?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-svn: remember to check for clean indices on globbed refs, too
[alt-git.git] / git-svn.perl
blobace31021e745e36deadc19925bd6957530165a1e
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 $Git::SVN::Ra::_log_window_size = 100;
16
17 $Git::SVN::Log::TZ = $ENV{TZ};
18 $ENV{TZ} = 'UTC';
19 $| = 1; # unbuffer STDOUT
20
21 sub fatal (@) { print STDERR @_; exit 1 }
22 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
23 require SVN::Ra;
24 require SVN::Delta;
25 if ($SVN::Core::VERSION lt '1.1.0') {
26 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
27 }
28 push @Git::SVN::Ra::ISA, 'SVN::Ra';
29 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
30 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
31 use Carp qw/croak/;
32 use IO::File qw//;
33 use File::Basename qw/dirname basename/;
34 use File::Path qw/mkpath/;
35 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
36 use IPC::Open3;
37 use Git;
38
39 BEGIN {
40 my $s;
41 foreach (qw/command command_oneline command_noisy command_output_pipe
42 command_input_pipe command_close_pipe/) {
43 $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
44 "*Git::SVN::Migration::$_ = ".
45 "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
46 }
47 eval $s;
48 }
49
50 my ($SVN);
51
52 $sha1 = qr/[a-f\d]{40}/;
53 $sha1_short = qr/[a-f\d]{4,40}/;
54 my ($_stdin, $_help, $_edit,
55 $_message, $_file,
56 $_template, $_shared,
57 $_version, $_fetch_all,
58 $_merge, $_strategy, $_dry_run,
59 $_prefix);
60 $Git::SVN::_follow_parent = 1;
61 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
62 'config-dir=s' => \$Git::SVN::Ra::config_dir,
63 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
64 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
65 'authors-file|A=s' => \$_authors,
66 'repack:i' => \$Git::SVN::_repack,
67 'noMetadata' => \$Git::SVN::_no_metadata,
68 'useSvmProps' => \$Git::SVN::_use_svm_props,
69 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
70 'quiet|q' => \$_q,
71 'repack-flags|repack-args|repack-opts=s' =>
72 \$Git::SVN::_repack_flags,
73 %remote_opts );
74
75 my ($_trunk, $_tags, $_branches);
76 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
77 'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
78 'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
79 %remote_opts );
80 my %cmt_opts = ( 'edit|e' => \$_edit,
81 'rmdir' => \$SVN::Git::Editor::_rmdir,
82 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
83 'l=i' => \$SVN::Git::Editor::_rename_limit,
84 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
85 );
86
87 my %cmd = (
88 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
89 { 'revision|r=s' => \$_revision,
90 'all|a' => \$_fetch_all,
91 %fc_opts } ],
92 init => [ \&cmd_init, "Initialize a repo for tracking" .
93 " (requires URL argument)",
94 \%init_opts ],
95 'multi-init' => [ \&cmd_multi_init,
96 "Deprecated alias for ".
97 "'$0 init -T<trunk> -b<branches> -t<tags>'",
98 \%init_opts ],
99 dcommit => [ \&cmd_dcommit,
100 'Commit several diffs to merge with upstream',
101 { 'merge|m|M' => \$_merge,
102 'strategy|s=s' => \$_strategy,
103 'dry-run|n' => \$_dry_run,
104 %cmt_opts, %fc_opts } ],
105 'set-tree' => [ \&cmd_set_tree,
106 "Set an SVN repository to a git tree-ish",
107 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
108 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
109 { 'revision|r=i' => \$_revision } ],
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 next if /^multi-/; # don't show deprecated commands
186 # prints out arguments as they should be passed:
187 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
188 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
189 "--$_" : "-$_" }
190 split /\|/,$_)," $x\n";
191 }
192 }
193 print $fd <<"";
194 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
195 arbitrary identifier if you're tracking multiple SVN branches/repositories in
196 one git repository and want to keep them separate. See git-svn(1) for more
197 information.
198
199 exit $exit;
200 }
201
202 sub version {
203 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
204 exit 0;
205 }
206
207 sub do_git_init_db {
208 unless (-d $ENV{GIT_DIR}) {
209 my @init_db = ('init');
210 push @init_db, "--template=$_template" if defined $_template;
211 if (defined $_shared) {
212 if ($_shared =~ /[a-z]/) {
213 push @init_db, "--shared=$_shared";
214 } else {
215 push @init_db, "--shared";
216 }
217 }
218 command_noisy(@init_db);
219 }
220 }
221
222 sub init_subdir {
223 my $repo_path = shift or return;
224 mkpath([$repo_path]) unless -d $repo_path;
225 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
226 $ENV{GIT_DIR} = $repo_path . "/.git";
227 }
228
229 sub cmd_init {
230 if (defined $_trunk || defined $_branches || defined $_tags) {
231 return cmd_multi_init(@_);
232 }
233 my $url = shift or die "SVN repository location required ",
234 "as a command-line argument\n";
235 init_subdir(@_);
236 do_git_init_db();
237
238 Git::SVN->init($url);
239 }
240
241 sub cmd_fetch {
242 if (grep /^\d+=./, @_) {
243 die "'<rev>=<commit>' fetch arguments are ",
244 "no longer supported.\n";
245 }
246 my ($remote) = @_;
247 if (@_ > 1) {
248 die "Usage: $0 fetch [--all|-a] [svn-remote]\n";
249 }
250 $remote ||= $Git::SVN::default_repo_id;
251 if ($_fetch_all) {
252 cmd_multi_fetch();
253 } else {
254 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
255 }
256 }
257
258 sub cmd_set_tree {
259 my (@commits) = @_;
260 if ($_stdin || !@commits) {
261 print "Reading from stdin...\n";
262 @commits = ();
263 while (<STDIN>) {
264 if (/\b($sha1_short)\b/o) {
265 unshift @commits, $1;
266 }
267 }
268 }
269 my @revs;
270 foreach my $c (@commits) {
271 my @tmp = command('rev-parse',$c);
272 if (scalar @tmp == 1) {
273 push @revs, $tmp[0];
274 } elsif (scalar @tmp > 1) {
275 push @revs, reverse(command('rev-list',@tmp));
276 } else {
277 fatal "Failed to rev-parse $c\n";
278 }
279 }
280 my $gs = Git::SVN->new;
281 my ($r_last, $cmt_last) = $gs->last_rev_commit;
282 $gs->fetch;
283 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
284 fatal "There are new revisions that were fetched ",
285 "and need to be merged (or acknowledged) ",
286 "before committing.\nlast rev: $r_last\n",
287 " current: $gs->{last_rev}\n";
288 }
289 $gs->set_tree($_) foreach @revs;
290 print "Done committing ",scalar @revs," revisions to SVN\n";
291 }
292
293 sub cmd_dcommit {
294 my $head = shift;
295 $head ||= 'HEAD';
296 my ($url, $rev, $uuid);
297 my ($fh, $ctx) = command_output_pipe('rev-list', $head);
298 my @refs;
299 my $c;
300 while (<$fh>) {
301 $c = $_;
302 chomp $c;
303 ($url, $rev, $uuid) = cmt_metadata($c);
304 last if (defined $url && defined $rev && defined $uuid);
305 unshift @refs, $c;
306 }
307 close $fh; # most likely breaking the pipe
308 unless (defined $url && defined $rev && defined $uuid) {
309 die "Unable to determine upstream SVN information from ",
310 "$head history:\n $ctx\n";
311 }
312 my $gs = Git::SVN->find_by_url($url) or
313 die "Can't determine fetch information for $url\n";
314 my $last_rev;
315 foreach my $d (@refs) {
316 if (!verify_ref("$d~1")) {
317 fatal "Commit $d\n",
318 "has no parent commit, and therefore ",
319 "nothing to diff against.\n",
320 "You should be working from a repository ",
321 "originally created by git-svn\n";
322 }
323 unless (defined $last_rev) {
324 (undef, $last_rev, undef) = cmt_metadata("$d~1");
325 unless (defined $last_rev) {
326 fatal "Unable to extract revision information ",
327 "from commit $d~1\n";
328 }
329 }
330 if ($_dry_run) {
331 print "diff-tree $d~1 $d\n";
332 } else {
333 my %ed_opts = ( r => $last_rev,
334 log => get_commit_entry($d)->{log},
335 ra => Git::SVN::Ra->new($url),
336 tree_a => "$d~1",
337 tree_b => $d,
338 editor_cb => sub {
339 print "Committed r$_[0]\n";
340 $last_rev = $_[0]; },
341 svn_path => '');
342 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
343 print "No changes\n$d~1 == $d\n";
344 }
345 }
346 }
347 return if $_dry_run;
348 $gs->fetch;
349 # we always want to rebase against the current HEAD, not any
350 # head that was passed to us
351 my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
352 my @finish;
353 if (@diff) {
354 @finish = qw/rebase/;
355 push @finish, qw/--merge/ if $_merge;
356 push @finish, "--strategy=$_strategy" if $_strategy;
357 print STDERR "W: HEAD and ", $gs->refname, " differ, ",
358 "using @finish:\n", "@diff";
359 } else {
360 print "No changes between current HEAD and ",
361 $gs->refname, "\nResetting to the latest ",
362 $gs->refname, "\n";
363 @finish = qw/reset --mixed/;
364 }
365 command_noisy(@finish, $gs->refname);
366 }
367
368 sub cmd_show_ignore {
369 my $gs = Git::SVN->new;
370 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
371 $gs->traverse_ignore(\*STDOUT, '', $r);
372 }
373
374 sub cmd_multi_init {
375 my $url = shift;
376 unless (defined $_trunk || defined $_branches || defined $_tags) {
377 usage(1);
378 }
379 do_git_init_db();
380 $_prefix = '' unless defined $_prefix;
381 if (defined $url) {
382 $url =~ s#/+$##;
383 init_subdir(@_);
384 }
385 if (defined $_trunk) {
386 my $trunk_ref = $_prefix . 'trunk';
387 # try both old-style and new-style lookups:
388 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
389 unless ($gs_trunk) {
390 my ($trunk_url, $trunk_path) =
391 complete_svn_url($url, $_trunk);
392 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
393 undef, $trunk_ref);
394 }
395 }
396 return unless defined $_branches || defined $_tags;
397 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
398 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
399 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
400 }
401
402 sub cmd_multi_fetch {
403 my $remotes = Git::SVN::read_all_remotes();
404 foreach my $repo_id (sort keys %$remotes) {
405 if ($remotes->{$repo_id}->{url}) {
406 Git::SVN::fetch_all($repo_id, $remotes);
407 }
408 }
409 }
410
411 # this command is special because it requires no metadata
412 sub cmd_commit_diff {
413 my ($ta, $tb, $url) = @_;
414 my $usage = "Usage: $0 commit-diff -r<revision> ".
415 "<tree-ish> <tree-ish> [<URL>]\n";
416 fatal($usage) if (!defined $ta || !defined $tb);
417 my $svn_path;
418 if (!defined $url) {
419 my $gs = eval { Git::SVN->new };
420 if (!$gs) {
421 fatal("Needed URL or usable git-svn --id in ",
422 "the command-line\n", $usage);
423 }
424 $url = $gs->{url};
425 $svn_path = $gs->{path};
426 }
427 unless (defined $_revision) {
428 fatal("-r|--revision is a required argument\n", $usage);
429 }
430 if (defined $_message && defined $_file) {
431 fatal("Both --message/-m and --file/-F specified ",
432 "for the commit message.\n",
433 "I have no idea what you mean\n");
434 }
435 if (defined $_file) {
436 $_message = file_to_s($_file);
437 } else {
438 $_message ||= get_commit_entry($tb)->{log};
439 }
440 my $ra ||= Git::SVN::Ra->new($url);
441 $svn_path ||= $ra->{svn_path};
442 my $r = $_revision;
443 if ($r eq 'HEAD') {
444 $r = $ra->get_latest_revnum;
445 } elsif ($r !~ /^\d+$/) {
446 die "revision argument: $r not understood by git-svn\n";
447 }
448 my %ed_opts = ( r => $r,
449 log => $_message,
450 ra => $ra,
451 tree_a => $ta,
452 tree_b => $tb,
453 editor_cb => sub { print "Committed r$_[0]\n" },
454 svn_path => $svn_path );
455 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
456 print "No changes\n$ta == $tb\n";
457 }
458 }
459
460 ########################### utility functions #########################
461
462 sub complete_svn_url {
463 my ($url, $path) = @_;
464 $path =~ s#/+$##;
465 if ($path !~ m#^[a-z\+]+://#) {
466 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
467 fatal("E: '$path' is not a complete URL ",
468 "and a separate URL is not specified\n");
469 }
470 return ($url, $path);
471 }
472 return ($path, '');
473 }
474
475 sub complete_url_ls_init {
476 my ($ra, $repo_path, $switch, $pfx) = @_;
477 unless ($repo_path) {
478 print STDERR "W: $switch not specified\n";
479 return;
480 }
481 $repo_path =~ s#/+$##;
482 if ($repo_path =~ m#^[a-z\+]+://#) {
483 $ra = Git::SVN::Ra->new($repo_path);
484 $repo_path = '';
485 } else {
486 $repo_path =~ s#^/+##;
487 unless ($ra) {
488 fatal("E: '$repo_path' is not a complete URL ",
489 "and a separate URL is not specified\n");
490 }
491 }
492 my $url = $ra->{url};
493 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
494 my $k = "svn-remote.$gs->{repo_id}.url";
495 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
496 if ($orig_url && ($orig_url ne $gs->{url})) {
497 die "$k already set: $orig_url\n",
498 "wanted to set to: $gs->{url}\n";
499 }
500 command_oneline('config', $k, $gs->{url}) unless $orig_url;
501 my $remote_path = "$ra->{svn_path}/$repo_path/*";
502 $remote_path =~ s#/+#/#g;
503 $remote_path =~ s#^/##g;
504 my ($n) = ($switch =~ /^--(\w+)/);
505 if (length $pfx && $pfx !~ m#/$#) {
506 die "--prefix='$pfx' must have a trailing slash '/'\n";
507 }
508 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
509 "$remote_path:refs/remotes/$pfx*");
510 }
511
512 sub verify_ref {
513 my ($ref) = @_;
514 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
515 { STDERR => 0 }); };
516 }
517
518 sub get_tree_from_treeish {
519 my ($treeish) = @_;
520 # $treeish can be a symbolic ref, too:
521 my $type = command_oneline(qw/cat-file -t/, $treeish);
522 my $expected;
523 while ($type eq 'tag') {
524 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
525 }
526 if ($type eq 'commit') {
527 $expected = (grep /^tree /, command(qw/cat-file commit/,
528 $treeish))[0];
529 ($expected) = ($expected =~ /^tree ($sha1)$/o);
530 die "Unable to get tree from $treeish\n" unless $expected;
531 } elsif ($type eq 'tree') {
532 $expected = $treeish;
533 } else {
534 die "$treeish is a $type, expected tree, tag or commit\n";
535 }
536 return $expected;
537 }
538
539 sub get_commit_entry {
540 my ($treeish) = shift;
541 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
542 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
543 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
544 open my $log_fh, '>', $commit_editmsg or croak $!;
545
546 my $type = command_oneline(qw/cat-file -t/, $treeish);
547 if ($type eq 'commit' || $type eq 'tag') {
548 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
549 $type, $treeish);
550 my $in_msg = 0;
551 while (<$msg_fh>) {
552 if (!$in_msg) {
553 $in_msg = 1 if (/^\s*$/);
554 } elsif (/^git-svn-id: /) {
555 # skip this for now, we regenerate the
556 # correct one on re-fetch anyways
557 # TODO: set *:merge properties or like...
