git-svn.perlon commit Merge branch 'ft/gitweb-tar-with-gzip-n' into maint (b73f3d2)
   1#!/usr/bin/env perl
   2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
   3# License: GPL v2 or later
   4use 5.008;
   5use warnings;
   6use strict;
   7use vars qw/    $AUTHOR $VERSION
   8                $sha1 $sha1_short $_revision $_repository
   9                $_q $_authors $_authors_prog %users/;
  10$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
  11$VERSION = '@@GIT_VERSION@@';
  12
  13# From which subdir have we been invoked?
  14my $cmd_dir_prefix = eval {
  15        command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
  16} || '';
  17
  18my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
  19$ENV{GIT_DIR} ||= '.git';
  20$Git::SVN::default_repo_id = 'svn';
  21$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
  22$Git::SVN::Ra::_log_window_size = 100;
  23$Git::SVN::_minimize_url = 'unset';
  24
  25if (! exists $ENV{SVN_SSH}) {
  26        if (exists $ENV{GIT_SSH}) {
  27                $ENV{SVN_SSH} = $ENV{GIT_SSH};
  28                if ($^O eq 'msys') {
  29                        $ENV{SVN_SSH} =~ s/\\/\\\\/g;
  30                        $ENV{SVN_SSH} =~ s/(.*)/"$1"/;
  31                }
  32        }
  33}
  34
  35$Git::SVN::Log::TZ = $ENV{TZ};
  36$ENV{TZ} = 'UTC';
  37$| = 1; # unbuffer STDOUT
  38
  39sub fatal (@) { print STDERR "@_\n"; exit 1 }
  40sub _req_svn {
  41        require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
  42        require SVN::Ra;
  43        require SVN::Delta;
  44        if ($SVN::Core::VERSION lt '1.1.0') {
  45                fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
  46        }
  47}
  48my $can_compress = eval { require Compress::Zlib; 1};
  49push @Git::SVN::Ra::ISA, 'SVN::Ra';
  50push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
  51push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
  52use Carp qw/croak/;
  53use Digest::MD5;
  54use IO::File qw//;
  55use File::Basename qw/dirname basename/;
  56use File::Path qw/mkpath/;
  57use File::Spec;
  58use File::Find;
  59use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
  60use IPC::Open3;
  61use Git;
  62use Memoize;  # core since 5.8.0, Jul 2002
  63
  64BEGIN {
  65        # import functions from Git into our packages, en masse
  66        no strict 'refs';
  67        foreach (qw/command command_oneline command_noisy command_output_pipe
  68                    command_input_pipe command_close_pipe
  69                    command_bidi_pipe command_close_bidi_pipe/) {
  70                for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
  71                        Git::SVN::Migration Git::SVN::Log Git::SVN),
  72                        __PACKAGE__) {
  73                        *{"${package}::$_"} = \&{"Git::$_"};
  74                }
  75        }
  76        Memoize::memoize 'Git::config';
  77        Memoize::memoize 'Git::config_bool';
  78}
  79
  80my ($SVN);
  81
  82$sha1 = qr/[a-f\d]{40}/;
  83$sha1_short = qr/[a-f\d]{4,40}/;
  84my ($_stdin, $_help, $_edit,
  85        $_message, $_file, $_branch_dest,
  86        $_template, $_shared,
  87        $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
  88        $_merge, $_strategy, $_dry_run, $_local,
  89        $_prefix, $_no_checkout, $_url, $_verbose,
  90        $_git_format, $_commit_url, $_tag, $_merge_info);
  91$Git::SVN::_follow_parent = 1;
  92$_q ||= 0;
  93my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
  94                    'config-dir=s' => \$Git::SVN::Ra::config_dir,
  95                    'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
  96                    'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
  97my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
  98                'authors-file|A=s' => \$_authors,
  99                'authors-prog=s' => \$_authors_prog,
 100                'repack:i' => \$Git::SVN::_repack,
 101                'noMetadata' => \$Git::SVN::_no_metadata,
 102                'useSvmProps' => \$Git::SVN::_use_svm_props,
 103                'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
 104                'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
 105                'no-checkout' => \$_no_checkout,
 106                'quiet|q+' => \$_q,
 107                'repack-flags|repack-args|repack-opts=s' =>
 108                   \$Git::SVN::_repack_flags,
 109                'use-log-author' => \$Git::SVN::_use_log_author,
 110                'add-author-from' => \$Git::SVN::_add_author_from,
 111                'localtime' => \$Git::SVN::_localtime,
 112                %remote_opts );
 113
 114my ($_trunk, @_tags, @_branches, $_stdlayout);
 115my %icv;
 116my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
 117                  'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
 118                  'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
 119                  'stdlayout|s' => \$_stdlayout,
 120                  'minimize-url|m!' => \$Git::SVN::_minimize_url,
 121                  'no-metadata' => sub { $icv{noMetadata} = 1 },
 122                  'use-svm-props' => sub { $icv{useSvmProps} = 1 },
 123                  'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
 124                  'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
 125                  'rewrite-uuid=s' => sub { $icv{rewriteUUID} = $_[1] },
 126                  %remote_opts );
 127my %cmt_opts = ( 'edit|e' => \$_edit,
 128                'rmdir' => \$SVN::Git::Editor::_rmdir,
 129                'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
 130                'l=i' => \$SVN::Git::Editor::_rename_limit,
 131                'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
 132);
 133
 134my %cmd = (
 135        fetch => [ \&cmd_fetch, "Download new revisions from SVN",
 136                        { 'revision|r=s' => \$_revision,
 137                          'fetch-all|all' => \$_fetch_all,
 138                          'parent|p' => \$_fetch_parent,
 139                           %fc_opts } ],
 140        clone => [ \&cmd_clone, "Initialize and fetch revisions",
 141                        { 'revision|r=s' => \$_revision,
 142                           %fc_opts, %init_opts } ],
 143        init => [ \&cmd_init, "Initialize a repo for tracking" .
 144                          " (requires URL argument)",
 145                          \%init_opts ],
 146        'multi-init' => [ \&cmd_multi_init,
 147                          "Deprecated alias for ".
 148                          "'$0 init -T<trunk> -b<branches> -t<tags>'",
 149                          \%init_opts ],
 150        dcommit => [ \&cmd_dcommit,
 151                     'Commit several diffs to merge with upstream',
 152                        { 'merge|m|M' => \$_merge,
 153                          'strategy|s=s' => \$_strategy,
 154                          'verbose|v' => \$_verbose,
 155                          'dry-run|n' => \$_dry_run,
 156                          'fetch-all|all' => \$_fetch_all,
 157                          'commit-url=s' => \$_commit_url,
 158                          'revision|r=i' => \$_revision,
 159                          'no-rebase' => \$_no_rebase,
 160                          'mergeinfo=s' => \$_merge_info,
 161                        %cmt_opts, %fc_opts } ],
 162        branch => [ \&cmd_branch,
 163                    'Create a branch in the SVN repository',
 164                    { 'message|m=s' => \$_message,
 165                      'destination|d=s' => \$_branch_dest,
 166                      'dry-run|n' => \$_dry_run,
 167                      'tag|t' => \$_tag,
 168                      'username=s' => \$Git::SVN::Prompt::_username,
 169                      'commit-url=s' => \$_commit_url } ],
 170        tag => [ sub { $_tag = 1; cmd_branch(@_) },
 171                 'Create a tag in the SVN repository',
 172                 { 'message|m=s' => \$_message,
 173                   'destination|d=s' => \$_branch_dest,
 174                   'dry-run|n' => \$_dry_run,
 175                   'username=s' => \$Git::SVN::Prompt::_username,
 176                   'commit-url=s' => \$_commit_url } ],
 177        'set-tree' => [ \&cmd_set_tree,
 178                        "Set an SVN repository to a git tree-ish",
 179                        { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
 180        'create-ignore' => [ \&cmd_create_ignore,
 181                             'Create a .gitignore per svn:ignore',
 182                             { 'revision|r=i' => \$_revision
 183                             } ],
 184        'mkdirs' => [ \&cmd_mkdirs ,
 185                      "recreate empty directories after a checkout",
 186                      { 'revision|r=i' => \$_revision } ],
 187        'propget' => [ \&cmd_propget,
 188                       'Print the value of a property on a file or directory',
 189                       { 'revision|r=i' => \$_revision } ],
 190        'proplist' => [ \&cmd_proplist,
 191                       'List all properties of a file or directory',
 192                       { 'revision|r=i' => \$_revision } ],
 193        'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
 194                        { 'revision|r=i' => \$_revision
 195                        } ],
 196        'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
 197                        { 'revision|r=i' => \$_revision
 198                        } ],
 199        'multi-fetch' => [ \&cmd_multi_fetch,
 200                           "Deprecated alias for $0 fetch --all",
 201                           { 'revision|r=s' => \$_revision, %fc_opts } ],
 202        'migrate' => [ sub { },
 203                       # no-op, we automatically run this anyways,
 204                       'Migrate configuration/metadata/layout from
 205                        previous versions of git-svn',
 206                       { 'minimize' => \$Git::SVN::Migration::_minimize,
 207                         %remote_opts } ],
 208        'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
 209                        { 'limit=i' => \$Git::SVN::Log::limit,
 210                          'revision|r=s' => \$_revision,
 211                          'verbose|v' => \$Git::SVN::Log::verbose,
 212                          'incremental' => \$Git::SVN::Log::incremental,
 213                          'oneline' => \$Git::SVN::Log::oneline,
 214                          'show-commit' => \$Git::SVN::Log::show_commit,
 215                          'non-recursive' => \$Git::SVN::Log::non_recursive,
 216                          'authors-file|A=s' => \$_authors,
 217                          'color' => \$Git::SVN::Log::color,
 218                          'pager=s' => \$Git::SVN::Log::pager
 219                        } ],
 220        'find-rev' => [ \&cmd_find_rev,
 221                        "Translate between SVN revision numbers and tree-ish",
 222                        {} ],
 223        'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
 224                        { 'merge|m|M' => \$_merge,
 225                          'verbose|v' => \$_verbose,
 226                          'strategy|s=s' => \$_strategy,
 227                          'local|l' => \$_local,
 228                          'fetch-all|all' => \$_fetch_all,
 229                          'dry-run|n' => \$_dry_run,
 230                          %fc_opts } ],
 231        'commit-diff' => [ \&cmd_commit_diff,
 232                           'Commit a diff between two trees',
 233                        { 'message|m=s' => \$_message,
 234                          'file|F=s' => \$_file,
 235                          'revision|r=s' => \$_revision,
 236                        %cmt_opts } ],
 237        'info' => [ \&cmd_info,
 238                    "Show info about the latest SVN revision
 239                     on the current branch",
 240                    { 'url' => \$_url, } ],
 241        'blame' => [ \&Git::SVN::Log::cmd_blame,
 242                    "Show what revision and author last modified each line of a file",
 243                    { 'git-format' => \$_git_format } ],
 244        'reset' => [ \&cmd_reset,
 245                     "Undo fetches back to the specified SVN revision",
 246                     { 'revision|r=s' => \$_revision,
 247                       'parent|p' => \$_fetch_parent } ],
 248        'gc' => [ \&cmd_gc,
 249                  "Compress unhandled.log files in .git/svn and remove " .
 250                  "index files in .git/svn",
 251                {} ],
 252);
 253
 254my $cmd;
 255for (my $i = 0; $i < @ARGV; $i++) {
 256        if (defined $cmd{$ARGV[$i]}) {
 257                $cmd = $ARGV[$i];
 258                splice @ARGV, $i, 1;
 259                last;
 260        } elsif ($ARGV[$i] eq 'help') {
 261                $cmd = $ARGV[$i+1];
 262                usage(0);
 263        }
 264};
 265
 266# make sure we're always running at the top-level working directory
 267unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
 268        unless (-d $ENV{GIT_DIR}) {
 269                if ($git_dir_user_set) {
 270                        die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
 271                            "but it is not a directory\n";
 272                }
 273                my $git_dir = delete $ENV{GIT_DIR};
 274                my $cdup = undef;
 275                git_cmd_try {
 276                        $cdup = command_oneline(qw/rev-parse --show-cdup/);
 277                        $git_dir = '.' unless ($cdup);
 278                        chomp $cdup if ($cdup);
 279                        $cdup = "." unless ($cdup && length $cdup);
 280                } "Already at toplevel, but $git_dir not found\n";
 281                chdir $cdup or die "Unable to chdir up to '$cdup'\n";
 282                unless (-d $git_dir) {
 283                        die "$git_dir still not found after going to ",
 284                            "'$cdup'\n";
 285                }
 286                $ENV{GIT_DIR} = $git_dir;
 287        }
 288        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 289}
 290
 291my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
 292
 293read_git_config(\%opts);
 294if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
 295        Getopt::Long::Configure('pass_through');
 296}
 297my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
 298                    'minimize-connections' => \$Git::SVN::Migration::_minimize,
 299                    'id|i=s' => \$Git::SVN::default_ref_id,
 300                    'svn-remote|remote|R=s' => sub {
 301                       $Git::SVN::no_reuse_existing = 1;
 302                       $Git::SVN::default_repo_id = $_[1] });
 303exit 1 if (!$rv && $cmd && $cmd ne 'log');
 304
 305usage(0) if $_help;
 306version() if $_version;
 307usage(1) unless defined $cmd;
 308load_authors() if $_authors;
 309if (defined $_authors_prog) {
 310        $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
 311}
 312
 313unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
 314        Git::SVN::Migration::migration_check();
 315}
 316Git::SVN::init_vars();
 317eval {
 318        Git::SVN::verify_remotes_sanity();
 319        $cmd{$cmd}->[0]->(@ARGV);
 320};
 321fatal $@ if $@;
 322post_fetch_checkout();
 323exit 0;
 324
 325####################### primary functions ######################
 326sub usage {
 327        my $exit = shift || 0;
 328        my $fd = $exit ? \*STDERR : \*STDOUT;
 329        print $fd <<"";
 330git-svn - bidirectional operations between a single Subversion tree and git
 331Usage: git svn <command> [options] [arguments]\n
 332
 333        print $fd "Available commands:\n" unless $cmd;
 334
 335        foreach (sort keys %cmd) {
 336                next if $cmd && $cmd ne $_;
 337                next if /^multi-/; # don't show deprecated commands
 338                print $fd '  ',pack('A17',$_),$cmd{$_}->[1],"\n";
 339                foreach (sort keys %{$cmd{$_}->[2]}) {
 340                        # mixed-case options are for .git/config only
 341                        next if /[A-Z]/ && /^[a-z]+$/i;
 342                        # prints out arguments as they should be passed:
 343                        my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
 344                        print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
 345                                                        "--$_" : "-$_" }
 346                                                split /\|/,$_)," $x\n";
 347                }
 348        }
 349        print $fd <<"";
 350\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
 351arbitrary identifier if you're tracking multiple SVN branches/repositories in
 352one git repository and want to keep them separate.  See git-svn(1) for more
 353information.
 354
 355        exit $exit;
 356}
 357
 358sub version {
 359        ::_req_svn();
 360        print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
 361        exit 0;
 362}
 363
 364sub do_git_init_db {
 365        unless (-d $ENV{GIT_DIR}) {
 366                my @init_db = ('init');
 367                push @init_db, "--template=$_template" if defined $_template;
 368                if (defined $_shared) {
 369                        if ($_shared =~ /[a-z]/) {
 370                                push @init_db, "--shared=$_shared";
 371                        } else {
 372                                push @init_db, "--shared";
 373                        }
 374                }
 375                command_noisy(@init_db);
 376                $_repository = Git->repository(Repository => ".git");
 377        }
 378        my $set;
 379        my $pfx = "svn-remote.$Git::SVN::default_repo_id";
 380        foreach my $i (keys %icv) {
 381                die "'$set' and '$i' cannot both be set\n" if $set;
 382                next unless defined $icv{$i};
 383                command_noisy('config', "$pfx.$i", $icv{$i});
 384                $set = $i;
 385        }
 386        my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
 387        command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
 388                if defined $$ignore_regex;
 389}
 390
 391sub init_subdir {
 392        my $repo_path = shift or return;
 393        mkpath([$repo_path]) unless -d $repo_path;
 394        chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
 395        $ENV{GIT_DIR} = '.git';
 396        $_repository = Git->repository(Repository => $ENV{GIT_DIR});
 397}
 398
 399sub cmd_clone {
 400        my ($url, $path) = @_;
 401        if (!defined $path &&
 402            (defined $_trunk || @_branches || @_tags ||
 403             defined $_stdlayout) &&
 404            $url !~ m#^[a-z\+]+://#) {
 405                $path = $url;
 406        }
 407        $path = basename($url) if !defined $path || !length $path;
 408        my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
 409        cmd_init($url, $path);
 410        command_oneline('config', 'svn.authorsfile', $authors_absolute)
 411            if $_authors;
 412        Git::SVN::fetch_all($Git::SVN::default_repo_id);
 413}
 414
 415sub cmd_init {
 416        if (defined $_stdlayout) {
 417                $_trunk = 'trunk' if (!defined $_trunk);
 418                @_tags = 'tags' if (! @_tags);
 419                @_branches = 'branches' if (! @_branches);
 420        }
 421        if (defined $_trunk || @_branches || @_tags) {
 422                return cmd_multi_init(@_);
 423        }
 424        my $url = shift or die "SVN repository location required ",
 425                               "as a command-line argument\n";
 426        $url = canonicalize_url($url);
 427        init_subdir(@_);
 428        do_git_init_db();
 429
 430        if ($Git::SVN::_minimize_url eq 'unset') {
 431                $Git::SVN::_minimize_url = 0;
 432        }
 433
 434        Git::SVN->init($url);
 435}
 436
 437sub cmd_fetch {
 438        if (grep /^\d+=./, @_) {
 439                die "'<rev>=<commit>' fetch arguments are ",
 440                    "no longer supported.\n";
 441        }
 442        my ($remote) = @_;
 443        if (@_ > 1) {
 444                die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
 445        }
 446        $Git::SVN::no_reuse_existing = undef;
 447        if ($_fetch_parent) {
 448                my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 449                unless ($gs) {
 450                        die "Unable to determine upstream SVN information from ",
 451                            "working tree history\n";
 452                }
 453                # just fetch, don't checkout.
 454                $_no_checkout = 'true';
 455                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 456        } elsif ($_fetch_all) {
 457                cmd_multi_fetch();
 458        } else {
 459                $remote ||= $Git::SVN::default_repo_id;
 460                Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
 461        }
 462}
 463
 464sub cmd_set_tree {
 465        my (@commits) = @_;
 466        if ($_stdin || !@commits) {
 467                print "Reading from stdin...\n";
 468                @commits = ();
 469                while (<STDIN>) {
 470                        if (/\b($sha1_short)\b/o) {
 471                                unshift @commits, $1;
 472                        }
 473                }
 474        }
 475        my @revs;
 476        foreach my $c (@commits) {
 477                my @tmp = command('rev-parse',$c);
 478                if (scalar @tmp == 1) {
 479                        push @revs, $tmp[0];
 480                } elsif (scalar @tmp > 1) {
 481                        push @revs, reverse(command('rev-list',@tmp));
 482                } else {
 483                        fatal "Failed to rev-parse $c";
 484                }
 485        }
 486        my $gs = Git::SVN->new;
 487        my ($r_last, $cmt_last) = $gs->last_rev_commit;
 488        $gs->fetch;
 489        if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
 490                fatal "There are new revisions that were fetched ",
 491                      "and need to be merged (or acknowledged) ",
 492                      "before committing.\nlast rev: $r_last\n",
 493                      " current: $gs->{last_rev}";
 494        }
 495        $gs->set_tree($_) foreach @revs;
 496        print "Done committing ",scalar @revs," revisions to SVN\n";
 497        unlink $gs->{index};
 498}
 499
 500sub cmd_dcommit {
 501        my $head = shift;
 502        command_noisy(qw/update-index --refresh/);
 503        git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
 504                'Cannot dcommit with a dirty index.  Commit your changes first, '
 505                . "or stash them with `git stash'.\n";
 506        $head ||= 'HEAD';
 507
 508        my $old_head;
 509        if ($head ne 'HEAD') {
 510                $old_head = eval {
 511                        command_oneline([qw/symbolic-ref -q HEAD/])
 512                };
 513                if ($old_head) {
 514                        $old_head =~ s{^refs/heads/}{};
 515                } else {
 516                        $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
 517                }
 518                command(['checkout', $head], STDERR => 0);
 519        }
 520
 521        my @refs;
 522        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
 523        unless ($gs) {
 524                die "Unable to determine upstream SVN information from ",
 525                    "$head history.\nPerhaps the repository is empty.";
 526        }
 527
 528        if (defined $_commit_url) {
 529                $url = $_commit_url;
 530        } else {
 531                $url = eval { command_oneline('config', '--get',
 532                              "svn-remote.$gs->{repo_id}.commiturl") };
 533                if (!$url) {
 534                        $url = $gs->full_pushurl
 535                }
 536        }
 537
 538        my $last_rev = $_revision if defined $_revision;
 539        if ($url) {
 540                print "Committing to $url ...\n";
 541        }
 542        my ($linear_refs, $parents) = linearize_history($gs, \@refs);
 543        if ($_no_rebase && scalar(@$linear_refs) > 1) {
 544                warn "Attempting to commit more than one change while ",
 545                     "--no-rebase is enabled.\n",
 546                     "If these changes depend on each other, re-running ",
 547                     "without --no-rebase may be required."
 548        }
 549        my $expect_url = $url;
 550        Git::SVN::remove_username($expect_url);
 551        while (1) {
 552                my $d = shift @$linear_refs or last;
 553                unless (defined $last_rev) {
 554                        (undef, $last_rev, undef) = cmt_metadata("$d~1");
 555                        unless (defined $last_rev) {
 556                                fatal "Unable to extract revision information ",
 557                                      "from commit $d~1";
 558                        }
 559                }
 560                if ($_dry_run) {
 561                        print "diff-tree $d~1 $d\n";
 562                } else {
 563                        my $cmt_rev;
 564                        my %ed_opts = ( r => $last_rev,
 565                                        log => get_commit_entry($d)->{log},
 566                                        ra => Git::SVN::Ra->new($url),
 567                                        config => SVN::Core::config_get_config(
 568                                                $Git::SVN::Ra::config_dir
 569                                        ),
 570                                        tree_a => "$d~1",
 571                                        tree_b => $d,
 572                                        editor_cb => sub {
 573                                               print "Committed r$_[0]\n";
 574                                               $cmt_rev = $_[0];
 575                                        },
 576                                        mergeinfo => $_merge_info,
 577                                        svn_path => '');
 578                        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
 579                                print "No changes\n$d~1 == $d\n";
 580                        } elsif ($parents->{$d} && @{$parents->{$d}}) {
 581                                $gs->{inject_parents_dcommit}->{$cmt_rev} =
 582                                                               $parents->{$d};
 583                        }
 584                        $_fetch_all ? $gs->fetch_all : $gs->fetch;
 585                        $last_rev = $cmt_rev;
 586                        next if $_no_rebase;
 587
 588                        # we always want to rebase against the current HEAD,
 589                        # not any head that was passed to us
 590                        my @diff = command('diff-tree', $d,
 591                                           $gs->refname, '--');
 592                        my @finish;
 593                        if (@diff) {
 594                                @finish = rebase_cmd();
 595                                print STDERR "W: $d and ", $gs->refname,
 596                                             " differ, using @finish:\n",
 597                                             join("\n", @diff), "\n";
 598                        } else {
 599                                print "No changes between current HEAD and ",
 600                                      $gs->refname,
 601                                      "\nResetting to the latest ",
 602                                      $gs->refname, "\n";
 603                                @finish = qw/reset --mixed/;
 604                        }
 605                        command_noisy(@finish, $gs->refname);
 606                        if (@diff) {
 607                                @refs = ();
 608                                my ($url_, $rev_, $uuid_, $gs_) =
 609                                              working_head_info('HEAD', \@refs);
 610                                my ($linear_refs_, $parents_) =
 611                                              linearize_history($gs_, \@refs);
 612                                if (scalar(@$linear_refs) !=
 613                                    scalar(@$linear_refs_)) {
 614                                        fatal "# of revisions changed ",
 615                                          "\nbefore:\n",
 616                                          join("\n", @$linear_refs),
 617                                          "\n\nafter:\n",
 618                                          join("\n", @$linear_refs_), "\n",
 619                                          'If you are attempting to commit ',
 620                                          "merges, try running:\n\t",
 621                                          'git rebase --interactive',
 622                                          '--preserve-merges ',
 623                                          $gs->refname,
 624                                          "\nBefore dcommitting";
 625                                }
 626                                if ($url_ ne $expect_url) {
 627                                        if ($url_ eq $gs->metadata_url) {
 628                                                print
 629                                                  "Accepting rewritten URL:",
 630                                                  " $url_\n";
 631                                        } else {
 632                                                fatal
 633                                                  "URL mismatch after rebase:",
 634                                                  " $url_ != $expect_url";
 635                                        }
 636                                }
 637                                if ($uuid_ ne $uuid) {
 638                                        fatal "uuid mismatch after rebase: ",
 639                                              "$uuid_ != $uuid";
 640                                }
 641                                # remap parents
 642                                my (%p, @l, $i);
 643                                for ($i = 0; $i < scalar @$linear_refs; $i++) {
 644                                        my $new = $linear_refs_->[$i] or next;
 645                                        $p{$new} =
 646                                                $parents->{$linear_refs->[$i]};
 647                                        push @l, $new;
 648                                }
 649                                $parents = \%p;
 650                                $linear_refs = \@l;
 651                        }
 652                }
 653        }
 654
 655        if ($old_head) {
 656                my $new_head = command_oneline(qw/rev-parse HEAD/);
 657                my $new_is_symbolic = eval {
 658                        command_oneline(qw/symbolic-ref -q HEAD/);
 659                };
 660                if ($new_is_symbolic) {
 661                        print "dcommitted the branch ", $head, "\n";
 662                } else {
 663                        print "dcommitted on a detached HEAD because you gave ",
 664                              "a revision argument.\n",
 665                              "The rewritten commit is: ", $new_head, "\n";
 666                }
 667                command(['checkout', $old_head], STDERR => 0);
 668        }
 669
 670        unlink $gs->{index};
 671}
 672
 673sub cmd_branch {
 674        my ($branch_name, $head) = @_;
 675
 676        unless (defined $branch_name && length $branch_name) {
 677                die(($_tag ? "tag" : "branch") . " name required\n");
 678        }
 679        $head ||= 'HEAD';
 680
 681        my (undef, $rev, undef, $gs) = working_head_info($head);
 682        my $src = $gs->full_pushurl;
 683
 684        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
 685        my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
 686        my $glob;
 687        if ($#{$allglobs} == 0) {
 688                $glob = $allglobs->[0];
 689        } else {
 690                unless(defined $_branch_dest) {
 691                        die "Multiple ",
 692                            $_tag ? "tag" : "branch",
 693                            " paths defined for Subversion repository.\n",
 694                            "You must specify where you want to create the ",
 695                            $_tag ? "tag" : "branch",
 696                            " with the --destination argument.\n";
 697                }
 698                foreach my $g (@{$allglobs}) {
 699                        # SVN::Git::Editor could probably be moved to Git.pm..
