819d25e28937488e4e7432c5b262474eaa821580
1#!/usr/bin/env perl
2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3# License: GPL v2 or later
4use warnings;
5use strict;
6use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision
8 $_q $_authors %users/;
9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10$VERSION = '@@GIT_VERSION@@';
11$ENV{GIT_DIR} ||= '.git';
13$Git::SVN::default_repo_id = 'svn';
14$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
15$Git::SVN::Log::TZ = $ENV{TZ};
17$ENV{TZ} = 'UTC';
18$| = 1; # unbuffer STDOUT
19sub fatal (@) { print STDERR @_; exit 1 }
21require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
22require SVN::Ra;
23require SVN::Delta;
24if ($SVN::Core::VERSION lt '1.1.0') {
25 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
26}
27push @Git::SVN::Ra::ISA, 'SVN::Ra';
28push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
29push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
30use Carp qw/croak/;
31use IO::File qw//;
32use File::Basename qw/dirname basename/;
33use File::Path qw/mkpath/;
34use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
35use IPC::Open3;
36use Git;
37BEGIN {
39 my $s;
40 foreach (qw/command command_oneline command_noisy command_output_pipe
41 command_input_pipe command_close_pipe/) {
42 $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
43 "*Git::SVN::Migration::$_ = ".
44 "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
45 }
46 eval $s;
47}
48my ($SVN);
50my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
52$sha1 = qr/[a-f\d]{40}/;
53$sha1_short = qr/[a-f\d]{4,40}/;
54my ($_stdin, $_help, $_edit,
55 $_message, $_file,
56 $_template, $_shared,
57 $_version,
58 $_merge, $_strategy, $_dry_run,
59 $_prefix);
60my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
62 'config-dir=s' => \$Git::SVN::Ra::config_dir,
63 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
64my %fc_opts = ( 'follow-parent|follow' => \$Git::SVN::_follow_parent,
65 'authors-file|A=s' => \$_authors,
66 'repack:i' => \$Git::SVN::_repack,
67 'no-metadata' => \$Git::SVN::_no_metadata,
68 'quiet|q' => \$_q,
69 'repack-flags|repack-args|repack-opts=s' =>
70 \$Git::SVN::_repack_flags,
71 %remote_opts );
72my ($_trunk, $_tags, $_branches);
74my %multi_opts = ( 'trunk|T=s' => \$_trunk,
75 'tags|t=s' => \$_tags,
76 'branches|b=s' => \$_branches );
77my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
78my %cmt_opts = ( 'edit|e' => \$_edit,
79 'rmdir' => \$SVN::Git::Editor::_rmdir,
80 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
81 'l=i' => \$SVN::Git::Editor::_rename_limit,
82 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
83);
84my %cmd = (
86 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
87 { 'revision|r=s' => \$_revision, %fc_opts } ],
88 init => [ \&cmd_init, "Initialize a repo for tracking" .
89 " (requires URL argument)",
90 \%init_opts ],
91 dcommit => [ \&cmd_dcommit,
92 'Commit several diffs to merge with upstream',
93 { 'merge|m|M' => \$_merge,
94 'strategy|s=s' => \$_strategy,
95 'dry-run|n' => \$_dry_run,
96 %cmt_opts, %fc_opts } ],
97 'set-tree' => [ \&cmd_set_tree,
98 "Set an SVN repository to a git tree-ish",
99 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
100 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
101 { 'revision|r=i' => \$_revision } ],
102 'multi-init' => [ \&cmd_multi_init,
103 'Initialize multiple trees (like git-svnimport)',
104 { %multi_opts, %init_opts, %remote_opts,
105 'revision|r=i' => \$_revision,
106 'prefix=s' => \$_prefix,
107 } ],
108 'multi-fetch' => [ \&cmd_multi_fetch,
109 'Fetch multiple trees (like git-svnimport)',
110 \%fc_opts ],
111 'migrate' => [ sub { },
112 # no-op, we automatically run this anyways,
113 'Migrate configuration/metadata/layout from
114 previous versions of git-svn',
115 \%remote_opts ],
116 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
117 { 'limit=i' => \$Git::SVN::Log::limit,
118 'revision|r=s' => \$_revision,
119 'verbose|v' => \$Git::SVN::Log::verbose,
120 'incremental' => \$Git::SVN::Log::incremental,
121 'oneline' => \$Git::SVN::Log::oneline,
122 'show-commit' => \$Git::SVN::Log::show_commit,
123 'non-recursive' => \$Git::SVN::Log::non_recursive,
124 'authors-file|A=s' => \$_authors,
125 'color' => \$Git::SVN::Log::color,
126 'pager=s' => \$Git::SVN::Log::pager,
127 } ],
128 'commit-diff' => [ \&cmd_commit_diff,
129 'Commit a diff between two trees',
130 { 'message|m=s' => \$_message,
131 'file|F=s' => \$_file,
132 'revision|r=s' => \$_revision,
133 %cmt_opts } ],
134);
135my $cmd;
137for (my $i = 0; $i < @ARGV; $i++) {
138 if (defined $cmd{$ARGV[$i]}) {
139 $cmd = $ARGV[$i];
140 splice @ARGV, $i, 1;
141 last;
142 }
143};
144my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
146read_repo_config(\%opts);
148my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
149 'minimize-connections' => \$Git::SVN::Migration::_minimize,
150 'id|i=s' => \$Git::SVN::default_ref_id,
151 'svn-remote|remote|R=s' => \$Git::SVN::default_repo_id);
152exit 1 if (!$rv && $cmd ne 'log');
153usage(0) if $_help;
155version() if $_version;
156usage(1) unless defined $cmd;
157load_authors() if $_authors;
158unless ($cmd =~ /^(?:init|multi-init|commit-diff)$/) {
159 Git::SVN::Migration::migration_check();
160}
161Git::SVN::init_vars();
162eval {
163 Git::SVN::verify_remotes_sanity();
164 $cmd{$cmd}->[0]->(@ARGV);
165};
166fatal $@ if $@;
167exit 0;
168####################### primary functions ######################
170sub usage {
171 my $exit = shift || 0;
172 my $fd = $exit ? \*STDERR : \*STDOUT;
173 print $fd <<"";
174git-svn - bidirectional operations between a single Subversion tree and git
175Usage: $0 <command> [options] [arguments]\n
176 print $fd "Available commands:\n" unless $cmd;
178 foreach (sort keys %cmd) {
180 next if $cmd && $cmd ne $_;
181 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
182 foreach (keys %{$cmd{$_}->[2]}) {
183 # prints out arguments as they should be passed:
184 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
185 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
186 "--$_" : "-$_" }
187 split /\|/,$_)," $x\n";
188 }
189 }
190 print $fd <<"";
191\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
192arbitrary identifier if you're tracking multiple SVN branches/repositories in
193one git repository and want to keep them separate. See git-svn(1) for more
194information.
195 exit $exit;
197}
198sub version {
200 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
201 exit 0;
202}
203sub do_git_init_db {
205 unless (-d $ENV{GIT_DIR}) {
206 my @init_db = ('init');
207 push @init_db, "--template=$_template" if defined $_template;
208 push @init_db, "--shared" if defined $_shared;
209 command_noisy(@init_db);
210 }
211}
212sub cmd_init {
214 my $url = shift or die "SVN repository location required " .
215 "as a command-line argument\n";
216 if (my $repo_path = shift) {
217 unless (-d $repo_path) {
218 mkpath([$repo_path]);
219 }
220 chdir $repo_path or croak $!;
221 $ENV{GIT_DIR} = $repo_path . "/.git";
222 }
223 do_git_init_db();
224 Git::SVN->init($url);
226}
227sub cmd_fetch {
229 if (@_) {
230 die "Additional fetch arguments are no longer supported.\n",
231 "Use --follow-parent if you have moved/copied directories
232 instead.\n";
233 }
234 my $gs = Git::SVN->new;
235 $gs->fetch(parse_revision_argument());
236 if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
237 command_noisy(qw(update-ref refs/heads/master),
238 $gs->{last_commit});
239 }
240}
241sub cmd_set_tree {
243 my (@commits) = @_;
244 if ($_stdin || !@commits) {
245 print "Reading from stdin...\n";
246 @commits = ();
247 while (<STDIN>) {
248 if (/\b($sha1_short)\b/o) {
249 unshift @commits, $1;
250 }
251 }
252 }
253 my @revs;
254 foreach my $c (@commits) {
255 my @tmp = command('rev-parse',$c);
256 if (scalar @tmp == 1) {
257 push @revs, $tmp[0];
258 } elsif (scalar @tmp > 1) {
259 push @revs, reverse(command('rev-list',@tmp));
260 } else {
261 fatal "Failed to rev-parse $c\n";
262 }
263 }
264 my $gs = Git::SVN->new;
265 my ($r_last, $cmt_last) = $gs->last_rev_commit;
266 $gs->fetch;
267 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
268 fatal "There are new revisions that were fetched ",
269 "and need to be merged (or acknowledged) ",
270 "before committing.\nlast rev: $r_last\n",
271 " current: $gs->{last_rev}\n";
272 }
273 $gs->set_tree($_) foreach @revs;
274 print "Done committing ",scalar @revs," revisions to SVN\n";
275}
276sub cmd_dcommit {
278 my $head = shift;
279 my $gs = Git::SVN->new;
280 $head ||= 'HEAD';
281 my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
282 my $last_rev;
283 foreach my $d (reverse @refs) {
284 if (!verify_ref("$d~1")) {
285 fatal "Commit $d\n",
286 "has no parent commit, and therefore ",
287 "nothing to diff against.\n",
288 "You should be working from a repository ",
289 "originally created by git-svn\n";
290 }
291 unless (defined $last_rev) {
292 (undef, $last_rev, undef) = cmt_metadata("$d~1");
293 unless (defined $last_rev) {
294 fatal "Unable to extract revision information ",
295 "from commit $d~1\n";
296 }
297 }
298 if ($_dry_run) {
299 print "diff-tree $d~1 $d\n";
300 } else {
301 my %ed_opts = ( r => $last_rev,
302 log => get_commit_entry($d)->{log},
303 ra => $gs->ra,
304 tree_a => "$d~1",
305 tree_b => $d,
306 editor_cb => sub {
307 print "Committed r$_[0]\n";
308 $last_rev = $_[0]; },
309 svn_path => $gs->{path} );
310 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
311 print "No changes\n$d~1 == $d\n";
312 }
313 }
314 }
315 return if $_dry_run;
316 $gs->fetch;
317 # we always want to rebase against the current HEAD, not any
318 # head that was passed to us
319 my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
320 my @finish;
321 if (@diff) {
322 @finish = qw/rebase/;
323 push @finish, qw/--merge/ if $_merge;
324 push @finish, "--strategy=$_strategy" if $_strategy;
325 print STDERR "W: HEAD and ", $gs->refname, " differ, ",
326 "using @finish:\n", "@diff";
327 } else {
328 print "No changes between current HEAD and ",
329 $gs->refname, "\nResetting to the latest ",
330 $gs->refname, "\n";
331 @finish = qw/reset --mixed/;
332 }
333 command_noisy(@finish, $gs->refname);
334}
335sub cmd_show_ignore {
337 my $gs = Git::SVN->new;
338 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
339 $gs->traverse_ignore(\*STDOUT, '', $r);
340}
341sub cmd_multi_init {
343 my $url = shift;
344 unless (defined $_trunk || defined $_branches || defined $_tags) {
345 usage(1);
346 }
347 do_git_init_db();
348 $_prefix = '' unless defined $_prefix;
349 $url =~ s#/+$## if defined $url;
350 if (defined $_trunk) {
351 my $trunk_ref = $_prefix . 'trunk';
352 # try both old-style and new-style lookups:
353 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
354 unless ($gs_trunk) {
355 my ($trunk_url, $trunk_path) =
356 complete_svn_url($url, $_trunk);
357 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
358 undef, $trunk_ref);
359 }
360 }
361 return unless defined $_branches || defined $_tags;
362 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
363 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
364 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
365}
366sub cmd_multi_fetch {
368 my $remotes = Git::SVN::read_all_remotes();
369 foreach my $repo_id (sort keys %$remotes) {
370 if ($remotes->{$repo_id}->{url} &&
371 $remotes->{$repo_id}->{fetch}) {
372 Git::SVN::fetch_all($repo_id, $remotes);
373 }
374 }
375}
376# this command is special because it requires no metadata
378sub cmd_commit_diff {
379 my ($ta, $tb, $url) = @_;
380 my $usage = "Usage: $0 commit-diff -r<revision> ".
