Andrew's git - gitweb.git/blob

7563eea35215cd7f14ba5d7b43cf823702a29187
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use warnings;
   5use strict;
   6use vars qw/    $AUTHOR $VERSION
   7                $sha1 $sha1_short $_revision
   8                $_q $_authors %users/;
   9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  10$VERSION = '@@GIT_VERSION@@';
  11
  12$ENV{GIT_DIR} ||= '.git';
  13$Git::SVN::default_repo_id = 'svn';
  14$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  15$Git::SVN::Ra::_log_window_size = 100;
  16
  17$Git::SVN::Log::TZ = $ENV{TZ};
  18$ENV{TZ} = 'UTC';
  19$| = 1; # unbuffer STDOUT
  20
  21sub fatal (@) { print STDERR @_; exit 1 }
  22require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  23require SVN::Ra;
  24require SVN::Delta;
  25if ($SVN::Core::VERSION lt '1.1.0') {
  26        fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
  27}
  28push @Git::SVN::Ra::ISA, 'SVN::Ra';
  29push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  30push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  31use Carp qw/croak/;
  32use IO::File qw//;
  33use File::Basename qw/dirname basename/;
  34use File::Path qw/mkpath/;
  35use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
  36use IPC::Open3;
  37use Git;
  38
  39BEGIN {
  40        my $s;
  41        foreach (qw/command command_oneline command_noisy command_output_pipe
  42                    command_input_pipe command_close_pipe/) {
  43                $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
  44                      "*Git::SVN::Migration::$_ = ".
  45                      "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
  46        }
  47        eval $s;
  48}
  49
  50my ($SVN);
  51
  52$sha1 = qr/[a-f\d]{40}/;
  53$sha1_short = qr/[a-f\d]{4,40}/;
  54my ($_stdin, $_help, $_edit,
  55        $_message, $_file,
  56        $_template, $_shared,
  57        $_version, $_fetch_all,
  58        $_merge, $_strategy, $_dry_run,
  59        $_prefix, $_no_checkout, $_verbose);
  60$Git::SVN::_follow_parent = 1;
  61my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  62                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  63                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
  64my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  65                'authors-file|A=s' => \$_authors,
  66                'repack:i' => \$Git::SVN::_repack,
  67                'noMetadata' => \$Git::SVN::_no_metadata,
  68                'useSvmProps' => \$Git::SVN::_use_svm_props,
  69                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
  70                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
  71                'no-checkout' => \$_no_checkout,
  72                'quiet|q' => \$_q,
  73                'repack-flags|repack-args|repack-opts=s' =>
  74                   \$Git::SVN::_repack_flags,
  75                %remote_opts );
  76
  77my ($_trunk, $_tags, $_branches);
  78my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
  79                  'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
  80                  'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
  81                  %remote_opts );
  82my %cmt_opts = ( 'edit|e' => \$_edit,
  83                'rmdir' => \$SVN::Git::Editor::_rmdir,
  84                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
  85                'l=i' => \$SVN::Git::Editor::_rename_limit,
  86                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
  87);
  88
  89my %cmd = (
  90        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
  91                        { 'revision|r=s' => \$_revision,
  92                          'fetch-all|all' => \$_fetch_all,
  93                           %fc_opts } ],
  94        clone => [ \&cmd_clone, "Initialize and fetch revisions",
  95                        { 'revision|r=s' => \$_revision,
  96                           %fc_opts, %init_opts } ],
  97        init => [ \&cmd_init, "Initialize a repo for tracking" .
  98                          " (requires URL argument)",
  99                          \%init_opts ],
 100        'multi-init' => [ \&cmd_multi_init,
 101                          "Deprecated alias for ".
 102                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 103                          \%init_opts ],
 104        dcommit => [ \&cmd_dcommit,
 105                     'Commit several diffs to merge with upstream',
 106                        { 'merge|m|M' => \$_merge,
 107                          'strategy|s=s' => \$_strategy,
 108                          'verbose|v' => \$_verbose,
 109                          'dry-run|n' => \$_dry_run,
 110                          'fetch-all|all' => \$_fetch_all,
 111                        %cmt_opts, %fc_opts } ],
 112        'set-tree' => [ \&cmd_set_tree,
 113                        "Set an SVN repository to a git tree-ish",
 114                        { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
 115        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 116                        { 'revision|r=i' => \$_revision } ],
 117        'multi-fetch' => [ \&cmd_multi_fetch,
 118                           "Deprecated alias for $0 fetch --all",
 119                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 120        'migrate' => [ sub { },
 121                       # no-op, we automatically run this anyways,
 122                       'Migrate configuration/metadata/layout from
 123                        previous versions of git-svn',
 124                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 125                         %remote_opts } ],
 126        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 127                        { 'limit=i' => \$Git::SVN::Log::limit,
 128                          'revision|r=s' => \$_revision,
 129                          'verbose|v' => \$Git::SVN::Log::verbose,
 130                          'incremental' => \$Git::SVN::Log::incremental,
 131                          'oneline' => \$Git::SVN::Log::oneline,
 132                          'show-commit' => \$Git::SVN::Log::show_commit,
 133                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 134                          'authors-file|A=s' => \$_authors,
 135                          'color' => \$Git::SVN::Log::color,
 136                          'pager=s' => \$Git::SVN::Log::pager,
 137                        } ],
 138        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 139                        { 'merge|m|M' => \$_merge,
 140                          'verbose|v' => \$_verbose,
 141                          'strategy|s=s' => \$_strategy,
 142                          'fetch-all|all' => \$_fetch_all,
 143                          %fc_opts } ],
 144        'commit-diff' => [ \&cmd_commit_diff,
 145                           'Commit a diff between two trees',
 146                        { 'message|m=s' => \$_message,
 147                          'file|F=s' => \$_file,
 148                          'revision|r=s' => \$_revision,
 149                        %cmt_opts } ],
 150);
 151
 152my $cmd;
 153for (my $i = 0; $i < @ARGV; $i++) {
 154        if (defined $cmd{$ARGV[$i]}) {
 155                $cmd = $ARGV[$i];
 156                splice @ARGV, $i, 1;
 157                last;
 158        }
 159};
 160
 161my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 162
 163read_repo_config(\%opts);
 164my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 165                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 166                    'id|i=s' => \$Git::SVN::default_ref_id,
 167                    'svn-remote|remote|R=s' => \$Git::SVN::default_repo_id);
 168exit 1 if (!$rv && $cmd ne 'log');
 169
 170usage(0) if $_help;
 171version() if $_version;
 172usage(1) unless defined $cmd;
 173load_authors() if $_authors;
 174unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 175        Git::SVN::Migration::migration_check();
 176}
 177Git::SVN::init_vars();
 178eval {
 179        Git::SVN::verify_remotes_sanity();
 180        $cmd{$cmd}->[0]->(@ARGV);
 181};
 182fatal $@ if $@;
 183post_fetch_checkout();
 184exit 0;
 185
 186####################### primary functions ######################
 187sub usage {
 188        my $exit = shift || 0;
 189        my $fd = $exit ? \*STDERR : \*STDOUT;
 190        print $fd <<"";
 191git-svn - bidirectional operations between a single Subversion tree and git
 192Usage: $0 <command> [options] [arguments]\n
 193
 194        print $fd "Available commands:\n" unless $cmd;
 195
 196        foreach (sort keys %cmd) {
 197                next if $cmd && $cmd ne $_;
 198                next if /^multi-/; # don't show deprecated commands
 199                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 200                foreach (keys %{$cmd{$_}->[2]}) {
 201                        # prints out arguments as they should be passed:
 202                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 203                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 204                                                        "--$_" : "-$_" }
 205                                                split /\|/,$_)," $x\n";
 206                }
 207        }
 208        print $fd <<"";
 209\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 210arbitrary identifier if you're tracking multiple SVN branches/repositories in
 211one git repository and want to keep them separate.  See git-svn(1) for more
 212information.
 213
 214        exit $exit;
 215}
 216
 217sub version {
 218        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 219        exit 0;
 220}
 221
 222sub do_git_init_db {
 223        unless (-d $ENV{GIT_DIR}) {
 224                my @init_db = ('init');
 225                push @init_db, "--template=$_template" if defined $_template;
 226                if (defined $_shared) {
 227                        if ($_shared =~ /[a-z]/) {
 228                                push @init_db, "--shared=$_shared";
 229                        } else {
 230                                push @init_db, "--shared";
 231                        }
 232                }
 233                command_noisy(@init_db);
 234        }
 235}
 236
 237sub init_subdir {
 238        my $repo_path = shift or return;
 239        mkpath([$repo_path]) unless -d $repo_path;
 240        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 241        $ENV{GIT_DIR} = $repo_path . "/.git";
 242}
 243
 244sub cmd_clone {
 245        my ($url, $path) = @_;
 246        if (!defined $path &&
 247            (defined $_trunk || defined $_branches || defined $_tags) &&
 248            $url !~ m#^[a-z\+]+://#) {
 249                $path = $url;
 250        }
 251        warn "--path: $path\n" if defined $path;
 252        $path = basename($url) if !defined $path || !length $path;
 253        warn "++path: $path\n" if defined $path;
 254        mkpath([$path]);
 255        chdir $path or die "Couldn't chdir to $path\n";
 256        cmd_init(@_);
 257        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 258}
 259
 260sub cmd_init {
 261        if (defined $_trunk || defined $_branches || defined $_tags) {
 262                return cmd_multi_init(@_);
 263        }
 264        my $url = shift or die "SVN repository location required ",
 265                               "as a command-line argument\n";
 266        init_subdir(@_);
 267        do_git_init_db();
 268
 269        Git::SVN->init($url);
 270}
 271
 272sub cmd_fetch {
 273        if (grep /^\d+=./, @_) {
 274                die "'<rev>=<commit>' fetch arguments are ",
 275                    "no longer supported.\n";
 276        }
 277        my ($remote) = @_;
 278        if (@_ > 1) {
 279                die "Usage: $0 fetch [--all] [svn-remote]\n";
 280        }
 281        $remote ||= $Git::SVN::default_repo_id;
 282        if ($_fetch_all) {
 283                cmd_multi_fetch();
 284        } else {
 285                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 286        }
 287}
 288
 289sub cmd_set_tree {
 290        my (@commits) = @_;
 291        if ($_stdin || !@commits) {
 292                print "Reading from stdin...\n";
 293                @commits = ();
 294                while (<STDIN>) {
 295                        if (/\b($sha1_short)\b/o) {
 296                                unshift @commits, $1;
 297                        }
 298                }
 299        }
 300        my @revs;
 301        foreach my $c (@commits) {
 302                my @tmp = command('rev-parse',$c);
 303                if (scalar @tmp == 1) {
 304                        push @revs, $tmp[0];
 305                } elsif (scalar @tmp > 1) {
 306                        push @revs, reverse(command('rev-list',@tmp));
 307                } else {
 308                        fatal "Failed to rev-parse $c\n";
 309                }
 310        }
 311        my $gs = Git::SVN->new;
 312        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 313        $gs->fetch;
 314        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 315                fatal "There are new revisions that were fetched ",
 316                      "and need to be merged (or acknowledged) ",
 317                      "before committing.\nlast rev: $r_last\n",
 318                      " current: $gs->{last_rev}\n";
 319        }
 320        $gs->set_tree($_) foreach @revs;
 321        print "Done committing ",scalar @revs," revisions to SVN\n";
 322}
 323
 324sub cmd_dcommit {
 325        my $head = shift;
 326        $head ||= 'HEAD';
 327        my @refs;
 328        my ($url, $rev, $uuid) = working_head_info($head, \@refs);
 329        my $c = $refs[-1];
 330        unless (defined $url && defined $rev && defined $uuid) {
 331                die "Unable to determine upstream SVN information from ",
 332                    "$head history\n";
 333        }
 334        my $gs = Git::SVN->find_by_url($url);
 335        my $last_rev;
 336        foreach my $d (@refs) {
 337                if (!verify_ref("$d~1")) {
 338                        fatal "Commit $d\n",
 339                              "has no parent commit, and therefore ",
 340                              "nothing to diff against.\n",
 341                              "You should be working from a repository ",
 342                              "originally created by git-svn\n";
 343                }
 344                unless (defined $last_rev) {
 345                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 346                        unless (defined $last_rev) {
 347                                fatal "Unable to extract revision information ",
 348                                      "from commit $d~1\n";
 349                        }
 350                }
 351                if ($_dry_run) {
 352                        print "diff-tree $d~1 $d\n";
 353                } else {
 354                        my %ed_opts = ( r => $last_rev,
 355                                        log => get_commit_entry($d)->{log},
 356                                        ra => Git::SVN::Ra->new($url),
 357                                        tree_a => "$d~1",
 358                                        tree_b => $d,
 359                                        editor_cb => sub {
 360                                               print "Committed r$_[0]\n";
 361                                               $last_rev = $_[0]; },
 362                                        svn_path => '');
 363                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 364                                print "No changes\n$d~1 == $d\n";
 365                        }
 366                }
 367        }
 368        return if $_dry_run;
 369        unless ($gs) {
 370                warn "Could not determine fetch information for $url\n",
 371                     "Will not attempt to fetch and rebase commits.\n",
 372                     "This probably means you have useSvmProps and should\n",
 373                     "now resync your SVN::Mirror repository.\n";
 374                return;
 375        }
 376        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 377        # we always want to rebase against the current HEAD, not any
 378        # head that was passed to us
 379        my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
 380        my @finish;
 381        if (@diff) {
 382                @finish = rebase_cmd();
 383                print STDERR "W: HEAD and ", $gs->refname, " differ, ",
 384                             "using @finish:\n", "@diff";
 385        } else {
 386                print "No changes between current HEAD and ",
 387                      $gs->refname, "\nResetting to the latest ",
 388                      $gs->refname, "\n";
 389                @finish = qw/reset --mixed/;
 390        }
 391        command_noisy(@finish, $gs->refname);
 392}
 393
 394sub cmd_rebase {
 395        command_noisy(qw/update-index --refresh/);
 396        my $url = (working_head_info('HEAD'))[0];
 397        if (!defined $url) {
 398                die "Unable to determine upstream SVN information from ",
 399                    "working tree history\n";
 400        }
 401
 402        my $gs = Git::SVN->find_by_url($url);
 403        if (command(qw/diff-index HEAD --/)) {
 404                print STDERR "Cannot rebase with uncommited changes:\n";
 405                command_noisy('status');
 406                exit 1;
 407        }
 408        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 409        command_noisy(rebase_cmd(), $gs->refname);
 410}
 411
 412sub cmd_show_ignore {
 413        my $gs = Git::SVN->new;
 414        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 415        $gs->traverse_ignore(\*STDOUT, '', $r);
 416}
 417
 418sub cmd_multi_init {
 419        my $url = shift;
 420        unless (defined $_trunk || defined $_branches || defined $_tags) {
 421                usage(1);
 422        }
 423        do_git_init_db();
 424        $_prefix = '' unless defined $_prefix;
 425        if (defined $url) {
 426                $url =~ s#/+$##;
 427                init_subdir(@_);
 428        }
 429        if (defined $_trunk) {
 430                my $trunk_ref = $_prefix . 'trunk';
 431                # try both old-style and new-style lookups:
 432                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 433                unless ($gs_trunk) {
 434                        my ($trunk_url, $trunk_path) =
 435                                              complete_svn_url($url, $_trunk);
 436                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 437                                                   undef, $trunk_ref);
 438                }
 439        }
 440        return unless defined $_branches || defined $_tags;
 441        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 442        complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
 443        complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
 444}
 445
 446sub cmd_multi_fetch {
 447        my $remotes = Git::SVN::read_all_remotes();
 448        foreach my $repo_id (sort keys %$remotes) {
 449                if ($remotes->{$repo_id}->{url}) {
 450                        Git::SVN::fetch_all($repo_id, $remotes);
 451                }
 452        }
 453}
 454
 455# this command is special because it requires no metadata
 456sub cmd_commit_diff {
 457        my ($ta, $tb, $url) = @_;
 458        my $usage = "Usage: $0 commit-diff -r<revision> ".
