?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-svn: allow multi-fetch to fetch things chronologically
[git/dscho.git] / git-svn.perl
blob5d398ee65fa1afafc21798c189330bac7e3220ff
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 $SVN_URL
8 $GIT_SVN_INDEX $GIT_SVN
9 $GIT_DIR $GIT_SVN_DIR $REVDB
10 $_follow_parent $sha1 $sha1_short $_revision
11 $_cp_remote $_upgrade $_rmdir $_q $_cp_similarity
12 $_find_copies_harder $_l $_authors %users/;
13 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
14 $VERSION = '@@GIT_VERSION@@';
15
16 $ENV{GIT_DIR} ||= '.git';
17 $Git::SVN::default_repo_id = 'git-svn';
18 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
19
20 my $LC_ALL = $ENV{LC_ALL};
21 $Git::SVN::Log::TZ = $ENV{TZ};
22 # make sure the svn binary gives consistent output between locales and TZs:
23 $ENV{TZ} = 'UTC';
24 $ENV{LC_ALL} = 'C';
25 $| = 1; # unbuffer STDOUT
26
27 sub fatal (@) { print STDERR @_; exit 1 }
28 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
29 require SVN::Ra;
30 require SVN::Delta;
31 if ($SVN::Core::VERSION lt '1.1.0') {
32 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
33 }
34 push @Git::SVN::Ra::ISA, 'SVN::Ra';
35 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
36 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
37 use Carp qw/croak/;
38 use IO::File qw//;
39 use File::Basename qw/dirname basename/;
40 use File::Path qw/mkpath/;
41 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
42 use IPC::Open3;
43 use Git;
44
45 BEGIN {
46 my $s;
47 foreach (qw/command command_oneline command_noisy command_output_pipe
48 command_input_pipe command_close_pipe/) {
49 $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
50 "*Git::SVN::Migration::$_ = ".
51 "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
52 }
53 eval $s;
54 }
55
56 my ($SVN);
57
58 my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
59 $sha1 = qr/[a-f\d]{40}/;
60 $sha1_short = qr/[a-f\d]{4,40}/;
61 my ($_stdin, $_help, $_edit,
62 $_repack, $_repack_nr, $_repack_flags,
63 $_message, $_file, $_no_metadata,
64 $_template, $_shared,
65 $_version, $_upgrade,
66 $_merge, $_strategy, $_dry_run,
67 $_prefix);
68
69 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
70 'config-dir=s' => \$Git::SVN::Ra::config_dir,
71 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
72 my %fc_opts = ( 'follow-parent|follow' => \$_follow_parent,
73 'authors-file|A=s' => \$_authors,
74 'repack:i' => \$_repack,
75 'no-metadata' => \$_no_metadata,
76 'quiet|q' => \$_q,
77 'repack-flags|repack-args|repack-opts=s' => \$_repack_flags,
78 %remote_opts );
79
80 my ($_trunk, $_tags, $_branches);
81 my %multi_opts = ( 'trunk|T=s' => \$_trunk,
82 'tags|t=s' => \$_tags,
83 'branches|b=s' => \$_branches );
84 my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
85 my %cmt_opts = ( 'edit|e' => \$_edit,
86 'rmdir' => \$_rmdir,
87 'find-copies-harder' => \$_find_copies_harder,
88 'l=i' => \$_l,
89 'copy-similarity|C=i'=> \$_cp_similarity
90 );
91
92 my %cmd = (
93 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
94 { 'revision|r=s' => \$_revision, %fc_opts } ],
95 init => [ \&cmd_init, "Initialize a repo for tracking" .
96 " (requires URL argument)",
97 \%init_opts ],
98 dcommit => [ \&cmd_dcommit,
99 'Commit several diffs to merge with upstream',
100 { 'merge|m|M' => \$_merge,
101 'strategy|s=s' => \$_strategy,
102 'dry-run|n' => \$_dry_run,
103 %cmt_opts, %fc_opts } ],
104 'set-tree' => [ \&cmd_set_tree,
105 "Set an SVN repository to a git tree-ish",
106 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
107 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
108 { 'revision|r=i' => \$_revision } ],
109 rebuild => [ \&cmd_rebuild, "Rebuild git-svn metadata (after git clone)",
110 { 'copy-remote|remote=s' => \$_cp_remote,
111 'upgrade' => \$_upgrade } ],
112 'multi-init' => [ \&cmd_multi_init,
113 'Initialize multiple trees (like git-svnimport)',
114 { %multi_opts, %init_opts, %remote_opts,
115 'revision|r=i' => \$_revision,
116 'prefix=s' => \$_prefix,
117 } ],
118 'multi-fetch' => [ \&cmd_multi_fetch,
119 'Fetch multiple trees (like git-svnimport)',
120 \%fc_opts ],
121 'migrate' => [ sub { },
122 # no-op, we automatically run this anyways,
123 'Migrate configuration/metadata/layout from
124 previous versions of git-svn',
125 \%remote_opts ],
126 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
127 { 'limit=i' => \$Git::SVN::Log::limit,
128 'revision|r=s' => \$_revision,
129 'verbose|v' => \$Git::SVN::Log::verbose,
130 'incremental' => \$Git::SVN::Log::incremental,
131 'oneline' => \$Git::SVN::Log::oneline,
132 'show-commit' => \$Git::SVN::Log::show_commit,
133 'non-recursive' => \$Git::SVN::Log::non_recursive,
134 'authors-file|A=s' => \$_authors,
135 'color' => \$Git::SVN::Log::color,
136 'pager=s' => \$Git::SVN::Log::pager,
137 } ],
138 'commit-diff' => [ \&cmd_commit_diff,
139 'Commit a diff between two trees',
140 { 'message|m=s' => \$_message,
141 'file|F=s' => \$_file,
142 'revision|r=s' => \$_revision,
143 %cmt_opts } ],
144 );
145
146 my $cmd;
147 for (my $i = 0; $i < @ARGV; $i++) {
148 if (defined $cmd{$ARGV[$i]}) {
149 $cmd = $ARGV[$i];
150 splice @ARGV, $i, 1;
151 last;
152 }
153 };
154
155 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
156
157 read_repo_config(\%opts);
158 my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
159 'version|V' => \$_version,
160 'minimize-connections' =>
161 \$Git::SVN::Migration::_minimize,
162 'id|i=s' => \$Git::SVN::default_ref_id);
163 exit 1 if (!$rv && $cmd ne 'log');
164
165 usage(0) if $_help;
166 version() if $_version;
167 usage(1) unless defined $cmd;
168 load_authors() if $_authors;
169 unless ($cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/) {
170 Git::SVN::Migration::migration_check();
171 }
172 eval {
173 Git::SVN::verify_remotes_sanity();
174 $cmd{$cmd}->[0]->(@ARGV);
175 };
176 fatal $@ if $@;
177 exit 0;
178
179 ####################### primary functions ######################
180 sub usage {
181 my $exit = shift || 0;
182 my $fd = $exit ? \*STDERR : \*STDOUT;
183 print $fd <<"";
184 git-svn - bidirectional operations between a single Subversion tree and git
185 Usage: $0 <command> [options] [arguments]\n
186
187 print $fd "Available commands:\n" unless $cmd;
188
189 foreach (sort keys %cmd) {
190 next if $cmd && $cmd ne $_;
191 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
192 foreach (keys %{$cmd{$_}->[2]}) {
193 # prints out arguments as they should be passed:
194 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
195 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
196 "--$_" : "-$_" }
197 split /\|/,$_)," $x\n";
198 }
199 }
200 print $fd <<"";
201 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
202 arbitrary identifier if you're tracking multiple SVN branches/repositories in
203 one git repository and want to keep them separate. See git-svn(1) for more
204 information.
205
206 exit $exit;
207 }
208
209 sub version {
210 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
211 exit 0;
212 }
213
214 sub cmd_rebuild {
215 my $url = shift;
216 my $gs = $url ? Git::SVN->init($url)
217 : eval { Git::SVN->new };
218 $gs ||= Git::SVN->_new;
219 if (!verify_ref($gs->refname.'^0')) {
220 $gs->copy_remote_ref;
221 }
222
223 my ($rev_list, $ctx) = command_output_pipe("rev-list", $gs->refname);
224 my $latest;
225 my $svn_uuid;
226 while (<$rev_list>) {
227 chomp;
228 my $c = $_;
229 fatal "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
230 my ($url, $rev, $uuid) = cmt_metadata($c);
231
232 # ignore merges (from set-tree)
233 next if (!defined $rev || !$uuid);
234
235 # if we merged or otherwise started elsewhere, this is
236 # how we break out of it
237 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
238 ($gs->{url} && $url && ($url ne $gs->{url}))) {
239 next;
240 }
241
242 unless (defined $latest) {
243 if (!$gs->{url} && !$url) {
244 fatal "SVN repository location required\n";
245 }
246 $gs = Git::SVN->init($url);
247 $latest = $rev;
248 }
249 $gs->rev_db_set($rev, $c);
250 print "r$rev = $c\n";
251 }
252 command_close_pipe($rev_list, $ctx);
253 }
254
255 sub do_git_init_db {
256 unless (-d $ENV{GIT_DIR}) {
257 my @init_db = ('init');
258 push @init_db, "--template=$_template" if defined $_template;
259 push @init_db, "--shared" if defined $_shared;
260 command_noisy(@init_db);
261 }
262 }
263
264 sub cmd_init {
265 my $url = shift or die "SVN repository location required " .
266 "as a command-line argument\n";
267 if (my $repo_path = shift) {
268 unless (-d $repo_path) {
269 mkpath([$repo_path]);
270 }
271 chdir $repo_path or croak $!;
272 $ENV{GIT_DIR} = $repo_path . "/.git";
273 }
274 do_git_init_db();
275
276 Git::SVN->init($url);
277 }
278
279 sub cmd_fetch {
280 if (@_) {
281 die "Additional fetch arguments are no longer supported.\n",
282 "Use --follow-parent if you have moved/copied directories
283 instead.\n";
284 }
285 my $gs = Git::SVN->new;
286 $gs->fetch(parse_revision_argument());
287 if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
288 command_noisy(qw(update-ref refs/heads/master),
289 $gs->{last_commit});
290 }
291 }
292
293 sub cmd_set_tree {
294 my (@commits) = @_;
295 if ($_stdin || !@commits) {
296 print "Reading from stdin...\n";
297 @commits = ();
298 while (<STDIN>) {
299 if (/\b($sha1_short)\b/o) {
300 unshift @commits, $1;
301 }
302 }
303 }
304 my @revs;
305 foreach my $c (@commits) {
306 my @tmp = command('rev-parse',$c);
307 if (scalar @tmp == 1) {
308 push @revs, $tmp[0];
309 } elsif (scalar @tmp > 1) {
310 push @revs, reverse(command('rev-list',@tmp));
311 } else {
312 fatal "Failed to rev-parse $c\n";
313 }
314 }
315 my $gs = Git::SVN->new;
316 my ($r_last, $cmt_last) = $gs->last_rev_commit;
317 $gs->fetch;
318 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
319 fatal "There are new revisions that were fetched ",
320 "and need to be merged (or acknowledged) ",
321 "before committing.\nlast rev: $r_last\n",
322 " current: $gs->{last_rev}\n";
323 }
324 $gs->set_tree($_) foreach @revs;
325 print "Done committing ",scalar @revs," revisions to SVN\n";
326 }
327
328 sub cmd_dcommit {
329 my $head = shift;
330 my $gs = Git::SVN->new;
331 $head ||= 'HEAD';
332 my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
333 my $last_rev;
334 foreach my $d (reverse @refs) {
335 if (!verify_ref("$d~1")) {
336 fatal "Commit $d\n",
337 "has no parent commit, and therefore ",
338 "nothing to diff against.\n",
339 "You should be working from a repository ",
340 "originally created by git-svn\n";
341 }
342 unless (defined $last_rev) {
343 (undef, $last_rev, undef) = cmt_metadata("$d~1");
344 unless (defined $last_rev) {
345 fatal "Unable to extract revision information ",
346 "from commit $d~1\n";
347 }
348 }
349 if ($_dry_run) {
350 print "diff-tree $d~1 $d\n";
351 } else {
352 my %ed_opts = ( r => $last_rev,
353 log => get_commit_entry($d)->{log},
354 ra => $gs->ra,
355 tree_a => "$d~1",
356 tree_b => $d,
357 editor_cb => sub {
358 print "Committed r$_[0]\n";
359 $last_rev = $_[0]; },
360 svn_path => $gs->{path} );
361 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
362 print "No changes\n$d~1 == $d\n";
363 }
364 }
365 }
366 return if $_dry_run;
367 $gs->fetch;
368 # we always want to rebase against the current HEAD, not any
369 # head that was passed to us
370 my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
371 my @finish;
372 if (@diff) {
373 @finish = qw/rebase/;
374 push @finish, qw/--merge/ if $_merge;
375 push @finish, "--strategy=$_strategy" if $_strategy;
376 print STDERR "W: HEAD and ", $gs->refname, " differ, ",
377 "using @finish:\n", "@diff";
378 } else {
379 print "No changes between current HEAD and ",
380 $gs->refname, "\nResetting to the latest ",
381 $gs->refname, "\n";
382 @finish = qw/reset --mixed/;
383 }
384 command_noisy(@finish, $gs->refname);
385 }
386
387 sub cmd_show_ignore {
388 my $gs = Git::SVN->new;
389 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
390 $gs->traverse_ignore(\*STDOUT, '', $r);
391 }
392
393 sub cmd_multi_init {
394 my $url = shift;
395 unless (defined $_trunk || defined $_branches || defined $_tags) {
396 usage(1);
397 }
398 do_git_init_db();
399 $_prefix = '' unless defined $_prefix;
400 $url =~ s#/+$## if defined $url;
401 if (defined $_trunk) {
402 my $trunk_ref = $_prefix . 'trunk';
403 # try both old-style and new-style lookups:
404 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
405 unless ($gs_trunk) {
406 my ($trunk_url, $trunk_path) =
407 complete_svn_url($url, $_trunk);
408 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
409 undef, $trunk_ref);
410 }
411 }
412 return unless defined $_branches || defined $_tags;
413 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
414 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
415 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
416 }
417
418 sub cmd_multi_fetch {
419 my $remotes = Git::SVN::read_all_remotes();
420 foreach my $repo_id (sort keys %$remotes) {
421 my $url = $remotes->{$repo_id}->{url} or next;
422 my $fetch = $remotes->{$repo_id}->{fetch} or next;
423 Git::SVN::fetch_all($repo_id, $url, $fetch);
424 }
425 }
426
427 # this command is special because it requires no metadata
428 sub cmd_commit_diff {
429 my ($ta, $tb, $url) = @_;
430 my $usage = "Usage: $0 commit-diff -r<revision> ".