558 } else {
559 print $log_fh $_ or croak $!;
560 }
561 }
562 command_close_pipe($msg_fh, $ctx);
563 }
564 close $log_fh or croak $!;
565
566 if ($_edit || ($type eq 'tree')) {
567 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
568 # TODO: strip out spaces, comments, like git-commit.sh
569 system($editor, $commit_editmsg);
570 }
571 rename $commit_editmsg, $commit_msg or croak $!;
572 open $log_fh, '<', $commit_msg or croak $!;
573 { local $/; chomp($log_entry{log} = <$log_fh>); }
574 close $log_fh or croak $!;
575 unlink $commit_msg;
576 \%log_entry;
577 }
578
579 sub s_to_file {
580 my ($str, $file, $mode) = @_;
581 open my $fd,'>',$file or croak $!;
582 print $fd $str,"\n" or croak $!;
583 close $fd or croak $!;
584 chmod ($mode &~ umask, $file) if (defined $mode);
585 }
586
587 sub file_to_s {
588 my $file = shift;
589 open my $fd,'<',$file or croak "$!: file: $file\n";
590 local $/;
591 my $ret = <$fd>;
592 close $fd or croak $!;
593 $ret =~ s/\s*$//s;
594 return $ret;
595 }
596
597 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
598 sub load_authors {
599 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
600 my $log = $cmd eq 'log';
601 while (<$authors>) {
602 chomp;
603 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
604 my ($user, $name, $email) = ($1, $2, $3);
605 if ($log) {
606 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
607 } else {
608 $users{$user} = [$name, $email];
609 }
610 }
611 close $authors or croak $!;
612 }
613
614 # convert GetOpt::Long specs for use by git-config
615 sub read_repo_config {
616 return unless -d $ENV{GIT_DIR};
617 my $opts = shift;
618 my @config_only;
619 foreach my $o (keys %$opts) {
620 # if we have mixedCase and a long option-only, then
621 # it's a config-only variable that we don't need for
622 # the command-line.
623 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
624 my $v = $opts->{$o};
625 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
626 $key =~ s/-//g;
627 my $arg = 'git-config';
628 $arg .= ' --int' if ($o =~ /[:=]i$/);
629 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
630 if (ref $v eq 'ARRAY') {
631 chomp(my @tmp = `$arg --get-all svn.$key`);
632 @$v = @tmp if @tmp;
633 } else {
634 chomp(my $tmp = `$arg --get svn.$key`);
635 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
636 $$v = $tmp;
637 }
638 }
639 }
640 delete @$opts{@config_only} if @config_only;
641 }
642
643 sub extract_metadata {
644 my $id = shift or return (undef, undef, undef);
645 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
646 \s([a-f\d\-]+)$/x);
647 if (!defined $rev || !$uuid || !$url) {
648 # some of the original repositories I made had
649 # identifiers like this:
650 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
651 }
652 return ($url, $rev, $uuid);
653 }
654
655 sub cmt_metadata {
656 return extract_metadata((grep(/^git-svn-id: /,
657 command(qw/cat-file commit/, shift)))[-1]);
658 }
659
660 package Git::SVN;
661 use strict;
662 use warnings;
663 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
664 $_repack $_repack_flags $_use_svm_props/;
665 use Carp qw/croak/;
666 use File::Path qw/mkpath/;
667 use File::Copy qw/copy/;
668 use IPC::Open3;
669
670 my $_repack_nr;
671 # properties that we do not log:
672 my %SKIP_PROP;
673 BEGIN {
674 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
675 svn:special svn:executable
676 svn:entry:committed-rev
677 svn:entry:last-author
678 svn:entry:uuid
679 svn:entry:committed-date/;
680
681 # some options are read globally, but can be overridden locally
682 # per [svn-remote "..."] section. Command-line options will *NOT*
683 # override options set in an [svn-remote "..."] section
684 my $e;
685 foreach (qw/follow_parent no_metadata use_svm_props/) {
686 my $key = $_;
687 $key =~ tr/_//d;
688 $e .= "sub $_ {
689 my (\$self) = \@_;
690 return \$self->{-$_} if exists \$self->{-$_};
691 my \$k = \"svn-remote.\$self->{repo_id}\.$key\";
692 eval { command_oneline(qw/config --get/, \$k) };
693 if (\$@) {
694 \$self->{-$_} = \$Git::SVN::_$_;
695 } else {
696 my \$v = command_oneline(qw/config --bool/,\$k);
697 \$self->{-$_} = \$v eq 'false' ? 0 : 1;
698 }
699 return \$self->{-$_} }\n";
700 }
701 $e .= "1;\n";
702 eval $e or die $@;
703 }
704
705 my %LOCKFILES;
706 END { unlink keys %LOCKFILES if %LOCKFILES }
707
708 sub resolve_local_globs {
709 my ($url, $fetch, $glob_spec) = @_;
710 return unless defined $glob_spec;
711 my $ref = $glob_spec->{ref};
712 my $path = $glob_spec->{path};
713 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
714 next unless m#^refs/remotes/$ref->{regex}$#;
715 my $p = $1;
716 my $pathname = $path->full_path($p);
717 my $refname = $ref->full_path($p);
718 if (my $existing = $fetch->{$pathname}) {
719 if ($existing ne $refname) {
720 die "Refspec conflict:\n",
721 "existing: refs/remotes/$existing\n",
722 " globbed: refs/remotes/$refname\n";
723 }
724 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
725 $u =~ s!^\Q$url\E(/|$)!! or die
726 "refs/remotes/$refname: '$url' not found in '$u'\n";
727 if ($pathname ne $u) {
728 warn "W: Refspec glob conflict ",
729 "(ref: refs/remotes/$refname):\n",
730 "expected path: $pathname\n",
731 " real path: $u\n",
732 "Continuing ahead with $u\n";
733 next;
734 }
735 } else {
736 $fetch->{$pathname} = $refname;
737 }
738 }
739 }
740
741 sub parse_revision_argument {
742 my ($base, $head) = @_;
743 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
744 return ($base, $head);
745 }
746 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
747 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
748 return ($head, $head) if ($::_revision eq 'HEAD');
749 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
750 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
751 die "revision argument: $::_revision not understood by git-svn\n";
752 }
753
754 sub fetch_all {
755 my ($repo_id, $remotes) = @_;
756 my $remote = $remotes->{$repo_id};
757 my $fetch = $remote->{fetch};
758 my $url = $remote->{url};
759 my (@gs, @globs);
760 my $ra = Git::SVN::Ra->new($url);
761 my $uuid = $ra->get_uuid;
762 my $head = $ra->get_latest_revnum;
763 my $base = defined $fetch ? $head : 0;
764
765 # read the max revs for wildcard expansion (branches/*, tags/*)
766 foreach my $t (qw/branches tags/) {
767 defined $remote->{$t} or next;
768 push @globs, $remote->{$t};
769 my $max_rev = eval { tmp_config(qw/--int --get/,
770 "svn-remote.$repo_id.${t}-maxRev") };
771 if (defined $max_rev && ($max_rev < $base)) {
772 $base = $max_rev;
773 }
774 }
775
776 if ($fetch) {
777 foreach my $p (sort keys %$fetch) {
778 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
779 my $lr = $gs->rev_db_max;
780 if (defined $lr) {
781 $base = $lr if ($lr < $base);
782 }
783 push @gs, $gs;
784 }
785 }
786
787 ($base, $head) = parse_revision_argument($base, $head);
788 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
789 }
790
791 sub read_all_remotes {
792 my $r = {};
793 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
794 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
795 $r->{$1}->{fetch}->{$2} = $3;
796 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
797 $r->{$1}->{url} = $2;
798 } elsif (m!^(.+)\.(branches|tags)=
799 (.*):refs/remotes/(.+)\s*$/!x) {
800 my ($p, $g) = ($3, $4);
801 my $rs = $r->{$1}->{$2} = {
802 t => $2,
803 remote => $1,
804 path => Git::SVN::GlobSpec->new($p),
805 ref => Git::SVN::GlobSpec->new($g) };
806 if (length($rs->{ref}->{right}) != 0) {
807 die "The '*' glob character must be the last ",
808 "character of '$g'\n";
809 }
810 }
811 }
812 $r;
813 }
814
815 sub init_vars {
816 if (defined $_repack) {
817 $_repack = 1000 if ($_repack <= 0);
818 $_repack_nr = $_repack;
819 $_repack_flags ||= '-d';
820 }
821 }
822
823 sub verify_remotes_sanity {
824 return unless -d $ENV{GIT_DIR};
825 my %seen;
826 foreach (command(qw/config -l/)) {
827 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
828 if ($seen{$1}) {
829 die "Remote ref refs/remote/$1 is tracked by",
830 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
831 "Please resolve this ambiguity in ",
832 "your git configuration file before ",
833 "continuing\n";
834 }
835 $seen{$1} = $_;
836 }
837 }
838 }
839
840 # we allow more chars than remotes2config.sh...
841 sub sanitize_remote_name {
842 my ($name) = @_;
843 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
844 $name;
845 }
846
847 sub find_existing_remote {
848 my ($url, $remotes) = @_;
849 my $existing;
850 foreach my $repo_id (keys %$remotes) {
851 my $u = $remotes->{$repo_id}->{url} or next;
852 next if $u ne $url;
853 $existing = $repo_id;
854 last;
855 }
856 $existing;
857 }
858
859 sub init_remote_config {
860 my ($self, $url, $no_write) = @_;
861 $url =~ s!/+$!!; # strip trailing slash
862 my $r = read_all_remotes();
863 my $existing = find_existing_remote($url, $r);
864 if ($existing) {
865 unless ($no_write) {
866 print STDERR "Using existing ",
867 "[svn-remote \"$existing\"]\n";
868 }
869 $self->{repo_id} = $existing;
870 } else {
871 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
872 $existing = find_existing_remote($min_url, $r);
873 if ($existing) {
874 unless ($no_write) {
875 print STDERR "Using existing ",
876 "[svn-remote \"$existing\"]\n";
877 }
878 $self->{repo_id} = $existing;
879 }
880 if ($min_url ne $url) {
881 unless ($no_write) {
882 print STDERR "Using higher level of URL: ",
883 "$url => $min_url\n";
884 }
885 my $old_path = $self->{path};
886 $self->{path} = $url;
887 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
888 if (length $old_path) {
889 $self->{path} .= "/$old_path";
890 }
891 $url = $min_url;
892 }
893 }
894 my $orig_url;
895 if (!$existing) {
896 # verify that we aren't overwriting anything:
897 $orig_url = eval {
898 command_oneline('config', '--get',
899 "svn-remote.$self->{repo_id}.url")
900 };
901 if ($orig_url && ($orig_url ne $url)) {
902 die "svn-remote.$self->{repo_id}.url already set: ",
903 "$orig_url\nwanted to set to: $url\n";
904 }
905 }
906 my ($xrepo_id, $xpath) = find_ref($self->refname);
907 if (defined $xpath) {
908 die "svn-remote.$xrepo_id.fetch already set to track ",
909 "$xpath:refs/remotes/", $self->refname, "\n";
910 }
911 unless ($no_write) {
912 command_noisy('config',
913 "svn-remote.$self->{repo_id}.url", $url);
914 command_noisy('config', '--add',
915 "svn-remote.$self->{repo_id}.fetch",
916 "$self->{path}:".$self->refname);
917 }
918 $self->{url} = $url;
919 }
920
921 sub find_by_url { # repos_root and, path are optional
922 my ($class, $full_url, $repos_root, $path) = @_;
923 my $remotes = read_all_remotes();
924 if (defined $full_url && defined $repos_root && !defined $path) {
925 $path = $full_url;
926 $path =~ s#^\Q$repos_root\E(?:/|$)##;
927 }
928 foreach my $repo_id (keys %$remotes) {
929 my $u = $remotes->{$repo_id}->{url} or next;
930 next if defined $repos_root && $repos_root ne $u;
931
932 my $fetch = $remotes->{$repo_id}->{fetch} || {};
933 foreach (qw/branches tags/) {
934 resolve_local_globs($u, $fetch,
935 $remotes->{$repo_id}->{$_});
936 }
937 my $p = $path;
938 unless (defined $p) {
939 $p = $full_url;
940 $p =~ s#^\Q$u\E(?:/|$)## or next;
941 }
942 foreach my $f (keys %$fetch) {
943 next if $f ne $p;
944 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
945 }
946 }
947 undef;
948 }
949
950 sub init {
951 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
952 my $self = _new($class, $repo_id, $ref_id, $path);
953 if (defined $url) {
954 $self->init_remote_config($url, $no_write);
955 }
956 $self;
957 }
958
959 sub find_ref {
960 my ($ref_id) = @_;
961 foreach (command(qw/config -l/)) {
962 next unless m!^svn-remote\.(.+)\.fetch=
963 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
964 my ($repo_id, $path, $ref) = ($1, $2, $3);
965 if ($ref eq $ref_id) {
966 $path = '' if ($path =~ m#^\./?#);
967 return ($repo_id, $path);
968 }
969 }
970 (undef, undef, undef);
971 }
972
973 sub new {
974 my ($class, $ref_id, $repo_id, $path) = @_;
975 if (defined $ref_id && !defined $repo_id && !defined $path) {
976 ($repo_id, $path) = find_ref($ref_id);
977 if (!defined $repo_id) {
978 die "Could not find a \"svn-remote.*.fetch\" key ",
979 "in the repository configuration matching: ",
980 "refs/remotes/$ref_id\n";
981 }
982 }
983 my $self = _new($class, $repo_id, $ref_id, $path);
984 if (!defined $self->{path} || !length $self->{path}) {
985 my $fetch = command_oneline('config', '--get',
986 "svn-remote.$repo_id.fetch",
987 ":refs/remotes/$ref_id\$") or
988 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
989 "\":refs/remotes/$ref_id\$\" in config\n";
990 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
991 }
992 $self->{url} = command_oneline('config', '--get',
993 "svn-remote.$repo_id.url") or
994 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
995 if ((-z $self->db_path || ! -e $self->db_path) &&
996 ::verify_ref($self->refname.'^0')) {
997 $self->rebuild;
998 }
999 $self;
1000 }
1001
1002 sub refname { "refs/remotes/$_[0]->{ref_id}" }
1003
1004 sub svm_uuid {
1005 my ($self) = @_;
1006 return $self->{svm}->{uuid} if $self->svm;
1007 $self->ra;
1008 unless ($self->{svm}) {
1009 die "SVM UUID not cached, and reading remotely failed\n";
1010 }
1011 $self->{svm}->{uuid};
1012 }
1013
1014 sub svm {
1015 my ($self) = @_;
1016 return $self->{svm} if $self->{svm};
1017 my $svm;
1018 # see if we have it in our config, first:
1019 eval {
1020 my $section = "svn-remote.$self->{repo_id}";
1021 $svm = {
1022 source => tmp_config('--get', "$section.svm-source"),
1023 uuid => tmp_config('--get', "$section.svm-uuid"),
1024 }
1025 };
1026 $self->{svm} = $svm if ($svm && $svm->{source} && $svm->{uuid});
1027 $self->{svm};
1028 }
1029
1030 sub _set_svm_vars {
1031 my ($self, $ra) = @_;
1032 return $ra if $self->svm;
1033
1034 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1035 "(svm:source, svm:mirror, svm:mirror) ",
1036 "from the following URLs:\n" );
1037 sub read_svm_props {
1038 my ($self, $props) = @_;
1039 my $src = $props->{'svm:source'};
1040 my $mirror = $props->{'svm:mirror'};
1041 my $uuid = $props->{'svm:uuid'};
1042 return undef if (!$src || !$mirror || !$uuid);
1043
1044 chomp($src, $mirror, $uuid);
1045
1046 $uuid =~ m{^[0-9a-f\-]{30,}$}
1047 or die "doesn't look right - svm:uuid is '$uuid'\n";
1048 # don't know what a '!' is there for, also the
1049 # username is of no interest
1050 $src =~ s{/?!$}{$mirror};
1051 $src =~ s{/+$}{}; # no trailing slashes please
1052 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1053
1054 my $section = "svn-remote.$self->{repo_id}";
1055 tmp_config('--add', "$section.svm-source", $src);
1056 tmp_config('--add', "$section.svm-uuid", $uuid);
1057 $self->{svm} = { source => $src , uuid => $uuid };
1058 return 1;
1059 }
1060
1061 my $r = $ra->get_latest_revnum;
1062 my $path = $self->{path};
1063 my @tried_a = ($path);
1064 while (length $path) {
1065 if ($self->read_svm_props(($ra->get_dir($path, $r))[2])) {
1066 return $ra;
1067 }
1068 $path =~ s#/?[^/]+$## && push @tried_a, $path;
1069 }
1070 if ($self->read_svm_props(($ra->get_dir('', $r))[2])) {
1071 return $ra;
1072 }
1073
1074 if ($ra->{repos_root} eq $self->{url}) {
1075 die @err, map { " $self->{url}/$_\n" } @tried_a, "\n";
1076 }
1077
1078 # nope, make sure we're connected to the repository root:
1079 my $ok;
1080 my @tried_b;
1081 $path = $ra->{svn_path};
1082 $path =~ s#/?[^/]+$##; # we already tried this one above
1083 $ra = Git::SVN::Ra->new($ra->{repos_root});
1084 while (length $path) {
1085 $ok = $self->read_svm_props(($ra->get_dir($path, $r))[2]);
1086 last if $ok;
1087 $path =~ s#/?[^/]+$## && push @tried_b, $path;
1088 }
1089 $ok = $self->read_svm_props(($ra->get_dir('', $r))[2]) unless $ok;
1090 if (!$ok) {
1091 die @err, map { " $self->{url}/$_\n" } @tried_a, "\n",
1092 map { " $ra->{url}/$_\n" } @tried_b, "\n"
1093 }
1094 Git::SVN::Ra->new($self->{url});
1095 }
1096
1097 # this allows us to memoize our SVN::Ra UUID locally and avoid a
1098 # remote lookup (useful for 'git svn log').
