?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-svn: hopefully make 'fetch' more user-friendly
[git/dscho.git] / git-svn.perl
blob3eed62fc0bcacd74827b1e7c901d352f6b843986
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 $sha1 = qr/[a-f\d]{40}/;
52 $sha1_short = qr/[a-f\d]{4,40}/;
53 my ($_stdin, $_help, $_edit,
54 $_message, $_file,
55 $_template, $_shared,
56 $_version, $_fetch_all,
57 $_merge, $_strategy, $_dry_run,
58 $_prefix);
59 $Git::SVN::_follow_parent = 1;
60 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
61 'config-dir=s' => \$Git::SVN::Ra::config_dir,
62 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
63 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
64 'authors-file|A=s' => \$_authors,
65 'repack:i' => \$Git::SVN::_repack,
66 'noMetadata' => \$Git::SVN::_no_metadata,
67 'useSvmProps' => \$Git::SVN::_use_svm_props,
68 'quiet|q' => \$_q,
69 'repack-flags|repack-args|repack-opts=s' =>
70 \$Git::SVN::_repack_flags,
71 %remote_opts );
72
73 my ($_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,
88 'all|a' => \$_fetch_all,
89 %fc_opts } ],
90 init => [ \&cmd_init, "Initialize a repo for tracking" .
91 " (requires URL argument)",
92 \%init_opts ],
93 dcommit => [ \&cmd_dcommit,
94 'Commit several diffs to merge with upstream',
95 { 'merge|m|M' => \$_merge,
96 'strategy|s=s' => \$_strategy,
97 'dry-run|n' => \$_dry_run,
98 %cmt_opts, %fc_opts } ],
99 'set-tree' => [ \&cmd_set_tree,
100 "Set an SVN repository to a git tree-ish",
101 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
102 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
103 { 'revision|r=i' => \$_revision } ],
104 'multi-init' => [ \&cmd_multi_init,
105 'Initialize multiple trees (like git-svnimport)',
106 { %multi_opts, %init_opts, %remote_opts,
107 'revision|r=i' => \$_revision,
108 'prefix=s' => \$_prefix,
109 } ],
110 'multi-fetch' => [ \&cmd_multi_fetch,
111 "Deprecated alias for $0 fetch --all",
112 { 'revision|r=s' => \$_revision, %fc_opts } ],
113 'migrate' => [ sub { },
114 # no-op, we automatically run this anyways,
115 'Migrate configuration/metadata/layout from
116 previous versions of git-svn',
117 \%remote_opts ],
118 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
119 { 'limit=i' => \$Git::SVN::Log::limit,
120 'revision|r=s' => \$_revision,
121 'verbose|v' => \$Git::SVN::Log::verbose,
122 'incremental' => \$Git::SVN::Log::incremental,
123 'oneline' => \$Git::SVN::Log::oneline,
124 'show-commit' => \$Git::SVN::Log::show_commit,
125 'non-recursive' => \$Git::SVN::Log::non_recursive,
126 'authors-file|A=s' => \$_authors,
127 'color' => \$Git::SVN::Log::color,
128 'pager=s' => \$Git::SVN::Log::pager,
129 } ],
130 'commit-diff' => [ \&cmd_commit_diff,
131 'Commit a diff between two trees',
132 { 'message|m=s' => \$_message,
133 'file|F=s' => \$_file,
134 'revision|r=s' => \$_revision,
135 %cmt_opts } ],
136 );
137
138 my $cmd;
139 for (my $i = 0; $i < @ARGV; $i++) {
140 if (defined $cmd{$ARGV[$i]}) {
141 $cmd = $ARGV[$i];
142 splice @ARGV, $i, 1;
143 last;
144 }
145 };
146
147 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
148
149 read_repo_config(\%opts);
150 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
151 'minimize-connections' => \$Git::SVN::Migration::_minimize,
152 'id|i=s' => \$Git::SVN::default_ref_id,
153 'svn-remote|remote|R=s' => \$Git::SVN::default_repo_id);
154 exit 1 if (!$rv && $cmd ne 'log');
155
156 usage(0) if $_help;
157 version() if $_version;
158 usage(1) unless defined $cmd;
159 load_authors() if $_authors;
160 unless ($cmd =~ /^(?:init|multi-init|commit-diff)$/) {
161 Git::SVN::Migration::migration_check();
162 }
163 Git::SVN::init_vars();
164 eval {
165 Git::SVN::verify_remotes_sanity();
166 $cmd{$cmd}->[0]->(@ARGV);
167 };
168 fatal $@ if $@;
169 exit 0;
170
171 ####################### primary functions ######################
172 sub usage {
173 my $exit = shift || 0;
174 my $fd = $exit ? \*STDERR : \*STDOUT;
175 print $fd <<"";
176 git-svn - bidirectional operations between a single Subversion tree and git
177 Usage: $0 <command> [options] [arguments]\n
178
179 print $fd "Available commands:\n" unless $cmd;
180
181 foreach (sort keys %cmd) {
182 next if $cmd && $cmd ne $_;
183 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
184 foreach (keys %{$cmd{$_}->[2]}) {
185 # prints out arguments as they should be passed:
186 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
187 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
188 "--$_" : "-$_" }
189 split /\|/,$_)," $x\n";
190 }
191 }
192 print $fd <<"";
193 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
194 arbitrary identifier if you're tracking multiple SVN branches/repositories in
195 one git repository and want to keep them separate. See git-svn(1) for more
196 information.
197
198 exit $exit;
199 }
200
201 sub version {
202 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
203 exit 0;
204 }
205
206 sub do_git_init_db {
207 unless (-d $ENV{GIT_DIR}) {
208 my @init_db = ('init');
209 push @init_db, "--template=$_template" if defined $_template;
210 push @init_db, "--shared" if defined $_shared;
211 command_noisy(@init_db);
212 }
213 }
214
215 sub cmd_init {
216 my $url = shift or die "SVN repository location required " .
217 "as a command-line argument\n";
218 if (my $repo_path = shift) {
219 unless (-d $repo_path) {
220 mkpath([$repo_path]);
221 }
222 chdir $repo_path or croak $!;
223 $ENV{GIT_DIR} = $repo_path . "/.git";
224 }
225 do_git_init_db();
226
227 Git::SVN->init($url);
228 }
229
230 sub cmd_fetch {
231 if (grep /^\d+=./, @_) {
232 die "'<rev>=<commit>' fetch arguments are ",
233 "no longer supported.\n";
234 }
235 my ($remote) = @_;
236 if (@_ > 1) {
237 die "Usage: $0 fetch [--all|-a] [svn-remote]\n";
238 }
239 $remote ||= $Git::SVN::default_repo_id;
240 if ($_fetch_all) {
241 cmd_multi_fetch();
242 } else {
243 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
244 }
245 }
246
247 sub cmd_set_tree {
248 my (@commits) = @_;
249 if ($_stdin || !@commits) {
250 print "Reading from stdin...\n";
251 @commits = ();
252 while (<STDIN>) {
253 if (/\b($sha1_short)\b/o) {
254 unshift @commits, $1;
255 }
256 }
257 }
258 my @revs;
259 foreach my $c (@commits) {
260 my @tmp = command('rev-parse',$c);
261 if (scalar @tmp == 1) {
262 push @revs, $tmp[0];
263 } elsif (scalar @tmp > 1) {
264 push @revs, reverse(command('rev-list',@tmp));
265 } else {
266 fatal "Failed to rev-parse $c\n";
267 }
268 }
269 my $gs = Git::SVN->new;
270 my ($r_last, $cmt_last) = $gs->last_rev_commit;
271 $gs->fetch;
272 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
273 fatal "There are new revisions that were fetched ",
274 "and need to be merged (or acknowledged) ",
275 "before committing.\nlast rev: $r_last\n",
276 " current: $gs->{last_rev}\n";
277 }
278 $gs->set_tree($_) foreach @revs;
279 print "Done committing ",scalar @revs," revisions to SVN\n";
280 }
281
282 sub cmd_dcommit {
283 my $head = shift;
284 $head ||= 'HEAD';
285 my ($url, $rev, $uuid);
286 my ($fh, $ctx) = command_output_pipe('rev-list', $head);
287 my @refs;
288 my $c;
289 while (<$fh>) {
290 $c = $_;
291 chomp $c;
292 ($url, $rev, $uuid) = cmt_metadata($c);
293 last if (defined $url && defined $rev && defined $uuid);
294 unshift @refs, $c;
295 }
296 close $fh; # most likely breaking the pipe
297 unless (defined $url && defined $rev && defined $uuid) {
298 die "Unable to determine upstream SVN information from ",
299 "$head history:\n $ctx\n";
300 }
301 my $gs = Git::SVN->find_by_url($url) or
302 die "Can't determine fetch information for $url\n";
303 my $last_rev;
304 foreach my $d (@refs) {
305 if (!verify_ref("$d~1")) {
306 fatal "Commit $d\n",
307 "has no parent commit, and therefore ",
308 "nothing to diff against.\n",
309 "You should be working from a repository ",
310 "originally created by git-svn\n";
311 }
312 unless (defined $last_rev) {
313 (undef, $last_rev, undef) = cmt_metadata("$d~1");
314 unless (defined $last_rev) {
315 fatal "Unable to extract revision information ",
316 "from commit $d~1\n";
317 }
318 }
319 if ($_dry_run) {
320 print "diff-tree $d~1 $d\n";
321 } else {
322 my %ed_opts = ( r => $last_rev,
323 log => get_commit_entry($d)->{log},
324 ra => Git::SVN::Ra->new($url),
325 tree_a => "$d~1",
326 tree_b => $d,
327 editor_cb => sub {
328 print "Committed r$_[0]\n";
329 $last_rev = $_[0]; },
330 svn_path => '');
331 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
332 print "No changes\n$d~1 == $d\n";
333 }
334 }
335 }
336 return if $_dry_run;
337 $gs->fetch;
338 # we always want to rebase against the current HEAD, not any
339 # head that was passed to us
340 my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
341 my @finish;
342 if (@diff) {
343 @finish = qw/rebase/;
344 push @finish, qw/--merge/ if $_merge;
345 push @finish, "--strategy=$_strategy" if $_strategy;
346 print STDERR "W: HEAD and ", $gs->refname, " differ, ",
347 "using @finish:\n", "@diff";
348 } else {
349 print "No changes between current HEAD and ",
350 $gs->refname, "\nResetting to the latest ",
351 $gs->refname, "\n";
352 @finish = qw/reset --mixed/;
353 }
354 command_noisy(@finish, $gs->refname);
355 }
356
357 sub cmd_show_ignore {
358 my $gs = Git::SVN->new;
359 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
360 $gs->traverse_ignore(\*STDOUT, '', $r);
361 }
362
363 sub cmd_multi_init {
364 my $url = shift;
365 unless (defined $_trunk || defined $_branches || defined $_tags) {
366 usage(1);
367 }
368 do_git_init_db();
369 $_prefix = '' unless defined $_prefix;
370 $url =~ s#/+$## if defined $url;
371 if (defined $_trunk) {
372 my $trunk_ref = $_prefix . 'trunk';
373 # try both old-style and new-style lookups:
374 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
375 unless ($gs_trunk) {
376 my ($trunk_url, $trunk_path) =
377 complete_svn_url($url, $_trunk);
378 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
379 undef, $trunk_ref);
380 }
381 }
382 return unless defined $_branches || defined $_tags;
383 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
384 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
385 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
386 }
387
388 sub cmd_multi_fetch {
389 my $remotes = Git::SVN::read_all_remotes();
390 foreach my $repo_id (sort keys %$remotes) {
391 if ($remotes->{$repo_id}->{url}) {
392 Git::SVN::fetch_all($repo_id, $remotes);
393 }
394 }
395 }
396
397 # this command is special because it requires no metadata
398 sub cmd_commit_diff {
399 my ($ta, $tb, $url) = @_;
400 my $usage = "Usage: $0 commit-diff -r<revision> ".
401 "<tree-ish> <tree-ish> [<URL>]\n";
402 fatal($usage) if (!defined $ta || !defined $tb);
403 my $svn_path;
404 if (!defined $url) {
405 my $gs = eval { Git::SVN->new };
406 if (!$gs) {
407 fatal("Needed URL or usable git-svn --id in ",
408 "the command-line\n", $usage);
409 }
410 $url = $gs->{url};
411 $svn_path = $gs->{path};
412 }
413 unless (defined $_revision) {
414 fatal("-r|--revision is a required argument\n", $usage);
415 }
416 if (defined $_message && defined $_file) {
417 fatal("Both --message/-m and --file/-F specified ",
418 "for the commit message.\n",
419 "I have no idea what you mean\n");
420 }
421 if (defined $_file) {
422 $_message = file_to_s($_file);
423 } else {
424 $_message ||= get_commit_entry($tb)->{log};
425 }
426 my $ra ||= Git::SVN::Ra->new($url);
427 $svn_path ||= $ra->{svn_path};
428 my $r = $_revision;
429 if ($r eq 'HEAD') {
430 $r = $ra->get_latest_revnum;
431 } elsif ($r !~ /^\d+$/) {
432 die "revision argument: $r not understood by git-svn\n";
433 }
434 my %ed_opts = ( r => $r,
435 log => $_message,
436 ra => $ra,
437 tree_a => $ta,
438 tree_b => $tb,
439 editor_cb => sub { print "Committed r$_[0]\n" },
440 svn_path => $svn_path );
441 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
442 print "No changes\n$ta == $tb\n";
443 }
444 }
445
446 ########################### utility functions #########################
447
448 sub complete_svn_url {
449 my ($url, $path) = @_;
450 $path =~ s#/+$##;
451 if ($path !~ m#^[a-z\+]+://#) {
452 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
453 fatal("E: '$path' is not a complete URL ",
454 "and a separate URL is not specified\n");
455 }
456 return ($url, $path);
457 }
458 return ($path, '');
459 }
460
461 sub complete_url_ls_init {
462 my ($ra, $repo_path, $switch, $pfx) = @_;
463 unless ($repo_path) {
464 print STDERR "W: $switch not specified\n";
465 return;
466 }
467 $repo_path =~ s#/+$##;
468 if ($repo_path =~ m#^[a-z\+]+://#) {
469 $ra = Git::SVN::Ra->new($repo_path);
470 $repo_path = '';
471 } else {
472 $repo_path =~ s#^/+##;
473 unless ($ra) {
474 fatal("E: '$repo_path' is not a complete URL ",
475 "and a separate URL is not specified\n");
476 }
477 }
478 my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
479 my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
480 my $url = $ra->{url};
481 my $remote_id;
482 my $remote_path;
483 foreach my $d (sort keys %$dirent) {
484 next if ($dirent->{$d}->kind != $SVN::Node::dir);
485 my $path = "$repo_path/$d";
486 my $ref = "$pfx$d";
487 my $gs = eval { Git::SVN->new($ref) };
488 # don't try to init already existing refs
489 unless ($gs) {
490 print "init $url/$path => $ref\n";
491 $gs = Git::SVN->init($url, $path, undef, $ref, 1);
492 }
493 if ($gs) {
494 my $k = "svn-remote.$gs->{repo_id}.url";
495 my $orig_url = eval {
496 command_oneline(qw/config --get/, $k)
497 };
498 if ($orig_url && ($orig_url ne $gs->{url})) {
499 die "$k already set: $orig_url\n",
500 "wanted to set to: $gs->{url}\n";
501 }
502 unless ($orig_url) {
503 command_oneline('config', $k, $gs->{url});
504 }
505 $remote_id = $gs->{repo_id};
506 last;
507 }
508 }
509 if (defined $remote_id) {
510 $remote_path = "$ra->{svn_path}/$repo_path/*";
511 $remote_path =~ s#/+#/#g;
512 $remote_path =~ s#^/##g;
513 my ($n) = ($switch =~ /^--(\w+)/);
514 if (length $pfx && $pfx !~ m#/$#) {
515 die "--prefix='$pfx' must have a trailing slash '/'\n";
516 }
517 command_noisy('config', "svn-remote.$remote_id.$n",
518 "$remote_path:refs/remotes/$pfx*");
519 }
520 }
521
522 sub verify_ref {
523 my ($ref) = @_;
524 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
525 { STDERR => 0 }); };
526 }
527
528 sub get_tree_from_treeish {
529 my ($treeish) = @_;
530 # $treeish can be a symbolic ref, too:
531 my $type = command_oneline(qw/cat-file -t/, $treeish);
532 my $expected;
533 while ($type eq 'tag') {
534 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
535 }
536 if ($type eq 'commit') {
537 $expected = (grep /^tree /, command(qw/cat-file commit/,
538 $treeish))[0];
539 ($expected) = ($expected =~ /^tree ($sha1)$/o);
540 die "Unable to get tree from $treeish\n" unless $expected;
541 } elsif ($type eq 'tree') {
542 $expected = $treeish;
543 } else {
544 die "$treeish is a $type, expected tree, tag or commit\n";
545 }
546 return $expected;
547 }
548
549 sub get_commit_entry {
550 my ($treeish) = shift;
551 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
552 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
553 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
554 open my $log_fh, '>', $commit_editmsg or croak $!;
555
556 my $type = command_oneline(qw/cat-file -t/, $treeish);
557 if ($type eq 'commit' || $type eq 'tag') {
558 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
559 $type, $treeish);
560 my $in_msg = 0;
561 while (<$msg_fh>) {
562 if (!$in_msg) {
563 $in_msg = 1 if (/^\s*$/);
564 } elsif (/^git-svn-id: /) {
565 # skip this for now, we regenerate the
566 # correct one on re-fetch anyways
567 # TODO: set *:merge properties or like...
