?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-svn: handle multi-init without --trunk, UseSvmProps fixes
[git/dscho.git] / git-svn.perl
blobd55691216cca8f0070bd97898da6e0e8b134103f
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision
8 $_q $_authors %users/;
9 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10 $VERSION = '@@GIT_VERSION@@';
11
12 $ENV{GIT_DIR} ||= '.git';
13 $Git::SVN::default_repo_id = 'svn';
14 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
15
16 $Git::SVN::Log::TZ = $ENV{TZ};
17 $ENV{TZ} = 'UTC';
18 $| = 1; # unbuffer STDOUT
19
20 sub fatal (@) { print STDERR @_; exit 1 }
21 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
22 require SVN::Ra;
23 require SVN::Delta;
24 if ($SVN::Core::VERSION lt '1.1.0') {
25 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
26 }
27 push @Git::SVN::Ra::ISA, 'SVN::Ra';
28 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
29 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
30 use Carp qw/croak/;
31 use IO::File qw//;
32 use File::Basename qw/dirname basename/;
33 use File::Path qw/mkpath/;
34 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
35 use IPC::Open3;
36 use Git;
37
38 BEGIN {
39 my $s;
40 foreach (qw/command command_oneline command_noisy command_output_pipe
41 command_input_pipe command_close_pipe/) {
42 $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
43 "*Git::SVN::Migration::$_ = ".
44 "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
45 }
46 eval $s;
47 }
48
49 my ($SVN);
50
51 $sha1 = qr/[a-f\d]{40}/;
52 $sha1_short = qr/[a-f\d]{4,40}/;
53 my ($_stdin, $_help, $_edit,
54 $_message, $_file,
55 $_template, $_shared,
56 $_version,
57 $_merge, $_strategy, $_dry_run,
58 $_prefix);
59 $Git::SVN::_follow_parent = 1;
60 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
61 'config-dir=s' => \$Git::SVN::Ra::config_dir,
62 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
63 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
64 'authors-file|A=s' => \$_authors,
65 'repack:i' => \$Git::SVN::_repack,
66 'noMetadata' => \$Git::SVN::_no_metadata,
67 'useSvmProps' => \$Git::SVN::_use_svm_props,
68 'quiet|q' => \$_q,
69 'repack-flags|repack-args|repack-opts=s' =>
70 \$Git::SVN::_repack_flags,
71 %remote_opts );
72
73 my ($_trunk, $_tags, $_branches);
74 my %multi_opts = ( 'trunk|T=s' => \$_trunk,
75 'tags|t=s' => \$_tags,
76 'branches|b=s' => \$_branches );
77 my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
78 my %cmt_opts = ( 'edit|e' => \$_edit,
79 'rmdir' => \$SVN::Git::Editor::_rmdir,
80 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
81 'l=i' => \$SVN::Git::Editor::_rename_limit,
82 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
83 );
84
85 my %cmd = (
86 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
87 { 'revision|r=s' => \$_revision, %fc_opts } ],
88 init => [ \&cmd_init, "Initialize a repo for tracking" .
89 " (requires URL argument)",
90 \%init_opts ],
91 dcommit => [ \&cmd_dcommit,
92 'Commit several diffs to merge with upstream',
93 { 'merge|m|M' => \$_merge,
94 'strategy|s=s' => \$_strategy,
95 'dry-run|n' => \$_dry_run,
96 %cmt_opts, %fc_opts } ],
97 'set-tree' => [ \&cmd_set_tree,
98 "Set an SVN repository to a git tree-ish",
99 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
100 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
101 { 'revision|r=i' => \$_revision } ],
102 'multi-init' => [ \&cmd_multi_init,
103 'Initialize multiple trees (like git-svnimport)',
104 { %multi_opts, %init_opts, %remote_opts,
105 'revision|r=i' => \$_revision,
106 'prefix=s' => \$_prefix,
107 } ],
108 'multi-fetch' => [ \&cmd_multi_fetch,
109 'Fetch multiple trees (like git-svnimport)',
110 \%fc_opts ],
111 'migrate' => [ sub { },
112 # no-op, we automatically run this anyways,
113 'Migrate configuration/metadata/layout from
114 previous versions of git-svn',
115 \%remote_opts ],
116 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
117 { 'limit=i' => \$Git::SVN::Log::limit,
118 'revision|r=s' => \$_revision,
119 'verbose|v' => \$Git::SVN::Log::verbose,
120 'incremental' => \$Git::SVN::Log::incremental,
121 'oneline' => \$Git::SVN::Log::oneline,
122 'show-commit' => \$Git::SVN::Log::show_commit,
123 'non-recursive' => \$Git::SVN::Log::non_recursive,
124 'authors-file|A=s' => \$_authors,
125 'color' => \$Git::SVN::Log::color,
126 'pager=s' => \$Git::SVN::Log::pager,
127 } ],
128 'commit-diff' => [ \&cmd_commit_diff,
129 'Commit a diff between two trees',
130 { 'message|m=s' => \$_message,
131 'file|F=s' => \$_file,
132 'revision|r=s' => \$_revision,
133 %cmt_opts } ],
134 );
135
136 my $cmd;
137 for (my $i = 0; $i < @ARGV; $i++) {
138 if (defined $cmd{$ARGV[$i]}) {
139 $cmd = $ARGV[$i];
140 splice @ARGV, $i, 1;
141 last;
142 }
143 };
144
145 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
146
147 read_repo_config(\%opts);
148 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
149 'minimize-connections' => \$Git::SVN::Migration::_minimize,
150 'id|i=s' => \$Git::SVN::default_ref_id,
151 'svn-remote|remote|R=s' => \$Git::SVN::default_repo_id);
152 exit 1 if (!$rv && $cmd ne 'log');
153
154 usage(0) if $_help;
155 version() if $_version;
156 usage(1) unless defined $cmd;
157 load_authors() if $_authors;
158 unless ($cmd =~ /^(?:init|multi-init|commit-diff)$/) {
159 Git::SVN::Migration::migration_check();
160 }
161 Git::SVN::init_vars();
162 eval {
163 Git::SVN::verify_remotes_sanity();
164 $cmd{$cmd}->[0]->(@ARGV);
165 };
166 fatal $@ if $@;
167 exit 0;
168
169 ####################### primary functions ######################
170 sub usage {
171 my $exit = shift || 0;
172 my $fd = $exit ? \*STDERR : \*STDOUT;
173 print $fd <<"";
174 git-svn - bidirectional operations between a single Subversion tree and git
175 Usage: $0 <command> [options] [arguments]\n
176
177 print $fd "Available commands:\n" unless $cmd;
178
179 foreach (sort keys %cmd) {
180 next if $cmd && $cmd ne $_;
181 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
182 foreach (keys %{$cmd{$_}->[2]}) {
183 # prints out arguments as they should be passed:
184 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
185 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
186 "--$_" : "-$_" }
187 split /\|/,$_)," $x\n";
188 }
189 }
190 print $fd <<"";
191 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
192 arbitrary identifier if you're tracking multiple SVN branches/repositories in
193 one git repository and want to keep them separate. See git-svn(1) for more
194 information.
195
196 exit $exit;
197 }
198
199 sub version {
200 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
201 exit 0;
202 }
203
204 sub do_git_init_db {
205 unless (-d $ENV{GIT_DIR}) {
206 my @init_db = ('init');
207 push @init_db, "--template=$_template" if defined $_template;
208 push @init_db, "--shared" if defined $_shared;
209 command_noisy(@init_db);
210 }
211 }
212
213 sub cmd_init {
214 my $url = shift or die "SVN repository location required " .
215 "as a command-line argument\n";
216 if (my $repo_path = shift) {
217 unless (-d $repo_path) {
218 mkpath([$repo_path]);
219 }
220 chdir $repo_path or croak $!;
221 $ENV{GIT_DIR} = $repo_path . "/.git";
222 }
223 do_git_init_db();
224
225 Git::SVN->init($url);
226 }
227
228 sub cmd_fetch {
229 if (@_) {
230 die "Additional fetch arguments are no longer supported.\n",
231 "Use --follow-parent if you have moved/copied directories
232 instead.\n";
233 }
234 my $gs = Git::SVN->new;
235 $gs->fetch(parse_revision_argument());
236 if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
237 command_noisy(qw(update-ref refs/heads/master),
238 $gs->{last_commit});
239 }
240 }
241
242 sub cmd_set_tree {
243 my (@commits) = @_;
244 if ($_stdin || !@commits) {
245 print "Reading from stdin...\n";
246 @commits = ();
247 while (<STDIN>) {
248 if (/\b($sha1_short)\b/o) {
249 unshift @commits, $1;
250 }
251 }
252 }
253 my @revs;
254 foreach my $c (@commits) {
255 my @tmp = command('rev-parse',$c);
256 if (scalar @tmp == 1) {
257 push @revs, $tmp[0];
258 } elsif (scalar @tmp > 1) {
259 push @revs, reverse(command('rev-list',@tmp));
260 } else {
261 fatal "Failed to rev-parse $c\n";
262 }
263 }
264 my $gs = Git::SVN->new;
265 my ($r_last, $cmt_last) = $gs->last_rev_commit;
266 $gs->fetch;
267 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
268 fatal "There are new revisions that were fetched ",
269 "and need to be merged (or acknowledged) ",
270 "before committing.\nlast rev: $r_last\n",
271 " current: $gs->{last_rev}\n";
272 }
273 $gs->set_tree($_) foreach @revs;
274 print "Done committing ",scalar @revs," revisions to SVN\n";
275 }
276
277 sub cmd_dcommit {
278 my $head = shift;
279 my $gs = Git::SVN->new;
280 $head ||= 'HEAD';
281 my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
282 my $last_rev;
283 foreach my $d (reverse @refs) {
284 if (!verify_ref("$d~1")) {
285 fatal "Commit $d\n",
286 "has no parent commit, and therefore ",
287 "nothing to diff against.\n",
288 "You should be working from a repository ",
289 "originally created by git-svn\n";
290 }
291 unless (defined $last_rev) {
292 (undef, $last_rev, undef) = cmt_metadata("$d~1");
293 unless (defined $last_rev) {
294 fatal "Unable to extract revision information ",
295 "from commit $d~1\n";
296 }
297 }
298 if ($_dry_run) {
299 print "diff-tree $d~1 $d\n";
300 } else {
301 my %ed_opts = ( r => $last_rev,
302 log => get_commit_entry($d)->{log},
303 ra => $gs->ra,
304 tree_a => "$d~1",
305 tree_b => $d,
306 editor_cb => sub {
307 print "Committed r$_[0]\n";
308 $last_rev = $_[0]; },
309 svn_path => $gs->{path} );
310 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
311 print "No changes\n$d~1 == $d\n";
312 }
313 }
314 }
315 return if $_dry_run;
316 $gs->fetch;
317 # we always want to rebase against the current HEAD, not any
318 # head that was passed to us
319 my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
320 my @finish;
321 if (@diff) {
322 @finish = qw/rebase/;
323 push @finish, qw/--merge/ if $_merge;
324 push @finish, "--strategy=$_strategy" if $_strategy;
325 print STDERR "W: HEAD and ", $gs->refname, " differ, ",
326 "using @finish:\n", "@diff";
327 } else {
328 print "No changes between current HEAD and ",
329 $gs->refname, "\nResetting to the latest ",
330 $gs->refname, "\n";
331 @finish = qw/reset --mixed/;
332 }
333 command_noisy(@finish, $gs->refname);
334 }
335
336 sub cmd_show_ignore {
337 my $gs = Git::SVN->new;
338 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
339 $gs->traverse_ignore(\*STDOUT, '', $r);
340 }
341
342 sub cmd_multi_init {
343 my $url = shift;
344 unless (defined $_trunk || defined $_branches || defined $_tags) {
345 usage(1);
346 }
347 do_git_init_db();
348 $_prefix = '' unless defined $_prefix;
349 $url =~ s#/+$## if defined $url;
350 if (defined $_trunk) {
351 my $trunk_ref = $_prefix . 'trunk';
352 # try both old-style and new-style lookups:
353 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
354 unless ($gs_trunk) {
355 my ($trunk_url, $trunk_path) =
356 complete_svn_url($url, $_trunk);
357 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
358 undef, $trunk_ref);
359 }
360 }
361 return unless defined $_branches || defined $_tags;
362 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
363 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
364 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
365 }
366
367 sub cmd_multi_fetch {
368 my $remotes = Git::SVN::read_all_remotes();
369 foreach my $repo_id (sort keys %$remotes) {
370 if ($remotes->{$repo_id}->{url}) {
371 Git::SVN::fetch_all($repo_id, $remotes);
372 }
373 }
374 }
375
376 # this command is special because it requires no metadata
377 sub cmd_commit_diff {
378 my ($ta, $tb, $url) = @_;
379 my $usage = "Usage: $0 commit-diff -r<revision> ".