 700                        my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
 701                        if ($_branch_dest =~ /$re/) {
 702                                $glob = $g;
 703                                last;
 704                        }
 705                }
 706                unless (defined $glob) {
 707                        my $dest_re = qr/\b\Q$_branch_dest\E\b/;
 708                        foreach my $g (@{$allglobs}) {
 709                                $g->{path}->{left} =~ /$dest_re/ or next;
 710                                if (defined $glob) {
 711                                        die "Ambiguous destination: ",
 712                                            $_branch_dest, "\nmatches both '",
 713                                            $glob->{path}->{left}, "' and '",
 714                                            $g->{path}->{left}, "'\n";
 715                                }
 716                                $glob = $g;
 717                        }
 718                        unless (defined $glob) {
 719                                die "Unknown ",
 720                                    $_tag ? "tag" : "branch",
 721                                    " destination $_branch_dest\n";
 722                        }
 723                }
 724        }
 725        my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
 726        my $url;
 727        if (defined $_commit_url) {
 728                $url = $_commit_url;
 729        } else {
 730                $url = eval { command_oneline('config', '--get',
 731                        "svn-remote.$gs->{repo_id}.commiturl") };
 732                if (!$url) {
 733                        $url = $remote->{pushurl} || $remote->{url};
 734                }
 735        }
 736        my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
 737
 738        if ($dst =~ /^https:/ && $src =~ /^http:/) {
 739                $src=~s/^http:/https:/;
 740        }
 741
 742        ::_req_svn();
 743
 744        my $ctx = SVN::Client->new(
 745                auth    => Git::SVN::Ra::_auth_providers(),
 746                log_msg => sub {
 747                        ${ $_[0] } = defined $_message
 748                                ? $_message
 749                                : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
 750                                . $branch_name;
 751                },
 752        );
 753
 754        eval {
 755                $ctx->ls($dst, 'HEAD', 0);
 756        } and die "branch ${branch_name} already exists\n";
 757
 758        print "Copying ${src} at r${rev} to ${dst}...\n";
 759        $ctx->copy($src, $rev, $dst)
 760                unless $_dry_run;
 761
 762        $gs->fetch_all;
 763}
 764
 765sub cmd_find_rev {
 766        my $revision_or_hash = shift or die "SVN or git revision required ",
 767                                            "as a command-line argument\n";
 768        my $result;
 769        if ($revision_or_hash =~ /^r\d+$/) {
 770                my $head = shift;
 771                $head ||= 'HEAD';
 772                my @refs;
 773                my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
 774                unless ($gs) {
 775                        die "Unable to determine upstream SVN information from ",
 776                            "$head history\n";
 777                }
 778                my $desired_revision = substr($revision_or_hash, 1);
 779                $result = $gs->rev_map_get($desired_revision, $uuid);
 780        } else {
 781                my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
 782                $result = $rev;
 783        }
 784        print "$result\n" if $result;
 785}
 786
 787sub cmd_rebase {
 788        command_noisy(qw/update-index --refresh/);
 789        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 790        unless ($gs) {
 791                die "Unable to determine upstream SVN information from ",
 792                    "working tree history\n";
 793        }
 794        if ($_dry_run) {
 795                print "Remote Branch: " . $gs->refname . "\n";
 796                print "SVN URL: " . $url . "\n";
 797                return;
 798        }
 799        if (command(qw/diff-index HEAD --/)) {
 800                print STDERR "Cannot rebase with uncommited changes:\n";
 801                command_noisy('status');
 802                exit 1;
 803        }
 804        unless ($_local) {
 805                # rebase will checkout for us, so no need to do it explicitly
 806                $_no_checkout = 'true';
 807                $_fetch_all ? $gs->fetch_all : $gs->fetch;
 808        }
 809        command_noisy(rebase_cmd(), $gs->refname);
 810        $gs->mkemptydirs;
 811}
 812
 813sub cmd_show_ignore {
 814        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 815        $gs ||= Git::SVN->new;
 816        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 817        $gs->prop_walk($gs->{path}, $r, sub {
 818                my ($gs, $path, $props) = @_;
 819                print STDOUT "\n# $path\n";
 820                my $s = $props->{'svn:ignore'} or return;
 821                $s =~ s/[\r\n]+/\n/g;
 822                $s =~ s/^\n+//;
 823                chomp $s;
 824                $s =~ s#^#$path#gm;
 825                print STDOUT "$s\n";
 826        });
 827}
 828
 829sub cmd_show_externals {
 830        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 831        $gs ||= Git::SVN->new;
 832        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 833        $gs->prop_walk($gs->{path}, $r, sub {
 834                my ($gs, $path, $props) = @_;
 835                print STDOUT "\n# $path\n";
 836                my $s = $props->{'svn:externals'} or return;
 837                $s =~ s/[\r\n]+/\n/g;
 838                chomp $s;
 839                $s =~ s#^#$path#gm;
 840                print STDOUT "$s\n";
 841        });
 842}
 843
 844sub cmd_create_ignore {
 845        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 846        $gs ||= Git::SVN->new;
 847        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 848        $gs->prop_walk($gs->{path}, $r, sub {
 849                my ($gs, $path, $props) = @_;
 850                # $path is of the form /path/to/dir/
 851                $path = '.' . $path;
 852                # SVN can have attributes on empty directories,
 853                # which git won't track
 854                mkpath([$path]) unless -d $path;
 855                my $ignore = $path . '.gitignore';
 856                my $s = $props->{'svn:ignore'} or return;
 857                open(GITIGNORE, '>', $ignore)
 858                  or fatal("Failed to open `$ignore' for writing: $!");
 859                $s =~ s/[\r\n]+/\n/g;
 860                $s =~ s/^\n+//;
 861                chomp $s;
 862                # Prefix all patterns so that the ignore doesn't apply
 863                # to sub-directories.
 864                $s =~ s#^#/#gm;
 865                print GITIGNORE "$s\n";
 866                close(GITIGNORE)
 867                  or fatal("Failed to close `$ignore': $!");
 868                command_noisy('add', '-f', $ignore);
 869        });
 870}
 871
 872sub cmd_mkdirs {
 873        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 874        $gs ||= Git::SVN->new;
 875        $gs->mkemptydirs($_revision);
 876}
 877
 878sub canonicalize_path {
 879        my ($path) = @_;
 880        my $dot_slash_added = 0;
 881        if (substr($path, 0, 1) ne "/") {
 882                $path = "./" . $path;
 883                $dot_slash_added = 1;
 884        }
 885        # File::Spec->canonpath doesn't collapse x/../y into y (for a
 886        # good reason), so let's do this manually.
 887        $path =~ s#/+#/#g;
 888        $path =~ s#/\.(?:/|$)#/#g;
 889        $path =~ s#/[^/]+/\.\.##g;
 890        $path =~ s#/$##g;
 891        $path =~ s#^\./## if $dot_slash_added;
 892        $path =~ s#^/##;
 893        $path =~ s#^\.$##;
 894        return $path;
 895}
 896
 897sub canonicalize_url {
 898        my ($url) = @_;
 899        $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
 900        return $url;
 901}
 902
 903# get_svnprops(PATH)
 904# ------------------
 905# Helper for cmd_propget and cmd_proplist below.
 906sub get_svnprops {
 907        my $path = shift;
 908        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
 909        $gs ||= Git::SVN->new;
 910
 911        # prefix THE PATH by the sub-directory from which the user
 912        # invoked us.
 913        $path = $cmd_dir_prefix . $path;
 914        fatal("No such file or directory: $path") unless -e $path;
 915        my $is_dir = -d $path ? 1 : 0;
 916        $path = $gs->{path} . '/' . $path;
 917
 918        # canonicalize the path (otherwise libsvn will abort or fail to
 919        # find the file)
 920        $path = canonicalize_path($path);
 921
 922        my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
 923        my $props;
 924        if ($is_dir) {
 925                (undef, undef, $props) = $gs->ra->get_dir($path, $r);
 926        }
 927        else {
 928                (undef, $props) = $gs->ra->get_file($path, $r, undef);
 929        }
 930        return $props;
 931}
 932
 933# cmd_propget (PROP, PATH)
 934# ------------------------
 935# Print the SVN property PROP for PATH.
 936sub cmd_propget {
 937        my ($prop, $path) = @_;
 938        $path = '.' if not defined $path;
 939        usage(1) if not defined $prop;
 940        my $props = get_svnprops($path);
 941        if (not defined $props->{$prop}) {
 942                fatal("`$path' does not have a `$prop' SVN property.");
 943        }
 944        print $props->{$prop} . "\n";
 945}
 946
 947# cmd_proplist (PATH)
 948# -------------------
 949# Print the list of SVN properties for PATH.
 950sub cmd_proplist {
 951        my $path = shift;
 952        $path = '.' if not defined $path;
 953        my $props = get_svnprops($path);
 954        print "Properties on '$path':\n";
 955        foreach (sort keys %{$props}) {
 956                print "  $_\n";
 957        }
 958}
 959
 960sub cmd_multi_init {
 961        my $url = shift;
 962        unless (defined $_trunk || @_branches || @_tags) {
 963                usage(1);
 964        }
 965
 966        $_prefix = '' unless defined $_prefix;
 967        if (defined $url) {
 968                $url = canonicalize_url($url);
 969                init_subdir(@_);
 970        }
 971        do_git_init_db();
 972        if (defined $_trunk) {
 973                $_trunk =~ s#^/+##;
 974                my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
 975                # try both old-style and new-style lookups:
 976                my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
 977                unless ($gs_trunk) {
 978                        my ($trunk_url, $trunk_path) =
 979                                              complete_svn_url($url, $_trunk);
 980                        $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
 981                                                   undef, $trunk_ref);
 982                }
 983        }
 984        return unless @_branches || @_tags;
 985        my $ra = $url ? Git::SVN::Ra->new($url) : undef;
 986        foreach my $path (@_branches) {
 987                complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
 988        }
 989        foreach my $path (@_tags) {
 990                complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
 991        }
 992}
 993
 994sub cmd_multi_fetch {
 995        $Git::SVN::no_reuse_existing = undef;
 996        my $remotes = Git::SVN::read_all_remotes();
 997        foreach my $repo_id (sort keys %$remotes) {
 998                if ($remotes->{$repo_id}->{url}) {
 999                        Git::SVN::fetch_all($repo_id, $remotes);
1000                }
1001        }
1002}
1003
1004# this command is special because it requires no metadata
1005sub cmd_commit_diff {
1006        my ($ta, $tb, $url) = @_;
1007        my $usage = "Usage: $0 commit-diff -r<revision> ".
1008                    "<tree-ish> <tree-ish> [<URL>]";
1009        fatal($usage) if (!defined $ta || !defined $tb);
1010        my $svn_path = '';
1011        if (!defined $url) {
1012                my $gs = eval { Git::SVN->new };
1013                if (!$gs) {
1014                        fatal("Needed URL or usable git-svn --id in ",
1015                              "the command-line\n", $usage);
1016                }
1017                $url = $gs->{url};
1018                $svn_path = $gs->{path};
1019        }
1020        unless (defined $_revision) {
1021                fatal("-r|--revision is a required argument\n", $usage);
1022        }
1023        if (defined $_message && defined $_file) {
1024                fatal("Both --message/-m and --file/-F specified ",
1025                      "for the commit message.\n",
1026                      "I have no idea what you mean");
1027        }
1028        if (defined $_file) {
1029                $_message = file_to_s($_file);
1030        } else {
1031                $_message ||= get_commit_entry($tb)->{log};
1032        }
1033        my $ra ||= Git::SVN::Ra->new($url);
1034        my $r = $_revision;
1035        if ($r eq 'HEAD') {
1036                $r = $ra->get_latest_revnum;
1037        } elsif ($r !~ /^\d+$/) {
1038                die "revision argument: $r not understood by git-svn\n";
1039        }
1040        my %ed_opts = ( r => $r,
1041                        log => $_message,
1042                        ra => $ra,
1043                        tree_a => $ta,
1044                        tree_b => $tb,
1045                        editor_cb => sub { print "Committed r$_[0]\n" },
1046                        svn_path => $svn_path );
1047        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1048                print "No changes\n$ta == $tb\n";
1049        }
1050}
1051
1052sub escape_uri_only {
1053        my ($uri) = @_;
1054        my @tmp;
1055        foreach (split m{/}, $uri) {
1056                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1057                push @tmp, $_;
1058        }
1059        join('/', @tmp);
1060}
1061
1062sub escape_url {
1063        my ($url) = @_;
1064        if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1065                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1066                $url = "$scheme://$domain$uri";
1067        }
1068        $url;
1069}
1070
1071sub cmd_info {
1072        my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1073        my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1074        if (exists $_[1]) {
1075                die "Too many arguments specified\n";
1076        }
1077
1078        my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1079
1080        if (!$file_type && !$diff_status) {
1081                print STDERR "svn: '$path' is not under version control\n";
1082                exit 1;
1083        }
1084
1085        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1086        unless ($gs) {
1087                die "Unable to determine upstream SVN information from ",
1088                    "working tree history\n";
1089        }
1090
1091        # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1092        $path = "." if $path eq "";
1093
1094        my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1095
1096        if ($_url) {
1097                print escape_url($full_url), "\n";
1098                return;
1099        }
1100
1101        my $result = "Path: $path\n";
1102        $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1103        $result .= "URL: " . escape_url($full_url) . "\n";
1104
1105        eval {
1106                my $repos_root = $gs->repos_root;
1107                Git::SVN::remove_username($repos_root);
1108                $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1109        };
1110        if ($@) {
1111                $result .= "Repository Root: (offline)\n";
1112        }
1113        ::_req_svn();
1114        $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1115                ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1116        $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1117
1118        $result .= "Node Kind: " .
1119                   ($file_type eq "dir" ? "directory" : "file") . "\n";
1120
1121        my $schedule = $diff_status eq "A"
1122                       ? "add"
1123                       : ($diff_status eq "D" ? "delete" : "normal");
1124        $result .= "Schedule: $schedule\n";
1125
1126        if ($diff_status eq "A") {
1127                print $result, "\n";
1128                return;
1129        }
1130
1131        my ($lc_author, $lc_rev, $lc_date_utc);
1132        my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1133        my $log = command_output_pipe(@args);
1134        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1135        while (<$log>) {
1136                if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1137                        $lc_author = $1;
1138                        $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1139                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
1140                        (undef, $lc_rev, undef) = ::extract_metadata($1);
1141                }
1142        }
1143        close $log;
1144
1145        Git::SVN::Log::set_local_timezone();
1146
1147        $result .= "Last Changed Author: $lc_author\n";
1148        $result .= "Last Changed Rev: $lc_rev\n";
1149        $result .= "Last Changed Date: " .
1150                   Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1151
1152        if ($file_type ne "dir") {
1153                my $text_last_updated_date =
1154                    ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1155                $result .=
1156                    "Text Last Updated: " .
1157                    Git::SVN::Log::format_svn_date($text_last_updated_date) .
1158                    "\n";
1159                my $checksum;
1160                if ($diff_status eq "D") {
1161                        my ($fh, $ctx) =
1162                            command_output_pipe(qw(cat-file blob), "HEAD:$path");
1163                        if ($file_type eq "link") {
1164                                my $file_name = <$fh>;
1165                                $checksum = md5sum("link $file_name");
1166                        } else {
1167                                $checksum = md5sum($fh);
1168                        }
1169                        command_close_pipe($fh, $ctx);
1170                } elsif ($file_type eq "link") {
1171                        my $file_name =
1172                            command(qw(cat-file blob), "HEAD:$path");
1173                        $checksum =
1174                            md5sum("link " . $file_name);
1175                } else {
1176                        open FILE, "<", $path or die $!;
1177                        $checksum = md5sum(\*FILE);
1178                        close FILE or die $!;
1179                }
1180                $result .= "Checksum: " . $checksum . "\n";
1181        }
1182
1183        print $result, "\n";
1184}
1185
1186sub cmd_reset {
1187        my $target = shift || $_revision or die "SVN revision required\n";
1188        $target = $1 if $target =~ /^r(\d+)$/;
1189        $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1190        my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1191        unless ($gs) {
1192                die "Unable to determine upstream SVN information from ".
1193                    "history\n";
1194        }
1195        my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1196        die "Cannot find SVN revision $target\n" unless defined($c);
1197        $gs->rev_map_set($r, $c, 'reset', $uuid);
1198        print "r$r = $c ($gs->{ref_id})\n";
1199}
1200
1201sub cmd_gc {
1202        if (!$can_compress) {
1203                warn "Compress::Zlib could not be found; unhandled.log " .
1204                     "files will not be compressed.\n";
1205        }
1206        find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1207}
1208
1209########################### utility functions #########################
1210
1211sub rebase_cmd {
1212        my @cmd = qw/rebase/;
1213        push @cmd, '-v' if $_verbose;
1214        push @cmd, qw/--merge/ if $_merge;
1215        push @cmd, "--strategy=$_strategy" if $_strategy;
1216        @cmd;
1217}
1218
1219sub post_fetch_checkout {
1220        return if $_no_checkout;
1221        my $gs = $Git::SVN::_head or return;
1222        return if verify_ref('refs/heads/master^0');
1223
1224        # look for "trunk" ref if it exists
1225        my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1226        my $fetch = $remote->{fetch};
1227        if ($fetch) {
1228                foreach my $p (keys %$fetch) {
1229                        basename($fetch->{$p}) eq 'trunk' or next;
1230                        $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1231                        last;
1232                }
1233        }
1234
1235        my $valid_head = verify_ref('HEAD^0');
1236        command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1237        return if ($valid_head || !verify_ref('HEAD^0'));
1238
1239        return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1240        my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1241        return if -f $index;
1242
1243        return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1244        return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1245        command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1246        print STDERR "Checked out HEAD:\n  ",
1247                     $gs->full_url, " r", $gs->last_rev, "\n";
1248        $gs->mkemptydirs($gs->last_rev);
1249}
1250
1251sub complete_svn_url {
1252        my ($url, $path) = @_;
1253        $path =~ s#/+$##;
1254        if ($path !~ m#^[a-z\+]+://#) {
1255                if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1256                        fatal("E: '$path' is not a complete URL ",
1257                              "and a separate URL is not specified");
1258                }
1259                return ($url, $path);
1260        }
1261        return ($path, '');
1262}
1263
1264sub complete_url_ls_init {
1265        my ($ra, $repo_path, $switch, $pfx) = @_;
1266        unless ($repo_path) {
1267                print STDERR "W: $switch not specified\n";
1268                return;
1269        }
1270        $repo_path =~ s#/+$##;
1271        if ($repo_path =~ m#^[a-z\+]+://#) {
1272                $ra = Git::SVN::Ra->new($repo_path);
1273                $repo_path = '';
1274        } else {
1275                $repo_path =~ s#^/+##;
1276                unless ($ra) {
1277                        fatal("E: '$repo_path' is not a complete URL ",
1278                              "and a separate URL is not specified");
1279                }
1280        }
1281        my $url = $ra->{url};
1282        my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1283        my $k = "svn-remote.$gs->{repo_id}.url";
1284        my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1285        if ($orig_url && ($orig_url ne $gs->{url})) {
1286                die "$k already set: $orig_url\n",
1287                    "wanted to set to: $gs->{url}\n";
1288        }
1289        command_oneline('config', $k, $gs->{url}) unless $orig_url;
1290        my $remote_path = "$gs->{path}/$repo_path";
1291        $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1292        $remote_path =~ s#/+#/#g;
1293        $remote_path =~ s#^/##g;
1294        $remote_path .= "/*" if $remote_path !~ /\*/;
1295        my ($n) = ($switch =~ /^--(\w+)/);
1296        if (length $pfx && $pfx !~ m#/$#) {
1297                die "--prefix='$pfx' must have a trailing slash '/'\n";
1298        }
1299        command_noisy('config',
1300                      '--add',
1301                      "svn-remote.$gs->{repo_id}.$n",
1302                      "$remote_path:refs/remotes/$pfx*" .
1303                        ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1304}
1305
1306sub verify_ref {
1307        my ($ref) = @_;
1308        eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1309                               { STDERR => 0 }); };
1310}
1311
1312sub get_tree_from_treeish {
1313        my ($treeish) = @_;
1314        # $treeish can be a symbolic ref, too:
1315        my $type = command_oneline(qw/cat-file -t/, $treeish);
1316        my $expected;
1317        while ($type eq 'tag') {
1318                ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1319        }
1320        if ($type eq 'commit') {
1321                $expected = (grep /^tree /, command(qw/cat-file commit/,
1322                                                    $treeish))[0];
1323                ($expected) = ($expected =~ /^tree ($sha1)$/o);
1324                die "Unable to get tree from $treeish\n" unless $expected;
1325        } elsif ($type eq 'tree') {
1326                $expected = $treeish;
1327        } else {
1328                die "$treeish is a $type, expected tree, tag or commit\n";
1329        }
1330        return $expected;
1331}
1332
1333sub get_commit_entry {
1334        my ($treeish) = shift;
1335        my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1336        my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1337        my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1338        open my $log_fh, '>', $commit_editmsg or croak $!;
1339
1340        my $type = command_oneline(qw/cat-file -t/, $treeish);
1341        if ($type eq 'commit' || $type eq 'tag') {
1342                my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1343                                                         $type, $treeish);
1344                my $in_msg = 0;
1345                my $author;
1346                my $saw_from = 0;
1347                my $msgbuf = "";
1348                while (<$msg_fh>) {
1349                        if (!$in_msg) {
1350                                $in_msg = 1 if (/^\s*$/);
1351                                $author = $1 if (/^author (.*>)/);
1352                        } elsif (/^git-svn-id: /) {
1353                                # skip this for now, we regenerate the
1354                                # correct one on re-fetch anyways
1355                                # TODO: set *:merge properties or like...
1356                        } else {
1357                                if (/^From:/ || /^Signed-off-by:/) {
1358                                        $saw_from = 1;
1359                                }
1360                                $msgbuf .= $_;
1361                        }
1362                }
1363                $msgbuf =~ s/\s+$//s;
1364                if ($Git::SVN::_add_author_from && defined($author)
1365                    && !$saw_from) {
1366                        $msgbuf .= "\n\nFrom: $author";
1367                }
1368                print $log_fh $msgbuf or croak $!;
1369                command_close_pipe($msg_fh, $ctx);
1370        }
1371        close $log_fh or croak $!;
1372
1373        if ($_edit || ($type eq 'tree')) {
1374                chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1375                system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1376        }
1377        rename $commit_editmsg, $commit_msg or croak $!;
1378        {
1379                require Encode;
1380                # SVN requires messages to be UTF-8 when entering the repo
1381                local $/;
1382                open $log_fh, '<', $commit_msg or croak $!;
1383                binmode $log_fh;
1384                chomp($log_entry{log} = <$log_fh>);
1385
1386                my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1387                my $msg = $log_entry{log};
1388
1389                eval { $msg = Encode::decode($enc, $msg, 1) };
1390                if ($@) {
1391                        die "Could not decode as $enc:\n", $msg,
1392                            "\nPerhaps you need to set i18n.commitencoding\n";
1393                }
1394
1395                eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1396                die "Could not encode as UTF-8:\n$msg\n" if $@;
1397
1398                $log_entry{log} = $msg;
1399
1400                close $log_fh or croak $!;
1401        }
1402        unlink $commit_msg;
1403        \%log_entry;
1404}
1405
1406sub s_to_file {
1407        my ($str, $file, $mode) = @_;
1408        open my $fd,'>',$file or croak $!;
1409        print $fd $str,"\n" or croak $!;
1410        close $fd or croak $!;
1411        chmod ($mode &~ umask, $file) if (defined $mode);
1412}
1413
1414sub file_to_s {
1415        my $file = shift;
1416        open my $fd,'<',$file or croak "$!: file: $file\n";
1417        local $/;
1418        my $ret = <$fd>;
1419        close $fd or croak $!;
1420        $ret =~ s/\s*$//s;
1421        return $ret;
1422}
1423
1424# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1425sub load_authors {
1426        open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1427        my $log = $cmd eq 'log';
1428        while (<$authors>) {
1429                chomp;
1430                next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1431                my ($user, $name, $email) = ($1, $2, $3);
1432                if ($log) {
1433                        $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1434                } else {
1435                        $users{$user} = [$name, $email];
1436                }
1437        }
1438        close $authors or croak $!;
1439}
1440
1441# convert GetOpt::Long specs for use by git-config
1442sub read_git_config {
1443        my $opts = shift;
1444        my @config_only;
1445        foreach my $o (keys %$opts) {
1446                # if we have mixedCase and a long option-only, then
1447                # it's a config-only variable that we don't need for
1448                # the command-line.
