?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-svn: implement auto-discovery of branches/tags
[git.git] / git-svn.perl
blob21d53054f6638198c47af64f22a5fcdab3cea249
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision
8 $_q $_authors %users/;
9 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10 $VERSION = '@@GIT_VERSION@@';
11
12 $ENV{GIT_DIR} ||= '.git';
13 $Git::SVN::default_repo_id = 'svn';
14 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
15
16 $Git::SVN::Log::TZ = $ENV{TZ};
17 $ENV{TZ} = 'UTC';
18 $| = 1; # unbuffer STDOUT
19
20 sub fatal (@) { print STDERR @_; exit 1 }
21 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
22 require SVN::Ra;
23 require SVN::Delta;
24 if ($SVN::Core::VERSION lt '1.1.0') {
25 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
26 }
27 push @Git::SVN::Ra::ISA, 'SVN::Ra';
28 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
29 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
30 use Carp qw/croak/;
31 use IO::File qw//;
32 use File::Basename qw/dirname basename/;
33 use File::Path qw/mkpath/;
34 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
35 use IPC::Open3;
36 use Git;
37
38 BEGIN {
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 $s;
47 }
48
49 my ($SVN);
50
51 my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
52 $sha1 = qr/[a-f\d]{40}/;
53 $sha1_short = qr/[a-f\d]{4,40}/;
54 my ($_stdin, $_help, $_edit,
55 $_message, $_file,
56 $_template, $_shared,
57 $_version,
58 $_merge, $_strategy, $_dry_run,
59 $_prefix);
60
61 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
62 'config-dir=s' => \$Git::SVN::Ra::config_dir,
63 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
64 my %fc_opts = ( 'follow-parent|follow' => \$Git::SVN::_follow_parent,
65 'authors-file|A=s' => \$_authors,
66 'repack:i' => \$Git::SVN::_repack,
67 'no-metadata' => \$Git::SVN::_no_metadata,
68 'quiet|q' => \$_q,
69 'repack-flags|repack-args|repack-opts=s' =>
70 \$Git::SVN::_repack_flags,
71 %remote_opts );
72
73 my ($_trunk, $_tags, $_branches);
74 my %multi_opts = ( 'trunk|T=s' => \$_trunk,
75 'tags|t=s' => \$_tags,
76 'branches|b=s' => \$_branches );
77 my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
78 my %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 );
84
85 my %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 );
135
136 my $cmd;
137 for (my $i = 0; $i < @ARGV; $i++) {
138 if (defined $cmd{$ARGV[$i]}) {
139 $cmd = $ARGV[$i];
140 splice @ARGV, $i, 1;
141 last;
142 }
143 };
144
145 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
146
147 read_repo_config(\%opts);
148 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
149 'minimize-connections' => \$Git::SVN::Migration::_minimize,
150 'id|i=s' => \$Git::SVN::default_ref_id,
151 'svn-remote|remote|R=s' => \$Git::SVN::default_repo_id);
152 exit 1 if (!$rv && $cmd ne 'log');
153
154 usage(0) if $_help;
155 version() if $_version;
156 usage(1) unless defined $cmd;
157 load_authors() if $_authors;
158 unless ($cmd =~ /^(?:init|multi-init|commit-diff)$/) {
159 Git::SVN::Migration::migration_check();
160 }
161 Git::SVN::init_vars();
162 eval {
163 Git::SVN::verify_remotes_sanity();
164 $cmd{$cmd}->[0]->(@ARGV);
165 };
166 fatal $@ if $@;
167 exit 0;
168
169 ####################### primary functions ######################
170 sub usage {
171 my $exit = shift || 0;
172 my $fd = $exit ? \*STDERR : \*STDOUT;
173 print $fd <<"";
174 git-svn - bidirectional operations between a single Subversion tree and git
175 Usage: $0 <command> [options] [arguments]\n
176
177 print $fd "Available commands:\n" unless $cmd;
178
179 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
192 arbitrary identifier if you're tracking multiple SVN branches/repositories in
193 one git repository and want to keep them separate. See git-svn(1) for more
194 information.
195
196 exit $exit;
197 }
198
199 sub version {
200 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
201 exit 0;
202 }
203
204 sub 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 }
212
213 sub 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} = $repo_path . "/.git";
222 }
223 do_git_init_db();
224
225 Git::SVN->init($url);
226 }
227
228 sub 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 $gs->{last_commit});
239 }
240 }
241
242 sub cmd_set_tree {
243 my (@commits) = @_;
244 if ($_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 }
276
277 sub 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 }
335
336 sub 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 }
341
342 sub 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 }
366
367 sub 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, $remotes);
373 }
374 }
375 }
376
377 # this command is special because it requires no metadata
378 sub cmd_commit_diff {
379 my ($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
426 ########################### utility functions #########################
427
428 sub 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: $_revision not understood by git-svn\n",
437 "Try using the command-line svn client instead\n";
438 }
439
440 sub complete_svn_url {
441 my ($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 }
452
453 sub 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/$pfx*");
497 }
498 }
499
500 sub verify_ref {
501 my ($ref) = @_;
502 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
503 { STDERR => 0 }); };
504 }
505
506 sub get_tree_from_treeish {
507 my ($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 $expected;
525 }
526
527 sub get_commit_entry {
528 my ($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
534 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 $!;
553
554 if ($_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 $commit_msg;
564 \%log_entry;
565 }
566
567 sub s_to_file {
568 my ($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 }
574
575 sub 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
585 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
586 sub 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
602 # convert GetOpt::Long specs for use by git-config
603 sub 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 }
624
625 sub 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 }
636
637 sub cmt_metadata {
638 return extract_metadata((grep(/^git-svn-id: /,
639 command(qw/cat-file commit/, shift)))[-1]);
640 }
641
642 package Git::SVN;
643 use strict;
644 use warnings;
645 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
646 $_repack $_repack_flags/;
647 use Carp qw/croak/;
648 use File::Path qw/mkpath/;
649 use File::Copy qw/copy/;
650 use IPC::Open3;
651
652 my $_repack_nr;
653 # properties that we do not log:
654 my %SKIP_PROP;
655 BEGIN {
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 }
663
664 my %LOCKFILES;
665 END { unlink keys %LOCKFILES if %LOCKFILES }
666
667 sub 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 }
701
702 sub 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
713 # 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
722 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
730 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 }
740
741 sub 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 }
763
764 sub init_vars {
765 if (defined $_repack) {
766 $_repack = 1000 if ($_repack <= 0);
767 $_repack_nr = $_repack;
768 $_repack_flags ||= '-d';
769 }
770 }
771
772 sub 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
789 # we allow more chars than remotes2config.sh...
790 sub sanitize_remote_name {
791 my ($name) = @_;
792 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
793 $name;
794 }
795
796 sub 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 }
807
808 sub 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 }
869
870 sub 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 }
878
879 sub 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 }
892
893 sub 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 }
920
921 sub refname { "refs/remotes/$_[0]->{ref_id}" }
922
923 sub ra {
924 my ($self) = shift;
925 Git::SVN::Ra->new($self->{url});
926 }
927
928 sub 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 }
937
938 sub 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 }
962
963 sub last_rev { ($_[0]->last_rev_commit)[0] }
964 sub last_commit { ($_[0]->last_rev_commit)[1] }
965
966 # returns the newest SVN revision number and newest commit SHA1
967 sub 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 }
1003
1004 sub 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 }
1010
1011 sub 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 }
1023
1024 sub assert_index_clean {
1025 my ($self, $treeish) = @_;
1026
1027 $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 }
1043
1044 sub 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 }
1070
1071 sub full_url {
1072 my ($self) = @_;
1073 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1074 }
1075
1076 sub 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
1095 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
1102 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
1124 $self->rev_db_set($log_entry->{revision}, $commit, 1);
1125
1126 $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 }
1138
1139 sub match_paths {
1140 my ($self, $paths, $r) = @_;
1141 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\/?/;
1142 if (grep /$self->{path_regex}/, keys %$paths) {
1143 return 1;
1144 }
1145 my $c = '';
1146 foreach (split m#/#, $self->{path}) {
1147 $c .= "/$_";
1148 next unless ($paths->{$c} && ($paths->{$c}->{action} eq 'A'));
1149 if ($self->ra->check_path($self->{path}, $r) ==
1150 $SVN::Node::dir) {
1151 return 1;
1152 }
1153 }
1154 return 0;
1155 }
1156
1157 sub find_parent_branch {
1158 my ($self, $paths, $rev) = @_;
1159 return undef unless $_follow_parent;
1160 unless (defined $paths) {