 459                    "<tree-ish> <tree-ish> [<URL>]\n";
 460        fatal($usage) if (!defined $ta || !defined $tb);
 461        my $svn_path;
 462        if (!defined $url) {
 463                my $gs = eval { Git::SVN->new };
 464                if (!$gs) {
 465                        fatal("Needed URL or usable git-svn --id in ",
 466                              "the command-line\n", $usage);
 467                }
 468                $url = $gs->{url};
 469                $svn_path = $gs->{path};
 470        }
 471        unless (defined $_revision) {
 472                fatal("-r|--revision is a required argument\n", $usage);
 473        }
 474        if (defined $_message && defined $_file) {
 475                fatal("Both --message/-m and --file/-F specified ",
 476                      "for the commit message.\n",
 477                      "I have no idea what you mean\n");
 478        }
 479        if (defined $_file) {
 480                $_message = file_to_s($_file);
 481        } else {
 482                $_message ||= get_commit_entry($tb)->{log};
 483        }
 484        my $ra ||= Git::SVN::Ra->new($url);
 485        $svn_path ||= $ra->{svn_path};
 486        my $r = $_revision;
 487        if ($r eq 'HEAD') {
 488                $r = $ra->get_latest_revnum;
 489        } elsif ($r !~ /^\d+$/) {
 490                die "revision argument: $r not understood by git-svn\n";
 491        }
 492        my %ed_opts = ( r => $r,
 493                        log => $_message,
 494                        ra => $ra,
 495                        tree_a => $ta,
 496                        tree_b => $tb,
 497                        editor_cb => sub { print "Committed r$_[0]\n" },
 498                        svn_path => $svn_path );
 499        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 500                print "No changes\n$ta == $tb\n";
 501        }
 502}
 503
 504########################### utility functions #########################
 505
 506sub rebase_cmd {
 507        my @cmd = qw/rebase/;
 508        push @cmd, '-v' if $_verbose;
 509        push @cmd, qw/--merge/ if $_merge;
 510        push @cmd, "--strategy=$_strategy" if $_strategy;
 511        @cmd;
 512}
 513
 514sub post_fetch_checkout {
 515        return if $_no_checkout;
 516        my $gs = $Git::SVN::_head or return;
 517        return if verify_ref('refs/heads/master^0');
 518
 519        my $valid_head = verify_ref('HEAD^0');
 520        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
 521        return if ($valid_head || !verify_ref('HEAD^0'));
 522
 523        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
 524        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
 525        return if -f $index;
 526
 527        chomp(my $bare = `git config --bool --get core.bare`);
 528        return if $bare eq 'true';
 529        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
 530        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
 531        print STDERR "Checked out HEAD:\n  ",
 532                     $gs->full_url, " r", $gs->last_rev, "\n";
 533}
 534
 535sub complete_svn_url {
 536        my ($url, $path) = @_;
 537        $path =~ s#/+$##;
 538        if ($path !~ m#^[a-z\+]+://#) {
 539                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
 540                        fatal("E: '$path' is not a complete URL ",
 541                              "and a separate URL is not specified\n");
 542                }
 543                return ($url, $path);
 544        }
 545        return ($path, '');
 546}
 547
 548sub complete_url_ls_init {
 549        my ($ra, $repo_path, $switch, $pfx) = @_;
 550        unless ($repo_path) {
 551                print STDERR "W: $switch not specified\n";
 552                return;
 553        }
 554        $repo_path =~ s#/+$##;
 555        if ($repo_path =~ m#^[a-z\+]+://#) {
 556                $ra = Git::SVN::Ra->new($repo_path);
 557                $repo_path = '';
 558        } else {
 559                $repo_path =~ s#^/+##;
 560                unless ($ra) {
 561                        fatal("E: '$repo_path' is not a complete URL ",
 562                              "and a separate URL is not specified\n");
 563                }
 564        }
 565        my $url = $ra->{url};
 566        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
 567        my $k = "svn-remote.$gs->{repo_id}.url";
 568        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
 569        if ($orig_url && ($orig_url ne $gs->{url})) {
 570                die "$k already set: $orig_url\n",
 571                    "wanted to set to: $gs->{url}\n";
 572        }
 573        command_oneline('config', $k, $gs->{url}) unless $orig_url;
 574        my $remote_path = "$ra->{svn_path}/$repo_path/*";
 575        $remote_path =~ s#/+#/#g;
 576        $remote_path =~ s#^/##g;
 577        my ($n) = ($switch =~ /^--(\w+)/);
 578        if (length $pfx && $pfx !~ m#/$#) {
 579                die "--prefix='$pfx' must have a trailing slash '/'\n";
 580        }
 581        command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
 582                                "$remote_path:refs/remotes/$pfx*");
 583}
 584
 585sub verify_ref {
 586        my ($ref) = @_;
 587        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
 588                               { STDERR => 0 }); };
 589}
 590
 591sub get_tree_from_treeish {
 592        my ($treeish) = @_;
 593        # $treeish can be a symbolic ref, too:
 594        my $type = command_oneline(qw/cat-file -t/, $treeish);
 595        my $expected;
 596        while ($type eq 'tag') {
 597                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
 598        }
 599        if ($type eq 'commit') {
 600                $expected = (grep /^tree /, command(qw/cat-file commit/,
 601                                                    $treeish))[0];
 602                ($expected) = ($expected =~ /^tree ($sha1)$/o);
 603                die "Unable to get tree from $treeish\n" unless $expected;
 604        } elsif ($type eq 'tree') {
 605                $expected = $treeish;
 606        } else {
 607                die "$treeish is a $type, expected tree, tag or commit\n";
 608        }
 609        return $expected;
 610}
 611
 612sub get_commit_entry {
 613        my ($treeish) = shift;
 614        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
 615        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
 616        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
 617        open my $log_fh, '>', $commit_editmsg or croak $!;
 618
 619        my $type = command_oneline(qw/cat-file -t/, $treeish);
 620        if ($type eq 'commit' || $type eq 'tag') {
 621                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
 622                                                         $type, $treeish);
 623                my $in_msg = 0;
 624                while (<$msg_fh>) {
 625                        if (!$in_msg) {
 626                                $in_msg = 1 if (/^\s*$/);
 627                        } elsif (/^git-svn-id: /) {
 628                                # skip this for now, we regenerate the
 629                                # correct one on re-fetch anyways
 630                                # TODO: set *:merge properties or like...
 631                        } else {
 632                                print $log_fh $_ or croak $!;
 633                        }
 634                }
 635                command_close_pipe($msg_fh, $ctx);
 636        }
 637        close $log_fh or croak $!;
 638
 639        if ($_edit || ($type eq 'tree')) {
 640                my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
 641                # TODO: strip out spaces, comments, like git-commit.sh
 642                system($editor, $commit_editmsg);
 643        }
 644        rename $commit_editmsg, $commit_msg or croak $!;
 645        open $log_fh, '<', $commit_msg or croak $!;
 646        { local $/; chomp($log_entry{log} = <$log_fh>); }
 647        close $log_fh or croak $!;
 648        unlink $commit_msg;
 649        \%log_entry;
 650}
 651
 652sub s_to_file {
 653        my ($str, $file, $mode) = @_;
 654        open my $fd,'>',$file or croak $!;
 655        print $fd $str,"\n" or croak $!;
 656        close $fd or croak $!;
 657        chmod ($mode &~ umask, $file) if (defined $mode);
 658}
 659
 660sub file_to_s {
 661        my $file = shift;
 662        open my $fd,'<',$file or croak "$!: file: $file\n";
 663        local $/;
 664        my $ret = <$fd>;
 665        close $fd or croak $!;
 666        $ret =~ s/\s*$//s;
 667        return $ret;
 668}
 669
 670# '<svn username> = real-name <email address>' mapping based on git-svnimport:
 671sub load_authors {
 672        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
 673        my $log = $cmd eq 'log';
 674        while (<$authors>) {
 675                chomp;
 676                next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
 677                my ($user, $name, $email) = ($1, $2, $3);
 678                if ($log) {
 679                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
 680                } else {
 681                        $users{$user} = [$name, $email];
 682                }
 683        }
 684        close $authors or croak $!;
 685}
 686
 687# convert GetOpt::Long specs for use by git-config
 688sub read_repo_config {
 689        return unless -d $ENV{GIT_DIR};
 690        my $opts = shift;
 691        my @config_only;
 692        foreach my $o (keys %$opts) {
 693                # if we have mixedCase and a long option-only, then
 694                # it's a config-only variable that we don't need for
 695                # the command-line.
 696                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
 697                my $v = $opts->{$o};
 698                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
 699                $key =~ s/-//g;
 700                my $arg = 'git-config';
 701                $arg .= ' --int' if ($o =~ /[:=]i$/);
 702                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
 703                if (ref $v eq 'ARRAY') {
 704                        chomp(my @tmp = `$arg --get-all svn.$key`);
 705                        @$v = @tmp if @tmp;
 706                } else {
 707                        chomp(my $tmp = `$arg --get svn.$key`);
 708                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
 709                                $$v = $tmp;
 710                        }
 711                }
 712        }
 713        delete @$opts{@config_only} if @config_only;
 714}
 715
 716sub extract_metadata {
 717        my $id = shift or return (undef, undef, undef);
 718        my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
 719                                                        \s([a-f\d\-]+)$/x);
 720        if (!defined $rev || !$uuid || !$url) {
 721                # some of the original repositories I made had
 722                # identifiers like this:
 723                ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
 724        }
 725        return ($url, $rev, $uuid);
 726}
 727
 728sub cmt_metadata {
 729        return extract_metadata((grep(/^git-svn-id: /,
 730                command(qw/cat-file commit/, shift)))[-1]);
 731}
 732
 733sub working_head_info {
 734        my ($head, $refs) = @_;
 735        my ($url, $rev, $uuid);
 736        my ($fh, $ctx) = command_output_pipe('rev-list', $head);
 737        while (<$fh>) {
 738                chomp;
 739                ($url, $rev, $uuid) = cmt_metadata($_);
 740                last if (defined $url && defined $rev && defined $uuid);
 741                unshift @$refs, $_ if $refs;
 742        }
 743        close $fh; # break the pipe
 744        ($url, $rev, $uuid);
 745}
 746
 747package Git::SVN;
 748use strict;
 749use warnings;
 750use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
 751            $_repack $_repack_flags $_use_svm_props $_head
 752            $_use_svnsync_props/;
 753use Carp qw/croak/;
 754use File::Path qw/mkpath/;
 755use File::Copy qw/copy/;
 756use IPC::Open3;
 757
 758my $_repack_nr;
 759# properties that we do not log:
 760my %SKIP_PROP;
 761BEGIN {
 762        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
 763                                        svn:special svn:executable
 764                                        svn:entry:committed-rev
 765                                        svn:entry:last-author
 766                                        svn:entry:uuid
 767                                        svn:entry:committed-date/;
 768
 769        # some options are read globally, but can be overridden locally
 770        # per [svn-remote "..."] section.  Command-line options will *NOT*
 771        # override options set in an [svn-remote "..."] section
 772        my $e;
 773        foreach (qw/follow_parent no_metadata use_svm_props
 774                    use_svnsync_props/) {
 775                my $key = $_;
 776                $key =~ tr/_//d;
 777                $e .= "sub $_ {
 778                        my (\$self) = \@_;
 779                        return \$self->{-$_} if exists \$self->{-$_};
 780                        my \$k = \"svn-remote.\$self->{repo_id}\.$key\";
 781                        eval { command_oneline(qw/config --get/, \$k) };
 782                        if (\$@) {
 783                                \$self->{-$_} = \$Git::SVN::_$_;
 784                        } else {
 785                                my \$v = command_oneline(qw/config --bool/,\$k);
 786                                \$self->{-$_} = \$v eq 'false' ? 0 : 1;
 787                        }
 788                        return \$self->{-$_} }\n";
 789        }
 790        $e .= "1;\n";
 791        eval $e or die $@;
 792}
 793
 794my %LOCKFILES;
 795END { unlink keys %LOCKFILES if %LOCKFILES }
 796
 797sub resolve_local_globs {
 798        my ($url, $fetch, $glob_spec) = @_;
 799        return unless defined $glob_spec;
 800        my $ref = $glob_spec->{ref};
 801        my $path = $glob_spec->{path};
 802        foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
 803                next unless m#^refs/remotes/$ref->{regex}$#;
 804                my $p = $1;
 805                my $pathname = $path->full_path($p);
 806                my $refname = $ref->full_path($p);
 807                if (my $existing = $fetch->{$pathname}) {
 808                        if ($existing ne $refname) {
 809                                die "Refspec conflict:\n",
 810                                    "existing: refs/remotes/$existing\n",
 811                                    " globbed: refs/remotes/$refname\n";
 812                        }
 813                        my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
 814                        $u =~ s!^\Q$url\E(/|$)!! or die
 815                          "refs/remotes/$refname: '$url' not found in '$u'\n";
 816                        if ($pathname ne $u) {
 817                                warn "W: Refspec glob conflict ",
 818                                     "(ref: refs/remotes/$refname):\n",
 819                                     "expected path: $pathname\n",
 820                                     "    real path: $u\n",
 821                                     "Continuing ahead with $u\n";
 822                                next;
 823                        }
 824                } else {
 825                        $fetch->{$pathname} = $refname;
 826                }
 827        }
 828}
 829
 830sub parse_revision_argument {
 831        my ($base, $head) = @_;
 832        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
 833                return ($base, $head);
 834        }
 835        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
 836        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
 837        return ($head, $head) if ($::_revision eq 'HEAD');
 838        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
 839        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
 840        die "revision argument: $::_revision not understood by git-svn\n";
 841}
 842
 843sub fetch_all {
 844        my ($repo_id, $remotes) = @_;
 845        if (ref $repo_id) {
 846                my $gs = $repo_id;
 847                $repo_id = undef;
 848                $repo_id = $gs->{repo_id};
 849        }
 850        $remotes ||= read_all_remotes();
 851        my $remote = $remotes->{$repo_id} or
 852                     die "[svn-remote \"$repo_id\"] unknown\n";
 853        my $fetch = $remote->{fetch};
 854        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
 855        my (@gs, @globs);
 856        my $ra = Git::SVN::Ra->new($url);
 857        my $uuid = $ra->get_uuid;
 858        my $head = $ra->get_latest_revnum;
 859        my $base = defined $fetch ? $head : 0;
 860
 861        # read the max revs for wildcard expansion (branches/*, tags/*)
 862        foreach my $t (qw/branches tags/) {
 863                defined $remote->{$t} or next;
 864                push @globs, $remote->{$t};
 865                my $max_rev = eval { tmp_config(qw/--int --get/,
 866                                         "svn-remote.$repo_id.${t}-maxRev") };
 867                if (defined $max_rev && ($max_rev < $base)) {
 868                        $base = $max_rev;
 869                } elsif (!defined $max_rev) {
 870                        $base = 0;
 871                }
 872        }
 873
 874        if ($fetch) {
 875                foreach my $p (sort keys %$fetch) {
 876                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
 877                        my $lr = $gs->rev_db_max;
 878                        if (defined $lr) {
 879                                $base = $lr if ($lr < $base);
 880                        }
 881                        push @gs, $gs;
 882                }
 883        }
 884
 885        ($base, $head) = parse_revision_argument($base, $head);
 886        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
 887}
 888
 889sub read_all_remotes {
 890        my $r = {};
 891        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
 892                if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
 893                        $r->{$1}->{fetch}->{$2} = $3;
 894                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
 895                        $r->{$1}->{url} = $2;
 896                } elsif (m!^(.+)\.(branches|tags)=
 897                           (.*):refs/remotes/(.+)\s*$/!x) {
 898                        my ($p, $g) = ($3, $4);
 899                        my $rs = $r->{$1}->{$2} = {
 900                                          t => $2,
 901                                          remote => $1,
 902                                          path => Git::SVN::GlobSpec->new($p),
 903                                          ref => Git::SVN::GlobSpec->new($g) };
 904                        if (length($rs->{ref}->{right}) != 0) {
 905                                die "The '*' glob character must be the last ",
 906                                    "character of '$g'\n";
 907                        }
 908                }
 909        }
 910        $r;
 911}
 912
 913sub init_vars {
 914        if (defined $_repack) {
 915                $_repack = 1000 if ($_repack <= 0);
 916                $_repack_nr = $_repack;
 917                $_repack_flags ||= '-d';
 918        }
 919}
 920
 921sub verify_remotes_sanity {
 922        return unless -d $ENV{GIT_DIR};
 923        my %seen;
 924        foreach (command(qw/config -l/)) {
 925                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
 926                        if ($seen{$1}) {
 927                                die "Remote ref refs/remote/$1 is tracked by",
 928                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
 929                                    "Please resolve this ambiguity in ",
 930                                    "your git configuration file before ",
 931                                    "continuing\n";
 932                        }
 933                        $seen{$1} = $_;
 934                }
 935        }
 936}
 937
 938# we allow more chars than remotes2config.sh...