568 } else {
569 print $log_fh $_ or croak $!;
570 }
571 }
572 command_close_pipe($msg_fh, $ctx);
573 }
574 close $log_fh or croak $!;
575
576 if ($_edit || ($type eq 'tree')) {
577 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
578 # TODO: strip out spaces, comments, like git-commit.sh
579 system($editor, $commit_editmsg);
580 }
581 rename $commit_editmsg, $commit_msg or croak $!;
582 open $log_fh, '<', $commit_msg or croak $!;
583 { local $/; chomp($log_entry{log} = <$log_fh>); }
584 close $log_fh or croak $!;
585 unlink $commit_msg;
586 \%log_entry;
587 }
588
589 sub s_to_file {
590 my ($str, $file, $mode) = @_;
591 open my $fd,'>',$file or croak $!;
592 print $fd $str,"\n" or croak $!;
593 close $fd or croak $!;
594 chmod ($mode &~ umask, $file) if (defined $mode);
595 }
596
597 sub file_to_s {
598 my $file = shift;
599 open my $fd,'<',$file or croak "$!: file: $file\n";
600 local $/;
601 my $ret = <$fd>;
602 close $fd or croak $!;
603 $ret =~ s/\s*$//s;
604 return $ret;
605 }
606
607 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
608 sub load_authors {
609 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
610 my $log = $cmd eq 'log';
611 while (<$authors>) {
612 chomp;
613 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
614 my ($user, $name, $email) = ($1, $2, $3);
615 if ($log) {
616 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
617 } else {
618 $users{$user} = [$name, $email];
619 }
620 }
621 close $authors or croak $!;
622 }
623
624 # convert GetOpt::Long specs for use by git-config
625 sub read_repo_config {
626 return unless -d $ENV{GIT_DIR};
627 my $opts = shift;
628 my @config_only;
629 foreach my $o (keys %$opts) {
630 # if we have mixedCase and a long option-only, then
631 # it's a config-only variable that we don't need for
632 # the command-line.
633 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
634 my $v = $opts->{$o};
635 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
636 $key =~ s/-//g;
637 my $arg = 'git-config';
638 $arg .= ' --int' if ($o =~ /[:=]i$/);
639 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
640 if (ref $v eq 'ARRAY') {
641 chomp(my @tmp = `$arg --get-all svn.$key`);
642 @$v = @tmp if @tmp;
643 } else {
644 chomp(my $tmp = `$arg --get svn.$key`);
645 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
646 $$v = $tmp;
647 }
648 }
649 }
650 delete @$opts{@config_only} if @config_only;
651 }
652
653 sub extract_metadata {
654 my $id = shift or return (undef, undef, undef);
655 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
656 \s([a-f\d\-]+)$/x);
657 if (!defined $rev || !$uuid || !$url) {
658 # some of the original repositories I made had
659 # identifiers like this:
660 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
661 }
662 return ($url, $rev, $uuid);
663 }
664
665 sub cmt_metadata {
666 return extract_metadata((grep(/^git-svn-id: /,
667 command(qw/cat-file commit/, shift)))[-1]);
668 }
669
670 package Git::SVN;
671 use strict;
672 use warnings;
673 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
674 $_repack $_repack_flags $_use_svm_props/;
675 use Carp qw/croak/;
676 use File::Path qw/mkpath/;
677 use File::Copy qw/copy/;
678 use IPC::Open3;
679
680 my $_repack_nr;
681 # properties that we do not log:
682 my %SKIP_PROP;
683 BEGIN {
684 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
685 svn:special svn:executable
686 svn:entry:committed-rev
687 svn:entry:last-author
688 svn:entry:uuid
689 svn:entry:committed-date/;
690
691 # some options are read globally, but can be overridden locally
692 # per [svn-remote "..."] section. Command-line options will *NOT*
693 # override options set in an [svn-remote "..."] section
694 my $e;
695 foreach (qw/follow_parent no_metadata use_svm_props/) {
696 my $key = $_;
697 $key =~ tr/_//d;
698 $e .= "sub $_ {
699 my (\$self) = \@_;
700 return \$self->{-$_} if exists \$self->{-$_};
701 my \$k = \"svn-remote.\$self->{repo_id}\.$key\";
702 eval { command_oneline(qw/config --get/, \$k) };
703 if (\$@) {
704 \$self->{-$_} = \$Git::SVN::_$_;
705 } else {
706 my \$v = command_oneline(qw/config --bool/,\$k);
707 \$self->{-$_} = \$v eq 'false' ? 0 : 1;
708 }
709 return \$self->{-$_} }\n";
710 }
711 $e .= "1;\n";
712 eval $e or die $@;
713 }
714
715 my %LOCKFILES;
716 END { unlink keys %LOCKFILES if %LOCKFILES }
717
718 sub resolve_local_globs {
719 my ($url, $fetch, $glob_spec) = @_;
720 return unless defined $glob_spec;
721 my $ref = $glob_spec->{ref};
722 my $path = $glob_spec->{path};
723 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
724 next unless m#^refs/remotes/$ref->{regex}$#;
725 my $p = $1;
726 my $pathname = $path->full_path($p);
727 my $refname = $ref->full_path($p);
728 if (my $existing = $fetch->{$pathname}) {
729 if ($existing ne $refname) {
730 die "Refspec conflict:\n",
731 "existing: refs/remotes/$existing\n",
732 " globbed: refs/remotes/$refname\n";
733 }
734 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
735 $u =~ s!^\Q$url\E(/|$)!! or die
736 "refs/remotes/$refname: '$url' not found in '$u'\n";
737 if ($pathname ne $u) {
738 warn "W: Refspec glob conflict ",
739 "(ref: refs/remotes/$refname):\n",
740 "expected path: $pathname\n",
741 " real path: $u\n",
742 "Continuing ahead with $u\n";
743 next;
744 }
745 } else {
746 $fetch->{$pathname} = $refname;
747 }
748 }
749 }
750
751 sub parse_revision_argument {
752 my ($base, $head) = @_;
753 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
754 return ($base, $head);
755 }
756 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
757 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
758 return ($head, $head) if ($::_revision eq 'HEAD');
759 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
760 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
761 die "revision argument: $::_revision not understood by git-svn\n";
762 }
763
764 sub fetch_all {
765 my ($repo_id, $remotes) = @_;
766 my $remote = $remotes->{$repo_id};
767 my $fetch = $remote->{fetch};
768 my $url = $remote->{url};
769 my (@gs, @globs);
770 my $ra = Git::SVN::Ra->new($url);
771 my $uuid = $ra->get_uuid;
772 my $head = $ra->get_latest_revnum;
773 my $base = $head;
774
775 # read the max revs for wildcard expansion (branches/*, tags/*)
776 foreach my $t (qw/branches tags/) {
777 defined $remote->{$t} or next;
778 push @globs, $remote->{$t};
779 my $max_rev = eval { tmp_config(qw/--int --get/,
780 "svn-remote.$repo_id.${t}-maxRev") };
781 if (defined $max_rev && ($max_rev < $base)) {
782 $base = $max_rev;
783 }
784 }
785
786 if ($fetch) {
787 foreach my $p (sort keys %$fetch) {
788 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
789 my $lr = $gs->rev_db_max;
790 if (defined $lr) {
791 $base = $lr if ($lr < $base);
792 }
793 push @gs, $gs;
794 }
795 }
796
797 ($base, $head) = parse_revision_argument($base, $head);
798 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
799 }
800
801 sub read_all_remotes {
802 my $r = {};
803 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
804 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
805 $r->{$1}->{fetch}->{$2} = $3;
806 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
807 $r->{$1}->{url} = $2;
808 } elsif (m!^(.+)\.(branches|tags)=
809 (.*):refs/remotes/(.+)\s*$/!x) {
810 my ($p, $g) = ($3, $4);
811 my $rs = $r->{$1}->{$2} = {
812 t => $2,
813 remote => $1,
814 path => Git::SVN::GlobSpec->new($p),
815 ref => Git::SVN::GlobSpec->new($g) };
816 if (length($rs->{ref}->{right}) != 0) {
817 die "The '*' glob character must be the last ",
818 "character of '$g'\n";
819 }
820 }
821 }
822 $r;
823 }
824
825 sub init_vars {
826 if (defined $_repack) {
827 $_repack = 1000 if ($_repack <= 0);
828 $_repack_nr = $_repack;
829 $_repack_flags ||= '-d';
830 }
831 }
832
833 sub verify_remotes_sanity {
834 return unless -d $ENV{GIT_DIR};
835 my %seen;
836 foreach (command(qw/config -l/)) {
837 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
838 if ($seen{$1}) {
839 die "Remote ref refs/remote/$1 is tracked by",
840 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
841 "Please resolve this ambiguity in ",
842 "your git configuration file before ",
843 "continuing\n";
844 }
845 $seen{$1} = $_;
846 }
847 }
848 }
849
850 # we allow more chars than remotes2config.sh...