380 "<tree-ish> <tree-ish> [<URL>]\n";
381 fatal($usage) if (!defined $ta || !defined $tb);
382 my $svn_path;
383 if (!defined $url) {
384 my $gs = eval { Git::SVN->new };
385 if (!$gs) {
386 fatal("Needed URL or usable git-svn --id in ",
387 "the command-line\n", $usage);
388 }
389 $url = $gs->{url};
390 $svn_path = $gs->{path};
391 }
392 unless (defined $_revision) {
393 fatal("-r|--revision is a required argument\n", $usage);
394 }
395 if (defined $_message && defined $_file) {
396 fatal("Both --message/-m and --file/-F specified ",
397 "for the commit message.\n",
398 "I have no idea what you mean\n");
399 }
400 if (defined $_file) {
401 $_message = file_to_s($_file);
402 } else {
403 $_message ||= get_commit_entry($tb)->{log};
404 }
405 my $ra ||= Git::SVN::Ra->new($url);
406 $svn_path ||= $ra->{svn_path};
407 my $r = $_revision;
408 if ($r eq 'HEAD') {
409 $r = $ra->get_latest_revnum;
410 } elsif ($r !~ /^\d+$/) {
411 die "revision argument: $r not understood by git-svn\n";
412 }
413 my %ed_opts = ( r => $r,
414 log => $_message,
415 ra => $ra,
416 tree_a => $ta,
417 tree_b => $tb,
418 editor_cb => sub { print "Committed r$_[0]\n" },
419 svn_path => $svn_path );
420 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
421 print "No changes\n$ta == $tb\n";
422 }
423 }
424
425 ########################### utility functions #########################
426
427 sub parse_revision_argument {
428 if (!defined $_revision || $_revision eq 'BASE:HEAD') {
429 return (undef, undef);
430 }
431 return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
432 return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
433 return (undef, $1) if ($_revision =~ /^BASE:(\d+)$/);
434 return ($1, undef) if ($_revision =~ /^(\d+):HEAD$/);
435 die "revision argument: $_revision not understood by git-svn\n",
436 "Try using the command-line svn client instead\n";
437 }
438
439 sub complete_svn_url {
440 my ($url, $path) = @_;
441 $path =~ s#/+$##;
442 if ($path !~ m#^[a-z\+]+://#) {
443 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
444 fatal("E: '$path' is not a complete URL ",
445 "and a separate URL is not specified\n");
446 }
447 return ($url, $path);
448 }
449 return ($path, '');
450 }
451
452 sub complete_url_ls_init {
453 my ($ra, $repo_path, $switch, $pfx) = @_;
454 unless ($repo_path) {
455 print STDERR "W: $switch not specified\n";
456 return;
457 }
458 $repo_path =~ s#/+$##;
459 if ($repo_path =~ m#^[a-z\+]+://#) {
460 $ra = Git::SVN::Ra->new($repo_path);
461 $repo_path = '';
462 } else {
463 $repo_path =~ s#^/+##;
464 unless ($ra) {
465 fatal("E: '$repo_path' is not a complete URL ",
466 "and a separate URL is not specified\n");
467 }
468 }
469 my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
470 my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
471 my $url = $ra->{url};
472 my $remote_id;
473 my $remote_path;
474 foreach my $d (sort keys %$dirent) {
475 next if ($dirent->{$d}->kind != $SVN::Node::dir);
476 my $path = "$repo_path/$d";
477 my $ref = "$pfx$d";
478 my $gs = eval { Git::SVN->new($ref) };
479 # don't try to init already existing refs
480 unless ($gs) {
481 print "init $url/$path => $ref\n";
482 $gs = Git::SVN->init($url, $path, undef, $ref, 1);
483 }
484 if ($gs) {
485 my $k = "svn-remote.$gs->{repo_id}.url";
486 my $orig_url = eval {
487 command_oneline(qw/config --get/, $k)
488 };
489 if ($orig_url && ($orig_url ne $gs->{url})) {
490 die "$k already set: $orig_url\n",
491 "wanted to set to: $gs->{url}\n";
492 }
493 unless ($orig_url) {
494 command_oneline('config', $k, $gs->{url});
495 }
496 $remote_id = $gs->{repo_id};
497 last;
498 }
499 }
500 if (defined $remote_id) {
501 $remote_path = "$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='$pfx' must have a trailing slash '/'\n";
507 }
508 command_noisy('config', "svn-remote.$remote_id.$n",
509 "$remote_path:refs/remotes/$pfx*");
510 }
511 }
512
513 sub verify_ref {
514 my ($ref) = @_;
515 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
516 { STDERR => 0 }); };
517 }
518
519 sub get_tree_from_treeish {
520 my ($treeish) = @_;
521 # $treeish can be a symbolic ref, too:
522 my $type = command_oneline(qw/cat-file -t/, $treeish);
523 my $expected;
524 while ($type eq 'tag') {
525 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
526 }
527 if ($type eq 'commit') {
528 $expected = (grep /^tree /, command(qw/cat-file commit/,
529 $treeish))[0];
530 ($expected) = ($expected =~ /^tree ($sha1)$/o);
531 die "Unable to get tree from $treeish\n" unless $expected;
532 } elsif ($type eq 'tree') {
533 $expected = $treeish;
534 } else {
535 die "$treeish is a $type, expected tree, tag or commit\n";
536 }
537 return $expected;
538 }
539
540 sub get_commit_entry {
541 my ($treeish) = shift;
542 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
543 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
544 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
545 open my $log_fh, '>', $commit_editmsg or croak $!;
546
547 my $type = command_oneline(qw/cat-file -t/, $treeish);
548 if ($type eq 'commit' || $type eq 'tag') {
549 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
550 $type, $treeish);
551 my $in_msg = 0;
552 while (<$msg_fh>) {
553 if (!$in_msg) {
554 $in_msg = 1 if (/^\s*$/);
555 } elsif (/^git-svn-id: /) {
556 # skip this for now, we regenerate the
557 # correct one on re-fetch anyways
558 # TODO: set *:merge properties or like...
559 } else {
560 print $log_fh $_ or croak $!;
561 }
562 }
563 command_close_pipe($msg_fh, $ctx);
564 }
565 close $log_fh or croak $!;
566
567 if ($_edit || ($type eq 'tree')) {
568 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
569 # TODO: strip out spaces, comments, like git-commit.sh
570 system($editor, $commit_editmsg);
571 }
572 rename $commit_editmsg, $commit_msg or croak $!;
573 open $log_fh, '<', $commit_msg or croak $!;
574 { local $/; chomp($log_entry{log} = <$log_fh>); }
575 close $log_fh or croak $!;
576 unlink $commit_msg;
577 \%log_entry;
578 }
579
580 sub s_to_file {
581 my ($str, $file, $mode) = @_;
582 open my $fd,'>',$file or croak $!;
583 print $fd $str,"\n" or croak $!;
584 close $fd or croak $!;
585 chmod ($mode &~ umask, $file) if (defined $mode);
586 }
587
588 sub file_to_s {
589 my $file = shift;
590 open my $fd,'<',$file or croak "$!: file: $file\n";
591 local $/;
592 my $ret = <$fd>;
593 close $fd or croak $!;
594 $ret =~ s/\s*$//s;
595 return $ret;
596 }
597
598 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
599 sub load_authors {
600 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
601 my $log = $cmd eq 'log';
602 while (<$authors>) {
603 chomp;
604 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
605 my ($user, $name, $email) = ($1, $2, $3);
606 if ($log) {
607 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
608 } else {
609 $users{$user} = [$name, $email];
610 }
611 }
612 close $authors or croak $!;
613 }
614
615 # convert GetOpt::Long specs for use by git-config
616 sub read_repo_config {
617 return unless -d $ENV{GIT_DIR};
618 my $opts = shift;
619 my @config_only;
620 foreach my $o (keys %$opts) {
621 # if we have mixedCase and a long option-only, then
622 # it's a config-only variable that we don't need for
623 # the command-line.
624 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
625 my $v = $opts->{$o};
626 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
627 $key =~ s/-//g;
628 my $arg = 'git-config';
629 $arg .= ' --int' if ($o =~ /[:=]i$/);
630 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
631 if (ref $v eq 'ARRAY') {
632 chomp(my @tmp = `$arg --get-all svn.$key`);
633 @$v = @tmp if @tmp;
634 } else {
635 chomp(my $tmp = `$arg --get svn.$key`);
636 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
637 $$v = $tmp;
638 }
639 }
640 }
641 delete @$opts{@config_only} if @config_only;
642 }
643
644 sub extract_metadata {
645 my $id = shift or return (undef, undef, undef);
646 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
647 \s([a-f\d\-]+)$/x);
648 if (!defined $rev || !$uuid || !$url) {
649 # some of the original repositories I made had
650 # identifiers like this:
651 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
652 }
653 return ($url, $rev, $uuid);
654 }
655
656 sub cmt_metadata {
657 return extract_metadata((grep(/^git-svn-id: /,
658 command(qw/cat-file commit/, shift)))[-1]);
659 }
660
661 package Git::SVN;
662 use strict;
663 use warnings;
664 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
665 $_repack $_repack_flags $_use_svm_props/;
666 use Carp qw/croak/;
667 use File::Path qw/mkpath/;
668 use File::Copy qw/copy/;
669 use IPC::Open3;
670
671 my $_repack_nr;
672 # properties that we do not log:
673 my %SKIP_PROP;
674 BEGIN {
675 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
676 svn:special svn:executable
677 svn:entry:committed-rev
678 svn:entry:last-author
679 svn:entry:uuid
680 svn:entry:committed-date/;
681
682 # some options are read globally, but can be overridden locally
683 # per [svn-remote "..."] section. Command-line options will *NOT*
684 # override options set in an [svn-remote "..."] section
685 my $e;
686 foreach (qw/follow_parent no_metadata use_svm_props/) {
687 my $key = $_;
688 $key =~ tr/_//d;
689 $e .= "sub $_ {
690 my (\$self) = \@_;
691 return \$self->{-$_} if exists \$self->{-$_};
692 my \$k = \"svn-remote.\$self->{repo_id}\.$key\";
693 eval { command_oneline(qw/config --get/, \$k) };
694 if (\$@) {
695 \$self->{-$_} = \$Git::SVN::_$_;
696 } else {
697 my \$v = command_oneline(qw/config --bool/,\$k);
698 \$self->{-$_} = \$v eq 'false' ? 0 : 1;
699 }
700 return \$self->{-$_} }\n";
701 }
702 $e .= "1;\n";
703 eval $e or die $@;
704 }
705
706 my %LOCKFILES;
707 END { unlink keys %LOCKFILES if %LOCKFILES }
708
709 sub resolve_local_globs {
710 my ($url, $fetch, $glob_spec) = @_;
711 return unless defined $glob_spec;
712 my $ref = $glob_spec->{ref};
713 my $path = $glob_spec->{path};
714 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
715 next unless m#^refs/remotes/$ref->{regex}$#;
716 my $p = $1;
717 my $pathname = $path->full_path($p);
718 my $refname = $ref->full_path($p);
719 if (my $existing = $fetch->{$pathname}) {
720 if ($existing ne $refname) {
721 die "Refspec conflict:\n",
722 "existing: refs/remotes/$existing\n",
723 " globbed: refs/remotes/$refname\n";
724 }
725 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
726 $u =~ s!^\Q$url\E(/|$)!! or die
727 "refs/remotes/$refname: '$url' not found in '$u'\n";
728 if ($pathname ne $u) {
729 warn "W: Refspec glob conflict ",
730 "(ref: refs/remotes/$refname):\n",
731 "expected path: $pathname\n",
732 " real path: $u\n",
733 "Continuing ahead with $u\n";
734 next;
735 }
736 } else {
737 $fetch->{$pathname} = $refname;
738 }
739 }
740 }
741
742 sub fetch_all {
743 my ($repo_id, $remotes) = @_;
744 my $remote = $remotes->{$repo_id};
745 my $fetch = $remote->{fetch};
746 my $url = $remote->{url};
747 my (@gs, @globs);
748 my $ra = Git::SVN::Ra->new($url);
749 my $uuid = $ra->get_uuid;
750 my $head = $ra->get_latest_revnum;
751 my $base = $head;
752
753 # read the max revs for wildcard expansion (branches/*, tags/*)
754 foreach my $t (qw/branches tags/) {
755 defined $remote->{$t} or next;
756 push @globs, $remote->{$t};
757 my $max_rev = eval { tmp_config(qw/--int --get/,
758 "svn-remote.$repo_id.${t}-maxRev") };
759 if (defined $max_rev && ($max_rev < $base)) {
760 $base = $max_rev;
761 }
762 }
763
764 if ($fetch) {
765 foreach my $p (sort keys %$fetch) {
766 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
767 my $lr = $gs->rev_db_max;
768 if (defined $lr) {
769 $base = $lr if ($lr < $base);
770 }
771 push @gs, $gs;
772 }
773 }
774 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
775 }
776
777 sub read_all_remotes {
778 my $r = {};
779 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
780 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
781 $r->{$1}->{fetch}->{$2} = $3;
782 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
783 $r->{$1}->{url} = $2;
784 } elsif (m!^(.+)\.(branches|tags)=
785 (.*):refs/remotes/(.+)\s*$/!x) {
786 my ($p, $g) = ($3, $4);
787 my $rs = $r->{$1}->{$2} = {
788 t => $2,
789 remote => $1,
790 path => Git::SVN::GlobSpec->new($p),
791 ref => Git::SVN::GlobSpec->new($g) };
792 if (length($rs->{ref}->{right}) != 0) {
793 die "The '*' glob character must be the last ",
794 "character of '$g'\n";
795 }
796 }
797 }
798 $r;
799 }
800
801 sub init_vars {
802 if (defined $_repack) {
803 $_repack = 1000 if ($_repack <= 0);
804 $_repack_nr = $_repack;
805 $_repack_flags ||= '-d';
806 }
807 }
808
809 sub verify_remotes_sanity {
810 return unless -d $ENV{GIT_DIR};
811 my %seen;
812 foreach (command(qw/config -l/)) {
813 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
814 if ($seen{$1}) {
815 die "Remote ref refs/remote/$1 is tracked by",
816 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
817 "Please resolve this ambiguity in ",
818 "your git configuration file before ",
819 "continuing\n";
820 }
821 $seen{$1} = $_;
822 }
823 }
824 }
825
826 # we allow more chars than remotes2config.sh...
