?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-svn: make (multi-)fetch safer but slower
[git/jnareb-git.git] / git-svn.perl
blobefc55156639316259a246e50d8ae7a51439b8155
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision
8 $_q $_authors %users/;
9 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10 $VERSION = '@@GIT_VERSION@@';
11
12 $ENV{GIT_DIR} ||= '.git';
13 $Git::SVN::default_repo_id = 'git-svn';
14 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
15
16 $Git::SVN::Log::TZ = $ENV{TZ};
17 $ENV{TZ} = 'UTC';
18 $| = 1; # unbuffer STDOUT
19
20 sub fatal (@) { print STDERR @_; exit 1 }
21 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
22 require SVN::Ra;
23 require SVN::Delta;
24 if ($SVN::Core::VERSION lt '1.1.0') {
25 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
26 }
27 push @Git::SVN::Ra::ISA, 'SVN::Ra';
28 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
29 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
30 use Carp qw/croak/;
31 use IO::File qw//;
32 use File::Basename qw/dirname basename/;
33 use File::Path qw/mkpath/;
34 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
35 use IPC::Open3;
36 use Git;
37
38 BEGIN {
39 my $s;
40 foreach (qw/command command_oneline command_noisy command_output_pipe
41 command_input_pipe command_close_pipe/) {
42 $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
43 "*Git::SVN::Migration::$_ = ".
44 "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
45 }
46 eval $s;
47 }
48
49 my ($SVN);
50
51 my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
52 $sha1 = qr/[a-f\d]{40}/;
53 $sha1_short = qr/[a-f\d]{4,40}/;
54 my ($_stdin, $_help, $_edit,
55 $_repack, $_repack_nr, $_repack_flags,
56 $_message, $_file,
57 $_template, $_shared,
58 $_version,
59 $_merge, $_strategy, $_dry_run,
60 $_prefix);
61
62 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
63 'config-dir=s' => \$Git::SVN::Ra::config_dir,
64 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
65 my %fc_opts = ( 'follow-parent|follow' => \$Git::SVN::_follow_parent,
66 'authors-file|A=s' => \$_authors,
67 'repack:i' => \$_repack,
68 'no-metadata' => \$Git::SVN::_no_metadata,
69 'quiet|q' => \$_q,
70 'repack-flags|repack-args|repack-opts=s' => \$_repack_flags,
71 %remote_opts );
72
73 my ($_trunk, $_tags, $_branches);
74 my %multi_opts = ( 'trunk|T=s' => \$_trunk,
75 'tags|t=s' => \$_tags,
76 'branches|b=s' => \$_branches );
77 my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
78 my %cmt_opts = ( 'edit|e' => \$_edit,
79 'rmdir' => \$SVN::Git::Editor::_rmdir,
80 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
81 'l=i' => \$SVN::Git::Editor::_rename_limit,
82 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
83 );
84
85 my %cmd = (
86 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
87 { 'revision|r=s' => \$_revision, %fc_opts } ],
88 init => [ \&cmd_init, "Initialize a repo for tracking" .
89 " (requires URL argument)",
90 \%init_opts ],
91 dcommit => [ \&cmd_dcommit,
92 'Commit several diffs to merge with upstream',
93 { 'merge|m|M' => \$_merge,
94 'strategy|s=s' => \$_strategy,
95 'dry-run|n' => \$_dry_run,
96 %cmt_opts, %fc_opts } ],
97 'set-tree' => [ \&cmd_set_tree,
98 "Set an SVN repository to a git tree-ish",
99 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
100 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
101 { 'revision|r=i' => \$_revision } ],
102 'multi-init' => [ \&cmd_multi_init,
103 'Initialize multiple trees (like git-svnimport)',
104 { %multi_opts, %init_opts, %remote_opts,
105 'revision|r=i' => \$_revision,
106 'prefix=s' => \$_prefix,
107 } ],
108 'multi-fetch' => [ \&cmd_multi_fetch,
109 'Fetch multiple trees (like git-svnimport)',
110 \%fc_opts ],
111 'migrate' => [ sub { },
112 # no-op, we automatically run this anyways,
113 'Migrate configuration/metadata/layout from
114 previous versions of git-svn',
115 \%remote_opts ],
116 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
117 { 'limit=i' => \$Git::SVN::Log::limit,
118 'revision|r=s' => \$_revision,
119 'verbose|v' => \$Git::SVN::Log::verbose,
120 'incremental' => \$Git::SVN::Log::incremental,
121 'oneline' => \$Git::SVN::Log::oneline,
122 'show-commit' => \$Git::SVN::Log::show_commit,
123 'non-recursive' => \$Git::SVN::Log::non_recursive,
124 'authors-file|A=s' => \$_authors,
125 'color' => \$Git::SVN::Log::color,
126 'pager=s' => \$Git::SVN::Log::pager,
127 } ],
128 'commit-diff' => [ \&cmd_commit_diff,
129 'Commit a diff between two trees',
130 { 'message|m=s' => \$_message,
131 'file|F=s' => \$_file,
132 'revision|r=s' => \$_revision,
133 %cmt_opts } ],
134 );
135
136 my $cmd;
137 for (my $i = 0; $i < @ARGV; $i++) {
138 if (defined $cmd{$ARGV[$i]}) {
139 $cmd = $ARGV[$i];
140 splice @ARGV, $i, 1;
141 last;
142 }
143 };
144
145 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
146
147 read_repo_config(\%opts);
148 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
149 'minimize-connections' => \$Git::SVN::Migration::_minimize,
150 'id|i=s' => \$Git::SVN::default_ref_id,
151 'svn-remote|remote|R=s' => \$Git::SVN::default_repo_id);
152 exit 1 if (!$rv && $cmd ne 'log');
153
154 usage(0) if $_help;
155 version() if $_version;
156 usage(1) unless defined $cmd;
157 load_authors() if $_authors;
158 unless ($cmd =~ /^(?:init|multi-init|commit-diff)$/) {
159 Git::SVN::Migration::migration_check();
160 }
161 eval {
162 Git::SVN::verify_remotes_sanity();
163 $cmd{$cmd}->[0]->(@ARGV);
164 };
165 fatal $@ if $@;
166 exit 0;
167
168 ####################### primary functions ######################
169 sub usage {
170 my $exit = shift || 0;
171 my $fd = $exit ? \*STDERR : \*STDOUT;
172 print $fd <<"";
173 git-svn - bidirectional operations between a single Subversion tree and git
174 Usage: $0 <command> [options] [arguments]\n
175
176 print $fd "Available commands:\n" unless $cmd;
177
178 foreach (sort keys %cmd) {
179 next if $cmd && $cmd ne $_;
180 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
181 foreach (keys %{$cmd{$_}->[2]}) {
182 # prints out arguments as they should be passed:
183 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
184 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
185 "--$_" : "-$_" }
186 split /\|/,$_)," $x\n";
187 }
188 }
189 print $fd <<"";
190 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
191 arbitrary identifier if you're tracking multiple SVN branches/repositories in
192 one git repository and want to keep them separate. See git-svn(1) for more
193 information.
