub $_ {
2842                        my \$self = shift;
2843                        my \$pool = SVN::Pool->new;
2844                        my \@ret = \$self->SUPER::$_(\@_,\$pool);
2845                        \$pool->clear;
2846                        wantarray ? \@ret : \$ret[0]; }\n";
2847        }
2848
2849        # get_dir needs $pool held in cache for dirents to work,
2850        # check_path is cacheable and rev_proplist is close enough
2851        # for our purposes.
2852        foreach (qw/check_path get_dir rev_proplist/) {
2853                $e .= "my \%${_}_cache; my \$${_}_rev = 0; sub $_ {
2854                        my \$self = shift;
2855                        my \$r = pop;
2856                        my \$k = join(\"\\0\", \@_);
2857                        if (my \$x = \$${_}_cache{\$r}->{\$k}) {
2858                                return wantarray ? \@\$x : \$x->[0];
2859                        }
2860                        my \$pool = SVN::Pool->new;
2861                        my \@ret = \$self->SUPER::$_(\@_, \$r, \$pool);
2862                        if (\$r != \$${_}_rev) {
2863                                \%${_}_cache = ( pool => [] );
2864                                \$${_}_rev = \$r;
2865                        }
2866                        \$${_}_cache{\$r}->{\$k} = \\\@ret;
2867                        push \@{\$${_}_cache{pool}}, \$pool;
2868                        wantarray ? \@ret : \$ret[0]; }\n";
2869        }
2870        $e .= "\n1;";
2871        eval $e or die $@;
2872}
2873
2874sub new {
2875        my ($class, $url) = @_;
2876        $url =~ s!/+$!!;
2877        return $RA if ($RA && $RA->{url} eq $url);
2878
2879        SVN::_Core::svn_config_ensure($config_dir, undef);
2880        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2881            SVN::Client::get_simple_provider(),
2882            SVN::Client::get_ssl_server_trust_file_provider(),
2883            SVN::Client::get_simple_prompt_provider(
2884              \&Git::SVN::Prompt::simple, 2),
2885            SVN::Client::get_ssl_client_cert_prompt_provider(
2886              \&Git::SVN::Prompt::ssl_client_cert, 2),
2887            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2888              \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2889            SVN::Client::get_username_provider(),
2890            SVN::Client::get_ssl_server_trust_prompt_provider(
2891              \&Git::SVN::Prompt::ssl_server_trust),
2892            SVN::Client::get_username_prompt_provider(
2893              \&Git::SVN::Prompt::username, 2),
2894          ]);
2895        my $config = SVN::Core::config_get_config($config_dir);
2896        my $self = SVN::Ra->new(url => $url, auth => $baton,
2897                              config => $config,
2898                              pool => SVN::Pool->new,
2899                              auth_provider_callbacks => $callbacks);
2900        $self->{svn_path} = $url;
2901        $self->{repos_root} = $self->get_repos_root;
2902        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
2903        $RA = bless $self, $class;
2904}
2905
2906sub DESTROY {
2907        # do not call the real DESTROY since we store ourselves in $RA
2908}
2909
2910sub get_log {
2911        my ($self, @args) = @_;
2912        my $pool = SVN::Pool->new;
2913        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2914        my $ret = $self->SUPER::get_log(@args, $pool);
2915        $pool->clear;
2916        $ret;
2917}
2918
2919sub get_commit_editor {
2920        my ($self, $log, $cb, $pool) = @_;
2921        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2922        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2923}
2924
2925sub gs_do_update {
2926        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2927        my $new = ($rev_a == $rev_b);
2928        my $path = $gs->{path};
2929
2930        my $pool = SVN::Pool->new;
2931        $editor->set_path_strip($path);
2932        my (@pc) = split m#/#, $path;
2933        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2934                                        1, $editor, $pool);
2935        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2936
2937        # Since we can't rely on svn_ra_reparent being available, we'll
2938        # just have to do some magic with set_path to make it so
2939        # we only want a partial path.