1099 sub ra_uuid {
1100 my ($self) = @_;
1101 unless ($self->{ra_uuid}) {
1102 my $key = "svn-remote.$self->{repo_id}.uuid";
1103 my $uuid = eval { tmp_config('--get', $key) };
1104 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1105 $self->{ra_uuid} = $uuid;
1106 } else {
1107 die "ra_uuid called without URL\n" unless $self->{url};
1108 $self->{ra_uuid} = $self->ra->get_uuid;
1109 tmp_config('--add', $key, $self->{ra_uuid});
1110 }
1111 }
1112 $self->{ra_uuid};
1113 }
1114
1115 sub ra {
1116 my ($self) = shift;
1117 my $ra = Git::SVN::Ra->new($self->{url});
1118 if ($self->use_svm_props && !$self->{svm}) {
1119 if ($self->no_metadata) {
1120 die "Can't have both 'noMetadata' and ",
1121 "'useSvmProps' options set!\n";
1122 }
1123 $ra = $self->_set_svm_vars($ra);
1124 $self->{-want_revprops} = 1;
1125 }
1126 $ra;
1127 }
1128
1129 sub rel_path {
1130 my ($self) = @_;
1131 my $repos_root = $self->ra->{repos_root};
1132 return $self->{path} if ($self->{url} eq $repos_root);
1133 die "BUG: rel_path failed! repos_root: $repos_root, Ra URL: ",
1134 $self->ra->{url}, " path: $self->{path}, URL: $self->{url}\n";
1135 }
1136
1137 sub traverse_ignore {
1138 my ($self, $fh, $path, $r) = @_;
1139 $path =~ s#^/+##g;
1140 my $ra = $self->ra;
1141 my ($dirent, undef, $props) = $ra->get_dir($path, $r);
1142 my $p = $path;
1143 $p =~ s#^\Q$ra->{svn_path}\E/##;
1144 print $fh length $p ? "\n# $p\n" : "\n# /\n";
1145 if (my $s = $props->{'svn:ignore'}) {
1146 $s =~ s/[\r\n]+/\n/g;
1147 chomp $s;
1148 if (length $p == 0) {
1149 $s =~ s#\n#\n/$p#g;
1150 print $fh "/$s\n";
1151 } else {
1152 $s =~ s#\n#\n/$p/#g;
1153 print $fh "/$p/$s\n";
1154 }
1155 }
1156 foreach (sort keys %$dirent) {
1157 next if $dirent->{$_}->kind != $SVN::Node::dir;
1158 $self->traverse_ignore($fh, "$path/$_", $r);
1159 }
1160 }
1161
1162 sub last_rev { ($_[0]->last_rev_commit)[0] }
1163 sub last_commit { ($_[0]->last_rev_commit)[1] }
1164
1165 # returns the newest SVN revision number and newest commit SHA1
1166 sub last_rev_commit {
1167 my ($self) = @_;
1168 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1169 return ($self->{last_rev}, $self->{last_commit});
1170 }
1171 my $c = ::verify_ref($self->refname.'^0');
1172 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1173 my $rev = (::cmt_metadata($c))[1];
1174 if (defined $rev) {
1175 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1176 return ($rev, $c);
1177 }
1178 }
1179 my $db_path = $self->db_path;
1180 unless (-e $db_path) {
1181 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1182 return (undef, undef);
1183 }
1184 my $offset = -41; # from tail
1185 my $rl;
1186 open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1187 sysseek($fh, $offset, 2); # don't care for errors
1188 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1189 chomp $rl;
1190 while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1191 $offset -= 41;
1192 sysseek($fh, $offset, 2); # don't care for errors
1193 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1194 chomp $rl;
1195 }
1196 if ($c && $c ne $rl) {
1197 die "$db_path and ", $self->refname,
1198 " inconsistent!:\n$c != $rl\n";
1199 }
1200 my $rev = sysseek($fh, 0, 1) or croak $!;
1201 $rev = ($rev - 41) / 41;
1202 close $fh or croak $!;
1203 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1204 return ($rev, $c);
1205 }
1206
1207 sub get_fetch_range {
1208 my ($self, $min, $max) = @_;
1209 $max ||= $self->ra->get_latest_revnum;
1210 $min ||= $self->rev_db_max;
1211 (++$min, $max);
1212 }
1213
1214 sub tmp_config {
1215 my (@args) = @_;
1216 my $config = "$ENV{GIT_DIR}/svn/config";
1217 my $old_config = $ENV{GIT_CONFIG};
1218 $ENV{GIT_CONFIG} = $config;
1219 $@ = undef;
1220 my @ret = eval {
1221 unless (-f $config) {
1222 mkfile($config);
1223 open my $fh, '>', $config or
1224 die "Can't open $config: $!\n";
1225 print $fh "; This file is used internally by ",
1226 "git-svn\n" or die
1227 "Couldn't write to $config: $!\n";
1228 print $fh "; You should not have to edit it\n" or
1229 die "Couldn't write to $config: $!\n";
1230 close $fh or die "Couldn't close $config: $!\n";
1231 }
1232 command('config', @args);
1233 };
1234 my $err = $@;
1235 if (defined $old_config) {
1236 $ENV{GIT_CONFIG} = $old_config;
1237 } else {
1238 delete $ENV{GIT_CONFIG};
1239 }
1240 die $err if $err;
1241 wantarray ? @ret : $ret[0];
1242 }
1243
1244 sub tmp_index_do {
1245 my ($self, $sub) = @_;
1246 my $old_index = $ENV{GIT_INDEX_FILE};
1247 $ENV{GIT_INDEX_FILE} = $self->{index};
1248 $@ = undef;
1249 my @ret = eval {
1250 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1251 mkpath([$dir]) unless -d $dir;
1252 &$sub;
1253 };
1254 my $err = $@;
1255 if (defined $old_index) {
1256 $ENV{GIT_INDEX_FILE} = $old_index;
1257 } else {
1258 delete $ENV{GIT_INDEX_FILE};
1259 }
1260 die $err if $err;
1261 wantarray ? @ret : $ret[0];
1262 }
1263
1264 sub assert_index_clean {
1265 my ($self, $treeish) = @_;
1266
1267 $self->tmp_index_do(sub {
1268 command_noisy('read-tree', $treeish) unless -e $self->{index};
1269 my $x = command_oneline('write-tree');
1270 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1271 /^tree ($::sha1)/mo);
1272 return if $y eq $x;
1273
1274 warn "Index mismatch: $y != $x\nrereading $treeish\n";
1275 unlink $self->{index} or die "unlink $self->{index}: $!\n";
1276 command_noisy('read-tree', $treeish);
1277 $x = command_oneline('write-tree');
1278 if ($y ne $x) {
1279 ::fatal "trees ($treeish) $y != $x\n",
1280 "Something is seriously wrong...\n";
1281 }
1282 });
1283 }
1284
1285 sub get_commit_parents {
1286 my ($self, $log_entry) = @_;
1287 my (%seen, @ret, @tmp);
1288 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1289 if (my $ip = $self->{inject_parents}) {
1290 if (my $commit = delete $ip->{$log_entry->{revision}}) {
1291 push @tmp, $commit;
1292 }
1293 }
1294 if (my $cur = ::verify_ref($self->refname.'^0')) {
1295 push @tmp, $cur;
1296 }
1297 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1298 while (my $p = shift @tmp) {
1299 next if $seen{$p};
1300 $seen{$p} = 1;
1301 push @ret, $p;
1302 # MAXPARENT is defined to 16 in commit-tree.c:
1303 last if @ret >= 16;
1304 }
1305 if (@tmp) {
1306 die "r$log_entry->{revision}: No room for parents:\n\t",
1307 join("\n\t", @tmp), "\n";
1308 }
1309 @ret;
1310 }
1311
1312 sub full_url {
1313 my ($self) = @_;
1314 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1315 }
1316
1317 sub do_git_commit {
1318 my ($self, $log_entry) = @_;
1319 my $lr = $self->last_rev;
1320 if (defined $lr && $lr >= $log_entry->{revision}) {
1321 die "Last fetched revision of ", $self->refname,
1322 " was r$lr, but we are about to fetch: ",
1323 "r$log_entry->{revision}!\n";
1324 }
1325 if (my $c = $self->rev_db_get($log_entry->{revision})) {
1326 croak "$log_entry->{revision} = $c already exists! ",
1327 "Why are we refetching it?\n";
1328 }
1329 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1330 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1331 $log_entry->{email};
1332 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1333
1334 my $tree = $log_entry->{tree};
1335 if (!defined $tree) {
1336 $tree = $self->tmp_index_do(sub {
1337 command_oneline('write-tree') });
1338 }
1339 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1340
1341 my @exec = ('git-commit-tree', $tree);
1342 foreach ($self->get_commit_parents($log_entry)) {
1343 push @exec, '-p', $_;
1344 }
1345 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1346 or croak $!;
1347 print $msg_fh $log_entry->{log} or croak $!;
1348 unless ($self->no_metadata) {
1349 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1350 or croak $!;
1351 }
1352 $msg_fh->flush == 0 or croak $!;
1353 close $msg_fh or croak $!;
1354 chomp(my $commit = do { local $/; <$out_fh> });
1355 close $out_fh or croak $!;
1356 waitpid $pid, 0;
1357 croak $? if $?;
1358 if ($commit !~ /^$::sha1$/o) {
1359 die "Failed to commit, invalid sha1: $commit\n";
1360 }
1361
1362 $self->rev_db_set($log_entry->{revision}, $commit, 1);
1363
1364 $self->{last_rev} = $log_entry->{revision};
1365 $self->{last_commit} = $commit;
1366 print "r$log_entry->{revision}";
1367 if (defined $log_entry->{svm_revision}) {
1368 print " (\@$log_entry->{svm_revision})";
1369 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1370 0, $self->svm_uuid);
1371 }
1372 print " = $commit ($self->{ref_id})\n";
1373 if (defined $_repack && (--$_repack_nr == 0)) {
1374 $_repack_nr = $_repack;
1375 # repack doesn't use any arguments with spaces in them, does it?