431 "<tree-ish> <tree-ish> [<URL>]\n";
432 fatal($usage) if (!defined $ta || !defined $tb);
433 my $svn_path;
434 if (!defined $url) {
435 my $gs = eval { Git::SVN->new };
436 if (!$gs) {
437 fatal("Needed URL or usable git-svn --id in ",
438 "the command-line\n", $usage);
439 }
440 $url = $gs->{url};
441 $svn_path = $gs->{path};
442 }
443 unless (defined $_revision) {
444 fatal("-r|--revision is a required argument\n", $usage);
445 }
446 if (defined $_message && defined $_file) {
447 fatal("Both --message/-m and --file/-F specified ",
448 "for the commit message.\n",
449 "I have no idea what you mean\n");
450 }
451 if (defined $_file) {
452 $_message = file_to_s($_file);
453 } else {
454 $_message ||= get_commit_entry($tb)->{log};
455 }
456 my $ra ||= Git::SVN::Ra->new($url);
457 $svn_path ||= $ra->{svn_path};
458 my $r = $_revision;
459 if ($r eq 'HEAD') {
460 $r = $ra->get_latest_revnum;
461 } elsif ($r !~ /^\d+$/) {
462 die "revision argument: $r not understood by git-svn\n";
463 }
464 my %ed_opts = ( r => $r,
465 log => $_message,
466 ra => $ra,
467 tree_a => $ta,
468 tree_b => $tb,
469 editor_cb => sub { print "Committed r$_[0]\n" },
470 svn_path => $svn_path );
471 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
472 print "No changes\n$ta == $tb\n";
473 }
474 }
475
476 ########################### utility functions #########################
477
478 sub parse_revision_argument {
479 if (!defined $_revision || $_revision eq 'BASE:HEAD') {
480 return (undef, undef);
481 }
482 return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
483 return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
484 return (undef, $1) if ($_revision =~ /^BASE:(\d+)$/);
485 return ($1, undef) if ($_revision =~ /^(\d+):HEAD$/);
486 die "revision argument: $_revision not understood by git-svn\n",
487 "Try using the command-line svn client instead\n";
488 }
489
490 sub complete_svn_url {
491 my ($url, $path) = @_;
492 $path =~ s#/+$##;
493 if ($path !~ m#^[a-z\+]+://#) {
494 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
495 fatal("E: '$path' is not a complete URL ",
496 "and a separate URL is not specified\n");
497 }
498 return ($url, $path);
499 }
500 return ($path, '');
501 }
502
503 sub complete_url_ls_init {
504 my ($ra, $repo_path, $switch, $pfx) = @_;
505 unless ($repo_path) {
506 print STDERR "W: $switch not specified\n";
507 return;
508 }
509 $repo_path =~ s#/+$##;
510 if ($repo_path =~ m#^[a-z\+]+://#) {
511 $ra = Git::SVN::Ra->new($repo_path);
512 $repo_path = '';
513 } else {
514 $repo_path =~ s#^/+##;
515 unless ($ra) {
516 fatal("E: '$repo_path' is not a complete URL ",
517 "and a separate URL is not specified\n");
518 }
519 }
520 my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
521 my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
522 my $url = $ra->{url};
523 foreach my $d (sort keys %$dirent) {
524 next if ($dirent->{$d}->kind != $SVN::Node::dir);
525 my $path = "$repo_path/$d";
526 my $ref = "$pfx$d";
527 my $gs = eval { Git::SVN->new($ref) };
528 # don't try to init already existing refs
529 unless ($gs) {
530 print "init $url/$path => $ref\n";
531 Git::SVN->init($url, $path, undef, $ref);
532 }
533 }
534 }
535
536 sub verify_ref {
537 my ($ref) = @_;
538 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
539 { STDERR => 0 }); };
540 }
541
542 sub get_tree_from_treeish {
543 my ($treeish) = @_;
544 # $treeish can be a symbolic ref, too:
545 my $type = command_oneline(qw/cat-file -t/, $treeish);
546 my $expected;
547 while ($type eq 'tag') {
548 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
549 }
550 if ($type eq 'commit') {
551 $expected = (grep /^tree /, command(qw/cat-file commit/,
552 $treeish))[0];
553 ($expected) = ($expected =~ /^tree ($sha1)$/o);
554 die "Unable to get tree from $treeish\n" unless $expected;
555 } elsif ($type eq 'tree') {
556 $expected = $treeish;
557 } else {
558 die "$treeish is a $type, expected tree, tag or commit\n";
559 }
560 return $expected;
561 }
562
563 sub get_commit_entry {
564 my ($treeish) = shift;
565 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
566 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
567 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
568 open my $log_fh, '>', $commit_editmsg or croak $!;
569
570 my $type = command_oneline(qw/cat-file -t/, $treeish);
571 if ($type eq 'commit' || $type eq 'tag') {
572 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
573 $type, $treeish);
574 my $in_msg = 0;
575 while (<$msg_fh>) {
576 if (!$in_msg) {
577 $in_msg = 1 if (/^\s*$/);
578 } elsif (/^git-svn-id: /) {
579 # skip this for now, we regenerate the
580 # correct one on re-fetch anyways
581 # TODO: set *:merge properties or like...
582 } else {
583 print $log_fh $_ or croak $!;
584 }
585 }
586 command_close_pipe($msg_fh, $ctx);
587 }
588 close $log_fh or croak $!;
589
590 if ($_edit || ($type eq 'tree')) {
591 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
592 # TODO: strip out spaces, comments, like git-commit.sh
593 system($editor, $commit_editmsg);
594 }
595 rename $commit_editmsg, $commit_msg or croak $!;
596 open $log_fh, '<', $commit_msg or croak $!;
597 { local $/; chomp($log_entry{log} = <$log_fh>); }
598 close $log_fh or croak $!;
599 unlink $commit_msg;
600 \%log_entry;
601 }
602
603 sub s_to_file {
604 my ($str, $file, $mode) = @_;
605 open my $fd,'>',$file or croak $!;
606 print $fd $str,"\n" or croak $!;
607 close $fd or croak $!;
608 chmod ($mode &~ umask, $file) if (defined $mode);
609 }
610
611 sub file_to_s {
612 my $file = shift;
613 open my $fd,'<',$file or croak "$!: file: $file\n";
614 local $/;
615 my $ret = <$fd>;
616 close $fd or croak $!;
617 $ret =~ s/\s*$//s;
618 return $ret;
619 }
620
621 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
622 sub load_authors {
623 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
624 my $log = $cmd eq 'log';
625 while (<$authors>) {
626 chomp;
627 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
628 my ($user, $name, $email) = ($1, $2, $3);
629 if ($log) {
630 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
631 } else {
632 $users{$user} = [$name, $email];
633 }
634 }
635 close $authors or croak $!;
636 }
637
638 # convert GetOpt::Long specs for use by git-config
639 sub read_repo_config {
640 return unless -d $ENV{GIT_DIR};
641 my $opts = shift;
642 foreach my $o (keys %$opts) {
643 my $v = $opts->{$o};
644 my ($key) = ($o =~ /^([a-z\-]+)/);
645 $key =~ s/-//g;
646 my $arg = 'git-config';
647 $arg .= ' --int' if ($o =~ /[:=]i$/);
648 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
649 if (ref $v eq 'ARRAY') {
650 chomp(my @tmp = `$arg --get-all svn.$key`);
651 @$v = @tmp if @tmp;
652 } else {
653 chomp(my $tmp = `$arg --get svn.$key`);
654 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
655 $$v = $tmp;
656 }
657 }
658 }
659 }
660
661 sub extract_metadata {
662 my $id = shift or return (undef, undef, undef);
663 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
664 \s([a-f\d\-]+)$/x);
665 if (!defined $rev || !$uuid || !$url) {
666 # some of the original repositories I made had
667 # identifiers like this:
668 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
669 }
670 return ($url, $rev, $uuid);
671 }
672
673 sub cmt_metadata {
674 return extract_metadata((grep(/^git-svn-id: /,
675 command(qw/cat-file commit/, shift)))[-1]);
676 }
677
678 package Git::SVN;
679 use strict;
680 use warnings;
681 use vars qw/$default_repo_id $default_ref_id/;
682 use Carp qw/croak/;
683 use File::Path qw/mkpath/;
684 use IPC::Open3;
685
686 # properties that we do not log:
687 my %SKIP_PROP;
688 BEGIN {
689 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
690 svn:special svn:executable
691 svn:entry:committed-rev
692 svn:entry:last-author
693 svn:entry:uuid
694 svn:entry:committed-date/;
695 }
696
697 sub fetch_all {
698 my ($repo_id, $url, $fetch) = @_;
699 my @gs;
700 my $ra = Git::SVN::Ra->new($url);
701 my $head = $ra->get_latest_revnum;
702 my $base = $head;
703 my $new_remote;
704 foreach my $p (sort keys %$fetch) {
705 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
706 my $lr = $gs->last_rev;
707 if (defined $lr) {
708 $base = $lr if ($lr < $base);
709 } else {
710 $new_remote = 1;
711 }
712 push @gs, $gs;
713 }
714 $base = 0 if $new_remote;
715 return if (++$base > $head);
716 $ra->gs_fetch_loop_common($base, $head, @gs);
717 }
718
719 sub read_all_remotes {
720 my $r = {};
721 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
722 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
723 $r->{$1}->{fetch}->{$2} = $3;
724 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
725 $r->{$1}->{url} = $2;
726 }
727 }
728 $r;
729 }
730
731 sub verify_remotes_sanity {
732 return unless -d $ENV{GIT_DIR};
733 my %seen;
734 foreach (command(qw/config -l/)) {
735 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
736 if ($seen{$1}) {
737 die "Remote ref refs/remote/$1 is tracked by",
738 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
739 "Please resolve this ambiguity in ",
740 "your git configuration file before ",
741 "continuing\n";
742 }
743 $seen{$1} = $_;
744 }
745 }
746 }
747
748 # we allow more chars than remotes2config.sh...