2940        my $sp = '';
2941        my $final = join('/', @pc);
2942        while (@pc) {
2943                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2944                $sp .= '/' if length $sp;
2945                $sp .= shift @pc;
2946        }
2947        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2948
2949        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2950
2951        $reporter->finish_report($pool);
2952        $pool->clear;
2953        $editor->{git_commit_ok};
2954}
2955
2956# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2957# svn_ra_reparent didn't work before 1.4)
2958sub gs_do_switch {
2959        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
2960        my $path = $gs->{path};
2961        my $pool = SVN::Pool->new;
2962
2963        my $full_url = $self->{url};
2964        my $old_url = $full_url;
2965        $full_url .= "/$path" if length $path;
2966        my ($ra, $reparented);
2967        if ($old_url ne $full_url) {
2968                if ($old_url !~ m#^svn(\+ssh)?://#) {
2969                        SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2970                                                  $pool);
2971                        $self->{url} = $full_url;
2972                        $reparented = 1;
2973                } else {
2974                        $ra = Git::SVN::Ra->new($full_url);
2975                }
2976        }
2977        $ra ||= $self;
2978        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
2979        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2980        $reporter->set_path('', $rev_a, 0, @lock, $pool);
2981        $reporter->finish_report($pool);
2982
2983        if ($reparented) {
2984                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2985                $self->{url} = $old_url;
2986        }
2987
2988        $pool->clear;
2989        $editor->{git_commit_ok};
2990}
2991
2992sub gs_fetch_loop_common {
2993        my ($self, $base, $head, $gsv, $globs) = @_;
2994        return if ($base > $head);
2995        my $inc = $_log_window_size;
2996        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2997        my %common;
2998        my $common_max = scalar @$gsv;
2999
3000        foreach my $gs (@$gsv) {
3001                my @tmp = split m#/#, $gs->{path};
3002                my $p = '';
3003                foreach (@tmp) {
3004                        $p .= length($p) ? "/$_" : $_;
3005                        $common{$p} ||= 0;
3006                        $common{$p}++;
3007                }
3008        }
3009        $globs ||= [];
3010        $common_max += scalar @$globs;
3011        foreach my $glob (@$globs) {
3012                my @tmp = split m#/#, $glob->{path}->{left};
3013                my $p = '';
3014                foreach (@tmp) {
3015                        $p .= length($p) ? "/$_" : $_;
3016                        $common{$p} ||= 0;
3017                        $common{$p}++;
3018                }
3019        }
3020
3021        my $longest_path = '';
3022        foreach (sort {length $b <=> length $a} keys %common) {
3023                if ($common{$_} == $common_max) {
3024                        $longest_path = $_;
3025                        last;
3026                }
3027        }
3028        while (1) {
3029                my %revs;
3030                my $err;
3031                my $err_handler = $SVN::Error::handler;
3032                $SVN::Error::handler = sub {
3033                        ($err) = @_;
3034                        skip_unknown_revs($err);
3035                };
3036                sub _cb {
3037                        my ($paths, $r, $author, $date, $log) = @_;
3038                        [ dup_changed_paths($paths),
3039                          { author => $author, date => $date, log => $log } ];
3040                }
3041                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3042                               sub { $revs{$_[1]} = _cb(@_) });
3043                if ($err && $max >= $head) {
3044                        print STDERR "Path '$longest_path' ",
3045                                     "was probably deleted:\n",
3046                                     $err->expanded_message,
3047                                     "\nWill attempt to follow ",
3048                                     "revisions r$min .. r$max ",
3049                                     "committed before the deletion\n";
3050                        my $hi = $max;
3051                        while (--$hi >= $min) {
3052                                my $ok;
3053                                $self->get_log([$longest_path], $min, $hi,
3054                                               0, 1, 1, sub {
3055                                               $ok ||= $_[1];
3056                                               $revs{$_[1]} = _cb(@_) });
3057                                if ($ok) {
3058                                        print STDERR "r$min .. r$ok OK\n";
3059                                        last;
3060                                }
3061                        }
3062                }
3063                $SVN::Error::handler = $err_handler;
3064
3065                my %exists = map { $_->{path} => $_ } @$gsv;
3066                foreach my $r (sort {$a <=> $b} keys %revs) {
3067                        my ($paths, $logged) = @{$revs{$r}};
3068
3069                        foreach my $gs ($self->match_globs(\%exists, $paths,
3070                                                           $globs, $r)) {
3071                                if ($gs->rev_db_max >= $r) {
3072                                        next;
3073                                }
3074                                next unless $gs->match_paths($paths, $r);
3075                                $gs->{logged_rev_props} = $logged;
3076                                if (my $last_commit = $gs->last_commit) {
3077                                        $gs->assert_index_clean($last_commit);
3078                                }
3079                                my $log_entry = $gs->do_fetch($paths, $r);
3080                                if ($log_entry) {
3081                                        $gs->do_git_commit($log_entry);
3082                                }
3083                        }
3084                        foreach my $g (@$globs) {
3085                                my $k = "svn-remote.$g->{remote}." .