381 "<tree-ish> <tree-ish> [<URL>]\n";
382 fatal($usage) if (!defined $ta || !defined $tb);
383 my $svn_path;
384 if (!defined $url) {
385 my $gs = eval { Git::SVN->new };
386 if (!$gs) {
387 fatal("Needed URL or usable git-svn --id in ",
388 "the command-line\n", $usage);
389 }
390 $url = $gs->{url};
391 $svn_path = $gs->{path};
392 }
393 unless (defined $_revision) {
394 fatal("-r|--revision is a required argument\n", $usage);
395 }
396 if (defined $_message && defined $_file) {
397 fatal("Both --message/-m and --file/-F specified ",
398 "for the commit message.\n",
399 "I have no idea what you mean\n");
400 }
401 if (defined $_file) {
402 $_message = file_to_s($_file);
403 } else {
404 $_message ||= get_commit_entry($tb)->{log};
405 }
406 my $ra ||= Git::SVN::Ra->new($url);
407 $svn_path ||= $ra->{svn_path};
408 my $r = $_revision;
409 if ($r eq 'HEAD') {
410 $r = $ra->get_latest_revnum;
411 } elsif ($r !~ /^\d+$/) {
412 die "revision argument: $r not understood by git-svn\n";
413 }
414 my %ed_opts = ( r => $r,
415 log => $_message,
416 ra => $ra,
417 tree_a => $ta,
418 tree_b => $tb,
419 editor_cb => sub { print "Committed r$_[0]\n" },
420 svn_path => $svn_path );
421 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
422 print "No changes\n$ta == $tb\n";
423 }
424}
425########################### utility functions #########################
427sub parse_revision_argument {
429 if (!defined $_revision || $_revision eq 'BASE:HEAD') {
430 return (undef, undef);
431 }
432 return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
433 return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
434 return (undef, $1) if ($_revision =~ /^BASE:(\d+)$/);
435 return ($1, undef) if ($_revision =~ /^(\d+):HEAD$/);
436 die "revision argument: $_revision not understood by git-svn\n",
437 "Try using the command-line svn client instead\n";
438}
439sub complete_svn_url {
441 my ($url, $path) = @_;
442 $path =~ s#/+$##;
443 if ($path !~ m#^[a-z\+]+://#) {
444 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
445 fatal("E: '$path' is not a complete URL ",
446 "and a separate URL is not specified\n");
447 }
448 return ($url, $path);
449 }
450 return ($path, '');
451}
452sub complete_url_ls_init {
454 my ($ra, $repo_path, $switch, $pfx) = @_;
455 unless ($repo_path) {
456 print STDERR "W: $switch not specified\n";
457 return;
458 }
459 $repo_path =~ s#/+$##;
460 if ($repo_path =~ m#^[a-z\+]+://#) {
461 $ra = Git::SVN::Ra->new($repo_path);
462 $repo_path = '';
463 } else {
464 $repo_path =~ s#^/+##;
465 unless ($ra) {
466 fatal("E: '$repo_path' is not a complete URL ",
467 "and a separate URL is not specified\n");
468 }
469 }
470 my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
471 my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
472 my $url = $ra->{url};
473 my $remote_id;
474 my $remote_path;
475 foreach my $d (sort keys %$dirent) {
476 next if ($dirent->{$d}->kind != $SVN::Node::dir);
477 my $path = "$repo_path/$d";
478 my $ref = "$pfx$d";
479 my $gs = eval { Git::SVN->new($ref) };
480 # don't try to init already existing refs
481 unless ($gs) {
482 print "init $url/$path => $ref\n";
483 $gs = Git::SVN->init($url, $path, undef, $ref, 1);
484 }
485 if ($gs) {
486 $remote_id = $gs->{repo_id};
487 last;
488 }
489 }
490 if (defined $remote_id) {
491 $remote_path = "$ra->{svn_path}/$repo_path/*";
492 $remote_path =~ s#/+#/#g;
493 $remote_path =~ s#^/##g;
494 my ($n) = ($switch =~ /^--(\w+)/);
495 command_noisy('config', "svn-remote.$remote_id.$n",
496 "$remote_path:refs/remotes/$pfx*");
497 }
498}
499sub verify_ref {
501 my ($ref) = @_;
502 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
503 { STDERR => 0 }); };
504}
505sub get_tree_from_treeish {
507 my ($treeish) = @_;
508 # $treeish can be a symbolic ref, too:
509 my $type = command_oneline(qw/cat-file -t/, $treeish);
510 my $expected;
511 while ($type eq 'tag') {
512 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
513 }
514 if ($type eq 'commit') {
515 $expected = (grep /^tree /, command(qw/cat-file commit/,
516 $treeish))[0];
517 ($expected) = ($expected =~ /^tree ($sha1)$/o);
518 die "Unable to get tree from $treeish\n" unless $expected;
519 } elsif ($type eq 'tree') {
520 $expected = $treeish;
521 } else {
522 die "$treeish is a $type, expected tree, tag or commit\n";
523 }
524 return $expected;
525}
526sub get_commit_entry {
528 my ($treeish) = shift;
529 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
530 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
531 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
532 open my $log_fh, '>', $commit_editmsg or croak $!;
533 my $type = command_oneline(qw/cat-file -t/, $treeish);
535 if ($type eq 'commit' || $type eq 'tag') {
536 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
537 $type, $treeish);
538 my $in_msg = 0;
539 while (<$msg_fh>) {
540 if (!$in_msg) {
541 $in_msg = 1 if (/^\s*$/);
542 } elsif (/^git-svn-id: /) {
543 # skip this for now, we regenerate the
544 # correct one on re-fetch anyways
545 # TODO: set *:merge properties or like...
546 } else {
547 print $log_fh $_ or croak $!;
548 }
549 }
550 command_close_pipe($msg_fh, $ctx);
551 }
552 close $log_fh or croak $!;
553 if ($_edit || ($type eq 'tree')) {
555 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
556 # TODO: strip out spaces, comments, like git-commit.sh
557 system($editor, $commit_editmsg);
558 }
559 rename $commit_editmsg, $commit_msg or croak $!;
560 open $log_fh, '<', $commit_msg or croak $!;
561 { local $/; chomp($log_entry{log} = <$log_fh>); }
562 close $log_fh or croak $!;
563 unlink $commit_msg;
564 \%log_entry;
565}
566sub s_to_file {
568 my ($str, $file, $mode) = @_;
569 open my $fd,'>',$file or croak $!;
570 print $fd $str,"\n" or croak $!;
571 close $fd or croak $!;
572 chmod ($mode &~ umask, $file) if (defined $mode);
573}
574sub file_to_s {
576 my $file = shift;
577 open my $fd,'<',$file or croak "$!: file: $file\n";
578 local $/;
579 my $ret = <$fd>;
580 close $fd or croak $!;
581 $ret =~ s/\s*$//s;
582 return $ret;
583}
584# '<svn username> = real-name <email address>' mapping based on git-svnimport:
586sub load_authors {
587 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
588 my $log = $cmd eq 'log';
589 while (<$authors>) {
590 chomp;
591 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
592 my ($user, $name, $email) = ($1, $2, $3);
593 if ($log) {
594 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
595 } else {
596 $users{$user} = [$name, $email];
597 }
598 }
599 close $authors or croak $!;
600}
601# convert GetOpt::Long specs for use by git-config
603sub read_repo_config {
604 return unless -d $ENV{GIT_DIR};
605 my $opts = shift;
606 foreach my $o (keys %$opts) {
607 my $v = $opts->{$o};
608 my ($key) = ($o =~ /^([a-z\-]+)/);
609 $key =~ s/-//g;
610 my $arg = 'git-config';
611 $arg .= ' --int' if ($o =~ /[:=]i$/);
612 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
613 if (ref $v eq 'ARRAY') {
614 chomp(my @tmp = `$arg --get-all svn.$key`);
615 @$v = @tmp if @tmp;
616 } else {
617 chomp(my $tmp = `$arg --get svn.$key`);
618 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
619 $$v = $tmp;
620 }
621 }
622 }
623}
624sub extract_metadata {
626 my $id = shift or return (undef, undef, undef);
627 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
628 \s([a-f\d\-]+)$/x);
629 if (!defined $rev || !$uuid || !$url) {
630 # some of the original repositories I made had
631 # identifiers like this:
632 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
633 }
634 return ($url, $rev, $uuid);
635}
636sub cmt_metadata {
638 return extract_metadata((grep(/^git-svn-id: /,
639 command(qw/cat-file commit/, shift)))[-1]);
640}
641package Git::SVN;
643use strict;
644use warnings;
645use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
646 $_repack $_repack_flags/;
647use Carp qw/croak/;
648use File::Path qw/mkpath/;
649use File::Copy qw/copy/;
650use IPC::Open3;
651my $_repack_nr;
653# properties that we do not log:
654my %SKIP_PROP;
655BEGIN {
656 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
657 svn:special svn:executable
658 svn:entry:committed-rev
659 svn:entry:last-author
660 svn:entry:uuid
661 svn:entry:committed-date/;
662}
663my %LOCKFILES;
665END { unlink keys %LOCKFILES if %LOCKFILES }
666sub resolve_local_globs {
668 my ($url, $fetch, $glob_spec) = @_;
669 return unless defined $glob_spec;
670 my $ref = $glob_spec->{ref};
671 my $path = $glob_spec->{path};
672 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
673 next unless m#^refs/remotes/$ref->{regex}$#;
674 my $p = $1;
675 my $pathname = $path->full_path($p);
676 my $refname = $ref->full_path($p);
677 if (my $existing = $fetch->{$pathname}) {
678 if ($existing ne $refname) {
679 die "Refspec conflict:\n",
680 "existing: refs/remotes/$existing\n",
681 " globbed: refs/remotes/$refname\n";
682 }
683 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
684 $u =~ s!^\Q$url\E/?!! or die
685 "refs/remotes/$refname: '$url' not found in '$u'\n";
686 if ($pathname ne $u) {
687 warn "W: Refspec glob conflict ",
688 "(ref: refs/remotes/$refname):\n",
689 "expected path: $pathname\n",
690 " real path: $u\n",
691 "Continuing ahead with $u\n";
692 next;
693 }
694 } else {
695 warn "Globbed ($path->{glob}:$ref->{glob}): ",
696 "$pathname == $refname\n";
697 $fetch->{$pathname} = $refname;
698 }
699 }
700}
701sub fetch_all {
703 my ($repo_id, $remotes) = @_;
704 my $remote = $remotes->{$repo_id};
705 my $fetch = $remote->{fetch};
706 my $url = $remote->{url};
707 my (@gs, @globs);
708 my $ra = Git::SVN::Ra->new($url);
709 my $uuid = $ra->uuid;
710 my $head = $ra->get_latest_revnum;
711 my $base = $head;
712 # read the max revs for wildcard expansion (branches/*, tags/*)
714 foreach my $t (qw/branches tags/) {
715 defined $remote->{$t} or next;
716 push @globs, $remote->{$t};
717 my $f = "$ENV{GIT_DIR}/svn/.$uuid.$t";
718 if (open my $fh, '<', $f) {
719 chomp(my $max_rev = <$fh>);
720 close $fh or die "Error closing $f: $!\n";
721 if ($max_rev !~ /^\d+$/) {
723 die "$max_rev (in $f) is not an integer!\n";
724 }
725 $remote->{$t}->{max_rev} = $max_rev;
726 $base = $max_rev if ($max_rev < $base);
727 }
728 }
729 foreach my $p (sort keys %$fetch) {
731 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
732 my $lr = $gs->rev_db_max;
733 if (defined $lr) {
734 $base = $lr if ($lr < $base);
735 }
736 push @gs, $gs;
737 }
738 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
739}
740sub read_all_remotes {
742 my $r = {};
743 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
744 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
745 $r->{$1}->{fetch}->{$2} = $3;
746 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
747 $r->{$1}->{url} = $2;
748 } elsif (m!^(.+)\.(branches|tags)=
749 (.*):refs/remotes/(.+)\s*$/!x) {
750 my ($p, $g) = ($3, $4);
751 my $rs = $r->{$1}->{$2} = {
752 t => $2,
753 path => Git::SVN::GlobSpec->new($p),
754 ref => Git::SVN::GlobSpec->new($g) };
755 if (length($rs->{ref}->{right}) != 0) {
756 die "The '*' glob character must be the last ",
757 "character of '$g'\n";
758 }
759 }
760 }
761 $r;
762}
763sub init_vars {
765 if (defined $_repack) {
766 $_repack = 1000 if ($_repack <= 0);
767 $_repack_nr = $_repack;
768 $_repack_flags ||= '-d';
769 }
770}
771sub verify_remotes_sanity {
773 return unless -d $ENV{GIT_DIR};
774 my %seen;
775 foreach (command(qw/config -l/)) {
776 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
777 if ($seen{$1}) {
778 die "Remote ref refs/remote/$1 is tracked by",
779 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
780 "Please resolve this ambiguity in ",
781 "your git configuration file before ",
782 "continuing\n";
783 }
784 $seen{$1} = $_;
785 }
786 }
787}
788# we allow more chars than remotes2config.sh...
790sub sanitize_remote_name {
791 my ($name) = @_;
792 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
793 $name;
794}
795sub find_existing_remote {
797 my ($url, $remotes) = @_;
798 my $existing;
799 foreach my $repo_id (keys %$remotes) {
800 my $u = $remotes->{$repo_id}->{url} or next;
801 next if $u ne $url;
802 $existing = $repo_id;
803 last;
804 }
805 $existing;
806}
807sub init_remote_config {
809 my ($self, $url, $no_write) = @_;
810 $url =~ s!/+$!!; # strip trailing slash
811 my $r = read_all_remotes();
812 my $existing = find_existing_remote($url, $r);
813 if ($existing) {
814 unless ($no_write) {
815 print STDERR "Using existing ",
816 "[svn-remote \"$existing\"]\n";
817 }
818 $self->{repo_id} = $existing;
819 } else {
820 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
821 $existing = find_existing_remote($min_url, $r);
822 if ($existing) {
823 unless ($no_write) {
824 print STDERR "Using existing ",
825 "[svn-remote \"$existing\"]\n";
826 }
827 $self->{repo_id} = $existing;
828 }
829 if ($min_url ne $url) {
830 unless ($no_write) {
831 print STDERR "Using higher level of URL: ",
832 "$url => $min_url\n";
833 }
834 my $old_path = $self->{path};
835 $self->{path} = $url;
836 $self->{path} =~ s!^\Q$min_url\E/*!!;
837 if (length $old_path) {
838 $self->{path} .= "/$old_path";
839 }
840 $url = $min_url;
841 }
842 }
843 my $orig_url;
844 if (!$existing) {
845 # verify that we aren't overwriting anything:
846 $orig_url = eval {
847 command_oneline('config', '--get',
848 "svn-remote.$self->{repo_id}.url")
849 };
850 if ($orig_url && ($orig_url ne $url)) {
851 die "svn-remote.$self->{repo_id}.url already set: ",
852 "$orig_url\nwanted to set to: $url\n";
853 }
854 }
855 my ($xrepo_id, $xpath) = find_ref($self->refname);
856 if (defined $xpath) {
857 die "svn-remote.$xrepo_id.fetch already set to track ",
858 "$xpath:refs/remotes/", $self->refname, "\n";
859 }
860 unless ($no_write) {
861 command_noisy('config',
862 "svn-remote.$self->{repo_id}.url", $url);
863 command_noisy('config', '--add',
864 "svn-remote.$self->{repo_id}.fetch",
865 "$self->{path}:".$self->refname);
866 }
867 $self->{url} = $url;
868}
869sub init {
871 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
872 my $self = _new($class, $repo_id, $ref_id, $path);
873 if (defined $url) {
874 $self->init_remote_config($url, $no_write);
875 }
876 $self;
877}
878sub find_ref {
880 my ($ref_id) = @_;
881 foreach (command(qw/config -l/)) {
882 next unless m!^svn-remote\.(.+)\.fetch=
883 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
884 my ($repo_id, $path, $ref) = ($1, $2, $3);
885 if ($ref eq $ref_id) {
886 $path = '' if ($path =~ m#^\./?#);
887 return ($repo_id, $path);
888 }
889 }
890 (undef, undef, undef);
891}
892sub new {
894 my ($class, $ref_id, $repo_id, $path) = @_;
895 if (defined $ref_id && !defined $repo_id && !defined $path) {
896 ($repo_id, $path) = find_ref($ref_id);
897 if (!defined $repo_id) {
898 die "Could not find a \"svn-remote.*.fetch\" key ",
899 "in the repository configuration matching: ",
900 "refs/remotes/$ref_id\n";
901 }
902 }
903 my $self = _new($class, $repo_id, $ref_id, $path);
904 if (!defined $self->{path} || !length $self->{path}) {
905 my $fetch = command_oneline('config', '--get',
906 "svn-remote.$repo_id.fetch",
907 ":refs/remotes/$ref_id\$") or
908 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
909 "\":refs/remotes/$ref_id\$\" in config\n";
910 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
911 }
912 $self->{url} = command_oneline('config', '--get',
913 "svn-remote.$repo_id.url") or
914 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
915 if (-z $self->{db_path} && ::verify_ref($self->refname.'^0')) {
916 $self->rebuild;
917 }
918 $self;
919}
920sub refname { "refs/remotes/$_[0]->{ref_id}" }
922sub ra {
924 my ($self) = shift;
925 Git::SVN::Ra->new($self->{url});
926}
927sub rel_path {
929 my ($self) = @_;
930 my $repos_root = $self->ra->{repos_root};
931 return $self->{path} if ($self->{url} eq $repos_root);
932 my $url = $self->{url} .