1161 my $err_handler = $SVN::Error::handler;
1162 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1163 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1164 $paths =
1165 Git::SVN::Ra::dup_changed_paths($_[0]) });
1166 $SVN::Error::handler = $err_handler;
1167 }
1168 return undef unless defined $paths;
1169
1170 # look for a parent from another branch:
1171 my @b_path_components = split m#/#, $self->rel_path;
1172 my @a_path_components;
1173 my $i;
1174 while (@b_path_components) {
1175 $i = $paths->{'/'.join('/', @b_path_components)};
1176 last if $i;
1177 unshift(@a_path_components, pop(@b_path_components));
1178 }
1179 goto not_found unless defined $i;
1180 my $branch_from = $i->{copyfrom_path} or goto not_found;
1181 if (@a_path_components) {
1182 print STDERR "branch_from: $branch_from => ";
1183 $branch_from .= '/'.join('/', @a_path_components);
1184 print STDERR $branch_from, "\n";
1185 }
1186 my $r = $i->{copyfrom_rev};
1187 my $repos_root = $self->ra->{repos_root};
1188 my $url = $self->ra->{url};
1189 my $new_url = $repos_root . $branch_from;
1190 print STDERR "Found possible branch point: ",
1191 "$new_url => ", $self->full_url, ", $r\n";
1192 $branch_from =~ s#^/##;
1193 my $remotes = read_all_remotes();
1194 my $gs;
1195 foreach my $repo_id (keys %$remotes) {
1196 my $u = $remotes->{$repo_id}->{url} or next;
1197 next if $url ne $u;
1198 my $fetch = $remotes->{$repo_id}->{fetch};
1199 foreach (qw/branches tags/) {
1200 resolve_local_globs($url, $fetch,
1201 $remotes->{$repo_id}->{$_});
1202 }
1203 foreach my $f (keys %$fetch) {
1204 next if $f ne $branch_from;
1205 $gs = Git::SVN->new($fetch->{$f}, $repo_id, $f);
1206 last;
1207 }
1208 last if $gs;
1209 }
1210 unless ($gs) {
1211 my $ref_id = $self->{ref_id};
1212 $ref_id =~ s/\@\d+$//;
1213 $ref_id .= "\@$r";
1214 # just grow a tail if we're not unique enough :x
1215 $ref_id .= '-' while find_ref($ref_id);
1216 print STDERR "Initializing parent: $ref_id\n";
1217 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1218 }
1219 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1220 if ($_follow_parent && (!defined $r0 || !defined $parent)) {
1221 $gs->fetch(0, $r);
1222 ($r0, $parent) = $gs->last_rev_commit;
1223 }
1224 if (defined $r0 && defined $parent) {
1225 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1226 $self->assert_index_clean($parent);
1227 my $ed;
1228 if ($self->ra->can_do_switch) {
1229 print STDERR "Following parent with do_switch\n";
1230 # do_switch works with svn/trunk >= r22312, but that
1231 # is not included with SVN 1.4.3 (the latest version
1232 # at the moment), so we can't rely on it
1233 $self->{last_commit} = $parent;
1234 $ed = SVN::Git::Fetcher->new($self);
1235 $gs->ra->gs_do_switch($r0, $rev, $gs,
1236 $self->full_url, $ed)
1237 or die "SVN connection failed somewhere...\n";
1238 } else {
1239 print STDERR "Following parent with do_update\n";
1240 $ed = SVN::Git::Fetcher->new($self);
1241 $self->ra->gs_do_update($rev, $rev, $self, $ed)
1242 or die "SVN connection failed somewhere...\n";
1243 }
1244 print STDERR "Successfully followed parent\n";
1245 return $self->make_log_entry($rev, [$parent], $ed);
1246 }
1247 not_found:
1248 print STDERR "Branch parent for path: '/",
1249 $self->rel_path, "' @ r$rev not found:\n";
1250 return undef unless $paths;
1251 print STDERR "Changed paths:\n";
1252 foreach my $x (sort keys %$paths) {
1253 my $p = $paths->{$x};
1254 print STDERR "\t$p->{action}\t$x";
1255 if ($p->{copyfrom_path}) {
1256 print STDERR "(from $p->{copyfrom_path}: ",
1257 "$p->{copyfrom_rev})";
1258 }
1259 print STDERR "\n";
1260 }
1261 print STDERR '-'x72, "\n";
1262 return undef;
1263 }
1264
1265 sub do_fetch {
1266 my ($self, $paths, $rev) = @_;
1267 my $ed;
1268 my ($last_rev, @parents);
1269 if ($self->last_commit) {
1270 $ed = SVN::Git::Fetcher->new($self);
1271 $last_rev = $self->{last_rev};
1272 $ed->{c} = $self->{last_commit};
1273 @parents = ($self->{last_commit});
1274 } else {
1275 $last_rev = $rev;
1276 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1277 return $log_entry;
1278 }
1279 $ed = SVN::Git::Fetcher->new($self);
1280 }
1281 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1282 die "SVN connection failed somewhere...\n";
1283 }
1284 $self->make_log_entry($rev, \@parents, $ed);
1285 }
1286
1287 sub get_untracked {
1288 my ($self, $ed) = @_;
1289 my @out;
1290 my $h = $ed->{empty};
1291 foreach (sort keys %$h) {
1292 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1293 push @out, " $act: " . uri_encode($_);
1294 warn "W: $act: $_\n";
1295 }
1296 foreach my $t (qw/dir_prop file_prop/) {
1297 $h = $ed->{$t} or next;
1298 foreach my $path (sort keys %$h) {
1299 my $ppath = $path eq '' ? '.' : $path;
1300 foreach my $prop (sort keys %{$h->{$path}}) {
1301 next if $SKIP_PROP{$prop};
1302 my $v = $h->{$path}->{$prop};
1303 my $t_ppath_prop = "$t: " .
1304 uri_encode($ppath) . ' ' .
1305 uri_encode($prop);
1306 if (defined $v) {
1307 push @out, " +$t_ppath_prop " .
1308 uri_encode($v);
1309 } else {
1310 push @out, " -$t_ppath_prop";
1311 }
1312 }
1313 }
1314 }
1315 foreach my $t (qw/absent_file absent_directory/) {
1316 $h = $ed->{$t} or next;
1317 foreach my $parent (sort keys %$h) {
1318 foreach my $path (sort @{$h->{$parent}}) {
1319 push @out, " $t: " .
1320 uri_encode("$parent/$path");
1321 warn "W: $t: $parent/$path ",
1322 "Insufficient permissions?\n";
1323 }
1324 }
1325 }
1326 \@out;
1327 }
1328
1329 sub parse_svn_date {
1330 my $date = shift || return '+0000 1970-01-01 00:00:00';
1331 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1332 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1333 croak "Unable to parse date: $date\n";
1334 "+0000 $Y-$m-$d $H:$M:$S";
1335 }
1336
1337 sub check_author {
1338 my ($author) = @_;
1339 if (!defined $author || length $author == 0) {
1340 $author = '(no author)';
1341 }
1342 if (defined $::_authors && ! defined $::users{$author}) {
1343 die "Author: $author not defined in $::_authors file\n";
1344 }
1345 $author;
1346 }
1347
1348 sub make_log_entry {
1349 my ($self, $rev, $parents, $ed) = @_;
1350 my $untracked = $self->get_untracked($ed);
1351
1352 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1353 print $un "r$rev\n" or croak $!;
1354 print $un $_, "\n" foreach @$untracked;
1355 my %log_entry = ( parents => $parents || [], revision => $rev,
1356 log => '');
1357
1358 my $logged = delete $self->{logged_rev_props};
1359 if (!$logged || $self->{-want_extra_revprops}) {
1360 my $rp = $self->ra->rev_proplist($rev);
1361 foreach (sort keys %$rp) {
1362 my $v = $rp->{$_};
1363 if (/^svn:(author|date|log)$/) {
1364 $log_entry{$1} = $v;
1365 } else {
1366 print $un " rev_prop: ", uri_encode($_), ' ',
1367 uri_encode($v), "\n";
1368 }
1369 }
1370 } else {
1371 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1372 }
1373 close $un or croak $!;
1374
1375 $log_entry{date} = parse_svn_date($log_entry{date});
1376 $log_entry{author} = check_author($log_entry{author});
1377 $log_entry{log} .= "\n";
1378 \%log_entry;
1379 }
1380
1381 sub fetch {
1382 my ($self, $min_rev, $max_rev, @parents) = @_;
1383 my ($last_rev, $last_commit) = $self->last_rev_commit;
1384 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1385 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1386 }
1387
1388 sub set_tree_cb {
1389 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1390 # TODO: enable and test optimized commits:
1391 if (0 && $rev == ($self->{last_rev} + 1)) {
1392 $log_entry->{revision} = $rev;
1393 $log_entry->{author} = $author;
1394 $self->do_git_commit($log_entry, "$rev=$tree");
1395 } else {
1396 $self->{inject_parents} = { $rev => $tree };
1397 $self->fetch(undef, undef);
1398 }
1399 }
1400
1401 sub set_tree {
1402 my ($self, $tree) = (shift, shift);
1403 my $log_entry = ::get_commit_entry($tree);
1404 unless ($self->{last_rev}) {
1405 fatal("Must have an existing revision to commit\n");
1406 }
1407 my %ed_opts = ( r => $self->{last_rev},
1408 log => $log_entry->{log},
1409 ra => $self->ra,
1410 tree_a => $self->{last_commit},
1411 tree_b => $tree,
1412 editor_cb => sub {
1413 $self->set_tree_cb($log_entry, $tree, @_) },
1414 svn_path => $self->{path} );
1415 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1416 print "No changes\nr$self->{last_rev} = $tree\n";
1417 }
1418 }
1419
1420 sub rebuild {
1421 my ($self) = @_;
1422 print "Rebuilding $self->{db_path} ...\n";
1423 my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1424 my $latest;
1425 my $full_url = $self->full_url;
1426 my $svn_uuid;
1427 while (<$rev_list>) {
1428 chomp;
1429 my $c = $_;
1430 die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1431 my ($url, $rev, $uuid) = ::cmt_metadata($c);
1432
1433 # ignore merges (from set-tree)
1434 next if (!defined $rev || !$uuid);
1435
1436 # if we merged or otherwise started elsewhere, this is
1437 # how we break out of it
1438 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1439 ($full_url && $url && ($url ne $full_url))) {
1440 next;
1441 }
1442 $latest ||= $rev;
1443 $svn_uuid ||= $uuid;
1444
1445 $self->rev_db_set($rev, $c);
1446 print "r$rev = $c\n";
1447 }
1448 command_close_pipe($rev_list, $ctx);
1449 print "Done rebuilding $self->{db_path}\n";
1450 }
1451
1452 # rev_db:
1453 # Tie::File seems to be prone to offset errors if revisions get sparse,
1454 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
1455 # one of my favorite modules is out :< Next up would be one of the DBM
1456 # modules, but I'm not sure which is most portable... So I'll just
1457 # go with something that's plain-text, but still capable of
1458 # being randomly accessed. So here's my ultra-simple fixed-width
1459 # database. All records are 40 characters + "\n", so it's easy to seek
1460 # to a revision: (41 * rev) is the byte offset.
1461 # A record of 40 0s denotes an empty revision.
1462 # And yes, it's still pretty fast (faster than Tie::File).
