?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-svn: don't rely on do_switch + reparenting with svn(+ssh)://
[git/jrn.git] / git-svn.perl
blob4e357dfcef8bdec39932d565d77e271bbb97b979
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $SVN_URL
8 $GIT_SVN_INDEX $GIT_SVN
9 $GIT_DIR $GIT_SVN_DIR $REVDB
10 $_follow_parent $sha1 $sha1_short $_revision
11 $_cp_remote $_upgrade $_q
12 $_authors %users/;
13 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
14 $VERSION = '@@GIT_VERSION@@';
15
16 $ENV{GIT_DIR} ||= '.git';
17 $Git::SVN::default_repo_id = 'git-svn';
18 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
19
20 my $LC_ALL = $ENV{LC_ALL};
21 $Git::SVN::Log::TZ = $ENV{TZ};
22 # make sure the svn binary gives consistent output between locales and TZs:
23 $ENV{TZ} = 'UTC';
24 $ENV{LC_ALL} = 'C';
25 $| = 1; # unbuffer STDOUT
26
27 sub fatal (@) { print STDERR @_; exit 1 }
28 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
29 require SVN::Ra;
30 require SVN::Delta;
31 if ($SVN::Core::VERSION lt '1.1.0') {
32 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
33 }
34 push @Git::SVN::Ra::ISA, 'SVN::Ra';
35 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
36 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
37 use Carp qw/croak/;
38 use IO::File qw//;
39 use File::Basename qw/dirname basename/;
40 use File::Path qw/mkpath/;
41 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
42 use IPC::Open3;
43 use Git;
44
45 BEGIN {
46 my $s;
47 foreach (qw/command command_oneline command_noisy command_output_pipe
48 command_input_pipe command_close_pipe/) {
49 $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
50 "*Git::SVN::Migration::$_ = ".
51 "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
52 }
53 eval $s;
54 }
55
56 my ($SVN);
57
58 my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
59 $sha1 = qr/[a-f\d]{40}/;
60 $sha1_short = qr/[a-f\d]{4,40}/;
61 my ($_stdin, $_help, $_edit,
62 $_repack, $_repack_nr, $_repack_flags,
63 $_message, $_file, $_no_metadata,
64 $_template, $_shared,
65 $_version, $_upgrade,
66 $_merge, $_strategy, $_dry_run,
67 $_prefix);
68
69 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
70 'config-dir=s' => \$Git::SVN::Ra::config_dir,
71 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
72 my %fc_opts = ( 'follow-parent|follow' => \$_follow_parent,
73 'authors-file|A=s' => \$_authors,
74 'repack:i' => \$_repack,
75 'no-metadata' => \$_no_metadata,
76 'quiet|q' => \$_q,
77 'repack-flags|repack-args|repack-opts=s' => \$_repack_flags,
78 %remote_opts );
79
80 my ($_trunk, $_tags, $_branches);
81 my %multi_opts = ( 'trunk|T=s' => \$_trunk,
82 'tags|t=s' => \$_tags,
83 'branches|b=s' => \$_branches );
84 my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
85 my %cmt_opts = ( 'edit|e' => \$_edit,
86 'rmdir' => \$SVN::Git::Editor::_rmdir,
87 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
88 'l=i' => \$SVN::Git::Editor::_rename_limit,
89 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
90 );
91
92 my %cmd = (
93 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
94 { 'revision|r=s' => \$_revision, %fc_opts } ],
95 init => [ \&cmd_init, "Initialize a repo for tracking" .
96 " (requires URL argument)",
97 \%init_opts ],
98 dcommit => [ \&cmd_dcommit,
99 'Commit several diffs to merge with upstream',
100 { 'merge|m|M' => \$_merge,
101 'strategy|s=s' => \$_strategy,
102 'dry-run|n' => \$_dry_run,
103 %cmt_opts, %fc_opts } ],
104 'set-tree' => [ \&cmd_set_tree,
105 "Set an SVN repository to a git tree-ish",
106 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
107 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
108 { 'revision|r=i' => \$_revision } ],
109 rebuild => [ \&cmd_rebuild, "Rebuild git-svn metadata (after git clone)",
110 { 'copy-remote|remote=s' => \$_cp_remote,
111 'upgrade' => \$_upgrade } ],
112 'multi-init' => [ \&cmd_multi_init,
113 'Initialize multiple trees (like git-svnimport)',
114 { %multi_opts, %init_opts, %remote_opts,
115 'revision|r=i' => \$_revision,
116 'prefix=s' => \$_prefix,
117 } ],
118 'multi-fetch' => [ \&cmd_multi_fetch,
119 'Fetch multiple trees (like git-svnimport)',
120 \%fc_opts ],
121 'migrate' => [ sub { },
122 # no-op, we automatically run this anyways,
123 'Migrate configuration/metadata/layout from
124 previous versions of git-svn',
125 \%remote_opts ],
126 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
127 { 'limit=i' => \$Git::SVN::Log::limit,
128 'revision|r=s' => \$_revision,
129 'verbose|v' => \$Git::SVN::Log::verbose,
130 'incremental' => \$Git::SVN::Log::incremental,
131 'oneline' => \$Git::SVN::Log::oneline,
132 'show-commit' => \$Git::SVN::Log::show_commit,
133 'non-recursive' => \$Git::SVN::Log::non_recursive,
134 'authors-file|A=s' => \$_authors,
135 'color' => \$Git::SVN::Log::color,
136 'pager=s' => \$Git::SVN::Log::pager,
137 } ],
138 'commit-diff' => [ \&cmd_commit_diff,
139 'Commit a diff between two trees',
140 { 'message|m=s' => \$_message,
141 'file|F=s' => \$_file,
142 'revision|r=s' => \$_revision,
143 %cmt_opts } ],
144 );
145
146 my $cmd;
147 for (my $i = 0; $i < @ARGV; $i++) {
148 if (defined $cmd{$ARGV[$i]}) {
149 $cmd = $ARGV[$i];
150 splice @ARGV, $i, 1;
151 last;
152 }
153 };
154
155 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
156
157 read_repo_config(\%opts);
158 my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
159 'version|V' => \$_version,
160 'minimize-connections' =>
161 \$Git::SVN::Migration::_minimize,
162 'id|i=s' => \$Git::SVN::default_ref_id);
163 exit 1 if (!$rv && $cmd ne 'log');
164
165 usage(0) if $_help;
166 version() if $_version;
167 usage(1) unless defined $cmd;
168 load_authors() if $_authors;
169 unless ($cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/) {
170 Git::SVN::Migration::migration_check();
171 }
172 eval {
173 Git::SVN::verify_remotes_sanity();
174 $cmd{$cmd}->[0]->(@ARGV);
175 };
176 fatal $@ if $@;
177 exit 0;
178
179 ####################### primary functions ######################
180 sub usage {
181 my $exit = shift || 0;
182 my $fd = $exit ? \*STDERR : \*STDOUT;
183 print $fd <<"";
184 git-svn - bidirectional operations between a single Subversion tree and git
185 Usage: $0 <command> [options] [arguments]\n
186
187 print $fd "Available commands:\n" unless $cmd;
188
189 foreach (sort keys %cmd) {
190 next if $cmd && $cmd ne $_;
191 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
192 foreach (keys %{$cmd{$_}->[2]}) {
193 # prints out arguments as they should be passed:
194 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
195 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
196 "--$_" : "-$_" }
197 split /\|/,$_)," $x\n";
198 }
199 }
200 print $fd <<"";
201 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
202 arbitrary identifier if you're tracking multiple SVN branches/repositories in
203 one git repository and want to keep them separate. See git-svn(1) for more
204 information.
205
206 exit $exit;
207 }
208
209 sub version {
210 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
211 exit 0;
212 }
213
214 sub cmd_rebuild {
215 my $url = shift;
216 my $gs = $url ? Git::SVN->init($url)
217 : eval { Git::SVN->new };
218 $gs ||= Git::SVN->_new;
219 if (!verify_ref($gs->refname.'^0')) {
220 $gs->copy_remote_ref;
221 }
222
223 my ($rev_list, $ctx) = command_output_pipe("rev-list", $gs->refname);
224 my $latest;
225 my $svn_uuid;
226 while (<$rev_list>) {
227 chomp;
228 my $c = $_;
229 fatal "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
230 my ($url, $rev, $uuid) = cmt_metadata($c);
231
232 # ignore merges (from set-tree)
233 next if (!defined $rev || !$uuid);
234
235 # if we merged or otherwise started elsewhere, this is
236 # how we break out of it
237 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
238 ($gs->{url} && $url && ($url ne $gs->{url}))) {
239 next;
240 }
241
242 unless (defined $latest) {
243 if (!$gs->{url} && !$url) {
244 fatal "SVN repository location required\n";
245 }
246 $gs = Git::SVN->init($url);
247 $latest = $rev;
248 }
249 $gs->rev_db_set($rev, $c);
250 print "r$rev = $c\n";
251 }
252 command_close_pipe($rev_list, $ctx);
253 }
254
255 sub do_git_init_db {
256 unless (-d $ENV{GIT_DIR}) {
257 my @init_db = ('init');
258 push @init_db, "--template=$_template" if defined $_template;
259 push @init_db, "--shared" if defined $_shared;
260 command_noisy(@init_db);
261 }
262 }
263
264 sub cmd_init {
265 my $url = shift or die "SVN repository location required " .
266 "as a command-line argument\n";
267 if (my $repo_path = shift) {
268 unless (-d $repo_path) {
269 mkpath([$repo_path]);
270 }
271 chdir $repo_path or croak $!;
272 $ENV{GIT_DIR} = $repo_path . "/.git";
273 }
274 do_git_init_db();
275
276 Git::SVN->init($url);
277 }
278
279 sub cmd_fetch {
280 if (@_) {
281 die "Additional fetch arguments are no longer supported.\n",
282 "Use --follow-parent if you have moved/copied directories
283 instead.\n";
284 }
285 my $gs = Git::SVN->new;
286 $gs->fetch(parse_revision_argument());
287 if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
288 command_noisy(qw(update-ref refs/heads/master),
289 $gs->{last_commit});
290 }
291 }
292
293 sub cmd_set_tree {
294 my (@commits) = @_;
295 if ($_stdin || !@commits) {
296 print "Reading from stdin...\n";
297 @commits = ();
298 while (<STDIN>) {
299 if (/\b($sha1_short)\b/o) {
300 unshift @commits, $1;
301 }
302 }
303 }
304 my @revs;
305 foreach my $c (@commits) {
306 my @tmp = command('rev-parse',$c);
307 if (scalar @tmp == 1) {
308 push @revs, $tmp[0];
309 } elsif (scalar @tmp > 1) {
310 push @revs, reverse(command('rev-list',@tmp));
311 } else {
312 fatal "Failed to rev-parse $c\n";
313 }
314 }
315 my $gs = Git::SVN->new;
316 my ($r_last, $cmt_last) = $gs->last_rev_commit;
317 $gs->fetch;
318 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
319 fatal "There are new revisions that were fetched ",
320 "and need to be merged (or acknowledged) ",
321 "before committing.\nlast rev: $r_last\n",
322 " current: $gs->{last_rev}\n";
323 }
324 $gs->set_tree($_) foreach @revs;
325 print "Done committing ",scalar @revs," revisions to SVN\n";
326 }
327
328 sub cmd_dcommit {
329 my $head = shift;
330 my $gs = Git::SVN->new;
331 $head ||= 'HEAD';
332 my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
333 my $last_rev;
334 foreach my $d (reverse @refs) {
335 if (!verify_ref("$d~1")) {
336 fatal "Commit $d\n",
337 "has no parent commit, and therefore ",
338 "nothing to diff against.\n",
339 "You should be working from a repository ",
340 "originally created by git-svn\n";
341 }
342 unless (defined $last_rev) {
343 (undef, $last_rev, undef) = cmt_metadata("$d~1");
344 unless (defined $last_rev) {
345 fatal "Unable to extract revision information ",
346 "from commit $d~1\n";
347 }
348 }
349 if ($_dry_run) {
350 print "diff-tree $d~1 $d\n";
351 } else {
352 my %ed_opts = ( r => $last_rev,
353 log => get_commit_entry($d)->{log},
354 ra => $gs->ra,
355 tree_a => "$d~1",
356 tree_b => $d,
357 editor_cb => sub {
358 print "Committed r$_[0]\n";
359 $last_rev = $_[0]; },
360 svn_path => $gs->{path} );
361 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
362 print "No changes\n$d~1 == $d\n";
363 }
364 }
365 }
366 return if $_dry_run;
367 $gs->fetch;
368 # we always want to rebase against the current HEAD, not any
369 # head that was passed to us
370 my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
371 my @finish;
372 if (@diff) {
373 @finish = qw/rebase/;
374 push @finish, qw/--merge/ if $_merge;
375 push @finish, "--strategy=$_strategy" if $_strategy;
376 print STDERR "W: HEAD and ", $gs->refname, " differ, ",
377 "using @finish:\n", "@diff";
378 } else {
379 print "No changes between current HEAD and ",
380 $gs->refname, "\nResetting to the latest ",
381 $gs->refname, "\n";
382 @finish = qw/reset --mixed/;
383 }
384 command_noisy(@finish, $gs->refname);
385 }
386
387 sub cmd_show_ignore {
388 my $gs = Git::SVN->new;
389 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
390 $gs->traverse_ignore(\*STDOUT, '', $r);
391 }
392
393 sub cmd_multi_init {
394 my $url = shift;
395 unless (defined $_trunk || defined $_branches || defined $_tags) {
396 usage(1);
397 }
398 do_git_init_db();
399 $_prefix = '' unless defined $_prefix;
400 $url =~ s#/+$## if defined $url;
401 if (defined $_trunk) {
402 my $trunk_ref = $_prefix . 'trunk';
403 # try both old-style and new-style lookups:
404 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
405 unless ($gs_trunk) {
406 my ($trunk_url, $trunk_path) =
407 complete_svn_url($url, $_trunk);
408 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
409 undef, $trunk_ref);
410 }
411 }
412 return unless defined $_branches || defined $_tags;
413 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
414 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
415 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
416 }
417
418 sub cmd_multi_fetch {
419 my $remotes = Git::SVN::read_all_remotes();
420 foreach my $repo_id (sort keys %$remotes) {
421 my $url = $remotes->{$repo_id}->{url} or next;
422 my $fetch = $remotes->{$repo_id}->{fetch} or next;
423 Git::SVN::fetch_all($repo_id, $url, $fetch);
424 }
425 }
426
427 # this command is special because it requires no metadata
428 sub cmd_commit_diff {
429 my ($ta, $tb, $url) = @_;
430 my $usage = "Usage: $0 commit-diff -r<revision> ".