827 sub sanitize_remote_name {
828 my ($name) = @_;
829 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
830 $name;
831 }
832
833 sub find_existing_remote {
834 my ($url, $remotes) = @_;
835 my $existing;
836 foreach my $repo_id (keys %$remotes) {
837 my $u = $remotes->{$repo_id}->{url} or next;
838 next if $u ne $url;
839 $existing = $repo_id;
840 last;
841 }
842 $existing;
843 }
844
845 sub init_remote_config {
846 my ($self, $url, $no_write) = @_;
847 $url =~ s!/+$!!; # strip trailing slash
848 my $r = read_all_remotes();
849 my $existing = find_existing_remote($url, $r);
850 if ($existing) {
851 unless ($no_write) {
852 print STDERR "Using existing ",
853 "[svn-remote \"$existing\"]\n";
854 }
855 $self->{repo_id} = $existing;
856 } else {
857 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
858 $existing = find_existing_remote($min_url, $r);
859 if ($existing) {
860 unless ($no_write) {
861 print STDERR "Using existing ",
862 "[svn-remote \"$existing\"]\n";
863 }
864 $self->{repo_id} = $existing;
865 }
866 if ($min_url ne $url) {
867 unless ($no_write) {
868 print STDERR "Using higher level of URL: ",
869 "$url => $min_url\n";
870 }
871 my $old_path = $self->{path};
872 $self->{path} = $url;
873 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
874 if (length $old_path) {
875 $self->{path} .= "/$old_path";
876 }
877 $url = $min_url;
878 }
879 }
880 my $orig_url;
881 if (!$existing) {
882 # verify that we aren't overwriting anything:
883 $orig_url = eval {
884 command_oneline('config', '--get',
885 "svn-remote.$self->{repo_id}.url")
886 };
887 if ($orig_url && ($orig_url ne $url)) {
888 die "svn-remote.$self->{repo_id}.url already set: ",
889 "$orig_url\nwanted to set to: $url\n";
890 }
891 }
892 my ($xrepo_id, $xpath) = find_ref($self->refname);
893 if (defined $xpath) {
894 die "svn-remote.$xrepo_id.fetch already set to track ",
895 "$xpath:refs/remotes/", $self->refname, "\n";
896 }
897 unless ($no_write) {
898 command_noisy('config',
899 "svn-remote.$self->{repo_id}.url", $url);
900 command_noisy('config', '--add',
901 "svn-remote.$self->{repo_id}.fetch",
902 "$self->{path}:".$self->refname);
903 }
904 $self->{url} = $url;
905 }
906
907 sub init {
908 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
909 my $self = _new($class, $repo_id, $ref_id, $path);
910 if (defined $url) {
911 $self->init_remote_config($url, $no_write);
912 }
913 $self;
914 }
915
916 sub find_ref {
917 my ($ref_id) = @_;
918 foreach (command(qw/config -l/)) {
919 next unless m!^svn-remote\.(.+)\.fetch=
920 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
921 my ($repo_id, $path, $ref) = ($1, $2, $3);
922 if ($ref eq $ref_id) {
923 $path = '' if ($path =~ m#^\./?#);
924 return ($repo_id, $path);
925 }
926 }
927 (undef, undef, undef);
928 }
929
930 sub new {
931 my ($class, $ref_id, $repo_id, $path) = @_;
932 if (defined $ref_id && !defined $repo_id && !defined $path) {
933 ($repo_id, $path) = find_ref($ref_id);
934 if (!defined $repo_id) {
935 die "Could not find a \"svn-remote.*.fetch\" key ",
936 "in the repository configuration matching: ",
937 "refs/remotes/$ref_id\n";
938 }
939 }
940 my $self = _new($class, $repo_id, $ref_id, $path);
941 if (!defined $self->{path} || !length $self->{path}) {
942 my $fetch = command_oneline('config', '--get',
943 "svn-remote.$repo_id.fetch",
944 ":refs/remotes/$ref_id\$") or
945 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
946 "\":refs/remotes/$ref_id\$\" in config\n";
947 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
948 }
949 $self->{url} = command_oneline('config', '--get',
950 "svn-remote.$repo_id.url") or
951 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
952 if ((-z $self->db_path || ! -e $self->db_path) &&
953 ::verify_ref($self->refname.'^0')) {
954 $self->rebuild;
955 }
956 $self;
957 }
958
959 sub refname { "refs/remotes/$_[0]->{ref_id}" }
960
961 sub svm_uuid {
962 my ($self) = @_;
963 return $self->{svm}->{uuid} if $self->svm;
964 $self->ra;
965 unless ($self->{svm}) {
966 die "SVM UUID not cached, and reading remotely failed\n";
967 }
968 $self->{svm}->{uuid};
969 }
970
971 sub svm {
972 my ($self) = @_;
973 return $self->{svm} if $self->{svm};
974 my $svm;
975 # see if we have it in our config, first:
976 eval {
977 my $section = "svn-remote.$self->{repo_id}";
978 $svm = {
979 source => tmp_config('--get', "$section.svm-source"),
980 uuid => tmp_config('--get', "$section.svm-uuid"),
981 }
982 };
983 $self->{svm} = $svm if ($svm && $svm->{source} && $svm->{uuid});
984 $self->{svm};
985 }
986
987 sub _set_svm_vars {
988 my ($self, $ra) = @_;
989 return $ra if $self->svm;
990
991 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
992 "(svm:source, svm:mirror, svm:mirror) ",
993 "from the following URLs:\n" );
994 sub read_svm_props {
995 my ($self, $props) = @_;
996 my $src = $props->{'svm:source'};
997 my $mirror = $props->{'svm:mirror'};
998 my $uuid = $props->{'svm:uuid'};
999 return undef if (!$src || !$mirror || !$uuid);
1000
1001 chomp($src, $mirror, $uuid);
1002
1003 $uuid =~ m{^[0-9a-f\-]{30,}$}
1004 or die "doesn't look right - svm:uuid is '$uuid'\n";
1005 # don't know what a '!' is there for, also the
1006 # username is of no interest
1007 $src =~ s{/?!$}{$mirror};
1008 $src =~ s{/+$}{}; # no trailing slashes please
1009 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1010
1011 my $section = "svn-remote.$self->{repo_id}";
1012 tmp_config('--add', "$section.svm-source", $src);
1013 tmp_config('--add', "$section.svm-uuid", $uuid);
1014 $self->{svm} = { source => $src , uuid => $uuid };
1015 return 1;
1016 }
1017
1018 my $r = $ra->get_latest_revnum;
1019 my $path = $self->{path};
1020 my @tried_a = ($path);
1021 while (length $path) {
1022 if ($self->read_svm_props(($ra->get_dir($path, $r))[2])) {
1023 return $ra;
1024 }
1025 $path =~ s#/?[^/]+$## && push @tried_a, $path;
1026 }
1027 if ($self->read_svm_props(($ra->get_dir('', $r))[2])) {
1028 return $ra;
1029 }
1030
1031 if ($ra->{repos_root} eq $self->{url}) {
1032 die @err, map { " $self->{url}/$_\n" } @tried_a, "\n";
1033 }
1034
1035 # nope, make sure we're connected to the repository root:
1036 my $ok;
1037 my @tried_b;
1038 $path = $ra->{svn_path};
1039 $path =~ s#/?[^/]+$##; # we already tried this one above
1040 $ra = Git::SVN::Ra->new($ra->{repos_root});
1041 while (length $path) {
1042 $ok = $self->read_svm_props(($ra->get_dir($path, $r))[2]);
1043 last if $ok;
1044 $path =~ s#/?[^/]+$## && push @tried_b, $path;
1045 }
1046 $ok = $self->read_svm_props(($ra->get_dir('', $r))[2]) unless $ok;
1047 if (!$ok) {
1048 die @err, map { " $self->{url}/$_\n" } @tried_a, "\n",
1049 map { " $ra->{url}/$_\n" } @tried_b, "\n"
1050 }
1051 Git::SVN::Ra->new($self->{url});
1052 }
1053
1054 # this allows us to memoize our SVN::Ra UUID locally and avoid a
1055 # remote lookup (useful for 'git svn log').
1056 sub ra_uuid {
1057 my ($self) = @_;
1058 unless ($self->{ra_uuid}) {
1059 my $key = "svn-remote.$self->{repo_id}.uuid";
1060 my $uuid = eval { tmp_config('--get', $key) };
1061 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1062 $self->{ra_uuid} = $uuid;
1063 } else {
1064 die "ra_uuid called without URL\n" unless $self->{url};
1065 $self->{ra_uuid} = $self->ra->get_uuid;
1066 tmp_config('--add', $key, $self->{ra_uuid});
1067 }
1068 }
1069 $self->{ra_uuid};
1070 }
1071
1072 sub ra {
1073 my ($self) = shift;
1074 my $ra = Git::SVN::Ra->new($self->{url});
1075 if ($self->use_svm_props && !$self->{svm}) {
1076 if ($self->no_metadata) {
1077 die "Can't have both 'noMetadata' and ",
1078 "'useSvmProps' options set!\n";
1079 }
1080 $ra = $self->_set_svm_vars($ra);
1081 $self->{-want_revprops} = 1;
1082 }
1083 $ra;
1084 }
1085
1086 sub rel_path {
1087 my ($self) = @_;
1088 my $repos_root = $self->ra->{repos_root};
1089 return $self->{path} if ($self->{url} eq $repos_root);
1090 die "BUG: rel_path failed! repos_root: $repos_root, Ra URL: ",
1091 $self->ra->{url}, " path: $self->{path}, URL: $self->{url}\n";
1092 }
1093
1094 sub traverse_ignore {
1095 my ($self, $fh, $path, $r) = @_;
1096 $path =~ s#^/+##g;
1097 my $ra = $self->ra;
1098 my ($dirent, undef, $props) = $ra->get_dir($path, $r);
1099 my $p = $path;
1100 $p =~ s#^\Q$ra->{svn_path}\E/##;
1101 print $fh length $p ? "\n# $p\n" : "\n# /\n";
1102 if (my $s = $props->{'svn:ignore'}) {
1103 $s =~ s/[\r\n]+/\n/g;
1104 chomp $s;
1105 if (length $p == 0) {
1106 $s =~ s#\n#\n/$p#g;
1107 print $fh "/$s\n";
1108 } else {
1109 $s =~ s#\n#\n/$p/#g;
1110 print $fh "/$p/$s\n";
1111 }
1112 }
1113 foreach (sort keys %$dirent) {
1114 next if $dirent->{$_}->kind != $SVN::Node::dir;
1115 $self->traverse_ignore($fh, "$path/$_", $r);
1116 }
1117 }
1118
1119 sub last_rev { ($_[0]->last_rev_commit)[0] }
1120 sub last_commit { ($_[0]->last_rev_commit)[1] }
1121
1122 # returns the newest SVN revision number and newest commit SHA1
1123 sub last_rev_commit {
1124 my ($self) = @_;
1125 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1126 return ($self->{last_rev}, $self->{last_commit});
1127 }
1128 my $c = ::verify_ref($self->refname.'^0');
1129 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1130 my $rev = (::cmt_metadata($c))[1];
1131 if (defined $rev) {
1132 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1133 return ($rev, $c);
1134 }
1135 }
1136 my $db_path = $self->db_path;
1137 unless (-e $db_path) {
1138 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1139 return (undef, undef);
1140 }
1141 my $offset = -41; # from tail
1142 my $rl;
1143 open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1144 sysseek($fh, $offset, 2); # don't care for errors
1145 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1146 chomp $rl;
1147 while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1148 $offset -= 41;
1149 sysseek($fh, $offset, 2); # don't care for errors
1150 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1151 chomp $rl;
1152 }
1153 if ($c && $c ne $rl) {
1154 die "$db_path and ", $self->refname,
1155 " inconsistent!:\n$c != $rl\n";
1156 }
1157 my $rev = sysseek($fh, 0, 1) or croak $!;
1158 $rev = ($rev - 41) / 41;
1159 close $fh or croak $!;
1160 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1161 return ($rev, $c);
1162 }
1163
1164 sub get_fetch_range {
1165 my ($self, $min, $max) = @_;
1166 $max ||= $self->ra->get_latest_revnum;
1167 $min ||= $self->rev_db_max;
1168 (++$min, $max);
1169 }
1170
1171 sub tmp_config {
1172 my (@args) = @_;
1173 my $config = "$ENV{GIT_DIR}/svn/config";
1174 unless (-f $config) {
1175 open my $fh, '>', $config or
1176 die "Can't open $config: $!\n";
1177 print $fh "; This file is used internally by git-svn\n" or
1178 die "Couldn't write to $config: $!\n";
1179 print $fh "; You should not have to edit it\n" or
1180 die "Couldn't write to $config: $!\n";
1181 close $fh or die "Couldn't close $config: $!\n";
1182 }
1183 my $old_config = $ENV{GIT_CONFIG};
1184 $ENV{GIT_CONFIG} = $config;
1185 $@ = undef;
1186 my @ret = eval { command('config', @args) };
1187 my $err = $@;
1188 if (defined $old_config) {
1189 $ENV{GIT_CONFIG} = $old_config;
1190 } else {
1191 delete $ENV{GIT_CONFIG};
1192 }
1193 die $err if $err;
1194 wantarray ? @ret : $ret[0];
1195 }
1196
1197 sub tmp_index_do {
1198 my ($self, $sub) = @_;
1199 my $old_index = $ENV{GIT_INDEX_FILE};
1200 $ENV{GIT_INDEX_FILE} = $self->{index};
1201 $@ = undef;
1202 my @ret = eval { &$sub };
1203 my $err = $@;
1204 if (defined $old_index) {
1205 $ENV{GIT_INDEX_FILE} = $old_index;
1206 } else {
1207 delete $ENV{GIT_INDEX_FILE};
1208 }
1209 die $err if $err;
1210 wantarray ? @ret : $ret[0];
1211 }
1212
1213 sub assert_index_clean {
1214 my ($self, $treeish) = @_;
1215
1216 $self->tmp_index_do(sub {
1217 command_noisy('read-tree', $treeish) unless -e $self->{index};
1218 my $x = command_oneline('write-tree');
1219 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1220 /^tree ($::sha1)/mo);
1221 if ($y ne $x) {
1222 unlink $self->{index} or croak $!;
1223 command_noisy('read-tree', $treeish);
1224 }
1225 $x = command_oneline('write-tree');
1226 if ($y ne $x) {
1227 ::fatal "trees ($treeish) $y != $x\n",
1228 "Something is seriously wrong...\n";
1229 }
1230 });
1231 }
1232
1233 sub get_commit_parents {
1234 my ($self, $log_entry) = @_;
1235 my (%seen, @ret, @tmp);
1236 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1237 if (my $ip = $self->{inject_parents}) {
1238 if (my $commit = delete $ip->{$log_entry->{revision}}) {
1239 push @tmp, $commit;
1240 }
1241 }
1242 if (my $cur = ::verify_ref($self->refname.'^0')) {
1243 push @tmp, $cur;
1244 }
1245 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1246 while (my $p = shift @tmp) {
1247 next if $seen{$p};
1248 $seen{$p} = 1;
1249 push @ret, $p;
1250 # MAXPARENT is defined to 16 in commit-tree.c:
1251 last if @ret >= 16;
1252 }
1253 if (@tmp) {
1254 die "r$log_entry->{revision}: No room for parents:\n\t",
1255 join("\n\t", @tmp), "\n";
1256 }
1257 @ret;
1258 }
1259
1260 sub full_url {
1261 my ($self) = @_;
1262 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1263 }
1264
1265 sub do_git_commit {
1266 my ($self, $log_entry) = @_;
1267 my $lr = $self->last_rev;
1268 if (defined $lr && $lr >= $log_entry->{revision}) {
1269 die "Last fetched revision of ", $self->refname,
1270 " was r$lr, but we are about to fetch: ",
1271 "r$log_entry->{revision}!\n";
1272 }
1273 if (my $c = $self->rev_db_get($log_entry->{revision})) {
1274 croak "$log_entry->{revision} = $c already exists! ",
1275 "Why are we refetching it?\n";
1276 }
1277 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1278 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1279 $log_entry->{email};
1280 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1281
1282 my $tree = $log_entry->{tree};
1283 if (!defined $tree) {
1284 $tree = $self->tmp_index_do(sub {
1285 command_oneline('write-tree') });
1286 }
1287 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1288
1289 my @exec = ('git-commit-tree', $tree);
1290 foreach ($self->get_commit_parents($log_entry)) {
1291 push @exec, '-p', $_;
1292 }
1293 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1294 or croak $!;
1295 print $msg_fh $log_entry->{log} or croak $!;
1296 unless ($self->no_metadata) {
1297 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1298 or croak $!;
1299 }
1300 $msg_fh->flush == 0 or croak $!;
1301 close $msg_fh or croak $!;
1302 chomp(my $commit = do { local $/; <$out_fh> });
1303 close $out_fh or croak $!;
1304 waitpid $pid, 0;
1305 croak $? if $?;
1306 if ($commit !~ /^$::sha1$/o) {
1307 die "Failed to commit, invalid sha1: $commit\n";
1308 }
1309
1310 $self->rev_db_set($log_entry->{revision}, $commit, 1);
1311
1312 $self->{last_rev} = $log_entry->{revision};
1313 $self->{last_commit} = $commit;
1314 print "r$log_entry->{revision}";
1315 if (defined $log_entry->{svm_revision}) {
1316 print " (\@$log_entry->{svm_revision})";
1317 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1318 0, $self->svm_uuid);
1319 }
1320 print " = $commit ($self->{ref_id})\n";
1321 if (defined $_repack && (--$_repack_nr == 0)) {
1322 $_repack_nr = $_repack;
1323 # repack doesn't use any arguments with spaces in them, does it?