3086                                        "$g->{t}-maxRev";
3087                                Git::SVN::tmp_config($k, $r);
3088                        }
3089                }
3090                # pre-fill the .rev_db since it'll eventually get filled in
3091                # with '0' x40 if something new gets committed
3092                foreach my $gs (@$gsv) {
3093                        next if defined $gs->rev_db_get($max);
3094                        $gs->rev_db_set($max, 0 x40);
3095                }
3096                foreach my $g (@$globs) {
3097                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3098                        Git::SVN::tmp_config($k, $max);
3099                }
3100                last if $max >= $head;
3101                $min = $max + 1;
3102                $max += $inc;
3103                $max = $head if ($max > $head);
3104        }
3105}
3106
3107sub match_globs {
3108        my ($self, $exists, $paths, $globs, $r) = @_;
3109
3110        sub get_dir_check {
3111                my ($self, $exists, $g, $r) = @_;
3112                my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3113                return unless scalar @x == 3;
3114                my $dirents = $x[0];
3115                foreach my $de (keys %$dirents) {
3116                        next if $dirents->{$de}->kind != $SVN::Node::dir;
3117                        my $p = $g->{path}->full_path($de);
3118                        next if $exists->{$p};
3119                        next if (length $g->{path}->{right} &&
3120                                 ($self->check_path($p, $r) !=
3121                                  $SVN::Node::dir));
3122                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3123                                         $g->{ref}->full_path($de), 1);
3124                }
3125        }
3126        foreach my $g (@$globs) {
3127                if (my $path = $paths->{"/$g->{path}->{left}"}) {
3128                        if ($path->{action} =~ /^[AR]$/) {
3129                                get_dir_check($self, $exists, $g, $r);
3130                        }
3131                }
3132                foreach (keys %$paths) {
3133                        if (/$g->{path}->{left_regex}/ &&
3134                            !/$g->{path}->{regex}/) {
3135                                next if $paths->{$_}->{action} !~ /^[AR]$/;
3136                                get_dir_check($self, $exists, $g, $r);
3137                        }
3138                        next unless /$g->{path}->{regex}/;
3139                        my $p = $1;
3140                        my $pathname = $g->{path}->full_path($p);
3141                        next if $exists->{$pathname};
3142                        $exists->{$pathname} = Git::SVN->init(
3143                                              $self->{url}, $pathname, undef,
3144                                              $g->{ref}->full_path($p), 1);
3145                }
3146                my $c = '';
3147                foreach (split m#/#, $g->{path}->{left}) {
3148                        $c .= "/$_";
3149                        next unless ($paths->{$c} &&
3150                                     ($paths->{$c}->{action} =~ /^[AR]$/));
3151                        get_dir_check($self, $exists, $g, $r);
3152                }
3153        }
3154        values %$exists;
3155}
3156
3157sub minimize_url {
3158        my ($self) = @_;
3159        return $self->{url} if ($self->{url} eq $self->{repos_root});
3160        my $url = $self->{repos_root};
3161        my @components = split(m!/!, $self->{svn_path});
3162        my $c = '';
3163        do {
3164                $url .= "/$c" if length $c;
3165                eval { (ref $self)->new($url)->get_latest_revnum };
3166        } while ($@ && ($c = shift @components));
3167        $url;
3168}
3169
3170sub can_do_switch {
3171        my $self = shift;
3172        unless (defined $can_do_switch) {
3173                my $pool = SVN::Pool->new;
3174                my $rep = eval {
3175                        $self->do_switch(1, '', 0, $self->{url},
3176                                         SVN::Delta::Editor->new, $pool);
3177                };
3178                if ($@) {
3179                        $can_do_switch = 0;
3180                } else {
3181                        $rep->abort_report($pool);
3182                        $can_do_switch = 1;
3183                }
3184                $pool->clear;
3185        }
3186        $can_do_switch;
3187}
3188
3189sub skip_unknown_revs {
3190        my ($err) = @_;
3191        my $errno = $err->apr_err();
3192        # Maybe the branch we're tracking didn't
3193        # exist when the repo started, so it's
3194        # not an error if it doesn't, just continue
3195        #
3196        # Wonderfully consistent library, eh?
3197        # 160013 - svn:// and file://
3198        # 175002 - http(s)://
3199        # 175007 - http(s):// (this repo required authorization, too...)
3200        #   More codes may be discovered later...
3201        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3202                warn "W: Ignoring error from SVN, path probably ",
3203                     "does not exist: ($errno): ",
3204                     $err->expanded_message,"\n";
3205                return;
3206        }
3207        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3208}
3209
3210# svn_log_changed_path_t objects passed to get_log are likely to be
3211# overwritten even if only the refs are copied to an external variable,
3212# so we should dup the structures in their entirety.  Using an externally
3213# passed pool (instead of our temporary and quickly cleared pool in
3214# Git::SVN::Ra) does not help matters at all...