851 sub sanitize_remote_name {
852 my ($name) = @_;
853 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
854 $name;
855 }
856
857 sub find_existing_remote {
858 my ($url, $remotes) = @_;
859 my $existing;
860 foreach my $repo_id (keys %$remotes) {
861 my $u = $remotes->{$repo_id}->{url} or next;
862 next if $u ne $url;
863 $existing = $repo_id;
864 last;
865 }
866 $existing;
867 }
868
869 sub init_remote_config {
870 my ($self, $url, $no_write) = @_;
871 $url =~ s!/+$!!; # strip trailing slash
872 my $r = read_all_remotes();
873 my $existing = find_existing_remote($url, $r);
874 if ($existing) {
875 unless ($no_write) {
876 print STDERR "Using existing ",
877 "[svn-remote \"$existing\"]\n";
878 }
879 $self->{repo_id} = $existing;
880 } else {
881 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
882 $existing = find_existing_remote($min_url, $r);
883 if ($existing) {
884 unless ($no_write) {
885 print STDERR "Using existing ",
886 "[svn-remote \"$existing\"]\n";
887 }
888 $self->{repo_id} = $existing;
889 }
890 if ($min_url ne $url) {
891 unless ($no_write) {
892 print STDERR "Using higher level of URL: ",
893 "$url => $min_url\n";
894 }
895 my $old_path = $self->{path};
896 $self->{path} = $url;
897 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
898 if (length $old_path) {
899 $self->{path} .= "/$old_path";
900 }
901 $url = $min_url;
902 }
903 }
904 my $orig_url;
905 if (!$existing) {
906 # verify that we aren't overwriting anything:
907 $orig_url = eval {
908 command_oneline('config', '--get',
909 "svn-remote.$self->{repo_id}.url")
910 };
911 if ($orig_url && ($orig_url ne $url)) {
912 die "svn-remote.$self->{repo_id}.url already set: ",
913 "$orig_url\nwanted to set to: $url\n";
914 }
915 }
916 my ($xrepo_id, $xpath) = find_ref($self->refname);
917 if (defined $xpath) {
918 die "svn-remote.$xrepo_id.fetch already set to track ",
919 "$xpath:refs/remotes/", $self->refname, "\n";
920 }
921 unless ($no_write) {
922 command_noisy('config',
923 "svn-remote.$self->{repo_id}.url", $url);
924 command_noisy('config', '--add',
925 "svn-remote.$self->{repo_id}.fetch",
926 "$self->{path}:".$self->refname);
927 }
928 $self->{url} = $url;
929 }
930
931 sub find_by_url { # repos_root and, path are optional
932 my ($class, $full_url, $repos_root, $path) = @_;
933 my $remotes = read_all_remotes();
934 if (defined $full_url && defined $repos_root && !defined $path) {
935 $path = $full_url;
936 $path =~ s#^\Q$repos_root\E(?:/|$)##;
937 }
938 foreach my $repo_id (keys %$remotes) {
939 my $u = $remotes->{$repo_id}->{url} or next;
940 next if defined $repos_root && $repos_root ne $u;
941
942 my $fetch = $remotes->{$repo_id}->{fetch} || {};
943 foreach (qw/branches tags/) {
944 resolve_local_globs($u, $fetch,
945 $remotes->{$repo_id}->{$_});
946 }
947 my $p = $path;
948 unless (defined $p) {
949 $p = $full_url;
950 $p =~ s#^\Q$u\E(?:/|$)## or next;
951 }
952 foreach my $f (keys %$fetch) {
953 next if $f ne $p;
954 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
955 }
956 }
957 undef;
958 }
959
960 sub init {
961 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
962 my $self = _new($class, $repo_id, $ref_id, $path);
963 if (defined $url) {
964 $self->init_remote_config($url, $no_write);
965 }
966 $self;
967 }
968
969 sub find_ref {
970 my ($ref_id) = @_;
971 foreach (command(qw/config -l/)) {
972 next unless m!^svn-remote\.(.+)\.fetch=
973 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
974 my ($repo_id, $path, $ref) = ($1, $2, $3);
975 if ($ref eq $ref_id) {
976 $path = '' if ($path =~ m#^\./?#);
977 return ($repo_id, $path);
978 }
979 }
980 (undef, undef, undef);
981 }
982
983 sub new {
984 my ($class, $ref_id, $repo_id, $path) = @_;
985 if (defined $ref_id && !defined $repo_id && !defined $path) {
986 ($repo_id, $path) = find_ref($ref_id);
987 if (!defined $repo_id) {
988 die "Could not find a \"svn-remote.*.fetch\" key ",
989 "in the repository configuration matching: ",
990 "refs/remotes/$ref_id\n";
991 }
992 }
993 my $self = _new($class, $repo_id, $ref_id, $path);
994 if (!defined $self->{path} || !length $self->{path}) {
995 my $fetch = command_oneline('config', '--get',
996 "svn-remote.$repo_id.fetch",
997 ":refs/remotes/$ref_id\$") or
998 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
999 "\":refs/remotes/$ref_id\$\" in config\n";
1000 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1001 }
1002 $self->{url} = command_oneline('config', '--get',
1003 "svn-remote.$repo_id.url") or
1004 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1005 if ((-z $self->db_path || ! -e $self->db_path) &&
1006 ::verify_ref($self->refname.'^0')) {
1007 $self->rebuild;
1008 }
1009 $self;
1010 }
1011
1012 sub refname { "refs/remotes/$_[0]->{ref_id}" }
1013
1014 sub svm_uuid {
1015 my ($self) = @_;
1016 return $self->{svm}->{uuid} if $self->svm;
1017 $self->ra;
1018 unless ($self->{svm}) {
1019 die "SVM UUID not cached, and reading remotely failed\n";
1020 }
1021 $self->{svm}->{uuid};
1022 }
1023
1024 sub svm {
1025 my ($self) = @_;
1026 return $self->{svm} if $self->{svm};
1027 my $svm;
1028 # see if we have it in our config, first:
1029 eval {
1030 my $section = "svn-remote.$self->{repo_id}";
1031 $svm = {
1032 source => tmp_config('--get', "$section.svm-source"),
1033 uuid => tmp_config('--get', "$section.svm-uuid"),
1034 }
1035 };
1036 $self->{svm} = $svm if ($svm && $svm->{source} && $svm->{uuid});
1037 $self->{svm};
1038 }
1039
1040 sub _set_svm_vars {
1041 my ($self, $ra) = @_;
1042 return $ra if $self->svm;
1043
1044 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1045 "(svm:source, svm:mirror, svm:mirror) ",
1046 "from the following URLs:\n" );
1047 sub read_svm_props {
1048 my ($self, $props) = @_;
1049 my $src = $props->{'svm:source'};
1050 my $mirror = $props->{'svm:mirror'};
1051 my $uuid = $props->{'svm:uuid'};
1052 return undef if (!$src || !$mirror || !$uuid);
1053
1054 chomp($src, $mirror, $uuid);
1055
1056 $uuid =~ m{^[0-9a-f\-]{30,}$}
1057 or die "doesn't look right - svm:uuid is '$uuid'\n";
1058 # don't know what a '!' is there for, also the
1059 # username is of no interest
1060 $src =~ s{/?!$}{$mirror};
1061 $src =~ s{/+$}{}; # no trailing slashes please
1062 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1063
1064 my $section = "svn-remote.$self->{repo_id}";
1065 tmp_config('--add', "$section.svm-source", $src);
1066 tmp_config('--add', "$section.svm-uuid", $uuid);
1067 $self->{svm} = { source => $src , uuid => $uuid };
1068 return 1;
1069 }
1070
1071 my $r = $ra->get_latest_revnum;
1072 my $path = $self->{path};
1073 my @tried_a = ($path);
1074 while (length $path) {
1075 if ($self->read_svm_props(($ra->get_dir($path, $r))[2])) {
1076 return $ra;
1077 }
1078 $path =~ s#/?[^/]+$## && push @tried_a, $path;
1079 }
1080 if ($self->read_svm_props(($ra->get_dir('', $r))[2])) {
1081 return $ra;
1082 }
1083
1084 if ($ra->{repos_root} eq $self->{url}) {
1085 die @err, map { " $self->{url}/$_\n" } @tried_a, "\n";
1086 }
1087
1088 # nope, make sure we're connected to the repository root:
1089 my $ok;
1090 my @tried_b;
1091 $path = $ra->{svn_path};
1092 $path =~ s#/?[^/]+$##; # we already tried this one above
1093 $ra = Git::SVN::Ra->new($ra->{repos_root});
1094 while (length $path) {
1095 $ok = $self->read_svm_props(($ra->get_dir($path, $r))[2]);
1096 last if $ok;
1097 $path =~ s#/?[^/]+$## && push @tried_b, $path;
1098 }
1099 $ok = $self->read_svm_props(($ra->get_dir('', $r))[2]) unless $ok;
1100 if (!$ok) {
1101 die @err, map { " $self->{url}/$_\n" } @tried_a, "\n",
1102 map { " $ra->{url}/$_\n" } @tried_b, "\n"
1103 }
1104 Git::SVN::Ra->new($self->{url});
1105 }
1106
1107 # this allows us to memoize our SVN::Ra UUID locally and avoid a
1108 # remote lookup (useful for 'git svn log').
1109 sub ra_uuid {
1110 my ($self) = @_;
1111 unless ($self->{ra_uuid}) {
1112 my $key = "svn-remote.$self->{repo_id}.uuid";
1113 my $uuid = eval { tmp_config('--get', $key) };
1114 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1115 $self->{ra_uuid} = $uuid;
1116 } else {
1117 die "ra_uuid called without URL\n" unless $self->{url};
1118 $self->{ra_uuid} = $self->ra->get_uuid;
1119 tmp_config('--add', $key, $self->{ra_uuid});
1120 }
1121 }
1122 $self->{ra_uuid};
1123 }
1124
1125 sub ra {
1126 my ($self) = shift;
1127 my $ra = Git::SVN::Ra->new($self->{url});
1128 if ($self->use_svm_props && !$self->{svm}) {
1129 if ($self->no_metadata) {
1130 die "Can't have both 'noMetadata' and ",
1131 "'useSvmProps' options set!\n";
1132 }
1133 $ra = $self->_set_svm_vars($ra);
1134 $self->{-want_revprops} = 1;
1135 }
1136 $ra;
1137 }
1138
1139 sub rel_path {
1140 my ($self) = @_;
1141 my $repos_root = $self->ra->{repos_root};
1142 return $self->{path} if ($self->{url} eq $repos_root);
1143 die "BUG: rel_path failed! repos_root: $repos_root, Ra URL: ",
1144 $self->ra->{url}, " path: $self->{path}, URL: $self->{url}\n";
1145 }
1146
1147 sub traverse_ignore {
1148 my ($self, $fh, $path, $r) = @_;
1149 $path =~ s#^/+##g;
1150 my $ra = $self->ra;
1151 my ($dirent, undef, $props) = $ra->get_dir($path, $r);
1152 my $p = $path;
1153 $p =~ s#^\Q$ra->{svn_path}\E/##;
1154 print $fh length $p ? "\n# $p\n" : "\n# /\n";
1155 if (my $s = $props->{'svn:ignore'}) {
1156 $s =~ s/[\r\n]+/\n/g;
1157 chomp $s;
1158 if (length $p == 0) {
1159 $s =~ s#\n#\n/$p#g;
1160 print $fh "/$s\n";
1161 } else {
1162 $s =~ s#\n#\n/$p/#g;
1163 print $fh "/$p/$s\n";
1164 }
1165 }
1166 foreach (sort keys %$dirent) {
1167 next if $dirent->{$_}->kind != $SVN::Node::dir;
1168 $self->traverse_ignore($fh, "$path/$_", $r);
1169 }
1170 }
1171
1172 sub last_rev { ($_[0]->last_rev_commit)[0] }
1173 sub last_commit { ($_[0]->last_rev_commit)[1] }
1174
1175 # returns the newest SVN revision number and newest commit SHA1
1176 sub last_rev_commit {
1177 my ($self) = @_;
1178 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1179 return ($self->{last_rev}, $self->{last_commit});
1180 }
1181 my $c = ::verify_ref($self->refname.'^0');
1182 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1183 my $rev = (::cmt_metadata($c))[1];
1184 if (defined $rev) {
1185 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1186 return ($rev, $c);
1187 }
1188 }
1189 my $db_path = $self->db_path;
1190 unless (-e $db_path) {
1191 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1192 return (undef, undef);
1193 }
1194 my $offset = -41; # from tail
1195 my $rl;
1196 open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1197 sysseek($fh, $offset, 2); # don't care for errors
1198 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1199 chomp $rl;
1200 while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1201 $offset -= 41;
1202 sysseek($fh, $offset, 2); # don't care for errors
1203 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1204 chomp $rl;
1205 }
1206 if ($c && $c ne $rl) {
1207 die "$db_path and ", $self->refname,
1208 " inconsistent!:\n$c != $rl\n";
1209 }
1210 my $rev = sysseek($fh, 0, 1) or croak $!;
1211 $rev = ($rev - 41) / 41;
1212 close $fh or croak $!;
1213 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1214 return ($rev, $c);
1215 }
1216
1217 sub get_fetch_range {
1218 my ($self, $min, $max) = @_;
1219 $max ||= $self->ra->get_latest_revnum;
1220 $min ||= $self->rev_db_max;
1221 (++$min, $max);
1222 }
1223
1224 sub tmp_config {
1225 my (@args) = @_;
1226 my $config = "$ENV{GIT_DIR}/svn/config";
1227 unless (-f $config) {
1228 open my $fh, '>', $config or
1229 die "Can't open $config: $!\n";
1230 print $fh "; This file is used internally by git-svn\n" or
1231 die "Couldn't write to $config: $!\n";
1232 print $fh "; You should not have to edit it\n" or
1233 die "Couldn't write to $config: $!\n";
1234 close $fh or die "Couldn't close $config: $!\n";
1235 }
1236 my $old_config = $ENV{GIT_CONFIG};
1237 $ENV{GIT_CONFIG} = $config;
1238 $@ = undef;
1239 my @ret = eval { command('config', @args) };
1240 my $err = $@;
1241 if (defined $old_config) {
1242 $ENV{GIT_CONFIG} = $old_config;
1243 } else {
1244 delete $ENV{GIT_CONFIG};
1245 }
1246 die $err if $err;
1247 wantarray ? @ret : $ret[0];
1248 }
1249
1250 sub tmp_index_do {
1251 my ($self, $sub) = @_;
1252 my $old_index = $ENV{GIT_INDEX_FILE};
1253 $ENV{GIT_INDEX_FILE} = $self->{index};
1254 $@ = undef;
1255 my @ret = eval { &$sub };
1256 my $err = $@;
1257 if (defined $old_index) {
1258 $ENV{GIT_INDEX_FILE} = $old_index;
1259 } else {
1260 delete $ENV{GIT_INDEX_FILE};
1261 }
1262 die $err if $err;
1263 wantarray ? @ret : $ret[0];
1264 }
1265
1266 sub assert_index_clean {
1267 my ($self, $treeish) = @_;
1268
1269 $self->tmp_index_do(sub {
1270 command_noisy('read-tree', $treeish) unless -e $self->{index};
1271 my $x = command_oneline('write-tree');
1272 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1273 /^tree ($::sha1)/mo);
1274 if ($y ne $x) {
1275 unlink $self->{index} or croak $!;
1276 command_noisy('read-tree', $treeish);
1277 }
1278 $x = command_oneline('write-tree');
1279 if ($y ne $x) {
1280 ::fatal "trees ($treeish) $y != $x\n",
1281 "Something is seriously wrong...\n";
1282 }
1283 });
1284 }
1285
1286 sub get_commit_parents {
1287 my ($self, $log_entry) = @_;
1288 my (%seen, @ret, @tmp);
1289 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1290 if (my $ip = $self->{inject_parents}) {
1291 if (my $commit = delete $ip->{$log_entry->{revision}}) {
1292 push @tmp, $commit;
1293 }
1294 }
1295 if (my $cur = ::verify_ref($self->refname.'^0')) {
1296 push @tmp, $cur;
1297 }
1298 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1299 while (my $p = shift @tmp) {
1300 next if $seen{$p};
1301 $seen{$p} = 1;
1302 push @ret, $p;
1303 # MAXPARENT is defined to 16 in commit-tree.c:
1304 last if @ret >= 16;
1305 }
1306 if (@tmp) {
1307 die "r$log_entry->{revision}: No room for parents:\n\t",
1308 join("\n\t", @tmp), "\n";
1309 }
1310 @ret;
1311 }
1312
1313 sub full_url {
1314 my ($self) = @_;
1315 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1316 }
1317
1318 sub do_git_commit {
1319 my ($self, $log_entry) = @_;
1320 my $lr = $self->last_rev;
1321 if (defined $lr && $lr >= $log_entry->{revision}) {
1322 die "Last fetched revision of ", $self->refname,
1323 " was r$lr, but we are about to fetch: ",
1324 "r$log_entry->{revision}!\n";
1325 }
1326 if (my $c = $self->rev_db_get($log_entry->{revision})) {
1327 croak "$log_entry->{revision} = $c already exists! ",
1328 "Why are we refetching it?\n";
1329 }
1330 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1331 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1332 $log_entry->{email};
1333 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1334
1335 my $tree = $log_entry->{tree};
1336 if (!defined $tree) {
1337 $tree = $self->tmp_index_do(sub {
1338 command_oneline('write-tree') });
1339 }
1340 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1341
1342 my @exec = ('git-commit-tree', $tree);
1343 foreach ($self->get_commit_parents($log_entry)) {
1344 push @exec, '-p', $_;
1345 }
1346 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1347 or croak $!;
1348 print $msg_fh $log_entry->{log} or croak $!;
1349 unless ($self->no_metadata) {
1350 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1351 or croak $!;
1352 }
1353 $msg_fh->flush == 0 or croak $!;
1354 close $msg_fh or croak $!;
1355 chomp(my $commit = do { local $/; <$out_fh> });
1356 close $out_fh or croak $!;
1357 waitpid $pid, 0;
1358 croak $? if $?;
1359 if ($commit !~ /^$::sha1$/o) {
1360 die "Failed to commit, invalid sha1: $commit\n";
1361 }
1362
1363 $self->rev_db_set($log_entry->{revision}, $commit, 1);
1364
1365 $self->{last_rev} = $log_entry->{revision};
1366 $self->{last_commit} = $commit;
1367 print "r$log_entry->{revision}";
1368 if (defined $log_entry->{svm_revision}) {
1369 print " (\@$log_entry->{svm_revision})";
1370 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1371 0, $self->svm_uuid);
1372 }
1373 print " = $commit ($self->{ref_id})\n";
1374 if (defined $_repack && (--$_repack_nr == 0)) {
1375 $_repack_nr = $_repack;
1376 # repack doesn't use any arguments with spaces in them, does it?