1376 print "Running git repack $_repack_flags ...\n";
1377 command_noisy('repack', split(/\s+/, $_repack_flags));
1378 print "Done repacking\n";
1379 }
1380 return $commit;
1381 }
1382
1383 sub match_paths {
1384 my ($self, $paths, $r) = @_;
1385 return 1 if $self->{path} eq '';
1386 if (my $path = $paths->{"/$self->{path}"}) {
1387 return ($path->{action} eq 'D') ? 0 : 1;
1388 }
1389 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1390 if (grep /$self->{path_regex}/, keys %$paths) {
1391 return 1;
1392 }
1393 my $c = '';
1394 foreach (split m#/#, $self->{path}) {
1395 $c .= "/$_";
1396 next unless ($paths->{$c} &&
1397 ($paths->{$c}->{action} =~ /^[AR]$/));
1398 if ($self->ra->check_path($self->{path}, $r) ==
1399 $SVN::Node::dir) {
1400 return 1;
1401 }
1402 }
1403 return 0;
1404 }
1405
1406 sub find_parent_branch {
1407 my ($self, $paths, $rev) = @_;
1408 return undef unless $self->follow_parent;
1409 unless (defined $paths) {
1410 my $err_handler = $SVN::Error::handler;
1411 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1412 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1413 $paths =
1414 Git::SVN::Ra::dup_changed_paths($_[0]) });
1415 $SVN::Error::handler = $err_handler;
1416 }
1417 return undef unless defined $paths;
1418
1419 # look for a parent from another branch:
1420 my @b_path_components = split m#/#, $self->rel_path;
1421 my @a_path_components;
1422 my $i;
1423 while (@b_path_components) {
1424 $i = $paths->{'/'.join('/', @b_path_components)};
1425 last if $i && defined $i->{copyfrom_path};
1426 unshift(@a_path_components, pop(@b_path_components));
1427 }
1428 return undef unless defined $i && defined $i->{copyfrom_path};
1429 my $branch_from = $i->{copyfrom_path};
1430 if (@a_path_components) {
1431 print STDERR "branch_from: $branch_from => ";
1432 $branch_from .= '/'.join('/', @a_path_components);
1433 print STDERR $branch_from, "\n";
1434 }
1435 my $r = $i->{copyfrom_rev};
1436 my $repos_root = $self->ra->{repos_root};
1437 my $url = $self->ra->{url};
1438 my $new_url = $repos_root . $branch_from;
1439 print STDERR "Found possible branch point: ",
1440 "$new_url => ", $self->full_url, ", $r\n";
1441 $branch_from =~ s#^/##;
1442 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
1443 unless ($gs) {
1444 my $ref_id = $self->{ref_id};
1445 $ref_id =~ s/\@\d+$//;
1446 $ref_id .= "\@$r";
1447 # just grow a tail if we're not unique enough :x
1448 $ref_id .= '-' while find_ref($ref_id);
1449 print STDERR "Initializing parent: $ref_id\n";
1450 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1451 }
1452 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1453 if (!defined $r0 || !defined $parent) {
1454 $gs->fetch(0, $r);
1455 ($r0, $parent) = $gs->last_rev_commit;
1456 }
1457 if (defined $r0 && defined $parent) {
1458 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1459 $self->assert_index_clean($parent);
1460 my $ed;
1461 if ($self->ra->can_do_switch) {
1462 print STDERR "Following parent with do_switch\n";
1463 # do_switch works with svn/trunk >= r22312, but that
1464 # is not included with SVN 1.4.3 (the latest version
1465 # at the moment), so we can't rely on it
1466 $self->{last_commit} = $parent;
1467 $ed = SVN::Git::Fetcher->new($self);
1468 $gs->ra->gs_do_switch($r0, $rev, $gs,
1469 $self->full_url, $ed)
1470 or die "SVN connection failed somewhere...\n";
1471 } else {
1472 print STDERR "Following parent with do_update\n";
1473 $ed = SVN::Git::Fetcher->new($self);
1474 $self->ra->gs_do_update($rev, $rev, $self, $ed)
1475 or die "SVN connection failed somewhere...\n";
1476 }
1477 print STDERR "Successfully followed parent\n";
1478 return $self->make_log_entry($rev, [$parent], $ed);
1479 }
1480 return undef;
1481 }
1482
1483 sub do_fetch {
1484 my ($self, $paths, $rev) = @_;
1485 my $ed;
1486 my ($last_rev, @parents);
1487 if (my $lc = $self->last_commit) {
1488 # we can have a branch that was deleted, then re-added
1489 # under the same name but copied from another path, in
1490 # which case we'll have multiple parents (we don't
1491 # want to break the original ref, nor lose copypath info):
1492 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1493 push @{$log_entry->{parents}}, $lc;
1494 return $log_entry;
1495 }
1496 $ed = SVN::Git::Fetcher->new($self);
1497 $last_rev = $self->{last_rev};
1498 $ed->{c} = $lc;
1499 @parents = ($lc);
1500 } else {
1501 $last_rev = $rev;
1502 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1503 return $log_entry;
1504 }
1505 $ed = SVN::Git::Fetcher->new($self);
1506 }
1507 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1508 die "SVN connection failed somewhere...\n";
1509 }
1510 $self->make_log_entry($rev, \@parents, $ed);
1511 }
1512
1513 sub get_untracked {
1514 my ($self, $ed) = @_;
1515 my @out;
1516 my $h = $ed->{empty};
1517 foreach (sort keys %$h) {
1518 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1519 push @out, " $act: " . uri_encode($_);
1520 warn "W: $act: $_\n";
1521 }
1522 foreach my $t (qw/dir_prop file_prop/) {
1523 $h = $ed->{$t} or next;
1524 foreach my $path (sort keys %$h) {
1525 my $ppath = $path eq '' ? '.' : $path;
1526 foreach my $prop (sort keys %{$h->{$path}}) {
1527 next if $SKIP_PROP{$prop};
1528 my $v = $h->{$path}->{$prop};
1529 my $t_ppath_prop = "$t: " .
1530 uri_encode($ppath) . ' ' .
1531 uri_encode($prop);
1532 if (defined $v) {
1533 push @out, " +$t_ppath_prop " .
1534 uri_encode($v);
1535 } else {
1536 push @out, " -$t_ppath_prop";
1537 }
1538 }
1539 }
1540 }
1541 foreach my $t (qw/absent_file absent_directory/) {
1542 $h = $ed->{$t} or next;
1543 foreach my $parent (sort keys %$h) {
1544 foreach my $path (sort @{$h->{$parent}}) {
1545 push @out, " $t: " .
1546 uri_encode("$parent/$path");
1547 warn "W: $t: $parent/$path ",
1548 "Insufficient permissions?\n";
1549 }
1550 }
1551 }
1552 \@out;
1553 }
1554
1555 sub parse_svn_date {
1556 my $date = shift || return '+0000 1970-01-01 00:00:00';
1557 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1558 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1559 croak "Unable to parse date: $date\n";
1560 "+0000 $Y-$m-$d $H:$M:$S";
1561 }
1562
1563 sub check_author {
1564 my ($author) = @_;
1565 if (!defined $author || length $author == 0) {
1566 $author = '(no author)';
1567 }
1568 if (defined $::_authors && ! defined $::users{$author}) {
1569 die "Author: $author not defined in $::_authors file\n";
1570 }
1571 $author;
1572 }
1573
1574 sub make_log_entry {
1575 my ($self, $rev, $parents, $ed) = @_;
1576 my $untracked = $self->get_untracked($ed);
1577
1578 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1579 print $un "r$rev\n" or croak $!;
1580 print $un $_, "\n" foreach @$untracked;
1581 my %log_entry = ( parents => $parents || [], revision => $rev,
1582 log => '');
1583
1584 my $headrev;
1585 my $logged = delete $self->{logged_rev_props};
1586 if (!$logged || $self->{-want_revprops}) {
1587 my $rp = $self->ra->rev_proplist($rev);
1588 foreach (sort keys %$rp) {
1589 my $v = $rp->{$_};
1590 if (/^svn:(author|date|log)$/) {
1591 $log_entry{$1} = $v;
1592 } elsif ($_ eq 'svm:headrev') {
1593 $headrev = $v;
1594 } else {
1595 print $un " rev_prop: ", uri_encode($_), ' ',
1596 uri_encode($v), "\n";
1597 }
1598 }
1599 } else {
1600 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1601 }
1602 close $un or croak $!;
1603
1604 $log_entry{date} = parse_svn_date($log_entry{date});
1605 $log_entry{log} .= "\n";
1606 my $author = $log_entry{author} = check_author($log_entry{author});
1607 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1608 : ($author, undef);
1609 if (defined $headrev && $self->use_svm_props) {
1610 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1611 if ($uuid ne $self->{svm}->{uuid}) {
1612 die "UUID mismatch on SVM path:\n",
1613 "expected: $self->{svm}->{uuid}\n",
1614 " got: $uuid\n";
1615 }
1616 my $full_url = $self->{svm}->{source};
1617 $full_url .= "/$self->{path}" if length $self->{path};
1618 $log_entry{metadata} = "$full_url\@$r $uuid";
1619 $log_entry{svm_revision} = $r;
1620 $email ||= "$author\@$uuid"
1621 } else {
1622 $log_entry{metadata} = $self->full_url . "\@$rev " .
1623 $self->ra->get_uuid;
1624 $email ||= "$author\@" . $self->ra->get_uuid;
1625 }
1626 $log_entry{name} = $name;
1627 $log_entry{email} = $email;
1628 \%log_entry;
1629 }
1630
1631 sub fetch {
1632 my ($self, $min_rev, $max_rev, @parents) = @_;
1633 my ($last_rev, $last_commit) = $self->last_rev_commit;
1634 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1635 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1636 }
1637
1638 sub set_tree_cb {
1639 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1640 $self->{inject_parents} = { $rev => $tree };
1641 $self->fetch(undef, undef);
1642 }
1643
1644 sub set_tree {
1645 my ($self, $tree) = (shift, shift);
1646 my $log_entry = ::get_commit_entry($tree);
1647 unless ($self->{last_rev}) {
1648 fatal("Must have an existing revision to commit\n");
1649 }
1650 my %ed_opts = ( r => $self->{last_rev},
1651 log => $log_entry->{log},
1652 ra => $self->ra,
1653 tree_a => $self->{last_commit},
1654 tree_b => $tree,
1655 editor_cb => sub {
1656 $self->set_tree_cb($log_entry, $tree, @_) },
1657 svn_path => $self->{path} );
1658 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1659 print "No changes\nr$self->{last_rev} = $tree\n";
1660 }
1661 }
1662
1663 sub rebuild {
1664 my ($self) = @_;
1665 my $db_path = $self->db_path;
1666 if (-f $self->{db_root}) {
1667 rename $self->{db_root}, $db_path or die
1668 "rename $self->{db_root} => $db_path failed: $!\n";
1669 my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
1670 symlink $base, $self->{db_root} or die
1671 "symlink $base => $self->{db_root} failed: $!\n";
1672 return;
1673 }
1674 print "Rebuilding $db_path ...\n";
1675 my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1676 my $latest;
1677 my $full_url = $self->full_url;
1678 my $svn_uuid;
1679 while (<$rev_list>) {
1680 chomp;
1681 my $c = $_;
1682 die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1683 my ($url, $rev, $uuid) = ::cmt_metadata($c);
1684
1685 # ignore merges (from set-tree)
1686 next if (!defined $rev || !$uuid);
1687
1688 # if we merged or otherwise started elsewhere, this is
1689 # how we break out of it
1690 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1691 ($full_url && $url && ($url ne $full_url))) {
1692 next;
1693 }
1694 $latest ||= $rev;
1695 $svn_uuid ||= $uuid;
1696
1697 $self->rev_db_set($rev, $c);
1698 print "r$rev = $c\n";
1699 }
1700 command_close_pipe($rev_list, $ctx);
1701 print "Done rebuilding $db_path\n";
1702 }
1703
1704 # rev_db:
1705 # Tie::File seems to be prone to offset errors if revisions get sparse,
1706 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
1707 # one of my favorite modules is out :< Next up would be one of the DBM
1708 # modules, but I'm not sure which is most portable... So I'll just
1709 # go with something that's plain-text, but still capable of
1710 # being randomly accessed. So here's my ultra-simple fixed-width
1711 # database. All records are 40 characters + "\n", so it's easy to seek
1712 # to a revision: (41 * rev) is the byte offset.
1713 # A record of 40 0s denotes an empty revision.
1714 # And yes, it's still pretty fast (faster than Tie::File).
1715 # These files are disposable unless noMetadata or useSvmProps is set
1716
1717 sub _rev_db_set {
1718 my ($fh, $rev, $commit) = @_;
1719 my $offset = $rev * 41;
1720 # assume that append is the common case:
1721 seek $fh, 0, 2 or croak $!;
1722 my $pos = tell $fh;
1723 if ($pos < $offset) {
1724 for (1 .. (($offset - $pos) / 41)) {
1725 print $fh (('0' x 40),"\n") or croak $!;
1726 }
1727 }
1728 seek $fh, $offset, 0 or croak $!;
1729 print $fh $commit,"\n" or croak $!;
1730 }
1731
1732 sub mkfile {
1733 my ($path) = @_;
1734 unless (-e $path) {
1735 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
1736 mkpath([$dir]) unless -d $dir;
1737 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
1738 close $fh or die "Couldn't close (create) $path: $!\n";
1739 }
1740 }
1741
1742 sub rev_db_set {
1743 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
1744 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
1745 my $db = $self->db_path($uuid);
1746 my $db_lock = "$db.lock";
1747 my $sig;
1748 if ($update_ref) {
1749 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1750 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
1751 }
1752 mkfile($db);
1753
1754 $LOCKFILES{$db_lock} = 1;
1755 my $sync;
1756 # both of these options make our .rev_db file very, very important
1757 # and we can't afford to lose it because rebuild() won't work
1758 if ($self->use_svm_props || $self->no_metadata) {
1759 $sync = 1;
1760 copy($db, $db_lock) or die "rev_db_set(@_): ",
1761 "Failed to copy: ",
1762 "$db => $db_lock ($!)\n";
1763 } else {
1764 rename $db, $db_lock or die "rev_db_set(@_): ",
1765 "Failed to rename: ",
1766 "$db => $db_lock ($!)\n";
1767 }
1768 open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
1769 _rev_db_set($fh, $rev, $commit);
1770 if ($sync) {
1771 $fh->flush or die "Couldn't flush $db_lock: $!\n";
1772 $fh->sync or die "Couldn't sync $db_lock: $!\n";
1773 }
1774 close $fh or croak $!;
1775 if ($update_ref) {
1776 command_noisy('update-ref', '-m', "r$rev",
1777 $self->refname, $commit);
1778 }
1779 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
1780 "$db_lock => $db ($!)\n";
1781 delete $LOCKFILES{$db_lock};
1782 if ($update_ref) {
1783 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1784 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
1785 kill $sig, $$ if defined $sig;
1786 }
1787 }
1788
1789 sub rev_db_max {
1790 my ($self) = @_;
1791 my $db_path = $self->db_path;
1792 my @stat = stat $db_path or return 0;
1793 ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
1794 my $max = $stat[7] / 41;
1795 (($max > 0) ? $max - 1 : 0);
1796 }
1797
1798 sub rev_db_get {
1799 my ($self, $rev, $uuid) = @_;
1800 my $ret;
1801 my $offset = $rev * 41;
1802 my $db_path = $self->db_path($uuid);
1803 return undef unless -e $db_path;
1804 open my $fh, '<', $db_path or croak $!;
1805 if (sysseek($fh, $offset, 0) == $offset) {
1806 my $read = sysread($fh, $ret, 40);
1807 $ret = undef if ($read != 40 || $ret eq ('0'x40));
1808 }
1809 close $fh or croak $!;
1810 $ret;
1811 }
1812
1813 sub find_rev_before {
1814 my ($self, $rev, $eq_ok) = @_;
1815 --$rev unless $eq_ok;
1816 while ($rev > 0) {
1817 if (my $c = $self->rev_db_get($rev)) {
1818 return ($rev, $c);
1819 }
1820 --$rev;
1821 }
1822 return (undef, undef);
1823 }
1824
1825 sub _new {
1826 my ($class, $repo_id, $ref_id, $path) = @_;
1827 unless (defined $repo_id && length $repo_id) {
1828 $repo_id = $Git::SVN::default_repo_id;
1829 }
1830 unless (defined $ref_id && length $ref_id) {
1831 $_[2] = $ref_id = $Git::SVN::default_ref_id;
1832 }
1833 $_[1] = $repo_id = sanitize_remote_name($repo_id);
1834 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1835 $_[3] = $path = '' unless (defined $path);
1836 mkpath(["$ENV{GIT_DIR}/svn"]);
1837 bless {
1838 ref_id => $ref_id, dir => $dir, index => "$dir/index",
1839 path => $path, config => "$ENV{GIT_DIR}/svn/config",
1840 db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
1841 }
1842
1843 sub db_path {
1844 my ($self, $uuid) = @_;
1845 $uuid ||= $self->ra_uuid;
1846 "$self->{db_root}.$uuid";
1847 }
1848
1849 sub uri_encode {
1850 my ($f) = @_;
1851 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1852 $f
1853 }
1854
1855 package Git::SVN::Prompt;
1856 use strict;
1857 use warnings;
1858 require SVN::Core;
1859 use vars qw/$_no_auth_cache $_username/;
1860
1861 sub simple {
1862 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1863 $may_save = undef if $_no_auth_cache;
1864 $default_username = $_username if defined $_username;
1865 if (defined $default_username && length $default_username) {
1866 if (defined $realm && length $realm) {
1867 print STDERR "Authentication realm: $realm\n";
1868 STDERR->flush;
1869 }
1870 $cred->username($default_username);
1871 } else {
1872 username($cred, $realm, $may_save, $pool);
1873 }
1874 $cred->password(_read_password("Password for '" .