3215sub dup_changed_paths {
3216        my ($paths) = @_;
3217        return undef unless $paths;
3218        my %ret;
3219        foreach my $p (keys %$paths) {
3220                my $i = $paths->{$p};
3221                my %s = map { $_ => $i->$_ }
3222                              qw/copyfrom_path copyfrom_rev action/;
3223                $ret{$p} = \%s;
3224        }
3225        \%ret;
3226}
3227
3228package Git::SVN::Log;
3229use strict;
3230use warnings;
3231use POSIX qw/strftime/;
3232use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3233            %rusers $show_commit $incremental/;
3234my $l_fmt;
3235
3236sub cmt_showable {
3237        my ($c) = @_;
3238        return 1 if defined $c->{r};
3239        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3240                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3241                my @log = command(qw/cat-file commit/, $c->{c});
3242                shift @log while ($log[0] ne "\n");
3243                shift @log;
3244                @{$c->{l}} = grep !/^git-svn-id: /, @log;
3245
3246                (undef, $c->{r}, undef) = ::extract_metadata(
3247                                (grep(/^git-svn-id: /, @log))[-1]);
3248        }
3249        return defined $c->{r};
3250}
3251
3252sub log_use_color {
3253        return 1 if $color;
3254        my ($dc, $dcvar);
3255        $dcvar = 'color.diff';
3256        $dc = `git-config --get $dcvar`;
3257        if ($dc eq '') {
3258                # nothing at all; fallback to "diff.color"
3259                $dcvar = 'diff.color';
3260                $dc = `git-config --get $dcvar`;
3261        }
3262        chomp($dc);
3263        if ($dc eq 'auto') {
3264                my $pc;
3265                $pc = `git-config --get color.pager`;
3266                if ($pc eq '') {
3267                        # does not have it -- fallback to pager.color
3268                        $pc = `git-config --bool --get pager.color`;
3269                }
3270                else {
3271                        $pc = `git-config --bool --get color.pager`;
3272                        if ($?) {
3273                                $pc = 'false';
3274                        }
3275                }
3276                chomp($pc);
3277                if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3278                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3279                }
3280                return 0;
3281        }
3282        return 0 if $dc eq 'never';
3283        return 1 if $dc eq 'always';
3284        chomp($dc = `git-config --bool --get $dcvar`);
3285        return ($dc eq 'true');
3286}
3287
3288sub git_svn_log_cmd {
3289        my ($r_min, $r_max, @args) = @_;
3290        my $head = 'HEAD';
3291        foreach my $x (@args) {
3292                last if $x eq '--';
3293                next unless ::verify_ref("$x^0");
3294                $head = $x;
3295                last;
3296        }
3297
3298        my $url = (::working_head_info($head))[0];
3299        my $gs = Git::SVN->find_by_url($url) || Git::SVN->_new;
3300        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3301                   $gs->refname);
3302        push @cmd, '-r' unless $non_recursive;
3303        push @cmd, qw/--raw --name-status/ if $verbose;
3304        push @cmd, '--color' if log_use_color();
3305        return @cmd unless defined $r_max;
3306        if ($r_max == $r_min) {
3307                push @cmd, '--max-count=1';
3308                if (my $c = $gs->rev_db_get($r_max)) {
3309                        push @cmd, $c;
3310                }
3311        } else {
3312                my ($c_min, $c_max);
3313                $c_max = $gs->rev_db_get($r_max);
3314                $c_min = $gs->rev_db_get($r_min);
3315                if (defined $c_min && defined $c_max) {
3316                        if ($r_max > $r_max) {
3317                                push @cmd, "$c_min..$c_max";
3318                        } else {
3319                                push @cmd, "$c_max..$c_min";
3320                        }
3321                } elsif ($r_max > $r_min) {
3322                        push @cmd, $c_max;
3323                } else {
3324                        push @cmd, $c_min;
3325                }
3326        }
3327        return @cmd;
3328}
3329
3330# adapted from pager.c
3331sub config_pager {
3332        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3333        if (!defined $pager) {
3334                $pager = 'less';
3335        } elsif (length $pager == 0 || $pager eq 'cat') {
3336                $pager = undef;
3337        }
3338}
3339
3340sub run_pager {
3341        return unless -t *STDOUT;
3342        pipe my $rfd, my $wfd or return;
3343        defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
3344        if (!$pid) {
3345                open STDOUT, '>&', $wfd or
3346                                     ::fatal "Can't redirect to stdout: $!\n";
3347                return;
3348        }
3349        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
3350        $ENV{LESS} ||= 'FRSX';
3351        exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
3352}
3353
3354sub tz_to_s_offset {
3355        my ($tz) = @_;
3356        $tz =~ s/(\d\d)$//;
3357        return ($1 * 60) + ($tz * 3600);
3358}
3359
3360sub get_author_info {
3361        my ($dest, $author, $t, $tz) = @_;
3362        $author =~ s/(?:^\s*|\s*$)//g;
3363        $dest->{a_raw} = $author;
3364        my $au;
3365        if ($::_authors) {
3366                $au = $rusers{$author} || undef;
3367        }
3368        if (!$au) {
3369                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3370        }
3371        $dest->{t} = $t;