1377 print "Running git repack $_repack_flags ...\n";
1378 command_noisy('repack', split(/\s+/, $_repack_flags));
1379 print "Done repacking\n";
1380 }
1381 return $commit;
1382 }
1383
1384 sub match_paths {
1385 my ($self, $paths, $r) = @_;
1386 return 1 if $self->{path} eq '';
1387 if (my $path = $paths->{"/$self->{path}"}) {
1388 return ($path->{action} eq 'D') ? 0 : 1;
1389 }
1390 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1391 if (grep /$self->{path_regex}/, keys %$paths) {
1392 return 1;
1393 }
1394 my $c = '';
1395 foreach (split m#/#, $self->{path}) {
1396 $c .= "/$_";
1397 next unless ($paths->{$c} &&
1398 ($paths->{$c}->{action} =~ /^[AR]$/));
1399 if ($self->ra->check_path($self->{path}, $r) ==
1400 $SVN::Node::dir) {
1401 return 1;
1402 }
1403 }
1404 return 0;
1405 }
1406
1407 sub find_parent_branch {
1408 my ($self, $paths, $rev) = @_;
1409 return undef unless $self->follow_parent;
1410 unless (defined $paths) {
1411 my $err_handler = $SVN::Error::handler;
1412 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1413 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1414 $paths =
1415 Git::SVN::Ra::dup_changed_paths($_[0]) });
1416 $SVN::Error::handler = $err_handler;
1417 }
1418 return undef unless defined $paths;
1419
1420 # look for a parent from another branch:
1421 my @b_path_components = split m#/#, $self->rel_path;
1422 my @a_path_components;
1423 my $i;
1424 while (@b_path_components) {
1425 $i = $paths->{'/'.join('/', @b_path_components)};
1426 last if $i && defined $i->{copyfrom_path};
1427 unshift(@a_path_components, pop(@b_path_components));
1428 }
1429 return undef unless defined $i && defined $i->{copyfrom_path};
1430 my $branch_from = $i->{copyfrom_path};
1431 if (@a_path_components) {
1432 print STDERR "branch_from: $branch_from => ";
1433 $branch_from .= '/'.join('/', @a_path_components);
1434 print STDERR $branch_from, "\n";
1435 }
1436 my $r = $i->{copyfrom_rev};
1437 my $repos_root = $self->ra->{repos_root};
1438 my $url = $self->ra->{url};
1439 my $new_url = $repos_root . $branch_from;
1440 print STDERR "Found possible branch point: ",
1441 "$new_url => ", $self->full_url, ", $r\n";
1442 $branch_from =~ s#^/##;
1443 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
1444 unless ($gs) {
1445 my $ref_id = $self->{ref_id};
1446 $ref_id =~ s/\@\d+$//;
1447 $ref_id .= "\@$r";
1448 # just grow a tail if we're not unique enough :x
1449 $ref_id .= '-' while find_ref($ref_id);
1450 print STDERR "Initializing parent: $ref_id\n";
1451 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1452 }
1453 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1454 if (!defined $r0 || !defined $parent) {
1455 $gs->fetch(0, $r);
1456 ($r0, $parent) = $gs->last_rev_commit;
1457 }
1458 if (defined $r0 && defined $parent) {
1459 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1460 $self->assert_index_clean($parent);
1461 my $ed;
1462 if ($self->ra->can_do_switch) {
1463 print STDERR "Following parent with do_switch\n";
1464 # do_switch works with svn/trunk >= r22312, but that
1465 # is not included with SVN 1.4.3 (the latest version
1466 # at the moment), so we can't rely on it
1467 $self->{last_commit} = $parent;
1468 $ed = SVN::Git::Fetcher->new($self);
1469 $gs->ra->gs_do_switch($r0, $rev, $gs,
1470 $self->full_url, $ed)
1471 or die "SVN connection failed somewhere...\n";
1472 } else {
1473 print STDERR "Following parent with do_update\n";
1474 $ed = SVN::Git::Fetcher->new($self);
1475 $self->ra->gs_do_update($rev, $rev, $self, $ed)
1476 or die "SVN connection failed somewhere...\n";
1477 }
1478 print STDERR "Successfully followed parent\n";
1479 return $self->make_log_entry($rev, [$parent], $ed);
1480 }
1481 return undef;
1482 }
1483
1484 sub do_fetch {
1485 my ($self, $paths, $rev) = @_;
1486 my $ed;
1487 my ($last_rev, @parents);
1488 if (my $lc = $self->last_commit) {
1489 # we can have a branch that was deleted, then re-added
1490 # under the same name but copied from another path, in
1491 # which case we'll have multiple parents (we don't
1492 # want to break the original ref, nor lose copypath info):
1493 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1494 push @{$log_entry->{parents}}, $lc;
1495 return $log_entry;
1496 }
1497 $ed = SVN::Git::Fetcher->new($self);
1498 $last_rev = $self->{last_rev};
1499 $ed->{c} = $lc;
1500 @parents = ($lc);
1501 } else {
1502 $last_rev = $rev;
1503 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1504 return $log_entry;
1505 }
1506 $ed = SVN::Git::Fetcher->new($self);
1507 }
1508 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1509 die "SVN connection failed somewhere...\n";
1510 }
1511 $self->make_log_entry($rev, \@parents, $ed);
1512 }
1513
1514 sub get_untracked {
1515 my ($self, $ed) = @_;
1516 my @out;
1517 my $h = $ed->{empty};
1518 foreach (sort keys %$h) {
1519 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1520 push @out, " $act: " . uri_encode($_);
1521 warn "W: $act: $_\n";
1522 }
1523 foreach my $t (qw/dir_prop file_prop/) {
1524 $h = $ed->{$t} or next;
1525 foreach my $path (sort keys %$h) {
1526 my $ppath = $path eq '' ? '.' : $path;
1527 foreach my $prop (sort keys %{$h->{$path}}) {
1528 next if $SKIP_PROP{$prop};
1529 my $v = $h->{$path}->{$prop};
1530 my $t_ppath_prop = "$t: " .
1531 uri_encode($ppath) . ' ' .
1532 uri_encode($prop);
1533 if (defined $v) {
1534 push @out, " +$t_ppath_prop " .
1535 uri_encode($v);
1536 } else {
1537 push @out, " -$t_ppath_prop";
1538 }
1539 }
1540 }
1541 }
1542 foreach my $t (qw/absent_file absent_directory/) {
1543 $h = $ed->{$t} or next;
1544 foreach my $parent (sort keys %$h) {
1545 foreach my $path (sort @{$h->{$parent}}) {
1546 push @out, " $t: " .
1547 uri_encode("$parent/$path");
1548 warn "W: $t: $parent/$path ",
1549 "Insufficient permissions?\n";
1550 }
1551 }
1552 }
1553 \@out;
1554 }
1555
1556 sub parse_svn_date {
1557 my $date = shift || return '+0000 1970-01-01 00:00:00';
1558 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1559 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1560 croak "Unable to parse date: $date\n";
1561 "+0000 $Y-$m-$d $H:$M:$S";
1562 }
1563
1564 sub check_author {
1565 my ($author) = @_;
1566 if (!defined $author || length $author == 0) {
1567 $author = '(no author)';
1568 }
1569 if (defined $::_authors && ! defined $::users{$author}) {
1570 die "Author: $author not defined in $::_authors file\n";
1571 }
1572 $author;
1573 }
1574
1575 sub make_log_entry {
1576 my ($self, $rev, $parents, $ed) = @_;
1577 my $untracked = $self->get_untracked($ed);
1578
1579 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1580 print $un "r$rev\n" or croak $!;
1581 print $un $_, "\n" foreach @$untracked;
1582 my %log_entry = ( parents => $parents || [], revision => $rev,
1583 log => '');
1584
1585 my $headrev;
1586 my $logged = delete $self->{logged_rev_props};
1587 if (!$logged || $self->{-want_revprops}) {
1588 my $rp = $self->ra->rev_proplist($rev);
1589 foreach (sort keys %$rp) {
1590 my $v = $rp->{$_};
1591 if (/^svn:(author|date|log)$/) {
1592 $log_entry{$1} = $v;
1593 } elsif ($_ eq 'svm:headrev') {
1594 $headrev = $v;
1595 } else {
1596 print $un " rev_prop: ", uri_encode($_), ' ',
1597 uri_encode($v), "\n";
1598 }
1599 }
1600 } else {
1601 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1602 }
1603 close $un or croak $!;
1604
1605 $log_entry{date} = parse_svn_date($log_entry{date});
1606 $log_entry{log} .= "\n";
1607 my $author = $log_entry{author} = check_author($log_entry{author});
1608 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1609 : ($author, undef);
1610 if (defined $headrev && $self->use_svm_props) {
1611 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1612 if ($uuid ne $self->{svm}->{uuid}) {
1613 die "UUID mismatch on SVM path:\n",
1614 "expected: $self->{svm}->{uuid}\n",
1615 " got: $uuid\n";
1616 }
1617 my $full_url = $self->{svm}->{source};
1618 $full_url .= "/$self->{path}" if length $self->{path};
1619 $log_entry{metadata} = "$full_url\@$r $uuid";
1620 $log_entry{svm_revision} = $r;
1621 $email ||= "$author\@$uuid"
1622 } else {
1623 $log_entry{metadata} = $self->full_url . "\@$rev " .
1624 $self->ra->get_uuid;
1625 $email ||= "$author\@" . $self->ra->get_uuid;
1626 }
1627 $log_entry{name} = $name;
1628 $log_entry{email} = $email;
1629 \%log_entry;
1630 }
1631
1632 sub fetch {
1633 my ($self, $min_rev, $max_rev, @parents) = @_;
1634 my ($last_rev, $last_commit) = $self->last_rev_commit;
1635 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1636 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1637 }
1638
1639 sub set_tree_cb {
1640 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1641 $self->{inject_parents} = { $rev => $tree };
1642 $self->fetch(undef, undef);
1643 }
1644
1645 sub set_tree {
1646 my ($self, $tree) = (shift, shift);
1647 my $log_entry = ::get_commit_entry($tree);
1648 unless ($self->{last_rev}) {
1649 fatal("Must have an existing revision to commit\n");
1650 }
1651 my %ed_opts = ( r => $self->{last_rev},
1652 log => $log_entry->{log},
1653 ra => $self->ra,
1654 tree_a => $self->{last_commit},
1655 tree_b => $tree,
1656 editor_cb => sub {
1657 $self->set_tree_cb($log_entry, $tree, @_) },
1658 svn_path => $self->{path} );
1659 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1660 print "No changes\nr$self->{last_rev} = $tree\n";
1661 }
1662 }
1663
1664 sub rebuild {
1665 my ($self) = @_;
1666 my $db_path = $self->db_path;
1667 if (-f $self->{db_root}) {
1668 rename $self->{db_root}, $db_path or die
1669 "rename $self->{db_root} => $db_path failed: $!\n";
1670 my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
1671 symlink $base, $self->{db_root} or die
1672 "symlink $base => $self->{db_root} failed: $!\n";
1673 return;
1674 }
1675 print "Rebuilding $db_path ...\n";
1676 my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1677 my $latest;
1678 my $full_url = $self->full_url;
1679 my $svn_uuid;
1680 while (<$rev_list>) {
1681 chomp;
1682 my $c = $_;
1683 die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1684 my ($url, $rev, $uuid) = ::cmt_metadata($c);
1685
1686 # ignore merges (from set-tree)
1687 next if (!defined $rev || !$uuid);
1688
1689 # if we merged or otherwise started elsewhere, this is
1690 # how we break out of it
1691 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1692 ($full_url && $url && ($url ne $full_url))) {
1693 next;
1694 }
1695 $latest ||= $rev;
1696 $svn_uuid ||= $uuid;
1697
1698 $self->rev_db_set($rev, $c);
1699 print "r$rev = $c\n";
1700 }
1701 command_close_pipe($rev_list, $ctx);
1702 print "Done rebuilding $db_path\n";
1703 }
1704
1705 # rev_db:
1706 # Tie::File seems to be prone to offset errors if revisions get sparse,
1707 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
1708 # one of my favorite modules is out :< Next up would be one of the DBM
1709 # modules, but I'm not sure which is most portable... So I'll just
1710 # go with something that's plain-text, but still capable of
1711 # being randomly accessed. So here's my ultra-simple fixed-width
1712 # database. All records are 40 characters + "\n", so it's easy to seek
1713 # to a revision: (41 * rev) is the byte offset.
1714 # A record of 40 0s denotes an empty revision.
1715 # And yes, it's still pretty fast (faster than Tie::File).
1716 # These files are disposable unless noMetadata or useSvmProps is set
1717
1718 sub _rev_db_set {
1719 my ($fh, $rev, $commit) = @_;
1720 my $offset = $rev * 41;
1721 # assume that append is the common case:
1722 seek $fh, 0, 2 or croak $!;
1723 my $pos = tell $fh;
1724 if ($pos < $offset) {
1725 for (1 .. (($offset - $pos) / 41)) {
1726 print $fh (('0' x 40),"\n") or croak $!;
1727 }
1728 }
1729 seek $fh, $offset, 0 or croak $!;
1730 print $fh $commit,"\n" or croak $!;
1731 }
1732
1733 sub mkfile {
1734 my ($path) = @_;
1735 unless (-e $path) {
1736 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
1737 mkpath([$dir]) unless -d $dir;
1738 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
1739 close $fh or die "Couldn't close (create) $path: $!\n";
1740 }
1741 }
1742
1743 sub rev_db_set {
1744 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
1745 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
1746 my $db = $self->db_path($uuid);
1747 my $db_lock = "$db.lock";
1748 my $sig;
1749 if ($update_ref) {
1750 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1751 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
1752 }
1753 mkfile($db);
1754
1755 $LOCKFILES{$db_lock} = 1;
1756 my $sync;
1757 # both of these options make our .rev_db file very, very important
1758 # and we can't afford to lose it because rebuild() won't work
1759 if ($self->use_svm_props || $self->no_metadata) {
1760 $sync = 1;
1761 copy($db, $db_lock) or die "rev_db_set(@_): ",
1762 "Failed to copy: ",
1763 "$db => $db_lock ($!)\n";
1764 } else {
1765 rename $db, $db_lock or die "rev_db_set(@_): ",
1766 "Failed to rename: ",
1767 "$db => $db_lock ($!)\n";
1768 }
1769 open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
1770 _rev_db_set($fh, $rev, $commit);
1771 if ($sync) {
1772 $fh->flush or die "Couldn't flush $db_lock: $!\n";
1773 $fh->sync or die "Couldn't sync $db_lock: $!\n";
1774 }
1775 close $fh or croak $!;
1776 if ($update_ref) {
1777 command_noisy('update-ref', '-m', "r$rev",
1778 $self->refname, $commit);
1779 }
1780 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
1781 "$db_lock => $db ($!)\n";
1782 delete $LOCKFILES{$db_lock};
1783 if ($update_ref) {
1784 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1785 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
1786 kill $sig, $$ if defined $sig;
1787 }
1788 }
1789
1790 sub rev_db_max {
1791 my ($self) = @_;
1792 my $db_path = $self->db_path;
1793 my @stat = stat $db_path or return 0;
1794 ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
1795 my $max = $stat[7] / 41;
1796 (($max > 0) ? $max - 1 : 0);
1797 }
1798
1799 sub rev_db_get {
1800 my ($self, $rev, $uuid) = @_;
1801 my $ret;
1802 my $offset = $rev * 41;
1803 my $db_path = $self->db_path($uuid);
1804 return undef unless -e $db_path;
1805 open my $fh, '<', $db_path or croak $!;
1806 if (sysseek($fh, $offset, 0) == $offset) {
1807 my $read = sysread($fh, $ret, 40);
1808 $ret = undef if ($read != 40 || $ret eq ('0'x40));
1809 }
1810 close $fh or croak $!;
1811 $ret;
1812 }
1813
1814 sub find_rev_before {
1815 my ($self, $rev, $eq_ok) = @_;
1816 --$rev unless $eq_ok;
1817 while ($rev > 0) {
1818 if (my $c = $self->rev_db_get($rev)) {
1819 return ($rev, $c);
1820 }
1821 --$rev;
1822 }
1823 return (undef, undef);
1824 }
1825
1826 sub _new {
1827 my ($class, $repo_id, $ref_id, $path) = @_;
1828 unless (defined $repo_id && length $repo_id) {
1829 $repo_id = $Git::SVN::default_repo_id;
1830 }
1831 unless (defined $ref_id && length $ref_id) {
1832 $_[2] = $ref_id = $Git::SVN::default_ref_id;
1833 }
1834 $_[1] = $repo_id = sanitize_remote_name($repo_id);
1835 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1836 $_[3] = $path = '' unless (defined $path);
1837 mkpath([$dir]);
1838 bless {
1839 ref_id => $ref_id, dir => $dir, index => "$dir/index",
1840 path => $path, config => "$ENV{GIT_DIR}/svn/config",
1841 db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
1842 }
1843
1844 sub db_path {
1845 my ($self, $uuid) = @_;
1846 $uuid ||= $self->ra_uuid;
1847 "$self->{db_root}.$uuid";
1848 }
1849
1850 sub uri_encode {
1851 my ($f) = @_;
1852 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1853 $f
1854 }
1855
1856 package Git::SVN::Prompt;
1857 use strict;
1858 use warnings;
1859 require SVN::Core;
1860 use vars qw/$_no_auth_cache $_username/;
1861
1862 sub simple {
1863 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1864 $may_save = undef if $_no_auth_cache;
1865 $default_username = $_username if defined $_username;
1866 if (defined $default_username && length $default_username) {
1867 if (defined $realm && length $realm) {
1868 print STDERR "Authentication realm: $realm\n";
1869 STDERR->flush;
1870 }
1871 $cred->username($default_username);
1872 } else {
1873 username($cred, $realm, $may_save, $pool);
1874 }
1875 $cred->password(_read_password("Password for '" .