1449                push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1450                my $v = $opts->{$o};
1451                my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1452                $key =~ s/-//g;
1453                my $arg = 'git config';
1454                $arg .= ' --int' if ($o =~ /[:=]i$/);
1455                $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1456                if (ref $v eq 'ARRAY') {
1457                        chomp(my @tmp = `$arg --get-all svn.$key`);
1458                        @$v = @tmp if @tmp;
1459                } else {
1460                        chomp(my $tmp = `$arg --get svn.$key`);
1461                        if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1462                                $$v = $tmp;
1463                        }
1464                }
1465        }
1466        delete @$opts{@config_only} if @config_only;
1467}
1468
1469sub extract_metadata {
1470        my $id = shift or return (undef, undef, undef);
1471        my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1472                                                        \s([a-f\d\-]+)$/ix);
1473        if (!defined $rev || !$uuid || !$url) {
1474                # some of the original repositories I made had
1475                # identifiers like this:
1476                ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1477        }
1478        return ($url, $rev, $uuid);
1479}
1480
1481sub cmt_metadata {
1482        return extract_metadata((grep(/^git-svn-id: /,
1483                command(qw/cat-file commit/, shift)))[-1]);
1484}
1485
1486sub cmt_sha2rev_batch {
1487        my %s2r;
1488        my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1489        my $list = shift;
1490
1491        foreach my $sha (@{$list}) {
1492                my $first = 1;
1493                my $size = 0;
1494                print $out $sha, "\n";
1495
1496                while (my $line = <$in>) {
1497                        if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1498                                last;
1499                        } elsif ($first &&
1500                               $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1501                                $first = 0;
1502                                $size = $1;
1503                                next;
1504                        } elsif ($line =~ /^(git-svn-id: )/) {
1505                                my (undef, $rev, undef) =
1506                                                      extract_metadata($line);
1507                                $s2r{$sha} = $rev;
1508                        }
1509
1510                        $size -= length($line);
1511                        last if ($size == 0);
1512                }
1513        }
1514
1515        command_close_bidi_pipe($pid, $in, $out, $ctx);
1516
1517        return \%s2r;
1518}
1519
1520sub working_head_info {
1521        my ($head, $refs) = @_;
1522        my @args = qw/log --no-color --no-decorate --first-parent
1523                      --pretty=medium/;
1524        my ($fh, $ctx) = command_output_pipe(@args, $head);
1525        my $hash;
1526        my %max;
1527        while (<$fh>) {
1528                if ( m{^commit ($::sha1)$} ) {
1529                        unshift @$refs, $hash if $hash and $refs;
1530                        $hash = $1;
1531                        next;
1532                }
1533                next unless s{^\s*(git-svn-id:)}{$1};
1534                my ($url, $rev, $uuid) = extract_metadata($_);
1535                if (defined $url && defined $rev) {
1536                        next if $max{$url} and $max{$url} < $rev;
1537                        if (my $gs = Git::SVN->find_by_url($url)) {
1538                                my $c = $gs->rev_map_get($rev, $uuid);
1539                                if ($c && $c eq $hash) {
1540                                        close $fh; # break the pipe
1541                                        return ($url, $rev, $uuid, $gs);
1542                                } else {
1543                                        $max{$url} ||= $gs->rev_map_max;
1544                                }
1545                        }
1546                }
1547        }
1548        command_close_pipe($fh, $ctx);
1549        (undef, undef, undef, undef);
1550}
1551
1552sub read_commit_parents {
1553        my ($parents, $c) = @_;
1554        chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1555        $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1556        @{$parents->{$c}} = split(/ /, $p);
1557}
1558
1559sub linearize_history {
1560        my ($gs, $refs) = @_;
1561        my %parents;
1562        foreach my $c (@$refs) {
1563                read_commit_parents(\%parents, $c);
1564        }
1565
1566        my @linear_refs;
1567        my %skip = ();
1568        my $last_svn_commit = $gs->last_commit;
1569        foreach my $c (reverse @$refs) {
1570                next if $c eq $last_svn_commit;
1571                last if $skip{$c};
1572
1573                unshift @linear_refs, $c;
1574                $skip{$c} = 1;
1575
1576                # we only want the first parent to diff against for linear
1577                # history, we save the rest to inject when we finalize the
1578                # svn commit
1579                my $fp_a = verify_ref("$c~1");
1580                my $fp_b = shift @{$parents{$c}} if $parents{$c};
1581                if (!$fp_a || !$fp_b) {
1582                        die "Commit $c\n",
1583                            "has no parent commit, and therefore ",
1584                            "nothing to diff against.\n",
1585                            "You should be working from a repository ",
1586                            "originally created by git-svn\n";
1587                }
1588                if ($fp_a ne $fp_b) {
1589                        die "$c~1 = $fp_a, however parsing commit $c ",
1590                            "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1591                }
1592
1593                foreach my $p (@{$parents{$c}}) {
1594                        $skip{$p} = 1;
1595                }
1596        }
1597        (\@linear_refs, \%parents);
1598}
1599
1600sub find_file_type_and_diff_status {
1601        my ($path) = @_;
1602        return ('dir', '') if $path eq '';
1603
1604        my $diff_output =
1605            command_oneline(qw(diff --cached --name-status --), $path) || "";
1606        my $diff_status = (split(' ', $diff_output))[0] || "";
1607
1608        my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1609
1610        return (undef, undef) if !$diff_status && !$ls_tree;
1611
1612        if ($diff_status eq "A") {
1613                return ("link", $diff_status) if -l $path;
1614                return ("dir", $diff_status) if -d $path;
1615                return ("file", $diff_status);
1616        }
1617
1618        my $mode = (split(' ', $ls_tree))[0] || "";
1619
1620        return ("link", $diff_status) if $mode eq "120000";
1621        return ("dir", $diff_status) if $mode eq "040000";
1622        return ("file", $diff_status);
1623}
1624
1625sub md5sum {
1626        my $arg = shift;
1627        my $ref = ref $arg;
1628        my $md5 = Digest::MD5->new();
1629        if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1630                $md5->addfile($arg) or croak $!;
1631        } elsif ($ref eq 'SCALAR') {
1632                $md5->add($$arg) or croak $!;
1633        } elsif (!$ref) {
1634                $md5->add($arg) or croak $!;
1635        } else {
1636                ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1637        }
1638        return $md5->hexdigest();
1639}
1640
1641sub gc_directory {
1642        if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1643                my $out_filename = $_ . ".gz";
1644                open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1645                binmode $in_fh;
1646                my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1647                                die "Unable to open $out_filename: $!\n";
1648
1649                my $res;
1650                while ($res = sysread($in_fh, my $str, 1024)) {
1651                        $gz->gzwrite($str) or
1652                                die "Unable to write: ".$gz->gzerror()."!\n";
1653                }
1654                unlink $_ or die "unlink $File::Find::name: $!\n";
1655        } elsif (-f $_ && basename($_) eq "index") {
1656                unlink $_ or die "unlink $_: $!\n";
1657        }
1658}
1659
1660package Git::SVN;
1661use strict;
1662use warnings;
1663use Fcntl qw/:DEFAULT :seek/;
1664use constant rev_map_fmt => 'NH40';
1665use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1666            $_repack $_repack_flags $_use_svm_props $_head
1667            $_use_svnsync_props $no_reuse_existing $_minimize_url
1668            $_use_log_author $_add_author_from $_localtime/;
1669use Carp qw/croak/;
1670use File::Path qw/mkpath/;
1671use File::Copy qw/copy/;
1672use IPC::Open3;
1673use Memoize;  # core since 5.8.0, Jul 2002
1674use Memoize::Storable;
1675
1676my ($_gc_nr, $_gc_period);
1677
1678# properties that we do not log:
1679my %SKIP_PROP;
1680BEGIN {
1681        %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1682                                        svn:special svn:executable
1683                                        svn:entry:committed-rev
1684                                        svn:entry:last-author
1685                                        svn:entry:uuid
1686                                        svn:entry:committed-date/;
1687
1688        # some options are read globally, but can be overridden locally
1689        # per [svn-remote "..."] section.  Command-line options will *NOT*
1690        # override options set in an [svn-remote "..."] section
1691        no strict 'refs';
1692        for my $option (qw/follow_parent no_metadata use_svm_props
1693                           use_svnsync_props/) {
1694                my $key = $option;
1695                $key =~ tr/_//d;
1696                my $prop = "-$option";
1697                *$option = sub {
1698                        my ($self) = @_;
1699                        return $self->{$prop} if exists $self->{$prop};
1700                        my $k = "svn-remote.$self->{repo_id}.$key";
1701                        eval { command_oneline(qw/config --get/, $k) };
1702                        if ($@) {
1703                                $self->{$prop} = ${"Git::SVN::_$option"};
1704                        } else {
1705                                my $v = command_oneline(qw/config --bool/,$k);
1706                                $self->{$prop} = $v eq 'false' ? 0 : 1;
1707                        }
1708                        return $self->{$prop};
1709                }
1710        }
1711}
1712
1713
1714my (%LOCKFILES, %INDEX_FILES);
1715END {
1716        unlink keys %LOCKFILES if %LOCKFILES;
1717        unlink keys %INDEX_FILES if %INDEX_FILES;
1718}
1719
1720sub resolve_local_globs {
1721        my ($url, $fetch, $glob_spec) = @_;
1722        return unless defined $glob_spec;
1723        my $ref = $glob_spec->{ref};
1724        my $path = $glob_spec->{path};
1725        foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1726                next unless m#^$ref->{regex}$#;
1727                my $p = $1;
1728                my $pathname = desanitize_refname($path->full_path($p));
1729                my $refname = desanitize_refname($ref->full_path($p));
1730                if (my $existing = $fetch->{$pathname}) {
1731                        if ($existing ne $refname) {
1732                                die "Refspec conflict:\n",
1733                                    "existing: $existing\n",
1734                                    " globbed: $refname\n";
1735                        }
1736                        my $u = (::cmt_metadata("$refname"))[0];
1737                        $u =~ s!^\Q$url\E(/|$)!! or die
1738                          "$refname: '$url' not found in '$u'\n";
1739                        if ($pathname ne $u) {
1740                                warn "W: Refspec glob conflict ",
1741                                     "(ref: $refname):\n",
1742                                     "expected path: $pathname\n",
1743                                     "    real path: $u\n",
1744                                     "Continuing ahead with $u\n";
1745                                next;
1746                        }
1747                } else {
1748                        $fetch->{$pathname} = $refname;
1749                }
1750        }
1751}
1752
1753sub parse_revision_argument {
1754        my ($base, $head) = @_;
1755        if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1756                return ($base, $head);
1757        }
1758        return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1759        return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1760        return ($head, $head) if ($::_revision eq 'HEAD');
1761        return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1762        return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1763        die "revision argument: $::_revision not understood by git-svn\n";
1764}
1765
1766sub fetch_all {
1767        my ($repo_id, $remotes) = @_;
1768        if (ref $repo_id) {
1769                my $gs = $repo_id;
1770                $repo_id = undef;
1771                $repo_id = $gs->{repo_id};
1772        }
1773        $remotes ||= read_all_remotes();
1774        my $remote = $remotes->{$repo_id} or
1775                     die "[svn-remote \"$repo_id\"] unknown\n";
1776        my $fetch = $remote->{fetch};
1777        my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1778        my (@gs, @globs);
1779        my $ra = Git::SVN::Ra->new($url);
1780        my $uuid = $ra->get_uuid;
1781        my $head = $ra->get_latest_revnum;
1782
1783        # ignore errors, $head revision may not even exist anymore
1784        eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
1785        warn "W: $@\n" if $@;
1786
1787        my $base = defined $fetch ? $head : 0;
1788
1789        # read the max revs for wildcard expansion (branches/*, tags/*)
1790        foreach my $t (qw/branches tags/) {
1791                defined $remote->{$t} or next;
1792                push @globs, @{$remote->{$t}};
1793
1794                my $max_rev = eval { tmp_config(qw/--int --get/,
1795                                         "svn-remote.$repo_id.${t}-maxRev") };
1796                if (defined $max_rev && ($max_rev < $base)) {
1797                        $base = $max_rev;
1798                } elsif (!defined $max_rev) {
1799                        $base = 0;
1800                }
1801        }
1802
1803        if ($fetch) {
1804                foreach my $p (sort keys %$fetch) {
1805                        my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1806                        my $lr = $gs->rev_map_max;
1807                        if (defined $lr) {
1808                                $base = $lr if ($lr < $base);
1809                        }
1810                        push @gs, $gs;
1811                }
1812        }
1813
1814        ($base, $head) = parse_revision_argument($base, $head);
1815        $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1816}
1817
1818sub read_all_remotes {
1819        my $r = {};
1820        my $use_svm_props = eval { command_oneline(qw/config --bool
1821            svn.useSvmProps/) };
1822        $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1823        my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1824        foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1825                if (m!^(.+)\.fetch=$svn_refspec$!) {
1826                        my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1827                        die("svn-remote.$remote: remote ref '$remote_ref' "
1828                            . "must start with 'refs/'\n")
1829                                unless $remote_ref =~ m{^refs/};
1830                        $local_ref = uri_decode($local_ref);
1831                        $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1832                        $r->{$remote}->{svm} = {} if $use_svm_props;
1833                } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1834                        $r->{$1}->{svm} = {};
1835                } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1836                        $r->{$1}->{url} = $2;
1837                } elsif (m!^(.+)\.pushurl=\s*(.*)\s*$!) {
1838                        $r->{$1}->{pushurl} = $2;
1839                } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1840                        my ($remote, $t, $local_ref, $remote_ref) =
1841                                                             ($1, $2, $3, $4);
1842                        die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1843                            . "must start with 'refs/'\n")
1844                                unless $remote_ref =~ m{^refs/};
1845                        $local_ref = uri_decode($local_ref);
1846                        my $rs = {
1847                            t => $t,
1848                            remote => $remote,
1849                            path => Git::SVN::GlobSpec->new($local_ref, 1),
1850                            ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
1851                        if (length($rs->{ref}->{right}) != 0) {
1852                                die "The '*' glob character must be the last ",
1853                                    "character of '$remote_ref'\n";
1854                        }
1855                        push @{ $r->{$remote}->{$t} }, $rs;
1856                }
1857        }
1858
1859        map {
1860                if (defined $r->{$_}->{svm}) {
1861                        my $svm;
1862                        eval {
1863                                my $section = "svn-remote.$_";
1864                                $svm = {
1865                                        source => tmp_config('--get',
1866                                            "$section.svm-source"),
1867                                        replace => tmp_config('--get',
1868                                            "$section.svm-replace"),
1869                                }
1870                        };
1871                        $r->{$_}->{svm} = $svm;
1872                }
1873        } keys %$r;
1874
1875        $r;
1876}
1877
1878sub init_vars {
1879        $_gc_nr = $_gc_period = 1000;
1880        if (defined $_repack || defined $_repack_flags) {
1881               warn "Repack options are obsolete; they have no effect.\n";
1882        }
1883}
1884
1885sub verify_remotes_sanity {
1886        return unless -d $ENV{GIT_DIR};
1887        my %seen;
1888        foreach (command(qw/config -l/)) {
1889                if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1890                        if ($seen{$1}) {
1891                                die "Remote ref refs/remote/$1 is tracked by",
1892                                    "\n  \"$_\"\nand\n  \"$seen{$1}\"\n",
1893                                    "Please resolve this ambiguity in ",
1894                                    "your git configuration file before ",
1895                                    "continuing\n";
1896                        }
1897                        $seen{$1} = $_;
1898                }
1899        }
1900}
1901
1902sub find_existing_remote {
1903        my ($url, $remotes) = @_;
1904        return undef if $no_reuse_existing;
1905        my $existing;
1906        foreach my $repo_id (keys %$remotes) {
1907                my $u = $remotes->{$repo_id}->{url} or next;
1908                next if $u ne $url;
1909                $existing = $repo_id;
1910                last;
1911        }
1912        $existing;
1913}
1914
1915sub init_remote_config {
1916        my ($self, $url, $no_write) = @_;
1917        $url =~ s!/+$!!; # strip trailing slash
1918        my $r = read_all_remotes();
1919        my $existing = find_existing_remote($url, $r);
1920        if ($existing) {
1921                unless ($no_write) {
1922                        print STDERR "Using existing ",
1923                                     "[svn-remote \"$existing\"]\n";
1924                }
1925                $self->{repo_id} = $existing;
1926        } elsif ($_minimize_url) {
1927                my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1928                $existing = find_existing_remote($min_url, $r);
1929                if ($existing) {
1930                        unless ($no_write) {
1931                                print STDERR "Using existing ",
1932                                             "[svn-remote \"$existing\"]\n";
1933                        }
1934                        $self->{repo_id} = $existing;
1935                }
1936                if ($min_url ne $url) {
1937                        unless ($no_write) {
1938                                print STDERR "Using higher level of URL: ",
1939                                             "$url => $min_url\n";
1940                        }
1941                        my $old_path = $self->{path};
1942                        $self->{path} = $url;
1943                        $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1944                        if (length $old_path) {
1945                                $self->{path} .= "/$old_path";
1946                        }
1947                        $url = $min_url;
1948                }
1949        }
1950        my $orig_url;
1951        if (!$existing) {
1952                # verify that we aren't overwriting anything:
1953                $orig_url = eval {
1954                        command_oneline('config', '--get',
1955                                        "svn-remote.$self->{repo_id}.url")
1956                };
1957                if ($orig_url && ($orig_url ne $url)) {
1958                        die "svn-remote.$self->{repo_id}.url already set: ",
1959                            "$orig_url\nwanted to set to: $url\n";
1960                }
1961        }
1962        my ($xrepo_id, $xpath) = find_ref($self->refname);
1963        if (!$no_write && defined $xpath) {
1964                die "svn-remote.$xrepo_id.fetch already set to track ",
1965                    "$xpath:", $self->refname, "\n";
1966        }
1967        unless ($no_write) {
1968                command_noisy('config',
1969                              "svn-remote.$self->{repo_id}.url", $url);
1970                $self->{path} =~ s{^/}{};
1971                $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1972                command_noisy('config', '--add',
1973                              "svn-remote.$self->{repo_id}.fetch",
1974                              "$self->{path}:".$self->refname);
1975        }
1976        $self->{url} = $url;
1977}
1978
1979sub find_by_url { # repos_root and, path are optional
1980        my ($class, $full_url, $repos_root, $path) = @_;
1981
1982        return undef unless defined $full_url;
1983        remove_username($full_url);
1984        remove_username($repos_root) if defined $repos_root;
1985        my $remotes = read_all_remotes();
1986        if (defined $full_url && defined $repos_root && !defined $path) {
1987                $path = $full_url;
1988                $path =~ s#^\Q$repos_root\E(?:/|$)##;
1989        }
1990        foreach my $repo_id (keys %$remotes) {
1991                my $u = $remotes->{$repo_id}->{url} or next;
1992                remove_username($u);
1993                next if defined $repos_root && $repos_root ne $u;
1994
1995                my $fetch = $remotes->{$repo_id}->{fetch} || {};
1996                foreach my $t (qw/branches tags/) {
1997                        foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1998                                resolve_local_globs($u, $fetch, $globspec);
1999                        }
2000                }
2001                my $p = $path;
2002                my $rwr = rewrite_root({repo_id => $repo_id});
2003                my $svm = $remotes->{$repo_id}->{svm}
2004                        if defined $remotes->{$repo_id}->{svm};
2005                unless (defined $p) {
2006                        $p = $full_url;
2007                        my $z = $u;
2008                        my $prefix = '';
2009                        if ($rwr) {
2010                                $z = $rwr;
2011                                remove_username($z);
2012                        } elsif (defined $svm) {
2013                                $z = $svm->{source};
2014                                $prefix = $svm->{replace};
2015                                $prefix =~ s#^\Q$u\E(?:/|$)##;
2016                                $prefix =~ s#/$##;
2017                        }
2018                        $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2019                }
2020                foreach my $f (keys %$fetch) {
2021                        next if $f ne $p;
2022                        return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2023                }
2024        }
2025        undef;
2026}
2027
2028sub init {
2029        my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2030        my $self = _new($class, $repo_id, $ref_id, $path);
2031        if (defined $url) {
2032                $self->init_remote_config($url, $no_write);
2033        }
2034        $self;
2035}
2036
2037sub find_ref {
2038        my ($ref_id) = @_;
2039        foreach (command(qw/config -l/)) {
2040                next unless m!^svn-remote\.(.+)\.fetch=
2041                              \s*(.*?)\s*:\s*(.+?)\s*$!x;
2042                my ($repo_id, $path, $ref) = ($1, $2, $3);
2043                if ($ref eq $ref_id) {
2044                        $path = '' if ($path =~ m#^\./?#);
2045                        return ($repo_id, $path);
2046                }
2047        }
2048        (undef, undef, undef);
2049}
2050
2051sub new {
2052        my ($class, $ref_id, $repo_id, $path) = @_;
2053        if (defined $ref_id && !defined $repo_id && !defined $path) {
2054                ($repo_id, $path) = find_ref($ref_id);
2055                if (!defined $repo_id) {
2056                        die "Could not find a \"svn-remote.*.fetch\" key ",
2057                            "in the repository configuration matching: ",
2058                            "$ref_id\n";
2059                }
2060        }
2061        my $self = _new($class, $repo_id, $ref_id, $path);
2062        if (!defined $self->{path} || !length $self->{path}) {
2063                my $fetch = command_oneline('config', '--get',
2064                                            "svn-remote.$repo_id.fetch",
2065                                            ":$ref_id\$") or
2066                     die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2067                         "\":$ref_id\$\" in config\n";
2068                ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2069        }
2070        $self->{path} =~ s{/+}{/}g;
2071        $self->{path} =~ s{\A/}{};
2072        $self->{path} =~ s{/\z}{};
2073        $self->{url} = command_oneline('config', '--get',
2074                                       "svn-remote.$repo_id.url") or
2075                  die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2076        $self->{pushurl} = eval { command_oneline('config', '--get',
2077                                  "svn-remote.$repo_id.pushurl") };
2078        $self->rebuild;
2079        $self;
2080}
2081
2082sub refname {
2083        my ($refname) = $_[0]->{ref_id} ;
2084
2085        # It cannot end with a slash /, we'll throw up on this because
2086        # SVN can't have directories with a slash in their name, either:
2087        if ($refname =~ m{/$}) {
2088                die "ref: '$refname' ends with a trailing slash, this is ",
2089                    "not permitted by git nor Subversion\n";
2090        }
2091
2092        # It cannot have ASCII control character space, tilde ~, caret ^,
2093        # colon :, question-mark ?, asterisk *, space, or open bracket [
2094        # anywhere.
2095        #
2096        # Additionally, % must be escaped because it is used for escaping
2097        # and we want our escaped refname to be reversible
2098        $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2099
2100        # no slash-separated component can begin with a dot .
2101        # /.* becomes /%2E*
2102        $refname =~ s{/\.}{/%2E}g;
2103
2104        # It cannot have two consecutive dots .. anywhere
2105        # .. becomes %2E%2E
2106        $refname =~ s{\.\.}{%2E%2E}g;
2107
2108        # trailing dots and .lock are not allowed
2109        # .$ becomes %2E and .lock becomes %2Elock
2110        $refname =~ s{\.(?=$|lock$)}{%2E};
2111
2112        # the sequence @{ is used to access the reflog
2113        # @{ becomes %40{
2114        $refname =~ s{\@\{}{%40\{}g;
2115
2116        return $refname;
2117}
2118
2119sub desanitize_refname {
2120        my ($refname) = @_;
2121        $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2122        return $refname;
2123}
2124
2125sub svm_uuid {
2126        my ($self) = @_;
2127        return $self->{svm}->{uuid} if $self->svm;
2128        $self->ra;
2129        unless ($self->{svm}) {
2130                die "SVM UUID not cached, and reading remotely failed\n";
2131        }
2132        $self->{svm}->{uuid};
2133}
2134
2135sub svm {
2136        my ($self) = @_;
2137        return $self->{svm} if $self->{svm};
2138        my $svm;
2139        # see if we have it in our config, first:
2140        eval {
2141                my $section = "svn-remote.$self->{repo_id}";
2142                $svm = {
2143                  source => tmp_config('--get', "$section.svm-source"),
2144                  uuid => tmp_config('--get', "$section.svm-uuid"),
2145                  replace => tmp_config('--get', "$section.svm-replace"),
2146                }
2147        };
2148        if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2149                $self->{svm} = $svm;
2150        }
2151        $self->{svm};
2152}
2153
2154sub _set_svm_vars {
2155        my ($self, $ra) = @_;
2156        return $ra if $self->svm;
2157
2158        my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2159                    "(svm:source, svm:uuid) ",
2160                    "from the following URLs:\n" );
2161        sub read_svm_props {
2162                my ($self, $ra, $path, $r) = @_;
2163                my $props = ($ra->get_dir($path, $r))[2];
2164                my $src = $props->{'svm:source'};
2165                my $uuid = $props->{'svm:uuid'};
2166                return undef if (!$src || !$uuid);
2167
2168                chomp($src, $uuid);
2169
2170                $uuid =~ m{^[0-9a-f\-]{30,}$}i
2171                    or die "doesn't look right - svm:uuid is '$uuid'\n";
2172
2173                # the '!' is used to mark the repos_root!/relative/path
2174                $src =~ s{/?!/?}{/};
2175                $src =~ s{/+$}{}; # no trailing slashes please
2176                # username is of no interest
2177                $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2178
2179                my $replace = $ra->{url};
2180                $replace .= "/$path" if length $path;
2181
2182                my $section = "svn-remote.$self->{repo_id}";
2183                tmp_config("$section.svm-source", $src);
2184                tmp_config("$section.svm-replace", $replace);
2185                tmp_config("$section.svm-uuid", $uuid);
2186                $self->{svm} = {
2187                        source => $src,
2188                        uuid => $uuid,
2189                        replace => $replace
2190                };
2191        }
2192
2193        my $r = $ra->get_latest_revnum;
2194        my $path = $self->{path};
2195        my %tried;
2196        while (length $path) {
2197                unless ($tried{"$self->{url}/$path"}) {
2198                        return $ra if $self->read_svm_props($ra, $path, $r);
2199                        $tried{"$self->{url}/$path"} = 1;
2200                }
2201                $path =~ s#/?[^/]+$##;
2202        }
2203        die "Path: '$path' should be ''\n" if $path ne '';
2204        return $ra if $self->read_svm_props($ra, $path, $r);
2205        $tried{"$self->{url}/$path"} = 1;
2206
2207        if ($ra->{repos_root} eq $self->{url}) {
2208                die @err, (map { "  $_\n" } keys %tried), "\n";
2209        }
2210
2211        # nope, make sure we're connected to the repository root:
2212        my $ok;
2213        my @tried_b;
2214        $path = $ra->{svn_path};
2215        $ra = Git::SVN::Ra->new($ra->{repos_root});
2216        while (length $path) {
2217                unless ($tried{"$ra->{url}/$path"}) {
2218                        $ok = $self->read_svm_props($ra, $path, $r);
2219                        last if $ok;
2220                        $tried{"$ra->{url}/$path"} = 1;
2221                }
2222                $path =~ s#/?[^/]+$##;
2223        }
2224        die "Path: '$path' should be ''\n" if $path ne '';
2225        $ok ||= $self->read_svm_props($ra, $path, $r);
2226        $tried{"$ra->{url}/$path"} = 1;
2227        if (!$ok) {
2228                die @err, (map { "  $_\n" } keys %tried), "\n";
2229        }
2230        Git::SVN::Ra->new($self->{url});
2231}
2232
2233sub svnsync {
2234        my ($self) = @_;
2235        return $self->{svnsync} if $self->{svnsync};
2236
2237        if ($self->no_metadata) {
2238                die "Can't have both 'noMetadata' and ",
2239                    "'useSvnsyncProps' options set!\n";
2240        }
2241        if ($self->rewrite_root) {
2242                die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2243                    "options set!\n";
2244        }
2245        if ($self->rewrite_uuid) {
2246                die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2247                    "options set!\n";
2248        }
2249
2250        my $svnsync;
2251        # see if we have it in our config, first:
2252        eval {
2253                my $section = "svn-remote.$self->{repo_id}";
2254
2255                my $url = tmp_config('--get', "$section.svnsync-url");
2256                ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2257                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2258
2259                my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2260                ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2261                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2262
2263                $svnsync = { url => $url, uuid => $uuid }
2264        };
2265        if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2266                return $self->{svnsync} = $svnsync;
2267        }
2268
2269        my $err = "useSvnsyncProps set, but failed to read " .
2270                  "svnsync property: svn:sync-from-";
2271        my $rp = $self->ra->rev_proplist(0);
2272
2273        my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2274        ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2275                   die "doesn't look right - svn:sync-from-url is '$url'\n";
2276
2277        my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2278        ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2279                   die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2280
2281        my $section = "svn-remote.$self->{repo_id}";
2282        tmp_config('--add', "$section.svnsync-uuid", $uuid);
2283        tmp_config('--add', "$section.svnsync-url", $url);
2284        return $self->{svnsync} = { url => $url, uuid => $uuid };
2285}
2286
2287# this allows us to memoize our SVN::Ra UUID locally and avoid a
2288# remote lookup (useful for 'git svn log').