431 "<tree-ish> <tree-ish> [<URL>]\n";
432 fatal($usage) if (!defined $ta || !defined $tb);
433 my $svn_path;
434 if (!defined $url) {
435 my $gs = eval { Git::SVN->new };
436 if (!$gs) {
437 fatal("Needed URL or usable git-svn --id in ",
438 "the command-line\n", $usage);
439 }
440 $url = $gs->{url};
441 $svn_path = $gs->{path};
442 }
443 unless (defined $_revision) {
444 fatal("-r|--revision is a required argument\n", $usage);
445 }
446 if (defined $_message && defined $_file) {
447 fatal("Both --message/-m and --file/-F specified ",
448 "for the commit message.\n",
449 "I have no idea what you mean\n");
450 }
451 if (defined $_file) {
452 $_message = file_to_s($_file);
453 } else {
454 $_message ||= get_commit_entry($tb)->{log};
455 }
456 my $ra ||= Git::SVN::Ra->new($url);
457 $svn_path ||= $ra->{svn_path};
458 my $r = $_revision;
459 if ($r eq 'HEAD') {
460 $r = $ra->get_latest_revnum;
461 } elsif ($r !~ /^\d+$/) {
462 die "revision argument: $r not understood by git-svn\n";
463 }
464 my %ed_opts = ( r => $r,
465 log => $_message,
466 ra => $ra,
467 tree_a => $ta,
468 tree_b => $tb,
469 editor_cb => sub { print "Committed r$_[0]\n" },
470 svn_path => $svn_path );
471 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
472 print "No changes\n$ta == $tb\n";
473 }
474 }
475
476 ########################### utility functions #########################
477
478 sub parse_revision_argument {
479 if (!defined $_revision || $_revision eq 'BASE:HEAD') {
480 return (undef, undef);
481 }
482 return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
483 return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
484 return (undef, $1) if ($_revision =~ /^BASE:(\d+)$/);
485 return ($1, undef) if ($_revision =~ /^(\d+):HEAD$/);
486 die "revision argument: $_revision not understood by git-svn\n",
487 "Try using the command-line svn client instead\n";
488 }
489
490 sub complete_svn_url {
491 my ($url, $path) = @_;
492 $path =~ s#/+$##;
493 if ($path !~ m#^[a-z\+]+://#) {
494 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
495 fatal("E: '$path' is not a complete URL ",
496 "and a separate URL is not specified\n");
497 }
498 return ($url, $path);
499 }
500 return ($path, '');
501 }
502
503 sub complete_url_ls_init {
504 my ($ra, $repo_path, $switch, $pfx) = @_;
505 unless ($repo_path) {
506 print STDERR "W: $switch not specified\n";
507 return;
508 }
509 $repo_path =~ s#/+$##;
510 if ($repo_path =~ m#^[a-z\+]+://#) {
511 $ra = Git::SVN::Ra->new($repo_path);
512 $repo_path = '';
513 } else {
514 $repo_path =~ s#^/+##;
515 unless ($ra) {
516 fatal("E: '$repo_path' is not a complete URL ",
517 "and a separate URL is not specified\n");
518 }
519 }
520 my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
521 my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
522 my $url = $ra->{url};
523 foreach my $d (sort keys %$dirent) {
524 next if ($dirent->{$d}->kind != $SVN::Node::dir);
525 my $path = "$repo_path/$d";
526 my $ref = "$pfx$d";
527 my $gs = eval { Git::SVN->new($ref) };
528 # don't try to init already existing refs
529 unless ($gs) {
530 print "init $url/$path => $ref\n";
531 Git::SVN->init($url, $path, undef, $ref);
532 }
533 }
534 }
535
536 sub verify_ref {
537 my ($ref) = @_;
538 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
539 { STDERR => 0 }); };
540 }
541
542 sub get_tree_from_treeish {
543 my ($treeish) = @_;
544 # $treeish can be a symbolic ref, too:
545 my $type = command_oneline(qw/cat-file -t/, $treeish);
546 my $expected;
547 while ($type eq 'tag') {
548 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
549 }
550 if ($type eq 'commit') {
551 $expected = (grep /^tree /, command(qw/cat-file commit/,
552 $treeish))[0];
553 ($expected) = ($expected =~ /^tree ($sha1)$/o);
554 die "Unable to get tree from $treeish\n" unless $expected;
555 } elsif ($type eq 'tree') {
556 $expected = $treeish;
557 } else {
558 die "$treeish is a $type, expected tree, tag or commit\n";
559 }
560 return $expected;
561 }
562
563 sub get_commit_entry {
564 my ($treeish) = shift;
565 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
566 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
567 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
568 open my $log_fh, '>', $commit_editmsg or croak $!;
569
570 my $type = command_oneline(qw/cat-file -t/, $treeish);
571 if ($type eq 'commit' || $type eq 'tag') {
572 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
573 $type, $treeish);
574 my $in_msg = 0;
575 while (<$msg_fh>) {
576 if (!$in_msg) {
577 $in_msg = 1 if (/^\s*$/);
578 } elsif (/^git-svn-id: /) {
579 # skip this for now, we regenerate the
580 # correct one on re-fetch anyways
581 # TODO: set *:merge properties or like...
582 } else {
583 print $log_fh $_ or croak $!;
584 }
585 }
586 command_close_pipe($msg_fh, $ctx);
587 }
588 close $log_fh or croak $!;
589
590 if ($_edit || ($type eq 'tree')) {
591 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
592 # TODO: strip out spaces, comments, like git-commit.sh
593 system($editor, $commit_editmsg);
594 }
595 rename $commit_editmsg, $commit_msg or croak $!;
596 open $log_fh, '<', $commit_msg or croak $!;
597 { local $/; chomp($log_entry{log} = <$log_fh>); }
598 close $log_fh or croak $!;
599 unlink $commit_msg;
600 \%log_entry;
601 }
602
603 sub s_to_file {
604 my ($str, $file, $mode) = @_;
605 open my $fd,'>',$file or croak $!;
606 print $fd $str,"\n" or croak $!;
607 close $fd or croak $!;
608 chmod ($mode &~ umask, $file) if (defined $mode);
609 }
610
611 sub file_to_s {
612 my $file = shift;
613 open my $fd,'<',$file or croak "$!: file: $file\n";
614 local $/;
615 my $ret = <$fd>;
616 close $fd or croak $!;
617 $ret =~ s/\s*$//s;
618 return $ret;
619 }
620
621 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
622 sub load_authors {
623 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
624 my $log = $cmd eq 'log';
625 while (<$authors>) {
626 chomp;
627 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
628 my ($user, $name, $email) = ($1, $2, $3);
629 if ($log) {
630 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
631 } else {
632 $users{$user} = [$name, $email];
633 }
634 }
635 close $authors or croak $!;
636 }
637
638 # convert GetOpt::Long specs for use by git-config
639 sub read_repo_config {
640 return unless -d $ENV{GIT_DIR};
641 my $opts = shift;
642 foreach my $o (keys %$opts) {
643 my $v = $opts->{$o};
644 my ($key) = ($o =~ /^([a-z\-]+)/);
645 $key =~ s/-//g;
646 my $arg = 'git-config';
647 $arg .= ' --int' if ($o =~ /[:=]i$/);
648 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
649 if (ref $v eq 'ARRAY') {
650 chomp(my @tmp = `$arg --get-all svn.$key`);
651 @$v = @tmp if @tmp;
652 } else {
653 chomp(my $tmp = `$arg --get svn.$key`);
654 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
655 $$v = $tmp;
656 }
657 }
658 }
659 }
660
661 sub extract_metadata {
662 my $id = shift or return (undef, undef, undef);
663 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
664 \s([a-f\d\-]+)$/x);
665 if (!defined $rev || !$uuid || !$url) {
666 # some of the original repositories I made had
667 # identifiers like this:
668 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
669 }
670 return ($url, $rev, $uuid);
671 }
672
673 sub cmt_metadata {
674 return extract_metadata((grep(/^git-svn-id: /,
675 command(qw/cat-file commit/, shift)))[-1]);
676 }
677
678 package Git::SVN;
679 use strict;
680 use warnings;
681 use vars qw/$default_repo_id $default_ref_id/;
682 use Carp qw/croak/;
683 use File::Path qw/mkpath/;
684 use IPC::Open3;
685
686 # properties that we do not log:
687 my %SKIP_PROP;
688 BEGIN {
689 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
690 svn:special svn:executable
691 svn:entry:committed-rev
692 svn:entry:last-author
693 svn:entry:uuid
694 svn:entry:committed-date/;
695 }
696
697 sub fetch_all {
698 my ($repo_id, $url, $fetch) = @_;
699 my @gs;
700 my $ra = Git::SVN::Ra->new($url);
701 my $head = $ra->get_latest_revnum;
702 my $base = $head;
703 my $new_remote;
704 foreach my $p (sort keys %$fetch) {
705 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
706 my $lr = $gs->last_rev;
707 if (defined $lr) {
708 $base = $lr if ($lr < $base);
709 } else {
710 $new_remote = 1;
711 }
712 push @gs, $gs;
713 }
714 $base = 0 if $new_remote;
715 return if (++$base > $head);
716 $ra->gs_fetch_loop_common($base, $head, @gs);
717 }
718
719 sub read_all_remotes {
720 my $r = {};
721 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
722 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
723 $r->{$1}->{fetch}->{$2} = $3;
724 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
725 $r->{$1}->{url} = $2;
726 }
727 }
728 $r;
729 }
730
731 sub verify_remotes_sanity {
732 return unless -d $ENV{GIT_DIR};
733 my %seen;
734 foreach (command(qw/config -l/)) {
735 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
736 if ($seen{$1}) {
737 die "Remote ref refs/remote/$1 is tracked by",
738 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
739 "Please resolve this ambiguity in ",
740 "your git configuration file before ",
741 "continuing\n";
742 }
743 $seen{$1} = $_;
744 }
745 }
746 }
747
748 # we allow more chars than remotes2config.sh...