1463 # These files are disposable unless --no-metadata is set
1464
1465 sub rev_db_set {
1466 my ($self, $rev, $commit, $update_ref) = @_;
1467 length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
1468 my ($db, $db_lock) = ($self->{db_path}, "$self->{db_path}.lock");
1469 my $sig;
1470 if ($update_ref) {
1471 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1472 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
1473 }
1474 $LOCKFILES{$db_lock} = 1;
1475 if ($_no_metadata) {
1476 copy($db, $db_lock) or die "rev_db_set(@_): ",
1477 "Failed to copy: ",
1478 "$db => $db_lock ($!)\n";
1479 } else {
1480 rename $db, $db_lock or die "rev_db_set(@_): ",
1481 "Failed to rename: ",
1482 "$db => $db_lock ($!)\n";
1483 }
1484 open my $fh, '+<', $db_lock or croak $!;
1485 my $offset = $rev * 41;
1486 # assume that append is the common case:
1487 seek $fh, 0, 2 or croak $!;
1488 my $pos = tell $fh;
1489 if ($pos < $offset) {
1490 for (1 .. (($offset - $pos) / 41)) {
1491 print $fh (('0' x 40),"\n") or croak $!;
1492 }
1493 }
1494 seek $fh, $offset, 0 or croak $!;
1495 print $fh $commit,"\n" or croak $!;
1496 close $fh or croak $!;
1497 if ($update_ref) {
1498 command_noisy('update-ref', '-m', "r$rev",
1499 $self->refname, $commit);
1500 }
1501 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
1502 "$db_lock => $db ($!)\n";
1503 delete $LOCKFILES{$db_lock};
1504 if ($update_ref) {
1505 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1506 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
1507 kill $sig, $$ if defined $sig;
1508 }
1509 }
1510
1511 sub rev_db_max {
1512 my ($self) = @_;
1513 my @stat = stat $self->{db_path} or
1514 die "Couldn't stat $self->{db_path}: $!\n";
1515 ($stat[7] % 41) == 0 or
1516 die "$self->{db_path} inconsistent size:$stat[7]\n";
1517 my $max = $stat[7] / 41;
1518 (($max > 0) ? $max - 1 : 0);
1519 }
1520
1521 sub rev_db_get {
1522 my ($self, $rev) = @_;
1523 my $ret;
1524 my $offset = $rev * 41;
1525 open my $fh, '<', $self->{db_path} or croak $!;
1526 if (sysseek($fh, $offset, 0) == $offset) {
1527 my $read = sysread($fh, $ret, 40);
1528 $ret = undef if ($read != 40 || $ret eq ('0'x40));
1529 }
1530 close $fh or croak $!;
1531 $ret;
1532 }
1533
1534 sub find_rev_before {
1535 my ($self, $rev, $eq_ok) = @_;
1536 --$rev unless $eq_ok;
1537 while ($rev > 0) {
1538 if (my $c = $self->rev_db_get($rev)) {
1539 return ($rev, $c);
1540 }
1541 --$rev;
1542 }
1543 return (undef, undef);
1544 }
1545
1546 sub _new {
1547 my ($class, $repo_id, $ref_id, $path) = @_;
1548 unless (defined $repo_id && length $repo_id) {
1549 $repo_id = $Git::SVN::default_repo_id;
1550 }
1551 unless (defined $ref_id && length $ref_id) {
1552 $_[2] = $ref_id = $Git::SVN::default_ref_id;
1553 }
1554 $_[1] = $repo_id = sanitize_remote_name($repo_id);
1555 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1556 $_[3] = $path = '' unless (defined $path);
1557 mkpath([$dir]);
1558 unless (-f "$dir/.rev_db") {
1559 open my $fh, '>>', "$dir/.rev_db" or croak $!;
1560 close $fh or croak $!;
1561 }
1562 bless { ref_id => $ref_id, dir => $dir, index => "$dir/index",
1563 path => $path,
1564 db_path => "$dir/.rev_db", repo_id => $repo_id }, $class;
1565 }
1566
1567 sub uri_encode {
1568 my ($f) = @_;
1569 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1570 $f
1571 }
1572
1573 package Git::SVN::Prompt;
1574 use strict;
1575 use warnings;
1576 require SVN::Core;
1577 use vars qw/$_no_auth_cache $_username/;
1578
1579 sub simple {
1580 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1581 $may_save = undef if $_no_auth_cache;
1582 $default_username = $_username if defined $_username;
1583 if (defined $default_username && length $default_username) {
1584 if (defined $realm && length $realm) {
1585 print STDERR "Authentication realm: $realm\n";
1586 STDERR->flush;
1587 }
1588 $cred->username($default_username);
1589 } else {
1590 username($cred, $realm, $may_save, $pool);
1591 }
1592 $cred->password(_read_password("Password for '" .
1593 $cred->username . "': ", $realm));
1594 $cred->may_save($may_save);
1595 $SVN::_Core::SVN_NO_ERROR;
1596 }
1597
1598 sub ssl_server_trust {
1599 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1600 $may_save = undef if $_no_auth_cache;
1601 print STDERR "Error validating server certificate for '$realm':\n";
1602 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1603 print STDERR " - The certificate is not issued by a trusted ",
1604 "authority. Use the\n",
1605 " fingerprint to validate the certificate manually!\n";
1606 }
1607 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1608 print STDERR " - The certificate hostname does not match.\n";
1609 }
1610 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1611 print STDERR " - The certificate is not yet valid.\n";
1612 }
1613 if ($failures & $SVN::Auth::SSL::EXPIRED) {
1614 print STDERR " - The certificate has expired.\n";
1615 }
1616 if ($failures & $SVN::Auth::SSL::OTHER) {
1617 print STDERR " - The certificate has an unknown error.\n";
1618 }
1619 printf STDERR
1620 "Certificate information:\n".
1621 " - Hostname: %s\n".
1622 " - Valid: from %s until %s\n".
1623 " - Issuer: %s\n".
1624 " - Fingerprint: %s\n",
1625 map $cert_info->$_, qw(hostname valid_from valid_until
1626 issuer_dname fingerprint);
1627 my $choice;
1628 prompt:
1629 print STDERR $may_save ?
1630 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1631 "(R)eject or accept (t)emporarily? ";
1632 STDERR->flush;
1633 $choice = lc(substr(<STDIN> || 'R', 0, 1));
1634 if ($choice =~ /^t$/i) {
1635 $cred->may_save(undef);
1636 } elsif ($choice =~ /^r$/i) {
1637 return -1;
1638 } elsif ($may_save && $choice =~ /^p$/i) {
1639 $cred->may_save($may_save);
1640 } else {
1641 goto prompt;
1642 }
1643 $cred->accepted_failures($failures);
1644 $SVN::_Core::SVN_NO_ERROR;
1645 }
1646
1647 sub ssl_client_cert {
1648 my ($cred, $realm, $may_save, $pool) = @_;
1649 $may_save = undef if $_no_auth_cache;
1650 print STDERR "Client certificate filename: ";
1651 STDERR->flush;
1652 chomp(my $filename = <STDIN>);
1653 $cred->cert_file($filename);
1654 $cred->may_save($may_save);
1655 $SVN::_Core::SVN_NO_ERROR;
1656 }
1657
1658 sub ssl_client_cert_pw {
1659 my ($cred, $realm, $may_save, $pool) = @_;
1660 $may_save = undef if $_no_auth_cache;
1661 $cred->password(_read_password("Password: ", $realm));
1662 $cred->may_save($may_save);
1663 $SVN::_Core::SVN_NO_ERROR;
1664 }
1665
1666 sub username {
1667 my ($cred, $realm, $may_save, $pool) = @_;
1668 $may_save = undef if $_no_auth_cache;
1669 if (defined $realm && length $realm) {
1670 print STDERR "Authentication realm: $realm\n";
1671 }
1672 my $username;
1673 if (defined $_username) {
1674 $username = $_username;
1675 } else {
1676 print STDERR "Username: ";
1677 STDERR->flush;
1678 chomp($username = <STDIN>);
1679 }
1680 $cred->username($username);
1681 $cred->may_save($may_save);
1682 $SVN::_Core::SVN_NO_ERROR;
1683 }
1684
1685 sub _read_password {
1686 my ($prompt, $realm) = @_;
1687 print STDERR $prompt;
1688 STDERR->flush;
1689 require Term::ReadKey;
1690 Term::ReadKey::ReadMode('noecho');
1691 my $password = '';
1692 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1693 last if $key =~ /[\012\015]/; # \n\r
1694 $password .= $key;
1695 }
1696 Term::ReadKey::ReadMode('restore');
1697 print STDERR "\n";
1698 STDERR->flush;
1699 $password;
1700 }
1701
1702 package main;
1703
1704 {
1705 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1706 $SVN::Node::dir.$SVN::Node::unknown.
1707 $SVN::Node::none.$SVN::Node::file.
1708 $SVN::Node::dir.$SVN::Node::unknown.
1709 $SVN::Auth::SSL::CNMISMATCH.
1710 $SVN::Auth::SSL::NOTYETVALID.
1711 $SVN::Auth::SSL::EXPIRED.
1712 $SVN::Auth::SSL::UNKNOWNCA.
1713 $SVN::Auth::SSL::OTHER;
1714 }
1715
1716 package SVN::Git::Fetcher;
1717 use vars qw/@ISA/;
1718 use strict;
1719 use warnings;
1720 use Carp qw/croak/;
1721 use IO::File qw//;
1722 use Digest::MD5;
1723
1724 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
1725 sub new {
1726 my ($class, $git_svn) = @_;
1727 my $self = SVN::Delta::Editor->new;
1728 bless $self, $class;
1729 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
1730 $self->{empty} = {};
1731 $self->{dir_prop} = {};
1732 $self->{file_prop} = {};
1733 $self->{absent_dir} = {};
1734 $self->{absent_file} = {};
1735 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
1736 $self;
1737 }
1738
1739 sub set_path_strip {
1740 my ($self, $path) = @_;
1741 $self->{path_strip} = qr/^\Q$path\E\/?/;
1742 }
1743
1744 sub open_root {
1745 { path => '' };
1746 }
1747
1748 sub open_directory {
1749 my ($self, $path, $pb, $rev) = @_;
1750 { path => $path };
1751 }
1752
1753 sub git_path {
1754 my ($self, $path) = @_;
1755 if ($self->{path_strip}) {
1756 $path =~ s!$self->{path_strip}!! or
1757 die "Failed to strip path '$path' ($self->{path_strip})\n";
1758 }
1759 $path;
1760 }
1761
1762 sub delete_entry {
1763 my ($self, $path, $rev, $pb) = @_;
1764
1765 my $gpath = $self->git_path($path);
1766 return undef if ($gpath eq '');
1767
1768 # remove entire directories.