1876 $cred->username . "': ", $realm));
1877 $cred->may_save($may_save);
1878 $SVN::_Core::SVN_NO_ERROR;
1879 }
1880
1881 sub ssl_server_trust {
1882 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1883 $may_save = undef if $_no_auth_cache;
1884 print STDERR "Error validating server certificate for '$realm':\n";
1885 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1886 print STDERR " - The certificate is not issued by a trusted ",
1887 "authority. Use the\n",
1888 " fingerprint to validate the certificate manually!\n";
1889 }
1890 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1891 print STDERR " - The certificate hostname does not match.\n";
1892 }
1893 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1894 print STDERR " - The certificate is not yet valid.\n";
1895 }
1896 if ($failures & $SVN::Auth::SSL::EXPIRED) {
1897 print STDERR " - The certificate has expired.\n";
1898 }
1899 if ($failures & $SVN::Auth::SSL::OTHER) {
1900 print STDERR " - The certificate has an unknown error.\n";
1901 }
1902 printf STDERR
1903 "Certificate information:\n".
1904 " - Hostname: %s\n".
1905 " - Valid: from %s until %s\n".
1906 " - Issuer: %s\n".
1907 " - Fingerprint: %s\n",
1908 map $cert_info->$_, qw(hostname valid_from valid_until
1909 issuer_dname fingerprint);
1910 my $choice;
1911 prompt:
1912 print STDERR $may_save ?
1913 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1914 "(R)eject or accept (t)emporarily? ";
1915 STDERR->flush;
1916 $choice = lc(substr(<STDIN> || 'R', 0, 1));
1917 if ($choice =~ /^t$/i) {
1918 $cred->may_save(undef);
1919 } elsif ($choice =~ /^r$/i) {
1920 return -1;
1921 } elsif ($may_save && $choice =~ /^p$/i) {
1922 $cred->may_save($may_save);
1923 } else {
1924 goto prompt;
1925 }
1926 $cred->accepted_failures($failures);
1927 $SVN::_Core::SVN_NO_ERROR;
1928 }
1929
1930 sub ssl_client_cert {
1931 my ($cred, $realm, $may_save, $pool) = @_;
1932 $may_save = undef if $_no_auth_cache;
1933 print STDERR "Client certificate filename: ";
1934 STDERR->flush;
1935 chomp(my $filename = <STDIN>);
1936 $cred->cert_file($filename);
1937 $cred->may_save($may_save);
1938 $SVN::_Core::SVN_NO_ERROR;
1939 }
1940
1941 sub ssl_client_cert_pw {
1942 my ($cred, $realm, $may_save, $pool) = @_;
1943 $may_save = undef if $_no_auth_cache;
1944 $cred->password(_read_password("Password: ", $realm));
1945 $cred->may_save($may_save);
1946 $SVN::_Core::SVN_NO_ERROR;
1947 }
1948
1949 sub username {
1950 my ($cred, $realm, $may_save, $pool) = @_;
1951 $may_save = undef if $_no_auth_cache;
1952 if (defined $realm && length $realm) {
1953 print STDERR "Authentication realm: $realm\n";
1954 }
1955 my $username;
1956 if (defined $_username) {
1957 $username = $_username;
1958 } else {
1959 print STDERR "Username: ";
1960 STDERR->flush;
1961 chomp($username = <STDIN>);
1962 }
1963 $cred->username($username);
1964 $cred->may_save($may_save);
1965 $SVN::_Core::SVN_NO_ERROR;
1966 }
1967
1968 sub _read_password {
1969 my ($prompt, $realm) = @_;
1970 print STDERR $prompt;
1971 STDERR->flush;
1972 require Term::ReadKey;
1973 Term::ReadKey::ReadMode('noecho');
1974 my $password = '';
1975 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1976 last if $key =~ /[\012\015]/; # \n\r
1977 $password .= $key;
1978 }
1979 Term::ReadKey::ReadMode('restore');
1980 print STDERR "\n";
1981 STDERR->flush;
1982 $password;
1983 }
1984
1985 package main;
1986
1987 {
1988 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1989 $SVN::Node::dir.$SVN::Node::unknown.
1990 $SVN::Node::none.$SVN::Node::file.
1991 $SVN::Node::dir.$SVN::Node::unknown.
1992 $SVN::Auth::SSL::CNMISMATCH.
1993 $SVN::Auth::SSL::NOTYETVALID.
1994 $SVN::Auth::SSL::EXPIRED.
1995 $SVN::Auth::SSL::UNKNOWNCA.
1996 $SVN::Auth::SSL::OTHER;
1997 }
1998
1999 package SVN::Git::Fetcher;
2000 use vars qw/@ISA/;
2001 use strict;
2002 use warnings;
2003 use Carp qw/croak/;
2004 use IO::File qw//;
2005 use Digest::MD5;
2006
2007 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
2008 sub new {
2009 my ($class, $git_svn) = @_;
2010 my $self = SVN::Delta::Editor->new;
2011 bless $self, $class;
2012 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2013 $self->{empty} = {};
2014 $self->{dir_prop} = {};
2015 $self->{file_prop} = {};
2016 $self->{absent_dir} = {};
2017 $self->{absent_file} = {};
2018 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2019 $self;
2020 }
2021
2022 sub set_path_strip {
2023 my ($self, $path) = @_;
2024 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2025 }
2026
2027 sub open_root {
2028 { path => '' };
2029 }
2030
2031 sub open_directory {
2032 my ($self, $path, $pb, $rev) = @_;
2033 { path => $path };
2034 }
2035
2036 sub git_path {
2037 my ($self, $path) = @_;
2038 if ($self->{path_strip}) {
2039 $path =~ s!$self->{path_strip}!! or
2040 die "Failed to strip path '$path' ($self->{path_strip})\n";
2041 }
2042 $path;
2043 }
2044
2045 sub delete_entry {
2046 my ($self, $path, $rev, $pb) = @_;
2047
2048 my $gpath = $self->git_path($path);
2049 return undef if ($gpath eq '');
2050
2051 # remove entire directories.
2052 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2053 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2054 -r --name-only -z/,
2055 $self->{c}, '--', $gpath);
2056 local $/ = "\0";
2057 while (<$ls>) {
2058 chomp;
2059 $self->{gii}->remove($_);
2060 print "\tD\t$_\n" unless $::_q;
2061 }
2062 print "\tD\t$gpath/\n" unless $::_q;
2063 command_close_pipe($ls, $ctx);
2064 $self->{empty}->{$path} = 0
2065 } else {
2066 $self->{gii}->remove($gpath);
2067 print "\tD\t$gpath\n" unless $::_q;
2068 }
2069 undef;
2070 }
2071
2072 sub open_file {
2073 my ($self, $path, $pb, $rev) = @_;
2074 my $gpath = $self->git_path($path);
2075 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2076 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2077 unless (defined $mode && defined $blob) {
2078 die "$path was not found in commit $self->{c} (r$rev)\n";
2079 }
2080 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2081 pool => SVN::Pool->new, action => 'M' };
2082 }
2083
2084 sub add_file {
2085 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2086 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2087 delete $self->{empty}->{$dir};
2088 { path => $path, mode_a => 100644, mode_b => 100644,
2089 pool => SVN::Pool->new, action => 'A' };
2090 }
2091
2092 sub add_directory {
2093 my ($self, $path, $cp_path, $cp_rev) = @_;
2094 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2095 delete $self->{empty}->{$dir};
2096 $self->{empty}->{$path} = 1;
2097 { path => $path };
2098 }
2099
2100 sub change_dir_prop {
2101 my ($self, $db, $prop, $value) = @_;
2102 $self->{dir_prop}->{$db->{path}} ||= {};
2103 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2104 undef;
2105 }
2106
2107 sub absent_directory {
2108 my ($self, $path, $pb) = @_;
2109 $self->{absent_dir}->{$pb->{path}} ||= [];
2110 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2111 undef;
2112 }
2113
2114 sub absent_file {
2115 my ($self, $path, $pb) = @_;
2116 $self->{absent_file}->{$pb->{path}} ||= [];
2117 push @{$self->{absent_file}->{$pb->{path}}}, $path;
2118 undef;
2119 }
2120
2121 sub change_file_prop {
2122 my ($self, $fb, $prop, $value) = @_;
2123 if ($prop eq 'svn:executable') {
2124 if ($fb->{mode_b} != 120000) {
2125 $fb->{mode_b} = defined $value ? 100755 : 100644;
2126 }
2127 } elsif ($prop eq 'svn:special') {
2128 $fb->{mode_b} = defined $value ? 120000 : 100644;
2129 } else {
2130 $self->{file_prop}->{$fb->{path}} ||= {};
2131 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2132 }
2133 undef;
2134 }
2135
2136 sub apply_textdelta {
2137 my ($self, $fb, $exp) = @_;
2138 my $fh = IO::File->new_tmpfile;
2139 $fh->autoflush(1);
2140 # $fh gets auto-closed() by SVN::TxDelta::apply(),
2141 # (but $base does not,) so dup() it for reading in close_file
2142 open my $dup, '<&', $fh or croak $!;
2143 my $base = IO::File->new_tmpfile;
2144 $base->autoflush(1);
2145 if ($fb->{blob}) {
2146 defined (my $pid = fork) or croak $!;
2147 if (!$pid) {
2148 open STDOUT, '>&', $base or croak $!;
2149 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2150 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2151 }
2152 waitpid $pid, 0;
2153 croak $? if $?;
2154
2155 if (defined $exp) {
2156 seek $base, 0, 0 or croak $!;
2157 my $md5 = Digest::MD5->new;
2158 $md5->addfile($base);
2159 my $got = $md5->hexdigest;
2160 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2161 "expected: $exp\n",
2162 " got: $got\n" if ($got ne $exp);
2163 }
2164 }
2165 seek $base, 0, 0 or croak $!;
2166 $fb->{fh} = $dup;
2167 $fb->{base} = $base;
2168 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2169 }
2170
2171 sub close_file {
2172 my ($self, $fb, $exp) = @_;
2173 my $hash;
2174 my $path = $self->git_path($fb->{path});
2175 if (my $fh = $fb->{fh}) {
2176 seek($fh, 0, 0) or croak $!;
2177 my $md5 = Digest::MD5->new;
2178 $md5->addfile($fh);
2179 my $got = $md5->hexdigest;
2180 die "Checksum mismatch: $path\n",
2181 "expected: $exp\n got: $got\n" if ($got ne $exp);
2182 seek($fh, 0, 0) or croak $!;
2183 if ($fb->{mode_b} == 120000) {
2184 read($fh, my $buf, 5) == 5 or croak $!;
2185 $buf eq 'link ' or die "$path has mode 120000",
2186 "but is not a link\n";
2187 }
2188 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2189 if (!$pid) {
2190 open STDIN, '<&', $fh or croak $!;
2191 exec qw/git-hash-object -w --stdin/ or croak $!;
2192 }
2193 chomp($hash = do { local $/; <$out> });
2194 close $out or croak $!;
2195 close $fh or croak $!;
2196 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2197 close $fb->{base} or croak $!;
2198 } else {
2199 $hash = $fb->{blob} or die "no blob information\n";
2200 }
2201 $fb->{pool}->clear;
2202 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2203 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2204 undef;
2205 }
2206
2207 sub abort_edit {
2208 my $self = shift;
2209 $self->{nr} = $self->{gii}->{nr};
2210 delete $self->{gii};
2211 $self->SUPER::abort_edit(@_);
2212 }
2213
2214 sub close_edit {
2215 my $self = shift;
2216 $self->{git_commit_ok} = 1;
2217 $self->{nr} = $self->{gii}->{nr};
2218 delete $self->{gii};
2219 $self->SUPER::close_edit(@_);
2220 }
2221
2222 package SVN::Git::Editor;
2223 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2224 use strict;
2225 use warnings;
2226 use Carp qw/croak/;
2227 use IO::File;
2228 use Digest::MD5;
2229
2230 sub new {
2231 my ($class, $opts) = @_;
2232 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2233 die "$_ required!\n" unless (defined $opts->{$_});
2234 }
2235
2236 my $pool = SVN::Pool->new;
2237 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2238 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2239 $opts->{r}, $mods);
2240
2241 # $opts->{ra} functions should not be used after this:
2242 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
2243 $opts->{editor_cb}, $pool);
2244 my $self = SVN::Delta::Editor->new(@ce, $pool);
2245 bless $self, $class;
2246 foreach (qw/svn_path r tree_a tree_b/) {
2247 $self->{$_} = $opts->{$_};
2248 }
2249 $self->{url} = $opts->{ra}->{url};
2250 $self->{mods} = $mods;
2251 $self->{types} = $types;
2252 $self->{pool} = $pool;
2253 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2254 $self->{rm} = { };
2255 $self->{path_prefix} = length $self->{svn_path} ?