749 sub sanitize_remote_name {
750 my ($name) = @_;
751 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
752 $name;
753 }
754
755 sub find_existing_remote {
756 my ($url, $remotes) = @_;
757 my $existing;
758 foreach my $repo_id (keys %$remotes) {
759 my $u = $remotes->{$repo_id}->{url} or next;
760 next if $u ne $url;
761 $existing = $repo_id;
762 last;
763 }
764 $existing;
765 }
766
767 sub init_remote_config {
768 my ($self, $url) = @_;
769 $url =~ s!/+$!!; # strip trailing slash
770 my $r = read_all_remotes();
771 my $existing = find_existing_remote($url, $r);
772 if ($existing) {
773 print STDERR "Using existing ",
774 "[svn-remote \"$existing\"]\n";
775 $self->{repo_id} = $existing;
776 } else {
777 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
778 $existing = find_existing_remote($min_url, $r);
779 if ($existing) {
780 print STDERR "Using existing ",
781 "[svn-remote \"$existing\"]\n";
782 $self->{repo_id} = $existing;
783 }
784 if ($min_url ne $url) {
785 print STDERR "Using higher level of URL: ",
786 "$url => $min_url\n";
787 my $old_path = $self->{path};
788 $self->{path} = $url;
789 $self->{path} =~ s!^\Q$min_url\E/*!!;
790 if (length $old_path) {
791 $self->{path} .= "/$old_path";
792 }
793 $url = $min_url;
794 }
795 }
796 my $orig_url;
797 if (!$existing) {
798 # verify that we aren't overwriting anything:
799 $orig_url = eval {
800 command_oneline('config', '--get',
801 "svn-remote.$self->{repo_id}.url")
802 };
803 if ($orig_url && ($orig_url ne $url)) {
804 die "svn-remote.$self->{repo_id}.url already set: ",
805 "$orig_url\nwanted to set to: $url\n";
806 }
807 }
808 my ($xrepo_id, $xpath) = find_ref($self->refname);
809 if (defined $xpath) {
810 die "svn-remote.$xrepo_id.fetch already set to track ",
811 "$xpath:refs/remotes/", $self->refname, "\n";
812 }
813 command_noisy('config',
814 "svn-remote.$self->{repo_id}.url", $url);
815 command_noisy('config', '--add',
816 "svn-remote.$self->{repo_id}.fetch",
817 "$self->{path}:".$self->refname);
818 $self->{url} = $url;
819 }
820
821 sub init {
822 my ($class, $url, $path, $repo_id, $ref_id) = @_;
823 my $self = _new($class, $repo_id, $ref_id, $path);
824 if (defined $url) {
825 $self->init_remote_config($url);
826 }
827 $self;
828 }
829
830 sub find_ref {
831 my ($ref_id) = @_;
832 foreach (command(qw/config -l/)) {
833 next unless m!^svn-remote\.(.+)\.fetch=
834 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
835 my ($repo_id, $path, $ref) = ($1, $2, $3);
836 if ($ref eq $ref_id) {
837 $path = '' if ($path =~ m#^\./?#);
838 return ($repo_id, $path);
839 }
840 }
841 (undef, undef, undef);
842 }
843
844 sub new {
845 my ($class, $ref_id, $repo_id, $path) = @_;
846 if (defined $ref_id && !defined $repo_id && !defined $path) {
847 ($repo_id, $path) = find_ref($ref_id);
848 if (!defined $repo_id) {
849 die "Could not find a \"svn-remote.*.fetch\" key ",
850 "in the repository configuration matching: ",
851 "refs/remotes/$ref_id\n";
852 }
853 }
854 my $self = _new($class, $repo_id, $ref_id, $path);
855 if (!defined $self->{path} || !length $self->{path}) {
856 my $fetch = command_oneline('config', '--get',
857 "svn-remote.$repo_id.fetch",
858 ":refs/remotes/$ref_id\$") or
859 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
860 "\":refs/remotes/$ref_id\$\" in config\n";
861 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
862 }
863 $self->{url} = command_oneline('config', '--get',
864 "svn-remote.$repo_id.url") or
865 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
866 $self;
867 }
868
869 sub refname { "refs/remotes/$_[0]->{ref_id}" }
870
871 sub ra {
872 my ($self) = shift;
873 Git::SVN::Ra->new($self->{url});
874 }
875
876 sub rel_path {
877 my ($self) = @_;
878 my $repos_root = $self->ra->{repos_root};
879 return $self->{path} if ($self->{url} eq $repos_root);
880 my $url = $self->{url} .
881 (length $self->{path} ? "/$self->{path}" : $self->{path});
882 $url =~ s!^\Q$repos_root\E/*!!g;
883 $url;
884 }
885
886 sub copy_remote_ref {
887 my ($self) = @_;
888 my $origin = $::_cp_remote ? $::_cp_remote : 'origin';
889 my $ref = $self->refname;
890 if (command('ls-remote', $origin, $ref)) {
891 command_noisy('fetch', $origin, "$ref:$ref");
892 } elsif ($::_cp_remote && !$::_upgrade) {
893 die "Unable to find remote reference: $ref on $origin\n";
894 }
895 }
896
897 sub traverse_ignore {
898 my ($self, $fh, $path, $r) = @_;
899 $path =~ s#^/+##g;
900 my $ra = $self->ra;
901 my ($dirent, undef, $props) = $ra->get_dir($path, $r);
902 my $p = $path;
903 $p =~ s#^\Q$ra->{svn_path}\E/##;
904 print $fh length $p ? "\n# $p\n" : "\n# /\n";
905 if (my $s = $props->{'svn:ignore'}) {
906 $s =~ s/[\r\n]+/\n/g;
907 chomp $s;
908 if (length $p == 0) {
909 $s =~ s#\n#\n/$p#g;
910 print $fh "/$s\n";
911 } else {
912 $s =~ s#\n#\n/$p/#g;
913 print $fh "/$p/$s\n";
914 }
915 }
916 foreach (sort keys %$dirent) {
917 next if $dirent->{$_}->kind != $SVN::Node::dir;
918 $self->traverse_ignore($fh, "$path/$_", $r);
919 }
920 }
921
922 sub last_rev { ($_[0]->last_rev_commit)[0] }
923 sub last_commit { ($_[0]->last_rev_commit)[1] }
924
925 # returns the newest SVN revision number and newest commit SHA1
926 sub last_rev_commit {
927 my ($self) = @_;
928 if (defined $self->{last_rev} && defined $self->{last_commit}) {
929 return ($self->{last_rev}, $self->{last_commit});
930 }
931 my $c = ::verify_ref($self->refname.'^0');
932 if ($c) {
933 my $rev = (::cmt_metadata($c))[1];
934 if (defined $rev) {
935 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
936 return ($rev, $c);
937 }
938 }
939 my $offset = -41; # from tail
940 my $rl;
941 open my $fh, '<', $self->{db_path} or
942 croak "$self->{db_path} not readable: $!\n";
943 seek $fh, $offset, 2;
944 $rl = readline $fh;
945 defined $rl or return (undef, undef);
946 chomp $rl;
947 while ($c ne $rl && tell $fh != 0) {
948 $offset -= 41;
949 seek $fh, $offset, 2;
950 $rl = readline $fh;
951 defined $rl or return (undef, undef);
952 chomp $rl;
953 }
954 my $rev = tell $fh;
955 croak $! if ($rev < 0);
956 $rev = ($rev - 41) / 41;
957 close $fh or croak $!;
958 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
959 return ($rev, $c);
960 }
961
962 sub get_fetch_range {
963 my ($self, $min, $max) = @_;
964 $max ||= $self->ra->get_latest_revnum;
965 $min ||= $self->last_rev || 0;
966 (++$min, $max);
967 }
968
969 sub tmp_index_do {
970 my ($self, $sub) = @_;
971 my $old_index = $ENV{GIT_INDEX_FILE};
972 $ENV{GIT_INDEX_FILE} = $self->{index};
973 my @ret = &$sub;
974 if ($old_index) {
975 $ENV{GIT_INDEX_FILE} = $old_index;
976 } else {
977 delete $ENV{GIT_INDEX_FILE};
978 }
979 wantarray ? @ret : $ret[0];
980 }
981
982 sub assert_index_clean {
983 my ($self, $treeish) = @_;
984
985 $self->tmp_index_do(sub {
986 command_noisy('read-tree', $treeish) unless -e $self->{index};
987 my $x = command_oneline('write-tree');
988 my ($y) = (command(qw/cat-file commit/, $treeish) =~
989 /^tree ($::sha1)/mo);
990 if ($y ne $x) {
991 unlink $self->{index} or croak $!;
992 command_noisy('read-tree', $treeish);
993 }
994 $x = command_oneline('write-tree');
995 if ($y ne $x) {
996 ::fatal "trees ($treeish) $y != $x\n",
997 "Something is seriously wrong...\n";
998 }
999 });
1000 }
1001
1002 sub get_commit_parents {
1003 my ($self, $log_entry) = @_;
1004 my (%seen, @ret, @tmp);
1005 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1006 if (my $ip = $self->{inject_parents}) {
1007 if (my $commit = delete $ip->{$log_entry->{revision}}) {
1008 push @tmp, $commit;
1009 }
1010 }
1011 if (my $cur = ::verify_ref($self->refname.'^0')) {
1012 push @tmp, $cur;
1013 }
1014 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1015 while (my $p = shift @tmp) {
1016 next if $seen{$p};
1017 $seen{$p} = 1;
1018 push @ret, $p;
1019 # MAXPARENT is defined to 16 in commit-tree.c:
1020 last if @ret >= 16;
1021 }
1022 if (@tmp) {
1023 die "r$log_entry->{revision}: No room for parents:\n\t",
1024 join("\n\t", @tmp), "\n";
1025 }
1026 @ret;
1027 }
1028
1029 sub full_url {
1030 my ($self) = @_;
1031 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1032 }
1033
1034 sub do_git_commit {
1035 my ($self, $log_entry) = @_;
1036 if (my $c = $self->rev_db_get($log_entry->{revision})) {
1037 croak "$log_entry->{revision} = $c already exists! ",
1038 "Why are we refetching it?\n";
1039 }
1040 my $author = $log_entry->{author};
1041 my ($name, $email) = (defined $::users{$author} ? @{$::users{$author}}
1042 : ($author, "$author\@".$self->ra->uuid));
1043 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1044 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1045 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1046
1047 my $tree = $log_entry->{tree};
1048 if (!defined $tree) {
1049 $tree = $self->tmp_index_do(sub {
1050 command_oneline('write-tree') });
1051 }
1052 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1053
1054 my @exec = ('git-commit-tree', $tree);
1055 foreach ($self->get_commit_parents($log_entry)) {
1056 push @exec, '-p', $_;
1057 }
1058 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1059 or croak $!;
1060 print $msg_fh $log_entry->{log} or croak $!;
1061 print $msg_fh "\ngit-svn-id: ", $self->full_url, '@',
1062 $log_entry->{revision}, ' ',
1063 $self->ra->uuid, "\n" or croak $!;
1064 $msg_fh->flush == 0 or croak $!;
1065 close $msg_fh or croak $!;
1066 chomp(my $commit = do { local $/; <$out_fh> });
1067 close $out_fh or croak $!;
1068 waitpid $pid, 0;
1069 croak $? if $?;
1070 if ($commit !~ /^$::sha1$/o) {
1071 die "Failed to commit, invalid sha1: $commit\n";
1072 }
1073
1074 command_noisy('update-ref',$self->refname, $commit);
1075 $self->rev_db_set($log_entry->{revision}, $commit);
1076
1077 $self->{last_rev} = $log_entry->{revision};
1078 $self->{last_commit} = $commit;
1079 print "r$log_entry->{revision} = $commit\n";
1080 return $commit;
1081 }
1082
1083 sub revisions_eq {
1084 my ($self, $r0, $r1) = @_;
1085 return 1 if $r0 == $r1;
1086 my $nr = 0;
1087 $self->ra->get_log([$self->{path}], $r0, $r1,
1088 0, 0, 1, sub { $nr++ });
1089 return 0 if ($nr > 1);
1090 return 1;
1091 }
1092
1093 sub find_parent_branch {
1094 my ($self, $paths, $rev) = @_;
1095 return undef unless $::_follow_parent;
1096 unless (defined $paths) {
1097 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
1098 sub { $paths = dup_changed_paths($_[0]) });
1099 }
1100 return undef unless defined $paths;
1101
1102 # look for a parent from another branch:
1103 my @b_path_components = split m#/#, $self->rel_path;
1104 my @a_path_components;
1105 my $i;
1106 while (@b_path_components) {
1107 $i = $paths->{'/'.join('/', @b_path_components)};
1108 last if $i;
1109 unshift(@a_path_components, pop(@b_path_components));
1110 }
1111 goto not_found unless defined $i;
1112 my $branch_from = $i->{copyfrom_path} or goto not_found;
1113 if (@a_path_components) {
1114 print STDERR "branch_from: $branch_from => ";
1115 $branch_from .= '/'.join('/', @a_path_components);
1116 print STDERR $branch_from, "\n";
1117 }
1118 my $r = $i->{copyfrom_rev};
1119 my $repos_root = $self->ra->{repos_root};
1120 my $url = $self->ra->{url};
1121 my $new_url = $repos_root . $branch_from;
1122 print STDERR "Found possible branch point: ",
1123 "$new_url => ", $self->full_url, ", $r\n";
1124 $branch_from =~ s#^/##;
1125 my $remotes = read_all_remotes();
1126 my $gs;
1127 foreach my $repo_id (keys %$remotes) {
1128 my $u = $remotes->{$repo_id}->{url} or next;
1129 next if $url ne $u;
1130 my $fetch = $remotes->{$repo_id}->{fetch};
1131 foreach my $f (keys %$fetch) {
1132 next if $f ne $branch_from;
1133 $gs = Git::SVN->new($fetch->{$f}, $repo_id, $f);
1134 last;
1135 }
1136 last if $gs;
1137 }
1138 unless ($gs) {
1139 my $ref_id = $self->{ref_id};
1140 $ref_id =~ s/\@\d+$//;
1141 $ref_id .= "\@$r";
1142 # just grow a tail if we're not unique enough :x
1143 $ref_id .= '-' while find_ref($ref_id);
1144 print STDERR "Initializing parent: $ref_id\n";
1145 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id);
1146 }
1147 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1148 if ($::_follow_parent && (!defined $r0 || !defined $parent)) {
1149 $gs->fetch(0, $r);
1150 ($r0, $parent) = $gs->last_rev_commit;
1151 }
1152 if (defined $r0 && defined $parent && $gs->revisions_eq($r0, $r)) {
1153 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1154 $self->assert_index_clean($parent);
1155 my $ed;
1156 if ($self->ra->can_do_switch) {
1157 print STDERR "Following parent with do_switch\n";
1158 # do_switch works with svn/trunk >= r22312, but that
1159 # is not included with SVN 1.4.3 (the latest version
1160 # at the moment), so we can't rely on it
1161 $self->{last_commit} = $parent;
1162 $ed = SVN::Git::Fetcher->new($self);
1163 $gs->ra->gs_do_switch($r0, $rev, $gs->{path}, 1,
1164 $self->full_url, $ed)
1165 or die "SVN connection failed somewhere...\n";
1166 } else {
1167 print STDERR "Following parent with do_update\n";
1168 $ed = SVN::Git::Fetcher->new($self);
1169 $self->ra->gs_do_update($rev, $rev, $self->{path},
1170 1, $ed)
1171 or die "SVN connection failed somewhere...\n";
1172 }
1173 print STDERR "Successfully followed parent\n";
1174 $ed->{new_fetch} = 1;
1175 return $self->make_log_entry($rev, [$parent], $ed);
1176 }
1177 not_found:
1178 print STDERR "Branch parent for path: '/",
1179 $self->rel_path, "' @ r$rev not found:\n";
1180 return undef unless $paths;
1181 print STDERR "Changed paths:\n";
1182 foreach my $x (sort keys %$paths) {
1183 my $p = $paths->{$x};
1184 print STDERR "\t$p->{action}\t$x";
1185 if ($p->{copyfrom_path}) {
1186 print STDERR "(from $p->{copyfrom_path}: ",
1187 "$p->{copyfrom_rev})";
1188 }
1189 print STDERR "\n";
1190 }
1191 print STDERR '-'x72, "\n";
1192 return undef;
1193 }
1194
1195 sub do_fetch {
1196 my ($self, $paths, $rev) = @_;
1197 my $ed;
1198 my ($last_rev, @parents);
1199 if ($self->{last_commit}) {
1200 $ed = SVN::Git::Fetcher->new($self);
1201 $last_rev = $self->{last_rev};
1202 $ed->{c} = $self->{last_commit};
1203 @parents = ($self->{last_commit});
1204 } else {
1205 $last_rev = $rev;
1206 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1207 return $log_entry;
1208 }
1209 $ed = SVN::Git::Fetcher->new($self);
1210 $ed->{new_fetch} = 1;
1211 }
1212 unless ($self->ra->gs_do_update($last_rev, $rev,
1213 $self->{path}, 1, $ed)) {
1214 die "SVN connection failed somewhere...\n";
1215 }
1216 $self->make_log_entry($rev, \@parents, $ed);
1217 }
1218
1219 sub get_untracked {
1220 my ($self, $ed) = @_;
1221 my @out;
1222 my $h = $ed->{empty};
1223 foreach (sort keys %$h) {
1224 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1225 push @out, " $act: " . uri_encode($_);
1226 warn "W: $act: $_\n";
1227 }
1228 foreach my $t (qw/dir_prop file_prop/) {
1229 $h = $ed->{$t} or next;
1230 foreach my $path (sort keys %$h) {
1231 my $ppath = $path eq '' ? '.' : $path;
1232 foreach my $prop (sort keys %{$h->{$path}}) {
1233 next if $SKIP_PROP{$prop};
1234 my $v = $h->{$path}->{$prop};
1235 my $t_ppath_prop = "$t: " .