 939sub sanitize_remote_name {
 940        my ($name) = @_;
 941        $name =~ tr{A-Za-z0-9:,/+-}{.}c;
 942        $name;
 943}
 944
 945sub find_existing_remote {
 946        my ($url, $remotes) = @_;
 947        my $existing;
 948        foreach my $repo_id (keys %$remotes) {
 949                my $u = $remotes->{$repo_id}->{url} or next;
 950                next if $u ne $url;
 951                $existing = $repo_id;
 952                last;
 953        }
 954        $existing;
 955}
 956
 957sub init_remote_config {
 958        my ($self, $url, $no_write) = @_;
 959        $url =~ s!/+$!!; # strip trailing slash
 960        my $r = read_all_remotes();
 961        my $existing = find_existing_remote($url, $r);
 962        if ($existing) {
 963                unless ($no_write) {
 964                        print STDERR "Using existing ",
 965                                     "[svn-remote \"$existing\"]\n";
 966                }
 967                $self->{repo_id} = $existing;
 968        } else {
 969                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
 970                $existing = find_existing_remote($min_url, $r);
 971                if ($existing) {
 972                        unless ($no_write) {
 973                                print STDERR "Using existing ",
 974                                             "[svn-remote \"$existing\"]\n";
 975                        }
 976                        $self->{repo_id} = $existing;
 977                }
 978                if ($min_url ne $url) {
 979                        unless ($no_write) {
 980                                print STDERR "Using higher level of URL: ",
 981                                             "$url => $min_url\n";
 982                        }
 983                        my $old_path = $self->{path};
 984                        $self->{path} = $url;
 985                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
 986                        if (length $old_path) {
 987                                $self->{path} .= "/$old_path";
 988                        }
 989                        $url = $min_url;
 990                }
 991        }
 992        my $orig_url;
 993        if (!$existing) {
 994                # verify that we aren't overwriting anything:
 995                $orig_url = eval {
 996                        command_oneline('config', '--get',
 997                                        "svn-remote.$self->{repo_id}.url")
 998                };
 999                if ($orig_url && ($orig_url ne $url)) {
1000                        die "svn-remote.$self->{repo_id}.url already set: ",
1001                            "$orig_url\nwanted to set to: $url\n";
1002                }
1003        }
1004        my ($xrepo_id, $xpath) = find_ref($self->refname);
1005        if (defined $xpath) {
1006                die "svn-remote.$xrepo_id.fetch already set to track ",
1007                    "$xpath:refs/remotes/", $self->refname, "\n";
1008        }
1009        unless ($no_write) {
1010                command_noisy('config',
1011                              "svn-remote.$self->{repo_id}.url", $url);
1012                command_noisy('config', '--add',
1013                              "svn-remote.$self->{repo_id}.fetch",
1014                              "$self->{path}:".$self->refname);
1015        }
1016        $self->{url} = $url;
1017}
1018
1019sub find_by_url { # repos_root and, path are optional
1020        my ($class, $full_url, $repos_root, $path) = @_;
1021        my $remotes = read_all_remotes();
1022        if (defined $full_url && defined $repos_root && !defined $path) {
1023                $path = $full_url;
1024                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1025        }
1026        foreach my $repo_id (keys %$remotes) {
1027                my $u = $remotes->{$repo_id}->{url} or next;
1028                next if defined $repos_root && $repos_root ne $u;
1029
1030                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1031                foreach (qw/branches tags/) {
1032                        resolve_local_globs($u, $fetch,
1033                                            $remotes->{$repo_id}->{$_});
1034                }
1035                my $p = $path;
1036                unless (defined $p) {
1037                        $p = $full_url;
1038                        $p =~ s#^\Q$u\E(?:/|$)## or next;
1039                }
1040                foreach my $f (keys %$fetch) {
1041                        next if $f ne $p;
1042                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1043                }
1044        }
1045        undef;
1046}
1047
1048sub init {
1049        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1050        my $self = _new($class, $repo_id, $ref_id, $path);
1051        if (defined $url) {
1052                $self->init_remote_config($url, $no_write);
1053        }
1054        $self;
1055}
1056
1057sub find_ref {
1058        my ($ref_id) = @_;
1059        foreach (command(qw/config -l/)) {
1060                next unless m!^svn-remote\.(.+)\.fetch=
1061                              \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1062                my ($repo_id, $path, $ref) = ($1, $2, $3);
1063                if ($ref eq $ref_id) {
1064                        $path = '' if ($path =~ m#^\./?#);
1065                        return ($repo_id, $path);
1066                }
1067        }
1068        (undef, undef, undef);
1069}
1070
1071sub new {
1072        my ($class, $ref_id, $repo_id, $path) = @_;
1073        if (defined $ref_id && !defined $repo_id && !defined $path) {
1074                ($repo_id, $path) = find_ref($ref_id);
1075                if (!defined $repo_id) {
1076                        die "Could not find a \"svn-remote.*.fetch\" key ",
1077                            "in the repository configuration matching: ",
1078                            "refs/remotes/$ref_id\n";
1079                }
1080        }
1081        my $self = _new($class, $repo_id, $ref_id, $path);
1082        if (!defined $self->{path} || !length $self->{path}) {
1083                my $fetch = command_oneline('config', '--get',
1084                                            "svn-remote.$repo_id.fetch",
1085                                            ":refs/remotes/$ref_id\$") or
1086                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1087                         "\":refs/remotes/$ref_id\$\" in config\n";
1088                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1089        }
1090        $self->{url} = command_oneline('config', '--get',
1091                                       "svn-remote.$repo_id.url") or
1092                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1093        $self->rebuild;
1094        $self;
1095}
1096
1097sub refname { "refs/remotes/$_[0]->{ref_id}" }
1098
1099sub svm_uuid {
1100        my ($self) = @_;
1101        return $self->{svm}->{uuid} if $self->svm;
1102        $self->ra;
1103        unless ($self->{svm}) {
1104                die "SVM UUID not cached, and reading remotely failed\n";
1105        }
1106        $self->{svm}->{uuid};
1107}
1108
1109sub svm {
1110        my ($self) = @_;
1111        return $self->{svm} if $self->{svm};
1112        my $svm;
1113        # see if we have it in our config, first:
1114        eval {
1115                my $section = "svn-remote.$self->{repo_id}";
1116                $svm = {
1117                  source => tmp_config('--get', "$section.svm-source"),
1118                  uuid => tmp_config('--get', "$section.svm-uuid"),
1119                }
1120        };
1121        $self->{svm} = $svm if ($svm && $svm->{source} && $svm->{uuid});
1122        $self->{svm};
1123}
1124
1125sub _set_svm_vars {
1126        my ($self, $ra) = @_;
1127        return $ra if $self->svm;
1128
1129        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1130                    "(svm:source, svm:mirror, svm:mirror) ",
1131                    "from the following URLs:\n" );
1132        sub read_svm_props {
1133                my ($self, $props) = @_;
1134                my $src = $props->{'svm:source'};
1135                my $mirror = $props->{'svm:mirror'};
1136                my $uuid = $props->{'svm:uuid'};
1137                return undef if (!$src || !$mirror || !$uuid);
1138
1139                chomp($src, $mirror, $uuid);
1140
1141                $uuid =~ m{^[0-9a-f\-]{30,}$}
1142                    or die "doesn't look right - svm:uuid is '$uuid'\n";
1143                # don't know what a '!' is there for, also the
1144                # username is of no interest
1145                $src =~ s{/?!$}{$mirror};
1146                $src =~ s{/+$}{}; # no trailing slashes please
1147                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1148
1149                my $section = "svn-remote.$self->{repo_id}";
1150                tmp_config('--add', "$section.svm-source", $src);
1151                tmp_config('--add', "$section.svm-uuid", $uuid);
1152                $self->{svm} = { source => $src , uuid => $uuid };
1153                return 1;
1154        }
1155
1156        my $r = $ra->get_latest_revnum;
1157        my $path = $self->{path};
1158        my @tried_a = ($path);
1159        while (length $path) {
1160                if ($self->read_svm_props(($ra->get_dir($path, $r))[2])) {
1161                        return $ra;
1162                }
1163                $path =~ s#/?[^/]+$## && push @tried_a, $path;
1164        }
1165        if ($self->read_svm_props(($ra->get_dir('', $r))[2])) {
1166                return $ra;
1167        }
1168
1169        if ($ra->{repos_root} eq $self->{url}) {
1170                die @err, map { "  $self->{url}/$_\n" } @tried_a, "\n";
1171        }
1172
1173        # nope, make sure we're connected to the repository root:
1174        my $ok;
1175        my @tried_b;
1176        $path = $ra->{svn_path};
1177        $path =~ s#/?[^/]+$##; # we already tried this one above
1178        $ra = Git::SVN::Ra->new($ra->{repos_root});
1179        while (length $path) {
1180                $ok = $self->read_svm_props(($ra->get_dir($path, $r))[2]);
1181                last if $ok;
1182                $path =~ s#/?[^/]+$## && push @tried_b, $path;
1183        }
1184        $ok = $self->read_svm_props(($ra->get_dir('', $r))[2]) unless $ok;
1185        if (!$ok) {
1186                die @err, map { "  $self->{url}/$_\n" } @tried_a, "\n",
1187                          map { "  $ra->{url}/$_\n" } @tried_b, "\n"
1188        }
1189        Git::SVN::Ra->new($self->{url});
1190}
1191
1192sub svnsync {
1193        my ($self) = @_;
1194        return $self->{svnsync} if $self->{svnsync};
1195
1196        if ($self->no_metadata) {
1197                die "Can't have both 'noMetadata' and ",
1198                    "'useSvnsyncProps' options set!\n";
1199        }
1200        if ($self->rewrite_root) {
1201                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1202                    "options set!\n";
1203        }
1204
1205        my $svnsync;
1206        # see if we have it in our config, first:
1207        eval {
1208                my $section = "svn-remote.$self->{repo_id}";
1209                $svnsync = {
1210                  url => tmp_config('--get', "$section.svnsync-url"),
1211                  uuid => tmp_config('--get', "$section.svnsync-uuid"),
1212                }
1213        };
1214        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1215                return $self->{svnsync} = $svnsync;
1216        }
1217
1218        my $err = "useSvnsyncProps set, but failed to read " .
1219                  "svnsync property: svn:sync-from-";
1220        my $rp = $self->ra->rev_proplist(0);
1221
1222        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1223        $url =~ m{^[a-z\+]+://} or
1224                   die "doesn't look right - svn:sync-from-url is '$url'\n";
1225
1226        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1227        $uuid =~ m{^[0-9a-f\-]{30,}$} or
1228                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1229
1230        my $section = "svn-remote.$self->{repo_id}";
1231        tmp_config('--add', "$section.svnsync-uuid", $uuid);
1232        tmp_config('--add', "$section.svnsync-url", $url);
1233        return $self->{svnsync} = { url => $url, uuid => $uuid };
1234}
1235
1236# this allows us to memoize our SVN::Ra UUID locally and avoid a
1237# remote lookup (useful for 'git svn log').
1238sub ra_uuid {
1239        my ($self) = @_;
1240        unless ($self->{ra_uuid}) {
1241                my $key = "svn-remote.$self->{repo_id}.uuid";
1242                my $uuid = eval { tmp_config('--get', $key) };
1243                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1244                        $self->{ra_uuid} = $uuid;
1245                } else {
1246                        die "ra_uuid called without URL\n" unless $self->{url};
1247                        $self->{ra_uuid} = $self->ra->get_uuid;
1248                        tmp_config('--add', $key, $self->{ra_uuid});
1249                }
1250        }
1251        $self->{ra_uuid};
1252}
1253
1254sub ra {
1255        my ($self) = shift;
1256        my $ra = Git::SVN::Ra->new($self->{url});
1257        if ($self->use_svm_props && !$self->{svm}) {
1258                if ($self->no_metadata) {
1259                        die "Can't have both 'noMetadata' and ",
1260                            "'useSvmProps' options set!\n";
1261                } elsif ($self->use_svnsync_props) {
1262                        die "Can't have both 'useSvnsyncProps' and ",
1263                            "'useSvmProps' options set!\n";
1264                }
1265                $ra = $self->_set_svm_vars($ra);
1266                $self->{-want_revprops} = 1;
1267        }
1268        $ra;
1269}
1270
1271sub rel_path {
1272        my ($self) = @_;
1273        my $repos_root = $self->ra->{repos_root};
1274        return $self->{path} if ($self->{url} eq $repos_root);
1275        die "BUG: rel_path failed! repos_root: $repos_root, Ra URL: ",
1276            $self->ra->{url}, " path: $self->{path},  URL: $self->{url}\n";
1277}
1278
1279sub traverse_ignore {
1280        my ($self, $fh, $path, $r) = @_;
1281        $path =~ s#^/+##g;
1282        my $ra = $self->ra;
1283        my ($dirent, undef, $props) = $ra->get_dir($path, $r);
1284        my $p = $path;
1285        $p =~ s#^\Q$ra->{svn_path}\E/##;
1286        print $fh length $p ? "\n# $p\n" : "\n# /\n";
1287        if (my $s = $props->{'svn:ignore'}) {
1288                $s =~ s/[\r\n]+/\n/g;
1289                chomp $s;
1290                if (length $p == 0) {
1291                        $s =~ s#\n#\n/$p#g;
1292                        print $fh "/$s\n";
1293                } else {
1294                        $s =~ s#\n#\n/$p/#g;
1295                        print $fh "/$p/$s\n";
1296                }
1297        }
1298        foreach (sort keys %$dirent) {
1299                next if $dirent->{$_}->kind != $SVN::Node::dir;
1300                $self->traverse_ignore($fh, "$path/$_", $r);
1301        }
1302}
1303
1304sub last_rev { ($_[0]->last_rev_commit)[0] }
1305sub last_commit { ($_[0]->last_rev_commit)[1] }
1306
1307# returns the newest SVN revision number and newest commit SHA1
1308sub last_rev_commit {
1309        my ($self) = @_;
1310        if (defined $self->{last_rev} && defined $self->{last_commit}) {
1311                return ($self->{last_rev}, $self->{last_commit});
1312        }
1313        my $c = ::verify_ref($self->refname.'^0');
1314        if ($c && !$self->use_svm_props && !$self->no_metadata) {
1315                my $rev = (::cmt_metadata($c))[1];
1316                if (defined $rev) {
1317                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1318                        return ($rev, $c);
1319                }
1320        }
1321        my $db_path = $self->db_path;
1322        unless (-e $db_path) {
1323                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1324                return (undef, undef);
1325        }
1326        my $offset = -41; # from tail
1327        my $rl;
1328        open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1329        sysseek($fh, $offset, 2); # don't care for errors
1330        sysread($fh, $rl, 41) == 41 or return (undef, undef);
1331        chomp $rl;
1332        while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1333                $offset -= 41;
1334                sysseek($fh, $offset, 2); # don't care for errors
1335                sysread($fh, $rl, 41) == 41 or return (undef, undef);
1336                chomp $rl;
1337        }
1338        if ($c && $c ne $rl) {
1339                die "$db_path and ", $self->refname,
1340                    " inconsistent!:\n$c != $rl\n";
1341        }
1342        my $rev = sysseek($fh, 0, 1) or croak $!;
1343        $rev =  ($rev - 41) / 41;
1344        close $fh or croak $!;
1345        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1346        return ($rev, $c);
1347}
1348
1349sub get_fetch_range {
1350        my ($self, $min, $max) = @_;
1351        $max ||= $self->ra->get_latest_revnum;
1352        $min ||= $self->rev_db_max;
1353        (++$min, $max);
1354}
1355
1356sub tmp_config {
1357        my (@args) = @_;
1358        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1359        my $config = "$ENV{GIT_DIR}/svn/.metadata";
1360        if (-e $old_def_config && ! -e $config) {
1361                rename $old_def_config, $config or
1362                       die "Failed rename $old_def_config => $config: $!\n";
1363        }
1364        my $old_config = $ENV{GIT_CONFIG};
1365        $ENV{GIT_CONFIG} = $config;
1366        $@ = undef;
1367        my @ret = eval {
1368                unless (-f $config) {
1369                        mkfile($config);
1370                        open my $fh, '>', $config or
1371                            die "Can't open $config: $!\n";
1372                        print $fh "; This file is used internally by ",
1373                                  "git-svn\n" or die
1374                                  "Couldn't write to $config: $!\n";
1375                        print $fh "; You should not have to edit it\n" or
1376                              die "Couldn't write to $config: $!\n";
1377                        close $fh or die "Couldn't close $config: $!\n";
1378                }
1379                command('config', @args);
1380        };
1381        my $err = $@;
1382        if (defined $old_config) {
1383                $ENV{GIT_CONFIG} = $old_config;
1384        } else {
1385                delete $ENV{GIT_CONFIG};
1386        }
1387        die $err if $err;
1388        wantarray ? @ret : $ret[0];
1389}
1390
1391sub tmp_index_do {
1392        my ($self, $sub) = @_;
1393        my $old_index = $ENV{GIT_INDEX_FILE};
1394        $ENV{GIT_INDEX_FILE} = $self->{index};
1395        $@ = undef;
1396        my @ret = eval {
1397                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1398                mkpath([$dir]) unless -d $dir;
1399                &$sub;
1400        };
1401        my $err = $@;
1402        if (defined $old_index) {
1403                $ENV{GIT_INDEX_FILE} = $old_index;
1404        } else {
1405                delete $ENV{GIT_INDEX_FILE};
1406        }
1407        die $err if $err;
1408        wantarray ? @ret : $ret[0];
1409}
1410
1411sub assert_index_clean {
1412        my ($self, $treeish) = @_;
1413
1414        $self->tmp_index_do(sub {
1415                command_noisy('read-tree', $treeish) unless -e $self->{index};
1416                my $x = command_oneline('write-tree');
1417                my ($y) = (command(qw/cat-file commit/, $treeish) =~
1418                           /^tree ($::sha1)/mo);
1419                return if $y eq $x;
1420
1421                warn "Index mismatch: $y != $x\nrereading $treeish\n";
1422                unlink $self->{index} or die "unlink $self->{index}: $!\n";
1423                command_noisy('read-tree', $treeish);
1424                $x = command_oneline('write-tree');
1425                if ($y ne $x) {
1426                        ::fatal "trees ($treeish) $y != $x\n",
1427                                "Something is seriously wrong...\n";
1428                }
1429        });
1430}
1431
1432sub get_commit_parents {
1433        my ($self, $log_entry) = @_;
1434        my (%seen, @ret, @tmp);
1435        # legacy support for 'set-tree'; this is only used by set_tree_cb:
1436        if (my $ip = $self->{inject_parents}) {
1437                if (my $commit = delete $ip->{$log_entry->{revision}}) {
1438                        push @tmp, $commit;
1439                }
1440        }
1441        if (my $cur = ::verify_ref($self->refname.'^0')) {
1442                push @tmp, $cur;
1443        }
1444        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1445        while (my $p = shift @tmp) {
1446                next if $seen{$p};
1447                $seen{$p} = 1;
1448                push @ret, $p;
1449                # MAXPARENT is defined to 16 in commit-tree.c:
1450                last if @ret >= 16;
1451        }
1452        if (@tmp) {
1453                die "r$log_entry->{revision}: No room for parents:\n\t",
1454                    join("\n\t", @tmp), "\n";
1455        }
1456        @ret;
1457}
1458
1459sub rewrite_root {
1460        my ($self) = @_;
1461        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
1462        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
1463        my $rwr = eval { command_oneline(qw/config --get/, $k) };
1464        if ($rwr) {
1465                $rwr =~ s#/+$##;
1466                if ($rwr !~ m#^[a-z\+]+://#) {
1467                        die "$rwr is not a valid URL (key: $k)\n";
1468                }
1469        }
1470        $self->{-rewrite_root} = $rwr;
1471}
1472
1473sub metadata_url {
1474        my ($self) = @_;
1475        ($self->rewrite_root || $self->{url}) .