933 (length $self->{path} ? "/$self->{path}" : $self->{path});
934 $url =~ s!^\Q$repos_root\E/*!!g;
935 $url;
936}
937sub traverse_ignore {
939 my ($self, $fh, $path, $r) = @_;
940 $path =~ s#^/+##g;
941 my $ra = $self->ra;
942 my ($dirent, undef, $props) = $ra->get_dir($path, $r);
943 my $p = $path;
944 $p =~ s#^\Q$ra->{svn_path}\E/##;
945 print $fh length $p ? "\n# $p\n" : "\n# /\n";
946 if (my $s = $props->{'svn:ignore'}) {
947 $s =~ s/[\r\n]+/\n/g;
948 chomp $s;
949 if (length $p == 0) {
950 $s =~ s#\n#\n/$p#g;
951 print $fh "/$s\n";
952 } else {
953 $s =~ s#\n#\n/$p/#g;
954 print $fh "/$p/$s\n";
955 }
956 }
957 foreach (sort keys %$dirent) {
958 next if $dirent->{$_}->kind != $SVN::Node::dir;
959 $self->traverse_ignore($fh, "$path/$_", $r);
960 }
961}
962sub last_rev { ($_[0]->last_rev_commit)[0] }
964sub last_commit { ($_[0]->last_rev_commit)[1] }
965# returns the newest SVN revision number and newest commit SHA1
967sub last_rev_commit {
968 my ($self) = @_;
969 if (defined $self->{last_rev} && defined $self->{last_commit}) {
970 return ($self->{last_rev}, $self->{last_commit});
971 }
972 my $c = ::verify_ref($self->refname.'^0');
973 if ($c) {
974 my $rev = (::cmt_metadata($c))[1];
975 if (defined $rev) {
976 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
977 return ($rev, $c);
978 }
979 }
980 my $offset = -41; # from tail
981 my $rl;
982 open my $fh, '<', $self->{db_path} or
983 croak "$self->{db_path} not readable: $!\n";
984 sysseek($fh, $offset, 2); # don't care for errors
985 sysread($fh, $rl, 41) == 41 or return (undef, undef);
986 chomp $rl;
987 while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
988 $offset -= 41;
989 sysseek($fh, $offset, 2); # don't care for errors
990 sysread($fh, $rl, 41) == 41 or return (undef, undef);
991 chomp $rl;
992 }
993 if ($c) {
994 die "$self->{db_path} and ", $self->refname,
995 " inconsistent!:\n$c != $rl\n";
996 }
997 my $rev = sysseek($fh, 0, 1) or croak $!;
998 $rev = ($rev - 41) / 41;
999 close $fh or croak $!;
1000 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1001 return ($rev, $c);
1002}
1003sub get_fetch_range {
1005 my ($self, $min, $max) = @_;
1006 $max ||= $self->ra->get_latest_revnum;
1007 $min ||= $self->rev_db_max;
1008 (++$min, $max);
1009}
1010sub tmp_index_do {
1012 my ($self, $sub) = @_;
1013 my $old_index = $ENV{GIT_INDEX_FILE};
1014 $ENV{GIT_INDEX_FILE} = $self->{index};
1015 my @ret = &$sub;
1016 if ($old_index) {
1017 $ENV{GIT_INDEX_FILE} = $old_index;
1018 } else {
1019 delete $ENV{GIT_INDEX_FILE};
1020 }
1021 wantarray ? @ret : $ret[0];
1022}
1023sub assert_index_clean {
1025 my ($self, $treeish) = @_;
1026 $self->tmp_index_do(sub {
1028 command_noisy('read-tree', $treeish) unless -e $self->{index};
1029 my $x = command_oneline('write-tree');
1030 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1031 /^tree ($::sha1)/mo);
1032 if ($y ne $x) {
1033 unlink $self->{index} or croak $!;
1034 command_noisy('read-tree', $treeish);
1035 }
1036 $x = command_oneline('write-tree');
1037 if ($y ne $x) {
1038 ::fatal "trees ($treeish) $y != $x\n",
1039 "Something is seriously wrong...\n";
1040 }
1041 });
1042}
1043sub get_commit_parents {
1045 my ($self, $log_entry) = @_;
1046 my (%seen, @ret, @tmp);
1047 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1048 if (my $ip = $self->{inject_parents}) {
1049 if (my $commit = delete $ip->{$log_entry->{revision}}) {
1050 push @tmp, $commit;
1051 }
1052 }
1053 if (my $cur = ::verify_ref($self->refname.'^0')) {
1054 push @tmp, $cur;
1055 }
1056 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1057 while (my $p = shift @tmp) {
1058 next if $seen{$p};
1059 $seen{$p} = 1;
1060 push @ret, $p;
1061 # MAXPARENT is defined to 16 in commit-tree.c:
1062 last if @ret >= 16;
1063 }
1064 if (@tmp) {
1065 die "r$log_entry->{revision}: No room for parents:\n\t",
1066 join("\n\t", @tmp), "\n";
1067 }
1068 @ret;
1069}
1070sub full_url {
1072 my ($self) = @_;
1073 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1074}
1075sub do_git_commit {
1077 my ($self, $log_entry) = @_;
1078 my $lr = $self->last_rev;
1079 if (defined $lr && $lr >= $log_entry->{revision}) {
1080 die "Last fetched revision of ", $self->refname,
1081 " was r$lr, but we are about to fetch: ",
1082 "r$log_entry->{revision}!\n";
1083 }
1084 if (my $c = $self->rev_db_get($log_entry->{revision})) {
1085 croak "$log_entry->{revision} = $c already exists! ",
1086 "Why are we refetching it?\n";
1087 }
1088 my $author = $log_entry->{author};
1089 my ($name, $email) = (defined $::users{$author} ? @{$::users{$author}}
1090 : ($author, "$author\@".$self->ra->uuid));
1091 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1092 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1093 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1094 my $tree = $log_entry->{tree};
1096 if (!defined $tree) {
1097 $tree = $self->tmp_index_do(sub {
1098 command_oneline('write-tree') });
1099 }
1100 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1101 my @exec = ('git-commit-tree', $tree);
1103 foreach ($self->get_commit_parents($log_entry)) {
1104 push @exec, '-p', $_;
1105 }
1106 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1107 or croak $!;
1108 print $msg_fh $log_entry->{log} or croak $!;
1109 unless ($_no_metadata) {
1110 print $msg_fh "\ngit-svn-id: ", $self->full_url, '@',
1111 $log_entry->{revision}, ' ',
1112 $self->ra->uuid, "\n" or croak $!;
1113 }
1114 $msg_fh->flush == 0 or croak $!;
1115 close $msg_fh or croak $!;
1116 chomp(my $commit = do { local $/; <$out_fh> });
1117 close $out_fh or croak $!;
1118 waitpid $pid, 0;
1119 croak $? if $?;
1120 if ($commit !~ /^$::sha1$/o) {
1121 die "Failed to commit, invalid sha1: $commit\n";
1122 }
1123 $self->rev_db_set($log_entry->{revision}, $commit, 1);
1125 $self->{last_rev} = $log_entry->{revision};
1127 $self->{last_commit} = $commit;
1128 print "r$log_entry->{revision} = $commit ($self->{ref_id})\n";
1129 if (defined $_repack && (--$_repack_nr == 0)) {
1130 $_repack_nr = $_repack;
1131 # repack doesn't use any arguments with spaces in them, does it?
1132 print "Running git repack $_repack_flags ...\n";
1133 command_noisy('repack', split(/\s+/, $_repack_flags));
1134 print "Done repacking\n";
1135 }
1136 return $commit;
1137}
1138sub match_paths {
1140 my ($self, $paths, $r) = @_;
1141 if (my $path = $paths->{"/$self->{path}"}) {
1142 return ($path->{action} eq 'D') ? 0 : 1;
1143 }
1144 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\/?/;
1145 if (grep /$self->{path_regex}/, keys %$paths) {
1146 return 1;
1147 }
1148 my $c = '';
1149 foreach (split m#/#, $self->{path}) {
1150 $c .= "/$_";
1151 next unless ($paths->{$c} && ($paths->{$c}->{action} eq 'A'));
1152 if ($self->ra->check_path($self->{path}, $r) ==
1153 $SVN::Node::dir) {
1154 return 1;
1155 }
1156 }
1157 return 0;
1158}
1159sub find_parent_branch {
1161 my ($self, $paths, $rev) = @_;
1162 return undef unless $_follow_parent;
1163 unless (defined $paths) {
1164 my $err_handler = $SVN::Error::handler;
1165 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1166 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1167 $paths =
1168 Git::SVN::Ra::dup_changed_paths($_[0]) });
1169 $SVN::Error::handler = $err_handler;
1170 }
1171 return undef unless defined $paths;
1172 # look for a parent from another branch:
1174 my @b_path_components = split m#/#, $self->rel_path;
1175 my @a_path_components;
1176 my $i;
1177 while (@b_path_components) {
1178 $i = $paths->{'/'.join('/', @b_path_components)};
1179 last if $i;
1180 unshift(@a_path_components, pop(@b_path_components));
1181 }
1182 goto not_found unless defined $i;
1183 my $branch_from = $i->{copyfrom_path} or goto not_found;
1184 if (@a_path_components) {
1185 print STDERR "branch_from: $branch_from => ";
1186 $branch_from .= '/'.join('/', @a_path_components);
1187 print STDERR $branch_from, "\n";
1188 }
1189 my $r = $i->{copyfrom_rev};
1190 my $repos_root = $self->ra->{repos_root};
1191 my $url = $self->ra->{url};
1192 my $new_url = $repos_root . $branch_from;
1193 print STDERR "Found possible branch point: ",
1194 "$new_url => ", $self->full_url, ", $r\n";
1195 $branch_from =~ s#^/##;
1196 my $remotes = read_all_remotes();
1197 my $gs;
1198 foreach my $repo_id (keys %$remotes) {
1199 my $u = $remotes->{$repo_id}->{url} or next;
1200 next if $url ne $u;
1201 my $fetch = $remotes->{$repo_id}->{fetch};
1202 foreach (qw/branches tags/) {
1203 resolve_local_globs($url, $fetch,
1204 $remotes->{$repo_id}->{$_});
1205 }
1206 foreach my $f (keys %$fetch) {
1207 next if $f ne $branch_from;
1208 $gs = Git::SVN->new($fetch->{$f}, $repo_id, $f);
1209 last;
1210 }
1211 last if $gs;
1212 }
1213 unless ($gs) {
1214 my $ref_id = $self->{ref_id};
1215 $ref_id =~ s/\@\d+$//;
1216 $ref_id .= "\@$r";
1217 # just grow a tail if we're not unique enough :x
1218 $ref_id .= '-' while find_ref($ref_id);
1219 print STDERR "Initializing parent: $ref_id\n";
1220 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1221 }
1222 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1223 if ($_follow_parent && (!defined $r0 || !defined $parent)) {
1224 $gs->fetch(0, $r);
1225 ($r0, $parent) = $gs->last_rev_commit;
1226 }
1227 if (defined $r0 && defined $parent) {
1228 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1229 $self->assert_index_clean($parent);
1230 my $ed;
1231 if ($self->ra->can_do_switch) {
1232 print STDERR "Following parent with do_switch\n";
1233 # do_switch works with svn/trunk >= r22312, but that
1234 # is not included with SVN 1.4.3 (the latest version
1235 # at the moment), so we can't rely on it
1236 $self->{last_commit} = $parent;
1237 $ed = SVN::Git::Fetcher->new($self);
1238 $gs->ra->gs_do_switch($r0, $rev, $gs,
1239 $self->full_url, $ed)
1240 or die "SVN connection failed somewhere...\n";
1241 } else {
1242 print STDERR "Following parent with do_update\n";
1243 $ed = SVN::Git::Fetcher->new($self);
1244 $self->ra->gs_do_update($rev, $rev, $self, $ed)
1245 or die "SVN connection failed somewhere...\n";
1246 }
1247 print STDERR "Successfully followed parent\n";
1248 return $self->make_log_entry($rev, [$parent], $ed);
1249 }
1250not_found:
1251 print STDERR "Branch parent for path: '/",
1252 $self->rel_path, "' @ r$rev not found:\n";
1253 return undef unless $paths;
1254 print STDERR "Changed paths:\n";
1255 foreach my $x (sort keys %$paths) {
1256 my $p = $paths->{$x};
1257 print STDERR "\t$p->{action}\t$x";
1258 if ($p->{copyfrom_path}) {
1259 print STDERR "(from $p->{copyfrom_path}: ",
1260 "$p->{copyfrom_rev})";
1261 }
1262 print STDERR "\n";
1263 }
1264 print STDERR '-'x72, "\n";
1265 return undef;
1266}
1267sub do_fetch {
1269 my ($self, $paths, $rev) = @_;
1270 my $ed;
1271 my ($last_rev, @parents);
1272 if (my $lc = $self->last_commit) {
1273 # we can have a branch that was deleted, then re-added
1274 # under the same name but copied from another path, in
1275 # which case we'll have multiple parents (we don't
1276 # want to break the original ref, nor lose copypath info):
1277 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1278 push @{$log_entry->{parents}}, $lc;
1279 return $log_entry;
1280 }
1281 $ed = SVN::Git::Fetcher->new($self);
1282 $last_rev = $self->{last_rev};
1283 $ed->{c} = $lc;
1284 @parents = ($lc);
1285 } else {
1286 $last_rev = $rev;
1287 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1288 return $log_entry;
1289 }
1290 $ed = SVN::Git::Fetcher->new($self);
1291 }
1292 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1293 die "SVN connection failed somewhere...\n";
1294 }
1295 $self->make_log_entry($rev, \@parents, $ed);
1296}
1297sub get_untracked {
1299 my ($self, $ed) = @_;
1300 my @out;
1301 my $h = $ed->{empty};
1302 foreach (sort keys %$h) {
1303 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1304 push @out, " $act: " . uri_encode($_);
1305 warn "W: $act: $_\n";
1306 }
1307 foreach my $t (qw/dir_prop file_prop/) {
1308 $h = $ed->{$t} or next;
1309 foreach my $path (sort keys %$h) {
1310 my $ppath = $path eq '' ? '.' : $path;
1311 foreach my $prop (sort keys %{$h->{$path}}) {
1312 next if $SKIP_PROP{$prop};
1313 my $v = $h->{$path}->{$prop};
1314 my $t_ppath_prop = "$t: " .