3372        $dest->{tz} = $tz;
3373        $dest->{a} = $au;
3374        # Date::Parse isn't in the standard Perl distro :(
3375        if ($tz =~ s/^\+//) {
3376                $t += tz_to_s_offset($tz);
3377        } elsif ($tz =~ s/^\-//) {
3378                $t -= tz_to_s_offset($tz);
3379        }
3380        $dest->{t_utc} = $t;
3381}
3382
3383sub process_commit {
3384        my ($c, $r_min, $r_max, $defer) = @_;
3385        if (defined $r_min && defined $r_max) {
3386                if ($r_min == $c->{r} && $r_min == $r_max) {
3387                        show_commit($c);
3388                        return 0;
3389                }
3390                return 1 if $r_min == $r_max;
3391                if ($r_min < $r_max) {
3392                        # we need to reverse the print order
3393                        return 0 if (defined $limit && --$limit < 0);
3394                        push @$defer, $c;
3395                        return 1;
3396                }
3397                if ($r_min != $r_max) {
3398                        return 1 if ($r_min < $c->{r});
3399                        return 1 if ($r_max > $c->{r});
3400                }
3401        }
3402        return 0 if (defined $limit && --$limit < 0);
3403        show_commit($c);
3404        return 1;
3405}
3406
3407sub show_commit {
3408        my $c = shift;
3409        if ($oneline) {
3410                my $x = "\n";
3411                if (my $l = $c->{l}) {
3412                        while ($l->[0] =~ /^\s*$/) { shift @$l }
3413                        $x = $l->[0];
3414                }
3415                $l_fmt ||= 'A' . length($c->{r});
3416                print 'r',pack($l_fmt, $c->{r}),' | ';
3417                print "$c->{c} | " if $show_commit;
3418                print $x;
3419        } else {
3420                show_commit_normal($c);
3421        }
3422}
3423
3424sub show_commit_changed_paths {
3425        my ($c) = @_;
3426        return unless $c->{changed};
3427        print "Changed paths:\n", @{$c->{changed}};
3428}
3429
3430sub show_commit_normal {
3431        my ($c) = @_;
3432        print '-' x72, "\nr$c->{r} | ";
3433        print "$c->{c} | " if $show_commit;
3434        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3435                                 localtime($c->{t_utc})), ' | ';
3436        my $nr_line = 0;
3437
3438        if (my $l = $c->{l}) {
3439                while ($l->[$#$l] eq "\n" && $#$l > 0
3440                                          && $l->[($#$l - 1)] eq "\n") {
3441                        pop @$l;
3442                }
3443                $nr_line = scalar @$l;
3444                if (!$nr_line) {
3445                        print "1 line\n\n\n";
3446                } else {
3447                        if ($nr_line == 1) {
3448                                $nr_line = '1 line';
3449                        } else {
3450                                $nr_line .= ' lines';
3451                        }
3452                        print $nr_line, "\n";
3453                        show_commit_changed_paths($c);
3454                        print "\n";
3455                        print $_ foreach @$l;
3456                }
3457        } else {
3458                print "1 line\n";
3459                show_commit_changed_paths($c);
3460                print "\n";
3461
3462        }
3463        foreach my $x (qw/raw stat diff/) {
3464                if ($c->{$x}) {
3465                        print "\n";
3466                        print $_ foreach @{$c->{$x}}
3467                }
3468        }
3469}
3470
3471sub cmd_show_log {
3472        my (@args) = @_;
3473        my ($r_min, $r_max);
3474        my $r_last = -1; # prevent dupes
3475        if (defined $TZ) {
3476                $ENV{TZ} = $TZ;
3477        } else {
3478                delete $ENV{TZ};
3479        }
3480        if (defined $::_revision) {
3481                if ($::_revision =~ /^(\d+):(\d+)$/) {
3482                        ($r_min, $r_max) = ($1, $2);
3483                } elsif ($::_revision =~ /^\d+$/) {
3484                        $r_min = $r_max = $::_revision;
3485                } else {
3486                        ::fatal "-r$::_revision is not supported, use ",
3487                                "standard \'git log\' arguments instead\n";
3488                }
3489        }
3490
3491        config_pager();
3492        @args = (git_svn_log_cmd($r_min, $r_max, @args), @args);
3493        my $log = command_output_pipe(@args);
3494        run_pager();
3495        my (@k, $c, $d, $stat);
3496        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3497        while (<$log>) {
3498                if (/^${esc_color}commit ($::sha1_short)/o) {
3499                        my $cmt = $1;
3500                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3501                                $r_last = $c->{r};
3502                                process_commit($c, $r_min, $r_max, \@k) or
3503                                                                goto out;
3504                        }
3505                        $d = undef;
3506                        $c = { c => $cmt };
3507                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3508                        get_author_info($c, $1, $2, $3);
3509                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3510                        # ignore
3511                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3512                        push @{$c->{raw}}, $_;
3513                } elsif (/^${esc_color}[ACRMDT]\t/) {
3514                        # we could add $SVN->{svn_path} here, but that requires
3515                        # remote access at the moment (repo_path_split)...