1476           (length $self->{path} ? '/' . $self->{path} : '');
1477}
1478
1479sub full_url {
1480        my ($self) = @_;
1481        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1482}
1483
1484sub do_git_commit {
1485        my ($self, $log_entry) = @_;
1486        my $lr = $self->last_rev;
1487        if (defined $lr && $lr >= $log_entry->{revision}) {
1488                die "Last fetched revision of ", $self->refname,
1489                    " was r$lr, but we are about to fetch: ",
1490                    "r$log_entry->{revision}!\n";
1491        }
1492        if (my $c = $self->rev_db_get($log_entry->{revision})) {
1493                croak "$log_entry->{revision} = $c already exists! ",
1494                      "Why are we refetching it?\n";
1495        }
1496        $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1497        $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1498                                                          $log_entry->{email};
1499        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1500
1501        my $tree = $log_entry->{tree};
1502        if (!defined $tree) {
1503                $tree = $self->tmp_index_do(sub {
1504                                            command_oneline('write-tree') });
1505        }
1506        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1507
1508        my @exec = ('git-commit-tree', $tree);
1509        foreach ($self->get_commit_parents($log_entry)) {
1510                push @exec, '-p', $_;
1511        }
1512        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1513                                                                   or croak $!;
1514        print $msg_fh $log_entry->{log} or croak $!;
1515        unless ($self->no_metadata) {
1516                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1517                              or croak $!;
1518        }
1519        $msg_fh->flush == 0 or croak $!;
1520        close $msg_fh or croak $!;
1521        chomp(my $commit = do { local $/; <$out_fh> });
1522        close $out_fh or croak $!;
1523        waitpid $pid, 0;
1524        croak $? if $?;
1525        if ($commit !~ /^$::sha1$/o) {
1526                die "Failed to commit, invalid sha1: $commit\n";
1527        }
1528
1529        $self->rev_db_set($log_entry->{revision}, $commit, 1);
1530
1531        $self->{last_rev} = $log_entry->{revision};
1532        $self->{last_commit} = $commit;
1533        print "r$log_entry->{revision}";
1534        if (defined $log_entry->{svm_revision}) {
1535                 print " (\@$log_entry->{svm_revision})";
1536                 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1537                                   0, $self->svm_uuid);
1538        }
1539        print " = $commit ($self->{ref_id})\n";
1540        if (defined $_repack && (--$_repack_nr == 0)) {
1541                $_repack_nr = $_repack;
1542                # repack doesn't use any arguments with spaces in them, does it?
1543                print "Running git repack $_repack_flags ...\n";
1544                command_noisy('repack', split(/\s+/, $_repack_flags));
1545                print "Done repacking\n";
1546        }
1547        return $commit;
1548}
1549
1550sub match_paths {
1551        my ($self, $paths, $r) = @_;
1552        return 1 if $self->{path} eq '';
1553        if (my $path = $paths->{"/$self->{path}"}) {
1554                return ($path->{action} eq 'D') ? 0 : 1;
1555        }
1556        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1557        if (grep /$self->{path_regex}/, keys %$paths) {
1558                return 1;
1559        }
1560        my $c = '';
1561        foreach (split m#/#, $self->{path}) {
1562                $c .= "/$_";
1563                next unless ($paths->{$c} &&
1564                             ($paths->{$c}->{action} =~ /^[AR]$/));
1565                if ($self->ra->check_path($self->{path}, $r) ==
1566                    $SVN::Node::dir) {
1567                        return 1;
1568                }
1569        }
1570        return 0;
1571}
1572
1573sub find_parent_branch {
1574        my ($self, $paths, $rev) = @_;
1575        return undef unless $self->follow_parent;
1576        unless (defined $paths) {
1577                my $err_handler = $SVN::Error::handler;
1578                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1579                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1580                                   $paths =
1581                                      Git::SVN::Ra::dup_changed_paths($_[0]) });
1582                $SVN::Error::handler = $err_handler;
1583        }
1584        return undef unless defined $paths;
1585
1586        # look for a parent from another branch:
1587        my @b_path_components = split m#/#, $self->rel_path;
1588        my @a_path_components;
1589        my $i;
1590        while (@b_path_components) {
1591                $i = $paths->{'/'.join('/', @b_path_components)};
1592                last if $i && defined $i->{copyfrom_path};
1593                unshift(@a_path_components, pop(@b_path_components));
1594        }
1595        return undef unless defined $i && defined $i->{copyfrom_path};
1596        my $branch_from = $i->{copyfrom_path};
1597        if (@a_path_components) {
1598                print STDERR "branch_from: $branch_from => ";
1599                $branch_from .= '/'.join('/', @a_path_components);
1600                print STDERR $branch_from, "\n";
1601        }
1602        my $r = $i->{copyfrom_rev};
1603        my $repos_root = $self->ra->{repos_root};
1604        my $url = $self->ra->{url};
1605        my $new_url = $repos_root . $branch_from;
1606        print STDERR  "Found possible branch point: ",
1607                      "$new_url => ", $self->full_url, ", $r\n";
1608        $branch_from =~ s#^/##;
1609        my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
1610        unless ($gs) {
1611                my $ref_id = $self->{ref_id};
1612                $ref_id =~ s/\@\d+$//;
1613                $ref_id .= "\@$r";
1614                # just grow a tail if we're not unique enough :x
1615                $ref_id .= '-' while find_ref($ref_id);
1616                print STDERR "Initializing parent: $ref_id\n";
1617                $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1618        }
1619        my ($r0, $parent) = $gs->find_rev_before($r, 1);
1620        if (!defined $r0 || !defined $parent) {
1621                $gs->fetch(0, $r);
1622                ($r0, $parent) = $gs->last_rev_commit;
1623        }
1624        if (defined $r0 && defined $parent) {
1625                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1626                $self->assert_index_clean($parent);
1627                my $ed;
1628                if ($self->ra->can_do_switch) {
1629                        print STDERR "Following parent with do_switch\n";
1630                        # do_switch works with svn/trunk >= r22312, but that
1631                        # is not included with SVN 1.4.3 (the latest version
1632                        # at the moment), so we can't rely on it
1633                        $self->{last_commit} = $parent;
1634                        $ed = SVN::Git::Fetcher->new($self);
1635                        $gs->ra->gs_do_switch($r0, $rev, $gs,
1636                                              $self->full_url, $ed)
1637                          or die "SVN connection failed somewhere...\n";
1638                } else {
1639                        print STDERR "Following parent with do_update\n";
1640                        $ed = SVN::Git::Fetcher->new($self);
1641                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
1642                          or die "SVN connection failed somewhere...\n";
1643                }
1644                print STDERR "Successfully followed parent\n";
1645                return $self->make_log_entry($rev, [$parent], $ed);
1646        }
1647        return undef;
1648}
1649
1650sub do_fetch {
1651        my ($self, $paths, $rev) = @_;
1652        my $ed;
1653        my ($last_rev, @parents);
1654        if (my $lc = $self->last_commit) {
1655                # we can have a branch that was deleted, then re-added
1656                # under the same name but copied from another path, in
1657                # which case we'll have multiple parents (we don't
1658                # want to break the original ref, nor lose copypath info):
1659                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1660                        push @{$log_entry->{parents}}, $lc;
1661                        return $log_entry;
1662                }
1663                $ed = SVN::Git::Fetcher->new($self);
1664                $last_rev = $self->{last_rev};
1665                $ed->{c} = $lc;
1666                @parents = ($lc);
1667        } else {
1668                $last_rev = $rev;
1669                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1670                        return $log_entry;
1671                }
1672                $ed = SVN::Git::Fetcher->new($self);
1673        }
1674        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1675                die "SVN connection failed somewhere...\n";
1676        }
1677        $self->make_log_entry($rev, \@parents, $ed);
1678}
1679
1680sub get_untracked {
1681        my ($self, $ed) = @_;
1682        my @out;
1683        my $h = $ed->{empty};
1684        foreach (sort keys %$h) {
1685                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1686                push @out, "  $act: " . uri_encode($_);
1687                warn "W: $act: $_\n";
1688        }
1689        foreach my $t (qw/dir_prop file_prop/) {
1690                $h = $ed->{$t} or next;
1691                foreach my $path (sort keys %$h) {
1692                        my $ppath = $path eq '' ? '.' : $path;
1693                        foreach my $prop (sort keys %{$h->{$path}}) {
1694                                next if $SKIP_PROP{$prop};
1695                                my $v = $h->{$path}->{$prop};
1696                                my $t_ppath_prop = "$t: " .
1697                                                    uri_encode($ppath) . ' ' .
1698                                                    uri_encode($prop);
1699                                if (defined $v) {
1700                                        push @out, "  +$t_ppath_prop " .
1701                                                   uri_encode($v);
1702                                } else {
1703                                        push @out, "  -$t_ppath_prop";
1704                                }
1705                        }
1706                }
1707        }
1708        foreach my $t (qw/absent_file absent_directory/) {
1709                $h = $ed->{$t} or next;
1710                foreach my $parent (sort keys %$h) {
1711                        foreach my $path (sort @{$h->{$parent}}) {
1712                                push @out, "  $t: " .
1713                                           uri_encode("$parent/$path");
1714                                warn "W: $t: $parent/$path ",
1715                                     "Insufficient permissions?\n";
1716                        }
1717                }
1718        }
1719        \@out;
1720}
1721
1722sub parse_svn_date {
1723        my $date = shift || return '+0000 1970-01-01 00:00:00';
1724        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1725                                            (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1726                                         croak "Unable to parse date: $date\n";
1727        "+0000 $Y-$m-$d $H:$M:$S";
1728}
1729
1730sub check_author {
1731        my ($author) = @_;
1732        if (!defined $author || length $author == 0) {
1733                $author = '(no author)';
1734        }
1735        if (defined $::_authors && ! defined $::users{$author}) {
1736                die "Author: $author not defined in $::_authors file\n";
1737        }
1738        $author;
1739}
1740
1741sub make_log_entry {
1742        my ($self, $rev, $parents, $ed) = @_;
1743        my $untracked = $self->get_untracked($ed);
1744
1745        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1746        print $un "r$rev\n" or croak $!;
1747        print $un $_, "\n" foreach @$untracked;
1748        my %log_entry = ( parents => $parents || [], revision => $rev,
1749                          log => '');
1750
1751        my $headrev;
1752        my $logged = delete $self->{logged_rev_props};
1753        if (!$logged || $self->{-want_revprops}) {
1754                my $rp = $self->ra->rev_proplist($rev);
1755                foreach (sort keys %$rp) {
1756                        my $v = $rp->{$_};
1757                        if (/^svn:(author|date|log)$/) {
1758                                $log_entry{$1} = $v;
1759                        } elsif ($_ eq 'svm:headrev') {
1760                                $headrev = $v;
1761                        } else {
1762                                print $un "  rev_prop: ", uri_encode($_), ' ',
1763                                          uri_encode($v), "\n";
1764                        }
1765                }
1766        } else {
1767                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1768        }
1769        close $un or croak $!;
1770
1771        $log_entry{date} = parse_svn_date($log_entry{date});
1772        $log_entry{log} .= "\n";
1773        my $author = $log_entry{author} = check_author($log_entry{author});
1774        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1775                                                       : ($author, undef);
1776        if (defined $headrev && $self->use_svm_props) {
1777                if ($self->rewrite_root) {
1778                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
1779                            "options set!\n";
1780                }
1781                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1782                if ($uuid ne $self->{svm}->{uuid}) {
1783                        die "UUID mismatch on SVM path:\n",
1784                            "expected: $self->{svm}->{uuid}\n",
1785                            "     got: $uuid\n";
1786                }
1787                my $full_url = $self->{svm}->{source};
1788                $full_url .= "/$self->{path}" if length $self->{path};
1789                $log_entry{metadata} = "$full_url\@$r $uuid";
1790                $log_entry{svm_revision} = $r;
1791                $email ||= "$author\@$uuid"
1792        } elsif ($self->use_svnsync_props) {
1793                my $full_url = $self->svnsync->{url};
1794                $full_url .= "/$self->{path}" if length $self->{path};
1795                my $uuid = $self->svnsync->{uuid};
1796                $log_entry{metadata} = "$full_url\@$rev $uuid";
1797                $email ||= "$author\@$uuid"
1798        } else {
1799                $log_entry{metadata} = $self->metadata_url. "\@$rev " .
1800                                       $self->ra->get_uuid;
1801                $email ||= "$author\@" . $self->ra->get_uuid;
1802        }
1803        $log_entry{name} = $name;
1804        $log_entry{email} = $email;
1805        \%log_entry;
1806}
1807
1808sub fetch {
1809        my ($self, $min_rev, $max_rev, @parents) = @_;
1810        my ($last_rev, $last_commit) = $self->last_rev_commit;
1811        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1812        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1813}
1814
1815sub set_tree_cb {
1816        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1817        $self->{inject_parents} = { $rev => $tree };
1818        $self->fetch(undef, undef);
1819}
1820
1821sub set_tree {
1822        my ($self, $tree) = (shift, shift);
1823        my $log_entry = ::get_commit_entry($tree);
1824        unless ($self->{last_rev}) {
1825                fatal("Must have an existing revision to commit\n");
1826        }
1827        my %ed_opts = ( r => $self->{last_rev},
1828                        log => $log_entry->{log},
1829                        ra => $self->ra,
1830                        tree_a => $self->{last_commit},
1831                        tree_b => $tree,
1832                        editor_cb => sub {
1833                               $self->set_tree_cb($log_entry, $tree, @_) },
1834                        svn_path => $self->{path} );
1835        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1836                print "No changes\nr$self->{last_rev} = $tree\n";
1837        }
1838}
1839
1840sub rebuild {
1841        my ($self) = @_;
1842        my $db_path = $self->db_path;
1843        return if (-e $db_path && ! -z $db_path);
1844        return unless ::verify_ref($self->refname.'^0');
1845        if (-f $self->{db_root}) {
1846                rename $self->{db_root}, $db_path or die
1847                     "rename $self->{db_root} => $db_path failed: $!\n";
1848                my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
1849                symlink $base, $self->{db_root} or die
1850                     "symlink $base => $self->{db_root} failed: $!\n";
1851                return;
1852        }
1853        print "Rebuilding $db_path ...\n";
1854        my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1855        my $latest;
1856        my $full_url = $self->full_url;
1857        my $svn_uuid;
1858        while (<$rev_list>) {
1859                chomp;
1860                my $c = $_;
1861                die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1862                my ($url, $rev, $uuid) = ::cmt_metadata($c);
1863
1864                # ignore merges (from set-tree)
1865                next if (!defined $rev || !$uuid);
1866
1867                # if we merged or otherwise started elsewhere, this is
1868                # how we break out of it
1869                if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1870                    ($full_url && $url && ($url ne $full_url))) {
1871                        next;
1872                }
1873                $latest ||= $rev;
1874                $svn_uuid ||= $uuid;
1875
1876                $self->rev_db_set($rev, $c);
1877                print "r$rev = $c\n";
1878        }
1879        command_close_pipe($rev_list, $ctx);
1880        print "Done rebuilding $db_path\n";
1881}
1882
1883# rev_db:
1884# Tie::File seems to be prone to offset errors if revisions get sparse,
1885# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
1886# one of my favorite modules is out :<  Next up would be one of the DBM
1887# modules, but I'm not sure which is most portable...  So I'll just
1888# go with something that's plain-text, but still capable of
1889# being randomly accessed.  So here's my ultra-simple fixed-width
1890# database.  All records are 40 characters + "\n", so it's easy to seek
1891# to a revision: (41 * rev) is the byte offset.