1769 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
1770 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
1771 -r --name-only -z/,
1772 $self->{c}, '--', $gpath);
1773 local $/ = "\0";
1774 while (<$ls>) {
1775 chomp;
1776 $self->{gii}->remove($_);
1777 print "\tD\t$_\n" unless $self->{q};
1778 }
1779 print "\tD\t$gpath/\n" unless $self->{q};
1780 command_close_pipe($ls, $ctx);
1781 $self->{empty}->{$path} = 0
1782 } else {
1783 $self->{gii}->remove($gpath);
1784 print "\tD\t$gpath\n" unless $self->{q};
1785 }
1786 undef;
1787 }
1788
1789 sub open_file {
1790 my ($self, $path, $pb, $rev) = @_;
1791 my $gpath = $self->git_path($path);
1792 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
1793 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
1794 unless (defined $mode && defined $blob) {
1795 die "$path was not found in commit $self->{c} (r$rev)\n";
1796 }
1797 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
1798 pool => SVN::Pool->new, action => 'M' };
1799 }
1800
1801 sub add_file {
1802 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
1803 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1804 delete $self->{empty}->{$dir};
1805 { path => $path, mode_a => 100644, mode_b => 100644,
1806 pool => SVN::Pool->new, action => 'A' };
1807 }
1808
1809 sub add_directory {
1810 my ($self, $path, $cp_path, $cp_rev) = @_;
1811 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1812 delete $self->{empty}->{$dir};
1813 $self->{empty}->{$path} = 1;
1814 { path => $path };
1815 }
1816
1817 sub change_dir_prop {
1818 my ($self, $db, $prop, $value) = @_;
1819 $self->{dir_prop}->{$db->{path}} ||= {};
1820 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
1821 undef;
1822 }
1823
1824 sub absent_directory {
1825 my ($self, $path, $pb) = @_;
1826 $self->{absent_dir}->{$pb->{path}} ||= [];
1827 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
1828 undef;
1829 }
1830
1831 sub absent_file {
1832 my ($self, $path, $pb) = @_;
1833 $self->{absent_file}->{$pb->{path}} ||= [];
1834 push @{$self->{absent_file}->{$pb->{path}}}, $path;
1835 undef;
1836 }
1837
1838 sub change_file_prop {
1839 my ($self, $fb, $prop, $value) = @_;
1840 if ($prop eq 'svn:executable') {
1841 if ($fb->{mode_b} != 120000) {
1842 $fb->{mode_b} = defined $value ? 100755 : 100644;
1843 }
1844 } elsif ($prop eq 'svn:special') {
1845 $fb->{mode_b} = defined $value ? 120000 : 100644;
1846 } else {
1847 $self->{file_prop}->{$fb->{path}} ||= {};
1848 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
1849 }
1850 undef;
1851 }
1852
1853 sub apply_textdelta {
1854 my ($self, $fb, $exp) = @_;
1855 my $fh = IO::File->new_tmpfile;
1856 $fh->autoflush(1);
1857 # $fh gets auto-closed() by SVN::TxDelta::apply(),
1858 # (but $base does not,) so dup() it for reading in close_file
1859 open my $dup, '<&', $fh or croak $!;
1860 my $base = IO::File->new_tmpfile;
1861 $base->autoflush(1);
1862 if ($fb->{blob}) {
1863 defined (my $pid = fork) or croak $!;
1864 if (!$pid) {
1865 open STDOUT, '>&', $base or croak $!;
1866 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
1867 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
1868 }
1869 waitpid $pid, 0;
1870 croak $? if $?;
1871
1872 if (defined $exp) {
1873 seek $base, 0, 0 or croak $!;
1874 my $md5 = Digest::MD5->new;
1875 $md5->addfile($base);
1876 my $got = $md5->hexdigest;
1877 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
1878 "expected: $exp\n",
1879 " got: $got\n" if ($got ne $exp);
1880 }
1881 }
1882 seek $base, 0, 0 or croak $!;
1883 $fb->{fh} = $dup;
1884 $fb->{base} = $base;
1885 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
1886 }
1887
1888 sub close_file {
1889 my ($self, $fb, $exp) = @_;
1890 my $hash;
1891 my $path = $self->git_path($fb->{path});
1892 if (my $fh = $fb->{fh}) {
1893 seek($fh, 0, 0) or croak $!;
1894 my $md5 = Digest::MD5->new;
1895 $md5->addfile($fh);
1896 my $got = $md5->hexdigest;
1897 die "Checksum mismatch: $path\n",
1898 "expected: $exp\n got: $got\n" if ($got ne $exp);
1899 seek($fh, 0, 0) or croak $!;
1900 if ($fb->{mode_b} == 120000) {
1901 read($fh, my $buf, 5) == 5 or croak $!;
1902 $buf eq 'link ' or die "$path has mode 120000",
1903 "but is not a link\n";
1904 }
1905 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
1906 if (!$pid) {
1907 open STDIN, '<&', $fh or croak $!;
1908 exec qw/git-hash-object -w --stdin/ or croak $!;
1909 }
1910 chomp($hash = do { local $/; <$out> });
1911 close $out or croak $!;
1912 close $fh or croak $!;
1913 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
1914 close $fb->{base} or croak $!;
1915 } else {
1916 $hash = $fb->{blob} or die "no blob information\n";
1917 }
1918 $fb->{pool}->clear;
1919 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
1920 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
1921 undef;
1922 }
1923
1924 sub abort_edit {
1925 my $self = shift;
1926 $self->{nr} = $self->{gii}->{nr};
1927 delete $self->{gii};
1928 $self->SUPER::abort_edit(@_);
1929 }
1930
1931 sub close_edit {
1932 my $self = shift;
1933 $self->{git_commit_ok} = 1;
1934 $self->{nr} = $self->{gii}->{nr};
1935 delete $self->{gii};
1936 $self->SUPER::close_edit(@_);
1937 }
1938
1939 package SVN::Git::Editor;
1940 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
1941 use strict;
1942 use warnings;
1943 use Carp qw/croak/;
1944 use IO::File;
1945 use Digest::MD5;
1946
1947 sub new {
1948 my ($class, $opts) = @_;
1949 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
1950 die "$_ required!\n" unless (defined $opts->{$_});
1951 }
1952
1953 my $pool = SVN::Pool->new;
1954 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
1955 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
1956 $opts->{r}, $mods);
1957
1958 # $opts->{ra} functions should not be used after this:
1959 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
1960 $opts->{editor_cb}, $pool);
1961 my $self = SVN::Delta::Editor->new(@ce, $pool);
1962 bless $self, $class;
1963 foreach (qw/svn_path r tree_a tree_b/) {
1964 $self->{$_} = $opts->{$_};
1965 }
1966 $self->{url} = $opts->{ra}->{url};
1967 $self->{mods} = $mods;
1968 $self->{types} = $types;
1969 $self->{pool} = $pool;
1970 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
1971 $self->{rm} = { };
1972 $self->{path_prefix} = length $self->{svn_path} ?
1973 "$self->{svn_path}/" : '';
1974 return $self;
1975 }
1976
1977 sub generate_diff {
1978 my ($tree_a, $tree_b) = @_;
1979 my @diff_tree = qw(diff-tree -z -r);
1980 if ($_cp_similarity) {
1981 push @diff_tree, "-C$_cp_similarity";
1982 } else {
1983 push @diff_tree, '-C';
1984 }
1985 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1986 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
1987 push @diff_tree, $tree_a, $tree_b;
1988 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
1989 local $/ = "\0";
1990 my $state = 'meta';
1991 my @mods;
1992 while (<$diff_fh>) {
1993 chomp $_; # this gets rid of the trailing "\0"
1994 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1995 $::sha1\s($::sha1)\s
1996 ([MTCRAD])\d*$/xo) {
1997 push @mods, { mode_a => $1, mode_b => $2,
1998 sha1_b => $3, chg => $4 };
1999 if ($4 =~ /^(?:C|R)$/) {
2000 $state = 'file_a';
2001 } else {
2002 $state = 'file_b';
2003 }
2004 } elsif ($state eq 'file_a') {
2005 my $x = $mods[$#mods] or croak "Empty array\n";
2006 if ($x->{chg} !~ /^(?:C|R)$/) {
2007 croak "Error parsing $_, $x->{chg}\n";
2008 }
2009 $x->{file_a} = $_;
2010 $state = 'file_b';
2011 } elsif ($state eq 'file_b') {
2012 my $x = $mods[$#mods] or croak "Empty array\n";
2013 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2014 croak "Error parsing $_, $x->{chg}\n";
2015 }
2016 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2017 croak "Error parsing $_, $x->{chg}\n";
2018 }
2019 $x->{file_b} = $_;
2020 $state = 'meta';
2021 } else {
2022 croak "Error parsing $_\n";
2023 }
2024 }
2025 command_close_pipe($diff_fh, $ctx);
2026 \@mods;
2027 }
2028
2029 sub check_diff_paths {
2030 my ($ra, $pfx, $rev, $mods) = @_;
2031 my %types;
2032 $pfx .= '/' if length $pfx;
2033
2034 sub type_diff_paths {
2035 my ($ra, $types, $path, $rev) = @_;
2036 my @p = split m#/+#, $path;
2037 my $c = shift @p;
2038 unless (defined $types->{$c}) {
2039 $types->{$c} = $ra->check_path($c, $rev);
2040 }
2041 while (@p) {
2042 $c .= '/' . shift @p;
2043 next if defined $types->{$c};
2044 $types->{$c} = $ra->check_path($c, $rev);
2045 }
2046 }
2047
2048 foreach my $m (@$mods) {
2049 foreach my $f (qw/file_a file_b/) {
2050 next unless defined $m->{$f};
2051 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2052 if (length $pfx.$dir && ! defined $types{$dir}) {
2053 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2054 }
2055 }
2056 }
2057 \%types;
2058 }
2059
2060 sub split_path {
2061 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2062 }
2063
2064 sub repo_path {
2065 my ($self, $path) = @_;
2066 $self->{path_prefix}.(defined $path ? $path : '');
2067 }
2068
2069 sub url_path {
2070 my ($self, $path) = @_;
2071 $self->{url} . '/' . $self->repo_path($path);
2072 }
2073
2074 sub rmdirs {
2075 my ($self) = @_;
2076 my $rm = $self->{rm};
2077 delete $rm->{''}; # we never delete the url we're tracking
2078 return unless %$rm;
2079
2080 foreach (keys %$rm) {
2081 my @d = split m#/#, $_;
2082 my $c = shift @d;
2083 $rm->{$c} = 1;
2084 while (@d) {
2085 $c .= '/' . shift @d;
2086 $rm->{$c} = 1;
2087 }
2088 }
2089 delete $rm->{$self->{svn_path}};
2090 delete $rm->{''}; # we never delete the url we're tracking
2091 return unless %$rm;
2092