1315 uri_encode($ppath) . ' ' .
1316 uri_encode($prop);
1317 if (defined $v) {
1318 push @out, " +$t_ppath_prop " .
1319 uri_encode($v);
1320 } else {
1321 push @out, " -$t_ppath_prop";
1322 }
1323 }
1324 }
1325 }
1326 foreach my $t (qw/absent_file absent_directory/) {
1327 $h = $ed->{$t} or next;
1328 foreach my $parent (sort keys %$h) {
1329 foreach my $path (sort @{$h->{$parent}}) {
1330 push @out, " $t: " .
1331 uri_encode("$parent/$path");
1332 warn "W: $t: $parent/$path ",
1333 "Insufficient permissions?\n";
1334 }
1335 }
1336 }
1337 \@out;
1338}
1339sub parse_svn_date {
1341 my $date = shift || return '+0000 1970-01-01 00:00:00';
1342 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1343 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1344 croak "Unable to parse date: $date\n";
1345 "+0000 $Y-$m-$d $H:$M:$S";
1346}
1347sub check_author {
1349 my ($author) = @_;
1350 if (!defined $author || length $author == 0) {
1351 $author = '(no author)';
1352 }
1353 if (defined $::_authors && ! defined $::users{$author}) {
1354 die "Author: $author not defined in $::_authors file\n";
1355 }
1356 $author;
1357}
1358sub make_log_entry {
1360 my ($self, $rev, $parents, $ed) = @_;
1361 my $untracked = $self->get_untracked($ed);
1362 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1364 print $un "r$rev\n" or croak $!;
1365 print $un $_, "\n" foreach @$untracked;
1366 my %log_entry = ( parents => $parents || [], revision => $rev,
1367 log => '');
1368 my $logged = delete $self->{logged_rev_props};
1370 if (!$logged || $self->{-want_extra_revprops}) {
1371 my $rp = $self->ra->rev_proplist($rev);
1372 foreach (sort keys %$rp) {
1373 my $v = $rp->{$_};
1374 if (/^svn:(author|date|log)$/) {
1375 $log_entry{$1} = $v;
1376 } else {
1377 print $un " rev_prop: ", uri_encode($_), ' ',
1378 uri_encode($v), "\n";
1379 }
1380 }
1381 } else {
1382 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1383 }
1384 close $un or croak $!;
1385 $log_entry{date} = parse_svn_date($log_entry{date});
1387 $log_entry{author} = check_author($log_entry{author});
1388 $log_entry{log} .= "\n";
1389 \%log_entry;
1390}
1391sub fetch {
1393 my ($self, $min_rev, $max_rev, @parents) = @_;
1394 my ($last_rev, $last_commit) = $self->last_rev_commit;
1395 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1396 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1397}
1398sub set_tree_cb {
1400 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1401 # TODO: enable and test optimized commits:
1402 if (0 && $rev == ($self->{last_rev} + 1)) {
1403 $log_entry->{revision} = $rev;
1404 $log_entry->{author} = $author;
1405 $self->do_git_commit($log_entry, "$rev=$tree");
1406 } else {
1407 $self->{inject_parents} = { $rev => $tree };
1408 $self->fetch(undef, undef);
1409 }
1410}
1411sub set_tree {
1413 my ($self, $tree) = (shift, shift);
1414 my $log_entry = ::get_commit_entry($tree);
1415 unless ($self->{last_rev}) {
1416 fatal("Must have an existing revision to commit\n");
1417 }
1418 my %ed_opts = ( r => $self->{last_rev},
1419 log => $log_entry->{log},
1420 ra => $self->ra,
1421 tree_a => $self->{last_commit},
1422 tree_b => $tree,
1423 editor_cb => sub {
1424 $self->set_tree_cb($log_entry, $tree, @_) },
1425 svn_path => $self->{path} );
1426 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1427 print "No changes\nr$self->{last_rev} = $tree\n";
1428 }
1429}
1430sub rebuild {
1432 my ($self) = @_;
1433 print "Rebuilding $self->{db_path} ...\n";
1434 my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1435 my $latest;
1436 my $full_url = $self->full_url;
1437 my $svn_uuid;
1438 while (<$rev_list>) {
1439 chomp;
1440 my $c = $_;
1441 die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1442 my ($url, $rev, $uuid) = ::cmt_metadata($c);
1443 # ignore merges (from set-tree)
1445 next if (!defined $rev || !$uuid);
1446 # if we merged or otherwise started elsewhere, this is
1448 # how we break out of it
1449 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1450 ($full_url && $url && ($url ne $full_url))) {
1451 next;
1452 }
1453 $latest ||= $rev;
1454 $svn_uuid ||= $uuid;
1455 $self->rev_db_set($rev, $c);
1457 print "r$rev = $c\n";
1458 }
1459 command_close_pipe($rev_list, $ctx);
1460 print "Done rebuilding $self->{db_path}\n";
1461}
1462# rev_db:
1464# Tie::File seems to be prone to offset errors if revisions get sparse,
1465# it's not that fast, either. Tie::File is also not in Perl 5.6. So
1466# one of my favorite modules is out :< Next up would be one of the DBM
1467# modules, but I'm not sure which is most portable... So I'll just
1468# go with something that's plain-text, but still capable of
1469# being randomly accessed. So here's my ultra-simple fixed-width
1470# database. All records are 40 characters + "\n", so it's easy to seek
1471# to a revision: (41 * rev) is the byte offset.
1472# A record of 40 0s denotes an empty revision.
1473# And yes, it's still pretty fast (faster than Tie::File).
1474# These files are disposable unless --no-metadata is set
1475sub rev_db_set {
1477 my ($self, $rev, $commit, $update_ref) = @_;
1478 length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
1479 my ($db, $db_lock) = ($self->{db_path}, "$self->{db_path}.lock");
1480 my $sig;
1481 if ($update_ref) {
1482 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1483 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
1484 }
1485 $LOCKFILES{$db_lock} = 1;
1486 if ($_no_metadata) {
1487 copy($db, $db_lock) or die "rev_db_set(@_): ",
1488 "Failed to copy: ",
1489 "$db => $db_lock ($!)\n";
1490 } else {
1491 rename $db, $db_lock or die "rev_db_set(@_): ",
1492 "Failed to rename: ",
1493 "$db => $db_lock ($!)\n";
1494 }
1495 open my $fh, '+<', $db_lock or croak $!;
1496 my $offset = $rev * 41;
1497 # assume that append is the common case:
1498 seek $fh, 0, 2 or croak $!;
1499 my $pos = tell $fh;
1500 if ($pos < $offset) {
1501 for (1 .. (($offset - $pos) / 41)) {
1502 print $fh (('0' x 40),"\n") or croak $!;
1503 }
1504 }
1505 seek $fh, $offset, 0 or croak $!;
1506 print $fh $commit,"\n" or croak $!;
1507 close $fh or croak $!;
1508 if ($update_ref) {
1509 command_noisy('update-ref', '-m', "r$rev",
1510 $self->refname, $commit);
1511 }
1512 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
1513 "$db_lock => $db ($!)\n";
1514 delete $LOCKFILES{$db_lock};
1515 if ($update_ref) {
1516 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1517 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
1518 kill $sig, $$ if defined $sig;
1519 }
1520}
1521sub rev_db_max {
1523 my ($self) = @_;
1524 my @stat = stat $self->{db_path} or
1525 die "Couldn't stat $self->{db_path}: $!\n";
1526 ($stat[7] % 41) == 0 or
1527 die "$self->{db_path} inconsistent size:$stat[7]\n";
1528 my $max = $stat[7] / 41;
1529 (($max > 0) ? $max - 1 : 0);
1530}
1531sub rev_db_get {
1533 my ($self, $rev) = @_;
1534 my $ret;
1535 my $offset = $rev * 41;
1536 open my $fh, '<', $self->{db_path} or croak $!;
1537 if (sysseek($fh, $offset, 0) == $offset) {
1538 my $read = sysread($fh, $ret, 40);
1539 $ret = undef if ($read != 40 || $ret eq ('0'x40));
1540 }
1541 close $fh or croak $!;
1542 $ret;
1543}
1544sub find_rev_before {
1546 my ($self, $rev, $eq_ok) = @_;
1547 --$rev unless $eq_ok;
1548 while ($rev > 0) {
1549 if (my $c = $self->rev_db_get($rev)) {
1550 return ($rev, $c);
1551 }
1552 --$rev;
1553 }
1554 return (undef, undef);
1555}
1556sub _new {
1558 my ($class, $repo_id, $ref_id, $path) = @_;
1559 unless (defined $repo_id && length $repo_id) {
1560 $repo_id = $Git::SVN::default_repo_id;
1561 }
1562 unless (defined $ref_id && length $ref_id) {
1563 $_[2] = $ref_id = $Git::SVN::default_ref_id;
1564 }
1565 $_[1] = $repo_id = sanitize_remote_name($repo_id);
1566 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1567 $_[3] = $path = '' unless (defined $path);
1568 mkpath([$dir]);
1569 unless (-f "$dir/.rev_db") {
1570 open my $fh, '>>', "$dir/.rev_db" or croak $!;
1571 close $fh or croak $!;
1572 }
1573 bless { ref_id => $ref_id, dir => $dir, index => "$dir/index",
1574 path => $path,
1575 db_path => "$dir/.rev_db", repo_id => $repo_id }, $class;
1576}
1577sub uri_encode {
1579 my ($f) = @_;
1580 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1581 $f
1582}
1583package Git::SVN::Prompt;
1585use strict;
1586use warnings;
1587require SVN::Core;
1588use vars qw/$_no_auth_cache $_username/;
1589sub simple {
1591 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1592 $may_save = undef if $_no_auth_cache;
1593 $default_username = $_username if defined $_username;
1594 if (defined $default_username && length $default_username) {
1595 if (defined $realm && length $realm) {
1596 print STDERR "Authentication realm: $realm\n";
1597 STDERR->flush;
1598 }
1599 $cred->username($default_username);
1600 } else {
1601 username($cred, $realm, $may_save, $pool);
1602 }
1603 $cred->password(_read_password("Password for '" .
1604 $cred->username . "': ", $realm));
1605 $cred->may_save($may_save);
1606 $SVN::_Core::SVN_NO_ERROR;
1607}
1608sub ssl_server_trust {
1610 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1611 $may_save = undef if $_no_auth_cache;
1612 print STDERR "Error validating server certificate for '$realm':\n";
1613 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1614 print STDERR " - The certificate is not issued by a trusted ",
1615 "authority. Use the\n",
1616 " fingerprint to validate the certificate manually!\n";
1617 }
1618 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1619 print STDERR " - The certificate hostname does not match.\n";
1620 }
1621 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1622 print STDERR " - The certificate is not yet valid.\n";
1623 }
1624 if ($failures & $SVN::Auth::SSL::EXPIRED) {
1625 print STDERR " - The certificate has expired.\n";
1626 }
1627 if ($failures & $SVN::Auth::SSL::OTHER) {
1628 print STDERR " - The certificate has an unknown error.\n";
1629 }
1630 printf STDERR
1631 "Certificate information:\n".
1632 " - Hostname: %s\n".
1633 " - Valid: from %s until %s\n".
1634 " - Issuer: %s\n".