1236 uri_encode($ppath) . ' ' .
1237 uri_encode($prop);
1238 if (defined $v) {
1239 push @out, " +$t_ppath_prop " .
1240 uri_encode($v);
1241 } else {
1242 push @out, " -$t_ppath_prop";
1243 }
1244 }
1245 }
1246 }
1247 foreach my $t (qw/absent_file absent_directory/) {
1248 $h = $ed->{$t} or next;
1249 foreach my $parent (sort keys %$h) {
1250 foreach my $path (sort @{$h->{$parent}}) {
1251 push @out, " $t: " .
1252 uri_encode("$parent/$path");
1253 warn "W: $t: $parent/$path ",
1254 "Insufficient permissions?\n";
1255 }
1256 }
1257 }
1258 \@out;
1259 }
1260
1261 sub parse_svn_date {
1262 my $date = shift || return '+0000 1970-01-01 00:00:00';
1263 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1264 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1265 croak "Unable to parse date: $date\n";
1266 "+0000 $Y-$m-$d $H:$M:$S";
1267 }
1268
1269 sub check_author {
1270 my ($author) = @_;
1271 if (!defined $author || length $author == 0) {
1272 $author = '(no author)';
1273 }
1274 if (defined $::_authors && ! defined $::users{$author}) {
1275 die "Author: $author not defined in $::_authors file\n";
1276 }
1277 $author;
1278 }
1279
1280 sub make_log_entry {
1281 my ($self, $rev, $parents, $ed) = @_;
1282 my $untracked = $self->get_untracked($ed);
1283
1284 return undef if (! $ed->{new_fetch} && ! $ed->{nr} && ! @$untracked);
1285
1286 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1287 print $un "r$rev\n" or croak $!;
1288 print $un $_, "\n" foreach @$untracked;
1289 my %log_entry = ( parents => $parents || [], revision => $rev,
1290 log => '');
1291 my $rp = $self->ra->rev_proplist($rev);
1292 foreach (sort keys %$rp) {
1293 my $v = $rp->{$_};
1294 if (/^svn:(author|date|log)$/) {
1295 $log_entry{$1} = $v;
1296 } else {
1297 print $un " rev_prop: ", uri_encode($_), ' ',
1298 uri_encode($v), "\n";
1299 }
1300 }
1301 close $un or croak $!;
1302
1303 $log_entry{date} = parse_svn_date($log_entry{date});
1304 $log_entry{author} = check_author($log_entry{author});
1305 $log_entry{log} .= "\n";
1306 \%log_entry;
1307 }
1308
1309 sub fetch {
1310 my ($self, $min_rev, $max_rev, @parents) = @_;
1311 my ($last_rev, $last_commit) = $self->last_rev_commit;
1312 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1313 return if ($base > $head);
1314 $self->ra->gs_fetch_loop_common($base, $head, $self);
1315 }
1316
1317 sub set_tree_cb {
1318 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1319 # TODO: enable and test optimized commits:
1320 if (0 && $rev == ($self->{last_rev} + 1)) {
1321 $log_entry->{revision} = $rev;
1322 $log_entry->{author} = $author;
1323 $self->do_git_commit($log_entry, "$rev=$tree");
1324 } else {
1325 $self->{inject_parents} = { $rev => $tree };
1326 $self->fetch(undef, undef);
1327 }
1328 }
1329
1330 sub set_tree {
1331 my ($self, $tree) = (shift, shift);
1332 my $log_entry = ::get_commit_entry($tree);
1333 unless ($self->{last_rev}) {
1334 fatal("Must have an existing revision to commit\n");
1335 }
1336 my %ed_opts = ( r => $self->{last_rev},
1337 log => $log_entry->{log},
1338 ra => $self->ra,
1339 tree_a => $self->{last_commit},
1340 tree_b => $tree,
1341 editor_cb => sub {
1342 $self->set_tree_cb($log_entry, $tree, @_) },
1343 svn_path => $self->{path} );
1344 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1345 print "No changes\nr$self->{last_rev} = $tree\n";
1346 }
1347 }
1348
1349 # rev_db:
1350 # Tie::File seems to be prone to offset errors if revisions get sparse,
1351 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
1352 # one of my favorite modules is out :< Next up would be one of the DBM
1353 # modules, but I'm not sure which is most portable... So I'll just
1354 # go with something that's plain-text, but still capable of
1355 # being randomly accessed. So here's my ultra-simple fixed-width
1356 # database. All records are 40 characters + "\n", so it's easy to seek
1357 # to a revision: (41 * rev) is the byte offset.
1358 # A record of 40 0s denotes an empty revision.
1359 # And yes, it's still pretty fast (faster than Tie::File).
1360
1361 sub rev_db_set {
1362 my ($self, $rev, $commit) = @_;
1363 length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
1364 open my $fh, '+<', $self->{db_path} or croak $!;
1365 my $offset = $rev * 41;
1366 # assume that append is the common case:
1367 seek $fh, 0, 2 or croak $!;
1368 my $pos = tell $fh;
1369 if ($pos < $offset) {
1370 print $fh (('0' x 40),"\n") x (($offset - $pos) / 41)
1371 or croak $!;
1372 }
1373 seek $fh, $offset, 0 or croak $!;
1374 print $fh $commit,"\n" or croak $!;
1375 close $fh or croak $!;
1376 }
1377
1378 sub rev_db_get {
1379 my ($self, $rev) = @_;
1380 my $ret;
1381 my $offset = $rev * 41;
1382 open my $fh, '<', $self->{db_path} or croak $!;
1383 if (seek $fh, $offset, 0) {
1384 $ret = readline $fh;
1385 if (defined $ret) {
1386 chomp $ret;
1387 $ret = undef if ($ret =~ /^0{40}$/);
1388 }
1389 }
1390 close $fh or croak $!;
1391 $ret;
1392 }
1393
1394 sub find_rev_before {
1395 my ($self, $rev, $eq_ok) = @_;
1396 --$rev unless $eq_ok;
1397 while ($rev > 0) {
1398 if (my $c = $self->rev_db_get($rev)) {
1399 return ($rev, $c);
1400 }
1401 --$rev;
1402 }
1403 return (undef, undef);
1404 }
1405
1406 sub _new {
1407 my ($class, $repo_id, $ref_id, $path) = @_;
1408 unless (defined $repo_id && length $repo_id) {
1409 $repo_id = $Git::SVN::default_repo_id;
1410 }
1411 unless (defined $ref_id && length $ref_id) {
1412 $_[2] = $ref_id = $Git::SVN::default_ref_id;
1413 }
1414 $_[1] = $repo_id = sanitize_remote_name($repo_id);
1415 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1416 $_[3] = $path = '' unless (defined $path);
1417 mkpath([$dir]);
1418 unless (-f "$dir/.rev_db") {
1419 open my $fh, '>>', "$dir/.rev_db" or croak $!;
1420 close $fh or croak $!;
1421 }
1422 bless { ref_id => $ref_id, dir => $dir, index => "$dir/index",
1423 path => $path,
1424 db_path => "$dir/.rev_db", repo_id => $repo_id }, $class;
1425 }
1426
1427 sub uri_encode {
1428 my ($f) = @_;
1429 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1430 $f
1431 }
1432
1433 package Git::SVN::Prompt;
1434 use strict;
1435 use warnings;
1436 require SVN::Core;
1437 use vars qw/$_no_auth_cache $_username/;
1438
1439 sub simple {
1440 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1441 $may_save = undef if $_no_auth_cache;
1442 $default_username = $_username if defined $_username;
1443 if (defined $default_username && length $default_username) {
1444 if (defined $realm && length $realm) {
1445 print STDERR "Authentication realm: $realm\n";
1446 STDERR->flush;
1447 }
1448 $cred->username($default_username);
1449 } else {
1450 username($cred, $realm, $may_save, $pool);
1451 }
1452 $cred->password(_read_password("Password for '" .
1453 $cred->username . "': ", $realm));
1454 $cred->may_save($may_save);
1455 $SVN::_Core::SVN_NO_ERROR;
1456 }
1457
1458 sub ssl_server_trust {
1459 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1460 $may_save = undef if $_no_auth_cache;
1461 print STDERR "Error validating server certificate for '$realm':\n";
1462 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1463 print STDERR " - The certificate is not issued by a trusted ",
1464 "authority. Use the\n",
1465 " fingerprint to validate the certificate manually!\n";
1466 }
1467 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1468 print STDERR " - The certificate hostname does not match.\n";
1469 }
1470 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1471 print STDERR " - The certificate is not yet valid.\n";
1472 }
1473 if ($failures & $SVN::Auth::SSL::EXPIRED) {
1474 print STDERR " - The certificate has expired.\n";
1475 }
1476 if ($failures & $SVN::Auth::SSL::OTHER) {
1477 print STDERR " - The certificate has an unknown error.\n";
1478 }
1479 printf STDERR
1480 "Certificate information:\n".
1481 " - Hostname: %s\n".
1482 " - Valid: from %s until %s\n".
1483 " - Issuer: %s\n".
1484 " - Fingerprint: %s\n",
1485 map $cert_info->$_, qw(hostname valid_from valid_until
1486 issuer_dname fingerprint);
1487 my $choice;
1488 prompt:
1489 print STDERR $may_save ?