2093 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2094 $self->{tree_b});
2095 local $/ = "\0";
2096 while (<$fh>) {
2097 chomp;
2098 my @dn = split m#/#, $_;
2099 while (pop @dn) {
2100 delete $rm->{join '/', @dn};
2101 }
2102 unless (%$rm) {
2103 close $fh;
2104 return;
2105 }
2106 }
2107 command_close_pipe($fh, $ctx);
2108
2109 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2110 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2111 $self->close_directory($bat->{$d}, $p);
2112 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2113 print "\tD+\t$d/\n" unless $::_q;
2114 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2115 delete $bat->{$d};
2116 }
2117 }
2118
2119 sub open_or_add_dir {
2120 my ($self, $full_path, $baton) = @_;
2121 my $t = $self->{types}->{$full_path};
2122 if (!defined $t) {
2123 die "$full_path not known in r$self->{r} or we have a bug!\n";
2124 }
2125 if ($t == $SVN::Node::none) {
2126 return $self->add_directory($full_path, $baton,
2127 undef, -1, $self->{pool});
2128 } elsif ($t == $SVN::Node::dir) {
2129 return $self->open_directory($full_path, $baton,
2130 $self->{r}, $self->{pool});
2131 }
2132 print STDERR "$full_path already exists in repository at ",
2133 "r$self->{r} and it is not a directory (",
2134 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2135 exit 1;
2136 }
2137
2138 sub ensure_path {
2139 my ($self, $path) = @_;
2140 my $bat = $self->{bat};
2141 my $repo_path = $self->repo_path($path);
2142 return $bat->{''} unless (length $repo_path);
2143 my @p = split m#/+#, $repo_path;
2144 my $c = shift @p;
2145 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2146 while (@p) {
2147 my $c0 = $c;
2148 $c .= '/' . shift @p;
2149 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2150 }
2151 return $bat->{$c};
2152 }
2153
2154 sub A {
2155 my ($self, $m) = @_;
2156 my ($dir, $file) = split_path($m->{file_b});
2157 my $pbat = $self->ensure_path($dir);
2158 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2159 undef, -1);
2160 print "\tA\t$m->{file_b}\n" unless $::_q;
2161 $self->chg_file($fbat, $m);
2162 $self->close_file($fbat,undef,$self->{pool});
2163 }
2164
2165 sub C {
2166 my ($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 $self->url_path($m->{file_a}), $self->{r});
2171 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2172 $self->chg_file($fbat, $m);
2173 $self->close_file($fbat,undef,$self->{pool});
2174 }
2175
2176 sub delete_entry {
2177 my ($self, $path, $pbat) = @_;
2178 my $rpath = $self->repo_path($path);
2179 my ($dir, $file) = split_path($rpath);
2180 $self->{rm}->{$dir} = 1;
2181 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2182 }
2183
2184 sub R {
2185 my ($self, $m) = @_;
2186 my ($dir, $file) = split_path($m->{file_b});
2187 my $pbat = $self->ensure_path($dir);
2188 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2189 $self->url_path($m->{file_a}), $self->{r});
2190 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2191 $self->chg_file($fbat, $m);
2192 $self->close_file($fbat,undef,$self->{pool});
2193
2194 ($dir, $file) = split_path($m->{file_a});
2195 $pbat = $self->ensure_path($dir);
2196 $self->delete_entry($m->{file_a}, $pbat);
2197 }
2198
2199 sub M {
2200 my ($self, $m) = @_;
2201 my ($dir, $file) = split_path($m->{file_b});
2202 my $pbat = $self->ensure_path($dir);
2203 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2204 $pbat,$self->{r},$self->{pool});
2205 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2206 $self->chg_file($fbat, $m);
2207 $self->close_file($fbat,undef,$self->{pool});
2208 }
2209
2210 sub T { shift->M(@_) }
2211
2212 sub change_file_prop {
2213 my ($self, $fbat, $pname, $pval) = @_;
2214 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2215 }
2216
2217 sub chg_file {
2218 my ($self, $fbat, $m) = @_;
2219 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2220 $self->change_file_prop($fbat,'svn:executable','*');
2221 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2222 $self->change_file_prop($fbat,'svn:executable',undef);
2223 }
2224 my $fh = IO::File->new_tmpfile or croak $!;
2225 if ($m->{mode_b} =~ /^120/) {
2226 print $fh 'link ' or croak $!;
2227 $self->change_file_prop($fbat,'svn:special','*');
2228 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2229 $self->change_file_prop($fbat,'svn:special',undef);
2230 }
2231 defined(my $pid = fork) or croak $!;
2232 if (!$pid) {
2233 open STDOUT, '>&', $fh or croak $!;
2234 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2235 }
2236 waitpid $pid, 0;
2237 croak $? if $?;
2238 $fh->flush == 0 or croak $!;
2239 seek $fh, 0, 0 or croak $!;
2240
2241 my $md5 = Digest::MD5->new;
2242 $md5->addfile($fh) or croak $!;
2243 seek $fh, 0, 0 or croak $!;
2244
2245 my $exp = $md5->hexdigest;
2246 my $pool = SVN::Pool->new;
2247 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2248 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2249 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2250 $pool->clear;
2251
2252 close $fh or croak $!;
2253 }
2254
2255 sub D {
2256 my ($self, $m) = @_;
2257 my ($dir, $file) = split_path($m->{file_b});
2258 my $pbat = $self->ensure_path($dir);
2259 print "\tD\t$m->{file_b}\n" unless $::_q;
2260 $self->delete_entry($m->{file_b}, $pbat);
2261 }
2262
2263 sub close_edit {
2264 my ($self) = @_;
2265 my ($p,$bat) = ($self->{pool}, $self->{bat});
2266 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2267 $self->close_directory($bat->{$_}, $p);
2268 }
2269 $self->SUPER::close_edit($p);
2270 $p->clear;
2271 }
2272
2273 sub abort_edit {
2274 my ($self) = @_;
2275 $self->SUPER::abort_edit($self->{pool});
2276 }
2277
2278 sub DESTROY {
2279 my $self = shift;
2280 $self->SUPER::DESTROY(@_);
2281 $self->{pool}->clear;
2282 }
2283
2284 # this drives the editor
2285 sub apply_diff {
2286 my ($self) = @_;
2287 my $mods = $self->{mods};
2288 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2289 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2290 my $f = $m->{chg};
2291 if (defined $o{$f}) {
2292 $self->$f($m);
2293 } else {
2294 fatal("Invalid change type: $f\n");
2295 }
2296 }
2297 $self->rmdirs if $_rmdir;
2298 if (@$mods == 0) {
2299 $self->abort_edit;
2300 } else {
2301 $self->close_edit;
2302 }
2303 return scalar @$mods;
2304 }
2305
2306 package Git::SVN::Ra;
2307 use vars qw/@ISA $config_dir/;
2308 use strict;
2309 use warnings;
2310 my ($can_do_switch);
2311 my $RA;
2312
2313 BEGIN {
2314 # enforce temporary pool usage for some simple functions
2315 my $e;
2316 foreach (qw/get_latest_revnum rev_proplist get_file
2317 check_path get_dir get_uuid get_repos_root/) {
2318 $e .= "sub $_ {
2319 my \$self = shift;
2320 my \$pool = SVN::Pool->new;
2321 my \@ret = \$self->SUPER::$_(\@_,\$pool);
2322 \$pool->clear;
2323 wantarray ? \@ret : \$ret[0]; }\n";
2324 }
2325 eval $e;
2326 }
2327
2328 sub new {
2329 my ($class, $url) = @_;
2330 $url =~ s!/+$!!;
2331 return $RA if ($RA && $RA->{url} eq $url);
2332
2333 SVN::_Core::svn_config_ensure($config_dir, undef);
2334 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2335 SVN::Client::get_simple_provider(),
2336 SVN::Client::get_ssl_server_trust_file_provider(),
2337 SVN::Client::get_simple_prompt_provider(
2338 \&Git::SVN::Prompt::simple, 2),
2339 SVN::Client::get_ssl_client_cert_prompt_provider(
2340 \&Git::SVN::Prompt::ssl_client_cert, 2),
2341 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2342 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2343 SVN::Client::get_username_provider(),
2344 SVN::Client::get_ssl_server_trust_prompt_provider(
2345 \&Git::SVN::Prompt::ssl_server_trust),
2346 SVN::Client::get_username_prompt_provider(
2347 \&Git::SVN::Prompt::username, 2),
2348 ]);
2349 my $config = SVN::Core::config_get_config($config_dir);
2350 my $self = SVN::Ra->new(url => $url, auth => $baton,
2351 config => $config,
2352 pool => SVN::Pool->new,
2353 auth_provider_callbacks => $callbacks);
2354 $self->{svn_path} = $url;
2355 $self->{repos_root} = $self->get_repos_root;
2356 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E/*##;
2357 $RA = bless $self, $class;
2358 }
2359
2360 sub DESTROY {
2361 # do not call the real DESTROY since we store ourselves in $RA
2362 }
2363
2364 sub get_log {
2365 my ($self, @args) = @_;
2366 my $pool = SVN::Pool->new;
2367 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2368 my $ret = $self->SUPER::get_log(@args, $pool);
2369 $pool->clear;
2370 $ret;
2371 }
2372
2373 sub get_commit_editor {
2374 my ($self, $log, $cb, $pool) = @_;
2375 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2376 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2377 }
2378
2379 sub uuid {
2380 my ($self) = @_;
2381 $self->{uuid} ||= $self->get_uuid;
2382 }
2383
2384 sub gs_do_update {
2385 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2386 my $new = ($rev_a == $rev_b);
2387 my $path = $gs->{path};
2388
2389 my $ta = $self->check_path($path, $rev_a);
2390 my $tb = $new ? $ta : $self->check_path($path, $rev_b);
2391 return 1 if ($tb != $SVN::Node::dir && $ta != $SVN::Node::dir);
2392 if ($ta == $SVN::Node::none) {
2393 $rev_a = $rev_b;
2394 $new = 1;
2395 }
2396
2397 my $pool = SVN::Pool->new;
2398 $editor->set_path_strip($path);
2399 my (@pc) = split m#/#, $path;
2400 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2401 1, $editor, $pool);
2402 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2403
2404 # Since we can't rely on svn_ra_reparent being available, we'll
2405 # just have to do some magic with set_path to make it so
2406 # we only want a partial path.