1635 " - Fingerprint: %s\n",
1636 map $cert_info->$_, qw(hostname valid_from valid_until
1637 issuer_dname fingerprint);
1638 my $choice;
1639prompt:
1640 print STDERR $may_save ?
1641 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1642 "(R)eject or accept (t)emporarily? ";
1643 STDERR->flush;
1644 $choice = lc(substr(<STDIN> || 'R', 0, 1));
1645 if ($choice =~ /^t$/i) {
1646 $cred->may_save(undef);
1647 } elsif ($choice =~ /^r$/i) {
1648 return -1;
1649 } elsif ($may_save && $choice =~ /^p$/i) {
1650 $cred->may_save($may_save);
1651 } else {
1652 goto prompt;
1653 }
1654 $cred->accepted_failures($failures);
1655 $SVN::_Core::SVN_NO_ERROR;
1656}
1657sub ssl_client_cert {
1659 my ($cred, $realm, $may_save, $pool) = @_;
1660 $may_save = undef if $_no_auth_cache;
1661 print STDERR "Client certificate filename: ";
1662 STDERR->flush;
1663 chomp(my $filename = <STDIN>);
1664 $cred->cert_file($filename);
1665 $cred->may_save($may_save);
1666 $SVN::_Core::SVN_NO_ERROR;
1667}
1668sub ssl_client_cert_pw {
1670 my ($cred, $realm, $may_save, $pool) = @_;
1671 $may_save = undef if $_no_auth_cache;
1672 $cred->password(_read_password("Password: ", $realm));
1673 $cred->may_save($may_save);
1674 $SVN::_Core::SVN_NO_ERROR;
1675}
1676sub username {
1678 my ($cred, $realm, $may_save, $pool) = @_;
1679 $may_save = undef if $_no_auth_cache;
1680 if (defined $realm && length $realm) {
1681 print STDERR "Authentication realm: $realm\n";
1682 }
1683 my $username;
1684 if (defined $_username) {
1685 $username = $_username;
1686 } else {
1687 print STDERR "Username: ";
1688 STDERR->flush;
1689 chomp($username = <STDIN>);
1690 }
1691 $cred->username($username);
1692 $cred->may_save($may_save);
1693 $SVN::_Core::SVN_NO_ERROR;
1694}
1695sub _read_password {
1697 my ($prompt, $realm) = @_;
1698 print STDERR $prompt;
1699 STDERR->flush;
1700 require Term::ReadKey;
1701 Term::ReadKey::ReadMode('noecho');
1702 my $password = '';
1703 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1704 last if $key =~ /[\012\015]/; # \n\r
1705 $password .= $key;
1706 }
1707 Term::ReadKey::ReadMode('restore');
1708 print STDERR "\n";
1709 STDERR->flush;
1710 $password;
1711}
1712package main;
1714{
1716 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1717 $SVN::Node::dir.$SVN::Node::unknown.
1718 $SVN::Node::none.$SVN::Node::file.
1719 $SVN::Node::dir.$SVN::Node::unknown.
1720 $SVN::Auth::SSL::CNMISMATCH.
1721 $SVN::Auth::SSL::NOTYETVALID.
1722 $SVN::Auth::SSL::EXPIRED.
1723 $SVN::Auth::SSL::UNKNOWNCA.
1724 $SVN::Auth::SSL::OTHER;
1725}
1726package SVN::Git::Fetcher;
1728use vars qw/@ISA/;
1729use strict;
1730use warnings;
1731use Carp qw/croak/;
1732use IO::File qw//;
1733use Digest::MD5;
1734# file baton members: path, mode_a, mode_b, pool, fh, blob, base
1736sub new {
1737 my ($class, $git_svn) = @_;
1738 my $self = SVN::Delta::Editor->new;
1739 bless $self, $class;
1740 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
1741 $self->{empty} = {};
1742 $self->{dir_prop} = {};
1743 $self->{file_prop} = {};
1744 $self->{absent_dir} = {};
1745 $self->{absent_file} = {};
1746 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
1747 $self;
1748}
1749sub set_path_strip {
1751 my ($self, $path) = @_;
1752 $self->{path_strip} = qr/^\Q$path\E\/?/;
1753}
1754sub open_root {
1756 { path => '' };
1757}
1758sub open_directory {
1760 my ($self, $path, $pb, $rev) = @_;
1761 { path => $path };
1762}
1763sub git_path {
1765 my ($self, $path) = @_;
1766 if ($self->{path_strip}) {
1767 $path =~ s!$self->{path_strip}!! or
1768 die "Failed to strip path '$path' ($self->{path_strip})\n";
1769 }
1770 $path;
1771}
1772sub delete_entry {
1774 my ($self, $path, $rev, $pb) = @_;
1775 my $gpath = $self->git_path($path);
1777 return undef if ($gpath eq '');
1778 # remove entire directories.
1780 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
1781 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
1782 -r --name-only -z/,
1783 $self->{c}, '--', $gpath);
1784 local $/ = "\0";
1785 while (<$ls>) {
1786 chomp;
1787 $self->{gii}->remove($_);
1788 print "\tD\t$_\n" unless $self->{q};
1789 }
1790 print "\tD\t$gpath/\n" unless $self->{q};
1791 command_close_pipe($ls, $ctx);
1792 $self->{empty}->{$path} = 0
1793 } else {
1794 $self->{gii}->remove($gpath);
1795 print "\tD\t$gpath\n" unless $self->{q};
1796 }
1797 undef;
1798}
1799sub open_file {
1801 my ($self, $path, $pb, $rev) = @_;
1802 my $gpath = $self->git_path($path);
1803 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
1804 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
1805 unless (defined $mode && defined $blob) {
1806 die "$path was not found in commit $self->{c} (r$rev)\n";
1807 }
1808 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
1809 pool => SVN::Pool->new, action => 'M' };
1810}
1811sub add_file {
1813 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
1814 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1815 delete $self->{empty}->{$dir};
1816 { path => $path, mode_a => 100644, mode_b => 100644,
1817 pool => SVN::Pool->new, action => 'A' };
1818}
1819sub add_directory {
1821 my ($self, $path, $cp_path, $cp_rev) = @_;
1822 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1823 delete $self->{empty}->{$dir};
1824 $self->{empty}->{$path} = 1;
1825 { path => $path };
1826}
1827sub change_dir_prop {
1829 my ($self, $db, $prop, $value) = @_;
1830 $self->{dir_prop}->{$db->{path}} ||= {};
1831 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
1832 undef;
1833}
1834sub absent_directory {
1836 my ($self, $path, $pb) = @_;
1837 $self->{absent_dir}->{$pb->{path}} ||= [];
1838 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
1839 undef;
1840}
1841sub absent_file {
1843 my ($self, $path, $pb) = @_;
1844 $self->{absent_file}->{$pb->{path}} ||= [];
1845 push @{$self->{absent_file}->{$pb->{path}}}, $path;
1846 undef;
1847}
1848sub change_file_prop {
1850 my ($self, $fb, $prop, $value) = @_;
1851 if ($prop eq 'svn:executable') {
1852 if ($fb->{mode_b} != 120000) {
1853 $fb->{mode_b} = defined $value ? 100755 : 100644;
1854 }
1855 } elsif ($prop eq 'svn:special') {
1856 $fb->{mode_b} = defined $value ? 120000 : 100644;
1857 } else {
1858 $self->{file_prop}->{$fb->{path}} ||= {};
1859 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
1860 }
1861 undef;
1862}
1863sub apply_textdelta {
1865 my ($self, $fb, $exp) = @_;
1866 my $fh = IO::File->new_tmpfile;
1867 $fh->autoflush(1);
1868 # $fh gets auto-closed() by SVN::TxDelta::apply(),
1869 # (but $base does not,) so dup() it for reading in close_file
1870 open my $dup, '<&', $fh or croak $!;
1871 my $base = IO::File->new_tmpfile;
1872 $base->autoflush(1);
1873 if ($fb->{blob}) {
1874 defined (my $pid = fork) or croak $!;
1875 if (!$pid) {
1876 open STDOUT, '>&', $base or croak $!;
1877 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
1878 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
1879 }
1880 waitpid $pid, 0;
1881 croak $? if $?;
1882 if (defined $exp) {
1884 seek $base, 0, 0 or croak $!;
1885 my $md5 = Digest::MD5->new;
1886 $md5->addfile($base);
1887 my $got = $md5->hexdigest;
1888 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
1889 "expected: $exp\n",
1890 " got: $got\n" if ($got ne $exp);
1891 }
1892 }
1893 seek $base, 0, 0 or croak $!;
1894 $fb->{fh} = $dup;
1895 $fb->{base} = $base;
1896 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
1897}
1898sub close_file {
1900 my ($self, $fb, $exp) = @_;
1901 my $hash;
1902 my $path = $self->git_path($fb->{path});
1903 if (my $fh = $fb->{fh}) {
1904 seek($fh, 0, 0) or croak $!;
1905 my $md5 = Digest::MD5->new;
1906 $md5->addfile($fh);
1907 my $got = $md5->hexdigest;
1908 die "Checksum mismatch: $path\n",
1909 "expected: $exp\n got: $got\n" if ($got ne $exp);
1910 seek($fh, 0, 0) or croak $!;
1911 if ($fb->{mode_b} == 120000) {
1912 read($fh, my $buf, 5) == 5 or croak $!;
1913 $buf eq 'link ' or die "$path has mode 120000",
1914 "but is not a link\n";
1915 }
1916 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
1917 if (!$pid) {
1918 open STDIN, '<&', $fh or croak $!;
1919 exec qw/git-hash-object -w --stdin/ or croak $!;
1920 }
1921 chomp($hash = do { local $/; <$out> });
1922 close $out or croak $!;
1923 close $fh or croak $!;
1924 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
1925 close $fb->{base} or croak $!;
1926 } else {
1927 $hash = $fb->{blob} or die "no blob information\n";
1928 }
1929 $fb->{pool}->clear;
1930 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
1931 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
1932 undef;
1933}
1934sub abort_edit {
1936 my $self = shift;
1937 $self->{nr} = $self->{gii}->{nr};
1938 delete $self->{gii};
1939 $self->SUPER::abort_edit(@_);
1940}
1941sub close_edit {
1943 my $self = shift;
1944 $self->{git_commit_ok} = 1;
1945 $self->{nr} = $self->{gii}->{nr};
1946 delete $self->{gii};
1947 $self->SUPER::close_edit(@_);
1948}
1949package SVN::Git::Editor;
1951use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
1952use strict;
1953use warnings;
1954use Carp qw/croak/;
1955use IO::File;
1956use Digest::MD5;
1957sub new {
1959 my ($class, $opts) = @_;
1960 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
1961 die "$_ required!\n" unless (defined $opts->{$_});
1962 }
1963 my $pool = SVN::Pool->new;
1965 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
1966 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
1967 $opts->{r}, $mods);
1968 # $opts->{ra} functions should not be used after this:
1970 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
1971 $opts->{editor_cb}, $pool);
1972 my $self = SVN::Delta::Editor->new(@ce, $pool);
1973 bless $self, $class;
1974 foreach (qw/svn_path r tree_a tree_b/) {
1975 $self->{$_} = $opts->{$_};
1976 }
1977 $self->{url} = $opts->{ra}->{url};
1978 $self->{mods} = $mods;
1979 $self->{types} = $types;
1980 $self->{pool} = $pool;
1981 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
1982 $self->{rm} = { };
1983 $self->{path_prefix} = length $self->{svn_path} ?