1490 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1491 "(R)eject or accept (t)emporarily? ";
1492 STDERR->flush;
1493 $choice = lc(substr(<STDIN> || 'R', 0, 1));
1494 if ($choice =~ /^t$/i) {
1495 $cred->may_save(undef);
1496 } elsif ($choice =~ /^r$/i) {
1497 return -1;
1498 } elsif ($may_save && $choice =~ /^p$/i) {
1499 $cred->may_save($may_save);
1500 } else {
1501 goto prompt;
1502 }
1503 $cred->accepted_failures($failures);
1504 $SVN::_Core::SVN_NO_ERROR;
1505 }
1506
1507 sub ssl_client_cert {
1508 my ($cred, $realm, $may_save, $pool) = @_;
1509 $may_save = undef if $_no_auth_cache;
1510 print STDERR "Client certificate filename: ";
1511 STDERR->flush;
1512 chomp(my $filename = <STDIN>);
1513 $cred->cert_file($filename);
1514 $cred->may_save($may_save);
1515 $SVN::_Core::SVN_NO_ERROR;
1516 }
1517
1518 sub ssl_client_cert_pw {
1519 my ($cred, $realm, $may_save, $pool) = @_;
1520 $may_save = undef if $_no_auth_cache;
1521 $cred->password(_read_password("Password: ", $realm));
1522 $cred->may_save($may_save);
1523 $SVN::_Core::SVN_NO_ERROR;
1524 }
1525
1526 sub username {
1527 my ($cred, $realm, $may_save, $pool) = @_;
1528 $may_save = undef if $_no_auth_cache;
1529 if (defined $realm && length $realm) {
1530 print STDERR "Authentication realm: $realm\n";
1531 }
1532 my $username;
1533 if (defined $_username) {
1534 $username = $_username;
1535 } else {
1536 print STDERR "Username: ";
1537 STDERR->flush;
1538 chomp($username = <STDIN>);
1539 }
1540 $cred->username($username);
1541 $cred->may_save($may_save);
1542 $SVN::_Core::SVN_NO_ERROR;
1543 }
1544
1545 sub _read_password {
1546 my ($prompt, $realm) = @_;
1547 print STDERR $prompt;
1548 STDERR->flush;
1549 require Term::ReadKey;
1550 Term::ReadKey::ReadMode('noecho');
1551 my $password = '';
1552 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1553 last if $key =~ /[\012\015]/; # \n\r
1554 $password .= $key;
1555 }
1556 Term::ReadKey::ReadMode('restore');
1557 print STDERR "\n";
1558 STDERR->flush;
1559 $password;
1560 }
1561
1562 package main;
1563
1564 {
1565 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1566 $SVN::Node::dir.$SVN::Node::unknown.
1567 $SVN::Node::none.$SVN::Node::file.
1568 $SVN::Node::dir.$SVN::Node::unknown.
1569 $SVN::Auth::SSL::CNMISMATCH.
1570 $SVN::Auth::SSL::NOTYETVALID.
1571 $SVN::Auth::SSL::EXPIRED.
1572 $SVN::Auth::SSL::UNKNOWNCA.
1573 $SVN::Auth::SSL::OTHER;
1574 }
1575
1576 package SVN::Git::Fetcher;
1577 use vars qw/@ISA/;
1578 use strict;
1579 use warnings;
1580 use Carp qw/croak/;
1581 use IO::File qw//;
1582 use Digest::MD5;
1583
1584 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
1585 sub new {
1586 my ($class, $git_svn) = @_;
1587 my $self = SVN::Delta::Editor->new;
1588 bless $self, $class;
1589 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
1590 $self->{empty} = {};
1591 $self->{dir_prop} = {};
1592 $self->{file_prop} = {};
1593 $self->{absent_dir} = {};
1594 $self->{absent_file} = {};
1595 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
1596 $self;
1597 }
1598
1599 sub set_path_strip {
1600 my ($self, $path) = @_;
1601 $self->{path_strip} = qr/^\Q$path\E\/?/;
1602 }
1603
1604 sub open_root {
1605 { path => '' };
1606 }
1607
1608 sub open_directory {
1609 my ($self, $path, $pb, $rev) = @_;
1610 { path => $path };
1611 }
1612
1613 sub git_path {
1614 my ($self, $path) = @_;
1615 if ($self->{path_strip}) {
1616 $path =~ s!$self->{path_strip}!! or
1617 die "Failed to strip path '$path' ($self->{path_strip})\n";
1618 }
1619 $path;
1620 }
1621
1622 sub delete_entry {
1623 my ($self, $path, $rev, $pb) = @_;
1624
1625 my $gpath = $self->git_path($path);
1626 # remove entire directories.
1627 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
1628 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
1629 -r --name-only -z/,
1630 $self->{c}, '--', $gpath);
1631 local $/ = "\0";
1632 while (<$ls>) {
1633 chomp;
1634 $self->{gii}->remove($_);
1635 print "\tD\t$_\n" unless $self->{q};
1636 }
1637 print "\tD\t$gpath/\n" unless $self->{q};
1638 command_close_pipe($ls, $ctx);
1639 $self->{empty}->{$path} = 0
1640 } else {
1641 $self->{gii}->remove($gpath);
1642 print "\tD\t$gpath\n" unless $self->{q};
1643 }
1644 undef;
1645 }
1646
1647 sub open_file {
1648 my ($self, $path, $pb, $rev) = @_;
1649 my $gpath = $self->git_path($path);
1650 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
1651 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
1652 unless (defined $mode && defined $blob) {
1653 die "$path was not found in commit $self->{c} (r$rev)\n";
1654 }
1655 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
1656 pool => SVN::Pool->new, action => 'M' };
1657 }
1658
1659 sub add_file {
1660 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
1661 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1662 delete $self->{empty}->{$dir};
1663 { path => $path, mode_a => 100644, mode_b => 100644,
1664 pool => SVN::Pool->new, action => 'A' };
1665 }
1666
1667 sub add_directory {
1668 my ($self, $path, $cp_path, $cp_rev) = @_;
1669 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1670 delete $self->{empty}->{$dir};
1671 $self->{empty}->{$path} = 1;
1672 { path => $path };
1673 }
1674
1675 sub change_dir_prop {
1676 my ($self, $db, $prop, $value) = @_;
1677 $self->{dir_prop}->{$db->{path}} ||= {};
1678 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
1679 undef;
1680 }
1681
1682 sub absent_directory {
1683 my ($self, $path, $pb) = @_;
1684 $self->{absent_dir}->{$pb->{path}} ||= [];
1685 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
1686 undef;
1687 }
1688
1689 sub absent_file {
1690 my ($self, $path, $pb) = @_;
1691 $self->{absent_file}->{$pb->{path}} ||= [];
1692 push @{$self->{absent_file}->{$pb->{path}}}, $path;
1693 undef;
1694 }
1695
1696 sub change_file_prop {
1697 my ($self, $fb, $prop, $value) = @_;
1698 if ($prop eq 'svn:executable') {
1699 if ($fb->{mode_b} != 120000) {
1700 $fb->{mode_b} = defined $value ? 100755 : 100644;
1701 }
1702 } elsif ($prop eq 'svn:special') {
1703 $fb->{mode_b} = defined $value ? 120000 : 100644;
1704 } else {
1705 $self->{file_prop}->{$fb->{path}} ||= {};
1706 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
1707 }
1708 undef;
1709 }
1710
1711 sub apply_textdelta {
1712 my ($self, $fb, $exp) = @_;
1713 my $fh = IO::File->new_tmpfile;
1714 $fh->autoflush(1);
1715 # $fh gets auto-closed() by SVN::TxDelta::apply(),
1716 # (but $base does not,) so dup() it for reading in close_file
1717 open my $dup, '<&', $fh or croak $!;
1718 my $base = IO::File->new_tmpfile;
1719 $base->autoflush(1);
1720 if ($fb->{blob}) {
1721 defined (my $pid = fork) or croak $!;
1722 if (!$pid) {
1723 open STDOUT, '>&', $base or croak $!;
1724 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
1725 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
1726 }
1727 waitpid $pid, 0;
1728 croak $? if $?;
1729
1730 if (defined $exp) {
1731 seek $base, 0, 0 or croak $!;
1732 my $md5 = Digest::MD5->new;
1733 $md5->addfile($base);
1734 my $got = $md5->hexdigest;
1735 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
1736 "expected: $exp\n",
1737 " got: $got\n" if ($got ne $exp);
1738 }
1739 }
1740 seek $base, 0, 0 or croak $!;
1741 $fb->{fh} = $dup;
1742 $fb->{base} = $base;
1743 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
1744 }
1745
1746 sub close_file {
1747 my ($self, $fb, $exp) = @_;
1748 my $hash;
1749 my $path = $self->git_path($fb->{path});
1750 if (my $fh = $fb->{fh}) {
1751 seek($fh, 0, 0) or croak $!;
1752 my $md5 = Digest::MD5->new;
1753 $md5->addfile($fh);
1754 my $got = $md5->hexdigest;
1755 die "Checksum mismatch: $path\n",
1756 "expected: $exp\n got: $got\n" if ($got ne $exp);
1757 seek($fh, 0, 0) or croak $!;
1758 if ($fb->{mode_b} == 120000) {
1759 read($fh, my $buf, 5) == 5 or croak $!;
1760 $buf eq 'link ' or die "$path has mode 120000",
1761 "but is not a link\n";
1762 }
1763 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
1764 if (!$pid) {
1765 open STDIN, '<&', $fh or croak $!;
1766 exec qw/git-hash-object -w --stdin/ or croak $!;
1767 }
1768 chomp($hash = do { local $/; <$out> });
1769 close $out or croak $!;
1770 close $fh or croak $!;
1771 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
1772 close $fb->{base} or croak $!;
1773 } else {
1774 $hash = $fb->{blob} or die "no blob information\n";
1775 }
1776 $fb->{pool}->clear;
1777 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
1778 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
1779 undef;
1780 }
1781
1782 sub abort_edit {
1783 my $self = shift;
1784 $self->{nr} = $self->{gii}->{nr};
1785 delete $self->{gii};
1786 $self->SUPER::abort_edit(@_);
1787 }
1788
1789 sub close_edit {
1790 my $self = shift;
1791 $self->{git_commit_ok} = 1;
1792 $self->{nr} = $self->{gii}->{nr};
1793 delete $self->{gii};
1794 $self->SUPER::close_edit(@_);
1795 }
1796
1797 package SVN::Git::Editor;
1798 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
1799 use strict;
1800 use warnings;
1801 use Carp qw/croak/;
1802 use IO::File;
1803 use Digest::MD5;
1804
1805 sub new {
1806 my ($class, $opts) = @_;
1807 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
1808 die "$_ required!\n" unless (defined $opts->{$_});
1809 }
1810
1811 my $pool = SVN::Pool->new;
1812 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
1813 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
1814 $opts->{r}, $mods);
1815
1816 # $opts->{ra} functions should not be used after this:
1817 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
1818 $opts->{editor_cb}, $pool);
1819 my $self = SVN::Delta::Editor->new(@ce, $pool);
1820 bless $self, $class;
1821 foreach (qw/svn_path r tree_a tree_b/) {
1822 $self->{$_} = $opts->{$_};
1823 }
1824 $self->{url} = $opts->{ra}->{url};
1825 $self->{mods} = $mods;
1826 $self->{types} = $types;
1827 $self->{pool} = $pool;
1828 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
1829 $self->{rm} = { };
1830 $self->{path_prefix} = length $self->{svn_path} ?