194
195 exit $exit;
196 }
197
198 sub version {
199 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
200 exit 0;
201 }
202
203 sub do_git_init_db {
204 unless (-d $ENV{GIT_DIR}) {
205 my @init_db = ('init');
206 push @init_db, "--template=$_template" if defined $_template;
207 push @init_db, "--shared" if defined $_shared;
208 command_noisy(@init_db);
209 }
210 }
211
212 sub cmd_init {
213 my $url = shift or die "SVN repository location required " .
214 "as a command-line argument\n";
215 if (my $repo_path = shift) {
216 unless (-d $repo_path) {
217 mkpath([$repo_path]);
218 }
219 chdir $repo_path or croak $!;
220 $ENV{GIT_DIR} = $repo_path . "/.git";
221 }
222 do_git_init_db();
223
224 Git::SVN->init($url);
225 }
226
227 sub cmd_fetch {
228 if (@_) {
229 die "Additional fetch arguments are no longer supported.\n",
230 "Use --follow-parent if you have moved/copied directories
231 instead.\n";
232 }
233 my $gs = Git::SVN->new;
234 $gs->fetch(parse_revision_argument());
235 if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
236 command_noisy(qw(update-ref refs/heads/master),
237 $gs->{last_commit});
238 }
239 }
240
241 sub cmd_set_tree {
242 my (@commits) = @_;
243 if ($_stdin || !@commits) {
244 print "Reading from stdin...\n";
245 @commits = ();
246 while (<STDIN>) {
247 if (/\b($sha1_short)\b/o) {
248 unshift @commits, $1;
249 }
250 }
251 }
252 my @revs;
253 foreach my $c (@commits) {
254 my @tmp = command('rev-parse',$c);
255 if (scalar @tmp == 1) {
256 push @revs, $tmp[0];
257 } elsif (scalar @tmp > 1) {
258 push @revs, reverse(command('rev-list',@tmp));
259 } else {
260 fatal "Failed to rev-parse $c\n";
261 }
262 }
263 my $gs = Git::SVN->new;
264 my ($r_last, $cmt_last) = $gs->last_rev_commit;
265 $gs->fetch;
266 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
267 fatal "There are new revisions that were fetched ",
268 "and need to be merged (or acknowledged) ",
269 "before committing.\nlast rev: $r_last\n",
270 " current: $gs->{last_rev}\n";
271 }
272 $gs->set_tree($_) foreach @revs;
273 print "Done committing ",scalar @revs," revisions to SVN\n";
274 }
275
276 sub cmd_dcommit {
277 my $head = shift;
278 my $gs = Git::SVN->new;
279 $head ||= 'HEAD';
280 my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
281 my $last_rev;
282 foreach my $d (reverse @refs) {
283 if (!verify_ref("$d~1")) {
284 fatal "Commit $d\n",
285 "has no parent commit, and therefore ",
286 "nothing to diff against.\n",
287 "You should be working from a repository ",
288 "originally created by git-svn\n";
289 }
290 unless (defined $last_rev) {
291 (undef, $last_rev, undef) = cmt_metadata("$d~1");
292 unless (defined $last_rev) {
293 fatal "Unable to extract revision information ",
294 "from commit $d~1\n";
295 }
296 }
297 if ($_dry_run) {
298 print "diff-tree $d~1 $d\n";
299 } else {
300 my %ed_opts = ( r => $last_rev,
301 log => get_commit_entry($d)->{log},
302 ra => $gs->ra,
303 tree_a => "$d~1",
304 tree_b => $d,
305 editor_cb => sub {
306 print "Committed r$_[0]\n";
307 $last_rev = $_[0]; },
308 svn_path => $gs->{path} );
309 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
310 print "No changes\n$d~1 == $d\n";
311 }
312 }
313 }
314 return if $_dry_run;
315 $gs->fetch;
316 # we always want to rebase against the current HEAD, not any
317 # head that was passed to us
318 my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
319 my @finish;
320 if (@diff) {
321 @finish = qw/rebase/;
322 push @finish, qw/--merge/ if $_merge;
323 push @finish, "--strategy=$_strategy" if $_strategy;
324 print STDERR "W: HEAD and ", $gs->refname, " differ, ",
325 "using @finish:\n", "@diff";
326 } else {
327 print "No changes between current HEAD and ",
328 $gs->refname, "\nResetting to the latest ",
329 $gs->refname, "\n";
330 @finish = qw/reset --mixed/;
331 }
332 command_noisy(@finish, $gs->refname);
333 }
334
335 sub cmd_show_ignore {
336 my $gs = Git::SVN->new;
337 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
338 $gs->traverse_ignore(\*STDOUT, '', $r);
339 }
340
341 sub cmd_multi_init {
342 my $url = shift;
343 unless (defined $_trunk || defined $_branches || defined $_tags) {
344 usage(1);
345 }
346 do_git_init_db();
347 $_prefix = '' unless defined $_prefix;
348 $url =~ s#/+$## if defined $url;
349 if (defined $_trunk) {
350 my $trunk_ref = $_prefix . 'trunk';
351 # try both old-style and new-style lookups:
352 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
353 unless ($gs_trunk) {
354 my ($trunk_url, $trunk_path) =
355 complete_svn_url($url, $_trunk);
356 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
357 undef, $trunk_ref);
358 }
359 }
360 return unless defined $_branches || defined $_tags;
361 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
362 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
363 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
364 }
365
366 sub cmd_multi_fetch {
367 my $remotes = Git::SVN::read_all_remotes();
368 foreach my $repo_id (sort keys %$remotes) {
369 my $url = $remotes->{$repo_id}->{url} or next;
370 my $fetch = $remotes->{$repo_id}->{fetch} or next;
371 Git::SVN::fetch_all($repo_id, $url, $fetch);
372 }
373 }
374
375 # this command is special because it requires no metadata
376 sub cmd_commit_diff {
377 my ($ta, $tb, $url) = @_;
378 my $usage = "Usage: $0 commit-diff -r<revision> ".