749 sub sanitize_remote_name {
750 my ($name) = @_;
751 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
752 $name;
753 }
754
755 sub find_existing_remote {
756 my ($url, $remotes) = @_;
757 my $existing;
758 foreach my $repo_id (keys %$remotes) {
759 my $u = $remotes->{$repo_id}->{url} or next;
760 next if $u ne $url;
761 $existing = $repo_id;
762 last;
763 }
764 $existing;
765 }
766
767 sub init_remote_config {
768 my ($self, $url) = @_;
769 $url =~ s!/+$!!; # strip trailing slash
770 my $r = read_all_remotes();
771 my $existing = find_existing_remote($url, $r);
772 if ($existing) {
773 print STDERR "Using existing ",
774 "[svn-remote \"$existing\"]\n";
775 $self->{repo_id} = $existing;
776 } else {
777 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
778 $existing = find_existing_remote($min_url, $r);
779 if ($existing) {
780 print STDERR "Using existing ",
781 "[svn-remote \"$existing\"]\n";
782 $self->{repo_id} = $existing;
783 }
784 if ($min_url ne $url) {
785 print STDERR "Using higher level of URL: ",
786 "$url => $min_url\n";
787 my $old_path = $self->{path};
788 $self->{path} = $url;
789 $self->{path} =~ s!^\Q$min_url\E/*!!;
790 if (length $old_path) {
791 $self->{path} .= "/$old_path";
792 }
793 $url = $min_url;
794 }
795 }
796 my $orig_url;
797 if (!$existing) {
798 # verify that we aren't overwriting anything:
799 $orig_url = eval {
800 command_oneline('config', '--get',
801 "svn-remote.$self->{repo_id}.url")
802 };
803 if ($orig_url && ($orig_url ne $url)) {
804 die "svn-remote.$self->{repo_id}.url already set: ",
805 "$orig_url\nwanted to set to: $url\n";
806 }
807 }
808 my ($xrepo_id, $xpath) = find_ref($self->refname);
809 if (defined $xpath) {
810 die "svn-remote.$xrepo_id.fetch already set to track ",
811 "$xpath:refs/remotes/", $self->refname, "\n";
812 }
813 command_noisy('config',
814 "svn-remote.$self->{repo_id}.url", $url);
815 command_noisy('config', '--add',
816 "svn-remote.$self->{repo_id}.fetch",
817 "$self->{path}:".$self->refname);
818 $self->{url} = $url;
819 }
820
821 sub init {
822 my ($class, $url, $path, $repo_id, $ref_id) = @_;
823 my $self = _new($class, $repo_id, $ref_id, $path);
824 if (defined $url) {
825 $self->init_remote_config($url);
826 }
827 $self;
828 }
829
830 sub find_ref {
831 my ($ref_id) = @_;
832 foreach (command(qw/config -l/)) {
833 next unless m!^svn-remote\.(.+)\.fetch=
834 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
835 my ($repo_id, $path, $ref) = ($1, $2, $3);
836 if ($ref eq $ref_id) {
837 $path = '' if ($path =~ m#^\./?#);
838 return ($repo_id, $path);
839 }
840 }
841 (undef, undef, undef);
842 }
843
844 sub new {
845 my ($class, $ref_id, $repo_id, $path) = @_;
846 if (defined $ref_id && !defined $repo_id && !defined $path) {
847 ($repo_id, $path) = find_ref($ref_id);
848 if (!defined $repo_id) {
849 die "Could not find a \"svn-remote.*.fetch\" key ",
850 "in the repository configuration matching: ",
851 "refs/remotes/$ref_id\n";
852 }
853 }
854 my $self = _new($class, $repo_id, $ref_id, $path);
855 if (!defined $self->{path} || !length $self->{path}) {
856 my $fetch = command_oneline('config', '--get',
857 "svn-remote.$repo_id.fetch",
858 ":refs/remotes/$ref_id\$") or
859 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
860 "\":refs/remotes/$ref_id\$\" in config\n";
861 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
862 }
863 $self->{url} = command_oneline('config', '--get',
864 "svn-remote.$repo_id.url") or
865 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
866 $self;
867 }
868
869 sub refname { "refs/remotes/$_[0]->{ref_id}" }
870
871 sub ra {
872 my ($self) = shift;
873 $self->{ra} ||= Git::SVN::Ra->new($self->{url});
874 }
875
876 sub rel_path {
877 my ($self) = @_;
878 my $repos_root = $self->ra->{repos_root};
879 return $self->{path} if ($self->{url} eq $repos_root);
880 my $url = $self->{url} .
881 (length $self->{path} ? "/$self->{path}" : $self->{path});
882 $url =~ s!^\Q$repos_root\E/*!!g;
883 $url;
884 }
885
886 sub copy_remote_ref {
887 my ($self) = @_;
888 my $origin = $::_cp_remote ? $::_cp_remote : 'origin';
889 my $ref = $self->refname;
890 if (command('ls-remote', $origin, $ref)) {
891 command_noisy('fetch', $origin, "$ref:$ref");
892 } elsif ($::_cp_remote && !$::_upgrade) {
893 die "Unable to find remote reference: $ref on $origin\n";
894 }
895 }
896
897 sub traverse_ignore {
898 my ($self, $fh, $path, $r) = @_;
899 $path =~ s#^/+##g;
900 my ($dirent, undef, $props) = $self->ra->get_dir($path, $r);
901 my $p = $path;
902 $p =~ s#^\Q$self->{ra}->{svn_path}\E/##;
903 print $fh length $p ? "\n# $p\n" : "\n# /\n";
904 if (my $s = $props->{'svn:ignore'}) {
905 $s =~ s/[\r\n]+/\n/g;
906 chomp $s;
907 if (length $p == 0) {
908 $s =~ s#\n#\n/$p#g;
909 print $fh "/$s\n";
910 } else {
911 $s =~ s#\n#\n/$p/#g;
912 print $fh "/$p/$s\n";
913 }
914 }
915 foreach (sort keys %$dirent) {
916 next if $dirent->{$_}->kind != $SVN::Node::dir;
917 $self->traverse_ignore($fh, "$path/$_", $r);
918 }
919 }
920
921 sub last_rev { ($_[0]->last_rev_commit)[0] }
922 sub last_commit { ($_[0]->last_rev_commit)[1] }
923
924 # returns the newest SVN revision number and newest commit SHA1
925 sub last_rev_commit {
926 my ($self) = @_;
927 if (defined $self->{last_rev} && defined $self->{last_commit}) {
928 return ($self->{last_rev}, $self->{last_commit});
929 }
930 my $c = ::verify_ref($self->refname.'^0');
931 if ($c) {
932 my $rev = (::cmt_metadata($c))[1];
933 if (defined $rev) {
934 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
935 return ($rev, $c);
936 }
937 }
938 my $offset = -41; # from tail
939 my $rl;
940 open my $fh, '<', $self->{db_path} or
941 croak "$self->{db_path} not readable: $!\n";
942 seek $fh, $offset, 2;
943 $rl = readline $fh;
944 defined $rl or return (undef, undef);
945 chomp $rl;
946 while ($c ne $rl && tell $fh != 0) {
947 $offset -= 41;
948 seek $fh, $offset, 2;
949 $rl = readline $fh;
950 defined $rl or return (undef, undef);
951 chomp $rl;
952 }
953 my $rev = tell $fh;
954 croak $! if ($rev < 0);
955 $rev = ($rev - 41) / 41;
956 close $fh or croak $!;
957 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
958 return ($rev, $c);
959 }
960
961 sub get_fetch_range {
962 my ($self, $min, $max) = @_;
963 $max ||= $self->ra->get_latest_revnum;
964 $min ||= $self->last_rev || 0;
965 (++$min, $max);
966 }
967
968 sub tmp_index_do {
969 my ($self, $sub) = @_;
970 my $old_index = $ENV{GIT_INDEX_FILE};
971 $ENV{GIT_INDEX_FILE} = $self->{index};
972 my @ret = &$sub;
973 if ($old_index) {
974 $ENV{GIT_INDEX_FILE} = $old_index;
975 } else {
976 delete $ENV{GIT_INDEX_FILE};
977 }
978 wantarray ? @ret : $ret[0];
979 }
980
981 sub assert_index_clean {
982 my ($self, $treeish) = @_;
983
984 $self->tmp_index_do(sub {
985 command_noisy('read-tree', $treeish) unless -e $self->{index};
986 my $x = command_oneline('write-tree');
987 my ($y) = (command(qw/cat-file commit/, $treeish) =~
988 /^tree ($::sha1)/mo);
989 if ($y ne $x) {
990 unlink $self->{index} or croak $!;
991 command_noisy('read-tree', $treeish);
992 }
993 $x = command_oneline('write-tree');
994 if ($y ne $x) {
995 ::fatal "trees ($treeish) $y != $x\n",
996 "Something is seriously wrong...\n";
997 }
998 });
999 }
1000
1001 sub get_commit_parents {
1002 my ($self, $log_entry) = @_;
1003 my (%seen, @ret, @tmp);
1004 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1005 if (my $ip = $self->{inject_parents}) {
1006 if (my $commit = delete $ip->{$log_entry->{revision}}) {
1007 push @tmp, $commit;
1008 }
1009 }
1010 if (my $cur = ::verify_ref($self->refname.'^0')) {
1011 push @tmp, $cur;
1012 }
1013 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1014 while (my $p = shift @tmp) {
1015 next if $seen{$p};
1016 $seen{$p} = 1;
1017 push @ret, $p;
1018 # MAXPARENT is defined to 16 in commit-tree.c:
1019 last if @ret >= 16;
1020 }
1021 if (@tmp) {
1022 die "r$log_entry->{revision}: No room for parents:\n\t",
1023 join("\n\t", @tmp), "\n";
1024 }
1025 @ret;
1026 }
1027
1028 sub full_url {
1029 my ($self) = @_;
1030 $self->ra->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1031 }
1032
1033 sub do_git_commit {
1034 my ($self, $log_entry) = @_;
1035 if (my $c = $self->rev_db_get($log_entry->{revision})) {
1036 croak "$log_entry->{revision} = $c already exists! ",
1037 "Why are we refetching it?\n";
1038 }
1039 my $author = $log_entry->{author};
1040 my ($name, $email) = (defined $::users{$author} ? @{$::users{$author}}
1041 : ($author, "$author\@".$self->ra->uuid));
1042 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1043 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1044 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1045
1046 my $tree = $log_entry->{tree};
1047 if (!defined $tree) {
1048 $tree = $self->tmp_index_do(sub {
1049 command_oneline('write-tree') });
1050 }
1051 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1052
1053 my @exec = ('git-commit-tree', $tree);
1054 foreach ($self->get_commit_parents($log_entry)) {
1055 push @exec, '-p', $_;
1056 }
1057 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1058 or croak $!;
1059 print $msg_fh $log_entry->{log} or croak $!;
1060 print $msg_fh "\ngit-svn-id: ", $self->full_url, '@',
1061 $log_entry->{revision}, ' ',
1062 $self->ra->uuid, "\n" or croak $!;
1063 $msg_fh->flush == 0 or croak $!;
1064 close $msg_fh or croak $!;
1065 chomp(my $commit = do { local $/; <$out_fh> });
1066 close $out_fh or croak $!;
1067 waitpid $pid, 0;
1068 croak $? if $?;
1069 if ($commit !~ /^$::sha1$/o) {
1070 die "Failed to commit, invalid sha1: $commit\n";
1071 }
1072
1073 command_noisy('update-ref',$self->refname, $commit);
1074 $self->rev_db_set($log_entry->{revision}, $commit);
1075
1076 $self->{last_rev} = $log_entry->{revision};
1077 $self->{last_commit} = $commit;
1078 print "r$log_entry->{revision} = $commit\n";
1079 return $commit;
1080 }
1081
1082 sub revisions_eq {
1083 my ($self, $r0, $r1) = @_;
1084 return 1 if $r0 == $r1;
1085 my $nr = 0;
1086 $self->ra->get_log([$self->{path}], $r0, $r1,
1087 0, 0, 1, sub { $nr++ });
1088 return 0 if ($nr > 1);
1089 return 1;
1090 }
1091
1092 sub find_parent_branch {
1093 my ($self, $paths, $rev) = @_;
1094 return undef unless $::_follow_parent;
1095 unless (defined $paths) {
1096 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
1097 sub { $paths = dup_changed_paths($_[0]) });
1098 }
1099 return undef unless defined $paths;
1100
1101 # look for a parent from another branch:
1102 my @b_path_components = split m#/#, $self->rel_path;
1103 my @a_path_components;
1104 my $i;
1105 while (@b_path_components) {
1106 $i = $paths->{'/'.join('/', @b_path_components)};
1107 last if $i;
1108 unshift(@a_path_components, pop(@b_path_components));
1109 }
1110 goto not_found unless defined $i;
1111 my $branch_from = $i->{copyfrom_path} or goto not_found;
1112 if (@a_path_components) {
1113 print STDERR "branch_from: $branch_from => ";
1114 $branch_from .= '/'.join('/', @a_path_components);
1115 print STDERR $branch_from, "\n";
1116 }
1117 my $r = $i->{copyfrom_rev};
1118 my $repos_root = $self->ra->{repos_root};
1119 my $url = $self->ra->{url};
1120 my $new_url = $repos_root . $branch_from;
1121 print STDERR "Found possible branch point: ",
1122 "$new_url => ", $self->full_url, ", $r\n";
1123 $branch_from =~ s#^/##;
1124 my $remotes = read_all_remotes();
1125 my $gs;