2256 "$self->{svn_path}/" : '';
2257 return $self;
2258 }
2259
2260 sub generate_diff {
2261 my ($tree_a, $tree_b) = @_;
2262 my @diff_tree = qw(diff-tree -z -r);
2263 if ($_cp_similarity) {
2264 push @diff_tree, "-C$_cp_similarity";
2265 } else {
2266 push @diff_tree, '-C';
2267 }
2268 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2269 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2270 push @diff_tree, $tree_a, $tree_b;
2271 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2272 local $/ = "\0";
2273 my $state = 'meta';
2274 my @mods;
2275 while (<$diff_fh>) {
2276 chomp $_; # this gets rid of the trailing "\0"
2277 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2278 $::sha1\s($::sha1)\s
2279 ([MTCRAD])\d*$/xo) {
2280 push @mods, { mode_a => $1, mode_b => $2,
2281 sha1_b => $3, chg => $4 };
2282 if ($4 =~ /^(?:C|R)$/) {
2283 $state = 'file_a';
2284 } else {
2285 $state = 'file_b';
2286 }
2287 } elsif ($state eq 'file_a') {
2288 my $x = $mods[$#mods] or croak "Empty array\n";
2289 if ($x->{chg} !~ /^(?:C|R)$/) {
2290 croak "Error parsing $_, $x->{chg}\n";
2291 }
2292 $x->{file_a} = $_;
2293 $state = 'file_b';
2294 } elsif ($state eq 'file_b') {
2295 my $x = $mods[$#mods] or croak "Empty array\n";
2296 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2297 croak "Error parsing $_, $x->{chg}\n";
2298 }
2299 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2300 croak "Error parsing $_, $x->{chg}\n";
2301 }
2302 $x->{file_b} = $_;
2303 $state = 'meta';
2304 } else {
2305 croak "Error parsing $_\n";
2306 }
2307 }
2308 command_close_pipe($diff_fh, $ctx);
2309 \@mods;
2310 }
2311
2312 sub check_diff_paths {
2313 my ($ra, $pfx, $rev, $mods) = @_;
2314 my %types;
2315 $pfx .= '/' if length $pfx;
2316
2317 sub type_diff_paths {
2318 my ($ra, $types, $path, $rev) = @_;
2319 my @p = split m#/+#, $path;
2320 my $c = shift @p;
2321 unless (defined $types->{$c}) {
2322 $types->{$c} = $ra->check_path($c, $rev);
2323 }
2324 while (@p) {
2325 $c .= '/' . shift @p;
2326 next if defined $types->{$c};
2327 $types->{$c} = $ra->check_path($c, $rev);
2328 }
2329 }
2330
2331 foreach my $m (@$mods) {
2332 foreach my $f (qw/file_a file_b/) {
2333 next unless defined $m->{$f};
2334 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2335 if (length $pfx.$dir && ! defined $types{$dir}) {
2336 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2337 }
2338 }
2339 }
2340 \%types;
2341 }
2342
2343 sub split_path {
2344 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2345 }
2346
2347 sub repo_path {
2348 my ($self, $path) = @_;
2349 $self->{path_prefix}.(defined $path ? $path : '');
2350 }
2351
2352 sub url_path {
2353 my ($self, $path) = @_;
2354 $self->{url} . '/' . $self->repo_path($path);
2355 }
2356
2357 sub rmdirs {
2358 my ($self) = @_;
2359 my $rm = $self->{rm};
2360 delete $rm->{''}; # we never delete the url we're tracking
2361 return unless %$rm;
2362
2363 foreach (keys %$rm) {
2364 my @d = split m#/#, $_;
2365 my $c = shift @d;
2366 $rm->{$c} = 1;
2367 while (@d) {
2368 $c .= '/' . shift @d;
2369 $rm->{$c} = 1;
2370 }
2371 }
2372 delete $rm->{$self->{svn_path}};
2373 delete $rm->{''}; # we never delete the url we're tracking
2374 return unless %$rm;
2375
2376 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2377 $self->{tree_b});
2378 local $/ = "\0";
2379 while (<$fh>) {
2380 chomp;
2381 my @dn = split m#/#, $_;
2382 while (pop @dn) {
2383 delete $rm->{join '/', @dn};
2384 }
2385 unless (%$rm) {
2386 close $fh;
2387 return;
2388 }
2389 }
2390 command_close_pipe($fh, $ctx);
2391
2392 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2393 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2394 $self->close_directory($bat->{$d}, $p);
2395 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2396 print "\tD+\t$d/\n" unless $::_q;
2397 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2398 delete $bat->{$d};
2399 }
2400 }
2401
2402 sub open_or_add_dir {
2403 my ($self, $full_path, $baton) = @_;
2404 my $t = $self->{types}->{$full_path};
2405 if (!defined $t) {
2406 die "$full_path not known in r$self->{r} or we have a bug!\n";
2407 }
2408 if ($t == $SVN::Node::none) {
2409 return $self->add_directory($full_path, $baton,
2410 undef, -1, $self->{pool});
2411 } elsif ($t == $SVN::Node::dir) {
2412 return $self->open_directory($full_path, $baton,
2413 $self->{r}, $self->{pool});
2414 }
2415 print STDERR "$full_path already exists in repository at ",
2416 "r$self->{r} and it is not a directory (",
2417 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2418 exit 1;
2419 }
2420
2421 sub ensure_path {
2422 my ($self, $path) = @_;
2423 my $bat = $self->{bat};
2424 my $repo_path = $self->repo_path($path);
2425 return $bat->{''} unless (length $repo_path);
2426 my @p = split m#/+#, $repo_path;
2427 my $c = shift @p;
2428 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2429 while (@p) {
2430 my $c0 = $c;
2431 $c .= '/' . shift @p;
2432 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2433 }
2434 return $bat->{$c};
2435 }
2436
2437 sub A {
2438 my ($self, $m) = @_;
2439 my ($dir, $file) = split_path($m->{file_b});
2440 my $pbat = $self->ensure_path($dir);
2441 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2442 undef, -1);
2443 print "\tA\t$m->{file_b}\n" unless $::_q;
2444 $self->chg_file($fbat, $m);
2445 $self->close_file($fbat,undef,$self->{pool});
2446 }
2447
2448 sub C {
2449 my ($self, $m) = @_;
2450 my ($dir, $file) = split_path($m->{file_b});
2451 my $pbat = $self->ensure_path($dir);
2452 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2453 $self->url_path($m->{file_a}), $self->{r});
2454 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2455 $self->chg_file($fbat, $m);
2456 $self->close_file($fbat,undef,$self->{pool});
2457 }
2458
2459 sub delete_entry {
2460 my ($self, $path, $pbat) = @_;
2461 my $rpath = $self->repo_path($path);
2462 my ($dir, $file) = split_path($rpath);
2463 $self->{rm}->{$dir} = 1;
2464 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2465 }
2466
2467 sub R {
2468 my ($self, $m) = @_;
2469 my ($dir, $file) = split_path($m->{file_b});
2470 my $pbat = $self->ensure_path($dir);
2471 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2472 $self->url_path($m->{file_a}), $self->{r});
2473 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2474 $self->chg_file($fbat, $m);
2475 $self->close_file($fbat,undef,$self->{pool});
2476
2477 ($dir, $file) = split_path($m->{file_a});
2478 $pbat = $self->ensure_path($dir);
2479 $self->delete_entry($m->{file_a}, $pbat);
2480 }
2481
2482 sub M {
2483 my ($self, $m) = @_;
2484 my ($dir, $file) = split_path($m->{file_b});
2485 my $pbat = $self->ensure_path($dir);
2486 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2487 $pbat,$self->{r},$self->{pool});
2488 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2489 $self->chg_file($fbat, $m);
2490 $self->close_file($fbat,undef,$self->{pool});
2491 }
2492
2493 sub T { shift->M(@_) }
2494
2495 sub change_file_prop {
2496 my ($self, $fbat, $pname, $pval) = @_;
2497 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2498 }
2499
2500 sub chg_file {
2501 my ($self, $fbat, $m) = @_;
2502 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2503 $self->change_file_prop($fbat,'svn:executable','*');
2504 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2505 $self->change_file_prop($fbat,'svn:executable',undef);
2506 }
2507 my $fh = IO::File->new_tmpfile or croak $!;
2508 if ($m->{mode_b} =~ /^120/) {
2509 print $fh 'link ' or croak $!;
2510 $self->change_file_prop($fbat,'svn:special','*');
2511 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2512 $self->change_file_prop($fbat,'svn:special',undef);
2513 }
2514 defined(my $pid = fork) or croak $!;
2515 if (!$pid) {
2516 open STDOUT, '>&', $fh or croak $!;
2517 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2518 }
2519 waitpid $pid, 0;
2520 croak $? if $?;
2521 $fh->flush == 0 or croak $!;
2522 seek $fh, 0, 0 or croak $!;
2523
2524 my $md5 = Digest::MD5->new;
2525 $md5->addfile($fh) or croak $!;
2526 seek $fh, 0, 0 or croak $!;
2527
2528 my $exp = $md5->hexdigest;
2529 my $pool = SVN::Pool->new;
2530 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2531 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2532 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2533 $pool->clear;
2534
2535 close $fh or croak $!;
2536 }
2537
2538 sub D {
2539 my ($self, $m) = @_;
2540 my ($dir, $file) = split_path($m->{file_b});
2541 my $pbat = $self->ensure_path($dir);
2542 print "\tD\t$m->{file_b}\n" unless $::_q;
2543 $self->delete_entry($m->{file_b}, $pbat);
2544 }
2545
2546 sub close_edit {
2547 my ($self) = @_;
2548 my ($p,$bat) = ($self->{pool}, $self->{bat});
2549 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2550 $self->close_directory($bat->{$_}, $p);
2551 }
2552 $self->SUPER::close_edit($p);
2553 $p->clear;
2554 }
2555
2556 sub abort_edit {
2557 my ($self) = @_;
2558 $self->SUPER::abort_edit($self->{pool});
2559 }
2560
2561 sub DESTROY {
2562 my $self = shift;
2563 $self->SUPER::DESTROY(@_);
2564 $self->{pool}->clear;
2565 }
2566
2567 # this drives the editor
2568 sub apply_diff {
2569 my ($self) = @_;
2570 my $mods = $self->{mods};
2571 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2572 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2573 my $f = $m->{chg};
2574 if (defined $o{$f}) {
2575 $self->$f($m);
2576 } else {
2577 fatal("Invalid change type: $f\n");
2578 }
2579 }
2580 $self->rmdirs if $_rmdir;
2581 if (@$mods == 0) {
2582 $self->abort_edit;
2583 } else {
2584 $self->close_edit;
2585 }
2586 return scalar @$mods;
2587 }
2588
2589 package Git::SVN::Ra;
2590 use vars qw/@ISA $config_dir/;
2591 use strict;
2592 use warnings;
2593 my ($can_do_switch);
2594 my $RA;
2595
2596 BEGIN {
2597 # enforce temporary pool usage for some simple functions
2598 my $e;
2599 foreach (qw/get_latest_revnum get_uuid get_repos_root/) {
2600 $e .= "sub $_ {
2601 my \$self = shift;
2602 my \$pool = SVN::Pool->new;
2603 my \@ret = \$self->SUPER::$_(\@_,\$pool);
2604 \$pool->clear;
2605 wantarray ? \@ret : \$ret[0]; }\n";
2606 }
2607
2608 # get_dir needs $pool held in cache for dirents to work,
2609 # check_path is cacheable and rev_proplist is close enough
2610 # for our purposes.
2611 foreach (qw/check_path get_dir rev_proplist/) {
2612 $e .= "my \%${_}_cache; my \$${_}_rev = 0; sub $_ {
2613 my \$self = shift;
2614 my \$r = pop;
2615 my \$k = join(\"\\0\", \@_);
2616 if (my \$x = \$${_}_cache{\$r}->{\$k}) {
2617 return wantarray ? \@\$x : \$x->[0];
2618 }
2619 my \$pool = SVN::Pool->new;
2620 my \@ret = \$self->SUPER::$_(\@_, \$r, \$pool);
2621 if (\$r != \$${_}_rev) {
2622 \%${_}_cache = ( pool => [] );
2623 \$${_}_rev = \$r;
2624 }
2625 \$${_}_cache{\$r}->{\$k} = \\\@ret;
2626 push \@{\$${_}_cache{pool}}, \$pool;
2627 wantarray ? \@ret : \$ret[0]; }\n";
2628 }
2629 $e .= "\n1;";
2630 eval $e or die $@;
2631 }
2632
2633 sub new {
2634 my ($class, $url) = @_;
2635 $url =~ s!/+$!!;
2636 return $RA if ($RA && $RA->{url} eq $url);
2637
2638 SVN::_Core::svn_config_ensure($config_dir, undef);
2639 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2640 SVN::Client::get_simple_provider(),
2641 SVN::Client::get_ssl_server_trust_file_provider(),
2642 SVN::Client::get_simple_prompt_provider(
2643 \&Git::SVN::Prompt::simple, 2),
2644 SVN::Client::get_ssl_client_cert_prompt_provider(
2645 \&Git::SVN::Prompt::ssl_client_cert, 2),
2646 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2647 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2648 SVN::Client::get_username_provider(),
2649 SVN::Client::get_ssl_server_trust_prompt_provider(
2650 \&Git::SVN::Prompt::ssl_server_trust),
2651 SVN::Client::get_username_prompt_provider(
2652 \&Git::SVN::Prompt::username, 2),
2653 ]);
2654 my $config = SVN::Core::config_get_config($config_dir);
2655 my $self = SVN::Ra->new(url => $url, auth => $baton,
2656 config => $config,
2657 pool => SVN::Pool->new,
2658 auth_provider_callbacks => $callbacks);
2659 $self->{svn_path} = $url;
2660 $self->{repos_root} = $self->get_repos_root;
2661 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
2662 $RA = bless $self, $class;
2663 }
2664
2665 sub DESTROY {
2666 # do not call the real DESTROY since we store ourselves in $RA
2667 }
2668
2669 sub get_log {
2670 my ($self, @args) = @_;
2671 my $pool = SVN::Pool->new;
2672 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2673 my $ret = $self->SUPER::get_log(@args, $pool);
2674 $pool->clear;
2675 $ret;
2676 }
2677
2678 sub get_commit_editor {
2679 my ($self, $log, $cb, $pool) = @_;
2680 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2681 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2682 }
2683
2684 sub gs_do_update {
2685 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2686 my $new = ($rev_a == $rev_b);
2687 my $path = $gs->{path};
2688
2689 my $pool = SVN::Pool->new;
2690 $editor->set_path_strip($path);
2691 my (@pc) = split m#/#, $path;
2692 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2693 1, $editor, $pool);
2694 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2695
2696 # Since we can't rely on svn_ra_reparent being available, we'll
2697 # just have to do some magic with set_path to make it so
2698 # we only want a partial path.