1831 "$self->{svn_path}/" : '';
1832 return $self;
1833 }
1834
1835 sub generate_diff {
1836 my ($tree_a, $tree_b) = @_;
1837 my @diff_tree = qw(diff-tree -z -r);
1838 if ($_cp_similarity) {
1839 push @diff_tree, "-C$_cp_similarity";
1840 } else {
1841 push @diff_tree, '-C';
1842 }
1843 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1844 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
1845 push @diff_tree, $tree_a, $tree_b;
1846 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
1847 local $/ = "\0";
1848 my $state = 'meta';
1849 my @mods;
1850 while (<$diff_fh>) {
1851 chomp $_; # this gets rid of the trailing "\0"
1852 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1853 $::sha1\s($::sha1)\s
1854 ([MTCRAD])\d*$/xo) {
1855 push @mods, { mode_a => $1, mode_b => $2,
1856 sha1_b => $3, chg => $4 };
1857 if ($4 =~ /^(?:C|R)$/) {
1858 $state = 'file_a';
1859 } else {
1860 $state = 'file_b';
1861 }
1862 } elsif ($state eq 'file_a') {
1863 my $x = $mods[$#mods] or croak "Empty array\n";
1864 if ($x->{chg} !~ /^(?:C|R)$/) {
1865 croak "Error parsing $_, $x->{chg}\n";
1866 }
1867 $x->{file_a} = $_;
1868 $state = 'file_b';
1869 } elsif ($state eq 'file_b') {
1870 my $x = $mods[$#mods] or croak "Empty array\n";
1871 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1872 croak "Error parsing $_, $x->{chg}\n";
1873 }
1874 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1875 croak "Error parsing $_, $x->{chg}\n";
1876 }
1877 $x->{file_b} = $_;
1878 $state = 'meta';
1879 } else {
1880 croak "Error parsing $_\n";
1881 }
1882 }
1883 command_close_pipe($diff_fh, $ctx);
1884 \@mods;
1885 }
1886
1887 sub check_diff_paths {
1888 my ($ra, $pfx, $rev, $mods) = @_;
1889 my %types;
1890 $pfx .= '/' if length $pfx;
1891
1892 sub type_diff_paths {
1893 my ($ra, $types, $path, $rev) = @_;
1894 my @p = split m#/+#, $path;
1895 my $c = shift @p;
1896 unless (defined $types->{$c}) {
1897 $types->{$c} = $ra->check_path($c, $rev);
1898 }
1899 while (@p) {
1900 $c .= '/' . shift @p;
1901 next if defined $types->{$c};
1902 $types->{$c} = $ra->check_path($c, $rev);
1903 }
1904 }
1905
1906 foreach my $m (@$mods) {
1907 foreach my $f (qw/file_a file_b/) {
1908 next unless defined $m->{$f};
1909 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
1910 if (length $pfx.$dir && ! defined $types{$dir}) {
1911 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
1912 }
1913 }
1914 }
1915 \%types;
1916 }
1917
1918 sub split_path {
1919 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
1920 }
1921
1922 sub repo_path {
1923 my ($self, $path) = @_;
1924 $self->{path_prefix}.(defined $path ? $path : '');
1925 }
1926
1927 sub url_path {
1928 my ($self, $path) = @_;
1929 $self->{url} . '/' . $self->repo_path($path);
1930 }
1931
1932 sub rmdirs {
1933 my ($self) = @_;
1934 my $rm = $self->{rm};
1935 delete $rm->{''}; # we never delete the url we're tracking
1936 return unless %$rm;
1937
1938 foreach (keys %$rm) {
1939 my @d = split m#/#, $_;
1940 my $c = shift @d;
1941 $rm->{$c} = 1;
1942 while (@d) {
1943 $c .= '/' . shift @d;
1944 $rm->{$c} = 1;
1945 }
1946 }
1947 delete $rm->{$self->{svn_path}};
1948 delete $rm->{''}; # we never delete the url we're tracking
1949 return unless %$rm;
1950
1951 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
1952 $self->{tree_b});
1953 local $/ = "\0";
1954 while (<$fh>) {
1955 chomp;
1956 my @dn = split m#/#, $_;
1957 while (pop @dn) {
1958 delete $rm->{join '/', @dn};
1959 }
1960 unless (%$rm) {
1961 close $fh;
1962 return;
1963 }
1964 }
1965 command_close_pipe($fh, $ctx);
1966
1967 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
1968 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
1969 $self->close_directory($bat->{$d}, $p);
1970 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
1971 print "\tD+\t$d/\n" unless $::_q;
1972 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
1973 delete $bat->{$d};
1974 }
1975 }
1976
1977 sub open_or_add_dir {
1978 my ($self, $full_path, $baton) = @_;
1979 my $t = $self->{types}->{$full_path};
1980 if (!defined $t) {
1981 die "$full_path not known in r$self->{r} or we have a bug!\n";
1982 }
1983 if ($t == $SVN::Node::none) {
1984 return $self->add_directory($full_path, $baton,
1985 undef, -1, $self->{pool});
1986 } elsif ($t == $SVN::Node::dir) {
1987 return $self->open_directory($full_path, $baton,
1988 $self->{r}, $self->{pool});
1989 }
1990 print STDERR "$full_path already exists in repository at ",
1991 "r$self->{r} and it is not a directory (",
1992 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
1993 exit 1;
1994 }
1995
1996 sub ensure_path {
1997 my ($self, $path) = @_;
1998 my $bat = $self->{bat};
1999 my $repo_path = $self->repo_path($path);
2000 return $bat->{''} unless (length $repo_path);
2001 my @p = split m#/+#, $repo_path;
2002 my $c = shift @p;
2003 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2004 while (@p) {
2005 my $c0 = $c;
2006 $c .= '/' . shift @p;
2007 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2008 }
2009 return $bat->{$c};
2010 }
2011
2012 sub A {
2013 my ($self, $m) = @_;
2014 my ($dir, $file) = split_path($m->{file_b});
2015 my $pbat = $self->ensure_path($dir);
2016 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2017 undef, -1);
2018 print "\tA\t$m->{file_b}\n" unless $::_q;
2019 $self->chg_file($fbat, $m);
2020 $self->close_file($fbat,undef,$self->{pool});
2021 }
2022
2023 sub C {
2024 my ($self, $m) = @_;
2025 my ($dir, $file) = split_path($m->{file_b});
2026 my $pbat = $self->ensure_path($dir);
2027 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2028 $self->url_path($m->{file_a}), $self->{r});
2029 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2030 $self->chg_file($fbat, $m);
2031 $self->close_file($fbat,undef,$self->{pool});
2032 }
2033
2034 sub delete_entry {
2035 my ($self, $path, $pbat) = @_;
2036 my $rpath = $self->repo_path($path);
2037 my ($dir, $file) = split_path($rpath);
2038 $self->{rm}->{$dir} = 1;
2039 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2040 }
2041
2042 sub R {
2043 my ($self, $m) = @_;
2044 my ($dir, $file) = split_path($m->{file_b});
2045 my $pbat = $self->ensure_path($dir);
2046 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2047 $self->url_path($m->{file_a}), $self->{r});
2048 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2049 $self->chg_file($fbat, $m);
2050 $self->close_file($fbat,undef,$self->{pool});
2051
2052 ($dir, $file) = split_path($m->{file_a});
2053 $pbat = $self->ensure_path($dir);
2054 $self->delete_entry($m->{file_a}, $pbat);
2055 }
2056
2057 sub M {
2058 my ($self, $m) = @_;
2059 my ($dir, $file) = split_path($m->{file_b});
2060 my $pbat = $self->ensure_path($dir);
2061 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2062 $pbat,$self->{r},$self->{pool});
2063 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2064 $self->chg_file($fbat, $m);
2065 $self->close_file($fbat,undef,$self->{pool});
2066 }
2067
2068 sub T { shift->M(@_) }
2069
2070 sub change_file_prop {
2071 my ($self, $fbat, $pname, $pval) = @_;
2072 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2073 }
2074
2075 sub chg_file {
2076 my ($self, $fbat, $m) = @_;
2077 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2078 $self->change_file_prop($fbat,'svn:executable','*');
2079 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2080 $self->change_file_prop($fbat,'svn:executable',undef);
2081 }
2082 my $fh = IO::File->new_tmpfile or croak $!;
2083 if ($m->{mode_b} =~ /^120/) {
2084 print $fh 'link ' or croak $!;
2085 $self->change_file_prop($fbat,'svn:special','*');
2086 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2087 $self->change_file_prop($fbat,'svn:special',undef);
2088 }
2089 defined(my $pid = fork) or croak $!;
2090 if (!$pid) {
2091 open STDOUT, '>&', $fh or croak $!;
2092 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2093 }
2094 waitpid $pid, 0;
2095 croak $? if $?;
2096 $fh->flush == 0 or croak $!;
2097 seek $fh, 0, 0 or croak $!;
2098
2099 my $md5 = Digest::MD5->new;
2100 $md5->addfile($fh) or croak $!;
2101 seek $fh, 0, 0 or croak $!;
2102
2103 my $exp = $md5->hexdigest;
2104 my $pool = SVN::Pool->new;
2105 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2106 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2107 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2108 $pool->clear;
2109
2110 close $fh or croak $!;
2111 }
2112
2113 sub D {
2114 my ($self, $m) = @_;
2115 my ($dir, $file) = split_path($m->{file_b});
2116 my $pbat = $self->ensure_path($dir);
2117 print "\tD\t$m->{file_b}\n" unless $::_q;
2118 $self->delete_entry($m->{file_b}, $pbat);
2119 }
2120
2121 sub close_edit {
2122 my ($self) = @_;
2123 my ($p,$bat) = ($self->{pool}, $self->{bat});
2124 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2125 $self->close_directory($bat->{$_}, $p);
2126 }
2127 $self->SUPER::close_edit($p);
2128 $p->clear;
2129 }
2130
2131 sub abort_edit {
2132 my ($self) = @_;
2133 $self->SUPER::abort_edit($self->{pool});
2134 }
2135
2136 sub DESTROY {
2137 my $self = shift;
2138 $self->SUPER::DESTROY(@_);
2139 $self->{pool}->clear;
2140 }
2141
2142 # this drives the editor
2143 sub apply_diff {
2144 my ($self) = @_;
2145 my $mods = $self->{mods};
2146 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2147 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2148 my $f = $m->{chg};
2149 if (defined $o{$f}) {
2150 $self->$f($m);
2151 } else {
2152 fatal("Invalid change type: $f\n");
2153 }
2154 }
2155 $self->rmdirs if $_rmdir;
2156 if (@$mods == 0) {
2157 $self->abort_edit;
2158 } else {
2159 $self->close_edit;
2160 }
2161 return scalar @$mods;
2162 }
2163
2164 package Git::SVN::Ra;
2165 use vars qw/@ISA $config_dir/;
2166 use strict;
2167 use warnings;
2168 my ($can_do_switch);
2169 my $RA;
2170
2171 BEGIN {
2172 # enforce temporary pool usage for some simple functions
2173 my $e;
2174 foreach (qw/get_latest_revnum rev_proplist get_file
2175 check_path get_dir get_uuid get_repos_root/) {
2176 $e .= "sub $_ {
2177 my \$self = shift;
2178 my \$pool = SVN::Pool->new;
2179 my \@ret = \$self->SUPER::$_(\@_,\$pool);
2180 \$pool->clear;
2181 wantarray ? \@ret : \$ret[0]; }\n";
2182 }
2183 eval $e;
2184 }
2185
2186 sub new {
2187 my ($class, $url) = @_;
2188 $url =~ s!/+$!!;
2189 return $RA if ($RA && $RA->{url} eq $url);
2190
2191 SVN::_Core::svn_config_ensure($config_dir, undef);
2192 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2193 SVN::Client::get_simple_provider(),
2194 SVN::Client::get_ssl_server_trust_file_provider(),
2195 SVN::Client::get_simple_prompt_provider(
2196 \&Git::SVN::Prompt::simple, 2),
2197 SVN::Client::get_ssl_client_cert_prompt_provider(
2198 \&Git::SVN::Prompt::ssl_client_cert, 2),
2199 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2200 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2201 SVN::Client::get_username_provider(),
2202 SVN::Client::get_ssl_server_trust_prompt_provider(
2203 \&Git::SVN::Prompt::ssl_server_trust),
2204 SVN::Client::get_username_prompt_provider(
2205 \&Git::SVN::Prompt::username, 2),
2206 ]);
2207 my $config = SVN::Core::config_get_config($config_dir);
2208 my $self = SVN::Ra->new(url => $url, auth => $baton,
2209 config => $config,
2210 pool => SVN::Pool->new,
2211 auth_provider_callbacks => $callbacks);
2212 $self->{svn_path} = $url;
2213 $self->{repos_root} = $self->get_repos_root;
2214 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E/*##;
2215 $RA = bless $self, $class;
2216 }
2217
2218 sub DESTROY {
2219 # do not call the real DESTROY since we store ourselves in $RA
2220 }
2221
2222 sub get_log {
2223 my ($self, @args) = @_;
2224 my $pool = SVN::Pool->new;
2225 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2226 my $ret = $self->SUPER::get_log(@args, $pool);
2227 $pool->clear;
2228 $ret;
2229 }
2230
2231 sub get_commit_editor {
2232 my ($self, $log, $cb, $pool) = @_;
2233 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2234 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2235 }
2236
2237 sub uuid {
2238 my ($self) = @_;
2239 $self->{uuid} ||= $self->get_uuid;
2240 }
2241
2242 sub gs_do_update {
2243 my ($self, $rev_a, $rev_b, $path, $recurse, $editor) = @_;
2244 my $pool = SVN::Pool->new;
2245 $editor->set_path_strip($path);
2246 my (@pc) = split m#/#, $path;
2247 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2248 $recurse, $editor, $pool);
2249 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2250
2251 # Since we can't rely on svn_ra_reparent being available, we'll
2252 # just have to do some magic with set_path to make it so
2253 # we only want a partial path.