1126 foreach my $repo_id (keys %$remotes) {
1127 my $u = $remotes->{$repo_id}->{url} or next;
1128 next if $url ne $u;
1129 my $fetch = $remotes->{$repo_id}->{fetch};
1130 foreach my $f (keys %$fetch) {
1131 next if $f ne $branch_from;
1132 $gs = Git::SVN->new($fetch->{$f}, $repo_id, $f);
1133 last;
1134 }
1135 last if $gs;
1136 }
1137 unless ($gs) {
1138 my $ref_id = $branch_from;
1139 $ref_id .= "\@$r" if find_ref($ref_id);
1140 # just grow a tail if we're not unique enough :x
1141 $ref_id .= '-' while find_ref($ref_id);
1142 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id);
1143 }
1144 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1145 if ($::_follow_parent && (!defined $r0 || !defined $parent)) {
1146 $gs->fetch(0, $r);
1147 ($r0, $parent) = $gs->last_rev_commit;
1148 }
1149 if (defined $r0 && defined $parent && $gs->revisions_eq($r0, $r)) {
1150 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1151 $self->assert_index_clean($parent);
1152 my $ed;
1153 if ($self->ra->can_do_switch) {
1154 print STDERR "Following parent with do_switch\n";
1155 # do_switch works with svn/trunk >= r22312, but that
1156 # is not included with SVN 1.4.2 (the latest version
1157 # at the moment), so we can't rely on it
1158 $self->{last_commit} = $parent;
1159 $ed = SVN::Git::Fetcher->new($self);
1160 $gs->ra->gs_do_switch($r0, $rev, $gs->{path}, 1,
1161 $self->full_url, $ed)
1162 or die "SVN connection failed somewhere...\n";
1163 } else {
1164 print STDERR "Following parent with do_update\n";
1165 $ed = SVN::Git::Fetcher->new($self);
1166 $self->ra->gs_do_update($rev, $rev, $self->{path},
1167 1, $ed)
1168 or die "SVN connection failed somewhere...\n";
1169 }
1170 return $self->make_log_entry($rev, [$parent], $ed);
1171 }
1172 not_found:
1173 print STDERR "Branch parent for path: '/",
1174 $self->rel_path, "' @ r$rev not found:\n";
1175 return undef unless $paths;
1176 print STDERR "Changed paths:\n";
1177 foreach my $x (sort keys %$paths) {
1178 my $p = $paths->{$x};
1179 print STDERR "\t$p->{action}\t$x";
1180 if ($p->{copyfrom_path}) {
1181 print STDERR "(from $p->{copyfrom_path}: ",
1182 "$p->{copyfrom_rev})";
1183 }
1184 print STDERR "\n";
1185 }
1186 print STDERR '-'x72, "\n";
1187 return undef;
1188 }
1189
1190 sub do_fetch {
1191 my ($self, $paths, $rev) = @_;
1192 my $ed;
1193 my ($last_rev, @parents);
1194 if ($self->{last_commit}) {
1195 $ed = SVN::Git::Fetcher->new($self);
1196 $last_rev = $self->{last_rev};
1197 $ed->{c} = $self->{last_commit};
1198 @parents = ($self->{last_commit});
1199 } else {
1200 $last_rev = $rev;
1201 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1202 return $log_entry;
1203 }
1204 $ed = SVN::Git::Fetcher->new($self);
1205 }
1206 unless ($self->ra->gs_do_update($last_rev, $rev,
1207 $self->{path}, 1, $ed)) {
1208 die "SVN connection failed somewhere...\n";
1209 }
1210 $self->make_log_entry($rev, \@parents, $ed);
1211 }
1212
1213 sub get_untracked {
1214 my ($self, $ed) = @_;
1215 my @out;
1216 my $h = $ed->{empty};
1217 foreach (sort keys %$h) {
1218 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1219 push @out, " $act: " . uri_encode($_);
1220 warn "W: $act: $_\n";
1221 }
1222 foreach my $t (qw/dir_prop file_prop/) {
1223 $h = $ed->{$t} or next;
1224 foreach my $path (sort keys %$h) {
1225 my $ppath = $path eq '' ? '.' : $path;
1226 foreach my $prop (sort keys %{$h->{$path}}) {
1227 next if $SKIP_PROP{$prop};
1228 my $v = $h->{$path}->{$prop};
1229 my $t_ppath_prop = "$t: " .
1230 uri_encode($ppath) . ' ' .
1231 uri_encode($prop);
1232 if (defined $v) {
1233 push @out, " +$t_ppath_prop " .
1234 uri_encode($v);
1235 } else {
1236 push @out, " -$t_ppath_prop";
1237 }
1238 }
1239 }
1240 }
1241 foreach my $t (qw/absent_file absent_directory/) {
1242 $h = $ed->{$t} or next;
1243 foreach my $parent (sort keys %$h) {
1244 foreach my $path (sort @{$h->{$parent}}) {
1245 push @out, " $t: " .
1246 uri_encode("$parent/$path");
1247 warn "W: $t: $parent/$path ",
1248 "Insufficient permissions?\n";
1249 }
1250 }
1251 }
1252 \@out;
1253 }
1254
1255 sub parse_svn_date {
1256 my $date = shift || return '+0000 1970-01-01 00:00:00';
1257 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1258 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1259 croak "Unable to parse date: $date\n";
1260 "+0000 $Y-$m-$d $H:$M:$S";
1261 }
1262
1263 sub check_author {
1264 my ($author) = @_;
1265 if (!defined $author || length $author == 0) {
1266 $author = '(no author)';
1267 }
1268 if (defined $::_authors && ! defined $::users{$author}) {
1269 die "Author: $author not defined in $::_authors file\n";
1270 }
1271 $author;
1272 }
1273
1274 sub make_log_entry {
1275 my ($self, $rev, $parents, $ed) = @_;
1276 my $untracked = $self->get_untracked($ed);
1277
1278 return undef if ($ed->{nr} == 0 && scalar @$untracked == 0);
1279
1280 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1281 print $un "r$rev\n" or croak $!;
1282 print $un $_, "\n" foreach @$untracked;
1283 my %log_entry = ( parents => $parents || [], revision => $rev,
1284 log => '');
1285 my $rp = $self->ra->rev_proplist($rev);
1286 foreach (sort keys %$rp) {
1287 my $v = $rp->{$_};
1288 if (/^svn:(author|date|log)$/) {
1289 $log_entry{$1} = $v;
1290 } else {
1291 print $un " rev_prop: ", uri_encode($_), ' ',
1292 uri_encode($v), "\n";
1293 }
1294 }
1295 close $un or croak $!;
1296
1297 $log_entry{date} = parse_svn_date($log_entry{date});
1298 $log_entry{author} = check_author($log_entry{author});
1299 $log_entry{log} .= "\n";
1300 \%log_entry;
1301 }
1302
1303 sub fetch {
1304 my ($self, $min_rev, $max_rev, @parents) = @_;
1305 my ($last_rev, $last_commit) = $self->last_rev_commit;
1306 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1307 return if ($base > $head);
1308 $self->ra->gs_fetch_loop_common($base, $head, $self);
1309 }
1310
1311 sub set_tree_cb {
1312 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1313 # TODO: enable and test optimized commits:
1314 if (0 && $rev == ($self->{last_rev} + 1)) {
1315 $log_entry->{revision} = $rev;
1316 $log_entry->{author} = $author;
1317 $self->do_git_commit($log_entry, "$rev=$tree");
1318 } else {
1319 $self->{inject_parents} = { $rev => $tree };
1320 $self->fetch(undef, undef);
1321 }
1322 }
1323
1324 sub set_tree {
1325 my ($self, $tree) = (shift, shift);
1326 my $log_entry = ::get_commit_entry($tree);
1327 unless ($self->{last_rev}) {
1328 fatal("Must have an existing revision to commit\n");
1329 }
1330 my %ed_opts = ( r => $self->{last_rev},
1331 log => $log_entry->{log},
1332 ra => $self->ra,
1333 tree_a => $self->{last_commit},
1334 tree_b => $tree,
1335 editor_cb => sub {
1336 $self->set_tree_cb($log_entry, $tree, @_) },
1337 svn_path => $self->{path} );
1338 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1339 print "No changes\nr$self->{last_rev} = $tree\n";
1340 }
1341 }
1342
1343 # rev_db:
1344 # Tie::File seems to be prone to offset errors if revisions get sparse,
1345 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
1346 # one of my favorite modules is out :< Next up would be one of the DBM
1347 # modules, but I'm not sure which is most portable... So I'll just
1348 # go with something that's plain-text, but still capable of
1349 # being randomly accessed. So here's my ultra-simple fixed-width
1350 # database. All records are 40 characters + "\n", so it's easy to seek
1351 # to a revision: (41 * rev) is the byte offset.
1352 # A record of 40 0s denotes an empty revision.
1353 # And yes, it's still pretty fast (faster than Tie::File).
1354
1355 sub rev_db_set {
1356 my ($self, $rev, $commit) = @_;
1357 length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
1358 open my $fh, '+<', $self->{db_path} or croak $!;
1359 my $offset = $rev * 41;
1360 # assume that append is the common case:
1361 seek $fh, 0, 2 or croak $!;
1362 my $pos = tell $fh;
1363 if ($pos < $offset) {
1364 print $fh (('0' x 40),"\n") x (($offset - $pos) / 41)
1365 or croak $!;
1366 }
1367 seek $fh, $offset, 0 or croak $!;
1368 print $fh $commit,"\n" or croak $!;
1369 close $fh or croak $!;
1370 }
1371
1372 sub rev_db_get {
1373 my ($self, $rev) = @_;
1374 my $ret;
1375 my $offset = $rev * 41;
1376 open my $fh, '<', $self->{db_path} or croak $!;
1377 if (seek $fh, $offset, 0) {
1378 $ret = readline $fh;
1379 if (defined $ret) {
1380 chomp $ret;
1381 $ret = undef if ($ret =~ /^0{40}$/);
1382 }
1383 }
1384 close $fh or croak $!;
1385 $ret;
1386 }
1387
1388 sub find_rev_before {
1389 my ($self, $rev, $eq_ok) = @_;
1390 --$rev unless $eq_ok;
1391 while ($rev > 0) {
1392 if (my $c = $self->rev_db_get($rev)) {
1393 return ($rev, $c);
1394 }
1395 --$rev;
1396 }
1397 return (undef, undef);
1398 }
1399
1400 sub _new {
1401 my ($class, $repo_id, $ref_id, $path) = @_;
1402 unless (defined $repo_id && length $repo_id) {
1403 $repo_id = $Git::SVN::default_repo_id;
1404 }
1405 unless (defined $ref_id && length $ref_id) {
1406 $_[2] = $ref_id = $Git::SVN::default_ref_id;
1407 }
1408 $_[1] = $repo_id = sanitize_remote_name($repo_id);
1409 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1410 $_[3] = $path = '' unless (defined $path);
1411 mkpath([$dir]);
1412 unless (-f "$dir/.rev_db") {
1413 open my $fh, '>>', "$dir/.rev_db" or croak $!;
1414 close $fh or croak $!;
1415 }
1416 bless { ref_id => $ref_id, dir => $dir, index => "$dir/index",
1417 path => $path,
1418 db_path => "$dir/.rev_db", repo_id => $repo_id }, $class;
1419 }
1420
1421 sub uri_encode {
1422 my ($f) = @_;
1423 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1424 $f
1425 }
1426
1427 package Git::SVN::Prompt;
1428 use strict;
1429 use warnings;
1430 require SVN::Core;
1431 use vars qw/$_no_auth_cache $_username/;
1432
1433 sub simple {
1434 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1435 $may_save = undef if $_no_auth_cache;
1436 $default_username = $_username if defined $_username;
1437 if (defined $default_username && length $default_username) {
1438 if (defined $realm && length $realm) {
1439 print STDERR "Authentication realm: $realm\n";
1440 STDERR->flush;
1441 }
1442 $cred->username($default_username);
1443 } else {
1444 username($cred, $realm, $may_save, $pool);
1445 }
1446 $cred->password(_read_password("Password for '" .
1447 $cred->username . "': ", $realm));
1448 $cred->may_save($may_save);
1449 $SVN::_Core::SVN_NO_ERROR;
1450 }
1451
1452 sub ssl_server_trust {
1453 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1454 $may_save = undef if $_no_auth_cache;
1455 print STDERR "Error validating server certificate for '$realm':\n";
1456 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1457 print STDERR " - The certificate is not issued by a trusted ",
1458 "authority. Use the\n",
1459 " fingerprint to validate the certificate manually!\n";
1460 }
1461 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1462 print STDERR " - The certificate hostname does not match.\n";
1463 }
1464 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1465 print STDERR " - The certificate is not yet valid.\n";
1466 }
1467 if ($failures & $SVN::Auth::SSL::EXPIRED) {
1468 print STDERR " - The certificate has expired.\n";
1469 }
1470 if ($failures & $SVN::Auth::SSL::OTHER) {
1471 print STDERR " - The certificate has an unknown error.\n";
1472 }
1473 printf STDERR
1474 "Certificate information:\n".
1475 " - Hostname: %s\n".
1476 " - Valid: from %s until %s\n".