1892# A record of 40 0s denotes an empty revision.
1893# And yes, it's still pretty fast (faster than Tie::File).
1894# These files are disposable unless noMetadata or useSvmProps is set
1895
1896sub _rev_db_set {
1897        my ($fh, $rev, $commit) = @_;
1898        my $offset = $rev * 41;
1899        # assume that append is the common case:
1900        seek $fh, 0, 2 or croak $!;
1901        my $pos = tell $fh;
1902        if ($pos < $offset) {
1903                for (1 .. (($offset - $pos) / 41)) {
1904                        print $fh (('0' x 40),"\n") or croak $!;
1905                }
1906        }
1907        seek $fh, $offset, 0 or croak $!;
1908        print $fh $commit,"\n" or croak $!;
1909}
1910
1911sub mkfile {
1912        my ($path) = @_;
1913        unless (-e $path) {
1914                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
1915                mkpath([$dir]) unless -d $dir;
1916                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
1917                close $fh or die "Couldn't close (create) $path: $!\n";
1918        }
1919}
1920
1921sub rev_db_set {
1922        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
1923        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
1924        my $db = $self->db_path($uuid);
1925        my $db_lock = "$db.lock";
1926        my $sig;
1927        if ($update_ref) {
1928                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1929                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
1930        }
1931        mkfile($db);
1932
1933        $LOCKFILES{$db_lock} = 1;
1934        my $sync;
1935        # both of these options make our .rev_db file very, very important
1936        # and we can't afford to lose it because rebuild() won't work
1937        if ($self->use_svm_props || $self->no_metadata) {
1938                $sync = 1;
1939                copy($db, $db_lock) or die "rev_db_set(@_): ",
1940                                           "Failed to copy: ",
1941                                           "$db => $db_lock ($!)\n";
1942        } else {
1943                rename $db, $db_lock or die "rev_db_set(@_): ",
1944                                            "Failed to rename: ",
1945                                            "$db => $db_lock ($!)\n";
1946        }
1947        open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
1948        _rev_db_set($fh, $rev, $commit);
1949        if ($sync) {
1950                $fh->flush or die "Couldn't flush $db_lock: $!\n";
1951                $fh->sync or die "Couldn't sync $db_lock: $!\n";
1952        }
1953        close $fh or croak $!;
1954        if ($update_ref) {
1955                $_head = $self;
1956                command_noisy('update-ref', '-m', "r$rev",
1957                              $self->refname, $commit);
1958        }
1959        rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
1960                                    "$db_lock => $db ($!)\n";
1961        delete $LOCKFILES{$db_lock};
1962        if ($update_ref) {
1963                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1964                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
1965                kill $sig, $$ if defined $sig;
1966        }
1967}
1968
1969sub rev_db_max {
1970        my ($self) = @_;
1971        $self->rebuild;
1972        my $db_path = $self->db_path;
1973        my @stat = stat $db_path or return 0;
1974        ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
1975        my $max = $stat[7] / 41;
1976        (($max > 0) ? $max - 1 : 0);
1977}
1978
1979sub rev_db_get {
1980        my ($self, $rev, $uuid) = @_;
1981        my $ret;
1982        my $offset = $rev * 41;
1983        my $db_path = $self->db_path($uuid);
1984        return undef unless -e $db_path;
1985        open my $fh, '<', $db_path or croak $!;
1986        if (sysseek($fh, $offset, 0) == $offset) {
1987                my $read = sysread($fh, $ret, 40);
1988                $ret = undef if ($read != 40 || $ret eq ('0'x40));
1989        }
1990        close $fh or croak $!;
1991        $ret;
1992}
1993
1994sub find_rev_before {
1995        my ($self, $rev, $eq_ok) = @_;
1996        --$rev unless $eq_ok;
1997        while ($rev > 0) {
1998                if (my $c = $self->rev_db_get($rev)) {
1999                        return ($rev, $c);
2000                }
2001                --$rev;
2002        }
2003        return (undef, undef);
2004}
2005
2006sub _new {
2007        my ($class, $repo_id, $ref_id, $path) = @_;
2008        unless (defined $repo_id && length $repo_id) {
2009                $repo_id = $Git::SVN::default_repo_id;
2010        }
2011        unless (defined $ref_id && length $ref_id) {
2012                $_[2] = $ref_id = $Git::SVN::default_ref_id;
2013        }
2014        $_[1] = $repo_id = sanitize_remote_name($repo_id);
2015        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2016        $_[3] = $path = '' unless (defined $path);
2017        mkpath(["$ENV{GIT_DIR}/svn"]);
2018        bless {
2019                ref_id => $ref_id, dir => $dir, index => "$dir/index",
2020                path => $path, config => "$ENV{GIT_DIR}/svn/config",
2021                db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2022}
2023
2024sub db_path {
2025        my ($self, $uuid) = @_;
2026        $uuid ||= $self->ra_uuid;
2027        "$self->{db_root}.$uuid";
2028}
2029
2030sub uri_encode {
2031        my ($f) = @_;
2032        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2033        $f
2034}
2035
2036package Git::SVN::Prompt;
2037use strict;
2038use warnings;
2039require SVN::Core;
2040use vars qw/$_no_auth_cache $_username/;
2041
2042sub simple {
2043        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2044        $may_save = undef if $_no_auth_cache;
2045        $default_username = $_username if defined $_username;
2046        if (defined $default_username && length $default_username) {
2047                if (defined $realm && length $realm) {
2048                        print STDERR "Authentication realm: $realm\n";
2049                        STDERR->flush;
2050                }
2051                $cred->username($default_username);
2052        } else {
2053                username($cred, $realm, $may_save, $pool);
2054        }
2055        $cred->password(_read_password("Password for '" .
2056                                       $cred->username . "': ", $realm));
2057        $cred->may_save($may_save);
2058        $SVN::_Core::SVN_NO_ERROR;
2059}
2060
2061sub ssl_server_trust {
2062        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2063        $may_save = undef if $_no_auth_cache;
2064        print STDERR "Error validating server certificate for '$realm':\n";
2065        if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2066                print STDERR " - The certificate is not issued by a trusted ",
2067                      "authority. Use the\n",
2068                      "   fingerprint to validate the certificate manually!\n";
2069        }
2070        if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2071                print STDERR " - The certificate hostname does not match.\n";
2072        }
2073        if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2074                print STDERR " - The certificate is not yet valid.\n";
2075        }
2076        if ($failures & $SVN::Auth::SSL::EXPIRED) {
2077                print STDERR " - The certificate has expired.\n";
2078        }
2079        if ($failures & $SVN::Auth::SSL::OTHER) {
2080                print STDERR " - The certificate has an unknown error.\n";
2081        }
2082        printf STDERR
2083                "Certificate information:\n".
2084                " - Hostname: %s\n".
2085                " - Valid: from %s until %s\n".
2086                " - Issuer: %s\n".
2087                " - Fingerprint: %s\n",
2088                map $cert_info->$_, qw(hostname valid_from valid_until
2089                                       issuer_dname fingerprint);
2090        my $choice;
2091prompt:
2092        print STDERR $may_save ?
2093              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2094              "(R)eject or accept (t)emporarily? ";
2095        STDERR->flush;
2096        $choice = lc(substr(<STDIN> || 'R', 0, 1));
2097        if ($choice =~ /^t$/i) {
2098                $cred->may_save(undef);
2099        } elsif ($choice =~ /^r$/i) {
2100                return -1;
2101        } elsif ($may_save && $choice =~ /^p$/i) {
2102                $cred->may_save($may_save);
2103        } else {
2104                goto prompt;
2105        }
2106        $cred->accepted_failures($failures);
2107        $SVN::_Core::SVN_NO_ERROR;
2108}
2109
2110sub ssl_client_cert {
2111        my ($cred, $realm, $may_save, $pool) = @_;
2112        $may_save = undef if $_no_auth_cache;
2113        print STDERR "Client certificate filename: ";
2114        STDERR->flush;
2115        chomp(my $filename = <STDIN>);
2116        $cred->cert_file($filename);
2117        $cred->may_save($may_save);
2118        $SVN::_Core::SVN_NO_ERROR;
2119}
2120
2121sub ssl_client_cert_pw {
2122        my ($cred, $realm, $may_save, $pool) = @_;
2123        $may_save = undef if $_no_auth_cache;
2124        $cred->password(_read_password("Password: ", $realm));
2125        $cred->may_save($may_save);
2126        $SVN::_Core::SVN_NO_ERROR;
2127}
2128
2129sub username {
2130        my ($cred, $realm, $may_save, $pool) = @_;
2131        $may_save = undef if $_no_auth_cache;
2132        if (defined $realm && length $realm) {
2133                print STDERR "Authentication realm: $realm\n";
2134        }
2135        my $username;
2136        if (defined $_username) {
2137                $username = $_username;
2138        } else {
2139                print STDERR "Username: ";
2140                STDERR->flush;
2141                chomp($username = <STDIN>);
2142        }
2143        $cred->username($username);
2144        $cred->may_save($may_save);
2145        $SVN::_Core::SVN_NO_ERROR;
2146}
2147
2148sub _read_password {
2149        my ($prompt, $realm) = @_;
2150        print STDERR $prompt;
2151        STDERR->flush;
2152        require Term::ReadKey;
2153        Term::ReadKey::ReadMode('noecho');
2154        my $password = '';
2155        while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2156                last if $key =~ /[\012\015]/; # \n\r
2157                $password .= $key;
2158        }
2159        Term::ReadKey::ReadMode('restore');
2160        print STDERR "\n";
2161        STDERR->flush;
2162        $password;
2163}
2164
2165package main;
2166
2167{
2168        my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2169                                $SVN::Node::dir.$SVN::Node::unknown.
2170                                $SVN::Node::none.$SVN::Node::file.
2171                                $SVN::Node::dir.$SVN::Node::unknown.
2172                                $SVN::Auth::SSL::CNMISMATCH.
2173                                $SVN::Auth::SSL::NOTYETVALID.
2174                                $SVN::Auth::SSL::EXPIRED.
2175                                $SVN::Auth::SSL::UNKNOWNCA.
2176                                $SVN::Auth::SSL::OTHER;
2177}
2178
2179package SVN::Git::Fetcher;
2180use vars qw/@ISA/;
2181use strict;
2182use warnings;
2183use Carp qw/croak/;
2184use IO::File qw//;
2185use Digest::MD5;
2186
2187# file baton members: path, mode_a, mode_b, pool, fh, blob, base
2188sub new {
2189        my ($class, $git_svn) = @_;
2190        my $self = SVN::Delta::Editor->new;
2191        bless $self, $class;
2192        $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2193        $self->{empty} = {};
2194        $self->{dir_prop} = {};
2195        $self->{file_prop} = {};
2196        $self->{absent_dir} = {};
2197        $self->{absent_file} = {};
2198        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2199        $self;
2200}
2201
2202sub set_path_strip {
2203        my ($self, $path) = @_;
2204        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2205}
2206
2207sub open_root {
2208        { path => '' };
2209}
2210
2211sub open_directory {
2212        my ($self, $path, $pb, $rev) = @_;
2213        { path => $path };
2214}
2215
2216sub git_path {
2217        my ($self, $path) = @_;
2218        if ($self->{path_strip}) {
2219                $path =~ s!$self->{path_strip}!! or
2220                  die "Failed to strip path '$path' ($self->{path_strip})\n";
2221        }
2222        $path;
2223}
2224
2225sub delete_entry {
2226        my ($self, $path, $rev, $pb) = @_;
2227
2228        my $gpath = $self->git_path($path);
2229        return undef if ($gpath eq '');
2230
2231        # remove entire directories.
2232        if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2233                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2234                                                     -r --name-only -z/,
2235                                                     $self->{c}, '--', $gpath);
2236                local $/ = "\0";
2237                while (<$ls>) {
2238                        chomp;
2239                        $self->{gii}->remove($_);
2240                        print "\tD\t$_\n" unless $::_q;
2241                }
2242                print "\tD\t$gpath/\n" unless $::_q;
2243                command_close_pipe($ls, $ctx);
2244                $self->{empty}->{$path} = 0
2245        } else {
2246                $self->{gii}->remove($gpath);
2247                print "\tD\t$gpath\n" unless $::_q;
2248        }
2249        undef;
2250}
2251
2252sub open_file {
2253        my ($self, $path, $pb, $rev) = @_;
2254        my $gpath = $self->git_path($path);
2255        my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2256                             =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2257        unless (defined $mode && defined $blob) {
2258                die "$path was not found in commit $self->{c} (r$rev)\n";
2259        }
2260        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2261          pool => SVN::Pool->new, action => 'M' };
2262}
2263
2264sub add_file {
2265        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2266        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2267        delete $self->{empty}->{$dir};
2268        { path => $path, mode_a => 100644, mode_b => 100644,
2269          pool => SVN::Pool->new, action => 'A' };
2270}
2271
2272sub add_directory {
2273        my ($self, $path, $cp_path, $cp_rev) = @_;
2274        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2275        delete $self->{empty}->{$dir};
2276        $self->{empty}->{$path} = 1;
2277        { path => $path };
2278}
2279
2280sub change_dir_prop {
2281        my ($self, $db, $prop, $value) = @_;
2282        $self->{dir_prop}->{$db->{path}} ||= {};
2283        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2284        undef;
2285}
2286
2287sub absent_directory {
2288        my ($self, $path, $pb) = @_;
2289        $self->{absent_dir}->{$pb->{path}} ||= [];
2290        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2291        undef;
2292}
2293
2294sub absent_file {
2295        my ($self, $path, $pb) = @_;
2296        $self->{absent_file}->{$pb->{path}} ||= [];
2297        push @{$self->{absent_file}->{$pb->{path}}}, $path;
2298        undef;
2299}
2300
2301sub change_file_prop {
2302        my ($self, $fb, $prop, $value) = @_;
2303        if ($prop eq 'svn:executable') {
2304                if ($fb->{mode_b} != 120000) {
2305                        $fb->{mode_b} = defined $value ? 100755 : 100644;
2306                }
2307        } elsif ($prop eq 'svn:special') {
2308                $fb->{mode_b} = defined $value ? 120000 : 100644;
2309        } else {
2310                $self->{file_prop}->{$fb->{path}} ||= {};
2311                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2312        }
2313        undef;
2314}
2315
2316sub apply_textdelta {
2317        my ($self, $fb, $exp) = @_;
2318        my $fh = IO::File->new_tmpfile;
2319        $fh->autoflush(1);
2320        # $fh gets auto-closed() by SVN::TxDelta::apply(),
2321        # (but $base does not,) so dup() it for reading in close_file
2322        open my $dup, '<&', $fh or croak $!;
2323        my $base = IO::File->new_tmpfile;
2324        $base->autoflush(1);
2325        if ($fb->{blob}) {
2326                defined (my $pid = fork) or croak $!;
2327                if (!$pid) {
2328                        open STDOUT, '>&', $base or croak $!;
2329                        print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2330                        exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2331                }
2332                waitpid $pid, 0;
2333                croak $? if $?;
2334
2335                if (defined $exp) {
2336                        seek $base, 0, 0 or croak $!;
2337                        my $md5 = Digest::MD5->new;
2338                        $md5->addfile($base);
2339                        my $got = $md5->hexdigest;
2340                        die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2341                            "expected: $exp\n",
2342                            "     got: $got\n" if ($got ne $exp);
2343                }
2344        }
2345        seek $base, 0, 0 or croak $!;
2346        $fb->{fh} = $dup;
2347        $fb->{base} = $base;
2348        [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2349}
2350
2351sub close_file {
2352        my ($self, $fb, $exp) = @_;
2353        my $hash;
2354        my $path = $self->git_path($fb->{path});
2355        if (my $fh = $fb->{fh}) {
2356                seek($fh, 0, 0) or croak $!;
2357                my $md5 = Digest::MD5->new;
2358                $md5->addfile($fh);
2359                my $got = $md5->hexdigest;
2360                die "Checksum mismatch: $path\n",
2361                    "expected: $exp\n    got: $got\n" if ($got ne $exp);
2362                seek($fh, 0, 0) or croak $!;
2363                if ($fb->{mode_b} == 120000) {
2364                        read($fh, my $buf, 5) == 5 or croak $!;
2365                        $buf eq 'link ' or die "$path has mode 120000",
2366                                               "but is not a link\n";
2367                }
2368                defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2369                if (!$pid) {
2370                        open STDIN, '<&', $fh or croak $!;
2371                        exec qw/git-hash-object -w --stdin/ or croak $!;
2372                }
2373                chomp($hash = do { local $/; <$out> });
2374                close $out or croak $!;
2375                close $fh or croak $!;
2376                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2377                close $fb->{base} or croak $!;
2378        } else {
2379                $hash = $fb->{blob} or die "no blob information\n";
2380        }
2381        $fb->{pool}->clear;
2382        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2383        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2384        undef;
2385}
2386
2387sub abort_edit {
2388        my $self = shift;
2389        $self->{nr} = $self->{gii}->{nr};
2390        delete $self->{gii};
2391        $self->SUPER::abort_edit(@_);
2392}
2393
2394sub close_edit {
2395        my $self = shift;
2396        $self->{git_commit_ok} = 1;
2397        $self->{nr} = $self->{gii}->{nr};
2398        delete $self->{gii};
2399        $self->SUPER::close_edit(@_);
2400}
2401
2402package SVN::Git::Editor;
2403use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2404use strict;
2405use warnings;
2406use Carp qw/croak/;
2407use IO::File;
2408use Digest::MD5;
2409
2410sub new {
2411        my ($class, $opts) = @_;
2412        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2413                die "$_ required!\n" unless (defined $opts->{$_});
2414        }
2415
2416        my $pool = SVN::Pool->new;
2417        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2418        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2419                                     $opts->{r}, $mods);
2420
2421        # $opts->{ra} functions should not be used after this:
2422        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
2423                                                $opts->{editor_cb}, $pool);
2424        my $self = SVN::Delta::Editor->new(@ce, $pool);
2425        bless $self, $class;
2426        foreach (qw/svn_path r tree_a tree_b/) {
2427                $self->{$_} = $opts->{$_};
2428        }
2429        $self->{url} = $opts->{ra}->{url};
2430        $self->{mods} = $mods;
2431        $self->{types} = $types;
2432        $self->{pool} = $pool;
2433        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2434        $self->{rm} = { };
2435        $self->{path_prefix} = length $self->{svn_path} ?