379 "<tree-ish> <tree-ish> [<URL>]\n";
380 fatal($usage) if (!defined $ta || !defined $tb);
381 my $svn_path;
382 if (!defined $url) {
383 my $gs = eval { Git::SVN->new };
384 if (!$gs) {
385 fatal("Needed URL or usable git-svn --id in ",
386 "the command-line\n", $usage);
387 }
388 $url = $gs->{url};
389 $svn_path = $gs->{path};
390 }
391 unless (defined $_revision) {
392 fatal("-r|--revision is a required argument\n", $usage);
393 }
394 if (defined $_message && defined $_file) {
395 fatal("Both --message/-m and --file/-F specified ",
396 "for the commit message.\n",
397 "I have no idea what you mean\n");
398 }
399 if (defined $_file) {
400 $_message = file_to_s($_file);
401 } else {
402 $_message ||= get_commit_entry($tb)->{log};
403 }
404 my $ra ||= Git::SVN::Ra->new($url);
405 $svn_path ||= $ra->{svn_path};
406 my $r = $_revision;
407 if ($r eq 'HEAD') {
408 $r = $ra->get_latest_revnum;
409 } elsif ($r !~ /^\d+$/) {
410 die "revision argument: $r not understood by git-svn\n";
411 }
412 my %ed_opts = ( r => $r,
413 log => $_message,
414 ra => $ra,
415 tree_a => $ta,
416 tree_b => $tb,
417 editor_cb => sub { print "Committed r$_[0]\n" },
418 svn_path => $svn_path );
419 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
420 print "No changes\n$ta == $tb\n";
421 }
422 }
423
424 ########################### utility functions #########################
425
426 sub parse_revision_argument {
427 if (!defined $_revision || $_revision eq 'BASE:HEAD') {
428 return (undef, undef);
429 }
430 return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
431 return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
432 return (undef, $1) if ($_revision =~ /^BASE:(\d+)$/);
433 return ($1, undef) if ($_revision =~ /^(\d+):HEAD$/);
434 die "revision argument: $_revision not understood by git-svn\n",
435 "Try using the command-line svn client instead\n";
436 }
437
438 sub complete_svn_url {
439 my ($url, $path) = @_;
440 $path =~ s#/+$##;
441 if ($path !~ m#^[a-z\+]+://#) {
442 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
443 fatal("E: '$path' is not a complete URL ",
444 "and a separate URL is not specified\n");
445 }
446 return ($url, $path);
447 }
448 return ($path, '');
449 }
450
451 sub complete_url_ls_init {
452 my ($ra, $repo_path, $switch, $pfx) = @_;
453 unless ($repo_path) {
454 print STDERR "W: $switch not specified\n";
455 return;
456 }
457 $repo_path =~ s#/+$##;
458 if ($repo_path =~ m#^[a-z\+]+://#) {
459 $ra = Git::SVN::Ra->new($repo_path);
460 $repo_path = '';
461 } else {
462 $repo_path =~ s#^/+##;
463 unless ($ra) {
464 fatal("E: '$repo_path' is not a complete URL ",
465 "and a separate URL is not specified\n");
466 }
467 }
468 my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
469 my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
470 my $url = $ra->{url};
471 foreach my $d (sort keys %$dirent) {
472 next if ($dirent->{$d}->kind != $SVN::Node::dir);
473 my $path = "$repo_path/$d";
474 my $ref = "$pfx$d";
475 my $gs = eval { Git::SVN->new($ref) };
476 # don't try to init already existing refs
477 unless ($gs) {
478 print "init $url/$path => $ref\n";
479 Git::SVN->init($url, $path, undef, $ref);
480 }
481 }
482 }
483
484 sub verify_ref {
485 my ($ref) = @_;
486 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
487 { STDERR => 0 }); };
488 }
489
490 sub get_tree_from_treeish {
491 my ($treeish) = @_;
492 # $treeish can be a symbolic ref, too:
493 my $type = command_oneline(qw/cat-file -t/, $treeish);
494 my $expected;
495 while ($type eq 'tag') {
496 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
497 }
498 if ($type eq 'commit') {
499 $expected = (grep /^tree /, command(qw/cat-file commit/,
500 $treeish))[0];
501 ($expected) = ($expected =~ /^tree ($sha1)$/o);
502 die "Unable to get tree from $treeish\n" unless $expected;
503 } elsif ($type eq 'tree') {
504 $expected = $treeish;
505 } else {
506 die "$treeish is a $type, expected tree, tag or commit\n";
507 }
508 return $expected;
509 }
510
511 sub get_commit_entry {
512 my ($treeish) = shift;
513 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
514 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
515 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
516 open my $log_fh, '>', $commit_editmsg or croak $!;
517
518 my $type = command_oneline(qw/cat-file -t/, $treeish);
519 if ($type eq 'commit' || $type eq 'tag') {
520 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
521 $type, $treeish);
522 my $in_msg = 0;
523 while (<$msg_fh>) {
524 if (!$in_msg) {
525 $in_msg = 1 if (/^\s*$/);
526 } elsif (/^git-svn-id: /) {
527 # skip this for now, we regenerate the
528 # correct one on re-fetch anyways
529 # TODO: set *:merge properties or like...