1477 " - Issuer: %s\n".
1478 " - Fingerprint: %s\n",
1479 map $cert_info->$_, qw(hostname valid_from valid_until
1480 issuer_dname fingerprint);
1481 my $choice;
1482 prompt:
1483 print STDERR $may_save ?
1484 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1485 "(R)eject or accept (t)emporarily? ";
1486 STDERR->flush;
1487 $choice = lc(substr(<STDIN> || 'R', 0, 1));
1488 if ($choice =~ /^t$/i) {
1489 $cred->may_save(undef);
1490 } elsif ($choice =~ /^r$/i) {
1491 return -1;
1492 } elsif ($may_save && $choice =~ /^p$/i) {
1493 $cred->may_save($may_save);
1494 } else {
1495 goto prompt;
1496 }
1497 $cred->accepted_failures($failures);
1498 $SVN::_Core::SVN_NO_ERROR;
1499 }
1500
1501 sub ssl_client_cert {
1502 my ($cred, $realm, $may_save, $pool) = @_;
1503 $may_save = undef if $_no_auth_cache;
1504 print STDERR "Client certificate filename: ";
1505 STDERR->flush;
1506 chomp(my $filename = <STDIN>);
1507 $cred->cert_file($filename);
1508 $cred->may_save($may_save);
1509 $SVN::_Core::SVN_NO_ERROR;
1510 }
1511
1512 sub ssl_client_cert_pw {
1513 my ($cred, $realm, $may_save, $pool) = @_;
1514 $may_save = undef if $_no_auth_cache;
1515 $cred->password(_read_password("Password: ", $realm));
1516 $cred->may_save($may_save);
1517 $SVN::_Core::SVN_NO_ERROR;
1518 }
1519
1520 sub username {
1521 my ($cred, $realm, $may_save, $pool) = @_;
1522 $may_save = undef if $_no_auth_cache;
1523 if (defined $realm && length $realm) {
1524 print STDERR "Authentication realm: $realm\n";
1525 }
1526 my $username;
1527 if (defined $_username) {
1528 $username = $_username;
1529 } else {
1530 print STDERR "Username: ";
1531 STDERR->flush;
1532 chomp($username = <STDIN>);
1533 }
1534 $cred->username($username);
1535 $cred->may_save($may_save);
1536 $SVN::_Core::SVN_NO_ERROR;
1537 }
1538
1539 sub _read_password {
1540 my ($prompt, $realm) = @_;
1541 print STDERR $prompt;
1542 STDERR->flush;
1543 require Term::ReadKey;
1544 Term::ReadKey::ReadMode('noecho');
1545 my $password = '';
1546 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1547 last if $key =~ /[\012\015]/; # \n\r
1548 $password .= $key;
1549 }
1550 Term::ReadKey::ReadMode('restore');
1551 print STDERR "\n";
1552 STDERR->flush;
1553 $password;
1554 }
1555
1556 package main;
1557
1558 sub uri_encode {
1559 my ($f) = @_;
1560 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1561 $f
1562 }
1563
1564 sub uri_decode {
1565 my ($f) = @_;
1566 $f =~ tr/+/ /;
1567 $f =~ s/%([A-F0-9]{2})/chr hex($1)/ge;
1568 $f
1569 }
1570
1571 {
1572 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1573 $SVN::Node::dir.$SVN::Node::unknown.
1574 $SVN::Node::none.$SVN::Node::file.
1575 $SVN::Node::dir.$SVN::Node::unknown.
1576 $SVN::Auth::SSL::CNMISMATCH.
1577 $SVN::Auth::SSL::NOTYETVALID.
1578 $SVN::Auth::SSL::EXPIRED.
1579 $SVN::Auth::SSL::UNKNOWNCA.
1580 $SVN::Auth::SSL::OTHER;
1581 }
1582
1583 package SVN::Git::Fetcher;
1584 use vars qw/@ISA/;
1585 use strict;
1586 use warnings;
1587 use Carp qw/croak/;
1588 use IO::File qw//;
1589
1590 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
1591 sub new {
1592 my ($class, $git_svn) = @_;
1593 my $self = SVN::Delta::Editor->new;
1594 bless $self, $class;
1595 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
1596 $self->{empty} = {};
1597 $self->{dir_prop} = {};
1598 $self->{file_prop} = {};
1599 $self->{absent_dir} = {};
1600 $self->{absent_file} = {};
1601 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
1602 require Digest::MD5;
1603 $self;
1604 }
1605
1606 sub set_path_strip {
1607 my ($self, $path) = @_;
1608 $self->{path_strip} = qr/^\Q$path\E\/?/;
1609 }
1610
1611 sub open_root {
1612 { path => '' };
1613 }
1614
1615 sub open_directory {
1616 my ($self, $path, $pb, $rev) = @_;
1617 { path => $path };
1618 }
1619
1620 sub git_path {
1621 my ($self, $path) = @_;
1622 $path =~ s!$self->{path_strip}!! if $self->{path_strip};
1623 $path;
1624 }
1625
1626 sub delete_entry {
1627 my ($self, $path, $rev, $pb) = @_;
1628
1629 my $gpath = $self->git_path($path);
1630 # remove entire directories.
1631 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
1632 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
1633 -r --name-only -z/,
1634 $self->{c}, '--', $gpath);
1635 local $/ = "\0";
1636 while (<$ls>) {
1637 chomp;
1638 $self->{gii}->remove($_);
1639 print "\tD\t$_\n" unless $self->{q};
1640 }
1641 print "\tD\t$gpath/\n" unless $self->{q};
1642 command_close_pipe($ls, $ctx);
1643 $self->{empty}->{$path} = 0
1644 } else {
1645 $self->{gii}->remove($gpath);
1646 print "\tD\t$gpath\n" unless $self->{q};
1647 }
1648 undef;
1649 }
1650
1651 sub open_file {
1652 my ($self, $path, $pb, $rev) = @_;
1653 my $gpath = $self->git_path($path);
1654 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
1655 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
1656 unless (defined $mode && defined $blob) {
1657 die "$path was not found in commit $self->{c} (r$rev)\n";
1658 }
1659 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
1660 pool => SVN::Pool->new, action => 'M' };
1661 }
1662
1663 sub add_file {
1664 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
1665 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1666 delete $self->{empty}->{$dir};
1667 { path => $path, mode_a => 100644, mode_b => 100644,
1668 pool => SVN::Pool->new, action => 'A' };
1669 }
1670
1671 sub add_directory {
1672 my ($self, $path, $cp_path, $cp_rev) = @_;
1673 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1674 delete $self->{empty}->{$dir};
1675 $self->{empty}->{$path} = 1;
1676 { path => $path };
1677 }
1678
1679 sub change_dir_prop {
1680 my ($self, $db, $prop, $value) = @_;
1681 $self->{dir_prop}->{$db->{path}} ||= {};
1682 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
1683 undef;
1684 }
1685
1686 sub absent_directory {
1687 my ($self, $path, $pb) = @_;
1688 $self->{absent_dir}->{$pb->{path}} ||= [];
1689 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
1690 undef;
1691 }
1692
1693 sub absent_file {
1694 my ($self, $path, $pb) = @_;
1695 $self->{absent_file}->{$pb->{path}} ||= [];
1696 push @{$self->{absent_file}->{$pb->{path}}}, $path;
1697 undef;
1698 }
1699
1700 sub change_file_prop {
1701 my ($self, $fb, $prop, $value) = @_;
1702 if ($prop eq 'svn:executable') {
1703 if ($fb->{mode_b} != 120000) {
1704 $fb->{mode_b} = defined $value ? 100755 : 100644;
1705 }
1706 } elsif ($prop eq 'svn:special') {
1707 $fb->{mode_b} = defined $value ? 120000 : 100644;
1708 } else {
1709 $self->{file_prop}->{$fb->{path}} ||= {};
1710 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
1711 }
1712 undef;
1713 }
1714
1715 sub apply_textdelta {
1716 my ($self, $fb, $exp) = @_;
1717 my $fh = IO::File->new_tmpfile;
1718 $fh->autoflush(1);
1719 # $fh gets auto-closed() by SVN::TxDelta::apply(),
1720 # (but $base does not,) so dup() it for reading in close_file
1721 open my $dup, '<&', $fh or croak $!;
1722 my $base = IO::File->new_tmpfile;
1723 $base->autoflush(1);
1724 if ($fb->{blob}) {
1725 defined (my $pid = fork) or croak $!;
1726 if (!$pid) {
1727 open STDOUT, '>&', $base or croak $!;
1728 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
1729 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
1730 }
1731 waitpid $pid, 0;
1732 croak $? if $?;
1733
1734 if (defined $exp) {
1735 seek $base, 0, 0 or croak $!;
1736 my $md5 = Digest::MD5->new;
1737 $md5->addfile($base);
1738 my $got = $md5->hexdigest;
1739 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
1740 "expected: $exp\n",
1741 " got: $got\n" if ($got ne $exp);
1742 }
1743 }
1744 seek $base, 0, 0 or croak $!;
1745 $fb->{fh} = $dup;
1746 $fb->{base} = $base;
1747 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
1748 }
1749
1750 sub close_file {
1751 my ($self, $fb, $exp) = @_;
1752 my $hash;
1753 my $path = $self->git_path($fb->{path});
1754 if (my $fh = $fb->{fh}) {
1755 seek($fh, 0, 0) or croak $!;
1756 my $md5 = Digest::MD5->new;
1757 $md5->addfile($fh);
1758 my $got = $md5->hexdigest;
1759 die "Checksum mismatch: $path\n",
1760 "expected: $exp\n got: $got\n" if ($got ne $exp);
1761 seek($fh, 0, 0) or croak $!;
1762 if ($fb->{mode_b} == 120000) {
1763 read($fh, my $buf, 5) == 5 or croak $!;
1764 $buf eq 'link ' or die "$path has mode 120000",
1765 "but is not a link\n";
1766 }
1767 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
1768 if (!$pid) {
1769 open STDIN, '<&', $fh or croak $!;
1770 exec qw/git-hash-object -w --stdin/ or croak $!;
1771 }
1772 chomp($hash = do { local $/; <$out> });
1773 close $out or croak $!;
1774 close $fh or croak $!;
1775 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
1776 close $fb->{base} or croak $!;
1777 } else {
1778 $hash = $fb->{blob} or die "no blob information\n";
1779 }
1780 $fb->{pool}->clear;
1781 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
1782 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
1783 undef;
1784 }
1785
1786 sub abort_edit {
1787 my $self = shift;
1788 $self->{nr} = $self->{gii}->{nr};
1789 delete $self->{gii};
1790 $self->SUPER::abort_edit(@_);
1791 }
1792
1793 sub close_edit {
1794 my $self = shift;
1795 $self->{git_commit_ok} = 1;
1796 $self->{nr} = $self->{gii}->{nr};
1797 delete $self->{gii};
1798 $self->SUPER::close_edit(@_);
1799 }
1800
1801 package SVN::Git::Editor;
1802 use vars qw/@ISA/;
1803 use strict;
1804 use warnings;
1805 use Carp qw/croak/;
1806 use IO::File;
1807
1808 sub new {
1809 my ($class, $opts) = @_;
1810 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
1811 die "$_ required!\n" unless (defined $opts->{$_});
1812 }
1813
1814 my $pool = SVN::Pool->new;
1815 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
1816 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
1817 $opts->{r}, $mods);
1818
1819 # $opts->{ra} functions should not be used after this:
1820 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
1821 $opts->{editor_cb}, $pool);
1822 my $self = SVN::Delta::Editor->new(@ce, $pool);
1823 bless $self, $class;
1824 foreach (qw/svn_path r tree_a tree_b/) {
1825 $self->{$_} = $opts->{$_};
1826 }
1827 $self->{url} = $opts->{ra}->{url};
1828 $self->{mods} = $mods;
1829 $self->{types} = $types;
1830 $self->{pool} = $pool;
1831 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
1832 $self->{rm} = { };
1833 $self->{path_prefix} = length $self->{svn_path} ?