1875 $cred->username . "': ", $realm));
1876 $cred->may_save($may_save);
1877 $SVN::_Core::SVN_NO_ERROR;
1878 }
1879
1880 sub ssl_server_trust {
1881 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1882 $may_save = undef if $_no_auth_cache;
1883 print STDERR "Error validating server certificate for '$realm':\n";
1884 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1885 print STDERR " - The certificate is not issued by a trusted ",
1886 "authority. Use the\n",
1887 " fingerprint to validate the certificate manually!\n";
1888 }
1889 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1890 print STDERR " - The certificate hostname does not match.\n";
1891 }
1892 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1893 print STDERR " - The certificate is not yet valid.\n";
1894 }
1895 if ($failures & $SVN::Auth::SSL::EXPIRED) {
1896 print STDERR " - The certificate has expired.\n";
1897 }
1898 if ($failures & $SVN::Auth::SSL::OTHER) {
1899 print STDERR " - The certificate has an unknown error.\n";
1900 }
1901 printf STDERR
1902 "Certificate information:\n".
1903 " - Hostname: %s\n".
1904 " - Valid: from %s until %s\n".
1905 " - Issuer: %s\n".
1906 " - Fingerprint: %s\n",
1907 map $cert_info->$_, qw(hostname valid_from valid_until
1908 issuer_dname fingerprint);
1909 my $choice;
1910 prompt:
1911 print STDERR $may_save ?
1912 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1913 "(R)eject or accept (t)emporarily? ";
1914 STDERR->flush;
1915 $choice = lc(substr(<STDIN> || 'R', 0, 1));
1916 if ($choice =~ /^t$/i) {
1917 $cred->may_save(undef);
1918 } elsif ($choice =~ /^r$/i) {
1919 return -1;
1920 } elsif ($may_save && $choice =~ /^p$/i) {
1921 $cred->may_save($may_save);
1922 } else {
1923 goto prompt;
1924 }
1925 $cred->accepted_failures($failures);
1926 $SVN::_Core::SVN_NO_ERROR;
1927 }
1928
1929 sub ssl_client_cert {
1930 my ($cred, $realm, $may_save, $pool) = @_;
1931 $may_save = undef if $_no_auth_cache;
1932 print STDERR "Client certificate filename: ";
1933 STDERR->flush;
1934 chomp(my $filename = <STDIN>);
1935 $cred->cert_file($filename);
1936 $cred->may_save($may_save);
1937 $SVN::_Core::SVN_NO_ERROR;
1938 }
1939
1940 sub ssl_client_cert_pw {
1941 my ($cred, $realm, $may_save, $pool) = @_;
1942 $may_save = undef if $_no_auth_cache;
1943 $cred->password(_read_password("Password: ", $realm));
1944 $cred->may_save($may_save);
1945 $SVN::_Core::SVN_NO_ERROR;
1946 }
1947
1948 sub username {
1949 my ($cred, $realm, $may_save, $pool) = @_;
1950 $may_save = undef if $_no_auth_cache;
1951 if (defined $realm && length $realm) {
1952 print STDERR "Authentication realm: $realm\n";
1953 }
1954 my $username;
1955 if (defined $_username) {
1956 $username = $_username;
1957 } else {
1958 print STDERR "Username: ";
1959 STDERR->flush;
1960 chomp($username = <STDIN>);
1961 }
1962 $cred->username($username);
1963 $cred->may_save($may_save);
1964 $SVN::_Core::SVN_NO_ERROR;
1965 }
1966
1967 sub _read_password {
1968 my ($prompt, $realm) = @_;
1969 print STDERR $prompt;
1970 STDERR->flush;
1971 require Term::ReadKey;
1972 Term::ReadKey::ReadMode('noecho');
1973 my $password = '';
1974 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1975 last if $key =~ /[\012\015]/; # \n\r
1976 $password .= $key;
1977 }
1978 Term::ReadKey::ReadMode('restore');
1979 print STDERR "\n";
1980 STDERR->flush;
1981 $password;
1982 }
1983
1984 package main;
1985
1986 {
1987 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1988 $SVN::Node::dir.$SVN::Node::unknown.
1989 $SVN::Node::none.$SVN::Node::file.
1990 $SVN::Node::dir.$SVN::Node::unknown.
1991 $SVN::Auth::SSL::CNMISMATCH.
1992 $SVN::Auth::SSL::NOTYETVALID.
1993 $SVN::Auth::SSL::EXPIRED.
1994 $SVN::Auth::SSL::UNKNOWNCA.
1995 $SVN::Auth::SSL::OTHER;
1996 }
1997
1998 package SVN::Git::Fetcher;
1999 use vars qw/@ISA/;
2000 use strict;
2001 use warnings;
2002 use Carp qw/croak/;
2003 use IO::File qw//;
2004 use Digest::MD5;
2005
2006 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
2007 sub new {
2008 my ($class, $git_svn) = @_;
2009 my $self = SVN::Delta::Editor->new;
2010 bless $self, $class;
2011 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2012 $self->{empty} = {};
2013 $self->{dir_prop} = {};
2014 $self->{file_prop} = {};
2015 $self->{absent_dir} = {};
2016 $self->{absent_file} = {};
2017 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2018 $self;
2019 }
2020
2021 sub set_path_strip {
2022 my ($self, $path) = @_;
2023 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2024 }
2025
2026 sub open_root {
2027 { path => '' };
2028 }
2029
2030 sub open_directory {
2031 my ($self, $path, $pb, $rev) = @_;
2032 { path => $path };
2033 }
2034
2035 sub git_path {
2036 my ($self, $path) = @_;
2037 if ($self->{path_strip}) {
2038 $path =~ s!$self->{path_strip}!! or
2039 die "Failed to strip path '$path' ($self->{path_strip})\n";
2040 }
2041 $path;
2042 }
2043
2044 sub delete_entry {
2045 my ($self, $path, $rev, $pb) = @_;
2046
2047 my $gpath = $self->git_path($path);
2048 return undef if ($gpath eq '');
2049
2050 # remove entire directories.
2051 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2052 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2053 -r --name-only -z/,
2054 $self->{c}, '--', $gpath);
2055 local $/ = "\0";
2056 while (<$ls>) {
2057 chomp;
2058 $self->{gii}->remove($_);
2059 print "\tD\t$_\n" unless $::_q;
2060 }
2061 print "\tD\t$gpath/\n" unless $::_q;
2062 command_close_pipe($ls, $ctx);
2063 $self->{empty}->{$path} = 0
2064 } else {
2065 $self->{gii}->remove($gpath);
2066 print "\tD\t$gpath\n" unless $::_q;
2067 }
2068 undef;
2069 }
2070
2071 sub open_file {
2072 my ($self, $path, $pb, $rev) = @_;
2073 my $gpath = $self->git_path($path);
2074 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2075 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2076 unless (defined $mode && defined $blob) {
2077 die "$path was not found in commit $self->{c} (r$rev)\n";
2078 }
2079 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2080 pool => SVN::Pool->new, action => 'M' };
2081 }
2082
2083 sub add_file {
2084 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2085 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2086 delete $self->{empty}->{$dir};
2087 { path => $path, mode_a => 100644, mode_b => 100644,
2088 pool => SVN::Pool->new, action => 'A' };
2089 }
2090
2091 sub add_directory {
2092 my ($self, $path, $cp_path, $cp_rev) = @_;
2093 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2094 delete $self->{empty}->{$dir};
2095 $self->{empty}->{$path} = 1;
2096 { path => $path };
2097 }
2098
2099 sub change_dir_prop {
2100 my ($self, $db, $prop, $value) = @_;
2101 $self->{dir_prop}->{$db->{path}} ||= {};
2102 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2103 undef;
2104 }
2105
2106 sub absent_directory {
2107 my ($self, $path, $pb) = @_;
2108 $self->{absent_dir}->{$pb->{path}} ||= [];
2109 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2110 undef;
2111 }
2112
2113 sub absent_file {
2114 my ($self, $path, $pb) = @_;
2115 $self->{absent_file}->{$pb->{path}} ||= [];
2116 push @{$self->{absent_file}->{$pb->{path}}}, $path;
2117 undef;
2118 }
2119
2120 sub change_file_prop {
2121 my ($self, $fb, $prop, $value) = @_;
2122 if ($prop eq 'svn:executable') {
2123 if ($fb->{mode_b} != 120000) {
2124 $fb->{mode_b} = defined $value ? 100755 : 100644;
2125 }
2126 } elsif ($prop eq 'svn:special') {
2127 $fb->{mode_b} = defined $value ? 120000 : 100644;
2128 } else {
2129 $self->{file_prop}->{$fb->{path}} ||= {};
2130 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2131 }
2132 undef;
2133 }
2134
2135 sub apply_textdelta {
2136 my ($self, $fb, $exp) = @_;
2137 my $fh = IO::File->new_tmpfile;
2138 $fh->autoflush(1);
2139 # $fh gets auto-closed() by SVN::TxDelta::apply(),
2140 # (but $base does not,) so dup() it for reading in close_file
2141 open my $dup, '<&', $fh or croak $!;
2142 my $base = IO::File->new_tmpfile;
2143 $base->autoflush(1);
2144 if ($fb->{blob}) {
2145 defined (my $pid = fork) or croak $!;
2146 if (!$pid) {
2147 open STDOUT, '>&', $base or croak $!;
2148 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2149 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2150 }
2151 waitpid $pid, 0;
2152 croak $? if $?;
2153
2154 if (defined $exp) {
2155 seek $base, 0, 0 or croak $!;
2156 my $md5 = Digest::MD5->new;
2157 $md5->addfile($base);
2158 my $got = $md5->hexdigest;
2159 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2160 "expected: $exp\n",
2161 " got: $got\n" if ($got ne $exp);
2162 }
2163 }
2164 seek $base, 0, 0 or croak $!;
2165 $fb->{fh} = $dup;
2166 $fb->{base} = $base;
2167 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2168 }
2169
2170 sub close_file {
2171 my ($self, $fb, $exp) = @_;
2172 my $hash;
2173 my $path = $self->git_path($fb->{path});
2174 if (my $fh = $fb->{fh}) {
2175 seek($fh, 0, 0) or croak $!;
2176 my $md5 = Digest::MD5->new;
2177 $md5->addfile($fh);
2178 my $got = $md5->hexdigest;
2179 die "Checksum mismatch: $path\n",
2180 "expected: $exp\n got: $got\n" if ($got ne $exp);
2181 seek($fh, 0, 0) or croak $!;
2182 if ($fb->{mode_b} == 120000) {
2183 read($fh, my $buf, 5) == 5 or croak $!;
2184 $buf eq 'link ' or die "$path has mode 120000",
2185 "but is not a link\n";
2186 }
2187 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2188 if (!$pid) {
2189 open STDIN, '<&', $fh or croak $!;
2190 exec qw/git-hash-object -w --stdin/ or croak $!;
2191 }
2192 chomp($hash = do { local $/; <$out> });
2193 close $out or croak $!;
2194 close $fh or croak $!;
2195 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2196 close $fb->{base} or croak $!;
2197 } else {
2198 $hash = $fb->{blob} or die "no blob information\n";
2199 }
2200 $fb->{pool}->clear;
2201 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2202 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2203 undef;
2204 }
2205
2206 sub abort_edit {
2207 my $self = shift;
2208 $self->{nr} = $self->{gii}->{nr};
2209 delete $self->{gii};
2210 $self->SUPER::abort_edit(@_);
2211 }
2212
2213 sub close_edit {
2214 my $self = shift;
2215 $self->{git_commit_ok} = 1;
2216 $self->{nr} = $self->{gii}->{nr};
2217 delete $self->{gii};
2218 $self->SUPER::close_edit(@_);
2219 }
2220
2221 package SVN::Git::Editor;
2222 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2223 use strict;
2224 use warnings;
2225 use Carp qw/croak/;
2226 use IO::File;
2227 use Digest::MD5;
2228
2229 sub new {
2230 my ($class, $opts) = @_;
2231 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2232 die "$_ required!\n" unless (defined $opts->{$_});
2233 }
2234
2235 my $pool = SVN::Pool->new;
2236 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2237 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2238 $opts->{r}, $mods);
2239
2240 # $opts->{ra} functions should not be used after this:
2241 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
2242 $opts->{editor_cb}, $pool);
2243 my $self = SVN::Delta::Editor->new(@ce, $pool);
2244 bless $self, $class;
2245 foreach (qw/svn_path r tree_a tree_b/) {
2246 $self->{$_} = $opts->{$_};
2247 }
2248 $self->{url} = $opts->{ra}->{url};
2249 $self->{mods} = $mods;
2250 $self->{types} = $types;
2251 $self->{pool} = $pool;
2252 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2253 $self->{rm} = { };
2254 $self->{path_prefix} = length $self->{svn_path} ?