2699 my $sp = '';
2700 my $final = join('/', @pc);
2701 while (@pc) {
2702 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2703 $sp .= '/' if length $sp;
2704 $sp .= shift @pc;
2705 }
2706 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2707
2708 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2709
2710 $reporter->finish_report($pool);
2711 $pool->clear;
2712 $editor->{git_commit_ok};
2713 }
2714
2715 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2716 # svn_ra_reparent didn't work before 1.4)
2717 sub gs_do_switch {
2718 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
2719 my $path = $gs->{path};
2720 my $pool = SVN::Pool->new;
2721
2722 my $full_url = $self->{url};
2723 my $old_url = $full_url;
2724 $full_url .= "/$path" if length $path;
2725 my ($ra, $reparented);
2726 if ($old_url ne $full_url) {
2727 if ($old_url !~ m#^svn(\+ssh)?://#) {
2728 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2729 $pool);
2730 $self->{url} = $full_url;
2731 $reparented = 1;
2732 } else {
2733 $ra = Git::SVN::Ra->new($full_url);
2734 }
2735 }
2736 $ra ||= $self;
2737 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
2738 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2739 $reporter->set_path('', $rev_a, 0, @lock, $pool);
2740 $reporter->finish_report($pool);
2741
2742 if ($reparented) {
2743 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2744 $self->{url} = $old_url;
2745 }
2746
2747 $pool->clear;
2748 $editor->{git_commit_ok};
2749 }
2750
2751 sub gs_fetch_loop_common {
2752 my ($self, $base, $head, $gsv, $globs) = @_;
2753 return if ($base > $head);
2754 my $inc = 1000;
2755 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2756 my %common;
2757 my $common_max = scalar @$gsv;
2758
2759 foreach my $gs (@$gsv) {
2760 if (my $last_commit = $gs->last_commit) {
2761 $gs->assert_index_clean($last_commit);
2762 }
2763 my @tmp = split m#/#, $gs->{path};
2764 my $p = '';
2765 foreach (@tmp) {
2766 $p .= length($p) ? "/$_" : $_;
2767 $common{$p} ||= 0;
2768 $common{$p}++;
2769 }
2770 }
2771 $globs ||= [];
2772 $common_max += scalar @$globs;
2773 foreach my $glob (@$globs) {
2774 my @tmp = split m#/#, $glob->{path}->{left};
2775 my $p = '';
2776 foreach (@tmp) {
2777 $p .= length($p) ? "/$_" : $_;
2778 $common{$p} ||= 0;
2779 $common{$p}++;
2780 }
2781 }
2782
2783 my $longest_path = '';
2784 foreach (sort {length $b <=> length $a} keys %common) {
2785 if ($common{$_} == $common_max) {
2786 $longest_path = $_;
2787 last;
2788 }
2789 }
2790 while (1) {
2791 my %revs;
2792 my $err;
2793 my $err_handler = $SVN::Error::handler;
2794 $SVN::Error::handler = sub {
2795 ($err) = @_;
2796 skip_unknown_revs($err);
2797 };
2798 sub _cb {
2799 my ($paths, $r, $author, $date, $log) = @_;
2800 [ dup_changed_paths($paths),
2801 { author => $author, date => $date, log => $log } ];
2802 }
2803 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
2804 sub { $revs{$_[1]} = _cb(@_) });
2805 if ($err && $max >= $head) {
2806 print STDERR "Path '$longest_path' ",
2807 "was probably deleted:\n",
2808 $err->expanded_message,
2809 "\nWill attempt to follow ",
2810 "revisions r$min .. r$max ",
2811 "committed before the deletion\n";
2812 my $hi = $max;
2813 while (--$hi >= $min) {
2814 my $ok;
2815 $self->get_log([$longest_path], $min, $hi,
2816 0, 1, 1, sub {
2817 $ok ||= $_[1];
2818 $revs{$_[1]} = _cb(@_) });
2819 if ($ok) {
2820 print STDERR "r$min .. r$ok OK\n";
2821 last;
2822 }
2823 }
2824 }
2825 $SVN::Error::handler = $err_handler;
2826
2827 my %exists = map { $_->{path} => $_ } @$gsv;
2828 foreach my $r (sort {$a <=> $b} keys %revs) {
2829 my ($paths, $logged) = @{$revs{$r}};
2830
2831 foreach my $gs ($self->match_globs(\%exists, $paths,
2832 $globs, $r)) {
2833 if ($gs->rev_db_max >= $r) {
2834 next;
2835 }
2836 next unless $gs->match_paths($paths, $r);
2837 $gs->{logged_rev_props} = $logged;
2838 my $log_entry = $gs->do_fetch($paths, $r);
2839 if ($log_entry) {
2840 $gs->do_git_commit($log_entry);
2841 }
2842 }
2843 foreach my $g (@$globs) {
2844 my $k = "svn-remote.$g->{remote}." .
2845 "$g->{t}-maxRev";
2846 Git::SVN::tmp_config($k, $r);
2847 }
2848 }
2849 # pre-fill the .rev_db since it'll eventually get filled in
2850 # with '0' x40 if something new gets committed
2851 foreach my $gs (@$gsv) {
2852 next if defined $gs->rev_db_get($max);
2853 $gs->rev_db_set($max, 0 x40);
2854 }
2855 foreach my $g (@$globs) {
2856 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
2857 Git::SVN::tmp_config($k, $max);
2858 }
2859 last if $max >= $head;
2860 $min = $max + 1;
2861 $max += $inc;
2862 $max = $head if ($max > $head);
2863 }
2864 }
2865
2866 sub match_globs {
2867 my ($self, $exists, $paths, $globs, $r) = @_;
2868
2869 sub get_dir_check {
2870 my ($self, $exists, $g, $r) = @_;
2871 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
2872 return unless scalar @x == 3;
2873 my $dirents = $x[0];
2874 foreach my $de (keys %$dirents) {
2875 next if $dirents->{$de}->kind != $SVN::Node::dir;
2876 my $p = $g->{path}->full_path($de);
2877 next if $exists->{$p};
2878 next if (length $g->{path}->{right} &&
2879 ($self->check_path($p, $r) !=
2880 $SVN::Node::dir));
2881 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
2882 $g->{ref}->full_path($de), 1);
2883 }
2884 }
2885 foreach my $g (@$globs) {
2886 if (my $path = $paths->{"/$g->{path}->{left}"}) {
2887 if ($path->{action} =~ /^[AR]$/) {
2888 get_dir_check($self, $exists, $g, $r);
2889 }
2890 }
2891 foreach (keys %$paths) {
2892 if (/$g->{path}->{left_regex}/) {
2893 next if $paths->{$_}->{action} !~ /^[AR]$/;
2894 get_dir_check($self, $exists, $g, $r);
2895 }
2896 next unless /$g->{path}->{regex}/;
2897 my $p = $1;
2898 my $pathname = $g->{path}->full_path($p);
2899 next if $exists->{$pathname};
2900 $exists->{$pathname} = Git::SVN->init(
2901 $self->{url}, $pathname, undef,
2902 $g->{ref}->full_path($p), 1);
2903 }
2904 my $c = '';
2905 foreach (split m#/#, $g->{path}->{left}) {
2906 $c .= "/$_";
2907 next unless ($paths->{$c} &&
2908 ($paths->{$c}->{action} =~ /^[AR]$/));
2909 get_dir_check($self, $exists, $g, $r);
2910 }
2911 }
2912 values %$exists;
2913 }
2914
2915 sub minimize_url {
2916 my ($self) = @_;
2917 return $self->{url} if ($self->{url} eq $self->{repos_root});
2918 my $url = $self->{repos_root};
2919 my @components = split(m!/!, $self->{svn_path});
2920 my $c = '';
2921 do {
2922 $url .= "/$c" if length $c;
2923 eval { (ref $self)->new($url)->get_latest_revnum };
2924 } while ($@ && ($c = shift @components));
2925 $url;
2926 }
2927
2928 sub can_do_switch {
2929 my $self = shift;
2930 unless (defined $can_do_switch) {
2931 my $pool = SVN::Pool->new;
2932 my $rep = eval {
2933 $self->do_switch(1, '', 0, $self->{url},
2934 SVN::Delta::Editor->new, $pool);
2935 };
2936 if ($@) {
2937 $can_do_switch = 0;
2938 } else {
2939 $rep->abort_report($pool);
2940 $can_do_switch = 1;
2941 }
2942 $pool->clear;
2943 }
2944 $can_do_switch;
2945 }
2946
2947 sub skip_unknown_revs {
2948 my ($err) = @_;
2949 my $errno = $err->apr_err();
2950 # Maybe the branch we're tracking didn't
2951 # exist when the repo started, so it's
2952 # not an error if it doesn't, just continue
2953 #
2954 # Wonderfully consistent library, eh?
2955 # 160013 - svn:// and file://
2956 # 175002 - http(s)://
2957 # 175007 - http(s):// (this repo required authorization, too...)
2958 # More codes may be discovered later...
2959 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2960 return;
2961 }
2962 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2963 }
2964
2965 # svn_log_changed_path_t objects passed to get_log are likely to be
2966 # overwritten even if only the refs are copied to an external variable,
2967 # so we should dup the structures in their entirety. Using an externally
2968 # passed pool (instead of our temporary and quickly cleared pool in
2969 # Git::SVN::Ra) does not help matters at all...
2970 sub dup_changed_paths {
2971 my ($paths) = @_;
2972 return undef unless $paths;
2973 my %ret;
2974 foreach my $p (keys %$paths) {
2975 my $i = $paths->{$p};
2976 my %s = map { $_ => $i->$_ }
2977 qw/copyfrom_path copyfrom_rev action/;
2978 $ret{$p} = \%s;
2979 }
2980 \%ret;
2981 }
2982
2983 package Git::SVN::Log;
2984 use strict;
2985 use warnings;
2986 use POSIX qw/strftime/;
2987 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2988 %rusers $show_commit $incremental/;
2989 my $l_fmt;
2990
2991 sub cmt_showable {
2992 my ($c) = @_;
2993 return 1 if defined $c->{r};
2994 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2995 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2996 my @log = command(qw/cat-file commit/, $c->{c});
2997 shift @log while ($log[0] ne "\n");
2998 shift @log;
2999 @{$c->{l}} = grep !/^git-svn-id: /, @log;
3000
3001 (undef, $c->{r}, undef) = ::extract_metadata(
3002 (grep(/^git-svn-id: /, @log))[-1]);
3003 }
3004 return defined $c->{r};
3005 }
3006
3007 sub log_use_color {
3008 return 1 if $color;
3009 my ($dc, $dcvar);
3010 $dcvar = 'color.diff';
3011 $dc = `git-config --get $dcvar`;
3012 if ($dc eq '') {
3013 # nothing at all; fallback to "diff.color"
3014 $dcvar = 'diff.color';
3015 $dc = `git-config --get $dcvar`;
3016 }
3017 chomp($dc);
3018 if ($dc eq 'auto') {
3019 my $pc;
3020 $pc = `git-config --get color.pager`;
3021 if ($pc eq '') {
3022 # does not have it -- fallback to pager.color
3023 $pc = `git-config --bool --get pager.color`;
3024 }
3025 else {
3026 $pc = `git-config --bool --get color.pager`;
3027 if ($?) {
3028 $pc = 'false';
3029 }
3030 }
3031 chomp($pc);
3032 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3033 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3034 }
3035 return 0;
3036 }
3037 return 0 if $dc eq 'never';
3038 return 1 if $dc eq 'always';
3039 chomp($dc = `git-config --bool --get $dcvar`);
3040 return ($dc eq 'true');
3041 }
3042
3043 sub git_svn_log_cmd {
3044 my ($r_min, $r_max, @args) = @_;
3045 my $head = 'HEAD';
3046 foreach my $x (@args) {
3047 last if $x eq '--';
3048 next unless ::verify_ref("$x^0");
3049 $head = $x;
3050 last;
3051 }
3052
3053 my $url;
3054 my ($fh, $ctx) = command_output_pipe('rev-list', $head);
3055 while (<$fh>) {
3056 chomp;
3057 $url = (::cmt_metadata($_))[0];
3058 last if defined $url;
3059 }
3060 close $fh; # break the pipe
3061
3062 my $gs = Git::SVN->find_by_url($url) || Git::SVN->_new;
3063 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3064 $gs->refname);
3065 push @cmd, '-r' unless $non_recursive;
3066 push @cmd, qw/--raw --name-status/ if $verbose;
3067 push @cmd, '--color' if log_use_color();
3068 return @cmd unless defined $r_max;
3069 if ($r_max == $r_min) {
3070 push @cmd, '--max-count=1';
3071 if (my $c = $gs->rev_db_get($r_max)) {
3072 push @cmd, $c;
3073 }
3074 } else {
3075 my ($c_min, $c_max);
3076 $c_max = $gs->rev_db_get($r_max);
3077 $c_min = $gs->rev_db_get($r_min);
3078 if (defined $c_min && defined $c_max) {
3079 if ($r_max > $r_max) {
3080 push @cmd, "$c_min..$c_max";
3081 } else {
3082 push @cmd, "$c_max..$c_min";
3083 }
3084 } elsif ($r_max > $r_min) {
3085 push @cmd, $c_max;
3086 } else {
3087 push @cmd, $c_min;
3088 }
3089 }
3090 return @cmd;
3091 }
3092
3093 # adapted from pager.c
3094 sub config_pager {
3095 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3096 if (!defined $pager) {
3097 $pager = 'less';
3098 } elsif (length $pager == 0 || $pager eq 'cat') {
3099 $pager = undef;
3100 }
3101 }
3102
3103 sub run_pager {
3104 return unless -t *STDOUT;
3105 pipe my $rfd, my $wfd or return;
3106 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
3107 if (!$pid) {
3108 open STDOUT, '>&', $wfd or
3109 ::fatal "Can't redirect to stdout: $!\n";
3110 return;
3111 }
3112 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
3113 $ENV{LESS} ||= 'FRSX';
3114 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
3115 }
3116
3117 sub tz_to_s_offset {
3118 my ($tz) = @_;
3119 $tz =~ s/(\d\d)$//;
3120 return ($1 * 60) + ($tz * 3600);
3121 }
3122
3123 sub get_author_info {
3124 my ($dest, $author, $t, $tz) = @_;
3125 $author =~ s/(?:^\s*|\s*$)//g;
3126 $dest->{a_raw} = $author;
3127 my $au;
3128 if ($::_authors) {
3129 $au = $rusers{$author} || undef;
3130 }
3131 if (!$au) {
3132 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3133 }
3134 $dest->{t} = $t;
3135 $dest->{tz} = $tz;
3136 $dest->{a} = $au;
3137 # Date::Parse isn't in the standard Perl distro :(
3138 if ($tz =~ s/^\+//) {
3139 $t += tz_to_s_offset($tz);
3140 } elsif ($tz =~ s/^\-//) {
3141 $t -= tz_to_s_offset($tz);
3142 }
3143 $dest->{t_utc} = $t;
3144 }
3145
3146 sub process_commit {
3147 my ($c, $r_min, $r_max, $defer) = @_;
3148 if (defined $r_min && defined $r_max) {
3149 if ($r_min == $c->{r} && $r_min == $r_max) {
3150 show_commit($c);
3151 return 0;
3152 }
3153 return 1 if $r_min == $r_max;
3154 if ($r_min < $r_max) {
3155 # we need to reverse the print order
3156 return 0 if (defined $limit && --$limit < 0);
3157 push @$defer, $c;
3158 return 1;
3159 }
3160 if ($r_min != $r_max) {
3161 return 1 if ($r_min < $c->{r});
3162 return 1 if ($r_max > $c->{r});
3163 }
3164 }
3165 return 0 if (defined $limit && --$limit < 0);
3166 show_commit($c);
3167 return 1;
3168 }
3169
3170 sub show_commit {
3171 my $c = shift;
3172 if ($oneline) {
3173 my $x = "\n";
3174 if (my $l = $c->{l}) {
3175 while ($l->[0] =~ /^\s*$/) { shift @$l }
3176 $x = $l->[0];
3177 }
3178 $l_fmt ||= 'A' . length($c->{r});
3179 print 'r',pack($l_fmt, $c->{r}),' | ';
3180 print "$c->{c} | " if $show_commit;
3181 print $x;
3182 } else {
3183 show_commit_normal($c);
3184 }
3185 }
3186
3187 sub show_commit_changed_paths {
3188 my ($c) = @_;
3189 return unless $c->{changed};
3190 print "Changed paths:\n", @{$c->{changed}};
3191 }
3192
3193 sub show_commit_normal {
3194 my ($c) = @_;
3195 print '-' x72, "\nr$c->{r} | ";
3196 print "$c->{c} | " if $show_commit;
3197 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3198 localtime($c->{t_utc})), ' | ';
3199 my $nr_line = 0;
3200
3201 if (my $l = $c->{l}) {
3202 while ($l->[$#$l] eq "\n" && $#$l > 0
3203 && $l->[($#$l - 1)] eq "\n") {
3204 pop @$l;
3205 }
3206 $nr_line = scalar @$l;
3207 if (!$nr_line) {
3208 print "1 line\n\n\n";
3209 } else {
3210 if ($nr_line == 1) {
3211 $nr_line = '1 line';
3212 } else {
3213 $nr_line .= ' lines';
3214 }
3215 print $nr_line, "\n";
3216 show_commit_changed_paths($c);
3217 print "\n";
3218 print $_ foreach @$l;
3219 }
3220 } else {
3221 print "1 line\n";
3222 show_commit_changed_paths($c);
3223 print "\n";
3224
3225 }
3226 foreach my $x (qw/raw diff/) {
3227 if ($c->{$x}) {
3228 print "\n";
3229 print $_ foreach @{$c->{$x}}
3230 }
3231 }
3232 }
3233
3234 sub cmd_show_log {
3235 my (@args) = @_;
3236 my ($r_min, $r_max);
3237 my $r_last = -1; # prevent dupes
3238 if (defined $TZ) {
3239 $ENV{TZ} = $TZ;
3240 } else {
3241 delete $ENV{TZ};
3242 }
3243 if (defined $::_revision) {
3244 if ($::_revision =~ /^(\d+):(\d+)$/) {
3245 ($r_min, $r_max) = ($1, $2);
3246 } elsif ($::_revision =~ /^\d+$/) {
3247 $r_min = $r_max = $::_revision;
3248 } else {
3249 ::fatal "-r$::_revision is not supported, use ",
3250 "standard \'git log\' arguments instead\n";
3251 }
3252 }
3253
3254 config_pager();
3255 @args = (git_svn_log_cmd($r_min, $r_max, @args), @args);
3256 my $log = command_output_pipe(@args);
3257 run_pager();
3258 my (@k, $c, $d);
3259 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3260 while (<$log>) {
3261 if (/^${esc_color}commit ($::sha1_short)/o) {
3262 my $cmt = $1;
3263 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3264 $r_last = $c->{r};
3265 process_commit($c, $r_min, $r_max, \@k) or
3266 goto out;
3267 }
3268 $d = undef;
3269 $c = { c => $cmt };
3270 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3271 get_author_info($c, $1, $2, $3);
3272 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3273 # ignore
3274 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3275 push @{$c->{raw}}, $_;
3276 } elsif (/^${esc_color}[ACRMDT]\t/) {
3277 # we could add $SVN->{svn_path} here, but that requires
3278 # remote access at the moment (repo_path_split)...