2407 my $sp = '';
2408 my $final = join('/', @pc);
2409 while (@pc) {
2410 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2411 $sp .= '/' if length $sp;
2412 $sp .= shift @pc;
2413 }
2414 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2415
2416 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2417
2418 $reporter->finish_report($pool);
2419 $pool->clear;
2420 $editor->{git_commit_ok};
2421 }
2422
2423 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2424 # svn_ra_reparent didn't work before 1.4)
2425 sub gs_do_switch {
2426 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
2427 my $path = $gs->{path};
2428 my $pool = SVN::Pool->new;
2429
2430 my $full_url = $self->{url};
2431 my $old_url = $full_url;
2432 $full_url .= "/$path" if length $path;
2433 my ($ra, $reparented);
2434 if ($old_url ne $full_url) {
2435 if ($old_url !~ m#^svn(\+ssh)?://#) {
2436 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2437 $pool);
2438 $self->{url} = $full_url;
2439 $reparented = 1;
2440 } else {
2441 $ra = Git::SVN::Ra->new($full_url);
2442 }
2443 }
2444 $ra ||= $self;
2445 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
2446 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2447 $reporter->set_path('', $rev_a, 0, @lock, $pool);
2448 $reporter->finish_report($pool);
2449
2450 if ($reparented) {
2451 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2452 $self->{url} = $old_url;
2453 }
2454
2455 $pool->clear;
2456 $editor->{git_commit_ok};
2457 }
2458
2459 sub gs_fetch_loop_common {
2460 my ($self, $base, $head, $gsv, $globs) = @_;
2461 return if ($base > $head);
2462 my $inc = 1000;
2463 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2464 my %common;
2465 my $common_max = scalar @$gsv;
2466
2467 foreach my $gs (@$gsv) {
2468 if (my $last_commit = $gs->last_commit) {
2469 $gs->assert_index_clean($last_commit);
2470 }
2471 my @tmp = split m#/#, $gs->{path};
2472 my $p = '';
2473 foreach (@tmp) {
2474 $p .= length($p) ? "/$_" : $_;
2475 $common{$p} ||= 0;
2476 $common{$p}++;
2477 }
2478 }
2479 $globs ||= [];
2480 $common_max += scalar @$globs;
2481 foreach my $glob (@$globs) {
2482 my @tmp = split m#/#, $glob->{path}->{left};
2483 my $p = '';
2484 foreach (@tmp) {
2485 $p .= length($p) ? "/$_" : $_;
2486 $common{$p} ||= 0;
2487 $common{$p}++;
2488 }
2489 }
2490
2491 my $longest_path = '';
2492 foreach (sort {length $b <=> length $a} keys %common) {
2493 if ($common{$_} == $common_max) {
2494 $longest_path = $_;
2495 last;
2496 }
2497 }
2498 while (1) {
2499 my %revs;
2500 my $err;
2501 my $err_handler = $SVN::Error::handler;
2502 $SVN::Error::handler = sub {
2503 ($err) = @_;
2504 skip_unknown_revs($err);
2505 };
2506 sub _cb {
2507 my ($paths, $r, $author, $date, $log) = @_;
2508 [ dup_changed_paths($paths),
2509 { author => $author, date => $date, log => $log } ];
2510 }
2511 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
2512 sub { $revs{$_[1]} = _cb(@_) });
2513 if ($err && $max >= $head) {
2514 print STDERR "Path '$longest_path' ",
2515 "was probably deleted:\n",
2516 $err->expanded_message,
2517 "\nWill attempt to follow ",
2518 "revisions r$min .. r$max ",
2519 "committed before the deletion\n";
2520 my $hi = $max;
2521 while (--$hi >= $min) {
2522 my $ok;
2523 $self->get_log([$longest_path], $min, $hi,
2524 0, 1, 1, sub {
2525 $ok ||= $_[1];
2526 $revs{$_[1]} = _cb(@_) });
2527 if ($ok) {
2528 print STDERR "r$min .. r$ok OK\n";
2529 last;
2530 }
2531 }
2532 }
2533 $SVN::Error::handler = $err_handler;
2534
2535 my %exists = map { $_->{path} => $_ } @$gsv;
2536 foreach my $r (sort {$a <=> $b} keys %revs) {
2537 my ($paths, $logged) = @{$revs{$r}};
2538
2539 foreach my $gs ($self->match_globs(\%exists, $paths,
2540 $globs, $r)) {
2541 if ($gs->rev_db_max >= $r) {
2542 next;
2543 }
2544 next unless $gs->match_paths($paths, $r);
2545 $gs->{logged_rev_props} = $logged;
2546 my $log_entry = $gs->do_fetch($paths, $r);
2547 if ($log_entry) {
2548 $gs->do_git_commit($log_entry);
2549 }
2550 }
2551 foreach my $g (@$globs) {
2552 my $f = "$ENV{GIT_DIR}/svn/." .
2553 $self->uuid . ".$g->{t}";
2554 open my $fh, '>', "$f.tmp" or
2555 die "Can't open $f.tmp for writing: $!";
2556 print $fh "$r\n" or
2557 die "Couldn't write to $f: $!\n";
2558 close $fh or die "Error closing $f: $!\n";
2559 rename "$f.tmp", $f or
2560 die "Couldn't rename ",
2561 "$f.tmp => $f: $!\n";
2562 }
2563 }
2564 # pre-fill the .rev_db since it'll eventually get filled in
2565 # with '0' x40 if something new gets committed
2566 foreach my $gs (@$gsv) {
2567 next if defined $gs->rev_db_get($max);
2568 $gs->rev_db_set($max, 0 x40);
2569 }
2570 last if $max >= $head;
2571 $min = $max + 1;
2572 $max += $inc;
2573 $max = $head if ($max > $head);
2574 }
2575 }
2576
2577 sub match_globs {
2578 my ($self, $exists, $paths, $globs, $r) = @_;
2579 foreach my $g (@$globs) {
2580 foreach (keys %$paths) {
2581 next unless /$g->{path}->{regex}/;
2582 my $p = $1;
2583 my $pathname = $g->{path}->full_path($p);
2584 next if $exists->{$pathname};
2585 $exists->{$pathname} = Git::SVN->init(
2586 $self->{url}, $pathname, undef,
2587 $g->{ref}->full_path($p), 1);
2588 }
2589 my $c = '';
2590 foreach (split m#/#, $g->{path}->{left}) {
2591 $c .= "/$_";
2592 next unless ($paths->{$c} &&
2593 ($paths->{$c}->{action} eq 'A'));
2594 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
2595 next unless scalar @x == 3;
2596 my $dirents = $x[0];
2597 foreach my $de (keys %$dirents) {
2598 next if $dirents->{$de}->kind !=
2599 $SVN::Node::dir;
2600 my $p = $g->{path}->full_path($de);
2601 next if $exists->{$p};
2602 next if (length $g->{path}->{right} &&
2603 ($self->check_path($p, $r) !=
2604 $SVN::Node::dir));
2605 $exists->{$p} = Git::SVN->init($self->{url},
2606 $p, undef,
2607 $g->{ref}->full_path($de), 1);
2608 }
2609 }
2610 }
2611 values %$exists;
2612 }
2613
2614 sub minimize_url {
2615 my ($self) = @_;
2616 return $self->{url} if ($self->{url} eq $self->{repos_root});
2617 my $url = $self->{repos_root};
2618 my @components = split(m!/!, $self->{svn_path});
2619 my $c = '';
2620 do {
2621 $url .= "/$c" if length $c;
2622 eval { (ref $self)->new($url)->get_latest_revnum };
2623 } while ($@ && ($c = shift @components));
2624 $url;
2625 }
2626
2627 sub can_do_switch {
2628 my $self = shift;
2629 unless (defined $can_do_switch) {
2630 my $pool = SVN::Pool->new;
2631 my $rep = eval {
2632 $self->do_switch(1, '', 0, $self->{url},
2633 SVN::Delta::Editor->new, $pool);
2634 };
2635 if ($@) {
2636 $can_do_switch = 0;
2637 } else {
2638 $rep->abort_report($pool);
2639 $can_do_switch = 1;
2640 }
2641 $pool->clear;
2642 }
2643 $can_do_switch;
2644 }
2645
2646 sub skip_unknown_revs {
2647 my ($err) = @_;
2648 my $errno = $err->apr_err();
2649 # Maybe the branch we're tracking didn't
2650 # exist when the repo started, so it's
2651 # not an error if it doesn't, just continue
2652 #
2653 # Wonderfully consistent library, eh?
2654 # 160013 - svn:// and file://
2655 # 175002 - http(s)://
2656 # 175007 - http(s):// (this repo required authorization, too...)
2657 # More codes may be discovered later...
2658 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2659 return;
2660 }
2661 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2662 }
2663
2664 # svn_log_changed_path_t objects passed to get_log are likely to be
2665 # overwritten even if only the refs are copied to an external variable,
2666 # so we should dup the structures in their entirety. Using an externally
2667 # passed pool (instead of our temporary and quickly cleared pool in
2668 # Git::SVN::Ra) does not help matters at all...