2289sub ra_uuid {
2290        my ($self) = @_;
2291        unless ($self->{ra_uuid}) {
2292                my $key = "svn-remote.$self->{repo_id}.uuid";
2293                my $uuid = eval { tmp_config('--get', $key) };
2294                if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2295                        $self->{ra_uuid} = $uuid;
2296                } else {
2297                        die "ra_uuid called without URL\n" unless $self->{url};
2298                        $self->{ra_uuid} = $self->ra->get_uuid;
2299                        tmp_config('--add', $key, $self->{ra_uuid});
2300                }
2301        }
2302        $self->{ra_uuid};
2303}
2304
2305sub _set_repos_root {
2306        my ($self, $repos_root) = @_;
2307        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2308        $repos_root ||= $self->ra->{repos_root};
2309        tmp_config($k, $repos_root);
2310        $repos_root;
2311}
2312
2313sub repos_root {
2314        my ($self) = @_;
2315        my $k = "svn-remote.$self->{repo_id}.reposRoot";
2316        eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2317}
2318
2319sub ra {
2320        my ($self) = shift;
2321        my $ra = Git::SVN::Ra->new($self->{url});
2322        $self->_set_repos_root($ra->{repos_root});
2323        if ($self->use_svm_props && !$self->{svm}) {
2324                if ($self->no_metadata) {
2325                        die "Can't have both 'noMetadata' and ",
2326                            "'useSvmProps' options set!\n";
2327                } elsif ($self->use_svnsync_props) {
2328                        die "Can't have both 'useSvnsyncProps' and ",
2329                            "'useSvmProps' options set!\n";
2330                }
2331                $ra = $self->_set_svm_vars($ra);
2332                $self->{-want_revprops} = 1;
2333        }
2334        $ra;
2335}
2336
2337# prop_walk(PATH, REV, SUB)
2338# -------------------------
2339# Recursively traverse PATH at revision REV and invoke SUB for each
2340# directory that contains a SVN property.  SUB will be invoked as
2341# follows:  &SUB(gs, path, props);  where `gs' is this instance of
2342# Git::SVN, `path' the path to the directory where the properties
2343# `props' were found.  The `path' will be relative to point of checkout,
2344# that is, if url://repo/trunk is the current Git branch, and that
2345# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2346# as `path' (note the trailing `/').
2347sub prop_walk {
2348        my ($self, $path, $rev, $sub) = @_;
2349
2350        $path =~ s#^/##;
2351        my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2352        $path =~ s#^/*#/#g;
2353        my $p = $path;
2354        # Strip the irrelevant part of the path.
2355        $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2356        # Ensure the path is terminated by a `/'.
2357        $p =~ s#/*$#/#;
2358
2359        # The properties contain all the internal SVN stuff nobody
2360        # (usually) cares about.
2361        my $interesting_props = 0;
2362        foreach (keys %{$props}) {
2363                # If it doesn't start with `svn:', it must be a
2364                # user-defined property.
2365                ++$interesting_props and next if $_ !~ /^svn:/;
2366                # FIXME: Fragile, if SVN adds new public properties,
2367                # this needs to be updated.
2368                ++$interesting_props if /^svn:(?:ignore|keywords|executable
2369                                                 |eol-style|mime-type
2370                                                 |externals|needs-lock)$/x;
2371        }
2372        &$sub($self, $p, $props) if $interesting_props;
2373
2374        foreach (sort keys %$dirent) {
2375                next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2376                $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2377        }
2378}
2379
2380sub last_rev { ($_[0]->last_rev_commit)[0] }
2381sub last_commit { ($_[0]->last_rev_commit)[1] }
2382
2383# returns the newest SVN revision number and newest commit SHA1
2384sub last_rev_commit {
2385        my ($self) = @_;
2386        if (defined $self->{last_rev} && defined $self->{last_commit}) {
2387                return ($self->{last_rev}, $self->{last_commit});
2388        }
2389        my $c = ::verify_ref($self->refname.'^0');
2390        if ($c && !$self->use_svm_props && !$self->no_metadata) {
2391                my $rev = (::cmt_metadata($c))[1];
2392                if (defined $rev) {
2393                        ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2394                        return ($rev, $c);
2395                }
2396        }
2397        my $map_path = $self->map_path;
2398        unless (-e $map_path) {
2399                ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2400                return (undef, undef);
2401        }
2402        my ($rev, $commit) = $self->rev_map_max(1);
2403        ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2404        return ($rev, $commit);
2405}
2406
2407sub get_fetch_range {
2408        my ($self, $min, $max) = @_;
2409        $max ||= $self->ra->get_latest_revnum;
2410        $min ||= $self->rev_map_max;
2411        (++$min, $max);
2412}
2413
2414sub tmp_config {
2415        my (@args) = @_;
2416        my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2417        my $config = "$ENV{GIT_DIR}/svn/.metadata";
2418        if (! -f $config && -f $old_def_config) {
2419                rename $old_def_config, $config or
2420                       die "Failed rename $old_def_config => $config: $!\n";
2421        }
2422        my $old_config = $ENV{GIT_CONFIG};
2423        $ENV{GIT_CONFIG} = $config;
2424        $@ = undef;
2425        my @ret = eval {
2426                unless (-f $config) {
2427                        mkfile($config);
2428                        open my $fh, '>', $config or
2429                            die "Can't open $config: $!\n";
2430                        print $fh "; This file is used internally by ",
2431                                  "git-svn\n" or die
2432                                  "Couldn't write to $config: $!\n";
2433                        print $fh "; You should not have to edit it\n" or
2434                              die "Couldn't write to $config: $!\n";
2435                        close $fh or die "Couldn't close $config: $!\n";
2436                }
2437                command('config', @args);
2438        };
2439        my $err = $@;
2440        if (defined $old_config) {
2441                $ENV{GIT_CONFIG} = $old_config;
2442        } else {
2443                delete $ENV{GIT_CONFIG};
2444        }
2445        die $err if $err;
2446        wantarray ? @ret : $ret[0];
2447}
2448
2449sub tmp_index_do {
2450        my ($self, $sub) = @_;
2451        my $old_index = $ENV{GIT_INDEX_FILE};
2452        $ENV{GIT_INDEX_FILE} = $self->{index};
2453        $@ = undef;
2454        my @ret = eval {
2455                my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2456                mkpath([$dir]) unless -d $dir;
2457                &$sub;
2458        };
2459        my $err = $@;
2460        if (defined $old_index) {
2461                $ENV{GIT_INDEX_FILE} = $old_index;
2462        } else {
2463                delete $ENV{GIT_INDEX_FILE};
2464        }
2465        die $err if $err;
2466        wantarray ? @ret : $ret[0];
2467}
2468
2469sub assert_index_clean {
2470        my ($self, $treeish) = @_;
2471
2472        $self->tmp_index_do(sub {
2473                command_noisy('read-tree', $treeish) unless -e $self->{index};
2474                my $x = command_oneline('write-tree');
2475                my ($y) = (command(qw/cat-file commit/, $treeish) =~
2476                           /^tree ($::sha1)/mo);
2477                return if $y eq $x;
2478
2479                warn "Index mismatch: $y != $x\nrereading $treeish\n";
2480                unlink $self->{index} or die "unlink $self->{index}: $!\n";
2481                command_noisy('read-tree', $treeish);
2482                $x = command_oneline('write-tree');
2483                if ($y ne $x) {
2484                        ::fatal "trees ($treeish) $y != $x\n",
2485                                "Something is seriously wrong...";
2486                }
2487        });
2488}
2489
2490sub get_commit_parents {
2491        my ($self, $log_entry) = @_;
2492        my (%seen, @ret, @tmp);
2493        # legacy support for 'set-tree'; this is only used by set_tree_cb:
2494        if (my $ip = $self->{inject_parents}) {
2495                if (my $commit = delete $ip->{$log_entry->{revision}}) {
2496                        push @tmp, $commit;
2497                }
2498        }
2499        if (my $cur = ::verify_ref($self->refname.'^0')) {
2500                push @tmp, $cur;
2501        }
2502        if (my $ipd = $self->{inject_parents_dcommit}) {
2503                if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2504                        push @tmp, @$commit;
2505                }
2506        }
2507        push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2508        while (my $p = shift @tmp) {
2509                next if $seen{$p};
2510                $seen{$p} = 1;
2511                push @ret, $p;
2512        }
2513        @ret;
2514}
2515
2516sub rewrite_root {
2517        my ($self) = @_;
2518        return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2519        my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2520        my $rwr = eval { command_oneline(qw/config --get/, $k) };
2521        if ($rwr) {
2522                $rwr =~ s#/+$##;
2523                if ($rwr !~ m#^[a-z\+]+://#) {
2524                        die "$rwr is not a valid URL (key: $k)\n";
2525                }
2526        }
2527        $self->{-rewrite_root} = $rwr;
2528}
2529
2530sub rewrite_uuid {
2531        my ($self) = @_;
2532        return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2533        my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2534        my $rwid = eval { command_oneline(qw/config --get/, $k) };
2535        if ($rwid) {
2536                $rwid =~ s#/+$##;
2537                if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2538                        die "$rwid is not a valid UUID (key: $k)\n";
2539                }
2540        }
2541        $self->{-rewrite_uuid} = $rwid;
2542}
2543
2544sub metadata_url {
2545        my ($self) = @_;
2546        ($self->rewrite_root || $self->{url}) .
2547           (length $self->{path} ? '/' . $self->{path} : '');
2548}
2549
2550sub full_url {
2551        my ($self) = @_;
2552        $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2553}
2554
2555sub full_pushurl {
2556        my ($self) = @_;
2557        if ($self->{pushurl}) {
2558                return $self->{pushurl} . (length $self->{path} ? '/' .
2559                       $self->{path} : '');
2560        } else {
2561                return $self->full_url;
2562        }
2563}
2564
2565sub set_commit_header_env {
2566        my ($log_entry) = @_;
2567        my %env;
2568        foreach my $ned (qw/NAME EMAIL DATE/) {
2569                foreach my $ac (qw/AUTHOR COMMITTER/) {
2570                        $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2571                }
2572        }
2573
2574        $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2575        $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2576        $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2577
2578        $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2579                                                ? $log_entry->{commit_name}
2580                                                : $log_entry->{name};
2581        $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2582                                                ? $log_entry->{commit_email}
2583                                                : $log_entry->{email};
2584        \%env;
2585}
2586
2587sub restore_commit_header_env {
2588        my ($env) = @_;
2589        foreach my $ned (qw/NAME EMAIL DATE/) {
2590                foreach my $ac (qw/AUTHOR COMMITTER/) {
2591                        my $k = "GIT_${ac}_${ned}";
2592                        if (defined $env->{$k}) {
2593                                $ENV{$k} = $env->{$k};
2594                        } else {
2595                                delete $ENV{$k};
2596                        }
2597                }
2598        }
2599}
2600
2601sub gc {
2602        command_noisy('gc', '--auto');
2603};
2604
2605sub do_git_commit {
2606        my ($self, $log_entry) = @_;
2607        my $lr = $self->last_rev;
2608        if (defined $lr && $lr >= $log_entry->{revision}) {
2609                die "Last fetched revision of ", $self->refname,
2610                    " was r$lr, but we are about to fetch: ",
2611                    "r$log_entry->{revision}!\n";
2612        }
2613        if (my $c = $self->rev_map_get($log_entry->{revision})) {
2614                croak "$log_entry->{revision} = $c already exists! ",
2615                      "Why are we refetching it?\n";
2616        }
2617        my $old_env = set_commit_header_env($log_entry);
2618        my $tree = $log_entry->{tree};
2619        if (!defined $tree) {
2620                $tree = $self->tmp_index_do(sub {
2621                                            command_oneline('write-tree') });
2622        }
2623        die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2624
2625        my @exec = ('git', 'commit-tree', $tree);
2626        foreach ($self->get_commit_parents($log_entry)) {
2627                push @exec, '-p', $_;
2628        }
2629        defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2630                                                                   or croak $!;
2631        binmode $msg_fh;
2632
2633        # we always get UTF-8 from SVN, but we may want our commits in
2634        # a different encoding.
2635        if (my $enc = Git::config('i18n.commitencoding')) {
2636                require Encode;
2637                Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2638        }
2639        print $msg_fh $log_entry->{log} or croak $!;
2640        restore_commit_header_env($old_env);
2641        unless ($self->no_metadata) {
2642                print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2643                              or croak $!;
2644        }
2645        $msg_fh->flush == 0 or croak $!;
2646        close $msg_fh or croak $!;
2647        chomp(my $commit = do { local $/; <$out_fh> });
2648        close $out_fh or croak $!;
2649        waitpid $pid, 0;
2650        croak $? if $?;
2651        if ($commit !~ /^$::sha1$/o) {
2652                die "Failed to commit, invalid sha1: $commit\n";
2653        }
2654
2655        $self->rev_map_set($log_entry->{revision}, $commit, 1);
2656
2657        $self->{last_rev} = $log_entry->{revision};
2658        $self->{last_commit} = $commit;
2659        print "r$log_entry->{revision}" unless $::_q > 1;
2660        if (defined $log_entry->{svm_revision}) {
2661                 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2662                 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2663                                   0, $self->svm_uuid);
2664        }
2665        print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2666        if (--$_gc_nr == 0) {
2667                $_gc_nr = $_gc_period;
2668                gc();
2669        }
2670        return $commit;
2671}
2672
2673sub match_paths {
2674        my ($self, $paths, $r) = @_;
2675        return 1 if $self->{path} eq '';
2676        if (my $path = $paths->{"/$self->{path}"}) {
2677                return ($path->{action} eq 'D') ? 0 : 1;
2678        }
2679        $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2680        if (grep /$self->{path_regex}/, keys %$paths) {
2681                return 1;
2682        }
2683        my $c = '';
2684        foreach (split m#/#, $self->{path}) {
2685                $c .= "/$_";
2686                next unless ($paths->{$c} &&
2687                             ($paths->{$c}->{action} =~ /^[AR]$/));
2688                if ($self->ra->check_path($self->{path}, $r) ==
2689                    $SVN::Node::dir) {
2690                        return 1;
2691                }
2692        }
2693        return 0;
2694}
2695
2696sub find_parent_branch {
2697        my ($self, $paths, $rev) = @_;
2698        return undef unless $self->follow_parent;
2699        unless (defined $paths) {
2700                my $err_handler = $SVN::Error::handler;
2701                $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2702                $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2703                                   sub { $paths = $_[0] });
2704                $SVN::Error::handler = $err_handler;
2705        }
2706        return undef unless defined $paths;
2707
2708        # look for a parent from another branch:
2709        my @b_path_components = split m#/#, $self->{path};
2710        my @a_path_components;
2711        my $i;
2712        while (@b_path_components) {
2713                $i = $paths->{'/'.join('/', @b_path_components)};
2714                last if $i && defined $i->{copyfrom_path};
2715                unshift(@a_path_components, pop(@b_path_components));
2716        }
2717        return undef unless defined $i && defined $i->{copyfrom_path};
2718        my $branch_from = $i->{copyfrom_path};
2719        if (@a_path_components) {
2720                print STDERR "branch_from: $branch_from => ";
2721                $branch_from .= '/'.join('/', @a_path_components);
2722                print STDERR $branch_from, "\n";
2723        }
2724        my $r = $i->{copyfrom_rev};
2725        my $repos_root = $self->ra->{repos_root};
2726        my $url = $self->ra->{url};
2727        my $new_url = $url . $branch_from;
2728        print STDERR  "Found possible branch point: ",
2729                      "$new_url => ", $self->full_url, ", $r\n"
2730                      unless $::_q > 1;
2731        $branch_from =~ s#^/##;
2732        my $gs = $self->other_gs($new_url, $url,
2733                                 $branch_from, $r, $self->{ref_id});
2734        my ($r0, $parent) = $gs->find_rev_before($r, 1);
2735        {
2736                my ($base, $head);
2737                if (!defined $r0 || !defined $parent) {
2738                        ($base, $head) = parse_revision_argument(0, $r);
2739                } else {
2740                        if ($r0 < $r) {
2741                                $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2742                                        0, 1, sub { $base = $_[1] - 1 });
2743                        }
2744                }
2745                if (defined $base && $base <= $r) {
2746                        $gs->fetch($base, $r);
2747                }
2748                ($r0, $parent) = $gs->find_rev_before($r, 1);
2749        }
2750        if (defined $r0 && defined $parent) {
2751                print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2752                             unless $::_q > 1;
2753                my $ed;
2754                if ($self->ra->can_do_switch) {
2755                        $self->assert_index_clean($parent);
2756                        print STDERR "Following parent with do_switch\n"
2757                                     unless $::_q > 1;
2758                        # do_switch works with svn/trunk >= r22312, but that
2759                        # is not included with SVN 1.4.3 (the latest version
2760                        # at the moment), so we can't rely on it
2761                        $self->{last_rev} = $r0;
2762                        $self->{last_commit} = $parent;
2763                        $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2764                        $gs->ra->gs_do_switch($r0, $rev, $gs,
2765                                              $self->full_url, $ed)
2766                          or die "SVN connection failed somewhere...\n";
2767                } elsif ($self->ra->trees_match($new_url, $r0,
2768                                                $self->full_url, $rev)) {
2769                        print STDERR "Trees match:\n",
2770                                     "  $new_url\@$r0\n",
2771                                     "  ${\$self->full_url}\@$rev\n",
2772                                     "Following parent with no changes\n"
2773                                     unless $::_q > 1;
2774                        $self->tmp_index_do(sub {
2775                            command_noisy('read-tree', $parent);
2776                        });
2777                        $self->{last_commit} = $parent;
2778                } else {
2779                        print STDERR "Following parent with do_update\n"
2780                                     unless $::_q > 1;
2781                        $ed = SVN::Git::Fetcher->new($self);
2782                        $self->ra->gs_do_update($rev, $rev, $self, $ed)
2783                          or die "SVN connection failed somewhere...\n";
2784                }
2785                print STDERR "Successfully followed parent\n" unless $::_q > 1;
2786                return $self->make_log_entry($rev, [$parent], $ed);
2787        }
2788        return undef;
2789}
2790
2791sub do_fetch {
2792        my ($self, $paths, $rev) = @_;
2793        my $ed;
2794        my ($last_rev, @parents);
2795        if (my $lc = $self->last_commit) {
2796                # we can have a branch that was deleted, then re-added
2797                # under the same name but copied from another path, in
2798                # which case we'll have multiple parents (we don't
2799                # want to break the original ref, nor lose copypath info):
2800                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2801                        push @{$log_entry->{parents}}, $lc;
2802                        return $log_entry;
2803                }
2804                $ed = SVN::Git::Fetcher->new($self);
2805                $last_rev = $self->{last_rev};
2806                $ed->{c} = $lc;
2807                @parents = ($lc);
2808        } else {
2809                $last_rev = $rev;
2810                if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2811                        return $log_entry;
2812                }
2813                $ed = SVN::Git::Fetcher->new($self);
2814        }
2815        unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2816                die "SVN connection failed somewhere...\n";
2817        }
2818        $self->make_log_entry($rev, \@parents, $ed);
2819}
2820
2821sub mkemptydirs {
2822        my ($self, $r) = @_;
2823
2824        sub scan {
2825                my ($r, $empty_dirs, $line) = @_;
2826                if (defined $r && $line =~ /^r(\d+)$/) {
2827                        return 0 if $1 > $r;
2828                } elsif ($line =~ /^  \+empty_dir: (.+)$/) {
2829                        $empty_dirs->{$1} = 1;
2830                } elsif ($line =~ /^  \-empty_dir: (.+)$/) {
2831                        my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
2832                        delete @$empty_dirs{@d};
2833                }
2834                1; # continue
2835        };
2836
2837        my %empty_dirs = ();
2838        my $gz_file = "$self->{dir}/unhandled.log.gz";
2839        if (-f $gz_file) {
2840                if (!$can_compress) {
2841                        warn "Compress::Zlib could not be found; ",
2842                             "empty directories in $gz_file will not be read\n";
2843                } else {
2844                        my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
2845                                die "Unable to open $gz_file: $!\n";
2846                        my $line;
2847                        while ($gz->gzreadline($line) > 0) {
2848                                scan($r, \%empty_dirs, $line) or last;
2849                        }
2850                        $gz->gzclose;
2851                }
2852        }
2853
2854        if (open my $fh, '<', "$self->{dir}/unhandled.log") {
2855                binmode $fh or croak "binmode: $!";
2856                while (<$fh>) {
2857                        scan($r, \%empty_dirs, $_) or last;
2858                }
2859                close $fh;
2860        }
2861
2862        my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2863        foreach my $d (sort keys %empty_dirs) {
2864                $d = uri_decode($d);
2865                $d =~ s/$strip//;
2866                next unless length($d);
2867                next if -d $d;
2868                if (-e $d) {
2869                        warn "$d exists but is not a directory\n";
2870                } else {
2871                        print "creating empty directory: $d\n";
2872                        mkpath([$d]);
2873                }
2874        }
2875}
2876
2877sub get_untracked {
2878        my ($self, $ed) = @_;
2879        my @out;
2880        my $h = $ed->{empty};
2881        foreach (sort keys %$h) {
2882                my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2883                push @out, "  $act: " . uri_encode($_);
2884                warn "W: $act: $_\n";
2885        }
2886        foreach my $t (qw/dir_prop file_prop/) {
2887                $h = $ed->{$t} or next;
2888                foreach my $path (sort keys %$h) {
2889                        my $ppath = $path eq '' ? '.' : $path;
2890                        foreach my $prop (sort keys %{$h->{$path}}) {
2891                                next if $SKIP_PROP{$prop};
2892                                my $v = $h->{$path}->{$prop};
2893                                my $t_ppath_prop = "$t: " .
2894                                                    uri_encode($ppath) . ' ' .
2895                                                    uri_encode($prop);
2896                                if (defined $v) {
2897                                        push @out, "  +$t_ppath_prop " .
2898                                                   uri_encode($v);
2899                                } else {
2900                                        push @out, "  -$t_ppath_prop";
2901                                }
2902                        }
2903                }
2904        }
2905        foreach my $t (qw/absent_file absent_directory/) {
2906                $h = $ed->{$t} or next;
2907                foreach my $parent (sort keys %$h) {
2908                        foreach my $path (sort @{$h->{$parent}}) {
2909                                push @out, "  $t: " .
2910                                           uri_encode("$parent/$path");
2911                                warn "W: $t: $parent/$path ",
2912                                     "Insufficient permissions?\n";
2913                        }
2914                }
2915        }
2916        \@out;
2917}
2918
2919# parse_svn_date(DATE)
2920# --------------------
2921# Given a date (in UTC) from Subversion, return a string in the format
2922# "<TZ Offset> <local date/time>" that Git will use.
2923#
2924# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2925# is true we'll convert it to the local timezone instead.
2926sub parse_svn_date {
2927        my $date = shift || return '+0000 1970-01-01 00:00:00';
2928        my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2929                                            (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2930                                         croak "Unable to parse date: $date\n";
2931        my $parsed_date;    # Set next.
2932
2933        if ($Git::SVN::_localtime) {
2934                # Translate the Subversion datetime to an epoch time.
2935                # Begin by switching ourselves to $date's timezone, UTC.
2936                my $old_env_TZ = $ENV{TZ};
2937                $ENV{TZ} = 'UTC';
2938
2939                my $epoch_in_UTC =
2940                    POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2941
2942                # Determine our local timezone (including DST) at the
2943                # time of $epoch_in_UTC.  $Git::SVN::Log::TZ stored the
2944                # value of TZ, if any, at the time we were run.
2945                if (defined $Git::SVN::Log::TZ) {
2946                        $ENV{TZ} = $Git::SVN::Log::TZ;
2947                } else {
2948                        delete $ENV{TZ};
2949                }
2950
2951                my $our_TZ =
2952                    POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2953
2954                # This converts $epoch_in_UTC into our local timezone.
2955                my ($sec, $min, $hour, $mday, $mon, $year,
2956                    $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2957
2958                $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2959                                       $our_TZ, $year + 1900, $mon + 1,
2960                                       $mday, $hour, $min, $sec);
2961
2962                # Reset us to the timezone in effect when we entered
2963                # this routine.
2964                if (defined $old_env_TZ) {
2965                        $ENV{TZ} = $old_env_TZ;
2966                } else {
2967                        delete $ENV{TZ};
2968                }
2969        } else {
2970                $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2971        }
2972
2973        return $parsed_date;
2974}
2975
2976sub other_gs {
2977        my ($self, $new_url, $url,
2978            $branch_from, $r, $old_ref_id) = @_;
2979        my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2980        unless ($gs) {
2981                my $ref_id = $old_ref_id;
2982                $ref_id =~ s/\@\d+-*$//;
2983                $ref_id .= "\@$r";
2984                # just grow a tail if we're not unique enough :x
2985                $ref_id .= '-' while find_ref($ref_id);
2986                my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2987                if ($u =~ s#^\Q$url\E(/|$)##) {
2988                        $p = $u;
2989                        $u = $url;
2990                        $repo_id = $self->{repo_id};
2991                }
2992                while (1) {
2993                        # It is possible to tag two different subdirectories at
2994                        # the same revision.  If the url for an existing ref
2995                        # does not match, we must either find a ref with a
2996                        # matching url or create a new ref by growing a tail.
2997                        $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2998                        my (undef, $max_commit) = $gs->rev_map_max(1);
2999                        last if (!$max_commit);
3000                        my ($url) = ::cmt_metadata($max_commit);
3001                        last if ($url eq $gs->full_url);
3002                        $ref_id .= '-';
3003                }
3004                print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
3005        }
3006        $gs
3007}
3008
3009sub call_authors_prog {
3010        my ($orig_author) = @_;
3011        $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
3012        my $author = `$::_authors_prog $orig_author`;
3013        if ($? != 0) {
3014                die "$::_authors_prog failed with exit code $?\n"
3015        }
3016        if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
3017                my ($name, $email) = ($1, $2);
3018                $email = undef if length $2 == 0;
3019                return [$name, $email];
3020        } else {
3021                die "Author: $orig_author: $::_authors_prog returned "
3022                        . "invalid author format: $author\n";
3023        }
3024}
3025
3026sub check_author {
3027        my ($author) = @_;
3028        if (!defined $author || length $author == 0) {
3029                $author = '(no author)';
3030        }
3031        if (!defined $::users{$author}) {
3032                if (defined $::_authors_prog) {
3033                        $::users{$author} = call_authors_prog($author);
3034                } elsif (defined $::_authors) {
3035                        die "Author: $author not defined in $::_authors file\n";
3036                }
3037        }
3038        $author;
3039}
3040
3041sub find_extra_svk_parents {
3042        my ($self, $ed, $tickets, $parents) = @_;
3043        # aha!  svk:merge property changed...
3044        my @tickets = split "\n", $tickets;
3045        my @known_parents;
3046        for my $ticket ( @tickets ) {
3047                my ($uuid, $path, $rev) = split /:/, $ticket;
3048                if ( $uuid eq $self->ra_uuid ) {
3049                        my $url = $self->{url};
3050                        my $repos_root = $url;
3051                        my $branch_from = $path;
3052                        $branch_from =~ s{^/}{};
3053                        my $gs = $self->other_gs($repos_root."/".$branch_from,
3054                                                 $url,
3055                                                 $branch_from,
3056                                                 $rev,
3057                                                 $self->{ref_id});
3058                        if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3059                                # wahey!  we found it, but it might be
3060                                # an old one (!)
3061                                push @known_parents, [ $rev, $commit ];
3062                        }
3063                }
3064        }
3065        # Ordering matters; highest-numbered commit merge tickets
3066        # first, as they may account for later merge ticket additions
3067        # or changes.