3516                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
3517                        push @{$c->{changed}}, $_;
3518                } elsif (/^${esc_color}diff /o) {
3519                        $d = 1;
3520                        push @{$c->{diff}}, $_;
3521                } elsif ($d) {
3522                        push @{$c->{diff}}, $_;
3523                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
3524                          $esc_color*[\+\-]*$esc_color$/x) {
3525                        $stat = 1;
3526                        push @{$c->{stat}}, $_;
3527                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
3528                        push @{$c->{stat}}, $_;
3529                        $stat = undef;
3530                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
3531                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3532                } elsif (s/^${esc_color}    //o) {
3533                        push @{$c->{l}}, $_;
3534                }
3535        }
3536        if ($c && defined $c->{r} && $c->{r} != $r_last) {
3537                $r_last = $c->{r};
3538                process_commit($c, $r_min, $r_max, \@k);
3539        }
3540        if (@k) {
3541                my $swap = $r_max;
3542                $r_max = $r_min;
3543                $r_min = $swap;
3544                process_commit($_, $r_min, $r_max) foreach reverse @k;
3545        }
3546out:
3547        close $log;
3548        print '-' x72,"\n" unless $incremental || $oneline;
3549}
3550
3551package Git::SVN::Migration;
3552# these version numbers do NOT correspond to actual version numbers
3553# of git nor git-svn.  They are just relative.
3554#
3555# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3556#
3557# v1 layout: .git/$id/info/url, refs/remotes/$id
3558#
3559# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3560#
3561# v3 layout: .git/svn/$id, refs/remotes/$id
3562#            - info/url may remain for backwards compatibility
3563#            - this is what we migrate up to this layout automatically,
3564#            - this will be used by git svn init on single branches
3565# v3.1 layout (auto migrated):
3566#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
3567#              for backwards compatibility
3568#
3569# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3570#            - this is only created for newly multi-init-ed
3571#              repositories.  Similar in spirit to the
3572#              --use-separate-remotes option in git-clone (now default)
3573#            - we do not automatically migrate to this (following
3574#              the example set by core git)
3575use strict;
3576use warnings;
3577use Carp qw/croak/;
3578use File::Path qw/mkpath/;
3579use File::Basename qw/dirname basename/;
3580use vars qw/$_minimize/;
3581
3582sub migrate_from_v0 {
3583        my $git_dir = $ENV{GIT_DIR};
3584        return undef unless -d $git_dir;
3585        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3586        my $migrated = 0;
3587        while (<$fh>) {
3588                chomp;
3589                my ($id, $orig_ref) = ($_, $_);
3590                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3591                next unless -f "$git_dir/$id/info/url";
3592                my $new_ref = "refs/remotes/$id";
3593                if (::verify_ref("$new_ref^0")) {
3594                        print STDERR "W: $orig_ref is probably an old ",
3595                                     "branch used by an ancient version of ",
3596                                     "git-svn.\n",
3597                                     "However, $new_ref also exists.\n",
3598                                     "We will not be able ",
3599                                     "to use this branch until this ",
3600                                     "ambiguity is resolved.\n";
3601                        next;
3602                }
3603                print STDERR "Migrating from v0 layout...\n" if !$migrated;
3604                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3605                command_noisy('update-ref', $new_ref, $orig_ref);
3606                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3607                $migrated++;
3608        }
3609        command_close_pipe($fh, $ctx);
3610        print STDERR "Done migrating from v0 layout...\n" if $migrated;
3611        $migrated;
3612}
3613
3614sub migrate_from_v1 {
3615        my $git_dir = $ENV{GIT_DIR};
3616        my $migrated = 0;
3617        return $migrated unless -d $git_dir;
3618        my $svn_dir = "$git_dir/svn";
3619
3620        # just in case somebody used 'svn' as their $id at some point...