530 } else {
531 print $log_fh $_ or croak $!;
532 }
533 }
534 command_close_pipe($msg_fh, $ctx);
535 }
536 close $log_fh or croak $!;
537
538 if ($_edit || ($type eq 'tree')) {
539 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
540 # TODO: strip out spaces, comments, like git-commit.sh
541 system($editor, $commit_editmsg);
542 }
543 rename $commit_editmsg, $commit_msg or croak $!;
544 open $log_fh, '<', $commit_msg or croak $!;
545 { local $/; chomp($log_entry{log} = <$log_fh>); }
546 close $log_fh or croak $!;
547 unlink $commit_msg;
548 \%log_entry;
549 }
550
551 sub s_to_file {
552 my ($str, $file, $mode) = @_;
553 open my $fd,'>',$file or croak $!;
554 print $fd $str,"\n" or croak $!;
555 close $fd or croak $!;
556 chmod ($mode &~ umask, $file) if (defined $mode);
557 }
558
559 sub file_to_s {
560 my $file = shift;
561 open my $fd,'<',$file or croak "$!: file: $file\n";
562 local $/;
563 my $ret = <$fd>;
564 close $fd or croak $!;
565 $ret =~ s/\s*$//s;
566 return $ret;
567 }
568
569 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
570 sub load_authors {
571 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
572 my $log = $cmd eq 'log';
573 while (<$authors>) {
574 chomp;
575 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
576 my ($user, $name, $email) = ($1, $2, $3);
577 if ($log) {
578 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
579 } else {
580 $users{$user} = [$name, $email];
581 }
582 }
583 close $authors or croak $!;
584 }
585
586 # convert GetOpt::Long specs for use by git-config
587 sub read_repo_config {
588 return unless -d $ENV{GIT_DIR};
589 my $opts = shift;
590 foreach my $o (keys %$opts) {
591 my $v = $opts->{$o};
592 my ($key) = ($o =~ /^([a-z\-]+)/);
593 $key =~ s/-//g;
594 my $arg = 'git-config';
595 $arg .= ' --int' if ($o =~ /[:=]i$/);
596 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
597 if (ref $v eq 'ARRAY') {
598 chomp(my @tmp = `$arg --get-all svn.$key`);
599 @$v = @tmp if @tmp;
600 } else {
601 chomp(my $tmp = `$arg --get svn.$key`);
602 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
603 $$v = $tmp;
604 }
605 }
606 }
607 }
608
609 sub extract_metadata {
610 my $id = shift or return (undef, undef, undef);
611 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
612 \s([a-f\d\-]+)$/x);
613 if (!defined $rev || !$uuid || !$url) {
614 # some of the original repositories I made had
615 # identifiers like this:
616 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
617 }
618 return ($url, $rev, $uuid);
619 }
620
621 sub cmt_metadata {
622 return extract_metadata((grep(/^git-svn-id: /,
623 command(qw/cat-file commit/, shift)))[-1]);
624 }
625
626 package Git::SVN;
627 use strict;
628 use warnings;
629 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent/;
630 use Carp qw/croak/;
631 use File::Path qw/mkpath/;
632 use IPC::Open3;
633
634 # properties that we do not log:
635 my %SKIP_PROP;
636 BEGIN {
637 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
638 svn:special svn:executable
639 svn:entry:committed-rev
640 svn:entry:last-author
641 svn:entry:uuid
642 svn:entry:committed-date/;
643 }
644
645 sub fetch_all {
646 my ($repo_id, $url, $fetch) = @_;
647 my @gs;
648 my $ra = Git::SVN::Ra->new($url);
649 my $head = $ra->get_latest_revnum;
650 my $base = $head;
651 my $new_remote;
652 foreach my $p (sort keys %$fetch) {
653 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
654 my $lr = $gs->last_rev;
655 if (defined $lr) {
656 $base = $lr if ($lr < $base);
657 } else {
658 $new_remote = 1;
659 }
660 push @gs, $gs;
661 }
662 $base = 0 if $new_remote;
663 return if (++$base > $head);
664 $ra->gs_fetch_loop_common($base, $head, @gs);
665 }
666
667 sub read_all_remotes {
668 my $r = {};
669 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
670 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
671 $r->{$1}->{fetch}->{$2} = $3;
672 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
673 $r->{$1}->{url} = $2;
674 }
675 }
676 $r;
677 }
678
679 sub verify_remotes_sanity {
680 return unless -d $ENV{GIT_DIR};
681 my %seen;
682 foreach (command(qw/config -l/)) {
683 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
684 if ($seen{$1}) {
685 die "Remote ref refs/remote/$1 is tracked by",
686 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
687 "Please resolve this ambiguity in ",
688 "your git configuration file before ",
689 "continuing\n";
690 }
691 $seen{$1} = $_;
692 }
693 }
694 }
695
696 # we allow more chars than remotes2config.sh...