1324 print "Running git repack $_repack_flags ...\n";
1325 command_noisy('repack', split(/\s+/, $_repack_flags));
1326 print "Done repacking\n";
1327 }
1328 return $commit;
1329 }
1330
1331 sub match_paths {
1332 my ($self, $paths, $r) = @_;
1333 return 1 if $self->{path} eq '';
1334 if (my $path = $paths->{"/$self->{path}"}) {
1335 return ($path->{action} eq 'D') ? 0 : 1;
1336 }
1337 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1338 if (grep /$self->{path_regex}/, keys %$paths) {
1339 return 1;
1340 }
1341 my $c = '';
1342 foreach (split m#/#, $self->{path}) {
1343 $c .= "/$_";
1344 next unless ($paths->{$c} &&
1345 ($paths->{$c}->{action} =~ /^[AR]$/));
1346 if ($self->ra->check_path($self->{path}, $r) ==
1347 $SVN::Node::dir) {
1348 return 1;
1349 }
1350 }
1351 return 0;
1352 }
1353
1354 sub find_parent_branch {
1355 my ($self, $paths, $rev) = @_;
1356 return undef unless $self->follow_parent;
1357 unless (defined $paths) {
1358 my $err_handler = $SVN::Error::handler;
1359 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1360 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1361 $paths =
1362 Git::SVN::Ra::dup_changed_paths($_[0]) });
1363 $SVN::Error::handler = $err_handler;
1364 }
1365 return undef unless defined $paths;
1366
1367 # look for a parent from another branch:
1368 my @b_path_components = split m#/#, $self->rel_path;
1369 my @a_path_components;
1370 my $i;
1371 while (@b_path_components) {
1372 $i = $paths->{'/'.join('/', @b_path_components)};
1373 last if $i && defined $i->{copyfrom_path};
1374 unshift(@a_path_components, pop(@b_path_components));
1375 }
1376 return undef unless defined $i && defined $i->{copyfrom_path};
1377 my $branch_from = $i->{copyfrom_path};
1378 if (@a_path_components) {
1379 print STDERR "branch_from: $branch_from => ";
1380 $branch_from .= '/'.join('/', @a_path_components);
1381 print STDERR $branch_from, "\n";
1382 }
1383 my $r = $i->{copyfrom_rev};
1384 my $repos_root = $self->ra->{repos_root};
1385 my $url = $self->ra->{url};
1386 my $new_url = $repos_root . $branch_from;
1387 print STDERR "Found possible branch point: ",
1388 "$new_url => ", $self->full_url, ", $r\n";
1389 $branch_from =~ s#^/##;
1390 my $remotes = read_all_remotes();
1391 my $gs;
1392 foreach my $repo_id (keys %$remotes) {
1393 my $u = $remotes->{$repo_id}->{url} or next;
1394 next if $url ne $u;
1395 my $fetch = $remotes->{$repo_id}->{fetch};
1396 foreach (qw/branches tags/) {
1397 resolve_local_globs($url, $fetch,
1398 $remotes->{$repo_id}->{$_});
1399 }
1400 foreach my $f (keys %$fetch) {
1401 next if $f ne $branch_from;
1402 $gs = Git::SVN->new($fetch->{$f}, $repo_id, $f);
1403 last;
1404 }
1405 last if $gs;
1406 }
1407 unless ($gs) {
1408 my $ref_id = $self->{ref_id};
1409 $ref_id =~ s/\@\d+$//;
1410 $ref_id .= "\@$r";
1411 # just grow a tail if we're not unique enough :x
1412 $ref_id .= '-' while find_ref($ref_id);
1413 print STDERR "Initializing parent: $ref_id\n";
1414 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1415 }
1416 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1417 if (!defined $r0 || !defined $parent) {
1418 $gs->fetch(0, $r);
1419 ($r0, $parent) = $gs->last_rev_commit;
1420 }
1421 if (defined $r0 && defined $parent) {
1422 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1423 $self->assert_index_clean($parent);
1424 my $ed;
1425 if ($self->ra->can_do_switch) {
1426 print STDERR "Following parent with do_switch\n";
1427 # do_switch works with svn/trunk >= r22312, but that
1428 # is not included with SVN 1.4.3 (the latest version
1429 # at the moment), so we can't rely on it
1430 $self->{last_commit} = $parent;
1431 $ed = SVN::Git::Fetcher->new($self);
1432 $gs->ra->gs_do_switch($r0, $rev, $gs,
1433 $self->full_url, $ed)
1434 or die "SVN connection failed somewhere...\n";
1435 } else {
1436 print STDERR "Following parent with do_update\n";
1437 $ed = SVN::Git::Fetcher->new($self);
1438 $self->ra->gs_do_update($rev, $rev, $self, $ed)
1439 or die "SVN connection failed somewhere...\n";
1440 }
1441 print STDERR "Successfully followed parent\n";
1442 return $self->make_log_entry($rev, [$parent], $ed);
1443 }
1444 return undef;
1445 }
1446
1447 sub do_fetch {
1448 my ($self, $paths, $rev) = @_;
1449 my $ed;
1450 my ($last_rev, @parents);
1451 if (my $lc = $self->last_commit) {
1452 # we can have a branch that was deleted, then re-added
1453 # under the same name but copied from another path, in
1454 # which case we'll have multiple parents (we don't
1455 # want to break the original ref, nor lose copypath info):
1456 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1457 push @{$log_entry->{parents}}, $lc;
1458 return $log_entry;
1459 }
1460 $ed = SVN::Git::Fetcher->new($self);
1461 $last_rev = $self->{last_rev};
1462 $ed->{c} = $lc;
1463 @parents = ($lc);
1464 } else {
1465 $last_rev = $rev;
1466 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1467 return $log_entry;
1468 }
1469 $ed = SVN::Git::Fetcher->new($self);
1470 }
1471 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1472 die "SVN connection failed somewhere...\n";
1473 }
1474 $self->make_log_entry($rev, \@parents, $ed);
1475 }
1476
1477 sub get_untracked {
1478 my ($self, $ed) = @_;
1479 my @out;
1480 my $h = $ed->{empty};
1481 foreach (sort keys %$h) {
1482 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1483 push @out, " $act: " . uri_encode($_);
1484 warn "W: $act: $_\n";
1485 }
1486 foreach my $t (qw/dir_prop file_prop/) {
1487 $h = $ed->{$t} or next;
1488 foreach my $path (sort keys %$h) {
1489 my $ppath = $path eq '' ? '.' : $path;
1490 foreach my $prop (sort keys %{$h->{$path}}) {
1491 next if $SKIP_PROP{$prop};
1492 my $v = $h->{$path}->{$prop};
1493 my $t_ppath_prop = "$t: " .
1494 uri_encode($ppath) . ' ' .
1495 uri_encode($prop);
1496 if (defined $v) {
1497 push @out, " +$t_ppath_prop " .
1498 uri_encode($v);
1499 } else {
1500 push @out, " -$t_ppath_prop";
1501 }
1502 }
1503 }
1504 }
1505 foreach my $t (qw/absent_file absent_directory/) {
1506 $h = $ed->{$t} or next;
1507 foreach my $parent (sort keys %$h) {
1508 foreach my $path (sort @{$h->{$parent}}) {
1509 push @out, " $t: " .
1510 uri_encode("$parent/$path");
1511 warn "W: $t: $parent/$path ",
1512 "Insufficient permissions?\n";
1513 }
1514 }
1515 }
1516 \@out;
1517 }
1518
1519 sub parse_svn_date {
1520 my $date = shift || return '+0000 1970-01-01 00:00:00';
1521 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1522 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1523 croak "Unable to parse date: $date\n";
1524 "+0000 $Y-$m-$d $H:$M:$S";
1525 }
1526
1527 sub check_author {
1528 my ($author) = @_;
1529 if (!defined $author || length $author == 0) {
1530 $author = '(no author)';
1531 }
1532 if (defined $::_authors && ! defined $::users{$author}) {
1533 die "Author: $author not defined in $::_authors file\n";
1534 }
1535 $author;
1536 }
1537
1538 sub make_log_entry {
1539 my ($self, $rev, $parents, $ed) = @_;
1540 my $untracked = $self->get_untracked($ed);
1541
1542 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1543 print $un "r$rev\n" or croak $!;
1544 print $un $_, "\n" foreach @$untracked;
1545 my %log_entry = ( parents => $parents || [], revision => $rev,
1546 log => '');
1547
1548 my $headrev;
1549 my $logged = delete $self->{logged_rev_props};
1550 if (!$logged || $self->{-want_revprops}) {
1551 my $rp = $self->ra->rev_proplist($rev);
1552 foreach (sort keys %$rp) {
1553 my $v = $rp->{$_};
1554 if (/^svn:(author|date|log)$/) {
1555 $log_entry{$1} = $v;
1556 } elsif ($_ eq 'svm:headrev') {
1557 $headrev = $v;
1558 } else {
1559 print $un " rev_prop: ", uri_encode($_), ' ',
1560 uri_encode($v), "\n";
1561 }
1562 }
1563 } else {
1564 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1565 }
1566 close $un or croak $!;
1567
1568 $log_entry{date} = parse_svn_date($log_entry{date});
1569 $log_entry{log} .= "\n";
1570 my $author = $log_entry{author} = check_author($log_entry{author});
1571 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1572 : ($author, undef);
1573 if (defined $headrev && $self->use_svm_props) {
1574 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1575 if ($uuid ne $self->{svm}->{uuid}) {
1576 die "UUID mismatch on SVM path:\n",
1577 "expected: $self->{svm}->{uuid}\n",
1578 " got: $uuid\n";
1579 }
1580 my $full_url = $self->{svm}->{source};
1581 $full_url .= "/$self->{path}" if length $self->{path};
1582 $log_entry{metadata} = "$full_url\@$r $uuid";
1583 $log_entry{svm_revision} = $r;
1584 $email ||= "$author\@$uuid"
1585 } else {
1586 $log_entry{metadata} = $self->full_url . "\@$rev " .
1587 $self->ra->get_uuid;
1588 $email ||= "$author\@" . $self->ra->get_uuid;
1589 }
1590 $log_entry{name} = $name;
1591 $log_entry{email} = $email;
1592 \%log_entry;
1593 }
1594
1595 sub fetch {
1596 my ($self, $min_rev, $max_rev, @parents) = @_;
1597 my ($last_rev, $last_commit) = $self->last_rev_commit;
1598 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1599 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1600 }
1601
1602 sub set_tree_cb {
1603 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1604 $self->{inject_parents} = { $rev => $tree };
1605 $self->fetch(undef, undef);
1606 }
1607
1608 sub set_tree {
1609 my ($self, $tree) = (shift, shift);
1610 my $log_entry = ::get_commit_entry($tree);
1611 unless ($self->{last_rev}) {
1612 fatal("Must have an existing revision to commit\n");
1613 }
1614 my %ed_opts = ( r => $self->{last_rev},
1615 log => $log_entry->{log},
1616 ra => $self->ra,
1617 tree_a => $self->{last_commit},
1618 tree_b => $tree,
1619 editor_cb => sub {
1620 $self->set_tree_cb($log_entry, $tree, @_) },
1621 svn_path => $self->{path} );
1622 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1623 print "No changes\nr$self->{last_rev} = $tree\n";
1624 }
1625 }
1626
1627 sub rebuild {
1628 my ($self) = @_;
1629 my $db_path = $self->db_path;
1630 if (-f $self->{db_root}) {
1631 rename $self->{db_root}, $db_path or die
1632 "rename $self->{db_root} => $db_path failed: $!\n";
1633 my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
1634 symlink $base, $self->{db_root} or die
1635 "symlink $base => $self->{db_root} failed: $!\n";
1636 return;
1637 }
1638 print "Rebuilding $db_path ...\n";
1639 my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1640 my $latest;
1641 my $full_url = $self->full_url;
1642 my $svn_uuid;
1643 while (<$rev_list>) {
1644 chomp;
1645 my $c = $_;
1646 die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1647 my ($url, $rev, $uuid) = ::cmt_metadata($c);
1648
1649 # ignore merges (from set-tree)
1650 next if (!defined $rev || !$uuid);
1651
1652 # if we merged or otherwise started elsewhere, this is
1653 # how we break out of it
1654 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1655 ($full_url && $url && ($url ne $full_url))) {
1656 next;
1657 }
1658 $latest ||= $rev;
1659 $svn_uuid ||= $uuid;
1660
1661 $self->rev_db_set($rev, $c);
1662 print "r$rev = $c\n";
1663 }
1664 command_close_pipe($rev_list, $ctx);
1665 print "Done rebuilding $db_path\n";
1666 }
1667
1668 # rev_db:
1669 # Tie::File seems to be prone to offset errors if revisions get sparse,
1670 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
1671 # one of my favorite modules is out :< Next up would be one of the DBM
1672 # modules, but I'm not sure which is most portable... So I'll just
1673 # go with something that's plain-text, but still capable of
1674 # being randomly accessed. So here's my ultra-simple fixed-width
1675 # database. All records are 40 characters + "\n", so it's easy to seek
1676 # to a revision: (41 * rev) is the byte offset.
1677 # A record of 40 0s denotes an empty revision.
1678 # And yes, it's still pretty fast (faster than Tie::File).
1679 # These files are disposable unless noMetadata or useSvmProps is set
1680
1681 sub _rev_db_set {
1682 my ($fh, $rev, $commit) = @_;
1683 my $offset = $rev * 41;
1684 # assume that append is the common case:
1685 seek $fh, 0, 2 or croak $!;
1686 my $pos = tell $fh;
1687 if ($pos < $offset) {
1688 for (1 .. (($offset - $pos) / 41)) {
1689 print $fh (('0' x 40),"\n") or croak $!;
1690 }
1691 }
1692 seek $fh, $offset, 0 or croak $!;
1693 print $fh $commit,"\n" or croak $!;
1694 }
1695
1696 sub mkfile {
1697 my ($path) = @_;
1698 unless (-e $path) {
1699 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
1700 mkpath([$dir]) unless -d $dir;
1701 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
1702 close $fh or die "Couldn't close (create) $path: $!\n";
1703 }
1704 }
1705
1706 sub rev_db_set {
1707 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
1708 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
1709 my $db = $self->db_path($uuid);
1710 my $db_lock = "$db.lock";
1711 my $sig;
1712 if ($update_ref) {
1713 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1714 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
1715 }
1716 mkfile($db);
1717
1718 $LOCKFILES{$db_lock} = 1;
1719 my $sync;
1720 # both of these options make our .rev_db file very, very important
1721 # and we can't afford to lose it because rebuild() won't work
1722 if ($self->use_svm_props || $self->no_metadata) {
1723 $sync = 1;
1724 copy($db, $db_lock) or die "rev_db_set(@_): ",
1725 "Failed to copy: ",
1726 "$db => $db_lock ($!)\n";
1727 } else {
1728 rename $db, $db_lock or die "rev_db_set(@_): ",
1729 "Failed to rename: ",
1730 "$db => $db_lock ($!)\n";
1731 }
1732 open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
1733 _rev_db_set($fh, $rev, $commit);
1734 if ($sync) {
1735 $fh->flush or die "Couldn't flush $db_lock: $!\n";
1736 $fh->sync or die "Couldn't sync $db_lock: $!\n";
1737 }
1738 close $fh or croak $!;
1739 if ($update_ref) {
1740 command_noisy('update-ref', '-m', "r$rev",
1741 $self->refname, $commit);
1742 }
1743 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
1744 "$db_lock => $db ($!)\n";
1745 delete $LOCKFILES{$db_lock};
1746 if ($update_ref) {
1747 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
1748 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
1749 kill $sig, $$ if defined $sig;
1750 }
1751 }
1752
1753 sub rev_db_max {
1754 my ($self) = @_;
1755 my $db_path = $self->db_path;
1756 my @stat = stat $db_path or return 0;
1757 ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
1758 my $max = $stat[7] / 41;
1759 (($max > 0) ? $max - 1 : 0);
1760 }
1761
1762 sub rev_db_get {
1763 my ($self, $rev, $uuid) = @_;
1764 my $ret;
1765 my $offset = $rev * 41;
1766 my $db_path = $self->db_path($uuid);
1767 return undef unless -e $db_path;
1768 open my $fh, '<', $db_path or croak $!;
1769 if (sysseek($fh, $offset, 0) == $offset) {
1770 my $read = sysread($fh, $ret, 40);
1771 $ret = undef if ($read != 40 || $ret eq ('0'x40));
1772 }
1773 close $fh or croak $!;
1774 $ret;
1775 }
1776
1777 sub find_rev_before {
1778 my ($self, $rev, $eq_ok) = @_;
1779 --$rev unless $eq_ok;
1780 while ($rev > 0) {
1781 if (my $c = $self->rev_db_get($rev)) {
1782 return ($rev, $c);
1783 }
1784 --$rev;
1785 }
1786 return (undef, undef);
1787 }
1788
1789 sub _new {
1790 my ($class, $repo_id, $ref_id, $path) = @_;
1791 unless (defined $repo_id && length $repo_id) {
1792 $repo_id = $Git::SVN::default_repo_id;
1793 }
1794 unless (defined $ref_id && length $ref_id) {
1795 $_[2] = $ref_id = $Git::SVN::default_ref_id;
1796 }
1797 $_[1] = $repo_id = sanitize_remote_name($repo_id);
1798 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1799 $_[3] = $path = '' unless (defined $path);
1800 mkpath([$dir]);
1801 bless {
1802 ref_id => $ref_id, dir => $dir, index => "$dir/index",
1803 path => $path, config => "$ENV{GIT_DIR}/svn/config",
1804 db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
1805 }
1806
1807 sub db_path {
1808 my ($self, $uuid) = @_;
1809 $uuid ||= $self->ra_uuid;
1810 "$self->{db_root}.$uuid";
1811 }
1812
1813 sub uri_encode {
1814 my ($f) = @_;
1815 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1816 $f
1817 }
1818
1819 package Git::SVN::Prompt;
1820 use strict;
1821 use warnings;
1822 require SVN::Core;
1823 use vars qw/$_no_auth_cache $_username/;
1824
1825 sub simple {
1826 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1827 $may_save = undef if $_no_auth_cache;
1828 $default_username = $_username if defined $_username;
1829 if (defined $default_username && length $default_username) {
1830 if (defined $realm && length $realm) {
1831 print STDERR "Authentication realm: $realm\n";
1832 STDERR->flush;
1833 }
1834 $cred->username($default_username);
1835 } else {
1836 username($cred, $realm, $may_save, $pool);
1837 }
1838 $cred->password(_read_password("Password for '" .