2669 sub dup_changed_paths {
2670 my ($paths) = @_;
2671 return undef unless $paths;
2672 my %ret;
2673 foreach my $p (keys %$paths) {
2674 my $i = $paths->{$p};
2675 my %s = map { $_ => $i->$_ }
2676 qw/copyfrom_path copyfrom_rev action/;
2677 $ret{$p} = \%s;
2678 }
2679 \%ret;
2680 }
2681
2682 package Git::SVN::Log;
2683 use strict;
2684 use warnings;
2685 use POSIX qw/strftime/;
2686 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2687 %rusers $show_commit $incremental/;
2688 my $l_fmt;
2689
2690 sub cmt_showable {
2691 my ($c) = @_;
2692 return 1 if defined $c->{r};
2693 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2694 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2695 my @log = command(qw/cat-file commit/, $c->{c});
2696 shift @log while ($log[0] ne "\n");
2697 shift @log;
2698 @{$c->{l}} = grep !/^git-svn-id: /, @log;
2699
2700 (undef, $c->{r}, undef) = ::extract_metadata(
2701 (grep(/^git-svn-id: /, @log))[-1]);
2702 }
2703 return defined $c->{r};
2704 }
2705
2706 sub log_use_color {
2707 return 1 if $color;
2708 my ($dc, $dcvar);
2709 $dcvar = 'color.diff';
2710 $dc = `git-config --get $dcvar`;
2711 if ($dc eq '') {
2712 # nothing at all; fallback to "diff.color"
2713 $dcvar = 'diff.color';
2714 $dc = `git-config --get $dcvar`;
2715 }
2716 chomp($dc);
2717 if ($dc eq 'auto') {
2718 my $pc;
2719 $pc = `git-config --get color.pager`;
2720 if ($pc eq '') {
2721 # does not have it -- fallback to pager.color
2722 $pc = `git-config --bool --get pager.color`;
2723 }
2724 else {
2725 $pc = `git-config --bool --get color.pager`;
2726 if ($?) {
2727 $pc = 'false';
2728 }
2729 }
2730 chomp($pc);
2731 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2732 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2733 }
2734 return 0;
2735 }
2736 return 0 if $dc eq 'never';
2737 return 1 if $dc eq 'always';
2738 chomp($dc = `git-config --bool --get $dcvar`);
2739 return ($dc eq 'true');
2740 }
2741
2742 sub git_svn_log_cmd {
2743 my ($r_min, $r_max) = @_;
2744 my $gs = Git::SVN->_new;
2745 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2746 $gs->refname);
2747 push @cmd, '-r' unless $non_recursive;
2748 push @cmd, qw/--raw --name-status/ if $verbose;
2749 push @cmd, '--color' if log_use_color();
2750 return @cmd unless defined $r_max;
2751 if ($r_max == $r_min) {
2752 push @cmd, '--max-count=1';
2753 if (my $c = $gs->rev_db_get($r_max)) {
2754 push @cmd, $c;
2755 }
2756 } else {
2757 my ($c_min, $c_max);
2758 $c_max = $gs->rev_db_get($r_max);
2759 $c_min = $gs->rev_db_get($r_min);
2760 if (defined $c_min && defined $c_max) {
2761 if ($r_max > $r_max) {
2762 push @cmd, "$c_min..$c_max";
2763 } else {
2764 push @cmd, "$c_max..$c_min";
2765 }
2766 } elsif ($r_max > $r_min) {
2767 push @cmd, $c_max;
2768 } else {
2769 push @cmd, $c_min;
2770 }
2771 }
2772 return @cmd;
2773 }
2774
2775 # adapted from pager.c
2776 sub config_pager {
2777 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2778 if (!defined $pager) {
2779 $pager = 'less';
2780 } elsif (length $pager == 0 || $pager eq 'cat') {
2781 $pager = undef;
2782 }
2783 }
2784
2785 sub run_pager {
2786 return unless -t *STDOUT;
2787 pipe my $rfd, my $wfd or return;
2788 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2789 if (!$pid) {
2790 open STDOUT, '>&', $wfd or
2791 ::fatal "Can't redirect to stdout: $!\n";
2792 return;
2793 }
2794 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2795 $ENV{LESS} ||= 'FRSX';
2796 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2797 }
2798
2799 sub tz_to_s_offset {
2800 my ($tz) = @_;
2801 $tz =~ s/(\d\d)$//;
2802 return ($1 * 60) + ($tz * 3600);
2803 }
2804
2805 sub get_author_info {
2806 my ($dest, $author, $t, $tz) = @_;
2807 $author =~ s/(?:^\s*|\s*$)//g;
2808 $dest->{a_raw} = $author;
2809 my $au;
2810 if ($::_authors) {
2811 $au = $rusers{$author} || undef;
2812 }
2813 if (!$au) {
2814 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2815 }
2816 $dest->{t} = $t;
2817 $dest->{tz} = $tz;
2818 $dest->{a} = $au;
2819 # Date::Parse isn't in the standard Perl distro :(
2820 if ($tz =~ s/^\+//) {
2821 $t += tz_to_s_offset($tz);
2822 } elsif ($tz =~ s/^\-//) {
2823 $t -= tz_to_s_offset($tz);
2824 }
2825 $dest->{t_utc} = $t;
2826 }
2827
2828 sub process_commit {
2829 my ($c, $r_min, $r_max, $defer) = @_;
2830 if (defined $r_min && defined $r_max) {
2831 if ($r_min == $c->{r} && $r_min == $r_max) {
2832 show_commit($c);
2833 return 0;
2834 }
2835 return 1 if $r_min == $r_max;
2836 if ($r_min < $r_max) {
2837 # we need to reverse the print order
2838 return 0 if (defined $limit && --$limit < 0);
2839 push @$defer, $c;
2840 return 1;
2841 }
2842 if ($r_min != $r_max) {
2843 return 1 if ($r_min < $c->{r});
2844 return 1 if ($r_max > $c->{r});
2845 }
2846 }
2847 return 0 if (defined $limit && --$limit < 0);
2848 show_commit($c);
2849 return 1;
2850 }
2851
2852 sub show_commit {
2853 my $c = shift;
2854 if ($oneline) {
2855 my $x = "\n";
2856 if (my $l = $c->{l}) {
2857 while ($l->[0] =~ /^\s*$/) { shift @$l }
2858 $x = $l->[0];
2859 }
2860 $l_fmt ||= 'A' . length($c->{r});
2861 print 'r',pack($l_fmt, $c->{r}),' | ';
2862 print "$c->{c} | " if $show_commit;
2863 print $x;
2864 } else {
2865 show_commit_normal($c);
2866 }
2867 }
2868
2869 sub show_commit_changed_paths {
2870 my ($c) = @_;
2871 return unless $c->{changed};
2872 print "Changed paths:\n", @{$c->{changed}};
2873 }
2874
2875 sub show_commit_normal {
2876 my ($c) = @_;
2877 print '-' x72, "\nr$c->{r} | ";
2878 print "$c->{c} | " if $show_commit;
2879 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2880 localtime($c->{t_utc})), ' | ';
2881 my $nr_line = 0;
2882
2883 if (my $l = $c->{l}) {
2884 while ($l->[$#$l] eq "\n" && $#$l > 0
2885 && $l->[($#$l - 1)] eq "\n") {
2886 pop @$l;
2887 }
2888 $nr_line = scalar @$l;
2889 if (!$nr_line) {
2890 print "1 line\n\n\n";
2891 } else {
2892 if ($nr_line == 1) {
2893 $nr_line = '1 line';
2894 } else {
2895 $nr_line .= ' lines';
2896 }
2897 print $nr_line, "\n";
2898 show_commit_changed_paths($c);
2899 print "\n";
2900 print $_ foreach @$l;
2901 }
2902 } else {
2903 print "1 line\n";
2904 show_commit_changed_paths($c);
2905 print "\n";
2906
2907 }
2908 foreach my $x (qw/raw diff/) {
2909 if ($c->{$x}) {
2910 print "\n";
2911 print $_ foreach @{$c->{$x}}
2912 }
2913 }
2914 }
2915
2916 sub cmd_show_log {
2917 my (@args) = @_;
2918 my ($r_min, $r_max);
2919 my $r_last = -1; # prevent dupes
2920 if (defined $TZ) {
2921 $ENV{TZ} = $TZ;
2922 } else {
2923 delete $ENV{TZ};
2924 }
2925 if (defined $::_revision) {
2926 if ($::_revision =~ /^(\d+):(\d+)$/) {
2927 ($r_min, $r_max) = ($1, $2);
2928 } elsif ($::_revision =~ /^\d+$/) {
2929 $r_min = $r_max = $::_revision;
2930 } else {
2931 ::fatal "-r$::_revision is not supported, use ",
2932 "standard \'git log\' arguments instead\n";
2933 }
2934 }
2935
2936 config_pager();
2937 @args = (git_svn_log_cmd($r_min, $r_max), @args);
2938 my $log = command_output_pipe(@args);
2939 run_pager();
2940 my (@k, $c, $d);
2941 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
2942 while (<$log>) {
2943 if (/^${esc_color}commit ($::sha1_short)/o) {
2944 my $cmt = $1;
2945 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
2946 $r_last = $c->{r};
2947 process_commit($c, $r_min, $r_max, \@k) or
2948 goto out;
2949 }
2950 $d = undef;
2951 $c = { c => $cmt };
2952 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
2953 get_author_info($c, $1, $2, $3);
2954 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
2955 # ignore
2956 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
2957 push @{$c->{raw}}, $_;
2958 } elsif (/^${esc_color}[ACRMDT]\t/) {
2959 # we could add $SVN->{svn_path} here, but that requires
2960 # remote access at the moment (repo_path_split)...
2961 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
2962 push @{$c->{changed}}, $_;
2963 } elsif (/^${esc_color}diff /o) {
2964 $d = 1;
2965 push @{$c->{diff}}, $_;
2966 } elsif ($d) {
2967 push @{$c->{diff}}, $_;
2968 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
2969 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
2970 } elsif (s/^${esc_color} //o) {
2971 push @{$c->{l}}, $_;
2972 }
2973 }
2974 if ($c && defined $c->{r} && $c->{r} != $r_last) {
2975 $r_last = $c->{r};
2976 process_commit($c, $r_min, $r_max, \@k);
2977 }
2978 if (@k) {
2979 my $swap = $r_max;
2980 $r_max = $r_min;
2981 $r_min = $swap;
2982 process_commit($_, $r_min, $r_max) foreach reverse @k;
2983 }
2984 out:
2985 close $log;
2986 print '-' x72,"\n" unless $incremental || $oneline;
2987 }
2988
2989 package Git::SVN::Migration;
2990 # these version numbers do NOT correspond to actual version numbers
2991 # of git nor git-svn. They are just relative.
2992 #
2993 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
2994 #
2995 # v1 layout: .git/$id/info/url, refs/remotes/$id
2996 #
2997 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
2998 #
2999 # v3 layout: .git/svn/$id, refs/remotes/$id
3000 # - info/url may remain for backwards compatibility
3001 # - this is what we migrate up to this layout automatically,
3002 # - this will be used by git svn init on single branches
3003 #
3004 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3005 # - this is only created for newly multi-init-ed
3006 # repositories. Similar in spirit to the
3007 # --use-separate-remotes option in git-clone (now default)
3008 # - we do not automatically migrate to this (following
3009 # the example set by core git)
3010 use strict;
3011 use warnings;
3012 use Carp qw/croak/;
3013 use File::Path qw/mkpath/;
3014 use File::Basename qw/dirname basename/;
3015 use vars qw/$_minimize/;
3016
3017 sub migrate_from_v0 {
3018 my $git_dir = $ENV{GIT_DIR};
3019 return undef unless -d $git_dir;
3020 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3021 my $migrated = 0;
3022 while (<$fh>) {
3023 chomp;
3024 my ($id, $orig_ref) = ($_, $_);
3025 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3026 next unless -f "$git_dir/$id/info/url";
3027 my $new_ref = "refs/remotes/$id";
3028 if (::verify_ref("$new_ref^0")) {
3029 print STDERR "W: $orig_ref is probably an old ",
3030 "branch used by an ancient version of ",
3031 "git-svn.\n",
3032 "However, $new_ref also exists.\n",
3033 "We will not be able ",
3034 "to use this branch until this ",
3035 "ambiguity is resolved.\n";
3036 next;
3037 }
3038 print STDERR "Migrating from v0 layout...\n" if !$migrated;
3039 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3040 command_noisy('update-ref', $new_ref, $orig_ref);
3041 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3042 $migrated++;
3043 }
3044 command_close_pipe($fh, $ctx);
3045 print STDERR "Done migrating from v0 layout...\n" if $migrated;
3046 $migrated;
3047 }
3048
3049 sub migrate_from_v1 {
3050 my $git_dir = $ENV{GIT_DIR};
3051 my $migrated = 0;
3052 return $migrated unless -d $git_dir;
3053 my $svn_dir = "$git_dir/svn";
3054
3055 # just in case somebody used 'svn' as their $id at some point...