2436                               "$self->{svn_path}/" : '';
2437        return $self;
2438}
2439
2440sub generate_diff {
2441        my ($tree_a, $tree_b) = @_;
2442        my @diff_tree = qw(diff-tree -z -r);
2443        if ($_cp_similarity) {
2444                push @diff_tree, "-C$_cp_similarity";
2445        } else {
2446                push @diff_tree, '-C';
2447        }
2448        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2449        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2450        push @diff_tree, $tree_a, $tree_b;
2451        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2452        local $/ = "\0";
2453        my $state = 'meta';
2454        my @mods;
2455        while (<$diff_fh>) {
2456                chomp $_; # this gets rid of the trailing "\0"
2457                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2458                                        $::sha1\s($::sha1)\s
2459                                        ([MTCRAD])\d*$/xo) {
2460                        push @mods, {   mode_a => $1, mode_b => $2,
2461                                        sha1_b => $3, chg => $4 };
2462                        if ($4 =~ /^(?:C|R)$/) {
2463                                $state = 'file_a';
2464                        } else {
2465                                $state = 'file_b';
2466                        }
2467                } elsif ($state eq 'file_a') {
2468                        my $x = $mods[$#mods] or croak "Empty array\n";
2469                        if ($x->{chg} !~ /^(?:C|R)$/) {
2470                                croak "Error parsing $_, $x->{chg}\n";
2471                        }
2472                        $x->{file_a} = $_;
2473                        $state = 'file_b';
2474                } elsif ($state eq 'file_b') {
2475                        my $x = $mods[$#mods] or croak "Empty array\n";
2476                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2477                                croak "Error parsing $_, $x->{chg}\n";
2478                        }
2479                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2480                                croak "Error parsing $_, $x->{chg}\n";
2481                        }
2482                        $x->{file_b} = $_;
2483                        $state = 'meta';
2484                } else {
2485                        croak "Error parsing $_\n";
2486                }
2487        }
2488        command_close_pipe($diff_fh, $ctx);
2489        \@mods;
2490}
2491
2492sub check_diff_paths {
2493        my ($ra, $pfx, $rev, $mods) = @_;
2494        my %types;
2495        $pfx .= '/' if length $pfx;
2496
2497        sub type_diff_paths {
2498                my ($ra, $types, $path, $rev) = @_;
2499                my @p = split m#/+#, $path;
2500                my $c = shift @p;
2501                unless (defined $types->{$c}) {
2502                        $types->{$c} = $ra->check_path($c, $rev);
2503                }
2504                while (@p) {
2505                        $c .= '/' . shift @p;
2506                        next if defined $types->{$c};
2507                        $types->{$c} = $ra->check_path($c, $rev);
2508                }
2509        }
2510
2511        foreach my $m (@$mods) {
2512                foreach my $f (qw/file_a file_b/) {
2513                        next unless defined $m->{$f};
2514                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2515                        if (length $pfx.$dir && ! defined $types{$dir}) {
2516                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2517                        }
2518                }
2519        }
2520        \%types;
2521}
2522
2523sub split_path {
2524        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2525}
2526
2527sub repo_path {
2528        my ($self, $path) = @_;
2529        $self->{path_prefix}.(defined $path ? $path : '');
2530}
2531
2532sub url_path {
2533        my ($self, $path) = @_;
2534        $self->{url} . '/' . $self->repo_path($path);
2535}
2536
2537sub rmdirs {
2538        my ($self) = @_;
2539        my $rm = $self->{rm};
2540        delete $rm->{''}; # we never delete the url we're tracking
2541        return unless %$rm;
2542
2543        foreach (keys %$rm) {
2544                my @d = split m#/#, $_;
2545                my $c = shift @d;
2546                $rm->{$c} = 1;
2547                while (@d) {
2548                        $c .= '/' . shift @d;
2549                        $rm->{$c} = 1;
2550                }
2551        }
2552        delete $rm->{$self->{svn_path}};
2553        delete $rm->{''}; # we never delete the url we're tracking
2554        return unless %$rm;
2555
2556        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2557                                             $self->{tree_b});
2558        local $/ = "\0";
2559        while (<$fh>) {
2560                chomp;
2561                my @dn = split m#/#, $_;
2562                while (pop @dn) {
2563                        delete $rm->{join '/', @dn};
2564                }
2565                unless (%$rm) {
2566                        close $fh;
2567                        return;
2568                }
2569        }
2570        command_close_pipe($fh, $ctx);
2571
2572        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2573        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2574                $self->close_directory($bat->{$d}, $p);
2575                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2576                print "\tD+\t$d/\n" unless $::_q;
2577                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2578                delete $bat->{$d};
2579        }
2580}
2581
2582sub open_or_add_dir {
2583        my ($self, $full_path, $baton) = @_;
2584        my $t = $self->{types}->{$full_path};
2585        if (!defined $t) {
2586                die "$full_path not known in r$self->{r} or we have a bug!\n";
2587        }
2588        if ($t == $SVN::Node::none) {
2589                return $self->add_directory($full_path, $baton,
2590                                                undef, -1, $self->{pool});
2591        } elsif ($t == $SVN::Node::dir) {
2592                return $self->open_directory($full_path, $baton,
2593                                                $self->{r}, $self->{pool});
2594        }
2595        print STDERR "$full_path already exists in repository at ",
2596                "r$self->{r} and it is not a directory (",
2597                ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2598        exit 1;
2599}
2600
2601sub ensure_path {
2602        my ($self, $path) = @_;
2603        my $bat = $self->{bat};
2604        my $repo_path = $self->repo_path($path);
2605        return $bat->{''} unless (length $repo_path);
2606        my @p = split m#/+#, $repo_path;
2607        my $c = shift @p;
2608        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2609        while (@p) {
2610                my $c0 = $c;
2611                $c .= '/' . shift @p;
2612                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2613        }
2614        return $bat->{$c};
2615}
2616
2617sub A {
2618        my ($self, $m) = @_;
2619        my ($dir, $file) = split_path($m->{file_b});
2620        my $pbat = $self->ensure_path($dir);
2621        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2622                                        undef, -1);
2623        print "\tA\t$m->{file_b}\n" unless $::_q;
2624        $self->chg_file($fbat, $m);
2625        $self->close_file($fbat,undef,$self->{pool});
2626}
2627
2628sub C {
2629        my ($self, $m) = @_;
2630        my ($dir, $file) = split_path($m->{file_b});
2631        my $pbat = $self->ensure_path($dir);
2632        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2633                                $self->url_path($m->{file_a}), $self->{r});
2634        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2635        $self->chg_file($fbat, $m);
2636        $self->close_file($fbat,undef,$self->{pool});
2637}
2638
2639sub delete_entry {
2640        my ($self, $path, $pbat) = @_;
2641        my $rpath = $self->repo_path($path);
2642        my ($dir, $file) = split_path($rpath);
2643        $self->{rm}->{$dir} = 1;
2644        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2645}
2646
2647sub R {
2648        my ($self, $m) = @_;
2649        my ($dir, $file) = split_path($m->{file_b});
2650        my $pbat = $self->ensure_path($dir);
2651        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2652                                $self->url_path($m->{file_a}), $self->{r});
2653        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2654        $self->chg_file($fbat, $m);
2655        $self->close_file($fbat,undef,$self->{pool});
2656
2657        ($dir, $file) = split_path($m->{file_a});
2658        $pbat = $self->ensure_path($dir);
2659        $self->delete_entry($m->{file_a}, $pbat);
2660}
2661
2662sub M {
2663        my ($self, $m) = @_;
2664        my ($dir, $file) = split_path($m->{file_b});
2665        my $pbat = $self->ensure_path($dir);
2666        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2667                                $pbat,$self->{r},$self->{pool});
2668        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2669        $self->chg_file($fbat, $m);
2670        $self->close_file($fbat,undef,$self->{pool});
2671}
2672
2673sub T { shift->M(@_) }
2674
2675sub change_file_prop {
2676        my ($self, $fbat, $pname, $pval) = @_;
2677        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2678}
2679
2680sub chg_file {
2681        my ($self, $fbat, $m) = @_;
2682        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2683                $self->change_file_prop($fbat,'svn:executable','*');
2684        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2685                $self->change_file_prop($fbat,'svn:executable',undef);
2686        }
2687        my $fh = IO::File->new_tmpfile or croak $!;
2688        if ($m->{mode_b} =~ /^120/) {
2689                print $fh 'link ' or croak $!;
2690                $self->change_file_prop($fbat,'svn:special','*');
2691        } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2692                $self->change_file_prop($fbat,'svn:special',undef);
2693        }
2694        defined(my $pid = fork) or croak $!;
2695        if (!$pid) {
2696                open STDOUT, '>&', $fh or croak $!;
2697                exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2698        }
2699        waitpid $pid, 0;
2700        croak $? if $?;
2701        $fh->flush == 0 or croak $!;
2702        seek $fh, 0, 0 or croak $!;
2703
2704        my $md5 = Digest::MD5->new;
2705        $md5->addfile($fh) or croak $!;
2706        seek $fh, 0, 0 or croak $!;
2707
2708        my $exp = $md5->hexdigest;
2709        my $pool = SVN::Pool->new;
2710        my $atd = $self->apply_textdelta($fbat, undef, $pool);
2711        my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2712        die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2713        $pool->clear;
2714
2715        close $fh or croak $!;
2716}
2717
2718sub D {
2719        my ($self, $m) = @_;
2720        my ($dir, $file) = split_path($m->{file_b});
2721        my $pbat = $self->ensure_path($dir);
2722        print "\tD\t$m->{file_b}\n" unless $::_q;
2723        $self->delete_entry($m->{file_b}, $pbat);
2724}
2725
2726sub close_edit {
2727        my ($self) = @_;
2728        my ($p,$bat) = ($self->{pool}, $self->{bat});
2729        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2730                $self->close_directory($bat->{$_}, $p);
2731        }
2732        $self->SUPER::close_edit($p);
2733        $p->clear;
2734}
2735
2736sub abort_edit {
2737        my ($self) = @_;
2738        $self->SUPER::abort_edit($self->{pool});
2739}
2740
2741sub DESTROY {
2742        my $self = shift;
2743        $self->SUPER::DESTROY(@_);
2744        $self->{pool}->clear;
2745}
2746
2747# this drives the editor
2748sub apply_diff {
2749        my ($self) = @_;
2750        my $mods = $self->{mods};
2751        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2752        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2753                my $f = $m->{chg};
2754                if (defined $o{$f}) {
2755                        $self->$f($m);
2756                } else {
2757                        fatal("Invalid change type: $f\n");
2758                }
2759        }
2760        $self->rmdirs if $_rmdir;
2761        if (@$mods == 0) {
2762                $self->abort_edit;
2763        } else {
2764                $self->close_edit;
2765        }
2766        return scalar @$mods;
2767}
2768
2769package Git::SVN::Ra;
2770use vars qw/@ISA $config_dir $_log_window_size/;
2771use strict;
2772use warnings;
2773my ($can_do_switch);
2774my $RA;
2775
2776BEGIN {
2777        # enforce temporary pool usage for some simple functions
2778        my $e;
2779        foreach (qw/get_latest_revnum get_uuid get_repos_root/) {
2780                $e .= "sub $_ {
2781                        my \$self = shift;
2782                        my \$pool = SVN::Pool->new;
2783                        my \@ret = \$self->SUPER::$_(\@_,\$pool);
2784                        \$pool->clear;
2785                        wantarray ? \@ret : \$ret[0]; }\n";
2786        }
2787
2788        # get_dir needs $pool held in cache for dirents to work,
2789        # check_path is cacheable and rev_proplist is close enough
2790        # for our purposes.
2791        foreach (qw/check_path get_dir rev_proplist/) {
2792                $e .= "my \%${_}_cache; my \$${_}_rev = 0; sub $_ {
2793                        my \$self = shift;
2794                        my \$r = pop;
2795                        my \$k = join(\"\\0\", \@_);
2796                        if (my \$x = \$${_}_cache{\$r}->{\$k}) {
2797                                return wantarray ? \@\$x : \$x->[0];
2798                        }
2799                        my \$pool = SVN::Pool->new;
2800                        my \@ret = \$self->SUPER::$_(\@_, \$r, \$pool);
2801                        if (\$r != \$${_}_rev) {
2802                                \%${_}_cache = ( pool => [] );
2803                                \$${_}_rev = \$r;
2804                        }
2805                        \$${_}_cache{\$r}->{\$k} = \\\@ret;
2806                        push \@{\$${_}_cache{pool}}, \$pool;
2807                        wantarray ? \@ret : \$ret[0]; }\n";
2808        }
2809        $e .= "\n1;";
2810        eval $e or die $@;
2811}
2812
2813sub new {
2814        my ($class, $url) = @_;
2815        $url =~ s!/+$!!;
2816        return $RA if ($RA && $RA->{url} eq $url);
2817
2818        SVN::_Core::svn_config_ensure($config_dir, undef);
2819        my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2820            SVN::Client::get_simple_provider(),
2821            SVN::Client::get_ssl_server_trust_file_provider(),
2822            SVN::Client::get_simple_prompt_provider(
2823              \&Git::SVN::Prompt::simple, 2),
2824            SVN::Client::get_ssl_client_cert_prompt_provider(
2825              \&Git::SVN::Prompt::ssl_client_cert, 2),
2826            SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2827              \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2828            SVN::Client::get_username_provider(),
2829            SVN::Client::get_ssl_server_trust_prompt_provider(
2830              \&Git::SVN::Prompt::ssl_server_trust),
2831            SVN::Client::get_username_prompt_provider(
2832              \&Git::SVN::Prompt::username, 2),
2833          ]);
2834        my $config = SVN::Core::config_get_config($config_dir);
2835        my $self = SVN::Ra->new(url => $url, auth => $baton,
2836                              config => $config,
2837                              pool => SVN::Pool->new,
2838                              auth_provider_callbacks => $callbacks);
2839        $self->{svn_path} = $url;
2840        $self->{repos_root} = $self->get_repos_root;
2841        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
2842        $RA = bless $self, $class;
2843}
2844
2845sub DESTROY {
2846        # do not call the real DESTROY since we store ourselves in $RA
2847}
2848
2849sub get_log {
2850        my ($self, @args) = @_;
2851        my $pool = SVN::Pool->new;
2852        splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2853        my $ret = $self->SUPER::get_log(@args, $pool);
2854        $pool->clear;
2855        $ret;
2856}
2857
2858sub get_commit_editor {
2859        my ($self, $log, $cb, $pool) = @_;
2860        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2861        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2862}
2863
2864sub gs_do_update {
2865        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2866        my $new = ($rev_a == $rev_b);
2867        my $path = $gs->{path};
2868
2869        my $pool = SVN::Pool->new;
2870        $editor->set_path_strip($path);
2871        my (@pc) = split m#/#, $path;
2872        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2873                                        1, $editor, $pool);
2874        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2875
2876        # Since we can't rely on svn_ra_reparent being available, we'll
2877        # just have to do some magic with set_path to make it so
2878        # we only want a partial path.