1839 $cred->username . "': ", $realm));
1840 $cred->may_save($may_save);
1841 $SVN::_Core::SVN_NO_ERROR;
1842 }
1843
1844 sub ssl_server_trust {
1845 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1846 $may_save = undef if $_no_auth_cache;
1847 print STDERR "Error validating server certificate for '$realm':\n";
1848 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1849 print STDERR " - The certificate is not issued by a trusted ",
1850 "authority. Use the\n",
1851 " fingerprint to validate the certificate manually!\n";
1852 }
1853 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1854 print STDERR " - The certificate hostname does not match.\n";
1855 }
1856 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1857 print STDERR " - The certificate is not yet valid.\n";
1858 }
1859 if ($failures & $SVN::Auth::SSL::EXPIRED) {
1860 print STDERR " - The certificate has expired.\n";
1861 }
1862 if ($failures & $SVN::Auth::SSL::OTHER) {
1863 print STDERR " - The certificate has an unknown error.\n";
1864 }
1865 printf STDERR
1866 "Certificate information:\n".
1867 " - Hostname: %s\n".
1868 " - Valid: from %s until %s\n".
1869 " - Issuer: %s\n".
1870 " - Fingerprint: %s\n",
1871 map $cert_info->$_, qw(hostname valid_from valid_until
1872 issuer_dname fingerprint);
1873 my $choice;
1874 prompt:
1875 print STDERR $may_save ?
1876 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1877 "(R)eject or accept (t)emporarily? ";
1878 STDERR->flush;
1879 $choice = lc(substr(<STDIN> || 'R', 0, 1));
1880 if ($choice =~ /^t$/i) {
1881 $cred->may_save(undef);
1882 } elsif ($choice =~ /^r$/i) {
1883 return -1;
1884 } elsif ($may_save && $choice =~ /^p$/i) {
1885 $cred->may_save($may_save);
1886 } else {
1887 goto prompt;
1888 }
1889 $cred->accepted_failures($failures);
1890 $SVN::_Core::SVN_NO_ERROR;
1891 }
1892
1893 sub ssl_client_cert {
1894 my ($cred, $realm, $may_save, $pool) = @_;
1895 $may_save = undef if $_no_auth_cache;
1896 print STDERR "Client certificate filename: ";
1897 STDERR->flush;
1898 chomp(my $filename = <STDIN>);
1899 $cred->cert_file($filename);
1900 $cred->may_save($may_save);
1901 $SVN::_Core::SVN_NO_ERROR;
1902 }
1903
1904 sub ssl_client_cert_pw {
1905 my ($cred, $realm, $may_save, $pool) = @_;
1906 $may_save = undef if $_no_auth_cache;
1907 $cred->password(_read_password("Password: ", $realm));
1908 $cred->may_save($may_save);
1909 $SVN::_Core::SVN_NO_ERROR;
1910 }
1911
1912 sub username {
1913 my ($cred, $realm, $may_save, $pool) = @_;
1914 $may_save = undef if $_no_auth_cache;
1915 if (defined $realm && length $realm) {
1916 print STDERR "Authentication realm: $realm\n";
1917 }
1918 my $username;
1919 if (defined $_username) {
1920 $username = $_username;
1921 } else {
1922 print STDERR "Username: ";
1923 STDERR->flush;
1924 chomp($username = <STDIN>);
1925 }
1926 $cred->username($username);
1927 $cred->may_save($may_save);
1928 $SVN::_Core::SVN_NO_ERROR;
1929 }
1930
1931 sub _read_password {
1932 my ($prompt, $realm) = @_;
1933 print STDERR $prompt;
1934 STDERR->flush;
1935 require Term::ReadKey;
1936 Term::ReadKey::ReadMode('noecho');
1937 my $password = '';
1938 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1939 last if $key =~ /[\012\015]/; # \n\r
1940 $password .= $key;
1941 }
1942 Term::ReadKey::ReadMode('restore');
1943 print STDERR "\n";
1944 STDERR->flush;
1945 $password;
1946 }
1947
1948 package main;
1949
1950 {
1951 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1952 $SVN::Node::dir.$SVN::Node::unknown.
1953 $SVN::Node::none.$SVN::Node::file.
1954 $SVN::Node::dir.$SVN::Node::unknown.
1955 $SVN::Auth::SSL::CNMISMATCH.
1956 $SVN::Auth::SSL::NOTYETVALID.
1957 $SVN::Auth::SSL::EXPIRED.
1958 $SVN::Auth::SSL::UNKNOWNCA.
1959 $SVN::Auth::SSL::OTHER;
1960 }
1961
1962 package SVN::Git::Fetcher;
1963 use vars qw/@ISA/;
1964 use strict;
1965 use warnings;
1966 use Carp qw/croak/;
1967 use IO::File qw//;
1968 use Digest::MD5;
1969
1970 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
1971 sub new {
1972 my ($class, $git_svn) = @_;
1973 my $self = SVN::Delta::Editor->new;
1974 bless $self, $class;
1975 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
1976 $self->{empty} = {};
1977 $self->{dir_prop} = {};
1978 $self->{file_prop} = {};
1979 $self->{absent_dir} = {};
1980 $self->{absent_file} = {};
1981 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
1982 $self;
1983 }
1984
1985 sub set_path_strip {
1986 my ($self, $path) = @_;
1987 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
1988 }
1989
1990 sub open_root {
1991 { path => '' };
1992 }
1993
1994 sub open_directory {
1995 my ($self, $path, $pb, $rev) = @_;
1996 { path => $path };
1997 }
1998
1999 sub git_path {
2000 my ($self, $path) = @_;
2001 if ($self->{path_strip}) {
2002 $path =~ s!$self->{path_strip}!! or
2003 die "Failed to strip path '$path' ($self->{path_strip})\n";
2004 }
2005 $path;
2006 }
2007
2008 sub delete_entry {
2009 my ($self, $path, $rev, $pb) = @_;
2010
2011 my $gpath = $self->git_path($path);
2012 return undef if ($gpath eq '');
2013
2014 # remove entire directories.
2015 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2016 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2017 -r --name-only -z/,
2018 $self->{c}, '--', $gpath);
2019 local $/ = "\0";
2020 while (<$ls>) {
2021 chomp;
2022 $self->{gii}->remove($_);
2023 print "\tD\t$_\n" unless $::_q;
2024 }
2025 print "\tD\t$gpath/\n" unless $::_q;
2026 command_close_pipe($ls, $ctx);
2027 $self->{empty}->{$path} = 0
2028 } else {
2029 $self->{gii}->remove($gpath);
2030 print "\tD\t$gpath\n" unless $::_q;
2031 }
2032 undef;
2033 }
2034
2035 sub open_file {
2036 my ($self, $path, $pb, $rev) = @_;
2037 my $gpath = $self->git_path($path);
2038 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2039 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2040 unless (defined $mode && defined $blob) {
2041 die "$path was not found in commit $self->{c} (r$rev)\n";
2042 }
2043 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2044 pool => SVN::Pool->new, action => 'M' };
2045 }
2046
2047 sub add_file {
2048 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2049 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2050 delete $self->{empty}->{$dir};
2051 { path => $path, mode_a => 100644, mode_b => 100644,
2052 pool => SVN::Pool->new, action => 'A' };
2053 }
2054
2055 sub add_directory {
2056 my ($self, $path, $cp_path, $cp_rev) = @_;
2057 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2058 delete $self->{empty}->{$dir};
2059 $self->{empty}->{$path} = 1;
2060 { path => $path };
2061 }
2062
2063 sub change_dir_prop {
2064 my ($self, $db, $prop, $value) = @_;
2065 $self->{dir_prop}->{$db->{path}} ||= {};
2066 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2067 undef;
2068 }
2069
2070 sub absent_directory {
2071 my ($self, $path, $pb) = @_;
2072 $self->{absent_dir}->{$pb->{path}} ||= [];
2073 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2074 undef;
2075 }
2076
2077 sub absent_file {
2078 my ($self, $path, $pb) = @_;
2079 $self->{absent_file}->{$pb->{path}} ||= [];
2080 push @{$self->{absent_file}->{$pb->{path}}}, $path;
2081 undef;
2082 }
2083
2084 sub change_file_prop {
2085 my ($self, $fb, $prop, $value) = @_;
2086 if ($prop eq 'svn:executable') {
2087 if ($fb->{mode_b} != 120000) {
2088 $fb->{mode_b} = defined $value ? 100755 : 100644;
2089 }
2090 } elsif ($prop eq 'svn:special') {
2091 $fb->{mode_b} = defined $value ? 120000 : 100644;
2092 } else {
2093 $self->{file_prop}->{$fb->{path}} ||= {};
2094 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2095 }
2096 undef;
2097 }
2098
2099 sub apply_textdelta {
2100 my ($self, $fb, $exp) = @_;
2101 my $fh = IO::File->new_tmpfile;
2102 $fh->autoflush(1);
2103 # $fh gets auto-closed() by SVN::TxDelta::apply(),
2104 # (but $base does not,) so dup() it for reading in close_file
2105 open my $dup, '<&', $fh or croak $!;
2106 my $base = IO::File->new_tmpfile;
2107 $base->autoflush(1);
2108 if ($fb->{blob}) {
2109 defined (my $pid = fork) or croak $!;
2110 if (!$pid) {
2111 open STDOUT, '>&', $base or croak $!;
2112 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2113 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2114 }
2115 waitpid $pid, 0;
2116 croak $? if $?;
2117
2118 if (defined $exp) {
2119 seek $base, 0, 0 or croak $!;
2120 my $md5 = Digest::MD5->new;
2121 $md5->addfile($base);
2122 my $got = $md5->hexdigest;
2123 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2124 "expected: $exp\n",
2125 " got: $got\n" if ($got ne $exp);
2126 }
2127 }
2128 seek $base, 0, 0 or croak $!;
2129 $fb->{fh} = $dup;
2130 $fb->{base} = $base;
2131 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2132 }
2133
2134 sub close_file {
2135 my ($self, $fb, $exp) = @_;
2136 my $hash;
2137 my $path = $self->git_path($fb->{path});
2138 if (my $fh = $fb->{fh}) {
2139 seek($fh, 0, 0) or croak $!;
2140 my $md5 = Digest::MD5->new;
2141 $md5->addfile($fh);
2142 my $got = $md5->hexdigest;
2143 die "Checksum mismatch: $path\n",
2144 "expected: $exp\n got: $got\n" if ($got ne $exp);
2145 seek($fh, 0, 0) or croak $!;
2146 if ($fb->{mode_b} == 120000) {
2147 read($fh, my $buf, 5) == 5 or croak $!;
2148 $buf eq 'link ' or die "$path has mode 120000",
2149 "but is not a link\n";
2150 }
2151 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2152 if (!$pid) {
2153 open STDIN, '<&', $fh or croak $!;
2154 exec qw/git-hash-object -w --stdin/ or croak $!;
2155 }
2156 chomp($hash = do { local $/; <$out> });
2157 close $out or croak $!;
2158 close $fh or croak $!;
2159 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2160 close $fb->{base} or croak $!;
2161 } else {
2162 $hash = $fb->{blob} or die "no blob information\n";
2163 }
2164 $fb->{pool}->clear;
2165 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2166 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2167 undef;
2168 }
2169
2170 sub abort_edit {
2171 my $self = shift;
2172 $self->{nr} = $self->{gii}->{nr};
2173 delete $self->{gii};
2174 $self->SUPER::abort_edit(@_);
2175 }
2176
2177 sub close_edit {
2178 my $self = shift;
2179 $self->{git_commit_ok} = 1;
2180 $self->{nr} = $self->{gii}->{nr};
2181 delete $self->{gii};
2182 $self->SUPER::close_edit(@_);
2183 }
2184
2185 package SVN::Git::Editor;
2186 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2187 use strict;
2188 use warnings;
2189 use Carp qw/croak/;
2190 use IO::File;
2191 use Digest::MD5;
2192
2193 sub new {
2194 my ($class, $opts) = @_;
2195 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2196 die "$_ required!\n" unless (defined $opts->{$_});
2197 }
2198
2199 my $pool = SVN::Pool->new;
2200 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2201 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2202 $opts->{r}, $mods);
2203
2204 # $opts->{ra} functions should not be used after this:
2205 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
2206 $opts->{editor_cb}, $pool);
2207 my $self = SVN::Delta::Editor->new(@ce, $pool);
2208 bless $self, $class;
2209 foreach (qw/svn_path r tree_a tree_b/) {
2210 $self->{$_} = $opts->{$_};
2211 }
2212 $self->{url} = $opts->{ra}->{url};
2213 $self->{mods} = $mods;
2214 $self->{types} = $types;
2215 $self->{pool} = $pool;
2216 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2217 $self->{rm} = { };
2218 $self->{path_prefix} = length $self->{svn_path} ?