1834 "$self->{svn_path}/" : '';
1835 require Digest::MD5;
1836 return $self;
1837 }
1838
1839 sub generate_diff {
1840 my ($tree_a, $tree_b) = @_;
1841 my @diff_tree = qw(diff-tree -z -r);
1842 if ($::_cp_similarity) {
1843 push @diff_tree, "-C$::_cp_similarity";
1844 } else {
1845 push @diff_tree, '-C';
1846 }
1847 push @diff_tree, '--find-copies-harder' if $::_find_copies_harder;
1848 push @diff_tree, "-l$::_l" if defined $::_l;
1849 push @diff_tree, $tree_a, $tree_b;
1850 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
1851 local $/ = "\0";
1852 my $state = 'meta';
1853 my @mods;
1854 while (<$diff_fh>) {
1855 chomp $_; # this gets rid of the trailing "\0"
1856 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1857 $::sha1\s($::sha1)\s
1858 ([MTCRAD])\d*$/xo) {
1859 push @mods, { mode_a => $1, mode_b => $2,
1860 sha1_b => $3, chg => $4 };
1861 if ($4 =~ /^(?:C|R)$/) {
1862 $state = 'file_a';
1863 } else {
1864 $state = 'file_b';
1865 }
1866 } elsif ($state eq 'file_a') {
1867 my $x = $mods[$#mods] or croak "Empty array\n";
1868 if ($x->{chg} !~ /^(?:C|R)$/) {
1869 croak "Error parsing $_, $x->{chg}\n";
1870 }
1871 $x->{file_a} = $_;
1872 $state = 'file_b';
1873 } elsif ($state eq 'file_b') {
1874 my $x = $mods[$#mods] or croak "Empty array\n";
1875 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1876 croak "Error parsing $_, $x->{chg}\n";
1877 }
1878 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1879 croak "Error parsing $_, $x->{chg}\n";
1880 }
1881 $x->{file_b} = $_;
1882 $state = 'meta';
1883 } else {
1884 croak "Error parsing $_\n";
1885 }
1886 }
1887 command_close_pipe($diff_fh, $ctx);
1888 \@mods;
1889 }
1890
1891 sub check_diff_paths {
1892 my ($ra, $pfx, $rev, $mods) = @_;
1893 my %types;
1894 $pfx .= '/' if length $pfx;
1895
1896 sub type_diff_paths {
1897 my ($ra, $types, $path, $rev) = @_;
1898 my @p = split m#/+#, $path;
1899 my $c = shift @p;
1900 unless (defined $types->{$c}) {
1901 $types->{$c} = $ra->check_path($c, $rev);
1902 }
1903 while (@p) {
1904 $c .= '/' . shift @p;
1905 next if defined $types->{$c};
1906 $types->{$c} = $ra->check_path($c, $rev);
1907 }
1908 }
1909
1910 foreach my $m (@$mods) {
1911 foreach my $f (qw/file_a file_b/) {
1912 next unless defined $m->{$f};
1913 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
1914 if (length $pfx.$dir && ! defined $types{$dir}) {
1915 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
1916 }
1917 }
1918 }
1919 \%types;
1920 }
1921
1922 sub split_path {
1923 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
1924 }
1925
1926 sub repo_path {
1927 my ($self, $path) = @_;
1928 $self->{path_prefix}.(defined $path ? $path : '');
1929 }
1930
1931 sub url_path {
1932 my ($self, $path) = @_;
1933 $self->{url} . '/' . $self->repo_path($path);
1934 }
1935
1936 sub rmdirs {
1937 my ($self) = @_;
1938 my $rm = $self->{rm};
1939 delete $rm->{''}; # we never delete the url we're tracking
1940 return unless %$rm;
1941
1942 foreach (keys %$rm) {
1943 my @d = split m#/#, $_;
1944 my $c = shift @d;
1945 $rm->{$c} = 1;
1946 while (@d) {
1947 $c .= '/' . shift @d;
1948 $rm->{$c} = 1;
1949 }
1950 }
1951 delete $rm->{$self->{svn_path}};
1952 delete $rm->{''}; # we never delete the url we're tracking
1953 return unless %$rm;
1954
1955 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
1956 $self->{tree_b});
1957 local $/ = "\0";
1958 while (<$fh>) {
1959 chomp;
1960 my @dn = split m#/#, $_;
1961 while (pop @dn) {
1962 delete $rm->{join '/', @dn};
1963 }
1964 unless (%$rm) {
1965 close $fh;
1966 return;
1967 }
1968 }
1969 command_close_pipe($fh, $ctx);
1970
1971 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
1972 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
1973 $self->close_directory($bat->{$d}, $p);
1974 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
1975 print "\tD+\t$d/\n" unless $::_q;
1976 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
1977 delete $bat->{$d};
1978 }
1979 }
1980
1981 sub open_or_add_dir {
1982 my ($self, $full_path, $baton) = @_;
1983 my $t = $self->{types}->{$full_path};
1984 if (!defined $t) {
1985 die "$full_path not known in r$self->{r} or we have a bug!\n";
1986 }
1987 if ($t == $SVN::Node::none) {
1988 return $self->add_directory($full_path, $baton,
1989 undef, -1, $self->{pool});
1990 } elsif ($t == $SVN::Node::dir) {
1991 return $self->open_directory($full_path, $baton,
1992 $self->{r}, $self->{pool});
1993 }
1994 print STDERR "$full_path already exists in repository at ",
1995 "r$self->{r} and it is not a directory (",
1996 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
1997 exit 1;
1998 }
1999
2000 sub ensure_path {
2001 my ($self, $path) = @_;
2002 my $bat = $self->{bat};
2003 my $repo_path = $self->repo_path($path);
2004 return $bat->{''} unless (length $repo_path);
2005 my @p = split m#/+#, $repo_path;
2006 my $c = shift @p;
2007 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2008 while (@p) {
2009 my $c0 = $c;
2010 $c .= '/' . shift @p;
2011 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2012 }
2013 return $bat->{$c};
2014 }
2015
2016 sub A {
2017 my ($self, $m) = @_;
2018 my ($dir, $file) = split_path($m->{file_b});
2019 my $pbat = $self->ensure_path($dir);
2020 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2021 undef, -1);
2022 print "\tA\t$m->{file_b}\n" unless $::_q;
2023 $self->chg_file($fbat, $m);
2024 $self->close_file($fbat,undef,$self->{pool});
2025 }
2026
2027 sub C {
2028 my ($self, $m) = @_;
2029 my ($dir, $file) = split_path($m->{file_b});
2030 my $pbat = $self->ensure_path($dir);
2031 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2032 $self->url_path($m->{file_a}), $self->{r});
2033 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2034 $self->chg_file($fbat, $m);
2035 $self->close_file($fbat,undef,$self->{pool});
2036 }
2037
2038 sub delete_entry {
2039 my ($self, $path, $pbat) = @_;
2040 my $rpath = $self->repo_path($path);
2041 my ($dir, $file) = split_path($rpath);
2042 $self->{rm}->{$dir} = 1;
2043 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2044 }
2045
2046 sub R {
2047 my ($self, $m) = @_;
2048 my ($dir, $file) = split_path($m->{file_b});
2049 my $pbat = $self->ensure_path($dir);
2050 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2051 $self->url_path($m->{file_a}), $self->{r});
2052 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2053 $self->chg_file($fbat, $m);
2054 $self->close_file($fbat,undef,$self->{pool});
2055
2056 ($dir, $file) = split_path($m->{file_a});
2057 $pbat = $self->ensure_path($dir);
2058 $self->delete_entry($m->{file_a}, $pbat);
2059 }
2060
2061 sub M {
2062 my ($self, $m) = @_;
2063 my ($dir, $file) = split_path($m->{file_b});
2064 my $pbat = $self->ensure_path($dir);
2065 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2066 $pbat,$self->{r},$self->{pool});
2067 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2068 $self->chg_file($fbat, $m);
2069 $self->close_file($fbat,undef,$self->{pool});
2070 }
2071
2072 sub T { shift->M(@_) }
2073
2074 sub change_file_prop {
2075 my ($self, $fbat, $pname, $pval) = @_;
2076 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2077 }
2078
2079 sub chg_file {
2080 my ($self, $fbat, $m) = @_;
2081 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2082 $self->change_file_prop($fbat,'svn:executable','*');
2083 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2084 $self->change_file_prop($fbat,'svn:executable',undef);
2085 }
2086 my $fh = IO::File->new_tmpfile or croak $!;
2087 if ($m->{mode_b} =~ /^120/) {
2088 print $fh 'link ' or croak $!;
2089 $self->change_file_prop($fbat,'svn:special','*');
2090 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2091 $self->change_file_prop($fbat,'svn:special',undef);
2092 }
2093 defined(my $pid = fork) or croak $!;
2094 if (!$pid) {
2095 open STDOUT, '>&', $fh or croak $!;
2096 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2097 }
2098 waitpid $pid, 0;
2099 croak $? if $?;
2100 $fh->flush == 0 or croak $!;
2101 seek $fh, 0, 0 or croak $!;
2102
2103 my $md5 = Digest::MD5->new;
2104 $md5->addfile($fh) or croak $!;
2105 seek $fh, 0, 0 or croak $!;
2106
2107 my $exp = $md5->hexdigest;
2108 my $pool = SVN::Pool->new;
2109 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2110 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2111 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2112 $pool->clear;
2113
2114 close $fh or croak $!;
2115 }
2116
2117 sub D {
2118 my ($self, $m) = @_;
2119 my ($dir, $file) = split_path($m->{file_b});
2120 my $pbat = $self->ensure_path($dir);
2121 print "\tD\t$m->{file_b}\n" unless $::_q;
2122 $self->delete_entry($m->{file_b}, $pbat);
2123 }
2124
2125 sub close_edit {
2126 my ($self) = @_;
2127 my ($p,$bat) = ($self->{pool}, $self->{bat});
2128 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2129 $self->close_directory($bat->{$_}, $p);
2130 }
2131 $self->SUPER::close_edit($p);
2132 $p->clear;
2133 }
2134
2135 sub abort_edit {
2136 my ($self) = @_;
2137 $self->SUPER::abort_edit($self->{pool});
2138 }
2139
2140 sub DESTROY {
2141 my $self = shift;
2142 $self->SUPER::DESTROY(@_);
2143 $self->{pool}->clear;
2144 }
2145
2146 # this drives the editor
2147 sub apply_diff {
2148 my ($self) = @_;
2149 my $mods = $self->{mods};
2150 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2151 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2152 my $f = $m->{chg};
2153 if (defined $o{$f}) {
2154 $self->$f($m);
2155 } else {
2156 fatal("Invalid change type: $f\n");
2157 }
2158 }
2159 $self->rmdirs if $::_rmdir;
2160 if (@$mods == 0) {
2161 $self->abort_edit;
2162 } else {
2163 $self->close_edit;
2164 }
2165 return scalar @$mods;
2166 }
2167
2168 package Git::SVN::Ra;
2169 use vars qw/@ISA $config_dir/;
2170 use strict;
2171 use warnings;
2172 my ($can_do_switch);
2173 my %RA;
2174
2175 BEGIN {
2176 # enforce temporary pool usage for some simple functions
2177 my $e;
2178 foreach (qw/get_latest_revnum rev_proplist get_file
2179 check_path get_dir get_uuid get_repos_root/) {
2180 $e .= "sub $_ {
2181 my \$self = shift;
2182 my \$pool = SVN::Pool->new;
2183 my \@ret = \$self->SUPER::$_(\@_,\$pool);
2184 \$pool->clear;
2185 wantarray ? \@ret : \$ret[0]; }\n";
2186 }
2187 eval $e;
2188 }
2189
2190 sub new {
2191 my ($class, $url) = @_;
2192 $url =~ s!/+$!!;
2193 return $RA{$url} if $RA{$url};
2194
2195 SVN::_Core::svn_config_ensure($config_dir, undef);
2196 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2197 SVN::Client::get_simple_provider(),
2198 SVN::Client::get_ssl_server_trust_file_provider(),
2199 SVN::Client::get_simple_prompt_provider(
2200 \&Git::SVN::Prompt::simple, 2),
2201 SVN::Client::get_ssl_client_cert_prompt_provider(
2202 \&Git::SVN::Prompt::ssl_client_cert, 2),
2203 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2204 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2205 SVN::Client::get_username_provider(),
2206 SVN::Client::get_ssl_server_trust_prompt_provider(
2207 \&Git::SVN::Prompt::ssl_server_trust),
2208 SVN::Client::get_username_prompt_provider(
2209 \&Git::SVN::Prompt::username, 2),
2210 ]);
2211 my $config = SVN::Core::config_get_config($config_dir);
2212 my $self = SVN::Ra->new(url => $url, auth => $baton,
2213 config => $config,
2214 pool => SVN::Pool->new,
2215 auth_provider_callbacks => $callbacks);
2216 $self->{svn_path} = $url;
2217 $self->{repos_root} = $self->get_repos_root;
2218 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E/*##;
2219 $RA{$url} = bless $self, $class;
2220 }
2221
2222 sub DESTROY {
2223 # do not call the real DESTROY since we store ourselves in %RA
2224 }
2225
2226 sub get_log {
2227 my ($self, @args) = @_;
2228 my $pool = SVN::Pool->new;
2229 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2230 my $ret = $self->SUPER::get_log(@args, $pool);
2231 $pool->clear;
2232 $ret;
2233 }
2234
2235 sub get_commit_editor {
2236 my ($self, $log, $cb, $pool) = @_;
2237 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2238 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2239 }