2255 "$self->{svn_path}/" : '';
2256 return $self;
2257 }
2258
2259 sub generate_diff {
2260 my ($tree_a, $tree_b) = @_;
2261 my @diff_tree = qw(diff-tree -z -r);
2262 if ($_cp_similarity) {
2263 push @diff_tree, "-C$_cp_similarity";
2264 } else {
2265 push @diff_tree, '-C';
2266 }
2267 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2268 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2269 push @diff_tree, $tree_a, $tree_b;
2270 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2271 local $/ = "\0";
2272 my $state = 'meta';
2273 my @mods;
2274 while (<$diff_fh>) {
2275 chomp $_; # this gets rid of the trailing "\0"
2276 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2277 $::sha1\s($::sha1)\s
2278 ([MTCRAD])\d*$/xo) {
2279 push @mods, { mode_a => $1, mode_b => $2,
2280 sha1_b => $3, chg => $4 };
2281 if ($4 =~ /^(?:C|R)$/) {
2282 $state = 'file_a';
2283 } else {
2284 $state = 'file_b';
2285 }
2286 } elsif ($state eq 'file_a') {
2287 my $x = $mods[$#mods] or croak "Empty array\n";
2288 if ($x->{chg} !~ /^(?:C|R)$/) {
2289 croak "Error parsing $_, $x->{chg}\n";
2290 }
2291 $x->{file_a} = $_;
2292 $state = 'file_b';
2293 } elsif ($state eq 'file_b') {
2294 my $x = $mods[$#mods] or croak "Empty array\n";
2295 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2296 croak "Error parsing $_, $x->{chg}\n";
2297 }
2298 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2299 croak "Error parsing $_, $x->{chg}\n";
2300 }
2301 $x->{file_b} = $_;
2302 $state = 'meta';
2303 } else {
2304 croak "Error parsing $_\n";
2305 }
2306 }
2307 command_close_pipe($diff_fh, $ctx);
2308 \@mods;
2309 }
2310
2311 sub check_diff_paths {
2312 my ($ra, $pfx, $rev, $mods) = @_;
2313 my %types;
2314 $pfx .= '/' if length $pfx;
2315
2316 sub type_diff_paths {
2317 my ($ra, $types, $path, $rev) = @_;
2318 my @p = split m#/+#, $path;
2319 my $c = shift @p;
2320 unless (defined $types->{$c}) {
2321 $types->{$c} = $ra->check_path($c, $rev);
2322 }
2323 while (@p) {
2324 $c .= '/' . shift @p;
2325 next if defined $types->{$c};
2326 $types->{$c} = $ra->check_path($c, $rev);
2327 }
2328 }
2329
2330 foreach my $m (@$mods) {
2331 foreach my $f (qw/file_a file_b/) {
2332 next unless defined $m->{$f};
2333 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2334 if (length $pfx.$dir && ! defined $types{$dir}) {
2335 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2336 }
2337 }
2338 }
2339 \%types;
2340 }
2341
2342 sub split_path {
2343 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2344 }
2345
2346 sub repo_path {
2347 my ($self, $path) = @_;
2348 $self->{path_prefix}.(defined $path ? $path : '');
2349 }
2350
2351 sub url_path {
2352 my ($self, $path) = @_;
2353 $self->{url} . '/' . $self->repo_path($path);
2354 }
2355
2356 sub rmdirs {
2357 my ($self) = @_;
2358 my $rm = $self->{rm};
2359 delete $rm->{''}; # we never delete the url we're tracking
2360 return unless %$rm;
2361
2362 foreach (keys %$rm) {
2363 my @d = split m#/#, $_;
2364 my $c = shift @d;
2365 $rm->{$c} = 1;
2366 while (@d) {
2367 $c .= '/' . shift @d;
2368 $rm->{$c} = 1;
2369 }
2370 }
2371 delete $rm->{$self->{svn_path}};
2372 delete $rm->{''}; # we never delete the url we're tracking
2373 return unless %$rm;
2374
2375 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2376 $self->{tree_b});
2377 local $/ = "\0";
2378 while (<$fh>) {
2379 chomp;
2380 my @dn = split m#/#, $_;
2381 while (pop @dn) {
2382 delete $rm->{join '/', @dn};
2383 }
2384 unless (%$rm) {
2385 close $fh;
2386 return;
2387 }
2388 }
2389 command_close_pipe($fh, $ctx);
2390
2391 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2392 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2393 $self->close_directory($bat->{$d}, $p);
2394 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2395 print "\tD+\t$d/\n" unless $::_q;
2396 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2397 delete $bat->{$d};
2398 }
2399 }
2400
2401 sub open_or_add_dir {
2402 my ($self, $full_path, $baton) = @_;
2403 my $t = $self->{types}->{$full_path};
2404 if (!defined $t) {
2405 die "$full_path not known in r$self->{r} or we have a bug!\n";
2406 }
2407 if ($t == $SVN::Node::none) {
2408 return $self->add_directory($full_path, $baton,
2409 undef, -1, $self->{pool});
2410 } elsif ($t == $SVN::Node::dir) {
2411 return $self->open_directory($full_path, $baton,
2412 $self->{r}, $self->{pool});
2413 }
2414 print STDERR "$full_path already exists in repository at ",
2415 "r$self->{r} and it is not a directory (",
2416 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2417 exit 1;
2418 }
2419
2420 sub ensure_path {
2421 my ($self, $path) = @_;
2422 my $bat = $self->{bat};
2423 my $repo_path = $self->repo_path($path);
2424 return $bat->{''} unless (length $repo_path);
2425 my @p = split m#/+#, $repo_path;
2426 my $c = shift @p;
2427 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2428 while (@p) {
2429 my $c0 = $c;
2430 $c .= '/' . shift @p;
2431 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2432 }
2433 return $bat->{$c};
2434 }
2435
2436 sub A {
2437 my ($self, $m) = @_;
2438 my ($dir, $file) = split_path($m->{file_b});
2439 my $pbat = $self->ensure_path($dir);
2440 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2441 undef, -1);
2442 print "\tA\t$m->{file_b}\n" unless $::_q;
2443 $self->chg_file($fbat, $m);
2444 $self->close_file($fbat,undef,$self->{pool});
2445 }
2446
2447 sub C {
2448 my ($self, $m) = @_;
2449 my ($dir, $file) = split_path($m->{file_b});
2450 my $pbat = $self->ensure_path($dir);
2451 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2452 $self->url_path($m->{file_a}), $self->{r});
2453 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2454 $self->chg_file($fbat, $m);
2455 $self->close_file($fbat,undef,$self->{pool});
2456 }
2457
2458 sub delete_entry {
2459 my ($self, $path, $pbat) = @_;
2460 my $rpath = $self->repo_path($path);
2461 my ($dir, $file) = split_path($rpath);
2462 $self->{rm}->{$dir} = 1;
2463 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2464 }
2465
2466 sub R {
2467 my ($self, $m) = @_;
2468 my ($dir, $file) = split_path($m->{file_b});
2469 my $pbat = $self->ensure_path($dir);
2470 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2471 $self->url_path($m->{file_a}), $self->{r});
2472 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2473 $self->chg_file($fbat, $m);
2474 $self->close_file($fbat,undef,$self->{pool});
2475
2476 ($dir, $file) = split_path($m->{file_a});
2477 $pbat = $self->ensure_path($dir);
2478 $self->delete_entry($m->{file_a}, $pbat);
2479 }
2480
2481 sub M {
2482 my ($self, $m) = @_;
2483 my ($dir, $file) = split_path($m->{file_b});
2484 my $pbat = $self->ensure_path($dir);
2485 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2486 $pbat,$self->{r},$self->{pool});
2487 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2488 $self->chg_file($fbat, $m);
2489 $self->close_file($fbat,undef,$self->{pool});
2490 }
2491
2492 sub T { shift->M(@_) }
2493
2494 sub change_file_prop {
2495 my ($self, $fbat, $pname, $pval) = @_;
2496 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2497 }
2498
2499 sub chg_file {
2500 my ($self, $fbat, $m) = @_;
2501 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2502 $self->change_file_prop($fbat,'svn:executable','*');
2503 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2504 $self->change_file_prop($fbat,'svn:executable',undef);
2505 }
2506 my $fh = IO::File->new_tmpfile or croak $!;
2507 if ($m->{mode_b} =~ /^120/) {
2508 print $fh 'link ' or croak $!;
2509 $self->change_file_prop($fbat,'svn:special','*');
2510 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2511 $self->change_file_prop($fbat,'svn:special',undef);
2512 }
2513 defined(my $pid = fork) or croak $!;
2514 if (!$pid) {
2515 open STDOUT, '>&', $fh or croak $!;
2516 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2517 }
2518 waitpid $pid, 0;
2519 croak $? if $?;
2520 $fh->flush == 0 or croak $!;
2521 seek $fh, 0, 0 or croak $!;
2522
2523 my $md5 = Digest::MD5->new;
2524 $md5->addfile($fh) or croak $!;
2525 seek $fh, 0, 0 or croak $!;
2526
2527 my $exp = $md5->hexdigest;
2528 my $pool = SVN::Pool->new;
2529 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2530 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2531 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2532 $pool->clear;
2533
2534 close $fh or croak $!;
2535 }
2536
2537 sub D {
2538 my ($self, $m) = @_;
2539 my ($dir, $file) = split_path($m->{file_b});
2540 my $pbat = $self->ensure_path($dir);
2541 print "\tD\t$m->{file_b}\n" unless $::_q;
2542 $self->delete_entry($m->{file_b}, $pbat);
2543 }
2544
2545 sub close_edit {
2546 my ($self) = @_;
2547 my ($p,$bat) = ($self->{pool}, $self->{bat});
2548 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2549 $self->close_directory($bat->{$_}, $p);
2550 }
2551 $self->SUPER::close_edit($p);
2552 $p->clear;
2553 }
2554
2555 sub abort_edit {
2556 my ($self) = @_;
2557 $self->SUPER::abort_edit($self->{pool});
2558 }
2559
2560 sub DESTROY {
2561 my $self = shift;
2562 $self->SUPER::DESTROY(@_);
2563 $self->{pool}->clear;
2564 }
2565
2566 # this drives the editor
2567 sub apply_diff {
2568 my ($self) = @_;
2569 my $mods = $self->{mods};
2570 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2571 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2572 my $f = $m->{chg};
2573 if (defined $o{$f}) {
2574 $self->$f($m);
2575 } else {
2576 fatal("Invalid change type: $f\n");
2577 }
2578 }
2579 $self->rmdirs if $_rmdir;
2580 if (@$mods == 0) {
2581 $self->abort_edit;
2582 } else {
2583 $self->close_edit;
2584 }
2585 return scalar @$mods;
2586 }
2587
2588 package Git::SVN::Ra;
2589 use vars qw/@ISA $config_dir $_log_window_size/;
2590 use strict;
2591 use warnings;
2592 my ($can_do_switch);
2593 my $RA;
2594
2595 BEGIN {
2596 # enforce temporary pool usage for some simple functions
2597 my $e;
2598 foreach (qw/get_latest_revnum get_uuid get_repos_root/) {
2599 $e .= "sub $_ {
2600 my \$self = shift;
2601 my \$pool = SVN::Pool->new;
2602 my \@ret = \$self->SUPER::$_(\@_,\$pool);
2603 \$pool->clear;
2604 wantarray ? \@ret : \$ret[0]; }\n";
2605 }
2606
2607 # get_dir needs $pool held in cache for dirents to work,
2608 # check_path is cacheable and rev_proplist is close enough
2609 # for our purposes.
2610 foreach (qw/check_path get_dir rev_proplist/) {
2611 $e .= "my \%${_}_cache; my \$${_}_rev = 0; sub $_ {
2612 my \$self = shift;
2613 my \$r = pop;
2614 my \$k = join(\"\\0\", \@_);
2615 if (my \$x = \$${_}_cache{\$r}->{\$k}) {
2616 return wantarray ? \@\$x : \$x->[0];
2617 }
2618 my \$pool = SVN::Pool->new;
2619 my \@ret = \$self->SUPER::$_(\@_, \$r, \$pool);
2620 if (\$r != \$${_}_rev) {
2621 \%${_}_cache = ( pool => [] );
2622 \$${_}_rev = \$r;
2623 }
2624 \$${_}_cache{\$r}->{\$k} = \\\@ret;
2625 push \@{\$${_}_cache{pool}}, \$pool;
2626 wantarray ? \@ret : \$ret[0]; }\n";
2627 }
2628 $e .= "\n1;";
2629 eval $e or die $@;
2630 }
2631
2632 sub new {
2633 my ($class, $url) = @_;
2634 $url =~ s!/+$!!;
2635 return $RA if ($RA && $RA->{url} eq $url);
2636
2637 SVN::_Core::svn_config_ensure($config_dir, undef);
2638 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2639 SVN::Client::get_simple_provider(),
2640 SVN::Client::get_ssl_server_trust_file_provider(),
2641 SVN::Client::get_simple_prompt_provider(
2642 \&Git::SVN::Prompt::simple, 2),
2643 SVN::Client::get_ssl_client_cert_prompt_provider(
2644 \&Git::SVN::Prompt::ssl_client_cert, 2),
2645 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2646 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2647 SVN::Client::get_username_provider(),
2648 SVN::Client::get_ssl_server_trust_prompt_provider(
2649 \&Git::SVN::Prompt::ssl_server_trust),
2650 SVN::Client::get_username_prompt_provider(
2651 \&Git::SVN::Prompt::username, 2),
2652 ]);
2653 my $config = SVN::Core::config_get_config($config_dir);
2654 my $self = SVN::Ra->new(url => $url, auth => $baton,
2655 config => $config,
2656 pool => SVN::Pool->new,
2657 auth_provider_callbacks => $callbacks);
2658 $self->{svn_path} = $url;
2659 $self->{repos_root} = $self->get_repos_root;
2660 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
2661 $RA = bless $self, $class;
2662 }
2663
2664 sub DESTROY {
2665 # do not call the real DESTROY since we store ourselves in $RA
2666 }
2667
2668 sub get_log {
2669 my ($self, @args) = @_;
2670 my $pool = SVN::Pool->new;
2671 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2672 my $ret = $self->SUPER::get_log(@args, $pool);
2673 $pool->clear;
2674 $ret;
2675 }
2676
2677 sub get_commit_editor {
2678 my ($self, $log, $cb, $pool) = @_;
2679 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2680 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2681 }
2682
2683 sub gs_do_update {
2684 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2685 my $new = ($rev_a == $rev_b);
2686 my $path = $gs->{path};
2687
2688 my $pool = SVN::Pool->new;
2689 $editor->set_path_strip($path);
2690 my (@pc) = split m#/#, $path;
2691 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2692 1, $editor, $pool);
2693 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2694
2695 # Since we can't rely on svn_ra_reparent being available, we'll
2696 # just have to do some magic with set_path to make it so
2697 # we only want a partial path.