2254 my $sp = '';
2255 my $final = join('/', @pc);
2256 while (@pc) {
2257 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2258 $sp .= '/' if length $sp;
2259 $sp .= shift @pc;
2260 }
2261 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2262
2263 my $new = ($rev_a == $rev_b);
2264 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2265
2266 $reporter->finish_report($pool);
2267 $pool->clear;
2268 $editor->{git_commit_ok};
2269 }
2270
2271 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2272 # svn_ra_reparent didn't work before 1.4)
2273 sub gs_do_switch {
2274 my ($self, $rev_a, $rev_b, $path, $recurse, $url_b, $editor) = @_;
2275 my $pool = SVN::Pool->new;
2276
2277 my $full_url = $self->{url};
2278 my $old_url = $full_url;
2279 $full_url .= "/$path" if length $path;
2280 my ($ra, $reparented);
2281 if ($old_url ne $full_url) {
2282 if ($old_url !~ m#^svn(\+ssh)?://#) {
2283 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2284 $pool);
2285 $self->{url} = $full_url;
2286 $reparented = 1;
2287 } else {
2288 $ra = Git::SVN::Ra->new($full_url);
2289 }
2290 }
2291 $ra ||= $self;
2292 my $reporter = $ra->do_switch($rev_b, '',
2293 $recurse, $url_b, $editor, $pool);
2294 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2295 $reporter->set_path('', $rev_a, 0, @lock, $pool);
2296 $reporter->finish_report($pool);
2297
2298 if ($reparented) {
2299 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2300 $self->{url} = $old_url;
2301 }
2302
2303 $pool->clear;
2304 $editor->{git_commit_ok};
2305 }
2306
2307 sub gs_fetch_loop_common {
2308 my ($self, $base, $head, @gs) = @_;
2309 my $inc = 1000;
2310 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2311 my @paths = @gs == 1 ? ($gs[0]->{path}) : ('');
2312 foreach my $gs (@gs) {
2313 if (my $last_commit = $gs->last_commit) {
2314 $gs->assert_index_clean($last_commit);
2315 }
2316 $gs->{path_regex} = qr/^\/\Q$gs->{path}\E\/?/;
2317 }
2318 while (1) {
2319 my @revs;
2320 my $err;
2321 my $err_handler = $SVN::Error::handler;
2322 $SVN::Error::handler = sub {
2323 ($err) = @_;
2324 skip_unknown_revs($err);
2325 };
2326 $self->get_log(\@paths, $min, $max, 0, 1, 1,
2327 sub { push @revs, [ dup_changed_paths($_[0]), $_[1] ]; });
2328 $SVN::Error::handler = $err_handler;
2329
2330 if (! @revs && $err && $max >= $head) {
2331 print STDERR "Branch probably deleted:\n ",
2332 $err->expanded_message,
2333 "\nWill attempt to follow revisions ",
2334 "r$min .. r$max ",
2335 "committed before the deletion\n";
2336 @revs = map { [ undef, $_ ] } ($min .. $max);
2337 }
2338 foreach (@revs) {
2339 my ($paths, $r) = @$_;
2340 foreach my $gs (@gs) {
2341 if ($paths) {
2342 grep /$gs->{path_regex}/, keys %$paths
2343 or next;
2344 }
2345 next if defined $gs->rev_db_get($r);
2346 if (my $log_entry = $gs->do_fetch($paths, $r)) {
2347 $gs->do_git_commit($log_entry);
2348 }
2349 }
2350 }
2351 last if $max >= $head;
2352 $min = $max + 1;
2353 $max += $inc;
2354 $max = $head if ($max > $head);
2355 }
2356 }
2357
2358 sub minimize_url {
2359 my ($self) = @_;
2360 return $self->{url} if ($self->{url} eq $self->{repos_root});
2361 my $url = $self->{repos_root};
2362 my @components = split(m!/!, $self->{svn_path});
2363 my $c = '';
2364 do {
2365 $url .= "/$c" if length $c;
2366 eval { (ref $self)->new($url)->get_latest_revnum };
2367 } while ($@ && ($c = shift @components));
2368 $url;
2369 }
2370
2371 sub can_do_switch {
2372 my $self = shift;
2373 unless (defined $can_do_switch) {
2374 my $pool = SVN::Pool->new;
2375 my $rep = eval {
2376 $self->do_switch(1, '', 0, $self->{url},
2377 SVN::Delta::Editor->new, $pool);
2378 };
2379 if ($@) {
2380 $can_do_switch = 0;
2381 } else {
2382 $rep->abort_report($pool);
2383 $can_do_switch = 1;
2384 }
2385 $pool->clear;
2386 }
2387 $can_do_switch;
2388 }
2389
2390 sub skip_unknown_revs {
2391 my ($err) = @_;
2392 my $errno = $err->apr_err();
2393 # Maybe the branch we're tracking didn't
2394 # exist when the repo started, so it's
2395 # not an error if it doesn't, just continue
2396 #
2397 # Wonderfully consistent library, eh?
2398 # 160013 - svn:// and file://
2399 # 175002 - http(s)://
2400 # 175007 - http(s):// (this repo required authorization, too...)
2401 # More codes may be discovered later...
2402 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2403 return;
2404 }
2405 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2406 }
2407
2408 # svn_log_changed_path_t objects passed to get_log are likely to be
2409 # overwritten even if only the refs are copied to an external variable,
2410 # so we should dup the structures in their entirety. Using an externally
2411 # passed pool (instead of our temporary and quickly cleared pool in
2412 # Git::SVN::Ra) does not help matters at all...
2413 sub dup_changed_paths {
2414 my ($paths) = @_;
2415 return undef unless $paths;
2416 my %ret;
2417 foreach my $p (keys %$paths) {
2418 my $i = $paths->{$p};
2419 my %s = map { $_ => $i->$_ }
2420 qw/copyfrom_path copyfrom_rev action/;
2421 $ret{$p} = \%s;
2422 }
2423 \%ret;
2424 }
2425
2426 package Git::SVN::Log;
2427 use strict;
2428 use warnings;
2429 use POSIX qw/strftime/;
2430 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2431 %rusers $show_commit $incremental/;
2432 my $l_fmt;
2433
2434 sub cmt_showable {
2435 my ($c) = @_;
2436 return 1 if defined $c->{r};
2437 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2438 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2439 my @log = command(qw/cat-file commit/, $c->{c});
2440 shift @log while ($log[0] ne "\n");
2441 shift @log;
2442 @{$c->{l}} = grep !/^git-svn-id: /, @log;
2443
2444 (undef, $c->{r}, undef) = ::extract_metadata(
2445 (grep(/^git-svn-id: /, @log))[-1]);
2446 }
2447 return defined $c->{r};
2448 }
2449
2450 sub log_use_color {
2451 return 1 if $color;
2452 my ($dc, $dcvar);
2453 $dcvar = 'color.diff';
2454 $dc = `git-config --get $dcvar`;
2455 if ($dc eq '') {
2456 # nothing at all; fallback to "diff.color"
2457 $dcvar = 'diff.color';
2458 $dc = `git-config --get $dcvar`;
2459 }
2460 chomp($dc);
2461 if ($dc eq 'auto') {
2462 my $pc;
2463 $pc = `git-config --get color.pager`;
2464 if ($pc eq '') {
2465 # does not have it -- fallback to pager.color
2466 $pc = `git-config --bool --get pager.color`;
2467 }
2468 else {
2469 $pc = `git-config --bool --get color.pager`;
2470 if ($?) {
2471 $pc = 'false';
2472 }
2473 }
2474 chomp($pc);
2475 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2476 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2477 }
2478 return 0;
2479 }
2480 return 0 if $dc eq 'never';
2481 return 1 if $dc eq 'always';
2482 chomp($dc = `git-config --bool --get $dcvar`);
2483 return ($dc eq 'true');
2484 }
2485
2486 sub git_svn_log_cmd {
2487 my ($r_min, $r_max) = @_;
2488 my $gs = Git::SVN->_new;
2489 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2490 $gs->refname);
2491 push @cmd, '-r' unless $non_recursive;
2492 push @cmd, qw/--raw --name-status/ if $verbose;
2493 push @cmd, '--color' if log_use_color();
2494 return @cmd unless defined $r_max;
2495 if ($r_max == $r_min) {
2496 push @cmd, '--max-count=1';
2497 if (my $c = $gs->rev_db_get($r_max)) {
2498 push @cmd, $c;
2499 }
2500 } else {
2501 my ($c_min, $c_max);
2502 $c_max = $gs->rev_db_get($r_max);
2503 $c_min = $gs->rev_db_get($r_min);
2504 if (defined $c_min && defined $c_max) {
2505 if ($r_max > $r_max) {
2506 push @cmd, "$c_min..$c_max";
2507 } else {
2508 push @cmd, "$c_max..$c_min";
2509 }
2510 } elsif ($r_max > $r_min) {
2511 push @cmd, $c_max;
2512 } else {
2513 push @cmd, $c_min;
2514 }
2515 }
2516 return @cmd;
2517 }
2518
2519 # adapted from pager.c
2520 sub config_pager {
2521 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2522 if (!defined $pager) {
2523 $pager = 'less';
2524 } elsif (length $pager == 0 || $pager eq 'cat') {
2525 $pager = undef;
2526 }
2527 }
2528
2529 sub run_pager {
2530 return unless -t *STDOUT;
2531 pipe my $rfd, my $wfd or return;
2532 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2533 if (!$pid) {
2534 open STDOUT, '>&', $wfd or
2535 ::fatal "Can't redirect to stdout: $!\n";
2536 return;
2537 }
2538 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2539 $ENV{LESS} ||= 'FRSX';
2540 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2541 }
2542
2543 sub tz_to_s_offset {
2544 my ($tz) = @_;
2545 $tz =~ s/(\d\d)$//;
2546 return ($1 * 60) + ($tz * 3600);
2547 }
2548
2549 sub get_author_info {
2550 my ($dest, $author, $t, $tz) = @_;
2551 $author =~ s/(?:^\s*|\s*$)//g;
2552 $dest->{a_raw} = $author;
2553 my $au;
2554 if ($::_authors) {
2555 $au = $rusers{$author} || undef;
2556 }
2557 if (!$au) {
2558 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2559 }
2560 $dest->{t} = $t;
2561 $dest->{tz} = $tz;
2562 $dest->{a} = $au;
2563 # Date::Parse isn't in the standard Perl distro :(
2564 if ($tz =~ s/^\+//) {
2565 $t += tz_to_s_offset($tz);
2566 } elsif ($tz =~ s/^\-//) {
2567 $t -= tz_to_s_offset($tz);
2568 }
2569 $dest->{t_utc} = $t;
2570 }
2571
2572 sub process_commit {
2573 my ($c, $r_min, $r_max, $defer) = @_;
2574 if (defined $r_min && defined $r_max) {
2575 if ($r_min == $c->{r} && $r_min == $r_max) {
2576 show_commit($c);
2577 return 0;
2578 }
2579 return 1 if $r_min == $r_max;
2580 if ($r_min < $r_max) {
2581 # we need to reverse the print order
2582 return 0 if (defined $limit && --$limit < 0);
2583 push @$defer, $c;
2584 return 1;
2585 }
2586 if ($r_min != $r_max) {
2587 return 1 if ($r_min < $c->{r});
2588 return 1 if ($r_max > $c->{r});
2589 }
2590 }
2591 return 0 if (defined $limit && --$limit < 0);
2592 show_commit($c);
2593 return 1;
2594 }
2595
2596 sub show_commit {
2597 my $c = shift;
2598 if ($oneline) {
2599 my $x = "\n";
2600 if (my $l = $c->{l}) {
2601 while ($l->[0] =~ /^\s*$/) { shift @$l }
2602 $x = $l->[0];
2603 }
2604 $l_fmt ||= 'A' . length($c->{r});
2605 print 'r',pack($l_fmt, $c->{r}),' | ';
2606 print "$c->{c} | " if $show_commit;
2607 print $x;
2608 } else {
2609 show_commit_normal($c);
2610 }
2611 }
2612
2613 sub show_commit_changed_paths {
2614 my ($c) = @_;
2615 return unless $c->{changed};
2616 print "Changed paths:\n", @{$c->{changed}};
2617 }
2618
2619 sub show_commit_normal {
2620 my ($c) = @_;
2621 print '-' x72, "\nr$c->{r} | ";
2622 print "$c->{c} | " if $show_commit;
2623 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2624 localtime($c->{t_utc})), ' | ';
2625 my $nr_line = 0;
2626
2627 if (my $l = $c->{l}) {
2628 while ($l->[$#$l] eq "\n" && $#$l > 0
2629 && $l->[($#$l - 1)] eq "\n") {
2630 pop @$l;
2631 }
2632 $nr_line = scalar @$l;
2633 if (!$nr_line) {
2634 print "1 line\n\n\n";
2635 } else {
2636 if ($nr_line == 1) {
2637 $nr_line = '1 line';
2638 } else {
2639 $nr_line .= ' lines';
2640 }
2641 print $nr_line, "\n";
2642 show_commit_changed_paths($c);
2643 print "\n";
2644 print $_ foreach @$l;
2645 }
2646 } else {
2647 print "1 line\n";
2648 show_commit_changed_paths($c);
2649 print "\n";
2650
2651 }
2652 foreach my $x (qw/raw diff/) {
2653 if ($c->{$x}) {
2654 print "\n";
2655 print $_ foreach @{$c->{$x}}
2656 }
2657 }
2658 }
2659
2660 sub cmd_show_log {
2661 my (@args) = @_;
2662 my ($r_min, $r_max);
2663 my $r_last = -1; # prevent dupes
2664 if (defined $TZ) {
2665 $ENV{TZ} = $TZ;
2666 } else {
2667 delete $ENV{TZ};
2668 }
2669 if (defined $::_revision) {
2670 if ($::_revision =~ /^(\d+):(\d+)$/) {
2671 ($r_min, $r_max) = ($1, $2);
2672 } elsif ($::_revision =~ /^\d+$/) {
2673 $r_min = $r_max = $::_revision;
2674 } else {
2675 ::fatal "-r$::_revision is not supported, use ",
2676 "standard \'git log\' arguments instead\n";
2677 }
2678 }
2679
2680 config_pager();
2681 @args = (git_svn_log_cmd($r_min, $r_max), @args);
2682 my $log = command_output_pipe(@args);
2683 run_pager();
2684 my (@k, $c, $d);
2685 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
2686 while (<$log>) {
2687 if (/^${esc_color}commit ($::sha1_short)/o) {
2688 my $cmt = $1;
2689 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
2690 $r_last = $c->{r};
2691 process_commit($c, $r_min, $r_max, \@k) or
2692 goto out;
2693 }
2694 $d = undef;
2695 $c = { c => $cmt };
2696 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
2697 get_author_info($c, $1, $2, $3);
2698 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
2699 # ignore
2700 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
2701 push @{$c->{raw}}, $_;
2702 } elsif (/^${esc_color}[ACRMDT]\t/) {
2703 # we could add $SVN->{svn_path} here, but that requires
2704 # remote access at the moment (repo_path_split)...