2240
2241 sub uuid {
2242 my ($self) = @_;
2243 $self->{uuid} ||= $self->get_uuid;
2244 }
2245
2246 sub gs_do_update {
2247 my ($self, $rev_a, $rev_b, $path, $recurse, $editor) = @_;
2248 my $pool = SVN::Pool->new;
2249 $editor->set_path_strip($path);
2250 my $reporter = $self->do_update($rev_b, $path, $recurse,
2251 $editor, $pool);
2252 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2253 my $new = ($rev_a == $rev_b);
2254 $reporter->set_path('', $rev_a, $new, @lock, $pool);
2255 $reporter->finish_report($pool);
2256 $pool->clear;
2257 $editor->{git_commit_ok};
2258 }
2259
2260 sub gs_do_switch {
2261 my ($self, $rev_a, $rev_b, $path, $recurse, $url_b, $editor) = @_;
2262 my $pool = SVN::Pool->new;
2263 $editor->set_path_strip($path);
2264 my $reporter = $self->do_switch($rev_b, $path, $recurse,
2265 $url_b, $editor, $pool);
2266 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2267 $reporter->set_path('', $rev_a, 0, @lock, $pool);
2268 $reporter->finish_report($pool);
2269 $pool->clear;
2270 $editor->{git_commit_ok};
2271 }
2272
2273 sub gs_fetch_loop_common {
2274 my ($self, $base, $head, @gs) = @_;
2275 my $inc = 1000;
2276 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2277 my $err_handler = $SVN::Error::handler;
2278 my $err;
2279 $SVN::Error::handler = sub { ($err) = @_; skip_unknown_revs($err); };
2280 my @paths = @gs == 1 ? ($gs[0]->{path}) : ('');
2281 foreach my $gs (@gs) {
2282 if (my $last_commit = $gs->last_commit) {
2283 $gs->assert_index_clean($last_commit);
2284 }
2285 $gs->{path_regex} = qr/^\/\Q$gs->{path}\E\/?/;
2286 }
2287 while (1) {
2288 my @revs;
2289 $self->get_log(\@paths, $min, $max, 0, 1, 1,
2290 sub { push @revs, [ dup_changed_paths($_[0]), $_[1] ]; });
2291 if (! @revs && $err && $max >= $head) {
2292 print STDERR "Branch probably deleted:\n ",
2293 $err->expanded_message,
2294 "\nWill attempt to follow revisions ",
2295 "r$min .. r$max ",
2296 "committed before the deletion\n";
2297 @revs = map { [ undef, $_ ] } ($min .. $max);
2298 }
2299 foreach (@revs) {
2300 my ($paths, $r) = @$_;
2301 foreach my $gs (@gs) {
2302 if ($paths) {
2303 grep /$gs->{path_regex}/, keys %$paths
2304 or next;
2305 }
2306 next if defined $gs->rev_db_get($r);
2307 if (my $log_entry = $gs->do_fetch($paths, $r)) {
2308 $gs->do_git_commit($log_entry);
2309 }
2310 }
2311 }
2312 last if $max >= $head;
2313 $min = $max + 1;
2314 $max += $inc;
2315 $max = $head if ($max > $head);
2316 }
2317 $SVN::Error::handler = $err_handler;
2318 }
2319
2320 sub minimize_url {
2321 my ($self) = @_;
2322 return $self->{url} if ($self->{url} eq $self->{repos_root});
2323 my $url = $self->{repos_root};
2324 my @components = split(m!/!, $self->{svn_path});
2325 my $c = '';
2326 do {
2327 $url .= "/$c" if length $c;
2328 eval { (ref $self)->new($url)->get_latest_revnum };
2329 } while ($@ && ($c = shift @components));
2330 $url;
2331 }
2332
2333 sub can_do_switch {
2334 my $self = shift;
2335 unless (defined $can_do_switch) {
2336 my $pool = SVN::Pool->new;
2337 my $rep = eval {
2338 $self->do_switch(1, '', 0, $self->{url},
2339 SVN::Delta::Editor->new, $pool);
2340 };
2341 if ($@) {
2342 $can_do_switch = 0;
2343 } else {
2344 $rep->abort_report($pool);
2345 $can_do_switch = 1;
2346 }
2347 $pool->clear;
2348 }
2349 $can_do_switch;
2350 }
2351
2352 sub skip_unknown_revs {
2353 my ($err) = @_;
2354 my $errno = $err->apr_err();
2355 # Maybe the branch we're tracking didn't
2356 # exist when the repo started, so it's
2357 # not an error if it doesn't, just continue
2358 #
2359 # Wonderfully consistent library, eh?
2360 # 160013 - svn:// and file://
2361 # 175002 - http(s)://
2362 # 175007 - http(s):// (this repo required authorization, too...)
2363 # More codes may be discovered later...
2364 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2365 return;
2366 }
2367 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2368 }
2369
2370 # svn_log_changed_path_t objects passed to get_log are likely to be
2371 # overwritten even if only the refs are copied to an external variable,
2372 # so we should dup the structures in their entirety. Using an externally
2373 # passed pool (instead of our temporary and quickly cleared pool in
2374 # Git::SVN::Ra) does not help matters at all...
2375 sub dup_changed_paths {
2376 my ($paths) = @_;
2377 return undef unless $paths;
2378 my %ret;
2379 foreach my $p (keys %$paths) {
2380 my $i = $paths->{$p};
2381 my %s = map { $_ => $i->$_ }
2382 qw/copyfrom_path copyfrom_rev action/;
2383 $ret{$p} = \%s;
2384 }
2385 \%ret;
2386 }
2387
2388 package Git::SVN::Log;
2389 use strict;
2390 use warnings;
2391 use POSIX qw/strftime/;
2392 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2393 %rusers $show_commit $incremental/;
2394 my $l_fmt;
2395
2396 sub cmt_showable {
2397 my ($c) = @_;
2398 return 1 if defined $c->{r};
2399 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2400 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2401 my @log = command(qw/cat-file commit/, $c->{c});
2402 shift @log while ($log[0] ne "\n");
2403 shift @log;
2404 @{$c->{l}} = grep !/^git-svn-id: /, @log;
2405
2406 (undef, $c->{r}, undef) = ::extract_metadata(
2407 (grep(/^git-svn-id: /, @log))[-1]);
2408 }
2409 return defined $c->{r};
2410 }
2411
2412 sub log_use_color {
2413 return 1 if $color;
2414 my ($dc, $dcvar);
2415 $dcvar = 'color.diff';
2416 $dc = `git-config --get $dcvar`;
2417 if ($dc eq '') {
2418 # nothing at all; fallback to "diff.color"
2419 $dcvar = 'diff.color';
2420 $dc = `git-config --get $dcvar`;
2421 }
2422 chomp($dc);
2423 if ($dc eq 'auto') {
2424 my $pc;
2425 $pc = `git-config --get color.pager`;
2426 if ($pc eq '') {
2427 # does not have it -- fallback to pager.color
2428 $pc = `git-config --bool --get pager.color`;
2429 }
2430 else {
2431 $pc = `git-config --bool --get color.pager`;
2432 if ($?) {
2433 $pc = 'false';
2434 }
2435 }
2436 chomp($pc);
2437 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2438 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2439 }
2440 return 0;
2441 }
2442 return 0 if $dc eq 'never';
2443 return 1 if $dc eq 'always';
2444 chomp($dc = `git-config --bool --get $dcvar`);
2445 return ($dc eq 'true');
2446 }
2447
2448 sub git_svn_log_cmd {
2449 my ($r_min, $r_max) = @_;
2450 my $gs = Git::SVN->_new;
2451 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2452 $gs->refname);
2453 push @cmd, '-r' unless $non_recursive;
2454 push @cmd, qw/--raw --name-status/ if $verbose;
2455 push @cmd, '--color' if log_use_color();
2456 return @cmd unless defined $r_max;
2457 if ($r_max == $r_min) {
2458 push @cmd, '--max-count=1';
2459 if (my $c = $gs->rev_db_get($r_max)) {
2460 push @cmd, $c;
2461 }
2462 } else {
2463 my ($c_min, $c_max);
2464 $c_max = $gs->rev_db_get($r_max);
2465 $c_min = $gs->rev_db_get($r_min);
2466 if (defined $c_min && defined $c_max) {
2467 if ($r_max > $r_max) {
2468 push @cmd, "$c_min..$c_max";
2469 } else {
2470 push @cmd, "$c_max..$c_min";
2471 }
2472 } elsif ($r_max > $r_min) {
2473 push @cmd, $c_max;
2474 } else {
2475 push @cmd, $c_min;
2476 }
2477 }
2478 return @cmd;
2479 }
2480
2481 # adapted from pager.c
2482 sub config_pager {
2483 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2484 if (!defined $pager) {
2485 $pager = 'less';
2486 } elsif (length $pager == 0 || $pager eq 'cat') {
2487 $pager = undef;
2488 }
2489 }
2490
2491 sub run_pager {
2492 return unless -t *STDOUT;
2493 pipe my $rfd, my $wfd or return;
2494 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2495 if (!$pid) {
2496 open STDOUT, '>&', $wfd or
2497 ::fatal "Can't redirect to stdout: $!\n";
2498 return;
2499 }
2500 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2501 $ENV{LESS} ||= 'FRSX';
2502 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2503 }
2504
2505 sub tz_to_s_offset {
2506 my ($tz) = @_;
2507 $tz =~ s/(\d\d)$//;
2508 return ($1 * 60) + ($tz * 3600);
2509 }
2510
2511 sub get_author_info {
2512 my ($dest, $author, $t, $tz) = @_;
2513 $author =~ s/(?:^\s*|\s*$)//g;
2514 $dest->{a_raw} = $author;
2515 my $au;
2516 if ($::_authors) {
2517 $au = $rusers{$author} || undef;
2518 }
2519 if (!$au) {
2520 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2521 }
2522 $dest->{t} = $t;
2523 $dest->{tz} = $tz;
2524 $dest->{a} = $au;
2525 # Date::Parse isn't in the standard Perl distro :(
2526 if ($tz =~ s/^\+//) {
2527 $t += tz_to_s_offset($tz);
2528 } elsif ($tz =~ s/^\-//) {
2529 $t -= tz_to_s_offset($tz);
2530 }
2531 $dest->{t_utc} = $t;
2532 }
2533
2534 sub process_commit {
2535 my ($c, $r_min, $r_max, $defer) = @_;
2536 if (defined $r_min && defined $r_max) {
2537 if ($r_min == $c->{r} && $r_min == $r_max) {
2538 show_commit($c);
2539 return 0;
2540 }
2541 return 1 if $r_min == $r_max;
2542 if ($r_min < $r_max) {
2543 # we need to reverse the print order
2544 return 0 if (defined $limit && --$limit < 0);
2545 push @$defer, $c;
2546 return 1;
2547 }
2548 if ($r_min != $r_max) {
2549 return 1 if ($r_min < $c->{r});
2550 return 1 if ($r_max > $c->{r});
2551 }
2552 }
2553 return 0 if (defined $limit && --$limit < 0);
2554 show_commit($c);
2555 return 1;
2556 }
2557
2558 sub show_commit {
2559 my $c = shift;
2560 if ($oneline) {
2561 my $x = "\n";
2562 if (my $l = $c->{l}) {
2563 while ($l->[0] =~ /^\s*$/) { shift @$l }
2564 $x = $l->[0];
2565 }
2566 $l_fmt ||= 'A' . length($c->{r});
2567 print 'r',pack($l_fmt, $c->{r}),' | ';
2568 print "$c->{c} | " if $show_commit;
2569 print $x;
2570 } else {
2571 show_commit_normal($c);
2572 }
2573 }
2574
2575 sub show_commit_changed_paths {
2576 my ($c) = @_;
2577 return unless $c->{changed};
2578 print "Changed paths:\n", @{$c->{changed}};
2579 }
2580
2581 sub show_commit_normal {
2582 my ($c) = @_;
2583 print '-' x72, "\nr$c->{r} | ";
2584 print "$c->{c} | " if $show_commit;
2585 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2586 localtime($c->{t_utc})), ' | ';
2587 my $nr_line = 0;
2588
2589 if (my $l = $c->{l}) {
2590 while ($l->[$#$l] eq "\n" && $#$l > 0
2591 && $l->[($#$l - 1)] eq "\n") {
2592 pop @$l;
2593 }
2594 $nr_line = scalar @$l;
2595 if (!$nr_line) {
2596 print "1 line\n\n\n";
2597 } else {
2598 if ($nr_line == 1) {
2599 $nr_line = '1 line';
2600 } else {
2601 $nr_line .= ' lines';
2602 }
2603 print $nr_line, "\n";
2604 show_commit_changed_paths($c);
2605 print "\n";
2606 print $_ foreach @$l;
2607 }
2608 } else {
2609 print "1 line\n";
2610 show_commit_changed_paths($c);
2611 print "\n";
2612
2613 }
2614 foreach my $x (qw/raw diff/) {
2615 if ($c->{$x}) {
2616 print "\n";
2617 print $_ foreach @{$c->{$x}}
2618 }
2619 }
2620 }
2621
2622 sub cmd_show_log {
2623 my (@args) = @_;
2624 my ($r_min, $r_max);
2625 my $r_last = -1; # prevent dupes
2626 if (defined $TZ) {
2627 $ENV{TZ} = $TZ;
2628 } else {
2629 delete $ENV{TZ};
2630 }
2631 if (defined $::_revision) {
2632 if ($::_revision =~ /^(\d+):(\d+)$/) {
2633 ($r_min, $r_max) = ($1, $2);
2634 } elsif ($::_revision =~ /^\d+$/) {
2635 $r_min = $r_max = $::_revision;
2636 } else {
2637 ::fatal "-r$::_revision is not supported, use ",
2638 "standard \'git log\' arguments instead\n";
2639 }
2640 }
2641
2642 config_pager();
2643 @args = (git_svn_log_cmd($r_min, $r_max), @args);
2644 my $log = command_output_pipe(@args);
2645 run_pager();
2646 my (@k, $c, $d);
2647 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
2648 while (<$log>) {
2649 if (/^${esc_color}commit ($::sha1_short)/o) {
2650 my $cmt = $1;
2651 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
2652 $r_last = $c->{r};
2653 process_commit($c, $r_min, $r_max, \@k) or
2654 goto out;
2655 }
2656 $d = undef;
2657 $c = { c => $cmt };
2658 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
2659 get_author_info($c, $1, $2, $3);
2660 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
2661 # ignore
2662 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
2663 push @{$c->{raw}}, $_;
2664 } elsif (/^${esc_color}[ACRMDT]\t/) {
2665 # we could add $SVN->{svn_path} here, but that requires
2666 # remote access at the moment (repo_path_split)...