697 sub sanitize_remote_name {
698 my ($name) = @_;
699 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
700 $name;
701 }
702
703 sub find_existing_remote {
704 my ($url, $remotes) = @_;
705 my $existing;
706 foreach my $repo_id (keys %$remotes) {
707 my $u = $remotes->{$repo_id}->{url} or next;
708 next if $u ne $url;
709 $existing = $repo_id;
710 last;
711 }
712 $existing;
713 }
714
715 sub init_remote_config {
716 my ($self, $url) = @_;
717 $url =~ s!/+$!!; # strip trailing slash
718 my $r = read_all_remotes();
719 my $existing = find_existing_remote($url, $r);
720 if ($existing) {
721 print STDERR "Using existing ",
722 "[svn-remote \"$existing\"]\n";
723 $self->{repo_id} = $existing;
724 } else {
725 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
726 $existing = find_existing_remote($min_url, $r);
727 if ($existing) {
728 print STDERR "Using existing ",
729 "[svn-remote \"$existing\"]\n";
730 $self->{repo_id} = $existing;
731 }
732 if ($min_url ne $url) {
733 print STDERR "Using higher level of URL: ",
734 "$url => $min_url\n";
735 my $old_path = $self->{path};
736 $self->{path} = $url;
737 $self->{path} =~ s!^\Q$min_url\E/*!!;
738 if (length $old_path) {
739 $self->{path} .= "/$old_path";
740 }
741 $url = $min_url;
742 }
743 }
744 my $orig_url;
745 if (!$existing) {
746 # verify that we aren't overwriting anything:
747 $orig_url = eval {
748 command_oneline('config', '--get',
749 "svn-remote.$self->{repo_id}.url")
750 };
751 if ($orig_url && ($orig_url ne $url)) {
752 die "svn-remote.$self->{repo_id}.url already set: ",
753 "$orig_url\nwanted to set to: $url\n";
754 }
755 }
756 my ($xrepo_id, $xpath) = find_ref($self->refname);
757 if (defined $xpath) {
758 die "svn-remote.$xrepo_id.fetch already set to track ",
759 "$xpath:refs/remotes/", $self->refname, "\n";
760 }
761 command_noisy('config',
762 "svn-remote.$self->{repo_id}.url", $url);
763 command_noisy('config', '--add',
764 "svn-remote.$self->{repo_id}.fetch",
765 "$self->{path}:".$self->refname);
766 $self->{url} = $url;
767 }
768
769 sub init {
770 my ($class, $url, $path, $repo_id, $ref_id) = @_;
771 my $self = _new($class, $repo_id, $ref_id, $path);
772 if (defined $url) {
773 $self->init_remote_config($url);
774 }
775 $self;
776 }
777
778 sub find_ref {
779 my ($ref_id) = @_;
780 foreach (command(qw/config -l/)) {
781 next unless m!^svn-remote\.(.+)\.fetch=
782 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
783 my ($repo_id, $path, $ref) = ($1, $2, $3);
784 if ($ref eq $ref_id) {
785 $path = '' if ($path =~ m#^\./?#);
786 return ($repo_id, $path);
787 }
788 }
789 (undef, undef, undef);
790 }
791
792 sub new {
793 my ($class, $ref_id, $repo_id, $path) = @_;
794 if (defined $ref_id && !defined $repo_id && !defined $path) {
795 ($repo_id, $path) = find_ref($ref_id);
796 if (!defined $repo_id) {
797 die "Could not find a \"svn-remote.*.fetch\" key ",
798 "in the repository configuration matching: ",
799 "refs/remotes/$ref_id\n";
800 }
801 }
802 my $self = _new($class, $repo_id, $ref_id, $path);
803 if (!defined $self->{path} || !length $self->{path}) {
804 my $fetch = command_oneline('config', '--get',
805 "svn-remote.$repo_id.fetch",
806 ":refs/remotes/$ref_id\$") or
807 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
808 "\":refs/remotes/$ref_id\$\" in config\n";
809 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
810 }
811 $self->{url} = command_oneline('config', '--get',
812 "svn-remote.$repo_id.url") or
813 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
814 if (-z $self->{db_path} && ::verify_ref($self->refname.'^0')) {
815 $self->rebuild;
816 }
817 $self;
818 }
819
820 sub refname { "refs/remotes/$_[0]->{ref_id}" }
821
822 sub ra {
823 my ($self) = shift;
824 Git::SVN::Ra->new($self->{url});
825 }
826
827 sub rel_path {
828 my ($self) = @_;
829 my $repos_root = $self->ra->{repos_root};
830 return $self->{path} if ($self->{url} eq $repos_root);
831 my $url = $self->{url} .
832 (length $self->{path} ? "/$self->{path}" : $self->{path});
833 $url =~ s!^\Q$repos_root\E/*!!g;
834 $url;
835 }
836
837 sub traverse_ignore {
838 my ($self, $fh, $path, $r) = @_;
839 $path =~ s#^/+##g;
840 my $ra = $self->ra;
841 my ($dirent, undef, $props) = $ra->get_dir($path, $r);
842 my $p = $path;
843 $p =~ s#^\Q$ra->{svn_path}\E/##;
844 print $fh length $p ? "\n# $p\n" : "\n# /\n";
845 if (my $s = $props->{'svn:ignore'}) {
846 $s =~ s/[\r\n]+/\n/g;
847 chomp $s;
848 if (length $p == 0) {
849 $s =~ s#\n#\n/$p#g;
850 print $fh "/$s\n";
851 } else {
852 $s =~ s#\n#\n/$p/#g;
853 print $fh "/$p/$s\n";
854 }
855 }
856 foreach (sort keys %$dirent) {
857 next if $dirent->{$_}->kind != $SVN::Node::dir;
858 $self->traverse_ignore($fh, "$path/$_", $r);
859 }
860 }
861
862 sub last_rev { ($_[0]->last_rev_commit)[0] }
863 sub last_commit { ($_[0]->last_rev_commit)[1] }
864
865 # returns the newest SVN revision number and newest commit SHA1
866 sub last_rev_commit {
867 my ($self) = @_;
868 if (defined $self->{last_rev} && defined $self->{last_commit}) {