3621        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3622
3623        print STDERR "Migrating from a git-svn v1 layout...\n";
3624        mkpath([$svn_dir]);
3625        print STDERR "Data from a previous version of git-svn exists, but\n\t",
3626                     "$svn_dir\n\t(required for this version ",
3627                     "($::VERSION) of git-svn) does not. exist\n";
3628        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3629        while (<$fh>) {
3630                my $x = $_;
3631                next unless $x =~ s#^refs/remotes/##;
3632                chomp $x;
3633                next unless -f "$git_dir/$x/info/url";
3634                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3635                next unless $u;
3636                my $dn = dirname("$git_dir/svn/$x");
3637                mkpath([$dn]) unless -d $dn;
3638                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3639                        mkpath(["$git_dir/svn/svn"]);
3640                        print STDERR " - $git_dir/$x/info => ",
3641                                        "$git_dir/svn/$x/info\n";
3642                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3643                               croak "$!: $x";
3644                        # don't worry too much about these, they probably
3645                        # don't exist with repos this old (save for index,
3646                        # and we can easily regenerate that)
3647                        foreach my $f (qw/unhandled.log index .rev_db/) {
3648                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3649                        }
3650                } else {
3651                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3652                        rename "$git_dir/$x", "$git_dir/svn/$x" or
3653                               croak "$!: $x";
3654                }
3655                $migrated++;
3656        }
3657        command_close_pipe($fh, $ctx);
3658        print STDERR "Done migrating from a git-svn v1 layout\n";
3659        $migrated;
3660}
3661
3662sub read_old_urls {
3663        my ($l_map, $pfx, $path) = @_;
3664        my @dir;
3665        foreach (<$path/*>) {
3666                if (-r "$_/info/url") {
3667                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3668                        my $ref_id = $pfx . basename $_;
3669                        my $url = ::file_to_s("$_/info/url");
3670                        $l_map->{$ref_id} = $url;
3671                } elsif (-d $_) {
3672                        push @dir, $_;
3673                }
3674        }
3675        foreach (@dir) {
3676                my $x = $_;
3677                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3678                read_old_urls($l_map, $x, $_);
3679        }
3680}
3681
3682sub migrate_from_v2 {
3683        my @cfg = command(qw/config -l/);
3684        return if grep /^svn-remote\..+\.url=/, @cfg;
3685        my %l_map;
3686        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3687        my $migrated = 0;
3688
3689        foreach my $ref_id (sort keys %l_map) {
3690                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3691                if ($@) {
3692                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3693                }
3694                $migrated++;
3695        }
3696        $migrated;
3697}
3698
3699sub minimize_connections {
3700        my $r = Git::SVN::read_all_remotes();
3701        my $new_urls = {};
3702        my $root_repos = {};
3703        foreach my $repo_id (keys %$r) {
3704                my $url = $r->{$repo_id}->{url} or next;
3705                my $fetch = $r->{$repo_id}->{fetch} or next;
3706                my $ra = Git::SVN::Ra->new($url);
3707
3708                # skip existing cases where we already connect to the root
3709                if (($ra->{url} eq $ra->{repos_root}) ||
3710                    (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3711                     $repo_id)) {
3712                        $root_repos->{$ra->{url}} = $repo_id;
3713                        next;
3714                }
3715
3716                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3717                my $root_path = $ra->{url};
3718                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3719                foreach my $path (keys %$fetch) {
3720                        my $ref_id = $fetch->{$path};
3721                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3722
3723                        # make sure we can read when connecting to
3724                        # a higher level of a repository
3725                        my ($last_rev, undef) = $gs->last_rev_commit;
3726                        if (!defined $last_rev) {
3727                                $last_rev = eval {
3728                                        $root_ra->get_latest_revnum;
3729                                };
3730                                next if $@;
3731                        }
3732                        my $new = $root_path;
3733                        $new .= length $path ? "/$path" : '';
3734                        eval {
3735                                $root_ra->get_log([$new], $last_rev, $last_rev,
3736                                                  0, 0, 1, sub { });
3737                        };
3738                        next if $@;
3739                        $new_urls->{$ra->{repos_root}}->{$new} =
3740                                { ref_id => $ref_id,
3741                                  old_repo_id => $repo_id,
3742                                  old_path => $path };
3743                }
3744        }
3745
3746        my @emptied;
3747        foreach my $url (keys %$new_urls) {
3748                # see if we can re-use an existing [svn-remote "repo_id"]
3749                # instead of creating a(n ugly) new section:
3750                my $repo_id = $root_repos->{$url} ||
3751                              Git::SVN::sanitize_remote_name($url);
3752
3753                my $fetch = $new_urls->{$url};
3754                foreach my $path (keys %$fetch) {
3755                        my $x = $fetch->{$path};
3756                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3757                        my $pfx = "svn-remote.$x->{old_repo_id}";
3758
3759                        my $old_fetch = quotemeta("$x->{old_path}:".