2879        my $sp = '';
2880        my $final = join('/', @pc);
2881        while (@pc) {
2882                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2883                $sp .= '/' if length $sp;
2884                $sp .= shift @pc;
2885        }
2886        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2887
2888        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2889
2890        $reporter->finish_report($pool);
2891        $pool->clear;
2892        $editor->{git_commit_ok};
2893}
2894
2895# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2896# svn_ra_reparent didn't work before 1.4)
2897sub gs_do_switch {
2898        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
2899        my $path = $gs->{path};
2900        my $pool = SVN::Pool->new;
2901
2902        my $full_url = $self->{url};
2903        my $old_url = $full_url;
2904        $full_url .= "/$path" if length $path;
2905        my ($ra, $reparented);
2906        if ($old_url ne $full_url) {
2907                if ($old_url !~ m#^svn(\+ssh)?://#) {
2908                        SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2909                                                  $pool);
2910                        $self->{url} = $full_url;
2911                        $reparented = 1;
2912                } else {
2913                        $ra = Git::SVN::Ra->new($full_url);
2914                }
2915        }
2916        $ra ||= $self;
2917        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
2918        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2919        $reporter->set_path('', $rev_a, 0, @lock, $pool);
2920        $reporter->finish_report($pool);
2921
2922        if ($reparented) {
2923                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2924                $self->{url} = $old_url;
2925        }
2926
2927        $pool->clear;
2928        $editor->{git_commit_ok};
2929}
2930
2931sub gs_fetch_loop_common {
2932        my ($self, $base, $head, $gsv, $globs) = @_;
2933        return if ($base > $head);
2934        my $inc = $_log_window_size;
2935        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2936        my %common;
2937        my $common_max = scalar @$gsv;
2938
2939        foreach my $gs (@$gsv) {
2940                my @tmp = split m#/#, $gs->{path};
2941                my $p = '';
2942                foreach (@tmp) {
2943                        $p .= length($p) ? "/$_" : $_;
2944                        $common{$p} ||= 0;
2945                        $common{$p}++;
2946                }
2947        }
2948        $globs ||= [];
2949        $common_max += scalar @$globs;
2950        foreach my $glob (@$globs) {
2951                my @tmp = split m#/#, $glob->{path}->{left};
2952                my $p = '';
2953                foreach (@tmp) {
2954                        $p .= length($p) ? "/$_" : $_;
2955                        $common{$p} ||= 0;
2956                        $common{$p}++;
2957                }
2958        }
2959
2960        my $longest_path = '';
2961        foreach (sort {length $b <=> length $a} keys %common) {
2962                if ($common{$_} == $common_max) {
2963                        $longest_path = $_;
2964                        last;
2965                }
2966        }
2967        while (1) {
2968                my %revs;
2969                my $err;
2970                my $err_handler = $SVN::Error::handler;
2971                $SVN::Error::handler = sub {
2972                        ($err) = @_;
2973                        skip_unknown_revs($err);
2974                };
2975                sub _cb {
2976                        my ($paths, $r, $author, $date, $log) = @_;
2977                        [ dup_changed_paths($paths),
2978                          { author => $author, date => $date, log => $log } ];
2979                }
2980                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
2981                               sub { $revs{$_[1]} = _cb(@_) });
2982                if ($err && $max >= $head) {
2983                        print STDERR "Path '$longest_path' ",
2984                                     "was probably deleted:\n",
2985                                     $err->expanded_message,
2986                                     "\nWill attempt to follow ",
2987                                     "revisions r$min .. r$max ",
2988                                     "committed before the deletion\n";
2989                        my $hi = $max;
2990                        while (--$hi >= $min) {
2991                                my $ok;
2992                                $self->get_log([$longest_path], $min, $hi,
2993                                               0, 1, 1, sub {
2994                                               $ok ||= $_[1];
2995                                               $revs{$_[1]} = _cb(@_) });
2996                                if ($ok) {
2997                                        print STDERR "r$min .. r$ok OK\n";
2998                                        last;
2999                                }
3000                        }
3001                }
3002                $SVN::Error::handler = $err_handler;
3003
3004                my %exists = map { $_->{path} => $_ } @$gsv;
3005                foreach my $r (sort {$a <=> $b} keys %revs) {
3006                        my ($paths, $logged) = @{$revs{$r}};
3007
3008                        foreach my $gs ($self->match_globs(\%exists, $paths,
3009                                                           $globs, $r)) {
3010                                if ($gs->rev_db_max >= $r) {
3011                                        next;
3012                                }
3013                                next unless $gs->match_paths($paths, $r);
3014                                $gs->{logged_rev_props} = $logged;
3015                                if (my $last_commit = $gs->last_commit) {
3016                                        $gs->assert_index_clean($last_commit);
3017                                }
3018                                my $log_entry = $gs->do_fetch($paths, $r);
3019                                if ($log_entry) {
3020                                        $gs->do_git_commit($log_entry);
3021                                }
3022                        }
3023                        foreach my $g (@$globs) {
3024                                my $k = "svn-remote.$g->{remote}." .
3025                                        "$g->{t}-maxRev";
3026                                Git::SVN::tmp_config($k, $r);
3027                        }
3028                }
3029                # pre-fill the .rev_db since it'll eventually get filled in
3030                # with '0' x40 if something new gets committed
3031                foreach my $gs (@$gsv) {
3032                        next if defined $gs->rev_db_get($max);
3033                        $gs->rev_db_set($max, 0 x40);
3034                }
3035                foreach my $g (@$globs) {
3036                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3037                        Git::SVN::tmp_config($k, $max);
3038                }
3039                last if $max >= $head;
3040                $min = $max + 1;
3041                $max += $inc;
3042                $max = $head if ($max > $head);
3043        }
3044}
3045
3046sub match_globs {
3047        my ($self, $exists, $paths, $globs, $r) = @_;
3048
3049        sub get_dir_check {
3050                my ($self, $exists, $g, $r) = @_;
3051                my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3052                return unless scalar @x == 3;
3053                my $dirents = $x[0];
3054                foreach my $de (keys %$dirents) {
3055                        next if $dirents->{$de}->kind != $SVN::Node::dir;
3056                        my $p = $g->{path}->full_path($de);
3057                        next if $exists->{$p};
3058                        next if (length $g->{path}->{right} &&
3059                                 ($self->check_path($p, $r) !=
3060                                  $SVN::Node::dir));
3061                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3062                                         $g->{ref}->full_path($de), 1);
3063                }
3064        }
3065        foreach my $g (@$globs) {
3066                if (my $path = $paths->{"/$g->{path}->{left}"}) {
3067                        if ($path->{action} =~ /^[AR]$/) {
3068                                get_dir_check($self, $exists, $g, $r);
3069                        }
3070                }
3071                foreach (keys %$paths) {
3072                        if (/$g->{path}->{left_regex}/ &&
3073                            !/$g->{path}->{regex}/) {
3074                                next if $paths->{$_}->{action} !~ /^[AR]$/;
3075                                get_dir_check($self, $exists, $g, $r);
3076                        }
3077                        next unless /$g->{path}->{regex}/;
3078                        my $p = $1;
3079                        my $pathname = $g->{path}->full_path($p);
3080                        next if $exists->{$pathname};
3081                        $exists->{$pathname} = Git::SVN->init(
3082                                              $self->{url}, $pathname, undef,
3083                                              $g->{ref}->full_path($p), 1);
3084                }
3085                my $c = '';
3086                foreach (split m#/#, $g->{path}->{left}) {
3087                        $c .= "/$_";
3088                        next unless ($paths->{$c} &&
3089                                     ($paths->{$c}->{action} =~ /^[AR]$/));
3090                        get_dir_check($self, $exists, $g, $r);
3091                }
3092        }
3093        values %$exists;
3094}
3095
3096sub minimize_url {
3097        my ($self) = @_;
3098        return $self->{url} if ($self->{url} eq $self->{repos_root});
3099        my $url = $self->{repos_root};
3100        my @components = split(m!/!, $self->{svn_path});
3101        my $c = '';
3102        do {
3103                $url .= "/$c" if length $c;
3104                eval { (ref $self)->new($url)->get_latest_revnum };
3105        } while ($@ && ($c = shift @components));
3106        $url;
3107}
3108
3109sub can_do_switch {
3110        my $self = shift;
3111        unless (defined $can_do_switch) {
3112                my $pool = SVN::Pool->new;
3113                my $rep = eval {
3114                        $self->do_switch(1, '', 0, $self->{url},
3115                                         SVN::Delta::Editor->new, $pool);
3116                };
3117                if ($@) {
3118                        $can_do_switch = 0;
3119                } else {
3120                        $rep->abort_report($pool);
3121                        $can_do_switch = 1;
3122                }
3123                $pool->clear;
3124        }
3125        $can_do_switch;
3126}
3127
3128sub skip_unknown_revs {
3129        my ($err) = @_;
3130        my $errno = $err->apr_err();
3131        # Maybe the branch we're tracking didn't
3132        # exist when the repo started, so it's
3133        # not an error if it doesn't, just continue
3134        #
3135        # Wonderfully consistent library, eh?
3136        # 160013 - svn:// and file://
3137        # 175002 - http(s)://
3138        # 175007 - http(s):// (this repo required authorization, too...)
3139        #   More codes may be discovered later...
3140        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3141                warn "W: Ignoring error from SVN, path probably ",
3142                     "does not exist: ($errno): ",
3143                     $err->expanded_message,"\n";
3144                return;
3145        }
3146        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3147}
3148
3149# svn_log_changed_path_t objects passed to get_log are likely to be
3150# overwritten even if only the refs are copied to an external variable,
3151# so we should dup the structures in their entirety.  Using an externally
3152# passed pool (instead of our temporary and quickly cleared pool in
3153# Git::SVN::Ra) does not help matters at all...
3154sub dup_changed_paths {
3155        my ($paths) = @_;
3156        return undef unless $paths;
3157        my %ret;
3158        foreach my $p (keys %$paths) {
3159                my $i = $paths->{$p};
3160                my %s = map { $_ => $i->$_ }
3161                              qw/copyfrom_path copyfrom_rev action/;
3162                $ret{$p} = \%s;
3163        }
3164        \%ret;
3165}
3166
3167package Git::SVN::Log;
3168use strict;
3169use warnings;
3170use POSIX qw/strftime/;
3171use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3172            %rusers $show_commit $incremental/;
3173my $l_fmt;
3174
3175sub cmt_showable {
3176        my ($c) = @_;
3177        return 1 if defined $c->{r};
3178        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3179                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3180                my @log = command(qw/cat-file commit/, $c->{c});
3181                shift @log while ($log[0] ne "\n");
3182                shift @log;
3183                @{$c->{l}} = grep !/^git-svn-id: /, @log;
3184
3185                (undef, $c->{r}, undef) = ::extract_metadata(
3186                                (grep(/^git-svn-id: /, @log))[-1]);
3187        }
3188        return defined $c->{r};
3189}
3190
3191sub log_use_color {
3192        return 1 if $color;
3193        my ($dc, $dcvar);
3194        $dcvar = 'color.diff';
3195        $dc = `git-config --get $dcvar`;
3196        if ($dc eq '') {
3197                # nothing at all; fallback to "diff.color"
3198                $dcvar = 'diff.color';
3199                $dc = `git-config --get $dcvar`;
3200        }
3201        chomp($dc);
3202        if ($dc eq 'auto') {
3203                my $pc;
3204                $pc = `git-config --get color.pager`;
3205                if ($pc eq '') {
3206                        # does not have it -- fallback to pager.color
3207                        $pc = `git-config --bool --get pager.color`;
3208                }
3209                else {
3210                        $pc = `git-config --bool --get color.pager`;
3211                        if ($?) {
3212                                $pc = 'false';
3213                        }
3214                }
3215                chomp($pc);
3216                if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3217                        return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3218                }
3219                return 0;
3220        }
3221        return 0 if $dc eq 'never';
3222        return 1 if $dc eq 'always';
3223        chomp($dc = `git-config --bool --get $dcvar`);
3224        return ($dc eq 'true');
3225}
3226
3227sub git_svn_log_cmd {
3228        my ($r_min, $r_max, @args) = @_;
3229        my $head = 'HEAD';
3230        foreach my $x (@args) {
3231                last if $x eq '--';
3232                next unless ::verify_ref("$x^0");
3233                $head = $x;
3234                last;
3235        }
3236
3237        my $url = (::working_head_info($head))[0];
3238        my $gs = Git::SVN->find_by_url($url) || Git::SVN->_new;
3239        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3240                   $gs->refname);
3241        push @cmd, '-r' unless $non_recursive;
3242        push @cmd, qw/--raw --name-status/ if $verbose;
3243        push @cmd, '--color' if log_use_color();
3244        return @cmd unless defined $r_max;
3245        if ($r_max == $r_min) {
3246                push @cmd, '--max-count=1';
3247                if (my $c = $gs->rev_db_get($r_max)) {
3248                        push @cmd, $c;
3249                }
3250        } else {
3251                my ($c_min, $c_max);
3252                $c_max = $gs->rev_db_get($r_max);
3253                $c_min = $gs->rev_db_get($r_min);
3254                if (defined $c_min && defined $c_max) {
3255                        if ($r_max > $r_max) {
3256                                push @cmd, "$c_min..$c_max";
3257                        } else {
3258                                push @cmd, "$c_max..$c_min";
3259                        }
3260                } elsif ($r_max > $r_min) {
3261                        push @cmd, $c_max;
3262                } else {
3263                        push @cmd, $c_min;
3264                }
3265        }
3266        return @cmd;
3267}
3268
3269# adapted from pager.c
3270sub config_pager {
3271        $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3272        if (!defined $pager) {
3273                $pager = 'less';
3274        } elsif (length $pager == 0 || $pager eq 'cat') {
3275                $pager = undef;
3276        }
3277}
3278
3279sub run_pager {
3280        return unless -t *STDOUT;
3281        pipe my $rfd, my $wfd or return;
3282        defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
3283        if (!$pid) {
3284                open STDOUT, '>&', $wfd or
3285                                     ::fatal "Can't redirect to stdout: $!\n";
3286                return;
3287        }
3288        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
3289        $ENV{LESS} ||= 'FRSX';
3290        exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
3291}
3292
3293sub tz_to_s_offset {
3294        my ($tz) = @_;
3295        $tz =~ s/(\d\d)$//;
3296        return ($1 * 60) + ($tz * 3600);
3297}
3298
3299sub get_author_info {
3300        my ($dest, $author, $t, $tz) = @_;
3301        $author =~ s/(?:^\s*|\s*$)//g;
3302        $dest->{a_raw} = $author;
3303        my $au;
3304        if ($::_authors) {
3305                $au = $rusers{$author} || undef;
3306        }
3307        if (!$au) {
3308                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3309        }
3310        $dest->{t} = $t;
3311        $dest->{tz} = $tz;
3312        $dest->{a} = $au;
3313        # Date::Parse isn't in the standard Perl distro :(
3314        if ($tz =~ s/^\+//) {
3315                $t += tz_to_s_offset($tz);
3316        } elsif ($tz =~ s/^\-//) {
3317                $t -= tz_to_s_offset($tz);
3318        }
3319        $dest->{t_utc} = $t;
3320}
3321
3322sub process_commit {
3323        my ($c, $r_min, $r_max, $defer) = @_;
3324        if (defined $r_min && defined $r_max) {
3325                if ($r_min == $c->{r} && $r_min == $r_max) {
3326                        show_commit($c);
3327                        return 0;
3328                }
3329                return 1 if $r_min == $r_max;
3330                if ($r_min < $r_max) {
3331                        # we need to reverse the print order
3332                        return 0 if (defined $limit && --$limit < 0);
3333                        push @$defer, $c;
3334                        return 1;
3335                }
3336                if ($r_min != $r_max) {
3337                        return 1 if ($r_min < $c->{r});
3338                        return 1 if ($r_max > $c->{r});
3339                }
3340        }
3341        return 0 if (defined $limit && --$limit < 0);
3342        show_commit($c);
3343        return 1;
3344}
3345
3346sub show_commit {
3347        my $c = shift;
3348        if ($oneline) {
3349                my $x = "\n";
3350                if (my $l = $c->{l}) {
3351                        while ($l->[0] =~ /^\s*$/) { shift @$l }
3352                        $x = $l->[0];
3353                }
3354                $l_fmt ||= 'A' . length($c->{r});
3355                print 'r',pack($l_fmt, $c->{r}),' | ';
3356                print "$c->{c} | " if $show_commit;
3357                print $x;
3358        } else {
3359                show_commit_normal($c);
3360        }
3361}
3362
3363sub show_commit_changed_paths {
3364        my ($c) = @_;
3365        return unless $c->{changed};
3366        print "Changed paths:\n", @{$c->{changed}};
3367}
3368
3369sub show_commit_normal {
3370        my ($c) = @_;
3371        print '-' x72, "\nr$c->{r} | ";
3372        print "$c->{c} | " if $show_commit;
3373        print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3374                                 localtime($c->{t_utc})), ' | ';
3375        my $nr_line = 0;
3376
3377        if (my $l = $c->{l}) {
3378                while ($l->[$#$l] eq "\n" && $#$l > 0
3379                                          && $l->[($#$l - 1)] eq "\n") {
3380                        pop @$l;
3381                }
3382                $nr_line = scalar @$l;
3383                if (!$nr_line) {
3384                        print "1 line\n\n\n";
3385                } else {
3386                        if ($nr_line == 1) {
3387                                $nr_line = '1 line';
3388                        } else {
3389                                $nr_line .= ' lines';
3390                        }
3391                        print $nr_line, "\n";
3392                        show_commit_changed_paths($c);
3393                        print "\n";
3394                        print $_ foreach @$l;
3395                }
3396        } else {
3397                print "1 line\n";
3398                show_commit_changed_paths($c);
3399                print "\n";
3400
3401        }
3402        foreach my $x (qw/raw stat diff/) {
3403                if ($c->{$x}) {
3404                        print "\n";
3405                        print $_ foreach @{$c->{$x}}
3406                }
3407        }
3408}
3409
3410sub cmd_show_log {
3411        my (@args) = @_;
3412        my ($r_min, $r_max);
3413        my $r_last = -1; # prevent dupes
3414        if (defined $TZ) {
3415                $ENV{TZ} = $TZ;
3416        } else {
3417                delete $ENV{TZ};
3418        }
3419        if (defined $::_revision) {
3420                if ($::_revision =~ /^(\d+):(\d+)$/) {
3421                        ($r_min, $r_max) = ($1, $2);
3422                } elsif ($::_revision =~ /^\d+$/) {
3423                        $r_min = $r_max = $::_revision;
3424                } else {
3425                        ::fatal "-r$::_revision is not supported, use ",
3426                                "standard \'git log\' arguments instead\n";
3427                }
3428        }
3429
3430        config_pager();
3431        @args = (git_svn_log_cmd($r_min, $r_max, @args), @args);
3432        my $log = command_output_pipe(@args);
3433        run_pager();
3434        my (@k, $c, $d, $stat);
3435        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3436        while (<$log>) {
3437                if (/^${esc_color}commit ($::sha1_short)/o) {
3438                        my $cmt = $1;
3439                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3440                                $r_last = $c->{r};
3441                                process_commit($c, $r_min, $r_max, \@k) or
3442                                                                goto out;
3443                        }
3444                        $d = undef;
3445                        $c = { c => $cmt };
3446                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3447                        get_author_info($c, $1, $2, $3);
3448                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3449                        # ignore
3450                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3451                        push @{$c->{raw}}, $_;
3452                } elsif (/^${esc_color}[ACRMDT]\t/) {
3453                        # we could add $SVN->{svn_path} here, but that requires
3454                        # remote access at the moment (repo_path_split)...