1984 "$self->{svn_path}/" : '';
1985 return $self;
1986}
1987sub generate_diff {
1989 my ($tree_a, $tree_b) = @_;
1990 my @diff_tree = qw(diff-tree -z -r);
1991 if ($_cp_similarity) {
1992 push @diff_tree, "-C$_cp_similarity";
1993 } else {
1994 push @diff_tree, '-C';
1995 }
1996 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1997 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
1998 push @diff_tree, $tree_a, $tree_b;
1999 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2000 local $/ = "\0";
2001 my $state = 'meta';
2002 my @mods;
2003 while (<$diff_fh>) {
2004 chomp $_; # this gets rid of the trailing "\0"
2005 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2006 $::sha1\s($::sha1)\s
2007 ([MTCRAD])\d*$/xo) {
2008 push @mods, { mode_a => $1, mode_b => $2,
2009 sha1_b => $3, chg => $4 };
2010 if ($4 =~ /^(?:C|R)$/) {
2011 $state = 'file_a';
2012 } else {
2013 $state = 'file_b';
2014 }
2015 } elsif ($state eq 'file_a') {
2016 my $x = $mods[$#mods] or croak "Empty array\n";
2017 if ($x->{chg} !~ /^(?:C|R)$/) {
2018 croak "Error parsing $_, $x->{chg}\n";
2019 }
2020 $x->{file_a} = $_;
2021 $state = 'file_b';
2022 } elsif ($state eq 'file_b') {
2023 my $x = $mods[$#mods] or croak "Empty array\n";
2024 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2025 croak "Error parsing $_, $x->{chg}\n";
2026 }
2027 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2028 croak "Error parsing $_, $x->{chg}\n";
2029 }
2030 $x->{file_b} = $_;
2031 $state = 'meta';
2032 } else {
2033 croak "Error parsing $_\n";
2034 }
2035 }
2036 command_close_pipe($diff_fh, $ctx);
2037 \@mods;
2038}
2039sub check_diff_paths {
2041 my ($ra, $pfx, $rev, $mods) = @_;
2042 my %types;
2043 $pfx .= '/' if length $pfx;
2044 sub type_diff_paths {
2046 my ($ra, $types, $path, $rev) = @_;
2047 my @p = split m#/+#, $path;
2048 my $c = shift @p;
2049 unless (defined $types->{$c}) {
2050 $types->{$c} = $ra->check_path($c, $rev);
2051 }
2052 while (@p) {
2053 $c .= '/' . shift @p;
2054 next if defined $types->{$c};
2055 $types->{$c} = $ra->check_path($c, $rev);
2056 }
2057 }
2058 foreach my $m (@$mods) {
2060 foreach my $f (qw/file_a file_b/) {
2061 next unless defined $m->{$f};
2062 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2063 if (length $pfx.$dir && ! defined $types{$dir}) {
2064 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2065 }
2066 }
2067 }
2068 \%types;
2069}
2070sub split_path {
2072 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2073}
2074sub repo_path {
2076 my ($self, $path) = @_;
2077 $self->{path_prefix}.(defined $path ? $path : '');
2078}
2079sub url_path {
2081 my ($self, $path) = @_;
2082 $self->{url} . '/' . $self->repo_path($path);
2083}
2084sub rmdirs {
2086 my ($self) = @_;
2087 my $rm = $self->{rm};
2088 delete $rm->{''}; # we never delete the url we're tracking
2089 return unless %$rm;
2090 foreach (keys %$rm) {
2092 my @d = split m#/#, $_;
2093 my $c = shift @d;
2094 $rm->{$c} = 1;
2095 while (@d) {
2096 $c .= '/' . shift @d;
2097 $rm->{$c} = 1;
2098 }
2099 }
2100 delete $rm->{$self->{svn_path}};
2101 delete $rm->{''}; # we never delete the url we're tracking
2102 return unless %$rm;
2103 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2105 $self->{tree_b});
2106 local $/ = "\0";
2107 while (<$fh>) {
2108 chomp;
2109 my @dn = split m#/#, $_;
2110 while (pop @dn) {
2111 delete $rm->{join '/', @dn};
2112 }
2113 unless (%$rm) {
2114 close $fh;
2115 return;
2116 }
2117 }
2118 command_close_pipe($fh, $ctx);
2119 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2121 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2122 $self->close_directory($bat->{$d}, $p);
2123 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2124 print "\tD+\t$d/\n" unless $::_q;
2125 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2126 delete $bat->{$d};
2127 }
2128}
2129sub open_or_add_dir {
2131 my ($self, $full_path, $baton) = @_;
2132 my $t = $self->{types}->{$full_path};
2133 if (!defined $t) {
2134 die "$full_path not known in r$self->{r} or we have a bug!\n";
2135 }
2136 if ($t == $SVN::Node::none) {
2137 return $self->add_directory($full_path, $baton,
2138 undef, -1, $self->{pool});
2139 } elsif ($t == $SVN::Node::dir) {
2140 return $self->open_directory($full_path, $baton,
2141 $self->{r}, $self->{pool});
2142 }
2143 print STDERR "$full_path already exists in repository at ",
2144 "r$self->{r} and it is not a directory (",
2145 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2146 exit 1;
2147}
2148sub ensure_path {
2150 my ($self, $path) = @_;
2151 my $bat = $self->{bat};
2152 my $repo_path = $self->repo_path($path);
2153 return $bat->{''} unless (length $repo_path);
2154 my @p = split m#/+#, $repo_path;
2155 my $c = shift @p;
2156 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2157 while (@p) {
2158 my $c0 = $c;
2159 $c .= '/' . shift @p;
2160 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2161 }
2162 return $bat->{$c};
2163}
2164sub A {
2166 my ($self, $m) = @_;
2167 my ($dir, $file) = split_path($m->{file_b});
2168 my $pbat = $self->ensure_path($dir);
2169 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2170 undef, -1);
2171 print "\tA\t$m->{file_b}\n" unless $::_q;
2172 $self->chg_file($fbat, $m);
2173 $self->close_file($fbat,undef,$self->{pool});
2174}
2175sub C {
2177 my ($self, $m) = @_;
2178 my ($dir, $file) = split_path($m->{file_b});
2179 my $pbat = $self->ensure_path($dir);
2180 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2181 $self->url_path($m->{file_a}), $self->{r});
2182 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2183 $self->chg_file($fbat, $m);
2184 $self->close_file($fbat,undef,$self->{pool});
2185}
2186sub delete_entry {
2188 my ($self, $path, $pbat) = @_;
2189 my $rpath = $self->repo_path($path);
2190 my ($dir, $file) = split_path($rpath);
2191 $self->{rm}->{$dir} = 1;
2192 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2193}
2194sub R {
2196 my ($self, $m) = @_;
2197 my ($dir, $file) = split_path($m->{file_b});
2198 my $pbat = $self->ensure_path($dir);
2199 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2200 $self->url_path($m->{file_a}), $self->{r});
2201 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2202 $self->chg_file($fbat, $m);
2203 $self->close_file($fbat,undef,$self->{pool});
2204 ($dir, $file) = split_path($m->{file_a});
2206 $pbat = $self->ensure_path($dir);
2207 $self->delete_entry($m->{file_a}, $pbat);
2208}
2209sub M {
2211 my ($self, $m) = @_;
2212 my ($dir, $file) = split_path($m->{file_b});
2213 my $pbat = $self->ensure_path($dir);
2214 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2215 $pbat,$self->{r},$self->{pool});
2216 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2217 $self->chg_file($fbat, $m);
2218 $self->close_file($fbat,undef,$self->{pool});
2219}
2220sub T { shift->M(@_) }
2222sub change_file_prop {
2224 my ($self, $fbat, $pname, $pval) = @_;
2225 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2226}
2227sub chg_file {
2229 my ($self, $fbat, $m) = @_;
2230 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2231 $self->change_file_prop($fbat,'svn:executable','*');
2232 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2233 $self->change_file_prop($fbat,'svn:executable',undef);
2234 }
2235 my $fh = IO::File->new_tmpfile or croak $!;
2236 if ($m->{mode_b} =~ /^120/) {
2237 print $fh 'link ' or croak $!;
2238 $self->change_file_prop($fbat,'svn:special','*');
2239 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2240 $self->change_file_prop($fbat,'svn:special',undef);
2241 }
2242 defined(my $pid = fork) or croak $!;
2243 if (!$pid) {
2244 open STDOUT, '>&', $fh or croak $!;
2245 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2246 }
2247 waitpid $pid, 0;
2248 croak $? if $?;
2249 $fh->flush == 0 or croak $!;
2250 seek $fh, 0, 0 or croak $!;
2251 my $md5 = Digest::MD5->new;
2253 $md5->addfile($fh) or croak $!;
2254 seek $fh, 0, 0 or croak $!;
2255 my $exp = $md5->hexdigest;
2257 my $pool = SVN::Pool->new;
2258 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2259 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2260 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2261 $pool->clear;
2262 close $fh or croak $!;
2264}
2265sub D {
2267 my ($self, $m) = @_;
2268 my ($dir, $file) = split_path($m->{file_b});
2269 my $pbat = $self->ensure_path($dir);
2270 print "\tD\t$m->{file_b}\n" unless $::_q;
2271 $self->delete_entry($m->{file_b}, $pbat);
2272}
2273sub close_edit {
2275 my ($self) = @_;
2276 my ($p,$bat) = ($self->{pool}, $self->{bat});
2277 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2278 $self->close_directory($bat->{$_}, $p);
2279 }
2280 $self->SUPER::close_edit($p);
2281 $p->clear;
2282}
2283sub abort_edit {
2285 my ($self) = @_;
2286 $self->SUPER::abort_edit($self->{pool});
2287}
2288sub DESTROY {
2290 my $self = shift;
2291 $self->SUPER::DESTROY(@_);
2292 $self->{pool}->clear;
2293}
2294# this drives the editor
2296sub apply_diff {
2297 my ($self) = @_;
2298 my $mods = $self->{mods};
2299 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2300 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2301 my $f = $m->{chg};
2302 if (defined $o{$f}) {
2303 $self->$f($m);
2304 } else {
2305 fatal("Invalid change type: $f\n");
2306 }
2307 }
2308 $self->rmdirs if $_rmdir;
2309 if (@$mods == 0) {
2310 $self->abort_edit;
2311 } else {
2312 $self->close_edit;
2313 }
2314 return scalar @$mods;
2315}
2316package Git::SVN::Ra;
2318use vars qw/@ISA $config_dir/;
2319use strict;
2320use warnings;
2321my ($can_do_switch);
2322my $RA;
2323BEGIN {
2325 # enforce temporary pool usage for some simple functions
2326 my $e;
2327 foreach (qw/get_latest_revnum rev_proplist get_file
2328 check_path get_dir get_uuid get_repos_root/) {
2329 $e .= "sub $_ {
2330 my \$self = shift;
2331 my \$pool = SVN::Pool->new;
2332 my \@ret = \$self->SUPER::$_(\@_,\$pool);
2333 \$pool->clear;
2334 wantarray ? \@ret : \$ret[0]; }\n";
2335 }
2336 eval $e;
2337}
2338sub new {
2340 my ($class, $url) = @_;
2341 $url =~ s!/+$!!;
2342 return $RA if ($RA && $RA->{url} eq $url);
2343 SVN::_Core::svn_config_ensure($config_dir, undef);
2345 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2346 SVN::Client::get_simple_provider(),
2347 SVN::Client::get_ssl_server_trust_file_provider(),
2348 SVN::Client::get_simple_prompt_provider(
2349 \&Git::SVN::Prompt::simple, 2),
2350 SVN::Client::get_ssl_client_cert_prompt_provider(
2351 \&Git::SVN::Prompt::ssl_client_cert, 2),
2352 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2353 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2354 SVN::Client::get_username_provider(),
2355 SVN::Client::get_ssl_server_trust_prompt_provider(
2356 \&Git::SVN::Prompt::ssl_server_trust),
2357 SVN::Client::get_username_prompt_provider(
2358 \&Git::SVN::Prompt::username, 2),
2359 ]);
2360 my $config = SVN::Core::config_get_config($config_dir);
2361 my $self = SVN::Ra->new(url => $url, auth => $baton,
2362 config => $config,
2363 pool => SVN::Pool->new,
2364 auth_provider_callbacks => $callbacks);
2365 $self->{svn_path} = $url;
2366 $self->{repos_root} = $self->get_repos_root;
2367 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E/*##;
2368 $RA = bless $self, $class;
2369}
2370sub DESTROY {
2372 # do not call the real DESTROY since we store ourselves in $RA
2373}
2374sub get_log {
2376 my ($self, @args) = @_;
2377 my $pool = SVN::Pool->new;
2378 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2379 my $ret = $self->SUPER::get_log(@args, $pool);
2380 $pool->clear;
2381 $ret;
2382}
2383sub get_commit_editor {
2385 my ($self, $log, $cb, $pool) = @_;
2386 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2387 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2388}
2389sub uuid {
2391 my ($self) = @_;
2392 $self->{uuid} ||= $self->get_uuid;
2393}
2394sub gs_do_update {
2396 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2397 my $new = ($rev_a == $rev_b);
2398 my $path = $gs->{path};
2399 my $pool = SVN::Pool->new;
2401 $editor->set_path_strip($path);
2402 my (@pc) = split m#/#, $path;
2403 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2404 1, $editor, $pool);
2405 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2406 # Since we can't rely on svn_ra_reparent being available, we'll
2408 # just have to do some magic with set_path to make it so
2409 # we only want a partial path.
2410 my $sp = '';
2411 my $final = join('/', @pc);
2412 while (@pc) {
2413 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2414 $sp .= '/' if length $sp;
2415 $sp .= shift @pc;
2416 }
2417 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2418 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2420 $reporter->finish_report($pool);
2422 $pool->clear;
2423 $editor->{git_commit_ok};
2424}
2425# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2427# svn_ra_reparent didn't work before 1.4)
2428sub gs_do_switch {
2429 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
2430 my $path = $gs->{path};
2431 my $pool = SVN::Pool->new;
2432 my $full_url = $self->{url};
2434 my $old_url = $full_url;
2435 $full_url .= "/$path" if length $path;
2436 my ($ra, $reparented);
2437 if ($old_url ne $full_url) {
2438 if ($old_url !~ m#^svn(\+ssh)?://#) {
2439 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2440 $pool);
2441 $self->{url} = $full_url;
2442 $reparented = 1;
2443 } else {
2444 $ra = Git::SVN::Ra->new($full_url);
2445 }
2446 }
2447 $ra ||= $self;
2448 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
2449 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2450 $reporter->set_path('', $rev_a, 0, @lock, $pool);
2451 $reporter->finish_report($pool);
2452 if ($reparented) {
2454 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2455 $self->{url} = $old_url;
2456 }
2457 $pool->clear;
2459 $editor->{git_commit_ok};
2460}
2461sub gs_fetch_loop_common {
2463 my ($self, $base, $head, $gsv, $globs) = @_;
2464 return if ($base > $head);
2465 my $inc = 1000;
2466 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2467 my %common;
2468 my $common_max = scalar @$gsv;
2469 foreach my $gs (@$gsv) {
2471 if (my $last_commit = $gs->last_commit) {
2472 $gs->assert_index_clean($last_commit);
2473 }
2474 my @tmp = split m#/#, $gs->{path};
2475 my $p = '';
2476 foreach (@tmp) {
2477 $p .= length($p) ? "/$_" : $_;
2478 $common{$p} ||= 0;
2479 $common{$p}++;
2480 }
2481 }
2482 $globs ||= [];
2483 $common_max += scalar @$globs;
2484 foreach my $glob (@$globs) {
2485 my @tmp = split m#/#, $glob->{path}->{left};
2486 my $p = '';
2487 foreach (@tmp) {
2488 $p .= length($p) ? "/$_" : $_;
2489 $common{$p} ||= 0;
2490 $common{$p}++;
2491 }
2492 }
2493 my $longest_path = '';
2495 foreach (sort {length $b <=> length $a} keys %common) {
2496 if ($common{$_} == $common_max) {
2497 $longest_path = $_;
2498 last;
2499 }
2500 }
2501 while (1) {
2502 my %revs;
2503 my $err;
2504 my $err_handler = $SVN::Error::handler;
2505 $SVN::Error::handler = sub {
2506 ($err) = @_;
2507 skip_unknown_revs($err);
2508 };
2509 sub _cb {
2510 my ($paths, $r, $author, $date, $log) = @_;
2511 [ dup_changed_paths($paths),
2512 { author => $author, date => $date, log => $log } ];
2513 }
2514 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
2515 sub { $revs{$_[1]} = _cb(@_) });
2516 if ($err && $max >= $head) {
2517 print STDERR "Path '$longest_path' ",
2518 "was probably deleted:\n",
2519 $err->expanded_message,
2520 "\nWill attempt to follow ",
2521 "revisions r$min .. r$max ",
2522 "committed before the deletion\n";
2523 my $hi = $max;
2524 while (--$hi >= $min) {
2525 my $ok;
2526 $self->get_log([$longest_path], $min, $hi,
2527 0, 1, 1, sub {
2528 $ok ||= $_[1];
2529 $revs{$_[1]} = _cb(@_) });
2530 if ($ok) {
2531 print STDERR "r$min .. r$ok OK\n";
2532 last;
2533 }
2534 }
2535 }
2536 $SVN::Error::handler = $err_handler;
2537 my %exists = map { $_->{path} => $_ } @$gsv;
2539 foreach my $r (sort {$a <=> $b} keys %revs) {
2540 my ($paths, $logged) = @{$revs{$r}};
2541 foreach my $gs ($self->match_globs(\%exists, $paths,
2543 $globs, $r)) {
2544 if ($gs->rev_db_max >= $r) {
2545 next;
2546 }
2547 next unless $gs->match_paths($paths, $r);
2548 $gs->{logged_rev_props} = $logged;
2549 my $log_entry = $gs->do_fetch($paths, $r);
2550 if ($log_entry) {
2551 $gs->do_git_commit($log_entry);
2552 }
2553 }
2554 foreach my $g (@$globs) {
2555 my $f = "$ENV{GIT_DIR}/svn/." .