3056 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3057
3058 print STDERR "Migrating from a git-svn v1 layout...\n";
3059 mkpath([$svn_dir]);
3060 print STDERR "Data from a previous version of git-svn exists, but\n\t",
3061 "$svn_dir\n\t(required for this version ",
3062 "($::VERSION) of git-svn) does not. exist\n";
3063 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3064 while (<$fh>) {
3065 my $x = $_;
3066 next unless $x =~ s#^refs/remotes/##;
3067 chomp $x;
3068 next unless -f "$git_dir/$x/info/url";
3069 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3070 next unless $u;
3071 my $dn = dirname("$git_dir/svn/$x");
3072 mkpath([$dn]) unless -d $dn;
3073 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3074 mkpath(["$git_dir/svn/svn"]);
3075 print STDERR " - $git_dir/$x/info => ",
3076 "$git_dir/svn/$x/info\n";
3077 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3078 croak "$!: $x";
3079 # don't worry too much about these, they probably
3080 # don't exist with repos this old (save for index,
3081 # and we can easily regenerate that)
3082 foreach my $f (qw/unhandled.log index .rev_db/) {
3083 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3084 }
3085 } else {
3086 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3087 rename "$git_dir/$x", "$git_dir/svn/$x" or
3088 croak "$!: $x";
3089 }
3090 $migrated++;
3091 }
3092 command_close_pipe($fh, $ctx);
3093 print STDERR "Done migrating from a git-svn v1 layout\n";
3094 $migrated;
3095 }
3096
3097 sub read_old_urls {
3098 my ($l_map, $pfx, $path) = @_;
3099 my @dir;
3100 foreach (<$path/*>) {
3101 if (-r "$_/info/url") {
3102 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3103 my $ref_id = $pfx . basename $_;
3104 my $url = ::file_to_s("$_/info/url");
3105 $l_map->{$ref_id} = $url;
3106 } elsif (-d $_) {
3107 push @dir, $_;
3108 }
3109 }
3110 foreach (@dir) {
3111 my $x = $_;
3112 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3113 read_old_urls($l_map, $x, $_);
3114 }
3115 }
3116
3117 sub migrate_from_v2 {
3118 my @cfg = command(qw/config -l/);
3119 return if grep /^svn-remote\..+\.url=/, @cfg;
3120 my %l_map;
3121 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3122 my $migrated = 0;
3123
3124 foreach my $ref_id (sort keys %l_map) {
3125 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3126 if ($@) {
3127 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3128 }
3129 $migrated++;
3130 }
3131 $migrated;
3132 }
3133
3134 sub minimize_connections {
3135 my $r = Git::SVN::read_all_remotes();
3136 my $new_urls = {};
3137 my $root_repos = {};
3138 foreach my $repo_id (keys %$r) {
3139 my $url = $r->{$repo_id}->{url} or next;
3140 my $fetch = $r->{$repo_id}->{fetch} or next;
3141 my $ra = Git::SVN::Ra->new($url);
3142
3143 # skip existing cases where we already connect to the root
3144 if (($ra->{url} eq $ra->{repos_root}) ||
3145 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3146 $repo_id)) {
3147 $root_repos->{$ra->{url}} = $repo_id;
3148 next;
3149 }
3150
3151 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3152 my $root_path = $ra->{url};
3153 $root_path =~ s#^\Q$ra->{repos_root}\E/*##;
3154 foreach my $path (keys %$fetch) {
3155 my $ref_id = $fetch->{$path};
3156 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3157
3158 # make sure we can read when connecting to
3159 # a higher level of a repository
3160 my ($last_rev, undef) = $gs->last_rev_commit;
3161 if (!defined $last_rev) {
3162 $last_rev = eval {
3163 $root_ra->get_latest_revnum;
3164 };
3165 next if $@;
3166 }
3167 my $new = $root_path;
3168 $new .= length $path ? "/$path" : '';
3169 eval {
3170 $root_ra->get_log([$new], $last_rev, $last_rev,
3171 0, 0, 1, sub { });
3172 };
3173 next if $@;
3174 $new_urls->{$ra->{repos_root}}->{$new} =
3175 { ref_id => $ref_id,
3176 old_repo_id => $repo_id,
3177 old_path => $path };
3178 }
3179 }
3180
3181 my @emptied;
3182 foreach my $url (keys %$new_urls) {
3183 # see if we can re-use an existing [svn-remote "repo_id"]
3184 # instead of creating a(n ugly) new section:
3185 my $repo_id = $root_repos->{$url} ||
3186 Git::SVN::sanitize_remote_name($url);
3187
3188 my $fetch = $new_urls->{$url};
3189 foreach my $path (keys %$fetch) {
3190 my $x = $fetch->{$path};
3191 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3192 my $pfx = "svn-remote.$x->{old_repo_id}";
3193
3194 my $old_fetch = quotemeta("$x->{old_path}:".
3195 "refs/remotes/$x->{ref_id}");
3196 command_noisy(qw/config --unset/,
3197 "$pfx.fetch", '^'. $old_fetch . '$');
3198 delete $r->{$x->{old_repo_id}}->
3199 {fetch}->{$x->{old_path}};
3200 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3201 command_noisy(qw/config --unset/,
3202 "$pfx.url");
3203 push @emptied, $x->{old_repo_id}
3204 }
3205 }
3206 }
3207 if (@emptied) {
3208 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3209 "$ENV{GIT_DIR}/config";
3210 print STDERR <<EOF;
3211 The following [svn-remote] sections in your config file ($file) are empty
3212 and can be safely removed:
3213 EOF
3214 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3215 }
3216 }
3217
3218 sub migration_check {
3219 migrate_from_v0();
3220 migrate_from_v1();
3221 migrate_from_v2();
3222 minimize_connections() if $_minimize;
3223 }
3224
3225 package Git::IndexInfo;
3226 use strict;
3227 use warnings;
3228 use Git qw/command_input_pipe command_close_pipe/;
3229
3230 sub new {
3231 my ($class) = @_;
3232 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3233 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3234 }
3235
3236 sub remove {
3237 my ($self, $path) = @_;
3238 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3239 return ++$self->{nr};
3240 }
3241 undef;
3242 }
3243
3244 sub update {
3245 my ($self, $mode, $hash, $path) = @_;
3246 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3247 return ++$self->{nr};
3248 }
3249 undef;
3250 }
3251
3252 sub DESTROY {
3253 my ($self) = @_;
3254 command_close_pipe($self->{gui}, $self->{ctx});
3255 }
3256
3257 package Git::SVN::GlobSpec;
3258 use strict;
3259 use warnings;
3260
3261 sub new {
3262 my ($class, $glob) = @_;
3263 my $re = $glob;
3264 $re =~ s!/+$!!g; # no need for trailing slashes
3265 my $nr = ($re =~ s!^(.*/?)\*(/?.*)$!\(\[^/\]+\)!g);
3266 my ($left, $right) = ($1, $2);
3267 if ($nr > 1) {
3268 die "Only one '*' wildcard expansion ",
3269 "is supported (got $nr): '$glob'\n";
3270 } elsif ($nr == 0) {
3271 die "One '*' is needed for glob: '$glob'\n";
3272 }
3273 $re = quotemeta($left) . $re . quotemeta($right);
3274 $left =~ s!/+$!!g;
3275 $right =~ s!^/+!!g;
3276 bless { left => $left, right => $right,
3277 regex => qr/$re/, glob => $glob }, $class;
3278 }
3279
3280 sub full_path {
3281 my ($self, $path) = @_;
3282 return (length $self->{left} ? "$self->{left}/" : '') .
3283 $path . (length $self->{right} ? "/$self->{right}" : '');
3284 }
3285
3286 __END__
3287
3288 Data structures:
3289
3290
3291 $remotes = { # returned by read_all_remotes()
3292 'svn' => {
3293 # svn-remote.svn.url=https://svn.musicpd.org
3294 url => 'https://svn.musicpd.org',
3295 # svn-remote.svn.fetch=mpd/trunk:trunk
3296 fetch => {
3297 'mpd/trunk' => 'trunk',
3298 },
3299 # svn-remote.svn.tags=mpd/tags/*:tags/*
3300 tags => {
3301 path => {
3302 left => 'mpd/tags',
3303 right => '',
3304 regex => qr!mpd/tags/([^/]+)$!,
3305 glob => 'tags/*',
3306 },
3307 ref => {
3308 left => 'tags',
3309 right => '',
3310 regex => qr!tags/([^/]+)$!,
3311 glob => 'tags/*',
3312 },
3313 }
3314 }
3315 };
3316
3317 $log_entry hashref as returned by libsvn_log_entry()
3318 {
3319 log => 'whitespace-formatted log entry
3320 ', # trailing newline is preserved
3321 revision => '8', # integer
3322 date => '2004-02-24T17:01:44.108345Z', # commit date
3323 author => 'committer name'
3324 };
3325
3326
3327 # this is generated by generate_diff();
3328 @mods = array of diff-index line hashes, each element represents one line
3329 of diff-index output
3330
3331 diff-index line ($m hash)
3332 {
3333 mode_a => first column of diff-index output, no leading ':',
3334 mode_b => second column of diff-index output,
3335 sha1_b => sha1sum of the final blob,
3336 chg => change type [MCRADT],
3337 file_a => original file name of a file (iff chg is 'C' or 'R')
3338 file_b => new/current file name of a file (any chg)
3339 }
3340 ;
3341
3342 # retval of read_url_paths{,_all}();
3343 $l_map = {
3344 # repository root url
3345 'https://svn.musicpd.org' => {
3346 # repository path # GIT_SVN_ID
3347 'mpd/trunk' => 'trunk',
3348 'mpd/tags/0.11.5' => 'tags/0.11.5',
3349 },
3350 }
3351
3352 Notes:
3353 I don't trust the each() function on unless I created %hash myself
3354 because the internal iterator may not have started at base.
The Best VCS Around