3068        @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3069        for my $parent ( @known_parents ) {
3070                my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3071                my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3072                my $new;
3073                while ( <$msg_fh> ) {
3074                        $new=1;last;
3075                }
3076                command_close_pipe($msg_fh, $ctx);
3077                if ( $new ) {
3078                        print STDERR
3079                            "Found merge parent (svk:merge ticket): $parent\n";
3080                        push @$parents, $parent;
3081                }
3082        }
3083}
3084
3085sub lookup_svn_merge {
3086        my $uuid = shift;
3087        my $url = shift;
3088        my $merge = shift;
3089
3090        my ($source, $revs) = split ":", $merge;
3091        my $path = $source;
3092        $path =~ s{^/}{};
3093        my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3094        if ( !$gs ) {
3095                warn "Couldn't find revmap for $url$source\n";
3096                return;
3097        }
3098        my @ranges = split ",", $revs;
3099        my ($tip, $tip_commit);
3100        my @merged_commit_ranges;
3101        # find the tip
3102        for my $range ( @ranges ) {
3103                my ($bottom, $top) = split "-", $range;
3104                $top ||= $bottom;
3105                my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3106                my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3107
3108                unless ($top_commit and $bottom_commit) {
3109                        warn "W:unknown path/rev in svn:mergeinfo "
3110                                ."dirprop: $source:$range\n";
3111                        next;
3112                }
3113
3114                push @merged_commit_ranges,
3115                        "$bottom_commit^..$top_commit";
3116
3117                if ( !defined $tip or $top > $tip ) {
3118                        $tip = $top;
3119                        $tip_commit = $top_commit;
3120                }
3121        }
3122        return ($tip_commit, @merged_commit_ranges);
3123}
3124
3125sub _rev_list {
3126        my ($msg_fh, $ctx) = command_output_pipe(
3127                "rev-list", @_,
3128               );
3129        my @rv;
3130        while ( <$msg_fh> ) {
3131                chomp;
3132                push @rv, $_;
3133        }
3134        command_close_pipe($msg_fh, $ctx);
3135        @rv;
3136}
3137
3138sub check_cherry_pick {
3139        my $base = shift;
3140        my $tip = shift;
3141        my $parents = shift;
3142        my @ranges = @_;
3143        my %commits = map { $_ => 1 }
3144                _rev_list("--no-merges", $tip, "--not", $base, @$parents);
3145        for my $range ( @ranges ) {
3146                delete @commits{_rev_list($range)};
3147        }
3148        for my $commit (keys %commits) {
3149                if (has_no_changes($commit)) {
3150                        delete $commits{$commit};
3151                }
3152        }
3153        return (keys %commits);
3154}
3155
3156sub has_no_changes {
3157        my $commit = shift;
3158
3159        my @revs = split / /, command_oneline(
3160                qw(rev-list --parents -1 -m), $commit);
3161
3162        # Commits with no parents, e.g. the start of a partial branch,
3163        # have changes by definition.
3164        return 1 if (@revs < 2);
3165
3166        # Commits with multiple parents, e.g a merge, have no changes
3167        # by definition.
3168        return 0 if (@revs > 2);
3169
3170        return (command_oneline("rev-parse", "$commit^{tree}") eq
3171                command_oneline("rev-parse", "$commit~1^{tree}"));
3172}
3173
3174# The GIT_DIR environment variable is not always set until after the command
3175# line arguments are processed, so we can't memoize in a BEGIN block.
3176{
3177        my $memoized = 0;
3178
3179        sub memoize_svn_mergeinfo_functions {
3180                return if $memoized;
3181                $memoized = 1;
3182
3183                my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3184                mkpath([$cache_path]) unless -d $cache_path;
3185
3186                tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3187                    "$cache_path/lookup_svn_merge.db", 'nstore';
3188                memoize 'lookup_svn_merge',
3189                        SCALAR_CACHE => 'FAULT',
3190                        LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3191                ;
3192
3193                tie my %check_cherry_pick_cache => 'Memoize::Storable',
3194                    "$cache_path/check_cherry_pick.db", 'nstore';
3195                memoize 'check_cherry_pick',
3196                        SCALAR_CACHE => 'FAULT',
3197                        LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3198                ;
3199
3200                tie my %has_no_changes_cache => 'Memoize::Storable',
3201                    "$cache_path/has_no_changes.db", 'nstore';
3202                memoize 'has_no_changes',
3203                        SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3204                        LIST_CACHE => 'FAULT',
3205                ;
3206        }
3207
3208        sub unmemoize_svn_mergeinfo_functions {
3209                return if not $memoized;
3210                $memoized = 0;
3211
3212                Memoize::unmemoize 'lookup_svn_merge';
3213                Memoize::unmemoize 'check_cherry_pick';
3214                Memoize::unmemoize 'has_no_changes';
3215        }
3216
3217        Memoize::memoize 'Git::SVN::repos_root';
3218}
3219
3220END {
3221        # Force cache writeout explicitly instead of waiting for
3222        # global destruction to avoid segfault in Storable:
3223        # http://rt.cpan.org/Public/Bug/Display.html?id=36087
3224        unmemoize_svn_mergeinfo_functions();
3225}
3226
3227sub parents_exclude {
3228        my $parents = shift;
3229        my @commits = @_;
3230        return unless @commits;
3231
3232        my @excluded;
3233        my $excluded;
3234        do {
3235                my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3236                $excluded = command_oneline(@cmd);
3237                if ( $excluded ) {
3238                        my @new;
3239                        my $found;
3240                        for my $commit ( @commits ) {
3241                                if ( $commit eq $excluded ) {
3242                                        push @excluded, $commit;
3243                                        $found++;
3244                                        last;
3245                                }
3246                                else {
3247                                        push @new, $commit;
3248                                }
3249                        }
3250                        die "saw commit '$excluded' in rev-list output, "
3251                                ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3252                                        unless $found;
3253                        @commits = @new;
3254                }
3255        }
3256                while ($excluded and @commits);
3257
3258        return @excluded;
3259}
3260
3261
3262# note: this function should only be called if the various dirprops
3263# have actually changed
3264sub find_extra_svn_parents {
3265        my ($self, $ed, $mergeinfo, $parents) = @_;
3266        # aha!  svk:merge property changed...
3267
3268        memoize_svn_mergeinfo_functions();
3269
3270        # We first search for merged tips which are not in our
3271        # history.  Then, we figure out which git revisions are in
3272        # that tip, but not this revision.  If all of those revisions
3273        # are now marked as merge, we can add the tip as a parent.
3274        my @merges = split "\n", $mergeinfo;
3275        my @merge_tips;
3276        my $url = $self->{url};
3277        my $uuid = $self->ra_uuid;
3278        my %ranges;
3279        for my $merge ( @merges ) {
3280                my ($tip_commit, @ranges) =
3281                        lookup_svn_merge( $uuid, $url, $merge );
3282                unless (!$tip_commit or
3283                                grep { $_ eq $tip_commit } @$parents ) {
3284                        push @merge_tips, $tip_commit;
3285                        $ranges{$tip_commit} = \@ranges;
3286                } else {
3287                        push @merge_tips, undef;
3288                }
3289        }
3290
3291        my %excluded = map { $_ => 1 }
3292                parents_exclude($parents, grep { defined } @merge_tips);
3293
3294        # check merge tips for new parents
3295        my @new_parents;
3296        for my $merge_tip ( @merge_tips ) {
3297                my $spec = shift @merges;
3298                next unless $merge_tip and $excluded{$merge_tip};
3299
3300                my $ranges = $ranges{$merge_tip};
3301
3302                # check out 'new' tips
3303                my $merge_base;
3304                eval {
3305                        $merge_base = command_oneline(
3306                                "merge-base",
3307                                @$parents, $merge_tip,
3308                        );
3309                };
3310                if ($@) {
3311                        die "An error occurred during merge-base"
3312                                unless $@->isa("Git::Error::Command");
3313
3314                        warn "W: Cannot find common ancestor between ".
3315                             "@$parents and $merge_tip. Ignoring merge info.\n";
3316                        next;
3317                }
3318
3319                # double check that there are no missing non-merge commits
3320                my (@incomplete) = check_cherry_pick(
3321                        $merge_base, $merge_tip,
3322                        $parents,
3323                        @$ranges,
3324                       );
3325
3326                if ( @incomplete ) {
3327                        warn "W:svn cherry-pick ignored ($spec) - missing "
3328                                .@incomplete." commit(s) (eg $incomplete[0])\n";
3329                } else {
3330                        warn
3331                                "Found merge parent (svn:mergeinfo prop): ",
3332                                        $merge_tip, "\n";
3333                        push @new_parents, $merge_tip;
3334                }
3335        }
3336
3337        # cater for merges which merge commits from multiple branches
3338        if ( @new_parents > 1 ) {
3339                for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3340                        for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3341                                next if $i == $j;
3342                                next unless $new_parents[$i];
3343                                next unless $new_parents[$j];
3344                                my $revs = command_oneline(
3345                                        "rev-list", "-1",
3346                                        "$new_parents[$i]..$new_parents[$j]",
3347                                       );
3348                                if ( !$revs ) {
3349                                        undef($new_parents[$j]);
3350                                }
3351                        }
3352                }
3353        }
3354        push @$parents, grep { defined } @new_parents;
3355}
3356
3357sub make_log_entry {
3358        my ($self, $rev, $parents, $ed) = @_;
3359        my $untracked = $self->get_untracked($ed);
3360
3361        my @parents = @$parents;
3362        my $ps = $ed->{path_strip} || "";
3363        for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3364                my $props = $ed->{dir_prop}{$path};
3365                if ( $props->{"svk:merge"} ) {
3366                        $self->find_extra_svk_parents
3367                                ($ed, $props->{"svk:merge"}, \@parents);
3368                }
3369                if ( $props->{"svn:mergeinfo"} ) {
3370                        $self->find_extra_svn_parents
3371                                ($ed,
3372                                 $props->{"svn:mergeinfo"},
3373                                 \@parents);
3374                }
3375        }
3376
3377        open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3378        print $un "r$rev\n" or croak $!;
3379        print $un $_, "\n" foreach @$untracked;
3380        my %log_entry = ( parents => \@parents, revision => $rev,
3381                          log => '');
3382
3383        my $headrev;
3384        my $logged = delete $self->{logged_rev_props};
3385        if (!$logged || $self->{-want_revprops}) {
3386                my $rp = $self->ra->rev_proplist($rev);
3387                foreach (sort keys %$rp) {
3388                        my $v = $rp->{$_};
3389                        if (/^svn:(author|date|log)$/) {
3390                                $log_entry{$1} = $v;
3391                        } elsif ($_ eq 'svm:headrev') {
3392                                $headrev = $v;
3393                        } else {
3394                                print $un "  rev_prop: ", uri_encode($_), ' ',
3395                                          uri_encode($v), "\n";
3396                        }
3397                }
3398        } else {
3399                map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3400        }
3401        close $un or croak $!;
3402
3403        $log_entry{date} = parse_svn_date($log_entry{date});
3404        $log_entry{log} .= "\n";
3405        my $author = $log_entry{author} = check_author($log_entry{author});
3406        my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3407                                                       : ($author, undef);
3408
3409        my ($commit_name, $commit_email) = ($name, $email);
3410        if ($_use_log_author) {
3411                my $name_field;
3412                if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3413                        $name_field = $1;
3414                } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3415                        $name_field = $1;
3416                }
3417                if (!defined $name_field) {
3418                        if (!defined $email) {
3419                                $email = $name;
3420                        }
3421                } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3422                        ($name, $email) = ($1, $2);
3423                } elsif ($name_field =~ /(.*)@/) {
3424                        ($name, $email) = ($1, $name_field);
3425                } else {
3426                        ($name, $email) = ($name_field, $name_field);
3427                }
3428        }
3429        if (defined $headrev && $self->use_svm_props) {
3430                if ($self->rewrite_root) {
3431                        die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3432                            "options set!\n";
3433                }
3434                if ($self->rewrite_uuid) {
3435                        die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3436                            "options set!\n";
3437                }
3438                my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3439                # we don't want "SVM: initializing mirror for junk" ...
3440                return undef if $r == 0;
3441                my $svm = $self->svm;
3442                if ($uuid ne $svm->{uuid}) {
3443                        die "UUID mismatch on SVM path:\n",
3444                            "expected: $svm->{uuid}\n",
3445                            "     got: $uuid\n";
3446                }
3447                my $full_url = $self->full_url;
3448                $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3449                             die "Failed to replace '$svm->{replace}' with ",
3450                                 "'$svm->{source}' in $full_url\n";
3451                # throw away username for storing in records
3452                remove_username($full_url);
3453                $log_entry{metadata} = "$full_url\@$r $uuid";
3454                $log_entry{svm_revision} = $r;
3455                $email ||= "$author\@$uuid";
3456                $commit_email ||= "$author\@$uuid";
3457        } elsif ($self->use_svnsync_props) {
3458                my $full_url = $self->svnsync->{url};
3459                $full_url .= "/$self->{path}" if length $self->{path};
3460                remove_username($full_url);
3461                my $uuid = $self->svnsync->{uuid};
3462                $log_entry{metadata} = "$full_url\@$rev $uuid";
3463                $email ||= "$author\@$uuid";
3464                $commit_email ||= "$author\@$uuid";
3465        } else {
3466                my $url = $self->metadata_url;
3467                remove_username($url);
3468                my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3469                $log_entry{metadata} = "$url\@$rev " . $uuid;
3470                $email ||= "$author\@" . $uuid;
3471                $commit_email ||= "$author\@" . $uuid;
3472        }
3473        $log_entry{name} = $name;
3474        $log_entry{email} = $email;
3475        $log_entry{commit_name} = $commit_name;
3476        $log_entry{commit_email} = $commit_email;
3477        \%log_entry;
3478}
3479
3480sub fetch {
3481        my ($self, $min_rev, $max_rev, @parents) = @_;
3482        my ($last_rev, $last_commit) = $self->last_rev_commit;
3483        my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3484        $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3485}
3486
3487sub set_tree_cb {
3488        my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3489        $self->{inject_parents} = { $rev => $tree };
3490        $self->fetch(undef, undef);
3491}
3492
3493sub set_tree {
3494        my ($self, $tree) = (shift, shift);
3495        my $log_entry = ::get_commit_entry($tree);
3496        unless ($self->{last_rev}) {
3497                ::fatal("Must have an existing revision to commit");
3498        }
3499        my %ed_opts = ( r => $self->{last_rev},
3500                        log => $log_entry->{log},
3501                        ra => $self->ra,
3502                        tree_a => $self->{last_commit},
3503                        tree_b => $tree,
3504                        editor_cb => sub {
3505                               $self->set_tree_cb($log_entry, $tree, @_) },
3506                        svn_path => $self->{path} );
3507        if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3508                print "No changes\nr$self->{last_rev} = $tree\n";
3509        }
3510}
3511
3512sub rebuild_from_rev_db {
3513        my ($self, $path) = @_;
3514        my $r = -1;
3515        open my $fh, '<', $path or croak "open: $!";
3516        binmode $fh or croak "binmode: $!";
3517        while (<$fh>) {
3518                length($_) == 41 or croak "inconsistent size in ($_) != 41";
3519                chomp($_);
3520                ++$r;
3521                next if $_ eq ('0' x 40);
3522                $self->rev_map_set($r, $_);
3523                print "r$r = $_\n";
3524        }
3525        close $fh or croak "close: $!";
3526        unlink $path or croak "unlink: $!";
3527}
3528
3529sub rebuild {
3530        my ($self) = @_;
3531        my $map_path = $self->map_path;
3532        my $partial = (-e $map_path && ! -z $map_path);
3533        return unless ::verify_ref($self->refname.'^0');
3534        if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3535                my $rev_db = $self->rev_db_path;
3536                $self->rebuild_from_rev_db($rev_db);
3537                if ($self->use_svm_props) {
3538                        my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3539                        $self->rebuild_from_rev_db($svm_rev_db);
3540                }
3541                $self->unlink_rev_db_symlink;
3542                return;
3543        }
3544        print "Rebuilding $map_path ...\n" if (!$partial);
3545        my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3546                (undef, undef));
3547        my ($log, $ctx) =
3548            command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3549                                ($head ? "$head.." : "") . $self->refname,
3550                                '--');
3551        my $metadata_url = $self->metadata_url;
3552        remove_username($metadata_url);
3553        my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3554        my $c;
3555        while (<$log>) {
3556                if ( m{^commit ($::sha1)$} ) {
3557                        $c = $1;
3558                        next;
3559                }
3560                next unless s{^\s*(git-svn-id:)}{$1};
3561                my ($url, $rev, $uuid) = ::extract_metadata($_);
3562                remove_username($url);
3563
3564                # ignore merges (from set-tree)
3565                next if (!defined $rev || !$uuid);
3566
3567                # if we merged or otherwise started elsewhere, this is
3568                # how we break out of it
3569                if (($uuid ne $svn_uuid) ||
3570                    ($metadata_url && $url && ($url ne $metadata_url))) {
3571                        next;
3572                }
3573                if ($partial && $head) {
3574                        print "Partial-rebuilding $map_path ...\n";
3575                        print "Currently at $base_rev = $head\n";
3576                        $head = undef;
3577                }
3578
3579                $self->rev_map_set($rev, $c);
3580                print "r$rev = $c\n";
3581        }
3582        command_close_pipe($log, $ctx);
3583        print "Done rebuilding $map_path\n" if (!$partial || !$head);
3584        my $rev_db_path = $self->rev_db_path;
3585        if (-f $self->rev_db_path) {
3586                unlink $self->rev_db_path or croak "unlink: $!";
3587        }
3588        $self->unlink_rev_db_symlink;
3589}
3590
3591# rev_map:
3592# Tie::File seems to be prone to offset errors if revisions get sparse,
3593# it's not that fast, either.  Tie::File is also not in Perl 5.6.  So
3594# one of my favorite modules is out :<  Next up would be one of the DBM
3595# modules, but I'm not sure which is most portable...
3596#
3597# This is the replacement for the rev_db format, which was too big
3598# and inefficient for large repositories with a lot of sparse history
3599# (mainly tags)
3600#
3601# The format is this:
3602#   - 24 bytes for every record,
3603#     * 4 bytes for the integer representing an SVN revision number
3604#     * 20 bytes representing the sha1 of a git commit
3605#   - No empty padding records like the old format
3606#     (except the last record, which can be overwritten)
3607#   - new records are written append-only since SVN revision numbers
3608#     increase monotonically
3609#   - lookups on SVN revision number are done via a binary search
3610#   - Piping the file to xxd -c24 is a good way of dumping it for
3611#     viewing or editing (piped back through xxd -r), should the need
3612#     ever arise.
3613#   - The last record can be padding revision with an all-zero sha1
3614#     This is used to optimize fetch performance when using multiple
3615#     "fetch" directives in .git/config
3616#
3617# These files are disposable unless noMetadata or useSvmProps is set
3618
3619sub _rev_map_set {
3620        my ($fh, $rev, $commit) = @_;
3621
3622        binmode $fh or croak "binmode: $!";
3623        my $size = (stat($fh))[7];
3624        ($size % 24) == 0 or croak "inconsistent size: $size";
3625
3626        my $wr_offset = 0;
3627        if ($size > 0) {
3628                sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3629                my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3630                $read == 24 or croak "read only $read bytes (!= 24)";
3631                my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3632                if ($last_commit eq ('0' x40)) {
3633                        if ($size >= 48) {
3634                                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3635                                $read = sysread($fh, $buf, 24) or
3636                                    croak "read: $!";
3637                                $read == 24 or
3638                                    croak "read only $read bytes (!= 24)";
3639                                ($last_rev, $last_commit) =
3640                                    unpack(rev_map_fmt, $buf);
3641                                if ($last_commit eq ('0' x40)) {
3642                                        croak "inconsistent .rev_map\n";
3643                                }
3644                        }
3645                        if ($last_rev >= $rev) {
3646                                croak "last_rev is higher!: $last_rev >= $rev";
3647                        }
3648                        $wr_offset = -24;
3649                }
3650        }
3651        sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3652        syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3653          croak "write: $!";
3654}
3655
3656sub _rev_map_reset {
3657        my ($fh, $rev, $commit) = @_;
3658        my $c = _rev_map_get($fh, $rev);
3659        $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3660        my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3661        truncate $fh, $offset or croak "truncate: $!";
3662}
3663
3664sub mkfile {
3665        my ($path) = @_;
3666        unless (-e $path) {
3667                my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3668                mkpath([$dir]) unless -d $dir;
3669                open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3670                close $fh or die "Couldn't close (create) $path: $!\n";
3671        }
3672}
3673
3674sub rev_map_set {
3675        my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3676        defined $commit or die "missing arg3\n";
3677        length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3678        my $db = $self->map_path($uuid);
3679        my $db_lock = "$db.lock";
3680        my $sig;
3681        $update_ref ||= 0;
3682        if ($update_ref) {
3683                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3684                            $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3685        }
3686        mkfile($db);
3687
3688        $LOCKFILES{$db_lock} = 1;
3689        my $sync;
3690        # both of these options make our .rev_db file very, very important
3691        # and we can't afford to lose it because rebuild() won't work
3692        if ($self->use_svm_props || $self->no_metadata) {
3693                $sync = 1;
3694                copy($db, $db_lock) or die "rev_map_set(@_): ",
3695                                           "Failed to copy: ",
3696                                           "$db => $db_lock ($!)\n";
3697        } else {
3698                rename $db, $db_lock or die "rev_map_set(@_): ",
3699                                            "Failed to rename: ",
3700                                            "$db => $db_lock ($!)\n";
3701        }
3702
3703        sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3704             or croak "Couldn't open $db_lock: $!\n";
3705        $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3706                                 _rev_map_set($fh, $rev, $commit);
3707        if ($sync) {
3708                $fh->flush or die "Couldn't flush $db_lock: $!\n";
3709                $fh->sync or die "Couldn't sync $db_lock: $!\n";
3710        }
3711        close $fh or croak $!;
3712        if ($update_ref) {
3713                $_head = $self;
3714                my $note = "";
3715                $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3716                command_noisy('update-ref', '-m', "r$rev$note",
3717                              $self->refname, $commit);
3718        }
3719        rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3720                                    "$db_lock => $db ($!)\n";
3721        delete $LOCKFILES{$db_lock};
3722        if ($update_ref) {
3723                $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3724                            $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3725                kill $sig, $$ if defined $sig;
3726        }
3727}
3728
3729# If want_commit, this will return an array of (rev, commit) where
3730# commit _must_ be a valid commit in the archive.
3731# Otherwise, it'll return the max revision (whether or not the
3732# commit is valid or just a 0x40 placeholder).
3733sub rev_map_max {
3734        my ($self, $want_commit) = @_;
3735        $self->rebuild;
3736        my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3737        $want_commit ? ($r, $c) : $r;
3738}
3739
3740sub rev_map_max_norebuild {
3741        my ($self, $want_commit) = @_;
3742        my $map_path = $self->map_path;
3743        stat $map_path or return $want_commit ? (0, undef) : 0;
3744        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3745        binmode $fh or croak "binmode: $!";
3746        my $size = (stat($fh))[7];
3747        ($size % 24) == 0 or croak "inconsistent size: $size";
3748
3749        if ($size == 0) {
3750                close $fh or croak "close: $!";
3751                return $want_commit ? (0, undef) : 0;
3752        }
3753
3754        sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3755        sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3756        my ($r, $c) = unpack(rev_map_fmt, $buf);
3757        if ($want_commit && $c eq ('0' x40)) {
3758                if ($size < 48) {
3759                        return $want_commit ? (0, undef) : 0;
3760                }
3761                sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3762                sysread($fh, $buf, 24) == 24 or croak "read: $!";
3763                ($r, $c) = unpack(rev_map_fmt, $buf);
3764                if ($c eq ('0'x40)) {
3765                        croak "Penultimate record is all-zeroes in $map_path";
3766                }
3767        }
3768        close $fh or croak "close: $!";
3769        $want_commit ? ($r, $c) : $r;
3770}
3771
3772sub rev_map_get {
3773        my ($self, $rev, $uuid) = @_;
3774        my $map_path = $self->map_path($uuid);
3775        return undef unless -e $map_path;
3776
3777        sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3778        my $c = _rev_map_get($fh, $rev);
3779        close($fh) or croak "close: $!";
3780        $c
3781}
3782
3783sub _rev_map_get {
3784        my ($fh, $rev) = @_;
3785
3786        binmode $fh or croak "binmode: $!";
3787        my $size = (stat($fh))[7];
3788        ($size % 24) == 0 or croak "inconsistent size: $size";
3789
3790        if ($size == 0) {
3791                return undef;
3792        }
3793
3794        my ($l, $u) = (0, $size - 24);
3795        my ($r, $c, $buf);
3796
3797        while ($l <= $u) {
3798                my $i = int(($l/24 + $u/24) / 2) * 24;
3799                sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3800                sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3801                my ($r, $c) = unpack(rev_map_fmt, $buf);
3802
3803                if ($r < $rev) {
3804                        $l = $i + 24;
3805                } elsif ($r > $rev) {
3806                        $u = $i - 24;
3807                } else { # $r == $rev
3808                        return $c eq ('0' x 40) ? undef : $c;
3809                }
3810        }
3811        undef;
3812}
3813
3814# Finds the first svn revision that exists on (if $eq_ok is true) or
3815# before $rev for the current branch.  It will not search any lower
3816# than $min_rev.  Returns the git commit hash and svn revision number
3817# if found, else (undef, undef).
3818sub find_rev_before {
3819        my ($self, $rev, $eq_ok, $min_rev) = @_;
3820        --$rev unless $eq_ok;
3821        $min_rev ||= 1;
3822        my $max_rev = $self->rev_map_max;
3823        $rev = $max_rev if ($rev > $max_rev);
3824        while ($rev >= $min_rev) {
3825                if (my $c = $self->rev_map_get($rev)) {
3826                        return ($rev, $c);
3827                }
3828                --$rev;
3829        }
3830        return (undef, undef);
3831}
3832
3833# Finds the first svn revision that exists on (if $eq_ok is true) or
3834# after $rev for the current branch.  It will not search any higher
3835# than $max_rev.  Returns the git commit hash and svn revision number
3836# if found, else (undef, undef).