2219 "$self->{svn_path}/" : '';
2220 return $self;
2221 }
2222
2223 sub generate_diff {
2224 my ($tree_a, $tree_b) = @_;
2225 my @diff_tree = qw(diff-tree -z -r);
2226 if ($_cp_similarity) {
2227 push @diff_tree, "-C$_cp_similarity";
2228 } else {
2229 push @diff_tree, '-C';
2230 }
2231 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2232 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2233 push @diff_tree, $tree_a, $tree_b;
2234 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2235 local $/ = "\0";
2236 my $state = 'meta';
2237 my @mods;
2238 while (<$diff_fh>) {
2239 chomp $_; # this gets rid of the trailing "\0"
2240 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2241 $::sha1\s($::sha1)\s
2242 ([MTCRAD])\d*$/xo) {
2243 push @mods, { mode_a => $1, mode_b => $2,
2244 sha1_b => $3, chg => $4 };
2245 if ($4 =~ /^(?:C|R)$/) {
2246 $state = 'file_a';
2247 } else {
2248 $state = 'file_b';
2249 }
2250 } elsif ($state eq 'file_a') {
2251 my $x = $mods[$#mods] or croak "Empty array\n";
2252 if ($x->{chg} !~ /^(?:C|R)$/) {
2253 croak "Error parsing $_, $x->{chg}\n";
2254 }
2255 $x->{file_a} = $_;
2256 $state = 'file_b';
2257 } elsif ($state eq 'file_b') {
2258 my $x = $mods[$#mods] or croak "Empty array\n";
2259 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2260 croak "Error parsing $_, $x->{chg}\n";
2261 }
2262 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2263 croak "Error parsing $_, $x->{chg}\n";
2264 }
2265 $x->{file_b} = $_;
2266 $state = 'meta';
2267 } else {
2268 croak "Error parsing $_\n";
2269 }
2270 }
2271 command_close_pipe($diff_fh, $ctx);
2272 \@mods;
2273 }
2274
2275 sub check_diff_paths {
2276 my ($ra, $pfx, $rev, $mods) = @_;
2277 my %types;
2278 $pfx .= '/' if length $pfx;
2279
2280 sub type_diff_paths {
2281 my ($ra, $types, $path, $rev) = @_;
2282 my @p = split m#/+#, $path;
2283 my $c = shift @p;
2284 unless (defined $types->{$c}) {
2285 $types->{$c} = $ra->check_path($c, $rev);
2286 }
2287 while (@p) {
2288 $c .= '/' . shift @p;
2289 next if defined $types->{$c};
2290 $types->{$c} = $ra->check_path($c, $rev);
2291 }
2292 }
2293
2294 foreach my $m (@$mods) {
2295 foreach my $f (qw/file_a file_b/) {
2296 next unless defined $m->{$f};
2297 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2298 if (length $pfx.$dir && ! defined $types{$dir}) {
2299 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2300 }
2301 }
2302 }
2303 \%types;
2304 }
2305
2306 sub split_path {
2307 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2308 }
2309
2310 sub repo_path {
2311 my ($self, $path) = @_;
2312 $self->{path_prefix}.(defined $path ? $path : '');
2313 }
2314
2315 sub url_path {
2316 my ($self, $path) = @_;
2317 $self->{url} . '/' . $self->repo_path($path);
2318 }
2319
2320 sub rmdirs {
2321 my ($self) = @_;
2322 my $rm = $self->{rm};
2323 delete $rm->{''}; # we never delete the url we're tracking
2324 return unless %$rm;
2325
2326 foreach (keys %$rm) {
2327 my @d = split m#/#, $_;
2328 my $c = shift @d;
2329 $rm->{$c} = 1;
2330 while (@d) {
2331 $c .= '/' . shift @d;
2332 $rm->{$c} = 1;
2333 }
2334 }
2335 delete $rm->{$self->{svn_path}};
2336 delete $rm->{''}; # we never delete the url we're tracking
2337 return unless %$rm;
2338
2339 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2340 $self->{tree_b});
2341 local $/ = "\0";
2342 while (<$fh>) {
2343 chomp;
2344 my @dn = split m#/#, $_;
2345 while (pop @dn) {
2346 delete $rm->{join '/', @dn};
2347 }
2348 unless (%$rm) {
2349 close $fh;
2350 return;
2351 }
2352 }
2353 command_close_pipe($fh, $ctx);
2354
2355 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2356 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2357 $self->close_directory($bat->{$d}, $p);
2358 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2359 print "\tD+\t$d/\n" unless $::_q;
2360 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2361 delete $bat->{$d};
2362 }
2363 }
2364
2365 sub open_or_add_dir {
2366 my ($self, $full_path, $baton) = @_;
2367 my $t = $self->{types}->{$full_path};
2368 if (!defined $t) {
2369 die "$full_path not known in r$self->{r} or we have a bug!\n";
2370 }
2371 if ($t == $SVN::Node::none) {
2372 return $self->add_directory($full_path, $baton,
2373 undef, -1, $self->{pool});
2374 } elsif ($t == $SVN::Node::dir) {
2375 return $self->open_directory($full_path, $baton,
2376 $self->{r}, $self->{pool});
2377 }
2378 print STDERR "$full_path already exists in repository at ",
2379 "r$self->{r} and it is not a directory (",
2380 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2381 exit 1;
2382 }
2383
2384 sub ensure_path {
2385 my ($self, $path) = @_;
2386 my $bat = $self->{bat};
2387 my $repo_path = $self->repo_path($path);
2388 return $bat->{''} unless (length $repo_path);
2389 my @p = split m#/+#, $repo_path;
2390 my $c = shift @p;
2391 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2392 while (@p) {
2393 my $c0 = $c;
2394 $c .= '/' . shift @p;
2395 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2396 }
2397 return $bat->{$c};
2398 }
2399
2400 sub A {
2401 my ($self, $m) = @_;
2402 my ($dir, $file) = split_path($m->{file_b});
2403 my $pbat = $self->ensure_path($dir);
2404 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2405 undef, -1);
2406 print "\tA\t$m->{file_b}\n" unless $::_q;
2407 $self->chg_file($fbat, $m);
2408 $self->close_file($fbat,undef,$self->{pool});
2409 }
2410
2411 sub C {
2412 my ($self, $m) = @_;
2413 my ($dir, $file) = split_path($m->{file_b});
2414 my $pbat = $self->ensure_path($dir);
2415 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2416 $self->url_path($m->{file_a}), $self->{r});
2417 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2418 $self->chg_file($fbat, $m);
2419 $self->close_file($fbat,undef,$self->{pool});
2420 }
2421
2422 sub delete_entry {
2423 my ($self, $path, $pbat) = @_;
2424 my $rpath = $self->repo_path($path);
2425 my ($dir, $file) = split_path($rpath);
2426 $self->{rm}->{$dir} = 1;
2427 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2428 }
2429
2430 sub R {
2431 my ($self, $m) = @_;
2432 my ($dir, $file) = split_path($m->{file_b});
2433 my $pbat = $self->ensure_path($dir);
2434 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2435 $self->url_path($m->{file_a}), $self->{r});
2436 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2437 $self->chg_file($fbat, $m);
2438 $self->close_file($fbat,undef,$self->{pool});
2439
2440 ($dir, $file) = split_path($m->{file_a});
2441 $pbat = $self->ensure_path($dir);
2442 $self->delete_entry($m->{file_a}, $pbat);
2443 }
2444
2445 sub M {
2446 my ($self, $m) = @_;
2447 my ($dir, $file) = split_path($m->{file_b});
2448 my $pbat = $self->ensure_path($dir);
2449 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2450 $pbat,$self->{r},$self->{pool});
2451 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2452 $self->chg_file($fbat, $m);
2453 $self->close_file($fbat,undef,$self->{pool});
2454 }
2455
2456 sub T { shift->M(@_) }
2457
2458 sub change_file_prop {
2459 my ($self, $fbat, $pname, $pval) = @_;
2460 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2461 }
2462
2463 sub chg_file {
2464 my ($self, $fbat, $m) = @_;
2465 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2466 $self->change_file_prop($fbat,'svn:executable','*');
2467 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2468 $self->change_file_prop($fbat,'svn:executable',undef);
2469 }
2470 my $fh = IO::File->new_tmpfile or croak $!;
2471 if ($m->{mode_b} =~ /^120/) {
2472 print $fh 'link ' or croak $!;
2473 $self->change_file_prop($fbat,'svn:special','*');
2474 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2475 $self->change_file_prop($fbat,'svn:special',undef);
2476 }
2477 defined(my $pid = fork) or croak $!;
2478 if (!$pid) {
2479 open STDOUT, '>&', $fh or croak $!;
2480 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2481 }
2482 waitpid $pid, 0;
2483 croak $? if $?;
2484 $fh->flush == 0 or croak $!;
2485 seek $fh, 0, 0 or croak $!;
2486
2487 my $md5 = Digest::MD5->new;
2488 $md5->addfile($fh) or croak $!;
2489 seek $fh, 0, 0 or croak $!;
2490
2491 my $exp = $md5->hexdigest;
2492 my $pool = SVN::Pool->new;
2493 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2494 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2495 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2496 $pool->clear;
2497
2498 close $fh or croak $!;
2499 }
2500
2501 sub D {
2502 my ($self, $m) = @_;
2503 my ($dir, $file) = split_path($m->{file_b});
2504 my $pbat = $self->ensure_path($dir);
2505 print "\tD\t$m->{file_b}\n" unless $::_q;
2506 $self->delete_entry($m->{file_b}, $pbat);
2507 }
2508
2509 sub close_edit {
2510 my ($self) = @_;
2511 my ($p,$bat) = ($self->{pool}, $self->{bat});
2512 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2513 $self->close_directory($bat->{$_}, $p);
2514 }
2515 $self->SUPER::close_edit($p);
2516 $p->clear;
2517 }
2518
2519 sub abort_edit {
2520 my ($self) = @_;
2521 $self->SUPER::abort_edit($self->{pool});
2522 }
2523
2524 sub DESTROY {
2525 my $self = shift;
2526 $self->SUPER::DESTROY(@_);
2527 $self->{pool}->clear;
2528 }
2529
2530 # this drives the editor
2531 sub apply_diff {
2532 my ($self) = @_;
2533 my $mods = $self->{mods};
2534 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2535 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2536 my $f = $m->{chg};
2537 if (defined $o{$f}) {
2538 $self->$f($m);
2539 } else {
2540 fatal("Invalid change type: $f\n");
2541 }
2542 }
2543 $self->rmdirs if $_rmdir;
2544 if (@$mods == 0) {
2545 $self->abort_edit;
2546 } else {
2547 $self->close_edit;
2548 }
2549 return scalar @$mods;
2550 }
2551
2552 package Git::SVN::Ra;
2553 use vars qw/@ISA $config_dir/;
2554 use strict;
2555 use warnings;
2556 my ($can_do_switch);
2557 my $RA;
2558
2559 BEGIN {
2560 # enforce temporary pool usage for some simple functions
2561 my $e;
2562 foreach (qw/get_latest_revnum get_uuid get_repos_root/) {
2563 $e .= "sub $_ {
2564 my \$self = shift;
2565 my \$pool = SVN::Pool->new;
2566 my \@ret = \$self->SUPER::$_(\@_,\$pool);
2567 \$pool->clear;
2568 wantarray ? \@ret : \$ret[0]; }\n";
2569 }
2570
2571 # get_dir needs $pool held in cache for dirents to work,
2572 # check_path is cacheable and rev_proplist is close enough
2573 # for our purposes.
2574 foreach (qw/check_path get_dir rev_proplist/) {
2575 $e .= "my \%${_}_cache; my \$${_}_rev = 0; sub $_ {
2576 my \$self = shift;
2577 my \$r = pop;
2578 my \$k = join(\"\\0\", \@_);
2579 if (my \$x = \$${_}_cache{\$r}->{\$k}) {
2580 return wantarray ? \@\$x : \$x->[0];
2581 }
2582 my \$pool = SVN::Pool->new;
2583 my \@ret = \$self->SUPER::$_(\@_, \$r, \$pool);
2584 if (\$r != \$${_}_rev) {
2585 \%${_}_cache = ( pool => [] );
2586 \$${_}_rev = \$r;
2587 }
2588 \$${_}_cache{\$r}->{\$k} = \\\@ret;
2589 push \@{\$${_}_cache{pool}}, \$pool;
2590 wantarray ? \@ret : \$ret[0]; }\n";
2591 }
2592 $e .= "\n1;";
2593 eval $e or die $@;
2594 }
2595
2596 sub new {
2597 my ($class, $url) = @_;
2598 $url =~ s!/+$!!;
2599 return $RA if ($RA && $RA->{url} eq $url);
2600
2601 SVN::_Core::svn_config_ensure($config_dir, undef);
2602 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2603 SVN::Client::get_simple_provider(),
2604 SVN::Client::get_ssl_server_trust_file_provider(),
2605 SVN::Client::get_simple_prompt_provider(
2606 \&Git::SVN::Prompt::simple, 2),
2607 SVN::Client::get_ssl_client_cert_prompt_provider(
2608 \&Git::SVN::Prompt::ssl_client_cert, 2),
2609 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2610 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2611 SVN::Client::get_username_provider(),
2612 SVN::Client::get_ssl_server_trust_prompt_provider(
2613 \&Git::SVN::Prompt::ssl_server_trust),
2614 SVN::Client::get_username_prompt_provider(
2615 \&Git::SVN::Prompt::username, 2),
2616 ]);
2617 my $config = SVN::Core::config_get_config($config_dir);
2618 my $self = SVN::Ra->new(url => $url, auth => $baton,
2619 config => $config,
2620 pool => SVN::Pool->new,
2621 auth_provider_callbacks => $callbacks);
2622 $self->{svn_path} = $url;
2623 $self->{repos_root} = $self->get_repos_root;
2624 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
2625 $RA = bless $self, $class;
2626 }
2627
2628 sub DESTROY {
2629 # do not call the real DESTROY since we store ourselves in $RA
2630 }
2631
2632 sub get_log {
2633 my ($self, @args) = @_;
2634 my $pool = SVN::Pool->new;
2635 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2636 my $ret = $self->SUPER::get_log(@args, $pool);
2637 $pool->clear;
2638 $ret;
2639 }
2640
2641 sub get_commit_editor {
2642 my ($self, $log, $cb, $pool) = @_;
2643 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2644 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2645 }
2646
2647 sub gs_do_update {
2648 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2649 my $new = ($rev_a == $rev_b);
2650 my $path = $gs->{path};
2651
2652 my $pool = SVN::Pool->new;
2653 $editor->set_path_strip($path);
2654 my (@pc) = split m#/#, $path;
2655 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2656 1, $editor, $pool);
2657 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2658
2659 # Since we can't rely on svn_ra_reparent being available, we'll
2660 # just have to do some magic with set_path to make it so
2661 # we only want a partial path.