869 return ($self->{last_rev}, $self->{last_commit});
870 }
871 my $c = ::verify_ref($self->refname.'^0');
872 if ($c) {
873 my $rev = (::cmt_metadata($c))[1];
874 if (defined $rev) {
875 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
876 return ($rev, $c);
877 }
878 }
879 my $offset = -41; # from tail
880 my $rl;
881 open my $fh, '<', $self->{db_path} or
882 croak "$self->{db_path} not readable: $!\n";
883 seek $fh, $offset, 2;
884 $rl = readline $fh;
885 defined $rl or return (undef, undef);
886 chomp $rl;
887 while ($c ne $rl && tell $fh != 0) {
888 $offset -= 41;
889 seek $fh, $offset, 2;
890 $rl = readline $fh;
891 defined $rl or return (undef, undef);
892 chomp $rl;
893 }
894 my $rev = tell $fh;
895 croak $! if ($rev < 0);
896 $rev = ($rev - 41) / 41;
897 close $fh or croak $!;
898 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
899 return ($rev, $c);
900 }
901
902 sub get_fetch_range {
903 my ($self, $min, $max) = @_;
904 $max ||= $self->ra->get_latest_revnum;
905 $min ||= $self->last_rev || 0;
906 (++$min, $max);
907 }
908
909 sub tmp_index_do {
910 my ($self, $sub) = @_;
911 my $old_index = $ENV{GIT_INDEX_FILE};
912 $ENV{GIT_INDEX_FILE} = $self->{index};
913 my @ret = &$sub;
914 if ($old_index) {
915 $ENV{GIT_INDEX_FILE} = $old_index;
916 } else {
917 delete $ENV{GIT_INDEX_FILE};
918 }
919 wantarray ? @ret : $ret[0];
920 }
921
922 sub assert_index_clean {
923 my ($self, $treeish) = @_;
924
925 $self->tmp_index_do(sub {
926 command_noisy('read-tree', $treeish) unless -e $self->{index};
927 my $x = command_oneline('write-tree');
928 my ($y) = (command(qw/cat-file commit/, $treeish) =~
929 /^tree ($::sha1)/mo);
930 if ($y ne $x) {
931 unlink $self->{index} or croak $!;
932 command_noisy('read-tree', $treeish);
933 }
934 $x = command_oneline('write-tree');
935 if ($y ne $x) {
936 ::fatal "trees ($treeish) $y != $x\n",
937 "Something is seriously wrong...\n";
938 }
939 });
940 }
941
942 sub get_commit_parents {
943 my ($self, $log_entry) = @_;
944 my (%seen, @ret, @tmp);
945 # legacy support for 'set-tree'; this is only used by set_tree_cb:
946 if (my $ip = $self->{inject_parents}) {
947 if (my $commit = delete $ip->{$log_entry->{revision}}) {
948 push @tmp, $commit;
949 }
950 }
951 if (my $cur = ::verify_ref($self->refname.'^0')) {
952 push @tmp, $cur;
953 }
954 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
955 while (my $p = shift @tmp) {
956 next if $seen{$p};
957 $seen{$p} = 1;
958 push @ret, $p;
959 # MAXPARENT is defined to 16 in commit-tree.c:
960 last if @ret >= 16;
961 }
962 if (@tmp) {
963 die "r$log_entry->{revision}: No room for parents:\n\t",
964 join("\n\t", @tmp), "\n";
965 }
966 @ret;
967 }
968
969 sub full_url {
970 my ($self) = @_;
971 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
972 }
973
974 sub do_git_commit {
975 my ($self, $log_entry) = @_;
976 my $lr = $self->last_rev;
977 if (defined $lr && $lr >= $log_entry->{revision}) {
978 die "Last fetched revision of ", $self->refname,
979 " was r$lr, but we are about to fetch: ",
980 "r$log_entry->{revision}!\n";
981 }
982 if (my $c = $self->rev_db_get($log_entry->{revision})) {
983 croak "$log_entry->{revision} = $c already exists! ",
984 "Why are we refetching it?\n";
985 }
986 my $author = $log_entry->{author};
987 my ($name, $email) = (defined $::users{$author} ? @{$::users{$author}}
988 : ($author, "$author\@".$self->ra->uuid));
989 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
990 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
991 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
992
993 my $tree = $log_entry->{tree};
994 if (!defined $tree) {
995 $tree = $self->tmp_index_do(sub {
996 command_oneline('write-tree') });
997 }
998 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
999
1000 my @exec = ('git-commit-tree', $tree);
1001 foreach ($self->get_commit_parents($log_entry)) {
1002 push @exec, '-p', $_;
1003 }
1004 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1005 or croak $!;
1006 print $msg_fh $log_entry->{log} or croak $!;
1007 unless ($_no_metadata) {
1008 print $msg_fh "\ngit-svn-id: ", $self->full_url, '@',
1009 $log_entry->{revision}, ' ',
1010 $self->ra->uuid, "\n" or croak $!;
1011 }
1012 $msg_fh->flush == 0 or croak $!;
1013 close $msg_fh or croak $!;
1014 chomp(my $commit = do { local $/; <$out_fh> });
1015 close $out_fh or croak $!;
1016 waitpid $pid, 0;
1017 croak $? if $?;
1018 if ($commit !~ /^$::sha1$/o) {
1019 die "Failed to commit, invalid sha1: $commit\n";
1020 }
1021
1022 command_noisy('update-ref',$self->refname, $commit);
1023 $self->rev_db_set($log_entry->{revision}, $commit);
1024
1025 $self->{last_rev} = $log_entry->{revision};
1026 $self->{last_commit} = $commit;
1027 print "r$log_entry->{revision} = $commit ($self->{ref_id})\n";
1028 return $commit;
1029 }
1030
1031 sub revisions_eq {
1032 my ($self, $r0, $r1) = @_;
1033 return 1 if $r0 == $r1;
1034 my $nr = 0;
1035 $self->ra->get_log([$self->{path}], $r0, $r1,
1036 0, 0, 1, sub { $nr++ });
1037 return 0 if ($nr > 1);
1038 return 1;
1039 }
1040
1041 sub find_parent_branch {