3837sub find_rev_after {
3838        my ($self, $rev, $eq_ok, $max_rev) = @_;
3839        ++$rev unless $eq_ok;
3840        $max_rev ||= $self->rev_map_max;
3841        while ($rev <= $max_rev) {
3842                if (my $c = $self->rev_map_get($rev)) {
3843                        return ($rev, $c);
3844                }
3845                ++$rev;
3846        }
3847        return (undef, undef);
3848}
3849
3850sub _new {
3851        my ($class, $repo_id, $ref_id, $path) = @_;
3852        unless (defined $repo_id && length $repo_id) {
3853                $repo_id = $Git::SVN::default_repo_id;
3854        }
3855        unless (defined $ref_id && length $ref_id) {
3856                $_prefix = '' unless defined($_prefix);
3857                $_[2] = $ref_id =
3858                             "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3859        }
3860        $_[1] = $repo_id;
3861        my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3862
3863        # Older repos imported by us used $GIT_DIR/svn/foo instead of
3864        # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3865        if ($ref_id =~ m{^refs/remotes/(.*)}) {
3866                my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3867                if (-d $old_dir && ! -d $dir) {
3868                        $dir = $old_dir;
3869                }
3870        }
3871
3872        $_[3] = $path = '' unless (defined $path);
3873        mkpath([$dir]);
3874        bless {
3875                ref_id => $ref_id, dir => $dir, index => "$dir/index",
3876                path => $path, config => "$ENV{GIT_DIR}/svn/config",
3877                map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3878}
3879
3880# for read-only access of old .rev_db formats
3881sub unlink_rev_db_symlink {
3882        my ($self) = @_;
3883        my $link = $self->rev_db_path;
3884        $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3885        if (-l $link) {
3886                unlink $link or croak "unlink: $link failed!";
3887        }
3888}
3889
3890sub rev_db_path {
3891        my ($self, $uuid) = @_;
3892        my $db_path = $self->map_path($uuid);
3893        $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3894            or croak "map_path: $db_path does not contain '/.rev_map.' !";
3895        $db_path;
3896}
3897
3898# the new replacement for .rev_db
3899sub map_path {
3900        my ($self, $uuid) = @_;
3901        $uuid ||= $self->ra_uuid;
3902        "$self->{map_root}.$uuid";
3903}
3904
3905sub uri_encode {
3906        my ($f) = @_;
3907        $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3908        $f
3909}
3910
3911sub uri_decode {
3912        my ($f) = @_;
3913        $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3914        $f
3915}
3916
3917sub remove_username {
3918        $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3919}
3920
3921package Git::SVN::Prompt;
3922use strict;
3923use warnings;
3924require SVN::Core;
3925use vars qw/$_no_auth_cache $_username/;
3926
3927sub simple {
3928        my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3929        $may_save = undef if $_no_auth_cache;
3930        $default_username = $_username if defined $_username;
3931        if (defined $default_username && length $default_username) {
3932                if (defined $realm && length $realm) {
3933                        print STDERR "Authentication realm: $realm\n";
3934                        STDERR->flush;
3935                }
3936                $cred->username($default_username);
3937        } else {
3938                username($cred, $realm, $may_save, $pool);
3939        }
3940        $cred->password(_read_password("Password for '" .
3941                                       $cred->username . "': ", $realm));
3942        $cred->may_save($may_save);
3943        $SVN::_Core::SVN_NO_ERROR;
3944}
3945
3946sub ssl_server_trust {
3947        my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3948        $may_save = undef if $_no_auth_cache;
3949        print STDERR "Error validating server certificate for '$realm':\n";
3950        {
3951                no warnings 'once';
3952                # All variables SVN::Auth::SSL::* are used only once,
3953                # so we're shutting up Perl warnings about this.
3954                if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3955                        print STDERR " - The certificate is not issued ",
3956                            "by a trusted authority. Use the\n",
3957                            "   fingerprint to validate ",
3958                            "the certificate manually!\n";
3959                }
3960                if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3961                        print STDERR " - The certificate hostname ",
3962                            "does not match.\n";
3963                }
3964                if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3965                        print STDERR " - The certificate is not yet valid.\n";
3966                }
3967                if ($failures & $SVN::Auth::SSL::EXPIRED) {
3968                        print STDERR " - The certificate has expired.\n";
3969                }
3970                if ($failures & $SVN::Auth::SSL::OTHER) {
3971                        print STDERR " - The certificate has ",
3972                            "an unknown error.\n";
3973                }
3974        } # no warnings 'once'
3975        printf STDERR
3976                "Certificate information:\n".
3977                " - Hostname: %s\n".
3978                " - Valid: from %s until %s\n".
3979                " - Issuer: %s\n".
3980                " - Fingerprint: %s\n",
3981                map $cert_info->$_, qw(hostname valid_from valid_until
3982                                       issuer_dname fingerprint);
3983        my $choice;
3984prompt:
3985        print STDERR $may_save ?
3986              "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3987              "(R)eject or accept (t)emporarily? ";
3988        STDERR->flush;
3989        $choice = lc(substr(<STDIN> || 'R', 0, 1));
3990        if ($choice =~ /^t$/i) {
3991                $cred->may_save(undef);
3992        } elsif ($choice =~ /^r$/i) {
3993                return -1;
3994        } elsif ($may_save && $choice =~ /^p$/i) {
3995                $cred->may_save($may_save);
3996        } else {
3997                goto prompt;
3998        }
3999        $cred->accepted_failures($failures);
4000        $SVN::_Core::SVN_NO_ERROR;
4001}
4002
4003sub ssl_client_cert {
4004        my ($cred, $realm, $may_save, $pool) = @_;
4005        $may_save = undef if $_no_auth_cache;
4006        print STDERR "Client certificate filename: ";
4007        STDERR->flush;
4008        chomp(my $filename = <STDIN>);
4009        $cred->cert_file($filename);
4010        $cred->may_save($may_save);
4011        $SVN::_Core::SVN_NO_ERROR;
4012}
4013
4014sub ssl_client_cert_pw {
4015        my ($cred, $realm, $may_save, $pool) = @_;
4016        $may_save = undef if $_no_auth_cache;
4017        $cred->password(_read_password("Password: ", $realm));
4018        $cred->may_save($may_save);
4019        $SVN::_Core::SVN_NO_ERROR;
4020}
4021
4022sub username {
4023        my ($cred, $realm, $may_save, $pool) = @_;
4024        $may_save = undef if $_no_auth_cache;
4025        if (defined $realm && length $realm) {
4026                print STDERR "Authentication realm: $realm\n";
4027        }
4028        my $username;
4029        if (defined $_username) {
4030                $username = $_username;
4031        } else {
4032                print STDERR "Username: ";
4033                STDERR->flush;
4034                chomp($username = <STDIN>);
4035        }
4036        $cred->username($username);
4037        $cred->may_save($may_save);
4038        $SVN::_Core::SVN_NO_ERROR;
4039}
4040
4041sub _read_password {
4042        my ($prompt, $realm) = @_;
4043        my $password = '';
4044        if (exists $ENV{GIT_ASKPASS}) {
4045                open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
4046                $password = <PH>;
4047                $password =~ s/[\012\015]//; # \n\r
4048                close(PH);
4049        } else {
4050                print STDERR $prompt;
4051                STDERR->flush;
4052                require Term::ReadKey;
4053                Term::ReadKey::ReadMode('noecho');
4054                while (defined(my $key = Term::ReadKey::ReadKey(0))) {
4055                        last if $key =~ /[\012\015]/; # \n\r
4056                        $password .= $key;
4057                }
4058                Term::ReadKey::ReadMode('restore');
4059                print STDERR "\n";
4060                STDERR->flush;
4061        }
4062        $password;
4063}
4064
4065package SVN::Git::Fetcher;
4066use vars qw/@ISA/;
4067use strict;
4068use warnings;
4069use Carp qw/croak/;
4070use IO::File qw//;
4071use vars qw/$_ignore_regex/;
4072
4073# file baton members: path, mode_a, mode_b, pool, fh, blob, base
4074sub new {
4075        my ($class, $git_svn, $switch_path) = @_;
4076        my $self = SVN::Delta::Editor->new;
4077        bless $self, $class;
4078        if (exists $git_svn->{last_commit}) {
4079                $self->{c} = $git_svn->{last_commit};
4080                $self->{empty_symlinks} =
4081                                  _mark_empty_symlinks($git_svn, $switch_path);
4082        }
4083        $self->{ignore_regex} = eval { command_oneline('config', '--get',
4084                             "svn-remote.$git_svn->{repo_id}.ignore-paths") };
4085        $self->{empty} = {};
4086        $self->{dir_prop} = {};
4087        $self->{file_prop} = {};
4088        $self->{absent_dir} = {};
4089        $self->{absent_file} = {};
4090        $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4091        $self->{pathnameencoding} = Git::config('svn.pathnameencoding');
4092        $self;
4093}
4094
4095# this uses the Ra object, so it must be called before do_{switch,update},
4096# not inside them (when the Git::SVN::Fetcher object is passed) to
4097# do_{switch,update}
4098sub _mark_empty_symlinks {
4099        my ($git_svn, $switch_path) = @_;
4100        my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4101        return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4102
4103        my %ret;
4104        my ($rev, $cmt) = $git_svn->last_rev_commit;
4105        return {} unless ($rev && $cmt);
4106
4107        # allow the warning to be printed for each revision we fetch to
4108        # ensure the user sees it.  The user can also disable the workaround
4109        # on the repository even while git svn is running and the next
4110        # revision fetched will skip this expensive function.
4111        my $printed_warning;
4112        chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4113        my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4114        local $/ = "\0";
4115        my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4116        $pfx .= '/' if length($pfx);
4117        while (<$ls>) {
4118                chomp;
4119                s/\A100644 blob $empty_blob\t//o or next;
4120                unless ($printed_warning) {
4121                        print STDERR "Scanning for empty symlinks, ",
4122                                     "this may take a while if you have ",
4123                                     "many empty files\n",
4124                                     "You may disable this with `",
4125                                     "git config svn.brokenSymlinkWorkaround ",
4126                                     "false'.\n",
4127                                     "This may be done in a different ",
4128                                     "terminal without restarting ",
4129                                     "git svn\n";
4130                        $printed_warning = 1;
4131                }
4132                my $path = $_;
4133                my (undef, $props) =
4134                               $git_svn->ra->get_file($pfx.$path, $rev, undef);
4135                if ($props->{'svn:special'}) {
4136                        $ret{$path} = 1;
4137                }
4138        }
4139        command_close_pipe($ls, $ctx);
4140        \%ret;
4141}
4142
4143# returns true if a given path is inside a ".git" directory
4144sub in_dot_git {
4145        $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4146}
4147
4148# return value: 0 -- don't ignore, 1 -- ignore
4149sub is_path_ignored {
4150        my ($self, $path) = @_;
4151        return 1 if in_dot_git($path);
4152        return 1 if defined($self->{ignore_regex}) &&
4153                    $path =~ m!$self->{ignore_regex}!;
4154        return 0 unless defined($_ignore_regex);
4155        return 1 if $path =~ m!$_ignore_regex!o;
4156        return 0;
4157}
4158
4159sub set_path_strip {
4160        my ($self, $path) = @_;
4161        $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4162}
4163
4164sub open_root {
4165        { path => '' };
4166}
4167
4168sub open_directory {
4169        my ($self, $path, $pb, $rev) = @_;
4170        { path => $path };
4171}
4172
4173sub git_path {
4174        my ($self, $path) = @_;
4175        if (my $enc = $self->{pathnameencoding}) {
4176                require Encode;
4177                Encode::from_to($path, 'UTF-8', $enc);
4178        }
4179        if ($self->{path_strip}) {
4180                $path =~ s!$self->{path_strip}!! or
4181                  die "Failed to strip path '$path' ($self->{path_strip})\n";
4182        }
4183        $path;
4184}
4185
4186sub delete_entry {
4187        my ($self, $path, $rev, $pb) = @_;
4188        return undef if $self->is_path_ignored($path);
4189
4190        my $gpath = $self->git_path($path);
4191        return undef if ($gpath eq '');
4192
4193        # remove entire directories.
4194        my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4195                         =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4196        if ($tree) {
4197                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4198                                                     -r --name-only -z/,
4199                                                     $tree);
4200                local $/ = "\0";
4201                while (<$ls>) {
4202                        chomp;
4203                        my $rmpath = "$gpath/$_";
4204                        $self->{gii}->remove($rmpath);
4205                        print "\tD\t$rmpath\n" unless $::_q;
4206                }
4207                print "\tD\t$gpath/\n" unless $::_q;
4208                command_close_pipe($ls, $ctx);
4209        } else {
4210                $self->{gii}->remove($gpath);
4211                print "\tD\t$gpath\n" unless $::_q;
4212        }
4213        $self->{empty}->{$path} = 0;
4214        undef;
4215}
4216
4217sub open_file {
4218        my ($self, $path, $pb, $rev) = @_;
4219        my ($mode, $blob);
4220
4221        goto out if $self->is_path_ignored($path);
4222
4223        my $gpath = $self->git_path($path);
4224        ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4225                             =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4226        unless (defined $mode && defined $blob) {
4227                die "$path was not found in commit $self->{c} (r$rev)\n";
4228        }
4229        if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4230                $mode = '120000';
4231        }
4232out:
4233        { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4234          pool => SVN::Pool->new, action => 'M' };
4235}
4236
4237sub add_file {
4238        my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4239        my $mode;
4240
4241        if (!$self->is_path_ignored($path)) {
4242                my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4243                delete $self->{empty}->{$dir};
4244                $mode = '100644';
4245        }
4246        { path => $path, mode_a => $mode, mode_b => $mode,
4247          pool => SVN::Pool->new, action => 'A' };
4248}
4249
4250sub add_directory {
4251        my ($self, $path, $cp_path, $cp_rev) = @_;
4252        goto out if $self->is_path_ignored($path);
4253        my $gpath = $self->git_path($path);
4254        if ($gpath eq '') {
4255                my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4256                                                     -r --name-only -z/,
4257                                                     $self->{c});
4258                local $/ = "\0";
4259                while (<$ls>) {
4260                        chomp;
4261                        $self->{gii}->remove($_);
4262                        print "\tD\t$_\n" unless $::_q;
4263                }
4264                command_close_pipe($ls, $ctx);
4265                $self->{empty}->{$path} = 0;
4266        }
4267        my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4268        delete $self->{empty}->{$dir};
4269        $self->{empty}->{$path} = 1;
4270out:
4271        { path => $path };
4272}
4273
4274sub change_dir_prop {
4275        my ($self, $db, $prop, $value) = @_;
4276        return undef if $self->is_path_ignored($db->{path});
4277        $self->{dir_prop}->{$db->{path}} ||= {};
4278        $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4279        undef;
4280}
4281
4282sub absent_directory {
4283        my ($self, $path, $pb) = @_;
4284        return undef if $self->is_path_ignored($path);
4285        $self->{absent_dir}->{$pb->{path}} ||= [];
4286        push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4287        undef;
4288}
4289
4290sub absent_file {
4291        my ($self, $path, $pb) = @_;
4292        return undef if $self->is_path_ignored($path);
4293        $self->{absent_file}->{$pb->{path}} ||= [];
4294        push @{$self->{absent_file}->{$pb->{path}}}, $path;
4295        undef;
4296}
4297
4298sub change_file_prop {
4299        my ($self, $fb, $prop, $value) = @_;
4300        return undef if $self->is_path_ignored($fb->{path});
4301        if ($prop eq 'svn:executable') {
4302                if ($fb->{mode_b} != 120000) {
4303                        $fb->{mode_b} = defined $value ? 100755 : 100644;
4304                }
4305        } elsif ($prop eq 'svn:special') {
4306                $fb->{mode_b} = defined $value ? 120000 : 100644;
4307        } else {
4308                $self->{file_prop}->{$fb->{path}} ||= {};
4309                $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4310        }
4311        undef;
4312}
4313
4314sub apply_textdelta {
4315        my ($self, $fb, $exp) = @_;
4316        return undef if $self->is_path_ignored($fb->{path});
4317        my $fh = $::_repository->temp_acquire('svn_delta');
4318        # $fh gets auto-closed() by SVN::TxDelta::apply(),
4319        # (but $base does not,) so dup() it for reading in close_file
4320        open my $dup, '<&', $fh or croak $!;
4321        my $base = $::_repository->temp_acquire('git_blob');
4322
4323        if ($fb->{blob}) {
4324                my ($base_is_link, $size);
4325
4326                if ($fb->{mode_a} eq '120000' &&
4327                    ! $self->{empty_symlinks}->{$fb->{path}}) {
4328                        print $base 'link ' or die "print $!\n";
4329                        $base_is_link = 1;
4330                }
4331        retry:
4332                $size = $::_repository->cat_blob($fb->{blob}, $base);
4333                die "Failed to read object $fb->{blob}" if ($size < 0);
4334
4335                if (defined $exp) {
4336                        seek $base, 0, 0 or croak $!;
4337                        my $got = ::md5sum($base);
4338                        if ($got ne $exp) {
4339                                my $err = "Checksum mismatch: ".
4340                                       "$fb->{path} $fb->{blob}\n" .
4341                                       "expected: $exp\n" .
4342                                       "     got: $got\n";
4343                                if ($base_is_link) {
4344                                        warn $err,
4345                                             "Retrying... (possibly ",
4346                                             "a bad symlink from SVN)\n";
4347                                        $::_repository->temp_reset($base);
4348                                        $base_is_link = 0;
4349                                        goto retry;
4350                                }
4351                                die $err;
4352                        }
4353                }
4354        }
4355        seek $base, 0, 0 or croak $!;
4356        $fb->{fh} = $fh;
4357        $fb->{base} = $base;
4358        [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4359}
4360
4361sub close_file {
4362        my ($self, $fb, $exp) = @_;
4363        return undef if $self->is_path_ignored($fb->{path});
4364
4365        my $hash;
4366        my $path = $self->git_path($fb->{path});
4367        if (my $fh = $fb->{fh}) {
4368                if (defined $exp) {
4369                        seek($fh, 0, 0) or croak $!;
4370                        my $got = ::md5sum($fh);
4371                        if ($got ne $exp) {
4372                                die "Checksum mismatch: $path\n",
4373                                    "expected: $exp\n    got: $got\n";
4374                        }
4375                }
4376                if ($fb->{mode_b} == 120000) {
4377                        sysseek($fh, 0, 0) or croak $!;
4378                        my $rd = sysread($fh, my $buf, 5);
4379
4380                        if (!defined $rd) {
4381                                croak "sysread: $!\n";
4382                        } elsif ($rd == 0) {
4383                                warn "$path has mode 120000",
4384                                     " but it points to nothing\n",
4385                                     "converting to an empty file with mode",
4386                                     " 100644\n";
4387                                $fb->{mode_b} = '100644';
4388                        } elsif ($buf ne 'link ') {
4389                                warn "$path has mode 120000",
4390                                     " but is not a link\n";
4391                        } else {
4392                                my $tmp_fh = $::_repository->temp_acquire(
4393                                        'svn_hash');
4394                                my $res;
4395                                while ($res = sysread($fh, my $str, 1024)) {
4396                                        my $out = syswrite($tmp_fh, $str, $res);
4397                                        defined($out) && $out == $res
4398                                                or croak("write ",
4399                                                        Git::temp_path($tmp_fh),
4400                                                        ": $!\n");
4401                                }
4402                                defined $res or croak $!;
4403
4404                                ($fh, $tmp_fh) = ($tmp_fh, $fh);
4405                                Git::temp_release($tmp_fh, 1);
4406                        }
4407                }
4408
4409                $hash = $::_repository->hash_and_insert_object(
4410                                Git::temp_path($fh));
4411                $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4412
4413                Git::temp_release($fb->{base}, 1);
4414                Git::temp_release($fh, 1);
4415        } else {
4416                $hash = $fb->{blob} or die "no blob information\n";
4417        }
4418        $fb->{pool}->clear;
4419        $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4420        print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4421        undef;
4422}
4423
4424sub abort_edit {
4425        my $self = shift;
4426        $self->{nr} = $self->{gii}->{nr};
4427        delete $self->{gii};
4428        $self->SUPER::abort_edit(@_);
4429}
4430
4431sub close_edit {
4432        my $self = shift;
4433        $self->{git_commit_ok} = 1;
4434        $self->{nr} = $self->{gii}->{nr};
4435        delete $self->{gii};
4436        $self->SUPER::close_edit(@_);
4437}
4438
4439package SVN::Git::Editor;
4440use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4441use strict;
4442use warnings;
4443use Carp qw/croak/;
4444use IO::File;
4445
4446sub new {
4447        my ($class, $opts) = @_;
4448        foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4449                die "$_ required!\n" unless (defined $opts->{$_});
4450        }
4451
4452        my $pool = SVN::Pool->new;
4453        my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4454        my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4455                                     $opts->{r}, $mods);
4456
4457        # $opts->{ra} functions should not be used after this:
4458        my @ce  = $opts->{ra}->get_commit_editor($opts->{log},
4459                                                $opts->{editor_cb}, $pool);
4460        my $self = SVN::Delta::Editor->new(@ce, $pool);
4461        bless $self, $class;
4462        foreach (qw/svn_path r tree_a tree_b/) {
4463                $self->{$_} = $opts->{$_};
4464        }
4465        $self->{url} = $opts->{ra}->{url};
4466        $self->{mods} = $mods;
4467        $self->{types} = $types;
4468        $self->{pool} = $pool;
4469        $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4470        $self->{rm} = { };
4471        $self->{path_prefix} = length $self->{svn_path} ?
4472                               "$self->{svn_path}/" : '';
4473        $self->{config} = $opts->{config};
4474        $self->{mergeinfo} = $opts->{mergeinfo};
4475        return $self;
4476}
4477
4478sub generate_diff {
4479        my ($tree_a, $tree_b) = @_;
4480        my @diff_tree = qw(diff-tree -z -r);
4481        if ($_cp_similarity) {
4482                push @diff_tree, "-C$_cp_similarity";
4483        } else {
4484                push @diff_tree, '-C';
4485        }
4486        push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4487        push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4488        push @diff_tree, $tree_a, $tree_b;
4489        my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4490        local $/ = "\0";
4491        my $state = 'meta';
4492        my @mods;
4493        while (<$diff_fh>) {
4494                chomp $_; # this gets rid of the trailing "\0"
4495                if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4496                                        ($::sha1)\s($::sha1)\s
4497                                        ([MTCRAD])\d*$/xo) {
4498                        push @mods, {   mode_a => $1, mode_b => $2,
4499                                        sha1_a => $3, sha1_b => $4,
4500                                        chg => $5 };
4501                        if ($5 =~ /^(?:C|R)$/) {
4502                                $state = 'file_a';
4503                        } else {
4504                                $state = 'file_b';
4505                        }
4506                } elsif ($state eq 'file_a') {
4507                        my $x = $mods[$#mods] or croak "Empty array\n";
4508                        if ($x->{chg} !~ /^(?:C|R)$/) {
4509                                croak "Error parsing $_, $x->{chg}\n";
4510                        }
4511                        $x->{file_a} = $_;
4512                        $state = 'file_b';
4513                } elsif ($state eq 'file_b') {
4514                        my $x = $mods[$#mods] or croak "Empty array\n";
4515                        if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4516                                croak "Error parsing $_, $x->{chg}\n";
4517                        }
4518                        if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4519                                croak "Error parsing $_, $x->{chg}\n";
4520                        }
4521                        $x->{file_b} = $_;
4522                        $state = 'meta';
4523                } else {
4524                        croak "Error parsing $_\n";
4525                }
4526        }
4527        command_close_pipe($diff_fh, $ctx);
4528        \@mods;
4529}
4530
4531sub check_diff_paths {
4532        my ($ra, $pfx, $rev, $mods) = @_;
4533        my %types;
4534        $pfx .= '/' if length $pfx;
4535
4536        sub type_diff_paths {
4537                my ($ra, $types, $path, $rev) = @_;
4538                my @p = split m#/+#, $path;
4539                my $c = shift @p;
4540                unless (defined $types->{$c}) {
4541                        $types->{$c} = $ra->check_path($c, $rev);
4542                }
4543                while (@p) {
4544                        $c .= '/' . shift @p;
4545                        next if defined $types->{$c};
4546                        $types->{$c} = $ra->check_path($c, $rev);
4547                }
4548        }
4549
4550        foreach my $m (@$mods) {
4551                foreach my $f (qw/file_a file_b/) {
4552                        next unless defined $m->{$f};
4553                        my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4554                        if (length $pfx.$dir && ! defined $types{$dir}) {
4555                                type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4556                        }
4557                }
4558        }
4559        \%types;
4560}
4561
4562sub split_path {
4563        return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4564}
4565
4566sub repo_path {
4567        my ($self, $path) = @_;
4568        if (my $enc = $self->{pathnameencoding}) {
4569                require Encode;
4570                Encode::from_to($path, $enc, 'UTF-8');
4571        }
4572        $self->{path_prefix}.(defined $path ? $path : '');
4573}
4574
4575sub url_path {
4576        my ($self, $path) = @_;
4577        if ($self->{url} =~ m#^https?://#) {
4578                $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4579        }
4580        $self->{url} . '/' . $self->repo_path($path);
4581}
4582
4583sub rmdirs {
4584        my ($self) = @_;
4585        my $rm = $self->{rm};
4586        delete $rm->{''}; # we never delete the url we're tracking
4587        return unless %$rm;
4588
4589        foreach (keys %$rm) {
4590                my @d = split m#/#, $_;
4591                my $c = shift @d;
4592                $rm->{$c} = 1;
4593                while (@d) {
4594                        $c .= '/' . shift @d;
4595                        $rm->{$c} = 1;
4596                }
4597        }
4598        delete $rm->{$self->{svn_path}};
4599        delete $rm->{''}; # we never delete the url we're tracking
4600        return unless %$rm;
4601
4602        my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4603                                             $self->{tree_b});
4604        local $/ = "\0";
4605        while (<$fh>) {
4606                chomp;
4607                my @dn = split m#/#, $_;
4608                while (pop @dn) {
4609                        delete $rm->{join '/', @dn};
4610                }
4611                unless (%$rm) {
4612                        close $fh;
4613                        return;
4614                }
4615        }
4616        command_close_pipe($fh, $ctx);
4617
4618        my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4619        foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4620                $self->close_directory($bat->{$d}, $p);
4621                my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4622                print "\tD+\t$d/\n" unless $::_q;
4623                $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4624                delete $bat->{$d};
4625        }
4626}
4627
4628sub open_or_add_dir {
4629        my ($self, $full_path, $baton) = @_;
4630        my $t = $self->{types}->{$full_path};
4631        if (!defined $t) {
4632                die "$full_path not known in r$self->{r} or we have a bug!\n";
4633        }
4634        {
4635                no warnings 'once';
4636                # SVN::Node::none and SVN::Node::file are used only once,
4637                # so we're shutting up Perl's warnings about them.
4638                if ($t == $SVN::Node::none) {
4639                        return $self->add_directory($full_path, $baton,
4640                            undef, -1, $self->{pool});
4641                } elsif ($t == $SVN::Node::dir) {
4642                        return $self->open_directory($full_path, $baton,
4643                            $self->{r}, $self->{pool});
4644                } # no warnings 'once'
4645                print STDERR "$full_path already exists in repository at ",
4646                    "r$self->{r} and it is not a directory (",
4647                    ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4648        } # no warnings 'once'
4649        exit 1;
4650}
4651
4652sub ensure_path {
4653        my ($self, $path) = @_;
4654        my $bat = $self->{bat};
4655        my $repo_path = $self->repo_path($path);
4656        return $bat->{''} unless (length $repo_path);
4657        my @p = split m#/+#, $repo_path;
4658        my $c = shift @p;
4659        $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4660        while (@p) {
4661                my $c0 = $c;
4662                $c .= '/' . shift @p;
4663                $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4664        }
4665        return $bat->{$c};
4666}
4667
4668# Subroutine to convert a globbing pattern to a regular expression.