3455                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
3456                        push @{$c->{changed}}, $_;
3457                } elsif (/^${esc_color}diff /o) {
3458                        $d = 1;
3459                        push @{$c->{diff}}, $_;
3460                } elsif ($d) {
3461                        push @{$c->{diff}}, $_;
3462                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
3463                          $esc_color*[\+\-]*$esc_color$/x) {
3464                        $stat = 1;
3465                        push @{$c->{stat}}, $_;
3466                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
3467                        push @{$c->{stat}}, $_;
3468                        $stat = undef;
3469                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
3470                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3471                } elsif (s/^${esc_color}    //o) {
3472                        push @{$c->{l}}, $_;
3473                }
3474        }
3475        if ($c && defined $c->{r} && $c->{r} != $r_last) {
3476                $r_last = $c->{r};
3477                process_commit($c, $r_min, $r_max, \@k);
3478        }
3479        if (@k) {
3480                my $swap = $r_max;
3481                $r_max = $r_min;
3482                $r_min = $swap;
3483                process_commit($_, $r_min, $r_max) foreach reverse @k;
3484        }
3485out:
3486        close $log;
3487        print '-' x72,"\n" unless $incremental || $oneline;
3488}
3489
3490package Git::SVN::Migration;
3491# these version numbers do NOT correspond to actual version numbers
3492# of git nor git-svn.  They are just relative.
3493#
3494# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3495#
3496# v1 layout: .git/$id/info/url, refs/remotes/$id
3497#
3498# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3499#
3500# v3 layout: .git/svn/$id, refs/remotes/$id
3501#            - info/url may remain for backwards compatibility
3502#            - this is what we migrate up to this layout automatically,
3503#            - this will be used by git svn init on single branches
3504# v3.1 layout (auto migrated):
3505#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
3506#              for backwards compatibility
3507#
3508# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3509#            - this is only created for newly multi-init-ed
3510#              repositories.  Similar in spirit to the
3511#              --use-separate-remotes option in git-clone (now default)
3512#            - we do not automatically migrate to this (following
3513#              the example set by core git)
3514use strict;
3515use warnings;
3516use Carp qw/croak/;
3517use File::Path qw/mkpath/;
3518use File::Basename qw/dirname basename/;
3519use vars qw/$_minimize/;
3520
3521sub migrate_from_v0 {
3522        my $git_dir = $ENV{GIT_DIR};
3523        return undef unless -d $git_dir;
3524        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3525        my $migrated = 0;
3526        while (<$fh>) {
3527                chomp;
3528                my ($id, $orig_ref) = ($_, $_);
3529                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3530                next unless -f "$git_dir/$id/info/url";
3531                my $new_ref = "refs/remotes/$id";
3532                if (::verify_ref("$new_ref^0")) {
3533                        print STDERR "W: $orig_ref is probably an old ",
3534                                     "branch used by an ancient version of ",
3535                                     "git-svn.\n",
3536                                     "However, $new_ref also exists.\n",
3537                                     "We will not be able ",
3538                                     "to use this branch until this ",
3539                                     "ambiguity is resolved.\n";
3540                        next;
3541                }
3542                print STDERR "Migrating from v0 layout...\n" if !$migrated;
3543                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3544                command_noisy('update-ref', $new_ref, $orig_ref);
3545                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3546                $migrated++;
3547        }
3548        command_close_pipe($fh, $ctx);
3549        print STDERR "Done migrating from v0 layout...\n" if $migrated;
3550        $migrated;
3551}
3552
3553sub migrate_from_v1 {
3554        my $git_dir = $ENV{GIT_DIR};
3555        my $migrated = 0;
3556        return $migrated unless -d $git_dir;
3557        my $svn_dir = "$git_dir/svn";
3558
3559        # just in case somebody used 'svn' as their $id at some point...
3560        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3561
3562        print STDERR "Migrating from a git-svn v1 layout...\n";
3563        mkpath([$svn_dir]);
3564        print STDERR "Data from a previous version of git-svn exists, but\n\t",
3565                     "$svn_dir\n\t(required for this version ",
3566                     "($::VERSION) of git-svn) does not. exist\n";
3567        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3568        while (<$fh>) {
3569                my $x = $_;
3570                next unless $x =~ s#^refs/remotes/##;
3571                chomp $x;
3572                next unless -f "$git_dir/$x/info/url";
3573                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3574                next unless $u;
3575                my $dn = dirname("$git_dir/svn/$x");
3576                mkpath([$dn]) unless -d $dn;
3577                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3578                        mkpath(["$git_dir/svn/svn"]);
3579                        print STDERR " - $git_dir/$x/info => ",
3580                                        "$git_dir/svn/$x/info\n";
3581                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3582                               croak "$!: $x";
3583                        # don't worry too much about these, they probably
3584                        # don't exist with repos this old (save for index,
3585                        # and we can easily regenerate that)
3586                        foreach my $f (qw/unhandled.log index .rev_db/) {
3587                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3588                        }
3589                } else {
3590                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3591                        rename "$git_dir/$x", "$git_dir/svn/$x" or
3592                               croak "$!: $x";
3593                }
3594                $migrated++;
3595        }
3596        command_close_pipe($fh, $ctx);
3597        print STDERR "Done migrating from a git-svn v1 layout\n";
3598        $migrated;
3599}
3600
3601sub read_old_urls {
3602        my ($l_map, $pfx, $path) = @_;
3603        my @dir;
3604        foreach (<$path/*>) {
3605                if (-r "$_/info/url") {
3606                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3607                        my $ref_id = $pfx . basename $_;
3608                        my $url = ::file_to_s("$_/info/url");
3609                        $l_map->{$ref_id} = $url;
3610                } elsif (-d $_) {
3611                        push @dir, $_;
3612                }
3613        }
3614        foreach (@dir) {
3615                my $x = $_;
3616                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3617                read_old_urls($l_map, $x, $_);
3618        }
3619}
3620
3621sub migrate_from_v2 {
3622        my @cfg = command(qw/config -l/);
3623        return if grep /^svn-remote\..+\.url=/, @cfg;
3624        my %l_map;
3625        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3626        my $migrated = 0;
3627
3628        foreach my $ref_id (sort keys %l_map) {
3629                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3630                if ($@) {
3631                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3632                }
3633                $migrated++;
3634        }
3635        $migrated;
3636}
3637
3638sub minimize_connections {
3639        my $r = Git::SVN::read_all_remotes();
3640        my $new_urls = {};
3641        my $root_repos = {};
3642        foreach my $repo_id (keys %$r) {
3643                my $url = $r->{$repo_id}->{url} or next;
3644                my $fetch = $r->{$repo_id}->{fetch} or next;
3645                my $ra = Git::SVN::Ra->new($url);
3646
3647                # skip existing cases where we already connect to the root
3648                if (($ra->{url} eq $ra->{repos_root}) ||
3649                    (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3650                     $repo_id)) {
3651                        $root_repos->{$ra->{url}} = $repo_id;
3652                        next;
3653                }
3654
3655                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3656                my $root_path = $ra->{url};
3657                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3658                foreach my $path (keys %$fetch) {
3659                        my $ref_id = $fetch->{$path};
3660                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3661
3662                        # make sure we can read when connecting to
3663                        # a higher level of a repository
3664                        my ($last_rev, undef) = $gs->last_rev_commit;
3665                        if (!defined $last_rev) {
3666                                $last_rev = eval {
3667                                        $root_ra->get_latest_revnum;
3668                                };
3669                                next if $@;
3670                        }
3671                        my $new = $root_path;
3672                        $new .= length $path ? "/$path" : '';
3673                        eval {
3674                                $root_ra->get_log([$new], $last_rev, $last_rev,
3675                                                  0, 0, 1, sub { });
3676                        };
3677                        next if $@;
3678                        $new_urls->{$ra->{repos_root}}->{$new} =
3679                                { ref_id => $ref_id,
3680                                  old_repo_id => $repo_id,
3681                                  old_path => $path };
3682                }
3683        }
3684
3685        my @emptied;
3686        foreach my $url (keys %$new_urls) {
3687                # see if we can re-use an existing [svn-remote "repo_id"]
3688                # instead of creating a(n ugly) new section:
3689                my $repo_id = $root_repos->{$url} ||
3690                              Git::SVN::sanitize_remote_name($url);
3691
3692                my $fetch = $new_urls->{$url};
3693                foreach my $path (keys %$fetch) {
3694                        my $x = $fetch->{$path};
3695                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3696                        my $pfx = "svn-remote.$x->{old_repo_id}";
3697
3698                        my $old_fetch = quotemeta("$x->{old_path}:".
3699                                                  "refs/remotes/$x->{ref_id}");
3700                        command_noisy(qw/config --unset/,
3701                                      "$pfx.fetch", '^'. $old_fetch . '$');
3702                        delete $r->{$x->{old_repo_id}}->
3703                               {fetch}->{$x->{old_path}};
3704                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3705                                command_noisy(qw/config --unset/,
3706                                              "$pfx.url");
3707                                push @emptied, $x->{old_repo_id}
3708                        }
3709                }
3710        }
3711        if (@emptied) {
3712                my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3713                           "$ENV{GIT_DIR}/config";
3714                print STDERR <<EOF;
3715The following [svn-remote] sections in your config file ($file) are empty
3716and can be safely removed:
3717EOF
3718                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3719        }
3720}
3721
3722sub migration_check {
3723        migrate_from_v0();
3724        migrate_from_v1();
3725        migrate_from_v2();
3726        minimize_connections() if $_minimize;
3727}
3728
3729package Git::IndexInfo;
3730use strict;
3731use warnings;
3732use Git qw/command_input_pipe command_close_pipe/;
3733
3734sub new {
3735        my ($class) = @_;
3736        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3737        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3738}
3739
3740sub remove {
3741        my ($self, $path) = @_;
3742        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3743                return ++$self->{nr};
3744        }
3745        undef;
3746}
3747
3748sub update {
3749        my ($self, $mode, $hash, $path) = @_;
3750        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3751                return ++$self->{nr};
3752        }
3753        undef;
3754}
3755
3756sub DESTROY {
3757        my ($self) = @_;
3758        command_close_pipe($self->{gui}, $self->{ctx});
3759}
3760
3761package Git::SVN::GlobSpec;
3762use strict;
3763use warnings;
3764
3765sub new {
3766        my ($class, $glob) = @_;
3767        my $re = $glob;
3768        $re =~ s!/+$!!g; # no need for trailing slashes
3769        my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
3770        my ($left, $right) = ($1, $2);
3771        if ($nr > 1) {
3772                die "Only one '*' wildcard expansion ",
3773                    "is supported (got $nr): '$glob'\n";
3774        } elsif ($nr == 0) {
3775                die "One '*' is needed for glob: '$glob'\n";
3776        }
3777        $re = quotemeta($left) . $re . quotemeta($right);
3778        if (length $left && !($left =~ s!/+$!!g)) {
3779                die "Missing trailing '/' on left side of: '$glob' ($left)\n";
3780        }
3781        if (length $right && !($right =~ s!^/+!!g)) {
3782                die "Missing leading '/' on right side of: '$glob' ($right)\n";
3783        }
3784        my $left_re = qr/^\/\Q$left\E(\/|$)/;
3785        bless { left => $left, right => $right, left_regex => $left_re,
3786                regex => qr/$re/, glob => $glob }, $class;
3787}
3788
3789sub full_path {
3790        my ($self, $path) = @_;
3791        return (length $self->{left} ? "$self->{left}/" : '') .
3792               $path . (length $self->{right} ? "/$self->{right}" : '');
3793}
3794
3795__END__
3796
3797Data structures:
3798
3799
3800$remotes = { # returned by read_all_remotes()
3801        'svn' => {
3802                # svn-remote.svn.url=https://svn.musicpd.org
3803                url => 'https://svn.musicpd.org',
3804                # svn-remote.svn.fetch=mpd/trunk:trunk
3805                fetch => {
3806                        'mpd/trunk' => 'trunk',
3807                },
3808                # svn-remote.svn.tags=mpd/tags/*:tags/*
3809                tags => {
3810                        path => {
3811                                left => 'mpd/tags',
3812                                right => '',
3813                                regex => qr!mpd/tags/([^/]+)$!,
3814                                glob => 'tags/*',
3815                        },
3816                        ref => {
3817                                left => 'tags',
3818                                right => '',
3819                                regex => qr!tags/([^/]+)$!,
3820                                glob => 'tags/*',
3821                        },
3822                }
3823        }
3824};
3825
3826$log_entry hashref as returned by libsvn_log_entry()
3827{
3828        log => 'whitespace-formatted log entry
3829',                                              # trailing newline is preserved
3830        revision => '8',                        # integer
3831        date => '2004-02-24T17:01:44.108345Z',  # commit date
3832        author => 'committer name'
3833};
3834
3835
3836# this is generated by generate_diff();
3837@mods = array of diff-index line hashes, each element represents one line
3838        of diff-index output
3839
3840diff-index line ($m hash)
3841{
3842        mode_a => first column of diff-index output, no leading ':',
3843        mode_b => second column of diff-index output,
3844        sha1_b => sha1sum of the final blob,
3845        chg => change type [MCRADT],
3846        file_a => original file name of a file (iff chg is 'C' or 'R')
3847        file_b => new/current file name of a file (any chg)
3848}
3849;
3850
3851# retval of read_url_paths{,_all}();
3852$l_map = {
3853        # repository root url
3854        'https://svn.musicpd.org' => {
3855                # repository path               # GIT_SVN_ID
3856                'mpd/trunk'             =>      'trunk',
3857                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
3858        },
3859}
3860
3861Notes:
3862        I don't trust the each() function on unless I created %hash myself
3863        because the internal iterator may not have started at base.