1042 my ($self, $paths, $rev) = @_;
1043 return undef unless $_follow_parent;
1044 unless (defined $paths) {
1045 my $err_handler = $SVN::Error::handler;
1046 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1047 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1048 $paths =
1049 Git::SVN::Ra::dup_changed_paths($_[0]) });
1050 $SVN::Error::handler = $err_handler;
1051 }
1052 return undef unless defined $paths;
1053
1054 # look for a parent from another branch:
1055 my @b_path_components = split m#/#, $self->rel_path;
1056 my @a_path_components;
1057 my $i;
1058 while (@b_path_components) {
1059 $i = $paths->{'/'.join('/', @b_path_components)};
1060 last if $i;
1061 unshift(@a_path_components, pop(@b_path_components));
1062 }
1063 goto not_found unless defined $i;
1064 my $branch_from = $i->{copyfrom_path} or goto not_found;
1065 if (@a_path_components) {
1066 print STDERR "branch_from: $branch_from => ";
1067 $branch_from .= '/'.join('/', @a_path_components);
1068 print STDERR $branch_from, "\n";
1069 }
1070 my $r = $i->{copyfrom_rev};
1071 my $repos_root = $self->ra->{repos_root};
1072 my $url = $self->ra->{url};
1073 my $new_url = $repos_root . $branch_from;
1074 print STDERR "Found possible branch point: ",
1075 "$new_url => ", $self->full_url, ", $r\n";
1076 $branch_from =~ s#^/##;
1077 my $remotes = read_all_remotes();
1078 my $gs;
1079 foreach my $repo_id (keys %$remotes) {
1080 my $u = $remotes->{$repo_id}->{url} or next;
1081 next if $url ne $u;
1082 my $fetch = $remotes->{$repo_id}->{fetch};
1083 foreach my $f (keys %$fetch) {
1084 next if $f ne $branch_from;
1085 $gs = Git::SVN->new($fetch->{$f}, $repo_id, $f);
1086 last;
1087 }
1088 last if $gs;
1089 }
1090 unless ($gs) {
1091 my $ref_id = $self->{ref_id};
1092 $ref_id =~ s/\@\d+$//;
1093 $ref_id .= "\@$r";
1094 # just grow a tail if we're not unique enough :x
1095 $ref_id .= '-' while find_ref($ref_id);
1096 print STDERR "Initializing parent: $ref_id\n";
1097 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id);
1098 }
1099 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1100 if ($_follow_parent && (!defined $r0 || !defined $parent)) {
1101 $gs->fetch(0, $r);
1102 ($r0, $parent) = $gs->last_rev_commit;
1103 }
1104 if (defined $r0 && defined $parent && $gs->revisions_eq($r0, $r)) {
1105 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1106 $self->assert_index_clean($parent);
1107 my $ed;
1108 if ($self->ra->can_do_switch) {
1109 print STDERR "Following parent with do_switch\n";
1110 # do_switch works with svn/trunk >= r22312, but that
1111 # is not included with SVN 1.4.3 (the latest version
1112 # at the moment), so we can't rely on it
1113 $self->{last_commit} = $parent;
1114 $ed = SVN::Git::Fetcher->new($self);
1115 $gs->ra->gs_do_switch($r0, $rev, $gs,
1116 $self->full_url, $ed)
1117 or die "SVN connection failed somewhere...\n";
1118 } else {
1119 print STDERR "Following parent with do_update\n";
1120 $ed = SVN::Git::Fetcher->new($self);
1121 $self->ra->gs_do_update($rev, $rev, $self, $ed)
1122 or die "SVN connection failed somewhere...\n";
1123 }
1124 print STDERR "Successfully followed parent\n";
1125 return $self->make_log_entry($rev, [$parent], $ed);
1126 }
1127 not_found:
1128 print STDERR "Branch parent for path: '/",
1129 $self->rel_path, "' @ r$rev not found:\n";
1130 return undef unless $paths;
1131 print STDERR "Changed paths:\n";
1132 foreach my $x (sort keys %$paths) {
1133 my $p = $paths->{$x};
1134 print STDERR "\t$p->{action}\t$x";
1135 if ($p->{copyfrom_path}) {
1136 print STDERR "(from $p->{copyfrom_path}: ",
1137 "$p->{copyfrom_rev})";
1138 }
1139 print STDERR "\n";
1140 }
1141 print STDERR '-'x72, "\n";
1142 return undef;
1143 }
1144
1145 sub do_fetch {
1146 my ($self, $paths, $rev) = @_;
1147 my $ed;
1148 my ($last_rev, @parents);
1149 if ($self->{last_commit}) {
1150 $ed = SVN::Git::Fetcher->new($self);
1151 $last_rev = $self->{last_rev};
1152 $ed->{c} = $self->{last_commit};
1153 @parents = ($self->{last_commit});
1154 } else {
1155 $last_rev = $rev;
1156 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1157 return $log_entry;
1158 }
1159 $ed = SVN::Git::Fetcher->new($self);
1160 }
1161 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1162 die "SVN connection failed somewhere...\n";
1163 }
1164 $self->make_log_entry($rev, \@parents, $ed);
1165 }
1166
1167 sub get_untracked {
1168 my ($self, $ed) = @_;
1169 my @out;
1170 my $h = $ed->{empty};
1171 foreach (sort keys %$h) {
1172 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1173 push @out, " $act: " . uri_encode($_);
1174 warn "W: $act: $_\n";
1175 }
1176 foreach my $t (qw/dir_prop file_prop/) {
1177 $h = $ed->{$t} or next;
1178 foreach my $path (sort keys %$h) {
1179 my $ppath = $path eq '' ? '.' : $path;
1180 foreach my $prop (sort keys %{$h->{$path}}) {
1181 next if $SKIP_PROP{$prop};
1182 my $v = $h->{$path}->{$prop};
1183 my $t_ppath_prop = "$t: " .
1184 uri_encode($ppath) . ' ' .
1185 uri_encode($prop);
1186 if (defined $v) {
1187 push @out, " +$t_ppath_prop " .