2705 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
2706 push @{$c->{changed}}, $_;
2707 } elsif (/^${esc_color}diff /o) {
2708 $d = 1;
2709 push @{$c->{diff}}, $_;
2710 } elsif ($d) {
2711 push @{$c->{diff}}, $_;
2712 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
2713 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
2714 } elsif (s/^${esc_color} //o) {
2715 push @{$c->{l}}, $_;
2716 }
2717 }
2718 if ($c && defined $c->{r} && $c->{r} != $r_last) {
2719 $r_last = $c->{r};
2720 process_commit($c, $r_min, $r_max, \@k);
2721 }
2722 if (@k) {
2723 my $swap = $r_max;
2724 $r_max = $r_min;
2725 $r_min = $swap;
2726 process_commit($_, $r_min, $r_max) foreach reverse @k;
2727 }
2728 out:
2729 close $log;
2730 print '-' x72,"\n" unless $incremental || $oneline;
2731 }
2732
2733 package Git::SVN::Migration;
2734 # these version numbers do NOT correspond to actual version numbers
2735 # of git nor git-svn. They are just relative.
2736 #
2737 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
2738 #
2739 # v1 layout: .git/$id/info/url, refs/remotes/$id
2740 #
2741 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
2742 #
2743 # v3 layout: .git/svn/$id, refs/remotes/$id
2744 # - info/url may remain for backwards compatibility
2745 # - this is what we migrate up to this layout automatically,
2746 # - this will be used by git svn init on single branches
2747 #
2748 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
2749 # - this is only created for newly multi-init-ed
2750 # repositories. Similar in spirit to the
2751 # --use-separate-remotes option in git-clone (now default)
2752 # - we do not automatically migrate to this (following
2753 # the example set by core git)
2754 use strict;
2755 use warnings;
2756 use Carp qw/croak/;
2757 use File::Path qw/mkpath/;
2758 use File::Basename qw/dirname basename/;
2759 use vars qw/$_minimize/;
2760
2761 sub migrate_from_v0 {
2762 my $git_dir = $ENV{GIT_DIR};
2763 return undef unless -d $git_dir;
2764 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2765 my $migrated = 0;
2766 while (<$fh>) {
2767 chomp;
2768 my ($id, $orig_ref) = ($_, $_);
2769 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
2770 next unless -f "$git_dir/$id/info/url";
2771 my $new_ref = "refs/remotes/$id";
2772 if (::verify_ref("$new_ref^0")) {
2773 print STDERR "W: $orig_ref is probably an old ",
2774 "branch used by an ancient version of ",
2775 "git-svn.\n",
2776 "However, $new_ref also exists.\n",
2777 "We will not be able ",
2778 "to use this branch until this ",
2779 "ambiguity is resolved.\n";
2780 next;
2781 }
2782 print STDERR "Migrating from v0 layout...\n" if !$migrated;
2783 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
2784 command_noisy('update-ref', $new_ref, $orig_ref);
2785 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
2786 $migrated++;
2787 }
2788 command_close_pipe($fh, $ctx);
2789 print STDERR "Done migrating from v0 layout...\n" if $migrated;
2790 $migrated;
2791 }
2792
2793 sub migrate_from_v1 {
2794 my $git_dir = $ENV{GIT_DIR};
2795 my $migrated = 0;
2796 return $migrated unless -d $git_dir;
2797 my $svn_dir = "$git_dir/svn";
2798
2799 # just in case somebody used 'svn' as their $id at some point...
2800 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
2801
2802 print STDERR "Migrating from a git-svn v1 layout...\n";
2803 mkpath([$svn_dir]);
2804 print STDERR "Data from a previous version of git-svn exists, but\n\t",
2805 "$svn_dir\n\t(required for this version ",
2806 "($::VERSION) of git-svn) does not. exist\n";
2807 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2808 while (<$fh>) {
2809 my $x = $_;
2810 next unless $x =~ s#^refs/remotes/##;
2811 chomp $x;
2812 next unless -f "$git_dir/$x/info/url";
2813 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
2814 next unless $u;
2815 my $dn = dirname("$git_dir/svn/$x");
2816 mkpath([$dn]) unless -d $dn;
2817 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
2818 mkpath(["$git_dir/svn/svn"]);
2819 print STDERR " - $git_dir/$x/info => ",
2820 "$git_dir/svn/$x/info\n";
2821 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
2822 croak "$!: $x";
2823 # don't worry too much about these, they probably
2824 # don't exist with repos this old (save for index,
2825 # and we can easily regenerate that)
2826 foreach my $f (qw/unhandled.log index .rev_db/) {
2827 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
2828 }
2829 } else {
2830 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
2831 rename "$git_dir/$x", "$git_dir/svn/$x" or
2832 croak "$!: $x";
2833 }
2834 $migrated++;
2835 }
2836 command_close_pipe($fh, $ctx);
2837 print STDERR "Done migrating from a git-svn v1 layout\n";
2838 $migrated;
2839 }
2840
2841 sub read_old_urls {
2842 my ($l_map, $pfx, $path) = @_;
2843 my @dir;
2844 foreach (<$path/*>) {
2845 if (-r "$_/info/url") {
2846 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2847 my $ref_id = $pfx . basename $_;
2848 my $url = ::file_to_s("$_/info/url");
2849 $l_map->{$ref_id} = $url;
2850 } elsif (-d $_) {
2851 push @dir, $_;
2852 }
2853 }
2854 foreach (@dir) {
2855 my $x = $_;
2856 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
2857 read_old_urls($l_map, $x, $_);
2858 }
2859 }
2860
2861 sub migrate_from_v2 {
2862 my @cfg = command(qw/config -l/);
2863 return if grep /^svn-remote\..+\.url=/, @cfg;
2864 my %l_map;
2865 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
2866 my $migrated = 0;
2867
2868 foreach my $ref_id (sort keys %l_map) {
2869 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
2870 $migrated++;
2871 }
2872 $migrated;
2873 }
2874
2875 sub minimize_connections {
2876 my $r = Git::SVN::read_all_remotes();
2877 my $new_urls = {};
2878 my $root_repos = {};
2879 foreach my $repo_id (keys %$r) {
2880 my $url = $r->{$repo_id}->{url} or next;
2881 my $fetch = $r->{$repo_id}->{fetch} or next;
2882 my $ra = Git::SVN::Ra->new($url);
2883
2884 # skip existing cases where we already connect to the root
2885 if (($ra->{url} eq $ra->{repos_root}) ||
2886 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
2887 $repo_id)) {
2888 $root_repos->{$ra->{url}} = $repo_id;
2889 next;
2890 }
2891
2892 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
2893 my $root_path = $ra->{url};
2894 $root_path =~ s#^\Q$ra->{repos_root}\E/*##;
2895 foreach my $path (keys %$fetch) {
2896 my $ref_id = $fetch->{$path};
2897 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
2898
2899 # make sure we can read when connecting to
2900 # a higher level of a repository
2901 my ($last_rev, undef) = $gs->last_rev_commit;
2902 if (!defined $last_rev) {
2903 $last_rev = eval {
2904 $root_ra->get_latest_revnum;
2905 };
2906 next if $@;
2907 }
2908 my $new = $root_path;
2909 $new .= length $path ? "/$path" : '';
2910 eval {
2911 $root_ra->get_log([$new], $last_rev, $last_rev,
2912 0, 0, 1, sub { });
2913 };
2914 next if $@;
2915 $new_urls->{$ra->{repos_root}}->{$new} =
2916 { ref_id => $ref_id,
2917 old_repo_id => $repo_id,
2918 old_path => $path };
2919 }
2920 }
2921
2922 my @emptied;
2923 foreach my $url (keys %$new_urls) {
2924 # see if we can re-use an existing [svn-remote "repo_id"]
2925 # instead of creating a(n ugly) new section:
2926 my $repo_id = $root_repos->{$url} ||
2927 Git::SVN::sanitize_remote_name($url);
2928
2929 my $fetch = $new_urls->{$url};
2930 foreach my $path (keys %$fetch) {
2931 my $x = $fetch->{$path};
2932 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
2933 my $pfx = "svn-remote.$x->{old_repo_id}";
2934
2935 my $old_fetch = quotemeta("$x->{old_path}:".
2936 "refs/remotes/$x->{ref_id}");
2937 command_noisy(qw/config --unset/,
2938 "$pfx.fetch", '^'. $old_fetch . '$');
2939 delete $r->{$x->{old_repo_id}}->
2940 {fetch}->{$x->{old_path}};
2941 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
2942 command_noisy(qw/config --unset/,
2943 "$pfx.url");
2944 push @emptied, $x->{old_repo_id}
2945 }
2946 }
2947 }
2948 if (@emptied) {
2949 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
2950 "$ENV{GIT_DIR}/config";
2951 print STDERR <<EOF;
2952 The following [svn-remote] sections in your config file ($file) are empty
2953 and can be safely removed:
2954 EOF
2955 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
2956 }
2957 }
2958
2959 sub migration_check {
2960 migrate_from_v0();
2961 migrate_from_v1();
2962 migrate_from_v2();
2963 minimize_connections() if $_minimize;
2964 }
2965
2966 package Git::IndexInfo;
2967 use strict;
2968 use warnings;
2969 use Git qw/command_input_pipe command_close_pipe/;
2970
2971 sub new {
2972 my ($class) = @_;
2973 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
2974 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
2975 }
2976
2977 sub remove {
2978 my ($self, $path) = @_;
2979 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
2980 return ++$self->{nr};
2981 }
2982 undef;
2983 }
2984
2985 sub update {
2986 my ($self, $mode, $hash, $path) = @_;
2987 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
2988 return ++$self->{nr};
2989 }
2990 undef;
2991 }
2992
2993 sub DESTROY {
2994 my ($self) = @_;
2995 command_close_pipe($self->{gui}, $self->{ctx});
2996 }
2997
2998 __END__
2999
3000 Data structures:
3001
3002 $log_entry hashref as returned by libsvn_log_entry()
3003 {
3004 log => 'whitespace-formatted log entry
3005 ', # trailing newline is preserved
3006 revision => '8', # integer
3007 date => '2004-02-24T17:01:44.108345Z', # commit date
3008 author => 'committer name'
3009 };
3010
3011
3012 # this is generated by generate_diff();
3013 @mods = array of diff-index line hashes, each element represents one line
3014 of diff-index output
3015
3016 diff-index line ($m hash)
3017 {
3018 mode_a => first column of diff-index output, no leading ':',
3019 mode_b => second column of diff-index output,
3020 sha1_b => sha1sum of the final blob,
3021 chg => change type [MCRADT],
3022 file_a => original file name of a file (iff chg is 'C' or 'R')
3023 file_b => new/current file name of a file (any chg)
3024 }
3025 ;
3026
3027 # retval of read_url_paths{,_all}();
3028 $l_map = {
3029 # repository root url
3030 'https://svn.musicpd.org' => {
3031 # repository path # GIT_SVN_ID
3032 'mpd/trunk' => 'trunk',
3033 'mpd/tags/0.11.5' => 'tags/0.11.5',
3034 },
3035 }
3036
3037 Notes:
3038 I don't trust the each() function on unless I created %hash myself
3039 because the internal iterator may not have started at base.
Jonathan Nieder’s miscellaneous git branches