4669# From perl cookbook.
4670sub glob2pat {
4671        my $globstr = shift;
4672        my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4673        $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4674        return '^' . $globstr . '$';
4675}
4676
4677sub check_autoprop {
4678        my ($self, $pattern, $properties, $file, $fbat) = @_;
4679        # Convert the globbing pattern to a regular expression.
4680        my $regex = glob2pat($pattern);
4681        # Check if the pattern matches the file name.
4682        if($file =~ m/($regex)/) {
4683                # Parse the list of properties to set.
4684                my @props = split(/;/, $properties);
4685                foreach my $prop (@props) {
4686                        # Parse 'name=value' syntax and set the property.
4687                        if ($prop =~ /([^=]+)=(.*)/) {
4688                                my ($n,$v) = ($1,$2);
4689                                for ($n, $v) {
4690                                        s/^\s+//; s/\s+$//;
4691                                }
4692                                $self->change_file_prop($fbat, $n, $v);
4693                        }
4694                }
4695        }
4696}
4697
4698sub apply_autoprops {
4699        my ($self, $file, $fbat) = @_;
4700        my $conf_t = ${$self->{config}}{'config'};
4701        no warnings 'once';
4702        # Check [miscellany]/enable-auto-props in svn configuration.
4703        if (SVN::_Core::svn_config_get_bool(
4704                $conf_t,
4705                $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4706                $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4707                0)) {
4708                # Auto-props are enabled.  Enumerate them to look for matches.
4709                my $callback = sub {
4710                        $self->check_autoprop($_[0], $_[1], $file, $fbat);
4711                };
4712                SVN::_Core::svn_config_enumerate(
4713                        $conf_t,
4714                        $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4715                        $callback);
4716        }
4717}
4718
4719sub A {
4720        my ($self, $m) = @_;
4721        my ($dir, $file) = split_path($m->{file_b});
4722        my $pbat = $self->ensure_path($dir);
4723        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4724                                        undef, -1);
4725        print "\tA\t$m->{file_b}\n" unless $::_q;
4726        $self->apply_autoprops($file, $fbat);
4727        $self->chg_file($fbat, $m);
4728        $self->close_file($fbat,undef,$self->{pool});
4729}
4730
4731sub C {
4732        my ($self, $m) = @_;
4733        my ($dir, $file) = split_path($m->{file_b});
4734        my $pbat = $self->ensure_path($dir);
4735        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4736                                $self->url_path($m->{file_a}), $self->{r});
4737        print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4738        $self->chg_file($fbat, $m);
4739        $self->close_file($fbat,undef,$self->{pool});
4740}
4741
4742sub delete_entry {
4743        my ($self, $path, $pbat) = @_;
4744        my $rpath = $self->repo_path($path);
4745        my ($dir, $file) = split_path($rpath);
4746        $self->{rm}->{$dir} = 1;
4747        $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4748}
4749
4750sub R {
4751        my ($self, $m) = @_;
4752        my ($dir, $file) = split_path($m->{file_b});
4753        my $pbat = $self->ensure_path($dir);
4754        my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4755                                $self->url_path($m->{file_a}), $self->{r});
4756        print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4757        $self->apply_autoprops($file, $fbat);
4758        $self->chg_file($fbat, $m);
4759        $self->close_file($fbat,undef,$self->{pool});
4760
4761        ($dir, $file) = split_path($m->{file_a});
4762        $pbat = $self->ensure_path($dir);
4763        $self->delete_entry($m->{file_a}, $pbat);
4764}
4765
4766sub M {
4767        my ($self, $m) = @_;
4768        my ($dir, $file) = split_path($m->{file_b});
4769        my $pbat = $self->ensure_path($dir);
4770        my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4771                                $pbat,$self->{r},$self->{pool});
4772        print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4773        $self->chg_file($fbat, $m);
4774        $self->close_file($fbat,undef,$self->{pool});
4775}
4776
4777sub T { shift->M(@_) }
4778
4779sub change_file_prop {
4780        my ($self, $fbat, $pname, $pval) = @_;
4781        $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4782}
4783
4784sub change_dir_prop {
4785        my ($self, $pbat, $pname, $pval) = @_;
4786        $self->SUPER::change_dir_prop($pbat, $pname, $pval, $self->{pool});
4787}
4788
4789sub _chg_file_get_blob ($$$$) {
4790        my ($self, $fbat, $m, $which) = @_;
4791        my $fh = $::_repository->temp_acquire("git_blob_$which");
4792        if ($m->{"mode_$which"} =~ /^120/) {
4793                print $fh 'link ' or croak $!;
4794                $self->change_file_prop($fbat,'svn:special','*');
4795        } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4796                $self->change_file_prop($fbat,'svn:special',undef);
4797        }
4798        my $blob = $m->{"sha1_$which"};
4799        return ($fh,) if ($blob =~ /^0{40}$/);
4800        my $size = $::_repository->cat_blob($blob, $fh);
4801        croak "Failed to read object $blob" if ($size < 0);
4802        $fh->flush == 0 or croak $!;
4803        seek $fh, 0, 0 or croak $!;
4804
4805        my $exp = ::md5sum($fh);
4806        seek $fh, 0, 0 or croak $!;
4807        return ($fh, $exp);
4808}
4809
4810sub chg_file {
4811        my ($self, $fbat, $m) = @_;
4812        if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4813                $self->change_file_prop($fbat,'svn:executable','*');
4814        } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4815                $self->change_file_prop($fbat,'svn:executable',undef);
4816        }
4817        my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4818        my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4819        my $pool = SVN::Pool->new;
4820        my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4821        if (-s $fh_a) {
4822                my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4823                my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4824                if (defined $res) {
4825                        die "Unexpected result from send_txstream: $res\n",
4826                            "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4827                }
4828        } else {
4829                my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4830                die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4831                    if ($got ne $exp_b);
4832        }
4833        Git::temp_release($fh_b, 1);
4834        Git::temp_release($fh_a, 1);
4835        $pool->clear;
4836}
4837
4838sub D {
4839        my ($self, $m) = @_;
4840        my ($dir, $file) = split_path($m->{file_b});
4841        my $pbat = $self->ensure_path($dir);
4842        print "\tD\t$m->{file_b}\n" unless $::_q;
4843        $self->delete_entry($m->{file_b}, $pbat);
4844}
4845
4846sub close_edit {
4847        my ($self) = @_;
4848        my ($p,$bat) = ($self->{pool}, $self->{bat});
4849        foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4850                next if $_ eq '';
4851                $self->close_directory($bat->{$_}, $p);
4852        }
4853        $self->close_directory($bat->{''}, $p);
4854        $self->SUPER::close_edit($p);
4855        $p->clear;
4856}
4857
4858sub abort_edit {
4859        my ($self) = @_;
4860        $self->SUPER::abort_edit($self->{pool});
4861}
4862
4863sub DESTROY {
4864        my $self = shift;
4865        $self->SUPER::DESTROY(@_);
4866        $self->{pool}->clear;
4867}
4868
4869# this drives the editor
4870sub apply_diff {
4871        my ($self) = @_;
4872        my $mods = $self->{mods};
4873        my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4874        foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4875                my $f = $m->{chg};
4876                if (defined $o{$f}) {
4877                        $self->$f($m);
4878                } else {
4879                        fatal("Invalid change type: $f");
4880                }
4881        }
4882
4883        if (defined($self->{mergeinfo})) {
4884                $self->change_dir_prop($self->{bat}{''}, "svn:mergeinfo",
4885                                       $self->{mergeinfo});
4886        }
4887        $self->rmdirs if $_rmdir;
4888        if (@$mods == 0) {
4889                $self->abort_edit;
4890        } else {
4891                $self->close_edit;
4892        }
4893        return scalar @$mods;
4894}
4895
4896package Git::SVN::Ra;
4897use vars qw/@ISA $config_dir $_log_window_size/;
4898use strict;
4899use warnings;
4900my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4901
4902BEGIN {
4903        # enforce temporary pool usage for some simple functions
4904        no strict 'refs';
4905        for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4906                      get_file/) {
4907                my $SUPER = "SUPER::$f";
4908                *$f = sub {
4909                        my $self = shift;
4910                        my $pool = SVN::Pool->new;
4911                        my @ret = $self->$SUPER(@_,$pool);
4912                        $pool->clear;
4913                        wantarray ? @ret : $ret[0];
4914                };
4915        }
4916}
4917
4918sub _auth_providers () {
4919        [
4920          SVN::Client::get_simple_provider(),
4921          SVN::Client::get_ssl_server_trust_file_provider(),
4922          SVN::Client::get_simple_prompt_provider(
4923            \&Git::SVN::Prompt::simple, 2),
4924          SVN::Client::get_ssl_client_cert_file_provider(),
4925          SVN::Client::get_ssl_client_cert_prompt_provider(
4926            \&Git::SVN::Prompt::ssl_client_cert, 2),
4927          SVN::Client::get_ssl_client_cert_pw_file_provider(),
4928          SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4929            \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4930          SVN::Client::get_username_provider(),
4931          SVN::Client::get_ssl_server_trust_prompt_provider(
4932            \&Git::SVN::Prompt::ssl_server_trust),
4933          SVN::Client::get_username_prompt_provider(
4934            \&Git::SVN::Prompt::username, 2)
4935        ]
4936}
4937
4938sub escape_uri_only {
4939        my ($uri) = @_;
4940        my @tmp;
4941        foreach (split m{/}, $uri) {
4942                s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4943                push @tmp, $_;
4944        }
4945        join('/', @tmp);
4946}
4947
4948sub escape_url {
4949        my ($url) = @_;
4950        if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4951                my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4952                $url = "$scheme://$domain$uri";
4953        }
4954        $url;
4955}
4956
4957sub new {
4958        my ($class, $url) = @_;
4959        $url =~ s!/+$!!;
4960        return $RA if ($RA && $RA->{url} eq $url);
4961
4962        ::_req_svn();
4963
4964        SVN::_Core::svn_config_ensure($config_dir, undef);
4965        my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4966        my $config = SVN::Core::config_get_config($config_dir);
4967        $RA = undef;
4968        my $dont_store_passwords = 1;
4969        my $conf_t = ${$config}{'config'};
4970        {
4971                no warnings 'once';
4972                # The usage of $SVN::_Core::SVN_CONFIG_* variables
4973                # produces warnings that variables are used only once.
4974                # I had not found the better way to shut them up, so
4975                # the warnings of type 'once' are disabled in this block.
4976                if (SVN::_Core::svn_config_get_bool($conf_t,
4977                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4978                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4979                    1) == 0) {
4980                        SVN::_Core::svn_auth_set_parameter($baton,
4981                            $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4982                            bless (\$dont_store_passwords, "_p_void"));
4983                }
4984                if (SVN::_Core::svn_config_get_bool($conf_t,
4985                    $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4986                    $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4987                    1) == 0) {
4988                        $Git::SVN::Prompt::_no_auth_cache = 1;
4989                }
4990        } # no warnings 'once'
4991        my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4992                              config => $config,
4993                              pool => SVN::Pool->new,
4994                              auth_provider_callbacks => $callbacks);
4995        $self->{url} = $url;
4996        $self->{svn_path} = $url;
4997        $self->{repos_root} = $self->get_repos_root;
4998        $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4999        $self->{cache} = { check_path => { r => 0, data => {} },
5000                           get_dir => { r => 0, data => {} } };
5001        $RA = bless $self, $class;
5002}
5003
5004sub check_path {
5005        my ($self, $path, $r) = @_;
5006        my $cache = $self->{cache}->{check_path};
5007        if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
5008                return $cache->{data}->{$path};
5009        }
5010        my $pool = SVN::Pool->new;
5011        my $t = $self->SUPER::check_path($path, $r, $pool);
5012        $pool->clear;
5013        if ($r != $cache->{r}) {
5014                %{$cache->{data}} = ();
5015                $cache->{r} = $r;
5016        }
5017        $cache->{data}->{$path} = $t;
5018}
5019
5020sub get_dir {
5021        my ($self, $dir, $r) = @_;
5022        my $cache = $self->{cache}->{get_dir};
5023        if ($r == $cache->{r}) {
5024                if (my $x = $cache->{data}->{$dir}) {
5025                        return wantarray ? @$x : $x->[0];
5026                }
5027        }
5028        my $pool = SVN::Pool->new;
5029        my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
5030        my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
5031        $pool->clear;
5032        if ($r != $cache->{r}) {
5033                %{$cache->{data}} = ();
5034                $cache->{r} = $r;
5035        }
5036        $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
5037        wantarray ? (\%dirents, $r, $props) : \%dirents;
5038}
5039
5040sub DESTROY {
5041        # do not call the real DESTROY since we store ourselves in $RA
5042}
5043
5044# get_log(paths, start, end, limit,
5045#         discover_changed_paths, strict_node_history, receiver)
5046sub get_log {
5047        my ($self, @args) = @_;
5048        my $pool = SVN::Pool->new;
5049
5050        # svn_log_changed_path_t objects passed to get_log are likely to be
5051        # overwritten even if only the refs are copied to an external variable,
5052        # so we should dup the structures in their entirety.  Using an
5053        # externally passed pool (instead of our temporary and quickly cleared
5054        # pool in Git::SVN::Ra) does not help matters at all...
5055        my $receiver = pop @args;
5056        my $prefix = "/".$self->{svn_path};
5057        $prefix =~ s#/+($)##;
5058        my $prefix_regex = qr#^\Q$prefix\E#;
5059        push(@args, sub {
5060                my ($paths) = $_[0];
5061                return &$receiver(@_) unless $paths;
5062                $_[0] = ();
5063                foreach my $p (keys %$paths) {
5064                        my $i = $paths->{$p};
5065                        # Make path relative to our url, not repos_root
5066                        $p =~ s/$prefix_regex//;
5067                        my %s = map { $_ => $i->$_; }
5068                                qw/copyfrom_path copyfrom_rev action/;
5069                        if ($s{'copyfrom_path'}) {
5070                                $s{'copyfrom_path'} =~ s/$prefix_regex//;
5071                        }
5072                        $_[0]{$p} = \%s;
5073                }
5074                &$receiver(@_);
5075        });
5076
5077
5078        # the limit parameter was not supported in SVN 1.1.x, so we
5079        # drop it.  Therefore, the receiver callback passed to it
5080        # is made aware of this limitation by being wrapped if
5081        # the limit passed to is being wrapped.
5082        if ($SVN::Core::VERSION le '1.2.0') {
5083                my $limit = splice(@args, 3, 1);
5084                if ($limit > 0) {
5085                        my $receiver = pop @args;
5086                        push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
5087                }
5088        }
5089        my $ret = $self->SUPER::get_log(@args, $pool);
5090        $pool->clear;
5091        $ret;
5092}
5093
5094sub trees_match {
5095        my ($self, $url1, $rev1, $url2, $rev2) = @_;
5096        my $ctx = SVN::Client->new(auth => _auth_providers);
5097        my $out = IO::File->new_tmpfile;
5098
5099        # older SVN (1.1.x) doesn't take $pool as the last parameter for
5100        # $ctx->diff(), so we'll create a default one
5101        my $pool = SVN::Pool->new_default_sub;
5102
5103        $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5104        $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5105        $out->flush;
5106        my $ret = (($out->stat)[7] == 0);
5107        close $out or croak $!;
5108
5109        $ret;
5110}
5111
5112sub get_commit_editor {
5113        my ($self, $log, $cb, $pool) = @_;
5114        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
5115        $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5116}
5117
5118sub gs_do_update {
5119        my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5120        my $new = ($rev_a == $rev_b);
5121        my $path = $gs->{path};
5122
5123        if ($new && -e $gs->{index}) {
5124                unlink $gs->{index} or die
5125                  "Couldn't unlink index: $gs->{index}: $!\n";
5126        }
5127        my $pool = SVN::Pool->new;
5128        $editor->set_path_strip($path);
5129        my (@pc) = split m#/#, $path;
5130        my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5131                                        1, $editor, $pool);
5132        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5133
5134        # Since we can't rely on svn_ra_reparent being available, we'll
5135        # just have to do some magic with set_path to make it so
5136        # we only want a partial path.
5137        my $sp = '';
5138        my $final = join('/', @pc);
5139        while (@pc) {
5140                $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5141                $sp .= '/' if length $sp;
5142                $sp .= shift @pc;
5143        }
5144        die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5145
5146        $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5147
5148        $reporter->finish_report($pool);
5149        $pool->clear;
5150        $editor->{git_commit_ok};
5151}
5152
5153# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5154# svn_ra_reparent didn't work before 1.4)
5155sub gs_do_switch {
5156        my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5157        my $path = $gs->{path};
5158        my $pool = SVN::Pool->new;
5159
5160        my $full_url = $self->{url};
5161        my $old_url = $full_url;
5162        $full_url .= '/' . $path if length $path;
5163        my ($ra, $reparented);
5164
5165        if ($old_url =~ m#^svn(\+ssh)?://# ||
5166            ($full_url =~ m#^https?://# &&
5167             escape_url($full_url) ne $full_url)) {
5168                $_[0] = undef;
5169                $self = undef;
5170                $RA = undef;
5171                $ra = Git::SVN::Ra->new($full_url);
5172                $ra_invalid = 1;
5173        } elsif ($old_url ne $full_url) {
5174                SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5175                $self->{url} = $full_url;
5176                $reparented = 1;
5177        }
5178
5179        $ra ||= $self;
5180        $url_b = escape_url($url_b);
5181        my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5182        my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5183        $reporter->set_path('', $rev_a, 0, @lock, $pool);
5184        $reporter->finish_report($pool);
5185
5186        if ($reparented) {
5187                SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5188                $self->{url} = $old_url;
5189        }
5190
5191        $pool->clear;
5192        $editor->{git_commit_ok};
5193}
5194
5195sub longest_common_path {
5196        my ($gsv, $globs) = @_;
5197        my %common;
5198        my $common_max = scalar @$gsv;
5199
5200        foreach my $gs (@$gsv) {
5201                my @tmp = split m#/#, $gs->{path};
5202                my $p = '';
5203                foreach (@tmp) {
5204                        $p .= length($p) ? "/$_" : $_;
5205                        $common{$p} ||= 0;
5206                        $common{$p}++;
5207                }
5208        }
5209        $globs ||= [];
5210        $common_max += scalar @$globs;
5211        foreach my $glob (@$globs) {
5212                my @tmp = split m#/#, $glob->{path}->{left};
5213                my $p = '';
5214                foreach (@tmp) {
5215                        $p .= length($p) ? "/$_" : $_;
5216                        $common{$p} ||= 0;
5217                        $common{$p}++;
5218                }
5219        }
5220
5221        my $longest_path = '';
5222        foreach (sort {length $b <=> length $a} keys %common) {
5223                if ($common{$_} == $common_max) {
5224                        $longest_path = $_;
5225                        last;
5226                }
5227        }
5228        $longest_path;
5229}
5230
5231sub gs_fetch_loop_common {
5232        my ($self, $base, $head, $gsv, $globs) = @_;
5233        return if ($base > $head);
5234        my $inc = $_log_window_size;
5235        my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5236        my $longest_path = longest_common_path($gsv, $globs);
5237        my $ra_url = $self->{url};
5238        my $find_trailing_edge;
5239        while (1) {
5240                my %revs;
5241                my $err;
5242                my $err_handler = $SVN::Error::handler;
5243                $SVN::Error::handler = sub {
5244                        ($err) = @_;
5245                        skip_unknown_revs($err);
5246                };
5247                sub _cb {
5248                        my ($paths, $r, $author, $date, $log) = @_;
5249                        [ $paths,
5250                          { author => $author, date => $date, log => $log } ];
5251                }
5252                $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5253                               sub { $revs{$_[1]} = _cb(@_) });
5254                if ($err) {
5255                        print "Checked through r$max\r";
5256                } else {
5257                        $find_trailing_edge = 1;
5258                }
5259                if ($err and $find_trailing_edge) {
5260                        print STDERR "Path '$longest_path' ",
5261                                     "was probably deleted:\n",
5262                                     $err->expanded_message,
5263                                     "\nWill attempt to follow ",
5264                                     "revisions r$min .. r$max ",
5265                                     "committed before the deletion\n";
5266                        my $hi = $max;
5267                        while (--$hi >= $min) {
5268                                my $ok;
5269                                $self->get_log([$longest_path], $min, $hi,
5270                                               0, 1, 1, sub {
5271                                               $ok = $_[1];
5272                                               $revs{$_[1]} = _cb(@_) });
5273                                if ($ok) {
5274                                        print STDERR "r$min .. r$ok OK\n";
5275                                        last;
5276                                }
5277                        }
5278                        $find_trailing_edge = 0;
5279                }
5280                $SVN::Error::handler = $err_handler;
5281
5282                my %exists = map { $_->{path} => $_ } @$gsv;
5283                foreach my $r (sort {$a <=> $b} keys %revs) {
5284                        my ($paths, $logged) = @{$revs{$r}};
5285
5286                        foreach my $gs ($self->match_globs(\%exists, $paths,
5287                                                           $globs, $r)) {
5288                                if ($gs->rev_map_max >= $r) {
5289                                        next;
5290                                }
5291                                next unless $gs->match_paths($paths, $r);
5292                                $gs->{logged_rev_props} = $logged;
5293                                if (my $last_commit = $gs->last_commit) {
5294                                        $gs->assert_index_clean($last_commit);
5295                                }
5296                                my $log_entry = $gs->do_fetch($paths, $r);
5297                                if ($log_entry) {
5298                                        $gs->do_git_commit($log_entry);
5299                                }
5300                                $INDEX_FILES{$gs->{index}} = 1;
5301                        }
5302                        foreach my $g (@$globs) {
5303                                my $k = "svn-remote.$g->{remote}." .
5304                                        "$g->{t}-maxRev";
5305                                Git::SVN::tmp_config($k, $r);
5306                        }
5307                        if ($ra_invalid) {
5308                                $_[0] = undef;
5309                                $self = undef;
5310                                $RA = undef;
5311                                $self = Git::SVN::Ra->new($ra_url);
5312                                $ra_invalid = undef;
5313                        }
5314                }
5315                # pre-fill the .rev_db since it'll eventually get filled in
5316                # with '0' x40 if something new gets committed
5317                foreach my $gs (@$gsv) {
5318                        next if $gs->rev_map_max >= $max;
5319                        next if defined $gs->rev_map_get($max);
5320                        $gs->rev_map_set($max, 0 x40);
5321                }
5322                foreach my $g (@$globs) {
5323                        my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5324                        Git::SVN::tmp_config($k, $max);
5325                }
5326                last if $max >= $head;
5327                $min = $max + 1;
5328                $max += $inc;
5329                $max = $head if ($max > $head);
5330        }
5331        Git::SVN::gc();
5332}
5333
5334sub get_dir_globbed {
5335        my ($self, $left, $depth, $r) = @_;
5336
5337        my @x = eval { $self->get_dir($left, $r) };
5338        return unless scalar @x == 3;
5339        my $dirents = $x[0];
5340        my @finalents;
5341        foreach my $de (keys %$dirents) {
5342                next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5343                if ($depth > 1) {
5344                        my @args = ("$left/$de", $depth - 1, $r);
5345                        foreach my $dir ($self->get_dir_globbed(@args)) {
5346                                push @finalents, "$de/$dir";
5347                        }
5348                } else {
5349                        push @finalents, $de;
5350                }
5351        }
5352        @finalents;
5353}
5354
5355sub match_globs {
5356        my ($self, $exists, $paths, $globs, $r) = @_;
5357
5358        sub get_dir_check {
5359                my ($self, $exists, $g, $r) = @_;
5360
5361                my @dirs = $self->get_dir_globbed($g->{path}->{left},
5362                                                  $g->{path}->{depth},
5363                                                  $r);
5364
5365                foreach my $de (@dirs) {
5366                        my $p = $g->{path}->full_path($de);
5367                        next if $exists->{$p};
5368                        next if (length $g->{path}->{right} &&
5369                                 ($self->check_path($p, $r) !=
5370                                  $SVN::Node::dir));
5371                        next unless $p =~ /$g->{path}->{regex}/;
5372                        $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5373                                         $g->{ref}->full_path($de), 1);
5374                }
5375        }
5376        foreach my $g (@$globs) {
5377                if (my $path = $paths->{"/$g->{path}->{left}"}) {
5378                        if ($path->{action} =~ /^[AR]$/) {
5379                                get_dir_check($self, $exists, $g, $r);
5380                        }
5381                }
5382                foreach (keys %$paths) {
5383                        if (/$g->{path}->{left_regex}/ &&
5384                            !/$g->{path}->{regex}/) {
5385                                next if $paths->{$_}->{action} !~ /^[AR]$/;
5386                                get_dir_check($self, $exists, $g, $r);
5387                        }
5388                        next unless /$g->{path}->{regex}/;
5389                        my $p = $1;
5390                        my $pathname = $g->{path}->full_path($p);
5391                        next if $exists->{$pathname};
5392                        next if ($self->check_path($pathname, $r) !=
5393                                 $SVN::Node::dir);
5394                        $exists->{$pathname} = Git::SVN->init(
5395                                              $self->{url}, $pathname, undef,
5396                                              $g->{ref}->full_path($p), 1);
5397                }
5398                my $c = '';
5399                foreach (split m#/#, $g->{path}->{left}) {
5400                        $c .= "/$_";
5401                        next unless ($paths->{$c} &&
5402                                     ($paths->{$c}->{action} =~ /^[AR]$/));
5403                        get_dir_check($self, $exists, $g, $r);
5404                }
5405        }
5406        values %$exists;
5407}
5408
5409sub minimize_url {
5410        my ($self) = @_;
5411        return $self->{url} if ($self->{url} eq $self->{repos_root});
5412        my $url = $self->{repos_root};
5413        my @components = split(m!/!, $self->{svn_path});
5414        my $c = '';
5415        do {
5416                $url .= "/$c" if length $c;
5417                eval {
5418                        my $ra = (ref $self)->new($url);
5419                        my $latest = $ra->get_latest_revnum;
5420                        $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5421                };
5422        } while ($@ && ($c = shift @components));
5423        $url;
5424}
5425
5426sub can_do_switch {
5427        my $self = shift;
5428        unless (defined $can_do_switch) {
5429                my $pool = SVN::Pool->new;
5430                my $rep = eval {
5431                        $self->do_switch(1, '', 0, $self->{url},
5432                                         SVN::Delta::Editor->new, $pool);
5433                };
5434                if ($@) {
5435                        $can_do_switch = 0;
5436                } else {
5437                        $rep->abort_report($pool);
5438                        $can_do_switch = 1;
5439                }
5440                $pool->clear;
5441        }
5442        $can_do_switch;
5443}
5444
5445sub skip_unknown_revs {
5446        my ($err) = @_;
5447        my $errno = $err->apr_err();
5448        # Maybe the branch we're tracking didn't
5449        # exist when the repo started, so it's
5450        # not an error if it doesn't, just continue
5451        #
5452        # Wonderfully consistent library, eh?
5453        # 160013 - svn:// and file://
5454        # 175002 - http(s)://
5455        # 175007 - http(s):// (this repo required authorization, too...)
5456        #   More codes may be discovered later...