2698 my $sp = '';
2699 my $final = join('/', @pc);
2700 while (@pc) {
2701 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2702 $sp .= '/' if length $sp;
2703 $sp .= shift @pc;
2704 }
2705 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2706
2707 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2708
2709 $reporter->finish_report($pool);
2710 $pool->clear;
2711 $editor->{git_commit_ok};
2712 }
2713
2714 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2715 # svn_ra_reparent didn't work before 1.4)
2716 sub gs_do_switch {
2717 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
2718 my $path = $gs->{path};
2719 my $pool = SVN::Pool->new;
2720
2721 my $full_url = $self->{url};
2722 my $old_url = $full_url;
2723 $full_url .= "/$path" if length $path;
2724 my ($ra, $reparented);
2725 if ($old_url ne $full_url) {
2726 if ($old_url !~ m#^svn(\+ssh)?://#) {
2727 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2728 $pool);
2729 $self->{url} = $full_url;
2730 $reparented = 1;
2731 } else {
2732 $ra = Git::SVN::Ra->new($full_url);
2733 }
2734 }
2735 $ra ||= $self;
2736 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
2737 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2738 $reporter->set_path('', $rev_a, 0, @lock, $pool);
2739 $reporter->finish_report($pool);
2740
2741 if ($reparented) {
2742 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2743 $self->{url} = $old_url;
2744 }
2745
2746 $pool->clear;
2747 $editor->{git_commit_ok};
2748 }
2749
2750 sub gs_fetch_loop_common {
2751 my ($self, $base, $head, $gsv, $globs) = @_;
2752 return if ($base > $head);
2753 my $inc = $_log_window_size;
2754 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2755 my %common;
2756 my $common_max = scalar @$gsv;
2757
2758 foreach my $gs (@$gsv) {
2759 my @tmp = split m#/#, $gs->{path};
2760 my $p = '';
2761 foreach (@tmp) {
2762 $p .= length($p) ? "/$_" : $_;
2763 $common{$p} ||= 0;
2764 $common{$p}++;
2765 }
2766 }
2767 $globs ||= [];
2768 $common_max += scalar @$globs;
2769 foreach my $glob (@$globs) {
2770 my @tmp = split m#/#, $glob->{path}->{left};
2771 my $p = '';
2772 foreach (@tmp) {
2773 $p .= length($p) ? "/$_" : $_;
2774 $common{$p} ||= 0;
2775 $common{$p}++;
2776 }
2777 }
2778
2779 my $longest_path = '';
2780 foreach (sort {length $b <=> length $a} keys %common) {
2781 if ($common{$_} == $common_max) {
2782 $longest_path = $_;
2783 last;
2784 }
2785 }
2786 while (1) {
2787 my %revs;
2788 my $err;
2789 my $err_handler = $SVN::Error::handler;
2790 $SVN::Error::handler = sub {
2791 ($err) = @_;
2792 skip_unknown_revs($err);
2793 };
2794 sub _cb {
2795 my ($paths, $r, $author, $date, $log) = @_;
2796 [ dup_changed_paths($paths),
2797 { author => $author, date => $date, log => $log } ];
2798 }
2799 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
2800 sub { $revs{$_[1]} = _cb(@_) });
2801 if ($err && $max >= $head) {
2802 print STDERR "Path '$longest_path' ",
2803 "was probably deleted:\n",
2804 $err->expanded_message,
2805 "\nWill attempt to follow ",
2806 "revisions r$min .. r$max ",
2807 "committed before the deletion\n";
2808 my $hi = $max;
2809 while (--$hi >= $min) {
2810 my $ok;
2811 $self->get_log([$longest_path], $min, $hi,
2812 0, 1, 1, sub {
2813 $ok ||= $_[1];
2814 $revs{$_[1]} = _cb(@_) });
2815 if ($ok) {
2816 print STDERR "r$min .. r$ok OK\n";
2817 last;
2818 }
2819 }
2820 }
2821 $SVN::Error::handler = $err_handler;
2822
2823 my %exists = map { $_->{path} => $_ } @$gsv;
2824 foreach my $r (sort {$a <=> $b} keys %revs) {
2825 my ($paths, $logged) = @{$revs{$r}};
2826
2827 foreach my $gs ($self->match_globs(\%exists, $paths,
2828 $globs, $r)) {
2829 if ($gs->rev_db_max >= $r) {
2830 next;
2831 }
2832 next unless $gs->match_paths($paths, $r);
2833 $gs->{logged_rev_props} = $logged;
2834 if (my $last_commit = $gs->last_commit) {
2835 $gs->assert_index_clean($last_commit);
2836 }
2837 my $log_entry = $gs->do_fetch($paths, $r);
2838 if ($log_entry) {
2839 $gs->do_git_commit($log_entry);
2840 }
2841 }
2842 foreach my $g (@$globs) {
2843 my $k = "svn-remote.$g->{remote}." .
2844 "$g->{t}-maxRev";
2845 Git::SVN::tmp_config($k, $r);
2846 }
2847 }
2848 # pre-fill the .rev_db since it'll eventually get filled in
2849 # with '0' x40 if something new gets committed
2850 foreach my $gs (@$gsv) {
2851 next if defined $gs->rev_db_get($max);
2852 $gs->rev_db_set($max, 0 x40);
2853 }
2854 foreach my $g (@$globs) {
2855 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
2856 Git::SVN::tmp_config($k, $max);
2857 }
2858 last if $max >= $head;
2859 $min = $max + 1;
2860 $max += $inc;
2861 $max = $head if ($max > $head);
2862 }
2863 }
2864
2865 sub match_globs {
2866 my ($self, $exists, $paths, $globs, $r) = @_;
2867
2868 sub get_dir_check {
2869 my ($self, $exists, $g, $r) = @_;
2870 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
2871 return unless scalar @x == 3;
2872 my $dirents = $x[0];
2873 foreach my $de (keys %$dirents) {
2874 next if $dirents->{$de}->kind != $SVN::Node::dir;
2875 my $p = $g->{path}->full_path($de);
2876 next if $exists->{$p};
2877 next if (length $g->{path}->{right} &&
2878 ($self->check_path($p, $r) !=
2879 $SVN::Node::dir));
2880 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
2881 $g->{ref}->full_path($de), 1);
2882 }
2883 }
2884 foreach my $g (@$globs) {
2885 if (my $path = $paths->{"/$g->{path}->{left}"}) {
2886 if ($path->{action} =~ /^[AR]$/) {
2887 get_dir_check($self, $exists, $g, $r);
2888 }
2889 }
2890 foreach (keys %$paths) {
2891 if (/$g->{path}->{left_regex}/ &&
2892 !/$g->{path}->{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 || $errno == 160013) {
2960 warn "W: Ignoring error from SVN, path probably ",
2961 "does not exist: ($errno): ",
2962 $err->expanded_message,"\n";
2963 return;
2964 }
2965 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2966 }
2967
2968 # svn_log_changed_path_t objects passed to get_log are likely to be
2969 # overwritten even if only the refs are copied to an external variable,
2970 # so we should dup the structures in their entirety. Using an externally
2971 # passed pool (instead of our temporary and quickly cleared pool in
2972 # Git::SVN::Ra) does not help matters at all...
2973 sub dup_changed_paths {
2974 my ($paths) = @_;
2975 return undef unless $paths;
2976 my %ret;
2977 foreach my $p (keys %$paths) {
2978 my $i = $paths->{$p};
2979 my %s = map { $_ => $i->$_ }
2980 qw/copyfrom_path copyfrom_rev action/;
2981 $ret{$p} = \%s;
2982 }
2983 \%ret;
2984 }
2985
2986 package Git::SVN::Log;
2987 use strict;
2988 use warnings;
2989 use POSIX qw/strftime/;
2990 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2991 %rusers $show_commit $incremental/;
2992 my $l_fmt;
2993
2994 sub cmt_showable {
2995 my ($c) = @_;
2996 return 1 if defined $c->{r};
2997 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2998 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2999 my @log = command(qw/cat-file commit/, $c->{c});
3000 shift @log while ($log[0] ne "\n");
3001 shift @log;
3002 @{$c->{l}} = grep !/^git-svn-id: /, @log;
3003
3004 (undef, $c->{r}, undef) = ::extract_metadata(
3005 (grep(/^git-svn-id: /, @log))[-1]);
3006 }
3007 return defined $c->{r};
3008 }
3009
3010 sub log_use_color {
3011 return 1 if $color;
3012 my ($dc, $dcvar);
3013 $dcvar = 'color.diff';
3014 $dc = `git-config --get $dcvar`;
3015 if ($dc eq '') {
3016 # nothing at all; fallback to "diff.color"
3017 $dcvar = 'diff.color';
3018 $dc = `git-config --get $dcvar`;
3019 }
3020 chomp($dc);
3021 if ($dc eq 'auto') {
3022 my $pc;
3023 $pc = `git-config --get color.pager`;
3024 if ($pc eq '') {
3025 # does not have it -- fallback to pager.color
3026 $pc = `git-config --bool --get pager.color`;
3027 }
3028 else {
3029 $pc = `git-config --bool --get color.pager`;
3030 if ($?) {
3031 $pc = 'false';
3032 }
3033 }
3034 chomp($pc);
3035 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3036 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3037 }
3038 return 0;
3039 }
3040 return 0 if $dc eq 'never';
3041 return 1 if $dc eq 'always';
3042 chomp($dc = `git-config --bool --get $dcvar`);
3043 return ($dc eq 'true');
3044 }
3045
3046 sub git_svn_log_cmd {
3047 my ($r_min, $r_max, @args) = @_;
3048 my $head = 'HEAD';
3049 foreach my $x (@args) {
3050 last if $x eq '--';
3051 next unless ::verify_ref("$x^0");
3052 $head = $x;
3053 last;
3054 }
3055
3056 my $url;
3057 my ($fh, $ctx) = command_output_pipe('rev-list', $head);
3058 while (<$fh>) {
3059 chomp;
3060 $url = (::cmt_metadata($_))[0];
3061 last if defined $url;
3062 }
3063 close $fh; # break the pipe
3064
3065 my $gs = Git::SVN->find_by_url($url) || Git::SVN->_new;
3066 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3067 $gs->refname);
3068 push @cmd, '-r' unless $non_recursive;
3069 push @cmd, qw/--raw --name-status/ if $verbose;
3070 push @cmd, '--color' if log_use_color();
3071 return @cmd unless defined $r_max;
3072 if ($r_max == $r_min) {
3073 push @cmd, '--max-count=1';
3074 if (my $c = $gs->rev_db_get($r_max)) {
3075 push @cmd, $c;
3076 }
3077 } else {
3078 my ($c_min, $c_max);
3079 $c_max = $gs->rev_db_get($r_max);
3080 $c_min = $gs->rev_db_get($r_min);
3081 if (defined $c_min && defined $c_max) {
3082 if ($r_max > $r_max) {
3083 push @cmd, "$c_min..$c_max";
3084 } else {
3085 push @cmd, "$c_max..$c_min";
3086 }
3087 } elsif ($r_max > $r_min) {
3088 push @cmd, $c_max;
3089 } else {
3090 push @cmd, $c_min;
3091 }
3092 }
3093 return @cmd;
3094 }
3095
3096 # adapted from pager.c
3097 sub config_pager {
3098 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3099 if (!defined $pager) {
3100 $pager = 'less';
3101 } elsif (length $pager == 0 || $pager eq 'cat') {
3102 $pager = undef;
3103 }
3104 }
3105
3106 sub run_pager {
3107 return unless -t *STDOUT;
3108 pipe my $rfd, my $wfd or return;
3109 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
3110 if (!$pid) {
3111 open STDOUT, '>&', $wfd or
3112 ::fatal "Can't redirect to stdout: $!\n";
3113 return;
3114 }
3115 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
3116 $ENV{LESS} ||= 'FRSX';
3117 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
3118 }
3119
3120 sub tz_to_s_offset {
3121 my ($tz) = @_;
3122 $tz =~ s/(\d\d)$//;
3123 return ($1 * 60) + ($tz * 3600);
3124 }
3125
3126 sub get_author_info {
3127 my ($dest, $author, $t, $tz) = @_;
3128 $author =~ s/(?:^\s*|\s*$)//g;
3129 $dest->{a_raw} = $author;
3130 my $au;
3131 if ($::_authors) {
3132 $au = $rusers{$author} || undef;
3133 }
3134 if (!$au) {
3135 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3136 }
3137 $dest->{t} = $t;
3138 $dest->{tz} = $tz;
3139 $dest->{a} = $au;
3140 # Date::Parse isn't in the standard Perl distro :(
3141 if ($tz =~ s/^\+//) {
3142 $t += tz_to_s_offset($tz);
3143 } elsif ($tz =~ s/^\-//) {
3144 $t -= tz_to_s_offset($tz);
3145 }
3146 $dest->{t_utc} = $t;
3147 }
3148
3149 sub process_commit {
3150 my ($c, $r_min, $r_max, $defer) = @_;
3151 if (defined $r_min && defined $r_max) {
3152 if ($r_min == $c->{r} && $r_min == $r_max) {
3153 show_commit($c);
3154 return 0;
3155 }
3156 return 1 if $r_min == $r_max;
3157 if ($r_min < $r_max) {
3158 # we need to reverse the print order
3159 return 0 if (defined $limit && --$limit < 0);
3160 push @$defer, $c;
3161 return 1;
3162 }
3163 if ($r_min != $r_max) {
3164 return 1 if ($r_min < $c->{r});
3165 return 1 if ($r_max > $c->{r});
3166 }
3167 }
3168 return 0 if (defined $limit && --$limit < 0);
3169 show_commit($c);
3170 return 1;
3171 }
3172
3173 sub show_commit {
3174 my $c = shift;
3175 if ($oneline) {
3176 my $x = "\n";
3177 if (my $l = $c->{l}) {
3178 while ($l->[0] =~ /^\s*$/) { shift @$l }
3179 $x = $l->[0];
3180 }
3181 $l_fmt ||= 'A' . length($c->{r});
3182 print 'r',pack($l_fmt, $c->{r}),' | ';
3183 print "$c->{c} | " if $show_commit;
3184 print $x;
3185 } else {
3186 show_commit_normal($c);
3187 }
3188 }
3189
3190 sub show_commit_changed_paths {
3191 my ($c) = @_;
3192 return unless $c->{changed};
3193 print "Changed paths:\n", @{$c->{changed}};
3194 }
3195
3196 sub show_commit_normal {
3197 my ($c) = @_;
3198 print '-' x72, "\nr$c->{r} | ";
3199 print "$c->{c} | " if $show_commit;
3200 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3201 localtime($c->{t_utc})), ' | ';
3202 my $nr_line = 0;
3203
3204 if (my $l = $c->{l}) {
3205 while ($l->[$#$l] eq "\n" && $#$l > 0
3206 && $l->[($#$l - 1)] eq "\n") {
3207 pop @$l;
3208 }
3209 $nr_line = scalar @$l;
3210 if (!$nr_line) {
3211 print "1 line\n\n\n";
3212 } else {
3213 if ($nr_line == 1) {
3214 $nr_line = '1 line';
3215 } else {
3216 $nr_line .= ' lines';
3217 }
3218 print $nr_line, "\n";
3219 show_commit_changed_paths($c);
3220 print "\n";
3221 print $_ foreach @$l;
3222 }
3223 } else {
3224 print "1 line\n";
3225 show_commit_changed_paths($c);
3226 print "\n";
3227
3228 }
3229 foreach my $x (qw/raw diff/) {
3230 if ($c->{$x}) {
3231 print "\n";
3232 print $_ foreach @{$c->{$x}}
3233 }
3234 }
3235 }
3236
3237 sub cmd_show_log {
3238 my (@args) = @_;
3239 my ($r_min, $r_max);
3240 my $r_last = -1; # prevent dupes
3241 if (defined $TZ) {
3242 $ENV{TZ} = $TZ;
3243 } else {
3244 delete $ENV{TZ};
3245 }
3246 if (defined $::_revision) {
3247 if ($::_revision =~ /^(\d+):(\d+)$/) {
3248 ($r_min, $r_max) = ($1, $2);
3249 } elsif ($::_revision =~ /^\d+$/) {
3250 $r_min = $r_max = $::_revision;
3251 } else {
3252 ::fatal "-r$::_revision is not supported, use ",
3253 "standard \'git log\' arguments instead\n";
3254 }
3255 }
3256
3257 config_pager();
3258 @args = (git_svn_log_cmd($r_min, $r_max, @args), @args);
3259 my $log = command_output_pipe(@args);
3260 run_pager();
3261 my (@k, $c, $d);
3262 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3263 while (<$log>) {
3264 if (/^${esc_color}commit ($::sha1_short)/o) {
3265 my $cmt = $1;
3266 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3267 $r_last = $c->{r};
3268 process_commit($c, $r_min, $r_max, \@k) or
3269 goto out;
3270 }
3271 $d = undef;
3272 $c = { c => $cmt };
3273 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3274 get_author_info($c, $1, $2, $3);
3275 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3276 # ignore
3277 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3278 push @{$c->{raw}}, $_;
3279 } elsif (/^${esc_color}[ACRMDT]\t/) {
3280 # we could add $SVN->{svn_path} here, but that requires
3281 # remote access at the moment (repo_path_split)...