3279 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
3280 push @{$c->{changed}}, $_;
3281 } elsif (/^${esc_color}diff /o) {
3282 $d = 1;
3283 push @{$c->{diff}}, $_;
3284 } elsif ($d) {
3285 push @{$c->{diff}}, $_;
3286 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
3287 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3288 } elsif (s/^${esc_color} //o) {
3289 push @{$c->{l}}, $_;
3290 }
3291 }
3292 if ($c && defined $c->{r} && $c->{r} != $r_last) {
3293 $r_last = $c->{r};
3294 process_commit($c, $r_min, $r_max, \@k);
3295 }
3296 if (@k) {
3297 my $swap = $r_max;
3298 $r_max = $r_min;
3299 $r_min = $swap;
3300 process_commit($_, $r_min, $r_max) foreach reverse @k;
3301 }
3302 out:
3303 close $log;
3304 print '-' x72,"\n" unless $incremental || $oneline;
3305 }
3306
3307 package Git::SVN::Migration;
3308 # these version numbers do NOT correspond to actual version numbers
3309 # of git nor git-svn. They are just relative.
3310 #
3311 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3312 #
3313 # v1 layout: .git/$id/info/url, refs/remotes/$id
3314 #
3315 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3316 #
3317 # v3 layout: .git/svn/$id, refs/remotes/$id
3318 # - info/url may remain for backwards compatibility
3319 # - this is what we migrate up to this layout automatically,
3320 # - this will be used by git svn init on single branches
3321 # v3.1 layout (auto migrated):
3322 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
3323 # for backwards compatibility
3324 #
3325 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3326 # - this is only created for newly multi-init-ed
3327 # repositories. Similar in spirit to the
3328 # --use-separate-remotes option in git-clone (now default)
3329 # - we do not automatically migrate to this (following
3330 # the example set by core git)
3331 use strict;
3332 use warnings;
3333 use Carp qw/croak/;
3334 use File::Path qw/mkpath/;
3335 use File::Basename qw/dirname basename/;
3336 use vars qw/$_minimize/;
3337
3338 sub migrate_from_v0 {
3339 my $git_dir = $ENV{GIT_DIR};
3340 return undef unless -d $git_dir;
3341 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3342 my $migrated = 0;
3343 while (<$fh>) {
3344 chomp;
3345 my ($id, $orig_ref) = ($_, $_);
3346 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3347 next unless -f "$git_dir/$id/info/url";
3348 my $new_ref = "refs/remotes/$id";
3349 if (::verify_ref("$new_ref^0")) {
3350 print STDERR "W: $orig_ref is probably an old ",
3351 "branch used by an ancient version of ",
3352 "git-svn.\n",
3353 "However, $new_ref also exists.\n",
3354 "We will not be able ",
3355 "to use this branch until this ",
3356 "ambiguity is resolved.\n";
3357 next;
3358 }
3359 print STDERR "Migrating from v0 layout...\n" if !$migrated;
3360 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3361 command_noisy('update-ref', $new_ref, $orig_ref);
3362 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3363 $migrated++;
3364 }
3365 command_close_pipe($fh, $ctx);
3366 print STDERR "Done migrating from v0 layout...\n" if $migrated;
3367 $migrated;
3368 }
3369
3370 sub migrate_from_v1 {
3371 my $git_dir = $ENV{GIT_DIR};
3372 my $migrated = 0;
3373 return $migrated unless -d $git_dir;
3374 my $svn_dir = "$git_dir/svn";
3375
3376 # just in case somebody used 'svn' as their $id at some point...
3377 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3378
3379 print STDERR "Migrating from a git-svn v1 layout...\n";
3380 mkpath([$svn_dir]);
3381 print STDERR "Data from a previous version of git-svn exists, but\n\t",
3382 "$svn_dir\n\t(required for this version ",
3383 "($::VERSION) of git-svn) does not. exist\n";
3384 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3385 while (<$fh>) {
3386 my $x = $_;
3387 next unless $x =~ s#^refs/remotes/##;
3388 chomp $x;
3389 next unless -f "$git_dir/$x/info/url";
3390 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3391 next unless $u;
3392 my $dn = dirname("$git_dir/svn/$x");
3393 mkpath([$dn]) unless -d $dn;
3394 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3395 mkpath(["$git_dir/svn/svn"]);
3396 print STDERR " - $git_dir/$x/info => ",
3397 "$git_dir/svn/$x/info\n";
3398 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3399 croak "$!: $x";
3400 # don't worry too much about these, they probably
3401 # don't exist with repos this old (save for index,
3402 # and we can easily regenerate that)
3403 foreach my $f (qw/unhandled.log index .rev_db/) {
3404 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3405 }
3406 } else {
3407 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3408 rename "$git_dir/$x", "$git_dir/svn/$x" or
3409 croak "$!: $x";
3410 }
3411 $migrated++;
3412 }
3413 command_close_pipe($fh, $ctx);
3414 print STDERR "Done migrating from a git-svn v1 layout\n";
3415 $migrated;
3416 }
3417
3418 sub read_old_urls {
3419 my ($l_map, $pfx, $path) = @_;
3420 my @dir;
3421 foreach (<$path/*>) {
3422 if (-r "$_/info/url") {
3423 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3424 my $ref_id = $pfx . basename $_;
3425 my $url = ::file_to_s("$_/info/url");
3426 $l_map->{$ref_id} = $url;
3427 } elsif (-d $_) {
3428 push @dir, $_;
3429 }
3430 }
3431 foreach (@dir) {
3432 my $x = $_;
3433 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3434 read_old_urls($l_map, $x, $_);
3435 }
3436 }
3437
3438 sub migrate_from_v2 {
3439 my @cfg = command(qw/config -l/);
3440 return if grep /^svn-remote\..+\.url=/, @cfg;
3441 my %l_map;
3442 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3443 my $migrated = 0;
3444
3445 foreach my $ref_id (sort keys %l_map) {
3446 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3447 if ($@) {
3448 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3449 }
3450 $migrated++;
3451 }
3452 $migrated;
3453 }
3454
3455 sub minimize_connections {
3456 my $r = Git::SVN::read_all_remotes();
3457 my $new_urls = {};
3458 my $root_repos = {};
3459 foreach my $repo_id (keys %$r) {
3460 my $url = $r->{$repo_id}->{url} or next;
3461 my $fetch = $r->{$repo_id}->{fetch} or next;
3462 my $ra = Git::SVN::Ra->new($url);
3463
3464 # skip existing cases where we already connect to the root
3465 if (($ra->{url} eq $ra->{repos_root}) ||
3466 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3467 $repo_id)) {
3468 $root_repos->{$ra->{url}} = $repo_id;
3469 next;
3470 }
3471
3472 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3473 my $root_path = $ra->{url};
3474 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3475 foreach my $path (keys %$fetch) {
3476 my $ref_id = $fetch->{$path};
3477 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3478
3479 # make sure we can read when connecting to
3480 # a higher level of a repository
3481 my ($last_rev, undef) = $gs->last_rev_commit;
3482 if (!defined $last_rev) {
3483 $last_rev = eval {
3484 $root_ra->get_latest_revnum;
3485 };
3486 next if $@;
3487 }
3488 my $new = $root_path;
3489 $new .= length $path ? "/$path" : '';
3490 eval {
3491 $root_ra->get_log([$new], $last_rev, $last_rev,
3492 0, 0, 1, sub { });
3493 };
3494 next if $@;
3495 $new_urls->{$ra->{repos_root}}->{$new} =
3496 { ref_id => $ref_id,
3497 old_repo_id => $repo_id,
3498 old_path => $path };
3499 }
3500 }
3501
3502 my @emptied;
3503 foreach my $url (keys %$new_urls) {
3504 # see if we can re-use an existing [svn-remote "repo_id"]
3505 # instead of creating a(n ugly) new section:
3506 my $repo_id = $root_repos->{$url} ||
3507 Git::SVN::sanitize_remote_name($url);
3508
3509 my $fetch = $new_urls->{$url};
3510 foreach my $path (keys %$fetch) {
3511 my $x = $fetch->{$path};
3512 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3513 my $pfx = "svn-remote.$x->{old_repo_id}";
3514
3515 my $old_fetch = quotemeta("$x->{old_path}:".
3516 "refs/remotes/$x->{ref_id}");
3517 command_noisy(qw/config --unset/,
3518 "$pfx.fetch", '^'. $old_fetch . '$');
3519 delete $r->{$x->{old_repo_id}}->
3520 {fetch}->{$x->{old_path}};
3521 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3522 command_noisy(qw/config --unset/,
3523 "$pfx.url");
3524 push @emptied, $x->{old_repo_id}
3525 }
3526 }
3527 }
3528 if (@emptied) {
3529 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3530 "$ENV{GIT_DIR}/config";
3531 print STDERR <<EOF;
3532 The following [svn-remote] sections in your config file ($file) are empty
3533 and can be safely removed:
3534 EOF
3535 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3536 }
3537 }
3538
3539 sub migration_check {
3540 migrate_from_v0();
3541 migrate_from_v1();
3542 migrate_from_v2();
3543 minimize_connections() if $_minimize;
3544 }
3545
3546 package Git::IndexInfo;
3547 use strict;
3548 use warnings;
3549 use Git qw/command_input_pipe command_close_pipe/;
3550
3551 sub new {
3552 my ($class) = @_;
3553 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3554 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3555 }
3556
3557 sub remove {
3558 my ($self, $path) = @_;
3559 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3560 return ++$self->{nr};
3561 }
3562 undef;
3563 }
3564
3565 sub update {
3566 my ($self, $mode, $hash, $path) = @_;
3567 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3568 return ++$self->{nr};
3569 }
3570 undef;
3571 }
3572
3573 sub DESTROY {
3574 my ($self) = @_;
3575 command_close_pipe($self->{gui}, $self->{ctx});
3576 }
3577
3578 package Git::SVN::GlobSpec;
3579 use strict;
3580 use warnings;
3581
3582 sub new {
3583 my ($class, $glob) = @_;
3584 my $re = $glob;
3585 $re =~ s!/+$!!g; # no need for trailing slashes
3586 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
3587 my ($left, $right) = ($1, $2);
3588 if ($nr > 1) {
3589 die "Only one '*' wildcard expansion ",
3590 "is supported (got $nr): '$glob'\n";
3591 } elsif ($nr == 0) {
3592 die "One '*' is needed for glob: '$glob'\n";
3593 }
3594 $re = quotemeta($left) . $re . quotemeta($right);
3595 if (length $left && !($left =~ s!/+$!!g)) {
3596 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
3597 }
3598 if (length $right && !($right =~ s!^/+!!g)) {
3599 die "Missing leading '/' on right side of: '$glob' ($right)\n";
3600 }
3601 my $left_re = qr/^\/\Q$left\E(\/|$)/;
3602 bless { left => $left, right => $right, left_regex => $left_re,
3603 regex => qr/$re/, glob => $glob }, $class;
3604 }
3605
3606 sub full_path {
3607 my ($self, $path) = @_;
3608 return (length $self->{left} ? "$self->{left}/" : '') .
3609 $path . (length $self->{right} ? "/$self->{right}" : '');
3610 }
3611
3612 __END__
3613
3614 Data structures:
3615
3616
3617 $remotes = { # returned by read_all_remotes()
3618 'svn' => {
3619 # svn-remote.svn.url=https://svn.musicpd.org
3620 url => 'https://svn.musicpd.org',
3621 # svn-remote.svn.fetch=mpd/trunk:trunk
3622 fetch => {
3623 'mpd/trunk' => 'trunk',
3624 },
3625 # svn-remote.svn.tags=mpd/tags/*:tags/*
3626 tags => {
3627 path => {
3628 left => 'mpd/tags',
3629 right => '',
3630 regex => qr!mpd/tags/([^/]+)$!,
3631 glob => 'tags/*',
3632 },
3633 ref => {
3634 left => 'tags',
3635 right => '',
3636 regex => qr!tags/([^/]+)$!,
3637 glob => 'tags/*',
3638 },
3639 }
3640 }
3641 };
3642
3643 $log_entry hashref as returned by libsvn_log_entry()
3644 {
3645 log => 'whitespace-formatted log entry
3646 ', # trailing newline is preserved
3647 revision => '8', # integer
3648 date => '2004-02-24T17:01:44.108345Z', # commit date
3649 author => 'committer name'
3650 };
3651
3652
3653 # this is generated by generate_diff();
3654 @mods = array of diff-index line hashes, each element represents one line
3655 of diff-index output
3656
3657 diff-index line ($m hash)
3658 {
3659 mode_a => first column of diff-index output, no leading ':',
3660 mode_b => second column of diff-index output,
3661 sha1_b => sha1sum of the final blob,
3662 chg => change type [MCRADT],
3663 file_a => original file name of a file (iff chg is 'C' or 'R')
3664 file_b => new/current file name of a file (any chg)
3665 }
3666 ;
3667
3668 # retval of read_url_paths{,_all}();
3669 $l_map = {
3670 # repository root url
3671 'https://svn.musicpd.org' => {
3672 # repository path # GIT_SVN_ID
3673 'mpd/trunk' => 'trunk',
3674 'mpd/tags/0.11.5' => 'tags/0.11.5',
3675 },
3676 }
3677
3678 Notes:
3679 I don't trust the each() function on unless I created %hash myself
3680 because the internal iterator may not have started at base.
Dscho's miscellaneous git branches