2662 my $sp = '';
2663 my $final = join('/', @pc);
2664 while (@pc) {
2665 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2666 $sp .= '/' if length $sp;
2667 $sp .= shift @pc;
2668 }
2669 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2670
2671 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2672
2673 $reporter->finish_report($pool);
2674 $pool->clear;
2675 $editor->{git_commit_ok};
2676 }
2677
2678 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2679 # svn_ra_reparent didn't work before 1.4)
2680 sub gs_do_switch {
2681 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
2682 my $path = $gs->{path};
2683 my $pool = SVN::Pool->new;
2684
2685 my $full_url = $self->{url};
2686 my $old_url = $full_url;
2687 $full_url .= "/$path" if length $path;
2688 my ($ra, $reparented);
2689 if ($old_url ne $full_url) {
2690 if ($old_url !~ m#^svn(\+ssh)?://#) {
2691 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2692 $pool);
2693 $self->{url} = $full_url;
2694 $reparented = 1;
2695 } else {
2696 $ra = Git::SVN::Ra->new($full_url);
2697 }
2698 }
2699 $ra ||= $self;
2700 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
2701 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2702 $reporter->set_path('', $rev_a, 0, @lock, $pool);
2703 $reporter->finish_report($pool);
2704
2705 if ($reparented) {
2706 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2707 $self->{url} = $old_url;
2708 }
2709
2710 $pool->clear;
2711 $editor->{git_commit_ok};
2712 }
2713
2714 sub gs_fetch_loop_common {
2715 my ($self, $base, $head, $gsv, $globs) = @_;
2716 return if ($base > $head);
2717 my $inc = 1000;
2718 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2719 my %common;
2720 my $common_max = scalar @$gsv;
2721
2722 foreach my $gs (@$gsv) {
2723 if (my $last_commit = $gs->last_commit) {
2724 $gs->assert_index_clean($last_commit);
2725 }
2726 my @tmp = split m#/#, $gs->{path};
2727 my $p = '';
2728 foreach (@tmp) {
2729 $p .= length($p) ? "/$_" : $_;
2730 $common{$p} ||= 0;
2731 $common{$p}++;
2732 }
2733 }
2734 $globs ||= [];
2735 $common_max += scalar @$globs;
2736 foreach my $glob (@$globs) {
2737 my @tmp = split m#/#, $glob->{path}->{left};
2738 my $p = '';
2739 foreach (@tmp) {
2740 $p .= length($p) ? "/$_" : $_;
2741 $common{$p} ||= 0;
2742 $common{$p}++;
2743 }
2744 }
2745
2746 my $longest_path = '';
2747 foreach (sort {length $b <=> length $a} keys %common) {
2748 if ($common{$_} == $common_max) {
2749 $longest_path = $_;
2750 last;
2751 }
2752 }
2753 while (1) {
2754 my %revs;
2755 my $err;
2756 my $err_handler = $SVN::Error::handler;
2757 $SVN::Error::handler = sub {
2758 ($err) = @_;
2759 skip_unknown_revs($err);
2760 };
2761 sub _cb {
2762 my ($paths, $r, $author, $date, $log) = @_;
2763 [ dup_changed_paths($paths),
2764 { author => $author, date => $date, log => $log } ];
2765 }
2766 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
2767 sub { $revs{$_[1]} = _cb(@_) });
2768 if ($err && $max >= $head) {
2769 print STDERR "Path '$longest_path' ",
2770 "was probably deleted:\n",
2771 $err->expanded_message,
2772 "\nWill attempt to follow ",
2773 "revisions r$min .. r$max ",
2774 "committed before the deletion\n";
2775 my $hi = $max;
2776 while (--$hi >= $min) {
2777 my $ok;
2778 $self->get_log([$longest_path], $min, $hi,
2779 0, 1, 1, sub {
2780 $ok ||= $_[1];
2781 $revs{$_[1]} = _cb(@_) });
2782 if ($ok) {
2783 print STDERR "r$min .. r$ok OK\n";
2784 last;
2785 }
2786 }
2787 }
2788 $SVN::Error::handler = $err_handler;
2789
2790 my %exists = map { $_->{path} => $_ } @$gsv;
2791 foreach my $r (sort {$a <=> $b} keys %revs) {
2792 my ($paths, $logged) = @{$revs{$r}};
2793
2794 foreach my $gs ($self->match_globs(\%exists, $paths,
2795 $globs, $r)) {
2796 if ($gs->rev_db_max >= $r) {
2797 next;
2798 }
2799 next unless $gs->match_paths($paths, $r);
2800 $gs->{logged_rev_props} = $logged;
2801 my $log_entry = $gs->do_fetch($paths, $r);
2802 if ($log_entry) {
2803 $gs->do_git_commit($log_entry);
2804 }
2805 }
2806 foreach my $g (@$globs) {
2807 my $k = "svn-remote.$g->{remote}." .
2808 "$g->{t}-maxRev";
2809 Git::SVN::tmp_config($k, $r);
2810 }
2811 }
2812 # pre-fill the .rev_db since it'll eventually get filled in
2813 # with '0' x40 if something new gets committed
2814 foreach my $gs (@$gsv) {
2815 next if defined $gs->rev_db_get($max);
2816 $gs->rev_db_set($max, 0 x40);
2817 }
2818 foreach my $g (@$globs) {
2819 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
2820 Git::SVN::tmp_config($k, $max);
2821 }
2822 last if $max >= $head;
2823 $min = $max + 1;
2824 $max += $inc;
2825 $max = $head if ($max > $head);
2826 }
2827 }
2828
2829 sub match_globs {
2830 my ($self, $exists, $paths, $globs, $r) = @_;
2831
2832 sub get_dir_check {
2833 my ($self, $exists, $g, $r) = @_;
2834 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
2835 return unless scalar @x == 3;
2836 my $dirents = $x[0];
2837 foreach my $de (keys %$dirents) {
2838 next if $dirents->{$de}->kind != $SVN::Node::dir;
2839 my $p = $g->{path}->full_path($de);
2840 next if $exists->{$p};
2841 next if (length $g->{path}->{right} &&
2842 ($self->check_path($p, $r) !=
2843 $SVN::Node::dir));
2844 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
2845 $g->{ref}->full_path($de), 1);
2846 }
2847 }
2848 foreach my $g (@$globs) {
2849 if (my $path = $paths->{"/$g->{path}->{left}"}) {
2850 if ($path->{action} =~ /^[AR]$/) {
2851 get_dir_check($self, $exists, $g, $r);
2852 }
2853 }
2854 foreach (keys %$paths) {
2855 if (/$g->{path}->{left_regex}/) {
2856 next if $paths->{$_}->{action} !~ /^[AR]$/;
2857 get_dir_check($self, $exists, $g, $r);
2858 }
2859 next unless /$g->{path}->{regex}/;
2860 my $p = $1;
2861 my $pathname = $g->{path}->full_path($p);
2862 next if $exists->{$pathname};
2863 $exists->{$pathname} = Git::SVN->init(
2864 $self->{url}, $pathname, undef,
2865 $g->{ref}->full_path($p), 1);
2866 }
2867 my $c = '';
2868 foreach (split m#/#, $g->{path}->{left}) {
2869 $c .= "/$_";
2870 next unless ($paths->{$c} &&
2871 ($paths->{$c}->{action} =~ /^[AR]$/));
2872 get_dir_check($self, $exists, $g, $r);
2873 }
2874 }
2875 values %$exists;
2876 }
2877
2878 sub minimize_url {
2879 my ($self) = @_;
2880 return $self->{url} if ($self->{url} eq $self->{repos_root});
2881 my $url = $self->{repos_root};
2882 my @components = split(m!/!, $self->{svn_path});
2883 my $c = '';
2884 do {
2885 $url .= "/$c" if length $c;
2886 eval { (ref $self)->new($url)->get_latest_revnum };
2887 } while ($@ && ($c = shift @components));
2888 $url;
2889 }
2890
2891 sub can_do_switch {
2892 my $self = shift;
2893 unless (defined $can_do_switch) {
2894 my $pool = SVN::Pool->new;
2895 my $rep = eval {
2896 $self->do_switch(1, '', 0, $self->{url},
2897 SVN::Delta::Editor->new, $pool);
2898 };
2899 if ($@) {
2900 $can_do_switch = 0;
2901 } else {
2902 $rep->abort_report($pool);
2903 $can_do_switch = 1;
2904 }
2905 $pool->clear;
2906 }
2907 $can_do_switch;
2908 }
2909
2910 sub skip_unknown_revs {
2911 my ($err) = @_;
2912 my $errno = $err->apr_err();
2913 # Maybe the branch we're tracking didn't
2914 # exist when the repo started, so it's
2915 # not an error if it doesn't, just continue
2916 #
2917 # Wonderfully consistent library, eh?
2918 # 160013 - svn:// and file://
2919 # 175002 - http(s)://
2920 # 175007 - http(s):// (this repo required authorization, too...)
2921 # More codes may be discovered later...
2922 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2923 return;
2924 }
2925 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2926 }
2927
2928 # svn_log_changed_path_t objects passed to get_log are likely to be
2929 # overwritten even if only the refs are copied to an external variable,
2930 # so we should dup the structures in their entirety. Using an externally
2931 # passed pool (instead of our temporary and quickly cleared pool in
2932 # Git::SVN::Ra) does not help matters at all...
2933 sub dup_changed_paths {
2934 my ($paths) = @_;
2935 return undef unless $paths;
2936 my %ret;
2937 foreach my $p (keys %$paths) {
2938 my $i = $paths->{$p};
2939 my %s = map { $_ => $i->$_ }
2940 qw/copyfrom_path copyfrom_rev action/;
2941 $ret{$p} = \%s;
2942 }
2943 \%ret;
2944 }
2945
2946 package Git::SVN::Log;
2947 use strict;
2948 use warnings;
2949 use POSIX qw/strftime/;
2950 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2951 %rusers $show_commit $incremental/;
2952 my $l_fmt;
2953
2954 sub cmt_showable {
2955 my ($c) = @_;
2956 return 1 if defined $c->{r};
2957 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2958 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2959 my @log = command(qw/cat-file commit/, $c->{c});
2960 shift @log while ($log[0] ne "\n");
2961 shift @log;
2962 @{$c->{l}} = grep !/^git-svn-id: /, @log;
2963
2964 (undef, $c->{r}, undef) = ::extract_metadata(
2965 (grep(/^git-svn-id: /, @log))[-1]);
2966 }
2967 return defined $c->{r};
2968 }
2969
2970 sub log_use_color {
2971 return 1 if $color;
2972 my ($dc, $dcvar);
2973 $dcvar = 'color.diff';
2974 $dc = `git-config --get $dcvar`;
2975 if ($dc eq '') {
2976 # nothing at all; fallback to "diff.color"
2977 $dcvar = 'diff.color';
2978 $dc = `git-config --get $dcvar`;
2979 }
2980 chomp($dc);
2981 if ($dc eq 'auto') {
2982 my $pc;
2983 $pc = `git-config --get color.pager`;
2984 if ($pc eq '') {
2985 # does not have it -- fallback to pager.color
2986 $pc = `git-config --bool --get pager.color`;
2987 }
2988 else {
2989 $pc = `git-config --bool --get color.pager`;
2990 if ($?) {
2991 $pc = 'false';
2992 }
2993 }
2994 chomp($pc);
2995 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2996 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2997 }
2998 return 0;
2999 }
3000 return 0 if $dc eq 'never';
3001 return 1 if $dc eq 'always';
3002 chomp($dc = `git-config --bool --get $dcvar`);
3003 return ($dc eq 'true');
3004 }
3005
3006 sub git_svn_log_cmd {
3007 my ($r_min, $r_max) = @_;
3008 my $gs = Git::SVN->_new;
3009 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3010 $gs->refname);
3011 push @cmd, '-r' unless $non_recursive;
3012 push @cmd, qw/--raw --name-status/ if $verbose;
3013 push @cmd, '--color' if log_use_color();
3014 return @cmd unless defined $r_max;
3015 if ($r_max == $r_min) {
3016 push @cmd, '--max-count=1';
3017 if (my $c = $gs->rev_db_get($r_max)) {
3018 push @cmd, $c;
3019 }
3020 } else {
3021 my ($c_min, $c_max);
3022 $c_max = $gs->rev_db_get($r_max);
3023 $c_min = $gs->rev_db_get($r_min);
3024 if (defined $c_min && defined $c_max) {
3025 if ($r_max > $r_max) {
3026 push @cmd, "$c_min..$c_max";
3027 } else {
3028 push @cmd, "$c_max..$c_min";
3029 }
3030 } elsif ($r_max > $r_min) {
3031 push @cmd, $c_max;
3032 } else {
3033 push @cmd, $c_min;
3034 }
3035 }
3036 return @cmd;
3037 }
3038
3039 # adapted from pager.c
3040 sub config_pager {
3041 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3042 if (!defined $pager) {
3043 $pager = 'less';
3044 } elsif (length $pager == 0 || $pager eq 'cat') {
3045 $pager = undef;
3046 }
3047 }
3048
3049 sub run_pager {
3050 return unless -t *STDOUT;
3051 pipe my $rfd, my $wfd or return;
3052 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
3053 if (!$pid) {
3054 open STDOUT, '>&', $wfd or
3055 ::fatal "Can't redirect to stdout: $!\n";
3056 return;
3057 }
3058 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
3059 $ENV{LESS} ||= 'FRSX';
3060 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
3061 }
3062
3063 sub tz_to_s_offset {
3064 my ($tz) = @_;
3065 $tz =~ s/(\d\d)$//;
3066 return ($1 * 60) + ($tz * 3600);
3067 }
3068
3069 sub get_author_info {
3070 my ($dest, $author, $t, $tz) = @_;
3071 $author =~ s/(?:^\s*|\s*$)//g;
3072 $dest->{a_raw} = $author;
3073 my $au;
3074 if ($::_authors) {
3075 $au = $rusers{$author} || undef;
3076 }
3077 if (!$au) {
3078 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3079 }
3080 $dest->{t} = $t;
3081 $dest->{tz} = $tz;
3082 $dest->{a} = $au;
3083 # Date::Parse isn't in the standard Perl distro :(
3084 if ($tz =~ s/^\+//) {
3085 $t += tz_to_s_offset($tz);
3086 } elsif ($tz =~ s/^\-//) {
3087 $t -= tz_to_s_offset($tz);
3088 }
3089 $dest->{t_utc} = $t;
3090 }
3091
3092 sub process_commit {
3093 my ($c, $r_min, $r_max, $defer) = @_;
3094 if (defined $r_min && defined $r_max) {
3095 if ($r_min == $c->{r} && $r_min == $r_max) {
3096 show_commit($c);
3097 return 0;
3098 }
3099 return 1 if $r_min == $r_max;
3100 if ($r_min < $r_max) {
3101 # we need to reverse the print order
3102 return 0 if (defined $limit && --$limit < 0);
3103 push @$defer, $c;
3104 return 1;
3105 }
3106 if ($r_min != $r_max) {
3107 return 1 if ($r_min < $c->{r});
3108 return 1 if ($r_max > $c->{r});
3109 }
3110 }
3111 return 0 if (defined $limit && --$limit < 0);
3112 show_commit($c);
3113 return 1;
3114 }
3115
3116 sub show_commit {
3117 my $c = shift;
3118 if ($oneline) {
3119 my $x = "\n";
3120 if (my $l = $c->{l}) {
3121 while ($l->[0] =~ /^\s*$/) { shift @$l }
3122 $x = $l->[0];
3123 }
3124 $l_fmt ||= 'A' . length($c->{r});
3125 print 'r',pack($l_fmt, $c->{r}),' | ';
3126 print "$c->{c} | " if $show_commit;
3127 print $x;
3128 } else {
3129 show_commit_normal($c);
3130 }
3131 }
3132
3133 sub show_commit_changed_paths {
3134 my ($c) = @_;
3135 return unless $c->{changed};
3136 print "Changed paths:\n", @{$c->{changed}};
3137 }
3138
3139 sub show_commit_normal {
3140 my ($c) = @_;
3141 print '-' x72, "\nr$c->{r} | ";
3142 print "$c->{c} | " if $show_commit;
3143 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3144 localtime($c->{t_utc})), ' | ';
3145 my $nr_line = 0;
3146
3147 if (my $l = $c->{l}) {
3148 while ($l->[$#$l] eq "\n" && $#$l > 0
3149 && $l->[($#$l - 1)] eq "\n") {
3150 pop @$l;
3151 }
3152 $nr_line = scalar @$l;
3153 if (!$nr_line) {
3154 print "1 line\n\n\n";
3155 } else {
3156 if ($nr_line == 1) {
3157 $nr_line = '1 line';
3158 } else {
3159 $nr_line .= ' lines';
3160 }
3161 print $nr_line, "\n";
3162 show_commit_changed_paths($c);
3163 print "\n";
3164 print $_ foreach @$l;
3165 }
3166 } else {
3167 print "1 line\n";
3168 show_commit_changed_paths($c);
3169 print "\n";
3170
3171 }
3172 foreach my $x (qw/raw diff/) {
3173 if ($c->{$x}) {
3174 print "\n";
3175 print $_ foreach @{$c->{$x}}
3176 }
3177 }
3178 }
3179
3180 sub cmd_show_log {
3181 my (@args) = @_;
3182 my ($r_min, $r_max);
3183 my $r_last = -1; # prevent dupes
3184 if (defined $TZ) {
3185 $ENV{TZ} = $TZ;
3186 } else {
3187 delete $ENV{TZ};
3188 }
3189 if (defined $::_revision) {
3190 if ($::_revision =~ /^(\d+):(\d+)$/) {
3191 ($r_min, $r_max) = ($1, $2);
3192 } elsif ($::_revision =~ /^\d+$/) {
3193 $r_min = $r_max = $::_revision;
3194 } else {
3195 ::fatal "-r$::_revision is not supported, use ",
3196 "standard \'git log\' arguments instead\n";
3197 }
3198 }
3199
3200 config_pager();
3201 @args = (git_svn_log_cmd($r_min, $r_max), @args);
3202 my $log = command_output_pipe(@args);
3203 run_pager();
3204 my (@k, $c, $d);
3205 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3206 while (<$log>) {
3207 if (/^${esc_color}commit ($::sha1_short)/o) {
3208 my $cmt = $1;
3209 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3210 $r_last = $c->{r};
3211 process_commit($c, $r_min, $r_max, \@k) or
3212 goto out;
3213 }
3214 $d = undef;
3215 $c = { c => $cmt };
3216 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3217 get_author_info($c, $1, $2, $3);
3218 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3219 # ignore
3220 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3221 push @{$c->{raw}}, $_;
3222 } elsif (/^${esc_color}[ACRMDT]\t/) {
3223 # we could add $SVN->{svn_path} here, but that requires
3224 # remote access at the moment (repo_path_split)...