5457        if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5458                my $err_key = $err->expanded_message;
5459                # revision numbers change every time, filter them out
5460                $err_key =~ s/\d+/\0/g;
5461                $err_key = "$errno\0$err_key";
5462                unless ($ignored_err{$err_key}) {
5463                        warn "W: Ignoring error from SVN, path probably ",
5464                             "does not exist: ($errno): ",
5465                             $err->expanded_message,"\n";
5466                        warn "W: Do not be alarmed at the above message ",
5467                             "git-svn is just searching aggressively for ",
5468                             "old history.\n",
5469                             "This may take a while on large repositories\n";
5470                        $ignored_err{$err_key} = 1;
5471                }
5472                return;
5473        }
5474        die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5475}
5476
5477package Git::SVN::Log;
5478use strict;
5479use warnings;
5480use POSIX qw/strftime/;
5481use Time::Local;
5482use constant commit_log_separator => ('-' x 72) . "\n";
5483use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5484            %rusers $show_commit $incremental/;
5485my $l_fmt;
5486
5487sub cmt_showable {
5488        my ($c) = @_;
5489        return 1 if defined $c->{r};
5490
5491        # big commit message got truncated by the 16k pretty buffer in rev-list
5492        if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5493                                $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5494                @{$c->{l}} = ();
5495                my @log = command(qw/cat-file commit/, $c->{c});
5496
5497                # shift off the headers
5498                shift @log while ($log[0] ne '');
5499                shift @log;
5500
5501                # TODO: make $c->{l} not have a trailing newline in the future
5502                @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5503
5504                (undef, $c->{r}, undef) = ::extract_metadata(
5505                                (grep(/^git-svn-id: /, @log))[-1]);
5506        }
5507        return defined $c->{r};
5508}
5509
5510sub log_use_color {
5511        return $color || Git->repository->get_colorbool('color.diff');
5512}
5513
5514sub git_svn_log_cmd {
5515        my ($r_min, $r_max, @args) = @_;
5516        my $head = 'HEAD';
5517        my (@files, @log_opts);
5518        foreach my $x (@args) {
5519                if ($x eq '--' || @files) {
5520                        push @files, $x;
5521                } else {
5522                        if (::verify_ref("$x^0")) {
5523                                $head = $x;
5524                        } else {
5525                                push @log_opts, $x;
5526                        }
5527                }
5528        }
5529
5530        my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5531        $gs ||= Git::SVN->_new;
5532        my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5533                   $gs->refname);
5534        push @cmd, '-r' unless $non_recursive;
5535        push @cmd, qw/--raw --name-status/ if $verbose;
5536        push @cmd, '--color' if log_use_color();
5537        push @cmd, @log_opts;
5538        if (defined $r_max && $r_max == $r_min) {
5539                push @cmd, '--max-count=1';
5540                if (my $c = $gs->rev_map_get($r_max)) {
5541                        push @cmd, $c;
5542                }
5543        } elsif (defined $r_max) {
5544                if ($r_max < $r_min) {
5545                        ($r_min, $r_max) = ($r_max, $r_min);
5546                }
5547                my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5548                my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5549                # If there are no commits in the range, both $c_max and $c_min
5550                # will be undefined.  If there is at least 1 commit in the
5551                # range, both will be defined.
5552                return () if !defined $c_min || !defined $c_max;
5553                if ($c_min eq $c_max) {
5554                        push @cmd, '--max-count=1', $c_min;
5555                } else {
5556                        push @cmd, '--boundary', "$c_min..$c_max";
5557                }
5558        }
5559        return (@cmd, @files);
5560}
5561
5562# adapted from pager.c
5563sub config_pager {
5564        if (! -t *STDOUT) {
5565                $ENV{GIT_PAGER_IN_USE} = 'false';
5566                $pager = undef;
5567                return;
5568        }
5569        chomp($pager = command_oneline(qw(var GIT_PAGER)));
5570        if ($pager eq 'cat') {
5571                $pager = undef;
5572        }
5573        $ENV{GIT_PAGER_IN_USE} = defined($pager);
5574}
5575
5576sub run_pager {
5577        return unless defined $pager;
5578        pipe my ($rfd, $wfd) or return;
5579        defined(my $pid = fork) or ::fatal "Can't fork: $!";
5580        if (!$pid) {
5581                open STDOUT, '>&', $wfd or
5582                                     ::fatal "Can't redirect to stdout: $!";
5583                return;
5584        }
5585        open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5586        $ENV{LESS} ||= 'FRSX';
5587        exec $pager or ::fatal "Can't run pager: $! ($pager)";
5588}
5589
5590sub format_svn_date {
5591        # some systmes don't handle or mishandle %z, so be creative.
5592        my $t = shift || time;
5593        my $gm = timelocal(gmtime($t));
5594        my $sign = qw( + + - )[ $t <=> $gm ];
5595        my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5596        return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5597}
5598
5599sub parse_git_date {
5600        my ($t, $tz) = @_;
5601        # Date::Parse isn't in the standard Perl distro :(
5602        if ($tz =~ s/^\+//) {
5603                $t += tz_to_s_offset($tz);
5604        } elsif ($tz =~ s/^\-//) {
5605                $t -= tz_to_s_offset($tz);
5606        }
5607        return $t;
5608}
5609
5610sub set_local_timezone {
5611        if (defined $TZ) {
5612                $ENV{TZ} = $TZ;
5613        } else {
5614                delete $ENV{TZ};
5615        }
5616}
5617
5618sub tz_to_s_offset {
5619        my ($tz) = @_;
5620        $tz =~ s/(\d\d)$//;
5621        return ($1 * 60) + ($tz * 3600);
5622}
5623
5624sub get_author_info {
5625        my ($dest, $author, $t, $tz) = @_;
5626        $author =~ s/(?:^\s*|\s*$)//g;
5627        $dest->{a_raw} = $author;
5628        my $au;
5629        if ($::_authors) {
5630                $au = $rusers{$author} || undef;
5631        }
5632        if (!$au) {
5633                ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5634        }
5635        $dest->{t} = $t;
5636        $dest->{tz} = $tz;
5637        $dest->{a} = $au;
5638        $dest->{t_utc} = parse_git_date($t, $tz);
5639}
5640
5641sub process_commit {
5642        my ($c, $r_min, $r_max, $defer) = @_;
5643        if (defined $r_min && defined $r_max) {
5644                if ($r_min == $c->{r} && $r_min == $r_max) {
5645                        show_commit($c);
5646                        return 0;
5647                }
5648                return 1 if $r_min == $r_max;
5649                if ($r_min < $r_max) {
5650                        # we need to reverse the print order
5651                        return 0 if (defined $limit && --$limit < 0);
5652                        push @$defer, $c;
5653                        return 1;
5654                }
5655                if ($r_min != $r_max) {
5656                        return 1 if ($r_min < $c->{r});
5657                        return 1 if ($r_max > $c->{r});
5658                }
5659        }
5660        return 0 if (defined $limit && --$limit < 0);
5661        show_commit($c);
5662        return 1;
5663}
5664
5665sub show_commit {
5666        my $c = shift;
5667        if ($oneline) {
5668                my $x = "\n";
5669                if (my $l = $c->{l}) {
5670                        while ($l->[0] =~ /^\s*$/) { shift @$l }
5671                        $x = $l->[0];
5672                }
5673                $l_fmt ||= 'A' . length($c->{r});
5674                print 'r',pack($l_fmt, $c->{r}),' | ';
5675                print "$c->{c} | " if $show_commit;
5676                print $x;
5677        } else {
5678                show_commit_normal($c);
5679        }
5680}
5681
5682sub show_commit_changed_paths {
5683        my ($c) = @_;
5684        return unless $c->{changed};
5685        print "Changed paths:\n", @{$c->{changed}};
5686}
5687
5688sub show_commit_normal {
5689        my ($c) = @_;
5690        print commit_log_separator, "r$c->{r} | ";
5691        print "$c->{c} | " if $show_commit;
5692        print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5693        my $nr_line = 0;
5694
5695        if (my $l = $c->{l}) {
5696                while ($l->[$#$l] eq "\n" && $#$l > 0
5697                                          && $l->[($#$l - 1)] eq "\n") {
5698                        pop @$l;
5699                }
5700                $nr_line = scalar @$l;
5701                if (!$nr_line) {
5702                        print "1 line\n\n\n";
5703                } else {
5704                        if ($nr_line == 1) {
5705                                $nr_line = '1 line';
5706                        } else {
5707                                $nr_line .= ' lines';
5708                        }
5709                        print $nr_line, "\n";
5710                        show_commit_changed_paths($c);
5711                        print "\n";
5712                        print $_ foreach @$l;
5713                }
5714        } else {
5715                print "1 line\n";
5716                show_commit_changed_paths($c);
5717                print "\n";
5718
5719        }
5720        foreach my $x (qw/raw stat diff/) {
5721                if ($c->{$x}) {
5722                        print "\n";
5723                        print $_ foreach @{$c->{$x}}
5724                }
5725        }
5726}
5727
5728sub cmd_show_log {
5729        my (@args) = @_;
5730        my ($r_min, $r_max);
5731        my $r_last = -1; # prevent dupes
5732        set_local_timezone();
5733        if (defined $::_revision) {
5734                if ($::_revision =~ /^(\d+):(\d+)$/) {
5735                        ($r_min, $r_max) = ($1, $2);
5736                } elsif ($::_revision =~ /^\d+$/) {
5737                        $r_min = $r_max = $::_revision;
5738                } else {
5739                        ::fatal "-r$::_revision is not supported, use ",
5740                                "standard 'git log' arguments instead";
5741                }
5742        }
5743
5744        config_pager();
5745        @args = git_svn_log_cmd($r_min, $r_max, @args);
5746        if (!@args) {
5747                print commit_log_separator unless $incremental || $oneline;
5748                return;
5749        }
5750        my $log = command_output_pipe(@args);
5751        run_pager();
5752        my (@k, $c, $d, $stat);
5753        my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5754        while (<$log>) {
5755                if (/^${esc_color}commit (?:- )?($::sha1_short)/o) {
5756                        my $cmt = $1;
5757                        if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5758                                $r_last = $c->{r};
5759                                process_commit($c, $r_min, $r_max, \@k) or
5760                                                                goto out;
5761                        }
5762                        $d = undef;
5763                        $c = { c => $cmt };
5764                } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5765                        get_author_info($c, $1, $2, $3);
5766                } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5767                        # ignore
5768                } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5769                        push @{$c->{raw}}, $_;
5770                } elsif (/^${esc_color}[ACRMDT]\t/) {
5771                        # we could add $SVN->{svn_path} here, but that requires
5772                        # remote access at the moment (repo_path_split)...
5773                        s#^(${esc_color})([ACRMDT])\t#$1   $2 #o;
5774                        push @{$c->{changed}}, $_;
5775                } elsif (/^${esc_color}diff /o) {
5776                        $d = 1;
5777                        push @{$c->{diff}}, $_;
5778                } elsif ($d) {
5779                        push @{$c->{diff}}, $_;
5780                } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5781                          $esc_color*[\+\-]*$esc_color$/x) {
5782                        $stat = 1;
5783                        push @{$c->{stat}}, $_;
5784                } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5785                        push @{$c->{stat}}, $_;
5786                        $stat = undef;
5787                } elsif (/^${esc_color}    (git-svn-id:.+)$/o) {
5788                        ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5789                } elsif (s/^${esc_color}    //o) {
5790                        push @{$c->{l}}, $_;
5791                }
5792        }
5793        if ($c && defined $c->{r} && $c->{r} != $r_last) {
5794                $r_last = $c->{r};
5795                process_commit($c, $r_min, $r_max, \@k);
5796        }
5797        if (@k) {
5798                ($r_min, $r_max) = ($r_max, $r_min);
5799                process_commit($_, $r_min, $r_max) foreach reverse @k;
5800        }
5801out:
5802        close $log;
5803        print commit_log_separator unless $incremental || $oneline;
5804}
5805
5806sub cmd_blame {
5807        my $path = pop;
5808
5809        config_pager();
5810        run_pager();
5811
5812        my ($fh, $ctx, $rev);
5813
5814        if ($_git_format) {
5815                ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5816                while (my $line = <$fh>) {
5817                        if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5818                                # Uncommitted edits show up as a rev ID of
5819                                # all zeros, which we can't look up with
5820                                # cmt_metadata
5821                                if ($1 !~ /^0+$/) {
5822                                        (undef, $rev, undef) =
5823                                                ::cmt_metadata($1);
5824                                        $rev = '0' if (!$rev);
5825                                } else {
5826                                        $rev = '0';
5827                                }
5828                                $rev = sprintf('%-10s', $rev);
5829                                $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5830                        }
5831                        print $line;
5832                }
5833        } else {
5834                ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5835                                                  '--', $path);
5836                my ($sha1);
5837                my %authors;
5838                my @buffer;
5839                my %dsha; #distinct sha keys
5840
5841                while (my $line = <$fh>) {
5842                        push @buffer, $line;
5843                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5844                                $dsha{$1} = 1;
5845                        }
5846                }
5847
5848                my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5849
5850                foreach my $line (@buffer) {
5851                        if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5852                                $rev = $s2r->{$1};
5853                                $rev = '0' if (!$rev)
5854                        }
5855                        elsif ($line =~ /^author (.*)/) {
5856                                $authors{$rev} = $1;
5857                                $authors{$rev} =~ s/\s/_/g;
5858                        }
5859                        elsif ($line =~ /^\t(.*)$/) {
5860                                printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5861                        }
5862                }
5863        }
5864        command_close_pipe($fh, $ctx);
5865}
5866
5867package Git::SVN::Migration;
5868# these version numbers do NOT correspond to actual version numbers
5869# of git nor git-svn.  They are just relative.
5870#
5871# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5872#
5873# v1 layout: .git/$id/info/url, refs/remotes/$id
5874#
5875# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5876#
5877# v3 layout: .git/svn/$id, refs/remotes/$id
5878#            - info/url may remain for backwards compatibility
5879#            - this is what we migrate up to this layout automatically,
5880#            - this will be used by git svn init on single branches
5881# v3.1 layout (auto migrated):
5882#            - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5883#              for backwards compatibility
5884#
5885# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5886#            - this is only created for newly multi-init-ed
5887#              repositories.  Similar in spirit to the
5888#              --use-separate-remotes option in git-clone (now default)
5889#            - we do not automatically migrate to this (following
5890#              the example set by core git)
5891#
5892# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5893#            - newer, more-efficient format that uses 24-bytes per record
5894#              with no filler space.
5895#            - use xxd -c24 < .rev_map.$UUID to view and debug
5896#            - This is a one-way migration, repositories updated to the
5897#              new format will not be able to use old git-svn without
5898#              rebuilding the .rev_db.  Rebuilding the rev_db is not
5899#              possible if noMetadata or useSvmProps are set; but should
5900#              be no problem for users that use the (sensible) defaults.
5901use strict;
5902use warnings;
5903use Carp qw/croak/;
5904use File::Path qw/mkpath/;
5905use File::Basename qw/dirname basename/;
5906use vars qw/$_minimize/;
5907
5908sub migrate_from_v0 {
5909        my $git_dir = $ENV{GIT_DIR};
5910        return undef unless -d $git_dir;
5911        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5912        my $migrated = 0;
5913        while (<$fh>) {
5914                chomp;
5915                my ($id, $orig_ref) = ($_, $_);
5916                next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5917                next unless -f "$git_dir/$id/info/url";
5918                my $new_ref = "refs/remotes/$id";
5919                if (::verify_ref("$new_ref^0")) {
5920                        print STDERR "W: $orig_ref is probably an old ",
5921                                     "branch used by an ancient version of ",
5922                                     "git-svn.\n",
5923                                     "However, $new_ref also exists.\n",
5924                                     "We will not be able ",
5925                                     "to use this branch until this ",
5926                                     "ambiguity is resolved.\n";
5927                        next;
5928                }
5929                print STDERR "Migrating from v0 layout...\n" if !$migrated;
5930                print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5931                command_noisy('update-ref', $new_ref, $orig_ref);
5932                command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5933                $migrated++;
5934        }
5935        command_close_pipe($fh, $ctx);
5936        print STDERR "Done migrating from v0 layout...\n" if $migrated;
5937        $migrated;
5938}
5939
5940sub migrate_from_v1 {
5941        my $git_dir = $ENV{GIT_DIR};
5942        my $migrated = 0;
5943        return $migrated unless -d $git_dir;
5944        my $svn_dir = "$git_dir/svn";
5945
5946        # just in case somebody used 'svn' as their $id at some point...
5947        return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5948
5949        print STDERR "Migrating from a git-svn v1 layout...\n";
5950        mkpath([$svn_dir]);
5951        print STDERR "Data from a previous version of git-svn exists, but\n\t",
5952                     "$svn_dir\n\t(required for this version ",
5953                     "($::VERSION) of git-svn) does not exist.\n";
5954        my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5955        while (<$fh>) {
5956                my $x = $_;
5957                next unless $x =~ s#^refs/remotes/##;
5958                chomp $x;
5959                next unless -f "$git_dir/$x/info/url";
5960                my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5961                next unless $u;
5962                my $dn = dirname("$git_dir/svn/$x");
5963                mkpath([$dn]) unless -d $dn;
5964                if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5965                        mkpath(["$git_dir/svn/svn"]);
5966                        print STDERR " - $git_dir/$x/info => ",
5967                                        "$git_dir/svn/$x/info\n";
5968                        rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5969                               croak "$!: $x";
5970                        # don't worry too much about these, they probably
5971                        # don't exist with repos this old (save for index,
5972                        # and we can easily regenerate that)
5973                        foreach my $f (qw/unhandled.log index .rev_db/) {
5974                                rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5975                        }
5976                } else {
5977                        print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5978                        rename "$git_dir/$x", "$git_dir/svn/$x" or
5979                               croak "$!: $x";
5980                }
5981                $migrated++;
5982        }
5983        command_close_pipe($fh, $ctx);
5984        print STDERR "Done migrating from a git-svn v1 layout\n";
5985        $migrated;
5986}
5987
5988sub read_old_urls {
5989        my ($l_map, $pfx, $path) = @_;
5990        my @dir;
5991        foreach (<$path/*>) {
5992                if (-r "$_/info/url") {
5993                        $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5994                        my $ref_id = $pfx . basename $_;
5995                        my $url = ::file_to_s("$_/info/url");
5996                        $l_map->{$ref_id} = $url;
5997                } elsif (-d $_) {
5998                        push @dir, $_;
5999                }
6000        }
6001        foreach (@dir) {
6002                my $x = $_;
6003                $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
6004                read_old_urls($l_map, $x, $_);
6005        }
6006}
6007
6008sub migrate_from_v2 {
6009        my @cfg = command(qw/config -l/);
6010        return if grep /^svn-remote\..+\.url=/, @cfg;
6011        my %l_map;
6012        read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
6013        my $migrated = 0;
6014
6015        foreach my $ref_id (sort keys %l_map) {
6016                eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
6017                if ($@) {
6018                        Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
6019                }
6020                $migrated++;
6021        }
6022        $migrated;
6023}
6024
6025sub minimize_connections {
6026        my $r = Git::SVN::read_all_remotes();
6027        my $new_urls = {};
6028        my $root_repos = {};
6029        foreach my $repo_id (keys %$r) {
6030                my $url = $r->{$repo_id}->{url} or next;
6031                my $fetch = $r->{$repo_id}->{fetch} or next;
6032                my $ra = Git::SVN::Ra->new($url);
6033
6034                # skip existing cases where we already connect to the root
6035                if (($ra->{url} eq $ra->{repos_root}) ||
6036                    ($ra->{repos_root} eq $repo_id)) {
6037                        $root_repos->{$ra->{url}} = $repo_id;
6038                        next;
6039                }
6040
6041                my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
6042                my $root_path = $ra->{url};
6043                $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
6044                foreach my $path (keys %$fetch) {
6045                        my $ref_id = $fetch->{$path};
6046                        my $gs = Git::SVN->new($ref_id, $repo_id, $path);
6047
6048                        # make sure we can read when connecting to
6049                        # a higher level of a repository
6050                        my ($last_rev, undef) = $gs->last_rev_commit;
6051                        if (!defined $last_rev) {
6052                                $last_rev = eval {
6053                                        $root_ra->get_latest_revnum;
6054                                };
6055                                next if $@;
6056                        }
6057                        my $new = $root_path;
6058                        $new .= length $path ? "/$path" : '';
6059                        eval {
6060                                $root_ra->get_log([$new], $last_rev, $last_rev,
6061                                                  0, 0, 1, sub { });
6062                        };
6063                        next if $@;
6064                        $new_urls->{$ra->{repos_root}}->{$new} =
6065                                { ref_id => $ref_id,
6066                                  old_repo_id => $repo_id,
6067                                  old_path => $path };
6068                }
6069        }
6070
6071        my @emptied;
6072        foreach my $url (keys %$new_urls) {
6073                # see if we can re-use an existing [svn-remote "repo_id"]
6074                # instead of creating a(n ugly) new section:
6075                my $repo_id = $root_repos->{$url} || $url;
6076
6077                my $fetch = $new_urls->{$url};
6078                foreach my $path (keys %$fetch) {
6079                        my $x = $fetch->{$path};
6080                        Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
6081                        my $pfx = "svn-remote.$x->{old_repo_id}";
6082
6083                        my $old_fetch = quotemeta("$x->{old_path}:".
6084                                                  "$x->{ref_id}");
6085                        command_noisy(qw/config --unset/,
6086                                      "$pfx.fetch", '^'. $old_fetch . '$');
6087                        delete $r->{$x->{old_repo_id}}->
6088                               {fetch}->{$x->{old_path}};
6089                        if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6090                                command_noisy(qw/config --unset/,
6091                                              "$pfx.url");
6092                                push @emptied, $x->{old_repo_id}
6093                        }
6094                }
6095        }
6096        if (@emptied) {
6097                my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6098                print STDERR <<EOF;
6099The following [svn-remote] sections in your config file ($file) are empty
6100and can be safely removed:
6101EOF
6102                print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6103        }
6104}
6105
6106sub migration_check {
6107        migrate_from_v0();
6108        migrate_from_v1();
6109        migrate_from_v2();
6110        minimize_connections() if $_minimize;
6111}
6112
6113package Git::IndexInfo;
6114use strict;
6115use warnings;
6116use Git qw/command_input_pipe command_close_pipe/;
6117
6118sub new {
6119        my ($class) = @_;
6120        my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6121        bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6122}
6123
6124sub remove {
6125        my ($self, $path) = @_;
6126        if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6127                return ++$self->{nr};
6128        }
6129        undef;
6130}
6131
6132sub update {
6133        my ($self, $mode, $hash, $path) = @_;
6134        if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6135                return ++$self->{nr};
6136        }
6137        undef;
6138}
6139
6140sub DESTROY {
6141        my ($self) = @_;
6142        command_close_pipe($self->{gui}, $self->{ctx});
6143}
6144
6145package Git::SVN::GlobSpec;
6146use strict;
6147use warnings;
6148
6149sub new {
6150        my ($class, $glob, $pattern_ok) = @_;
6151        my $re = $glob;
6152        $re =~ s!/+$!!g; # no need for trailing slashes
6153        my (@left, @right, @patterns);
6154        my $state = "left";
6155        my $die_msg = "Only one set of wildcard directories " .
6156                                "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6157        for my $part (split(m|/|, $glob)) {
6158                if ($part =~ /\*/ && $part ne "*") {
6159                        die "Invalid pattern in '$glob': $part\n";
6160                } elsif ($pattern_ok && $part =~ /[{}]/ &&
6161                         $part !~ /^\{[^{}]+\}/) {
6162                        die "Invalid pattern in '$glob': $part\n";
6163                }
6164                if ($part eq "*") {
6165                        die $die_msg if $state eq "right";
6166                        $state = "pattern";
6167                        push(@patterns, "[^/]*");
6168                } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6169                        die $die_msg if $state eq "right";
6170                        $state = "pattern";
6171                        my $p = quotemeta($1);
6172                        $p =~ s/\\,/|/g;
6173                        push(@patterns, "(?:$p)");
6174                } else {
6175                        if ($state eq "left") {
6176                                push(@left, $part);
6177                        } else {
6178                                push(@right, $part);
6179                                $state = "right";
6180                        }
6181                }
6182        }
6183        my $depth = @patterns;
6184        if ($depth == 0) {
6185                die "One '*' is needed in glob: '$glob'\n";
6186        }
6187        my $left = join('/', @left);
6188        my $right = join('/', @right);
6189        $re = join('/', @patterns);
6190        $re = join('\/',
6191                   grep(length, quotemeta($left), "($re)", quotemeta($right)));
6192        my $left_re = qr/^\/\Q$left\E(\/|$)/;
6193        bless { left => $left, right => $right, left_regex => $left_re,
6194                regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6195}
6196
6197sub full_path {
6198        my ($self, $path) = @_;
6199        return (length $self->{left} ? "$self->{left}/" : '') .
6200               $path . (length $self->{right} ? "/$self->{right}" : '');
6201}
6202
6203__END__
6204
6205Data structures:
6206
6207
6208$remotes = { # returned by read_all_remotes()
6209        'svn' => {
6210                # svn-remote.svn.url=https://svn.musicpd.org
6211                url => 'https://svn.musicpd.org',
6212                # svn-remote.svn.fetch=mpd/trunk:trunk
6213                fetch => {
6214                        'mpd/trunk' => 'trunk',
6215                },
6216                # svn-remote.svn.tags=mpd/tags/*:tags/*
6217                tags => {
6218                        path => {
6219                                left => 'mpd/tags',
6220                                right => '',
6221                                regex => qr!mpd/tags/([^/]+)$!,
6222                                glob => 'tags/*',
6223                        },
6224                        ref => {
6225                                left => 'tags',
6226                                right => '',
6227                                regex => qr!tags/([^/]+)$!,
6228                                glob => 'tags/*',
6229                        },
6230                }
6231        }
6232};
6233
6234$log_entry hashref as returned by libsvn_log_entry()
6235{
6236        log => 'whitespace-formatted log entry
6237',                                              # trailing newline is preserved
6238        revision => '8',                        # integer
6239        date => '2004-02-24T17:01:44.108345Z',  # commit date
6240        author => 'committer name'
6241};
6242
6243
6244# this is generated by generate_diff();
6245@mods = array of diff-index line hashes, each element represents one line
6246        of diff-index output
6247
6248diff-index line ($m hash)
6249{
6250        mode_a => first column of diff-index output, no leading ':',
6251        mode_b => second column of diff-index output,
6252        sha1_b => sha1sum of the final blob,
6253        chg => change type [MCRADT],
6254        file_a => original file name of a file (iff chg is 'C' or 'R')
6255        file_b => new/current file name of a file (any chg)
6256}
6257;
6258
6259# retval of read_url_paths{,_all}();
6260$l_map = {
6261        # repository root url
6262        'https://svn.musicpd.org' => {
6263                # repository path               # GIT_SVN_ID
6264                'mpd/trunk'             =>      'trunk',
6265                'mpd/tags/0.11.5'       =>      'tags/0.11.5',
6266        },
6267}
6268
6269Notes:
6270        I don't trust the each() function on unless I created %hash myself
6271        because the internal iterator may not have started at base.