2667 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
2668 push @{$c->{changed}}, $_;
2669 } elsif (/^${esc_color}diff /o) {
2670 $d = 1;
2671 push @{$c->{diff}}, $_;
2672 } elsif ($d) {
2673 push @{$c->{diff}}, $_;
2674 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
2675 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
2676 } elsif (s/^${esc_color} //o) {
2677 push @{$c->{l}}, $_;
2678 }
2679 }
2680 if ($c && defined $c->{r} && $c->{r} != $r_last) {
2681 $r_last = $c->{r};
2682 process_commit($c, $r_min, $r_max, \@k);
2683 }
2684 if (@k) {
2685 my $swap = $r_max;
2686 $r_max = $r_min;
2687 $r_min = $swap;
2688 process_commit($_, $r_min, $r_max) foreach reverse @k;
2689 }
2690 out:
2691 close $log;
2692 print '-' x72,"\n" unless $incremental || $oneline;
2693 }
2694
2695 package Git::SVN::Migration;
2696 # these version numbers do NOT correspond to actual version numbers
2697 # of git nor git-svn. They are just relative.
2698 #
2699 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
2700 #
2701 # v1 layout: .git/$id/info/url, refs/remotes/$id
2702 #
2703 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
2704 #
2705 # v3 layout: .git/svn/$id, refs/remotes/$id
2706 # - info/url may remain for backwards compatibility
2707 # - this is what we migrate up to this layout automatically,
2708 # - this will be used by git svn init on single branches
2709 #
2710 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
2711 # - this is only created for newly multi-init-ed
2712 # repositories. Similar in spirit to the
2713 # --use-separate-remotes option in git-clone (now default)
2714 # - we do not automatically migrate to this (following
2715 # the example set by core git)
2716 use strict;
2717 use warnings;
2718 use Carp qw/croak/;
2719 use File::Path qw/mkpath/;
2720 use File::Basename qw/dirname basename/;
2721 use vars qw/$_minimize/;
2722
2723 sub migrate_from_v0 {
2724 my $git_dir = $ENV{GIT_DIR};
2725 return undef unless -d $git_dir;
2726 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2727 my $migrated = 0;
2728 while (<$fh>) {
2729 chomp;
2730 my ($id, $orig_ref) = ($_, $_);
2731 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
2732 next unless -f "$git_dir/$id/info/url";
2733 my $new_ref = "refs/remotes/$id";
2734 if (::verify_ref("$new_ref^0")) {
2735 print STDERR "W: $orig_ref is probably an old ",
2736 "branch used by an ancient version of ",
2737 "git-svn.\n",
2738 "However, $new_ref also exists.\n",
2739 "We will not be able ",
2740 "to use this branch until this ",
2741 "ambiguity is resolved.\n";
2742 next;
2743 }
2744 print STDERR "Migrating from v0 layout...\n" if !$migrated;
2745 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
2746 command_noisy('update-ref', $new_ref, $orig_ref);
2747 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
2748 $migrated++;
2749 }
2750 command_close_pipe($fh, $ctx);
2751 print STDERR "Done migrating from v0 layout...\n" if $migrated;
2752 $migrated;
2753 }
2754
2755 sub migrate_from_v1 {
2756 my $git_dir = $ENV{GIT_DIR};
2757 my $migrated = 0;
2758 return $migrated unless -d $git_dir;
2759 my $svn_dir = "$git_dir/svn";
2760
2761 # just in case somebody used 'svn' as their $id at some point...
2762 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
2763
2764 print STDERR "Migrating from a git-svn v1 layout...\n";
2765 mkpath([$svn_dir]);
2766 print STDERR "Data from a previous version of git-svn exists, but\n\t",
2767 "$svn_dir\n\t(required for this version ",
2768 "($::VERSION) of git-svn) does not. exist\n";
2769 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2770 while (<$fh>) {
2771 my $x = $_;
2772 next unless $x =~ s#^refs/remotes/##;
2773 chomp $x;
2774 next unless -f "$git_dir/$x/info/url";
2775 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
2776 next unless $u;
2777 my $dn = dirname("$git_dir/svn/$x");
2778 mkpath([$dn]) unless -d $dn;
2779 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
2780 mkpath(["$git_dir/svn/svn"]);
2781 print STDERR " - $git_dir/$x/info => ",
2782 "$git_dir/svn/$x/info\n";
2783 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
2784 croak "$!: $x";
2785 # don't worry too much about these, they probably
2786 # don't exist with repos this old (save for index,
2787 # and we can easily regenerate that)
2788 foreach my $f (qw/unhandled.log index .rev_db/) {
2789 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
2790 }
2791 } else {
2792 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
2793 rename "$git_dir/$x", "$git_dir/svn/$x" or
2794 croak "$!: $x";
2795 }
2796 $migrated++;
2797 }
2798 command_close_pipe($fh, $ctx);
2799 print STDERR "Done migrating from a git-svn v1 layout\n";
2800 $migrated;
2801 }
2802
2803 sub read_old_urls {
2804 my ($l_map, $pfx, $path) = @_;
2805 my @dir;
2806 foreach (<$path/*>) {
2807 if (-r "$_/info/url") {
2808 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2809 my $ref_id = $pfx . basename $_;
2810 my $url = ::file_to_s("$_/info/url");
2811 $l_map->{$ref_id} = $url;
2812 } elsif (-d $_) {
2813 push @dir, $_;
2814 }
2815 }
2816 foreach (@dir) {
2817 my $x = $_;
2818 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
2819 read_old_urls($l_map, $x, $_);
2820 }
2821 }
2822
2823 sub migrate_from_v2 {
2824 my @cfg = command(qw/config -l/);
2825 return if grep /^svn-remote\..+\.url=/, @cfg;
2826 my %l_map;
2827 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
2828 my $migrated = 0;
2829
2830 foreach my $ref_id (sort keys %l_map) {
2831 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
2832 $migrated++;
2833 }
2834 $migrated;
2835 }
2836
2837 sub minimize_connections {
2838 my $r = Git::SVN::read_all_remotes();
2839 my $new_urls = {};
2840 my $root_repos = {};
2841 foreach my $repo_id (keys %$r) {
2842 my $url = $r->{$repo_id}->{url} or next;
2843 my $fetch = $r->{$repo_id}->{fetch} or next;
2844 my $ra = Git::SVN::Ra->new($url);
2845
2846 # skip existing cases where we already connect to the root
2847 if (($ra->{url} eq $ra->{repos_root}) ||
2848 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
2849 $repo_id)) {
2850 $root_repos->{$ra->{url}} = $repo_id;
2851 next;
2852 }
2853
2854 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
2855 my $root_path = $ra->{url};
2856 $root_path =~ s#^\Q$ra->{repos_root}\E/*##;
2857 foreach my $path (keys %$fetch) {
2858 my $ref_id = $fetch->{$path};
2859 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
2860
2861 # make sure we can read when connecting to
2862 # a higher level of a repository
2863 my ($last_rev, undef) = $gs->last_rev_commit;
2864 if (!defined $last_rev) {
2865 $last_rev = eval {
2866 $root_ra->get_latest_revnum;
2867 };
2868 next if $@;
2869 }
2870 my $new = $root_path;
2871 $new .= length $path ? "/$path" : '';
2872 eval {
2873 $root_ra->get_log([$new], $last_rev, $last_rev,
2874 0, 0, 1, sub { });
2875 };
2876 next if $@;
2877 $new_urls->{$ra->{repos_root}}->{$new} =
2878 { ref_id => $ref_id,
2879 old_repo_id => $repo_id,
2880 old_path => $path };
2881 }
2882 }
2883
2884 my @emptied;
2885 foreach my $url (keys %$new_urls) {
2886 # see if we can re-use an existing [svn-remote "repo_id"]
2887 # instead of creating a(n ugly) new section:
2888 my $repo_id = $root_repos->{$url} ||
2889 Git::SVN::sanitize_remote_name($url);
2890
2891 my $fetch = $new_urls->{$url};
2892 foreach my $path (keys %$fetch) {
2893 my $x = $fetch->{$path};
2894 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
2895 my $pfx = "svn-remote.$x->{old_repo_id}";
2896
2897 my $old_fetch = quotemeta("$x->{old_path}:".
2898 "refs/remotes/$x->{ref_id}");
2899 command_noisy(qw/config --unset/,
2900 "$pfx.fetch", '^'. $old_fetch . '$');
2901 delete $r->{$x->{old_repo_id}}->
2902 {fetch}->{$x->{old_path}};
2903 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
2904 command_noisy(qw/config --unset/,
2905 "$pfx.url");
2906 push @emptied, $x->{old_repo_id}
2907 }
2908 }
2909 }
2910 if (@emptied) {
2911 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
2912 "$ENV{GIT_DIR}/config";
2913 print STDERR <<EOF;
2914 The following [svn-remote] sections in your config file ($file) are empty
2915 and can be safely removed:
2916 EOF
2917 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
2918 }
2919 }
2920
2921 sub migration_check {
2922 migrate_from_v0();
2923 migrate_from_v1();
2924 migrate_from_v2();
2925 minimize_connections() if $_minimize;
2926 }
2927
2928 package Git::IndexInfo;
2929 use strict;
2930 use warnings;
2931 use Git qw/command_input_pipe command_close_pipe/;
2932
2933 sub new {
2934 my ($class) = @_;
2935 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
2936 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
2937 }
2938
2939 sub remove {
2940 my ($self, $path) = @_;
2941 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
2942 return ++$self->{nr};
2943 }
2944 undef;
2945 }
2946
2947 sub update {
2948 my ($self, $mode, $hash, $path) = @_;
2949 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
2950 return ++$self->{nr};
2951 }
2952 undef;
2953 }
2954
2955 sub DESTROY {
2956 my ($self) = @_;
2957 command_close_pipe($self->{gui}, $self->{ctx});
2958 }
2959
2960 __END__
2961
2962 Data structures:
2963
2964 $log_entry hashref as returned by libsvn_log_entry()
2965 {
2966 log => 'whitespace-formatted log entry
2967 ', # trailing newline is preserved
2968 revision => '8', # integer
2969 date => '2004-02-24T17:01:44.108345Z', # commit date
2970 author => 'committer name'
2971 };
2972
2973
2974 # this is generated by generate_diff();
2975 @mods = array of diff-index line hashes, each element represents one line
2976 of diff-index output
2977
2978 diff-index line ($m hash)
2979 {
2980 mode_a => first column of diff-index output, no leading ':',
2981 mode_b => second column of diff-index output,
2982 sha1_b => sha1sum of the final blob,
2983 chg => change type [MCRADT],
2984 file_a => original file name of a file (iff chg is 'C' or 'R')
2985 file_b => new/current file name of a file (any chg)
2986 }
2987 ;
2988
2989 # retval of read_url_paths{,_all}();
2990 $l_map = {
2991 # repository root url
2992 'https://svn.musicpd.org' => {
2993 # repository path # GIT_SVN_ID
2994 'mpd/trunk' => 'trunk',
2995 'mpd/tags/0.11.5' => 'tags/0.11.5',
2996 },
2997 }
2998
2999 Notes:
3000 I don't trust the each() function on unless I created %hash myself
3001 because the internal iterator may not have started at base.
Dscho's miscellaneous git branches