2556 $self->uuid . ".$g->{t}";
2557 open my $fh, '>', "$f.tmp" or
2558 die "Can't open $f.tmp for writing: $!";
2559 print $fh "$r\n" or
2560 die "Couldn't write to $f: $!\n";
2561 close $fh or die "Error closing $f: $!\n";
2562 rename "$f.tmp", $f or
2563 die "Couldn't rename ",
2564 "$f.tmp => $f: $!\n";
2565 }
2566 }
2567 # pre-fill the .rev_db since it'll eventually get filled in
2568 # with '0' x40 if something new gets committed
2569 foreach my $gs (@$gsv) {
2570 next if defined $gs->rev_db_get($max);
2571 $gs->rev_db_set($max, 0 x40);
2572 }
2573 last if $max >= $head;
2574 $min = $max + 1;
2575 $max += $inc;
2576 $max = $head if ($max > $head);
2577 }
2578}
2579sub match_globs {
2581 my ($self, $exists, $paths, $globs, $r) = @_;
2582 foreach my $g (@$globs) {
2583 foreach (keys %$paths) {
2584 next unless /$g->{path}->{regex}/;
2585 my $p = $1;
2586 my $pathname = $g->{path}->full_path($p);
2587 next if $exists->{$pathname};
2588 $exists->{$pathname} = Git::SVN->init(
2589 $self->{url}, $pathname, undef,
2590 $g->{ref}->full_path($p), 1);
2591 }
2592 my $c = '';
2593 foreach (split m#/#, $g->{path}->{left}) {
2594 $c .= "/$_";
2595 next unless ($paths->{$c} &&
2596 ($paths->{$c}->{action} eq 'A'));
2597 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
2598 next unless scalar @x == 3;
2599 my $dirents = $x[0];
2600 foreach my $de (keys %$dirents) {
2601 next if $dirents->{$de}->kind !=
2602 $SVN::Node::dir;
2603 my $p = $g->{path}->full_path($de);
2604 next if $exists->{$p};
2605 next if (length $g->{path}->{right} &&
2606 ($self->check_path($p, $r) !=
2607 $SVN::Node::dir));
2608 $exists->{$p} = Git::SVN->init($self->{url},
2609 $p, undef,
2610 $g->{ref}->full_path($de), 1);
2611 }
2612 }
2613 }
2614 values %$exists;
2615}
2616sub minimize_url {
2618 my ($self) = @_;
2619 return $self->{url} if ($self->{url} eq $self->{repos_root});
2620 my $url = $self->{repos_root};
2621 my @components = split(m!/!, $self->{svn_path});
2622 my $c = '';
2623 do {
2624 $url .= "/$c" if length $c;
2625 eval { (ref $self)->new($url)->get_latest_revnum };
2626 } while ($@ && ($c = shift @components));
2627 $url;
2628}
2629sub can_do_switch {
2631 my $self = shift;
2632 unless (defined $can_do_switch) {
2633 my $pool = SVN::Pool->new;
2634 my $rep = eval {
2635 $self->do_switch(1, '', 0, $self->{url},
2636 SVN::Delta::Editor->new, $pool);
2637 };
2638 if ($@) {
2639 $can_do_switch = 0;
2640 } else {
2641 $rep->abort_report($pool);
2642 $can_do_switch = 1;
2643 }
2644 $pool->clear;
2645 }
2646 $can_do_switch;
2647}
2648sub skip_unknown_revs {
2650 my ($err) = @_;
2651 my $errno = $err->apr_err();
2652 # Maybe the branch we're tracking didn't
2653 # exist when the repo started, so it's
2654 # not an error if it doesn't, just continue
2655 #
2656 # Wonderfully consistent library, eh?
2657 # 160013 - svn:// and file://
2658 # 175002 - http(s)://
2659 # 175007 - http(s):// (this repo required authorization, too...)
2660 # More codes may be discovered later...
2661 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2662 return;
2663 }
2664 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2665}
2666# svn_log_changed_path_t objects passed to get_log are likely to be
2668# overwritten even if only the refs are copied to an external variable,
2669# so we should dup the structures in their entirety. Using an externally
2670# passed pool (instead of our temporary and quickly cleared pool in
2671# Git::SVN::Ra) does not help matters at all...
2672sub dup_changed_paths {
2673 my ($paths) = @_;
2674 return undef unless $paths;
2675 my %ret;
2676 foreach my $p (keys %$paths) {
2677 my $i = $paths->{$p};
2678 my %s = map { $_ => $i->$_ }
2679 qw/copyfrom_path copyfrom_rev action/;
2680 $ret{$p} = \%s;
2681 }
2682 \%ret;
2683}
2684package Git::SVN::Log;
2686use strict;
2687use warnings;
2688use POSIX qw/strftime/;
2689use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2690 %rusers $show_commit $incremental/;
2691my $l_fmt;
2692sub cmt_showable {
2694 my ($c) = @_;
2695 return 1 if defined $c->{r};
2696 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2697 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2698 my @log = command(qw/cat-file commit/, $c->{c});
2699 shift @log while ($log[0] ne "\n");
2700 shift @log;
2701 @{$c->{l}} = grep !/^git-svn-id: /, @log;
2702 (undef, $c->{r}, undef) = ::extract_metadata(
2704 (grep(/^git-svn-id: /, @log))[-1]);
2705 }
2706 return defined $c->{r};
2707}
2708sub log_use_color {
2710 return 1 if $color;
2711 my ($dc, $dcvar);
2712 $dcvar = 'color.diff';
2713 $dc = `git-config --get $dcvar`;
2714 if ($dc eq '') {
2715 # nothing at all; fallback to "diff.color"
2716 $dcvar = 'diff.color';
2717 $dc = `git-config --get $dcvar`;
2718 }
2719 chomp($dc);
2720 if ($dc eq 'auto') {
2721 my $pc;
2722 $pc = `git-config --get color.pager`;
2723 if ($pc eq '') {
2724 # does not have it -- fallback to pager.color
2725 $pc = `git-config --bool --get pager.color`;
2726 }
2727 else {
2728 $pc = `git-config --bool --get color.pager`;
2729 if ($?) {
2730 $pc = 'false';
2731 }
2732 }
2733 chomp($pc);
2734 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2735 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2736 }
2737 return 0;
2738 }
2739 return 0 if $dc eq 'never';
2740 return 1 if $dc eq 'always';
2741 chomp($dc = `git-config --bool --get $dcvar`);
2742 return ($dc eq 'true');
2743}
2744sub git_svn_log_cmd {
2746 my ($r_min, $r_max) = @_;
2747 my $gs = Git::SVN->_new;
2748 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2749 $gs->refname);
2750 push @cmd, '-r' unless $non_recursive;
2751 push @cmd, qw/--raw --name-status/ if $verbose;
2752 push @cmd, '--color' if log_use_color();
2753 return @cmd unless defined $r_max;
2754 if ($r_max == $r_min) {
2755 push @cmd, '--max-count=1';
2756 if (my $c = $gs->rev_db_get($r_max)) {
2757 push @cmd, $c;
2758 }
2759 } else {
2760 my ($c_min, $c_max);
2761 $c_max = $gs->rev_db_get($r_max);
2762 $c_min = $gs->rev_db_get($r_min);
2763 if (defined $c_min && defined $c_max) {
2764 if ($r_max > $r_max) {
2765 push @cmd, "$c_min..$c_max";
2766 } else {
2767 push @cmd, "$c_max..$c_min";
2768 }
2769 } elsif ($r_max > $r_min) {
2770 push @cmd, $c_max;
2771 } else {
2772 push @cmd, $c_min;
2773 }
2774 }
2775 return @cmd;
2776}
2777# adapted from pager.c
2779sub config_pager {
2780 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2781 if (!defined $pager) {
2782 $pager = 'less';
2783 } elsif (length $pager == 0 || $pager eq 'cat') {
2784 $pager = undef;
2785 }
2786}
2787sub run_pager {
2789 return unless -t *STDOUT;
2790 pipe my $rfd, my $wfd or return;
2791 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2792 if (!$pid) {
2793 open STDOUT, '>&', $wfd or
2794 ::fatal "Can't redirect to stdout: $!\n";
2795 return;
2796 }
2797 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2798 $ENV{LESS} ||= 'FRSX';
2799 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2800}
2801sub tz_to_s_offset {
2803 my ($tz) = @_;
2804 $tz =~ s/(\d\d)$//;
2805 return ($1 * 60) + ($tz * 3600);
2806}
2807sub get_author_info {
2809 my ($dest, $author, $t, $tz) = @_;
2810 $author =~ s/(?:^\s*|\s*$)//g;
2811 $dest->{a_raw} = $author;
2812 my $au;
2813 if ($::_authors) {
2814 $au = $rusers{$author} || undef;
2815 }
2816 if (!$au) {
2817 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2818 }
2819 $dest->{t} = $t;
2820 $dest->{tz} = $tz;
2821 $dest->{a} = $au;
2822 # Date::Parse isn't in the standard Perl distro :(
2823 if ($tz =~ s/^\+//) {
2824 $t += tz_to_s_offset($tz);
2825 } elsif ($tz =~ s/^\-//) {
2826 $t -= tz_to_s_offset($tz);
2827 }
2828 $dest->{t_utc} = $t;
2829}
2830sub process_commit {
2832 my ($c, $r_min, $r_max, $defer) = @_;
2833 if (defined $r_min && defined $r_max) {
2834 if ($r_min == $c->{r} && $r_min == $r_max) {
2835 show_commit($c);
2836 return 0;
2837 }
2838 return 1 if $r_min == $r_max;
2839 if ($r_min < $r_max) {
2840 # we need to reverse the print order
2841 return 0 if (defined $limit && --$limit < 0);
2842 push @$defer, $c;
2843 return 1;
2844 }
2845 if ($r_min != $r_max) {
2846 return 1 if ($r_min < $c->{r});
2847 return 1 if ($r_max > $c->{r});
2848 }
2849 }
2850 return 0 if (defined $limit && --$limit < 0);
2851 show_commit($c);
2852 return 1;
2853}
2854sub show_commit {
2856 my $c = shift;
2857 if ($oneline) {
2858 my $x = "\n";
2859 if (my $l = $c->{l}) {
2860 while ($l->[0] =~ /^\s*$/) { shift @$l }
2861 $x = $l->[0];
2862 }
2863 $l_fmt ||= 'A' . length($c->{r});
2864 print 'r',pack($l_fmt, $c->{r}),' | ';
2865 print "$c->{c} | " if $show_commit;
2866 print $x;
2867 } else {
2868 show_commit_normal($c);
2869 }
2870}
2871sub show_commit_changed_paths {
2873 my ($c) = @_;
2874 return unless $c->{changed};
2875 print "Changed paths:\n", @{$c->{changed}};
2876}
2877sub show_commit_normal {
2879 my ($c) = @_;
2880 print '-' x72, "\nr$c->{r} | ";
2881 print "$c->{c} | " if $show_commit;
2882 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2883 localtime($c->{t_utc})), ' | ';
2884 my $nr_line = 0;
2885 if (my $l = $c->{l}) {
2887 while ($l->[$#$l] eq "\n" && $#$l > 0
2888 && $l->[($#$l - 1)] eq "\n") {
2889 pop @$l;
2890 }
2891 $nr_line = scalar @$l;
2892 if (!$nr_line) {
2893 print "1 line\n\n\n";
2894 } else {
2895 if ($nr_line == 1) {
2896 $nr_line = '1 line';
2897 } else {
2898 $nr_line .= ' lines';
2899 }
2900 print $nr_line, "\n";
2901 show_commit_changed_paths($c);
2902 print "\n";
2903 print $_ foreach @$l;
2904 }
2905 } else {
2906 print "1 line\n";
2907 show_commit_changed_paths($c);
2908 print "\n";
2909 }
2911 foreach my $x (qw/raw diff/) {
2912 if ($c->{$x}) {
2913 print "\n";
2914 print $_ foreach @{$c->{$x}}
2915 }
2916 }
2917}
2918sub cmd_show_log {
2920 my (@args) = @_;
2921 my ($r_min, $r_max);
2922 my $r_last = -1; # prevent dupes
2923 if (defined $TZ) {
2924 $ENV{TZ} = $TZ;
2925 } else {
2926 delete $ENV{TZ};
2927 }
2928 if (defined $::_revision) {
2929 if ($::_revision =~ /^(\d+):(\d+)$/) {
2930 ($r_min, $r_max) = ($1, $2);
2931 } elsif ($::_revision =~ /^\d+$/) {
2932 $r_min = $r_max = $::_revision;
2933 } else {
2934 ::fatal "-r$::_revision is not supported, use ",
2935 "standard \'git log\' arguments instead\n";
2936 }
2937 }
2938 config_pager();
2940 @args = (git_svn_log_cmd($r_min, $r_max), @args);
2941 my $log = command_output_pipe(@args);
2942 run_pager();
2943 my (@k, $c, $d);
2944 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
2945 while (<$log>) {
2946 if (/^${esc_color}commit ($::sha1_short)/o) {
2947 my $cmt = $1;
2948 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
2949 $r_last = $c->{r};
2950 process_commit($c, $r_min, $r_max, \@k) or
2951 goto out;
2952 }
2953 $d = undef;
2954 $c = { c => $cmt };
2955 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
2956 get_author_info($c, $1, $2, $3);
2957 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
2958 # ignore
2959 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
2960 push @{$c->{raw}}, $_;
2961 } elsif (/^${esc_color}[ACRMDT]\t/) {
2962 # we could add $SVN->{svn_path} here, but that requires
2963 # remote access at the moment (repo_path_split)...