3225 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
3226 push @{$c->{changed}}, $_;
3227 } elsif (/^${esc_color}diff /o) {
3228 $d = 1;
3229 push @{$c->{diff}}, $_;
3230 } elsif ($d) {
3231 push @{$c->{diff}}, $_;
3232 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
3233 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3234 } elsif (s/^${esc_color} //o) {
3235 push @{$c->{l}}, $_;
3236 }
3237 }
3238 if ($c && defined $c->{r} && $c->{r} != $r_last) {
3239 $r_last = $c->{r};
3240 process_commit($c, $r_min, $r_max, \@k);
3241 }
3242 if (@k) {
3243 my $swap = $r_max;
3244 $r_max = $r_min;
3245 $r_min = $swap;
3246 process_commit($_, $r_min, $r_max) foreach reverse @k;
3247 }
3248 out:
3249 close $log;
3250 print '-' x72,"\n" unless $incremental || $oneline;
3251 }
3252
3253 package Git::SVN::Migration;
3254 # these version numbers do NOT correspond to actual version numbers
3255 # of git nor git-svn. They are just relative.
3256 #
3257 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3258 #
3259 # v1 layout: .git/$id/info/url, refs/remotes/$id
3260 #
3261 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3262 #
3263 # v3 layout: .git/svn/$id, refs/remotes/$id
3264 # - info/url may remain for backwards compatibility
3265 # - this is what we migrate up to this layout automatically,
3266 # - this will be used by git svn init on single branches
3267 # v3.1 layout (auto migrated):
3268 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
3269 # for backwards compatibility
3270 #
3271 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3272 # - this is only created for newly multi-init-ed
3273 # repositories. Similar in spirit to the
3274 # --use-separate-remotes option in git-clone (now default)
3275 # - we do not automatically migrate to this (following
3276 # the example set by core git)
3277 use strict;
3278 use warnings;
3279 use Carp qw/croak/;
3280 use File::Path qw/mkpath/;
3281 use File::Basename qw/dirname basename/;
3282 use vars qw/$_minimize/;
3283
3284 sub migrate_from_v0 {
3285 my $git_dir = $ENV{GIT_DIR};
3286 return undef unless -d $git_dir;
3287 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3288 my $migrated = 0;
3289 while (<$fh>) {
3290 chomp;
3291 my ($id, $orig_ref) = ($_, $_);
3292 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3293 next unless -f "$git_dir/$id/info/url";
3294 my $new_ref = "refs/remotes/$id";
3295 if (::verify_ref("$new_ref^0")) {
3296 print STDERR "W: $orig_ref is probably an old ",
3297 "branch used by an ancient version of ",
3298 "git-svn.\n",
3299 "However, $new_ref also exists.\n",
3300 "We will not be able ",
3301 "to use this branch until this ",
3302 "ambiguity is resolved.\n";
3303 next;
3304 }
3305 print STDERR "Migrating from v0 layout...\n" if !$migrated;
3306 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3307 command_noisy('update-ref', $new_ref, $orig_ref);
3308 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3309 $migrated++;
3310 }
3311 command_close_pipe($fh, $ctx);
3312 print STDERR "Done migrating from v0 layout...\n" if $migrated;
3313 $migrated;
3314 }
3315
3316 sub migrate_from_v1 {
3317 my $git_dir = $ENV{GIT_DIR};
3318 my $migrated = 0;
3319 return $migrated unless -d $git_dir;
3320 my $svn_dir = "$git_dir/svn";
3321
3322 # just in case somebody used 'svn' as their $id at some point...
3323 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3324
3325 print STDERR "Migrating from a git-svn v1 layout...\n";
3326 mkpath([$svn_dir]);
3327 print STDERR "Data from a previous version of git-svn exists, but\n\t",
3328 "$svn_dir\n\t(required for this version ",
3329 "($::VERSION) of git-svn) does not. exist\n";
3330 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3331 while (<$fh>) {
3332 my $x = $_;
3333 next unless $x =~ s#^refs/remotes/##;
3334 chomp $x;
3335 next unless -f "$git_dir/$x/info/url";
3336 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3337 next unless $u;
3338 my $dn = dirname("$git_dir/svn/$x");
3339 mkpath([$dn]) unless -d $dn;
3340 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3341 mkpath(["$git_dir/svn/svn"]);
3342 print STDERR " - $git_dir/$x/info => ",
3343 "$git_dir/svn/$x/info\n";
3344 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3345 croak "$!: $x";
3346 # don't worry too much about these, they probably
3347 # don't exist with repos this old (save for index,
3348 # and we can easily regenerate that)
3349 foreach my $f (qw/unhandled.log index .rev_db/) {
3350 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3351 }
3352 } else {
3353 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3354 rename "$git_dir/$x", "$git_dir/svn/$x" or
3355 croak "$!: $x";
3356 }
3357 $migrated++;
3358 }
3359 command_close_pipe($fh, $ctx);
3360 print STDERR "Done migrating from a git-svn v1 layout\n";
3361 $migrated;
3362 }
3363
3364 sub read_old_urls {
3365 my ($l_map, $pfx, $path) = @_;
3366 my @dir;
3367 foreach (<$path/*>) {
3368 if (-r "$_/info/url") {
3369 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3370 my $ref_id = $pfx . basename $_;
3371 my $url = ::file_to_s("$_/info/url");
3372 $l_map->{$ref_id} = $url;
3373 } elsif (-d $_) {
3374 push @dir, $_;
3375 }
3376 }
3377 foreach (@dir) {
3378 my $x = $_;
3379 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3380 read_old_urls($l_map, $x, $_);
3381 }
3382 }
3383
3384 sub migrate_from_v2 {
3385 my @cfg = command(qw/config -l/);
3386 return if grep /^svn-remote\..+\.url=/, @cfg;
3387 my %l_map;
3388 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3389 my $migrated = 0;
3390
3391 foreach my $ref_id (sort keys %l_map) {
3392 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3393 if ($@) {
3394 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3395 }
3396 $migrated++;
3397 }
3398 $migrated;
3399 }
3400
3401 sub minimize_connections {
3402 my $r = Git::SVN::read_all_remotes();
3403 my $new_urls = {};
3404 my $root_repos = {};
3405 foreach my $repo_id (keys %$r) {
3406 my $url = $r->{$repo_id}->{url} or next;
3407 my $fetch = $r->{$repo_id}->{fetch} or next;
3408 my $ra = Git::SVN::Ra->new($url);
3409
3410 # skip existing cases where we already connect to the root
3411 if (($ra->{url} eq $ra->{repos_root}) ||
3412 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3413 $repo_id)) {
3414 $root_repos->{$ra->{url}} = $repo_id;
3415 next;
3416 }
3417
3418 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3419 my $root_path = $ra->{url};
3420 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3421 foreach my $path (keys %$fetch) {
3422 my $ref_id = $fetch->{$path};
3423 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3424
3425 # make sure we can read when connecting to
3426 # a higher level of a repository
3427 my ($last_rev, undef) = $gs->last_rev_commit;
3428 if (!defined $last_rev) {
3429 $last_rev = eval {
3430 $root_ra->get_latest_revnum;
3431 };
3432 next if $@;
3433 }
3434 my $new = $root_path;
3435 $new .= length $path ? "/$path" : '';
3436 eval {
3437 $root_ra->get_log([$new], $last_rev, $last_rev,
3438 0, 0, 1, sub { });
3439 };
3440 next if $@;
3441 $new_urls->{$ra->{repos_root}}->{$new} =
3442 { ref_id => $ref_id,
3443 old_repo_id => $repo_id,
3444 old_path => $path };
3445 }
3446 }
3447
3448 my @emptied;
3449 foreach my $url (keys %$new_urls) {
3450 # see if we can re-use an existing [svn-remote "repo_id"]
3451 # instead of creating a(n ugly) new section:
3452 my $repo_id = $root_repos->{$url} ||
3453 Git::SVN::sanitize_remote_name($url);
3454
3455 my $fetch = $new_urls->{$url};
3456 foreach my $path (keys %$fetch) {
3457 my $x = $fetch->{$path};
3458 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3459 my $pfx = "svn-remote.$x->{old_repo_id}";
3460
3461 my $old_fetch = quotemeta("$x->{old_path}:".
3462 "refs/remotes/$x->{ref_id}");
3463 command_noisy(qw/config --unset/,
3464 "$pfx.fetch", '^'. $old_fetch . '$');
3465 delete $r->{$x->{old_repo_id}}->
3466 {fetch}->{$x->{old_path}};
3467 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
3468 command_noisy(qw/config --unset/,
3469 "$pfx.url");
3470 push @emptied, $x->{old_repo_id}
3471 }
3472 }
3473 }
3474 if (@emptied) {
3475 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
3476 "$ENV{GIT_DIR}/config";
3477 print STDERR <<EOF;
3478 The following [svn-remote] sections in your config file ($file) are empty
3479 and can be safely removed:
3480 EOF
3481 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
3482 }
3483 }
3484
3485 sub migration_check {
3486 migrate_from_v0();
3487 migrate_from_v1();
3488 migrate_from_v2();
3489 minimize_connections() if $_minimize;
3490 }
3491
3492 package Git::IndexInfo;
3493 use strict;
3494 use warnings;
3495 use Git qw/command_input_pipe command_close_pipe/;
3496
3497 sub new {
3498 my ($class) = @_;
3499 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
3500 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
3501 }
3502
3503 sub remove {
3504 my ($self, $path) = @_;
3505 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
3506 return ++$self->{nr};
3507 }
3508 undef;
3509 }
3510
3511 sub update {
3512 my ($self, $mode, $hash, $path) = @_;
3513 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
3514 return ++$self->{nr};
3515 }
3516 undef;
3517 }
3518
3519 sub DESTROY {
3520 my ($self) = @_;
3521 command_close_pipe($self->{gui}, $self->{ctx});
3522 }
3523
3524 package Git::SVN::GlobSpec;
3525 use strict;
3526 use warnings;
3527
3528 sub new {
3529 my ($class, $glob) = @_;
3530 my $re = $glob;
3531 $re =~ s!/+$!!g; # no need for trailing slashes
3532 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
3533 my ($left, $right) = ($1, $2);
3534 if ($nr > 1) {
3535 die "Only one '*' wildcard expansion ",
3536 "is supported (got $nr): '$glob'\n";
3537 } elsif ($nr == 0) {
3538 die "One '*' is needed for glob: '$glob'\n";
3539 }
3540 $re = quotemeta($left) . $re . quotemeta($right);
3541 if (length $left && !($left =~ s!/+$!!g)) {
3542 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
3543 }
3544 if (length $right && !($right =~ s!^/+!!g)) {
3545 die "Missing leading '/' on right side of: '$glob' ($right)\n";
3546 }
3547 my $left_re = qr/^\/\Q$left\E(\/|$)/;
3548 bless { left => $left, right => $right, left_regex => $left_re,
3549 regex => qr/$re/, glob => $glob }, $class;
3550 }
3551
3552 sub full_path {
3553 my ($self, $path) = @_;
3554 return (length $self->{left} ? "$self->{left}/" : '') .
3555 $path . (length $self->{right} ? "/$self->{right}" : '');
3556 }
3557
3558 __END__
3559
3560 Data structures:
3561
3562
3563 $remotes = { # returned by read_all_remotes()
3564 'svn' => {
3565 # svn-remote.svn.url=https://svn.musicpd.org
3566 url => 'https://svn.musicpd.org',
3567 # svn-remote.svn.fetch=mpd/trunk:trunk
3568 fetch => {
3569 'mpd/trunk' => 'trunk',
3570 },
3571 # svn-remote.svn.tags=mpd/tags/*:tags/*
3572 tags => {
3573 path => {
3574 left => 'mpd/tags',
3575 right => '',
3576 regex => qr!mpd/tags/([^/]+)$!,
3577 glob => 'tags/*',
3578 },
3579 ref => {
3580 left => 'tags',
3581 right => '',
3582 regex => qr!tags/([^/]+)$!,
3583 glob => 'tags/*',
3584 },
3585 }
3586 }
3587 };
3588
3589 $log_entry hashref as returned by libsvn_log_entry()
3590 {
3591 log => 'whitespace-formatted log entry
3592 ', # trailing newline is preserved
3593 revision => '8', # integer
3594 date => '2004-02-24T17:01:44.108345Z', # commit date
3595 author => 'committer name'
3596 };
3597
3598
3599 # this is generated by generate_diff();
3600 @mods = array of diff-index line hashes, each element represents one line
3601 of diff-index output
3602
3603 diff-index line ($m hash)
3604 {
3605 mode_a => first column of diff-index output, no leading ':',
3606 mode_b => second column of diff-index output,
3607 sha1_b => sha1sum of the final blob,
3608 chg => change type [MCRADT],
3609 file_a => original file name of a file (iff chg is 'C' or 'R')
3610 file_b => new/current file name of a file (any chg)
3611 }
3612 ;
3613
3614 # retval of read_url_paths{,_all}();
3615 $l_map = {
3616 # repository root url
3617 'https://svn.musicpd.org' => {
3618 # repository path # GIT_SVN_ID
3619 'mpd/trunk' => 'trunk',
3620 'mpd/tags/0.11.5' => 'tags/0.11.5',
3621 },
3622 }
3623
3624 Notes:
3625 I don't trust the each() function on unless I created %hash myself
3626 because the internal iterator may not have started at base.
Dscho's miscellaneous git branches