3282 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
3283 push @{$c->{changed}}, $_;
3284 } elsif (/^${esc_color}diff /o) {
3285 $d = 1;
3286 push @{$c->{diff}}, $_;
3287 } elsif ($d) {
3288 push @{$c->{diff}}, $_;
3289 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
3290 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3291 } elsif (s/^${esc_color} //o) {
3292 push @{$c->{l}}, $_;
3293 }
3294 }
3295 if ($c && defined $c->{r} && $c->{r} != $r_last) {
3296 $r_last = $c->{r};
3297 process_commit($c, $r_min, $r_max, \@k);
3298 }
3299 if (@k) {
3300 my $swap = $r_max;
3301 $r_max = $r_min;
3302 $r_min = $swap;
3303 process_commit($_, $r_min, $r_max) foreach reverse @k;
3304 }
3305 out:
3306 close $log;
3307 print '-' x72,"\n" unless $incremental || $oneline;
3308 }
3309
3310 package Git::SVN::Migration;
3311 # these version numbers do NOT correspond to actual version numbers
3312 # of git nor git-svn. They are just relative.
3313 #
3314 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3315 #
3316 # v1 layout: .git/$id/info/url, refs/remotes/$id
3317 #
3318 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3319 #
3320 # v3 layout: .git/svn/$id, refs/remotes/$id
3321 # - info/url may remain for backwards compatibility
3322 # - this is what we migrate up to this layout automatically,
3323 # - this will be used by git svn init on single branches
3324 # v3.1 layout (auto migrated):
3325 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
3326 # for backwards compatibility
3327 #
3328 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3329 # - this is only created for newly multi-init-ed
3330 # repositories. Similar in spirit to the
3331 # --use-separate-remotes option in git-clone (now default)
3332 # - we do not automatically migrate to this (following
3333 # the example set by core git)
3334 use strict;
3335 use warnings;
3336 use Carp qw/croak/;
3337 use File::Path qw/mkpath/;
3338 use File::Basename qw/dirname basename/;
3339 use vars qw/$_minimize/;
3340
3341 sub migrate_from_v0 {
3342 my $git_dir = $ENV{GIT_DIR};
3343 return undef unless -d $git_dir;
3344 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3345 my $migrated = 0;
3346 while (<$fh>) {
3347 chomp;
3348 my ($id, $orig_ref) = ($_, $_);
3349 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3350 next unless -f "$git_dir/$id/info/url";
3351 my $new_ref = "refs/remotes/$id";
3352 if (::verify_ref("$new_ref^0")) {
3353 print STDERR "W: $orig_ref is probably an old ",
3354 "branch used by an ancient version of ",
3355 "git-svn.\n",
3356 "However, $new_ref also exists.\n",
3357 "We will not be able ",
3358 "to use this branch until this ",
3359 "ambiguity is resolved.\n";
3360 next;
3361 }
3362 print STDERR "Migrating from v0 layout...\n" if !$migrated;
3363 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3364 command_noisy('update-ref', $new_ref, $orig_ref);
3365 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3366 $migrated++;
3367 }
3368 command_close_pipe($fh, $ctx);
3369 print STDERR "Done migrating from v0 layout...\n" if $migrated;
3370 $migrated;
3371 }
3372
3373 sub migrate_from_v1 {
3374 my $git_dir = $ENV{GIT_DIR};
3375 my $migrated = 0;
3376 return $migrated unless -d $git_dir;
3377 my $svn_dir = "$git_dir/svn";
3378
3379 # just in case somebody used 'svn' as their $id at some point...
3380 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3381
3382 print STDERR "Migrating from a git-svn v1 layout...\n";
3383 mkpath([$svn_dir]);
3384 print STDERR "Data from a previous version of git-svn exists, but\n\t",
3385 "$svn_dir\n\t(required for this version ",
3386 "($::VERSION) of git-svn) does not. exist\n";
3387 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3388 while (<$fh>) {
3389 my $x = $_;
3390 next unless $x =~ s#^refs/remotes/##;
3391 chomp $x;
3392 next unless -f "$git_dir/$x/info/url";
3393 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3394 next unless $u;
3395 my $dn = dirname("$git_dir/svn/$x");
3396 mkpath([$dn]) unless -d $dn;
3397 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3398 mkpath(["$git_dir/svn/svn"]);
3399 print STDERR " - $git_dir/$x/info => ",
3400 "$git_dir/svn/$x/info\n";
3401 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3402 croak "$!: $x";
3403 # don't worry too much about these, they probably
3404 # don't exist with repos this old (save for index,
3405 # and we can easily regenerate that)
3406 foreach my $f (qw/unhandled.log index .rev_db/) {
3407 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3408 }
3409 } else {
3410 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3411 rename "$git_dir/$x", "$git_dir/svn/$x" or
3412 croak "$!: $x";
3413 }
3414 $migrated++;
3415 }
3416 command_close_pipe($fh, $ctx);
3417 print STDERR "Done migrating from a git-svn v1 layout\n";
3418 $migrated;
3419 }
3420
3421 sub read_old_urls {
3422 my ($l_map, $pfx, $path) = @_;
3423 my @dir;
3424 foreach (<$path/*>) {
3425 if (-r "$_/info/url") {
3426 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3427 my $ref_id = $pfx . basename $_;
3428 my $url = ::file_to_s("$_/info/url");
3429 $l_map->{$ref_id} = $url;
3430 } elsif (-d $_) {
3431 push @dir, $_;
3432 }
3433 }
3434 foreach (@dir) {
3435 my $x = $_;
3436 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3437 read_old_urls($l_map, $x, $_);
3438 }
3439 }
3440
3441 sub migrate_from_v2 {
3442 my @cfg = command(qw/config -l/);
3443 return if grep /^svn-remote\..+\.url=/, @cfg;
3444 my %l_map;
3445 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3446 my $migrated = 0;
3447
3448 foreach my $ref_id (sort keys %l_map) {
3449 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3450 if ($@) {
3451 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3452 }
3453 $migrated++;
3454 }
3455 $migrated;
3456 }
3457
3458 sub minimize_connections {
3459 my $r = Git::SVN::read_all_remotes();
3460 my $new_urls = {};
3461 my $root_repos = {};
3462 foreach my $repo_id (keys %$r) {
3463 my $url = $r->{$repo_id}->{url} or next;
3464 my $fetch = $r->{$repo_id}->{fetch} or next;
3465 my $ra = Git::SVN::Ra->new($url);
3466
3467 # skip existing cases where we already connect to the root
3468 if (($ra->{url} eq $ra->{repos_root}) ||
3469 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3470 $repo_id)) {
3471 $root_repos->{$ra->{url}} = $repo_id;
3472 next;
3473 }
3474
3475 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3476 my $root_path = $ra->{url};
3477 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3478 foreach my $path (keys %$fetch) {
3479 my $ref_id = $fetch->{$path};
3480 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3481
3482 # make sure we can read when connecting to
3483 # a higher level of a repository
3484 my ($last_rev, undef) = $gs->last_rev_commit;
3485 if (!defined $last_rev) {
3486 $last_rev = eval {
3487 $root_ra->get_latest_revnum;
3488 };
3489 next if $@;
3490 }
3491 my $new = $root_path;
3492 $new .= length $path ? "/$path" : '';
3493 eval {
3494 $root_ra->get_log([$new], $last_rev, $last_rev,
3495 0, 0, 1, sub { });
3496 };
3497 next if $@;
3498 $new_urls->{$ra->{repos_root}}->{$new} =
3499 { ref_id => $ref_id,
3500 old_repo_id => $repo_id,
3501 old_path => $path };
3502 }
3503 }
3504
3505 my @emptied;
3506 foreach my $url (keys %$new_urls) {
3507 # see if we can re-use an existing [svn-remote "repo_id"]
3508 # instead of creating a(n ugly) new section:
3509 my $repo_id = $root_repos->{$url} ||
3510 Git::SVN::sanitize_remote_name($url);
3511
3512 my $fetch = $new_urls->{$url};
3513 foreach my $path (keys %$fetch) {
3514 my $x = $fetch->{$path};
3515 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3516 my $pfx = "svn-remote.$x->{old_repo_id}";
3517
3518 my $old_fetch = quotemeta("$x->{old_path}:".
3519 "refs/remotes/$x->{ref_id}");
3520 command_noisy(qw/config --unset/,
3521 "$pfx.fetch", '^'. $old_fetch . '$');
3522 delete $r->{$x->{old_repo_id}}->
3523 {fetch}->{$x->{old_path}};
3524 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3525 command_noisy(qw/config --unset/,
3526 "$pfx.url");
3527 push @emptied, $x->{old_repo_id}
3528 }
3529 }
3530 }
3531 if (@emptied) {
3532 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3533 "$ENV{GIT_DIR}/config";
3534 print STDERR <<EOF;
3535 The following [svn-remote] sections in your config file ($file) are empty
3536 and can be safely removed:
3537 EOF
3538 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3539 }
3540 }
3541
3542 sub migration_check {
3543 migrate_from_v0();
3544 migrate_from_v1();
3545 migrate_from_v2();
3546 minimize_connections() if $_minimize;
3547 }
3548
3549 package Git::IndexInfo;
3550 use strict;
3551 use warnings;
3552 use Git qw/command_input_pipe command_close_pipe/;
3553
3554 sub new {
3555 my ($class) = @_;
3556 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3557 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3558 }
3559
3560 sub remove {
3561 my ($self, $path) = @_;
3562 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3563 return ++$self->{nr};
3564 }
3565 undef;
3566 }
3567
3568 sub update {
3569 my ($self, $mode, $hash, $path) = @_;
3570 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3571 return ++$self->{nr};
3572 }
3573 undef;
3574 }
3575
3576 sub DESTROY {
3577 my ($self) = @_;
3578 command_close_pipe($self->{gui}, $self->{ctx});
3579 }
3580
3581 package Git::SVN::GlobSpec;
3582 use strict;
3583 use warnings;
3584
3585 sub new {
3586 my ($class, $glob) = @_;
3587 my $re = $glob;
3588 $re =~ s!/+$!!g; # no need for trailing slashes
3589 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
3590 my ($left, $right) = ($1, $2);
3591 if ($nr > 1) {
3592 die "Only one '*' wildcard expansion ",
3593 "is supported (got $nr): '$glob'\n";
3594 } elsif ($nr == 0) {
3595 die "One '*' is needed for glob: '$glob'\n";
3596 }
3597 $re = quotemeta($left) . $re . quotemeta($right);
3598 if (length $left && !($left =~ s!/+$!!g)) {
3599 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
3600 }
3601 if (length $right && !($right =~ s!^/+!!g)) {
3602 die "Missing leading '/' on right side of: '$glob' ($right)\n";
3603 }
3604 my $left_re = qr/^\/\Q$left\E(\/|$)/;
3605 bless { left => $left, right => $right, left_regex => $left_re,
3606 regex => qr/$re/, glob => $glob }, $class;
3607 }
3608
3609 sub full_path {
3610 my ($self, $path) = @_;
3611 return (length $self->{left} ? "$self->{left}/" : '') .
3612 $path . (length $self->{right} ? "/$self->{right}" : '');
3613 }
3614
3615 __END__
3616
3617 Data structures:
3618
3619
3620 $remotes = { # returned by read_all_remotes()
3621 'svn' => {
3622 # svn-remote.svn.url=https://svn.musicpd.org
3623 url => 'https://svn.musicpd.org',
3624 # svn-remote.svn.fetch=mpd/trunk:trunk
3625 fetch => {
3626 'mpd/trunk' => 'trunk',
3627 },
3628 # svn-remote.svn.tags=mpd/tags/*:tags/*
3629 tags => {
3630 path => {
3631 left => 'mpd/tags',
3632 right => '',
3633 regex => qr!mpd/tags/([^/]+)$!,
3634 glob => 'tags/*',
3635 },
3636 ref => {
3637 left => 'tags',
3638 right => '',
3639 regex => qr!tags/([^/]+)$!,
3640 glob => 'tags/*',
3641 },
3642 }
3643 }
3644 };
3645
3646 $log_entry hashref as returned by libsvn_log_entry()
3647 {
3648 log => 'whitespace-formatted log entry
3649 ', # trailing newline is preserved
3650 revision => '8', # integer
3651 date => '2004-02-24T17:01:44.108345Z', # commit date
3652 author => 'committer name'
3653 };
3654
3655
3656 # this is generated by generate_diff();
3657 @mods = array of diff-index line hashes, each element represents one line
3658 of diff-index output
3659
3660 diff-index line ($m hash)
3661 {
3662 mode_a => first column of diff-index output, no leading ':',
3663 mode_b => second column of diff-index output,
3664 sha1_b => sha1sum of the final blob,
3665 chg => change type [MCRADT],
3666 file_a => original file name of a file (iff chg is 'C' or 'R')
3667 file_b => new/current file name of a file (any chg)
3668 }
3669 ;
3670
3671 # retval of read_url_paths{,_all}();
3672 $l_map = {
3673 # repository root url
3674 'https://svn.musicpd.org' => {
3675 # repository path # GIT_SVN_ID
3676 'mpd/trunk' => 'trunk',
3677 'mpd/tags/0.11.5' => 'tags/0.11.5',
3678 },
3679 }
3680
3681 Notes:
3682 I don't trust the each() function on unless I created %hash myself
3683 because the internal iterator may not have started at base.
Alternate source for your git.git needs