2964 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
2965 push @{$c->{changed}}, $_;
2966 } elsif (/^${esc_color}diff /o) {
2967 $d = 1;
2968 push @{$c->{diff}}, $_;
2969 } elsif ($d) {
2970 push @{$c->{diff}}, $_;
2971 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
2972 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
2973 } elsif (s/^${esc_color} //o) {
2974 push @{$c->{l}}, $_;
2975 }
2976 }
2977 if ($c && defined $c->{r} && $c->{r} != $r_last) {
2978 $r_last = $c->{r};
2979 process_commit($c, $r_min, $r_max, \@k);
2980 }
2981 if (@k) {
2982 my $swap = $r_max;
2983 $r_max = $r_min;
2984 $r_min = $swap;
2985 process_commit($_, $r_min, $r_max) foreach reverse @k;
2986 }
2987out:
2988 close $log;
2989 print '-' x72,"\n" unless $incremental || $oneline;
2990}
2991package Git::SVN::Migration;
2993# these version numbers do NOT correspond to actual version numbers
2994# of git nor git-svn. They are just relative.
2995#
2996# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
2997#
2998# v1 layout: .git/$id/info/url, refs/remotes/$id
2999#
3000# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3001#
3002# v3 layout: .git/svn/$id, refs/remotes/$id
3003# - info/url may remain for backwards compatibility
3004# - this is what we migrate up to this layout automatically,
3005# - this will be used by git svn init on single branches
3006#
3007# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3008# - this is only created for newly multi-init-ed
3009# repositories. Similar in spirit to the
3010# --use-separate-remotes option in git-clone (now default)
3011# - we do not automatically migrate to this (following
3012# the example set by core git)
3013use strict;
3014use warnings;
3015use Carp qw/croak/;
3016use File::Path qw/mkpath/;
3017use File::Basename qw/dirname basename/;
3018use vars qw/$_minimize/;
3019sub migrate_from_v0 {
3021 my $git_dir = $ENV{GIT_DIR};
3022 return undef unless -d $git_dir;
3023 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3024 my $migrated = 0;
3025 while (<$fh>) {
3026 chomp;
3027 my ($id, $orig_ref) = ($_, $_);
3028 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3029 next unless -f "$git_dir/$id/info/url";
3030 my $new_ref = "refs/remotes/$id";
3031 if (::verify_ref("$new_ref^0")) {
3032 print STDERR "W: $orig_ref is probably an old ",
3033 "branch used by an ancient version of ",
3034 "git-svn.\n",
3035 "However, $new_ref also exists.\n",
3036 "We will not be able ",
3037 "to use this branch until this ",
3038 "ambiguity is resolved.\n";
3039 next;
3040 }
3041 print STDERR "Migrating from v0 layout...\n" if !$migrated;
3042 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3043 command_noisy('update-ref', $new_ref, $orig_ref);
3044 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3045 $migrated++;
3046 }
3047 command_close_pipe($fh, $ctx);
3048 print STDERR "Done migrating from v0 layout...\n" if $migrated;
3049 $migrated;
3050}
3051sub migrate_from_v1 {
3053 my $git_dir = $ENV{GIT_DIR};
3054 my $migrated = 0;
3055 return $migrated unless -d $git_dir;
3056 my $svn_dir = "$git_dir/svn";
3057 # just in case somebody used 'svn' as their $id at some point...
3059 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3060 print STDERR "Migrating from a git-svn v1 layout...\n";
3062 mkpath([$svn_dir]);
3063 print STDERR "Data from a previous version of git-svn exists, but\n\t",
3064 "$svn_dir\n\t(required for this version ",
3065 "($::VERSION) of git-svn) does not. exist\n";
3066 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3067 while (<$fh>) {
3068 my $x = $_;
3069 next unless $x =~ s#^refs/remotes/##;
3070 chomp $x;
3071 next unless -f "$git_dir/$x/info/url";
3072 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3073 next unless $u;
3074 my $dn = dirname("$git_dir/svn/$x");
3075 mkpath([$dn]) unless -d $dn;
3076 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3077 mkpath(["$git_dir/svn/svn"]);
3078 print STDERR " - $git_dir/$x/info => ",
3079 "$git_dir/svn/$x/info\n";
3080 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3081 croak "$!: $x";
3082 # don't worry too much about these, they probably
3083 # don't exist with repos this old (save for index,
3084 # and we can easily regenerate that)
3085 foreach my $f (qw/unhandled.log index .rev_db/) {
3086 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3087 }
3088 } else {
3089 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3090 rename "$git_dir/$x", "$git_dir/svn/$x" or
3091 croak "$!: $x";
3092 }
3093 $migrated++;
3094 }
3095 command_close_pipe($fh, $ctx);
3096 print STDERR "Done migrating from a git-svn v1 layout\n";
3097 $migrated;
3098}
3099sub read_old_urls {
3101 my ($l_map, $pfx, $path) = @_;
3102 my @dir;
3103 foreach (<$path/*>) {
3104 if (-r "$_/info/url") {
3105 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3106 my $ref_id = $pfx . basename $_;
3107 my $url = ::file_to_s("$_/info/url");
3108 $l_map->{$ref_id} = $url;
3109 } elsif (-d $_) {
3110 push @dir, $_;
3111 }
3112 }
3113 foreach (@dir) {
3114 my $x = $_;
3115 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3116 read_old_urls($l_map, $x, $_);
3117 }
3118}
3119sub migrate_from_v2 {
3121 my @cfg = command(qw/config -l/);
3122 return if grep /^svn-remote\..+\.url=/, @cfg;
3123 my %l_map;
3124 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3125 my $migrated = 0;
3126 foreach my $ref_id (sort keys %l_map) {
3128 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3129 if ($@) {
3130 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3131 }
3132 $migrated++;
3133 }
3134 $migrated;
3135}
3136sub minimize_connections {
3138 my $r = Git::SVN::read_all_remotes();
3139 my $new_urls = {};
3140 my $root_repos = {};
3141 foreach my $repo_id (keys %$r) {
3142 my $url = $r->{$repo_id}->{url} or next;
3143 my $fetch = $r->{$repo_id}->{fetch} or next;
3144 my $ra = Git::SVN::Ra->new($url);
3145 # skip existing cases where we already connect to the root
3147 if (($ra->{url} eq $ra->{repos_root}) ||
3148 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3149 $repo_id)) {
3150 $root_repos->{$ra->{url}} = $repo_id;
3151 next;
3152 }
3153 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3155 my $root_path = $ra->{url};
3156 $root_path =~ s#^\Q$ra->{repos_root}\E/*##;
3157 foreach my $path (keys %$fetch) {
3158 my $ref_id = $fetch->{$path};
3159 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3160 # make sure we can read when connecting to
3162 # a higher level of a repository
3163 my ($last_rev, undef) = $gs->last_rev_commit;
3164 if (!defined $last_rev) {
3165 $last_rev = eval {
3166 $root_ra->get_latest_revnum;
3167 };
3168 next if $@;
3169 }
3170 my $new = $root_path;
3171 $new .= length $path ? "/$path" : '';
3172 eval {
3173 $root_ra->get_log([$new], $last_rev, $last_rev,
3174 0, 0, 1, sub { });
3175 };
3176 next if $@;
3177 $new_urls->{$ra->{repos_root}}->{$new} =
3178 { ref_id => $ref_id,
3179 old_repo_id => $repo_id,
3180 old_path => $path };
3181 }
3182 }
3183 my @emptied;
3185 foreach my $url (keys %$new_urls) {
3186 # see if we can re-use an existing [svn-remote "repo_id"]
3187 # instead of creating a(n ugly) new section:
3188 my $repo_id = $root_repos->{$url} ||
3189 Git::SVN::sanitize_remote_name($url);
3190 my $fetch = $new_urls->{$url};
3192 foreach my $path (keys %$fetch) {
3193 my $x = $fetch->{$path};
3194 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3195 my $pfx = "svn-remote.$x->{old_repo_id}";
3196 my $old_fetch = quotemeta("$x->{old_path}:".
3198 "refs/remotes/$x->{ref_id}");
3199 command_noisy(qw/config --unset/,
3200 "$pfx.fetch", '^'. $old_fetch . '$');
3201 delete $r->{$x->{old_repo_id}}->
3202 {fetch}->{$x->{old_path}};
3203 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3204 command_noisy(qw/config --unset/,
3205 "$pfx.url");
3206 push @emptied, $x->{old_repo_id}
3207 }
3208 }
3209 }
3210 if (@emptied) {
3211 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3212 "$ENV{GIT_DIR}/config";
3213 print STDERR <<EOF;
3214The following [svn-remote] sections in your config file ($file) are empty
3215and can be safely removed:
3216EOF
3217 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3218 }
3219}
3220sub migration_check {
3222 migrate_from_v0();
3223 migrate_from_v1();
3224 migrate_from_v2();
3225 minimize_connections() if $_minimize;
3226}
3227package Git::IndexInfo;
3229use strict;
3230use warnings;
3231use Git qw/command_input_pipe command_close_pipe/;
3232sub new {
3234 my ($class) = @_;
3235 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3236 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3237}
3238sub remove {
3240 my ($self, $path) = @_;
3241 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3242 return ++$self->{nr};
3243 }
3244 undef;
3245}
3246sub update {
3248 my ($self, $mode, $hash, $path) = @_;
3249 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3250 return ++$self->{nr};
3251 }
3252 undef;
3253}
3254sub DESTROY {
3256 my ($self) = @_;
3257 command_close_pipe($self->{gui}, $self->{ctx});
3258}
3259package Git::SVN::GlobSpec;
3261use strict;
3262use warnings;
3263sub new {
3265 my ($class, $glob) = @_;
3266 my $re = $glob;
3267 $re =~ s!/+$!!g; # no need for trailing slashes
3268 my $nr = ($re =~ s!^(.*/?)\*(/?.*)$!\(\[^/\]+\)!g);
3269 my ($left, $right) = ($1, $2);
3270 if ($nr > 1) {
3271 die "Only one '*' wildcard expansion ",
3272 "is supported (got $nr): '$glob'\n";
3273 } elsif ($nr == 0) {
3274 die "One '*' is needed for glob: '$glob'\n";
3275 }
3276 $re = quotemeta($left) . $re . quotemeta($right);
3277 $left =~ s!/+$!!g;
3278 $right =~ s!^/+!!g;
3279 bless { left => $left, right => $right,
3280 regex => qr/$re/, glob => $glob }, $class;
3281}
3282sub full_path {
3284 my ($self, $path) = @_;
3285 return (length $self->{left} ? "$self->{left}/" : '') .
3286 $path . (length $self->{right} ? "/$self->{right}" : '');
3287}
3288__END__
3290Data structures:
3292$remotes = { # returned by read_all_remotes()
3295 'svn' => {
3296 # svn-remote.svn.url=https://svn.musicpd.org
3297 url => 'https://svn.musicpd.org',
3298 # svn-remote.svn.fetch=mpd/trunk:trunk
3299 fetch => {
3300 'mpd/trunk' => 'trunk',
3301 },
3302 # svn-remote.svn.tags=mpd/tags/*:tags/*
3303 tags => {
3304 path => {
3305 left => 'mpd/tags',
3306 right => '',
3307 regex => qr!mpd/tags/([^/]+)$!,
3308 glob => 'tags/*',
3309 },
3310 ref => {
3311 left => 'tags',
3312 right => '',
3313 regex => qr!tags/([^/]+)$!,
3314 glob => 'tags/*',
3315 },
3316 }
3317 }
3318};
3319$log_entry hashref as returned by libsvn_log_entry()
3321{
3322 log => 'whitespace-formatted log entry
3323', # trailing newline is preserved
3324 revision => '8', # integer
3325 date => '2004-02-24T17:01:44.108345Z', # commit date
3326 author => 'committer name'
3327};
3328# this is generated by generate_diff();
3331@mods = array of diff-index line hashes, each element represents one line
3332 of diff-index output
3333diff-index line ($m hash)
3335{
3336 mode_a => first column of diff-index output, no leading ':',
3337 mode_b => second column of diff-index output,
3338 sha1_b => sha1sum of the final blob,
3339 chg => change type [MCRADT],
3340 file_a => original file name of a file (iff chg is 'C' or 'R')
3341 file_b => new/current file name of a file (any chg)
3342}
3343;
3344# retval of read_url_paths{,_all}();
3346$l_map = {
3347 # repository root url
3348 'https://svn.musicpd.org' => {
3349 # repository path # GIT_SVN_ID
3350 'mpd/trunk' => 'trunk',
3351 'mpd/tags/0.11.5' => 'tags/0.11.5',
3352 },
3353}
3354Notes:
3356 I don't trust the each() function on unless I created %hash myself
3357 because the internal iterator may not have started at base.