3760                                                  "refs/remotes/$x->{ref_id}");
3761                        command_noisy(qw/config --unset/,
3762                                      "$pfx.fetch", '^'. $old_fetch . '$');
3763                        delete $r->{$x->{old_repo_id}}->
3764                               {fetch}->{$x->{old_path}};
3765                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3766                                command_noisy(qw/config --unset/,
3767                                              "$pfx.url");
3768                                push @emptied, $x->{old_repo_id}
3769                        }
3770                }
3771        }
3772        if (@emptied) {
3773                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3774                           "$ENV{GIT_DIR}/config";
3775                print STDERR <<EOF;
3776The following [svn-remote] sections in your config file ($file) are empty
3777and can be safely removed:
3778EOF
3779                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3780        }
3781}
3782
3783sub migration_check {
3784        migrate_from_v0();
3785        migrate_from_v1();
3786        migrate_from_v2();
3787        minimize_connections() if $_minimize;
3788}
3789
3790package Git::IndexInfo;
3791use strict;
3792use warnings;
3793use Git qw/command_input_pipe command_close_pipe/;
3794
3795sub new {
3796        my ($class) = @_;
3797        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3798        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3799}
3800
3801sub remove {
3802        my ($self, $path) = @_;
3803        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3804                return ++$self->{nr};
3805        }
3806        undef;
3807}
3808
3809sub update {
3810        my ($self, $mode, $hash, $path) = @_;
3811        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3812                return ++$self->{nr};
3813        }
3814        undef;
3815}
3816
3817sub DESTROY {
3818        my ($self) = @_;
3819        command_close_pipe($self->{gui}, $self->{ctx});
3820}
3821
3822package Git::SVN::GlobSpec;
3823use strict;
3824use warnings;
3825
3826sub new {
3827        my ($class, $glob) = @_;
3828        my $re = $glob;
3829        $re =~ s!/+$!!g; # no need for trailing slashes
3830        my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
3831        my ($left, $right) = ($1, $2);
3832        if ($nr > 1) {
3833                die "Only one '*' wildcard expansion ",
3834                    "is supported (got $nr): '$glob'\n";
3835        } elsif ($nr == 0) {
3836                die "One '*' is needed for glob: '$glob'\n";
3837        }
3838        $re = quotemeta($left) . $re . quotemeta($right);
3839        if (length $left && !($left =~ s!/+$!!g)) {
3840                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
3841        }
3842        if (length $right && !($right =~ s!^/+!!g)) {
3843                die "Missing leading '/' on right side of: '$glob' ($right)\n";
3844        }
3845        my $left_re = qr/^\/\Q$left\E(\/|$)/;
3846        bless { left => $left, right => $right, left_regex => $left_re,
3847                regex => qr/$re/, glob => $glob }, $class;
3848}
3849
3850sub full_path {
3851        my ($self, $path) = @_;
3852        return (length $self->{left} ? "$self->{left}/" : '') .
3853               $path . (length $self->{right} ? "/$self->{right}" : '');
3854}
3855
3856__END__
3857
3858Data structures:
3859
3860
3861$remotes = { # returned by read_all_remotes()
3862        'svn' => {
3863                # svn-remote.svn.url=https://svn.musicpd.org
3864                url => 'https://svn.musicpd.org',
3865                # svn-remote.svn.fetch=mpd/trunk:trunk
3866                fetch => {
3867                        'mpd/trunk' => 'trunk',
3868                },
3869                # svn-remote.svn.tags=mpd/tags/*:tags/*
3870                tags => {
3871                        path => {
3872                                left => 'mpd/tags',
3873                                right => '',
3874                                regex => qr!mpd/tags/([^/]+)$!,
3875                                glob => 'tags/*',
3876                        },
3877                        ref => {
3878                                left => 'tags',
3879                                right => '',
3880                                regex => qr!tags/([^/]+)$!,
3881                                glob => 'tags/*',
3882                        },
3883                }
3884        }
3885};
3886
3887$log_entry hashref as returned by libsvn_log_entry()
3888{
3889        log => 'whitespace-formatted log entry
3890',                                              # trailing newline is preserved
3891        revision => '8',                        # integer
3892        date => '2004-02-24T17:01:44.108345Z',  # commit date
3893        author => 'committer name'
3894};
3895
3896
3897# this is generated by generate_diff();
3898@mods = array of diff-index line hashes, each element represents one line
3899        of diff-index output
3900
3901diff-index line ($m hash)
3902{
3903        mode_a => first column of diff-index output, no leading ':',
3904        mode_b => second column of diff-index output,
3905        sha1_b => sha1sum of the final blob,
3906        chg => change type [MCRADT],
3907        file_a => original file name of a file (iff chg is 'C' or 'R')
3908        file_b => new/current file name of a file (any chg)
3909}
3910;
3911
3912# retval of read_url_paths{,_all}();
3913$l_map = {
3914        # repository root url
3915        'https://svn.musicpd.org' => {
3916                # repository path               # GIT_SVN_ID
3917                'mpd/trunk'             =>      'trunk',
3918                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3919        },
3920}
3921
3922Notes:
3923        I don't trust the each() function on unless I created %hash myself
3924        because the internal iterator may not have started at base.