1188 uri_encode($v);
1189 } else {
1190 push @out, " -$t_ppath_prop";
1191 }
1192 }
1193 }
1194 }
1195 foreach my $t (qw/absent_file absent_directory/) {
1196 $h = $ed->{$t} or next;
1197 foreach my $parent (sort keys %$h) {
1198 foreach my $path (sort @{$h->{$parent}}) {
1199 push @out, " $t: " .
1200 uri_encode("$parent/$path");
1201 warn "W: $t: $parent/$path ",
1202 "Insufficient permissions?\n";
1203 }
1204 }
1205 }
1206 \@out;
1207 }
1208
1209 sub parse_svn_date {
1210 my $date = shift || return '+0000 1970-01-01 00:00:00';
1211 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1212 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1213 croak "Unable to parse date: $date\n";
1214 "+0000 $Y-$m-$d $H:$M:$S";
1215 }
1216
1217 sub check_author {
1218 my ($author) = @_;
1219 if (!defined $author || length $author == 0) {
1220 $author = '(no author)';
1221 }
1222 if (defined $::_authors && ! defined $::users{$author}) {
1223 die "Author: $author not defined in $::_authors file\n";
1224 }
1225 $author;
1226 }
1227
1228 sub make_log_entry {
1229 my ($self, $rev, $parents, $ed) = @_;
1230 my $untracked = $self->get_untracked($ed);
1231
1232 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1233 print $un "r$rev\n" or croak $!;
1234 print $un $_, "\n" foreach @$untracked;
1235 my %log_entry = ( parents => $parents || [], revision => $rev,
1236 log => '');
1237 my $rp = $self->ra->rev_proplist($rev);
1238 foreach (sort keys %$rp) {
1239 my $v = $rp->{$_};
1240 if (/^svn:(author|date|log)$/) {
1241 $log_entry{$1} = $v;
1242 } else {
1243 print $un " rev_prop: ", uri_encode($_), ' ',
1244 uri_encode($v), "\n";
1245 }
1246 }
1247 close $un or croak $!;
1248
1249 $log_entry{date} = parse_svn_date($log_entry{date});
1250 $log_entry{author} = check_author($log_entry{author});
1251 $log_entry{log} .= "\n";
1252 \%log_entry;
1253 }
1254
1255 sub fetch {
1256 my ($self, $min_rev, $max_rev, @parents) = @_;
1257 my ($last_rev, $last_commit) = $self->last_rev_commit;
1258 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1259 return if ($base > $head);
1260 $self->ra->gs_fetch_loop_common($base, $head, $self);
1261 }
1262
1263 sub set_tree_cb {
1264 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1265 # TODO: enable and test optimized commits:
1266 if (0 && $rev == ($self->{last_rev} + 1)) {
1267 $log_entry->{revision} = $rev;
1268 $log_entry->{author} = $author;
1269 $self->do_git_commit($log_entry, "$rev=$tree");
1270 } else {
1271 $self->{inject_parents} = { $rev => $tree };
1272 $self->fetch(undef, undef);
1273 }
1274 }
1275
1276 sub set_tree {
1277 my ($self, $tree) = (shift, shift);
1278 my $log_entry = ::get_commit_entry($tree);
1279 unless ($self->{last_rev}) {
1280 fatal("Must have an existing revision to commit\n");
1281 }
1282 my %ed_opts = ( r => $self->{last_rev},
1283 log => $log_entry->{log},
1284 ra => $self->ra,
1285 tree_a => $self->{last_commit},
1286 tree_b => $tree,
1287 editor_cb => sub {
1288 $self->set_tree_cb($log_entry, $tree, @_) },
1289 svn_path => $self->{path} );
1290 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1291 print "No changes\nr$self->{last_rev} = $tree\n";
1292 }
1293 }
1294
1295 sub rebuild {
1296 my ($self) = @_;
1297 print "Rebuilding $self->{db_path} ...\n";
1298 my ($rev_list, $ctx) = command_output_pipe("rev-list", $self->refname);
1299 my $latest;
1300 my $full_url = $self->full_url;
1301 my $svn_uuid;
1302 while (<$rev_list>) {
1303 chomp;
1304 my $c = $_;
1305 die "Non-SHA1: $c\n" unless $c =~ /^$::sha1$/o;
1306 my ($url, $rev, $uuid) = ::cmt_metadata($c);
1307
1308 # ignore merges (from set-tree)
1309 next if (!defined $rev || !$uuid);
1310
1311 # if we merged or otherwise started elsewhere, this is
1312 # how we break out of it
1313 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
1314 ($full_url && $url && ($url ne $full_url))) {
1315 next;
1316 }
1317 $latest ||= $rev;
1318 $svn_uuid ||= $uuid;
1319
1320 $self->rev_db_set($rev, $c);
1321 print "r$rev = $c\n";
1322 }
1323 command_close_pipe($rev_list, $ctx);
1324 print "Done rebuilding $self->{db_path}\n";
1325 }
1326
1327 # rev_db:
1328 # Tie::File seems to be prone to offset errors if revisions get sparse,
1329 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
1330 # one of my favorite modules is out :< Next up would be one of the DBM
1331 # modules, but I'm not sure which is most portable... So I'll just
1332 # go with something that's plain-text, but still capable of
1333 # being randomly accessed. So here's my ultra-simple fixed-width
1334 # database. All records are 40 characters + "\n", so it's easy to seek
1335 # to a revision: (41 * rev) is the byte offset.
1336 # A record of 40 0s denotes an empty revision.
1337 # And yes, it's still pretty fast (faster than Tie::File).
1338
1339 sub rev_db_set {
1340 my ($self, $rev, $commit) = @_;
1341 length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
1342 open my $fh, '+<', $self->{db_path} or croak $!;
1343 my $offset = $rev * 41;
1344 # assume that append is the common case:
1345 seek $fh, 0, 2 or croak $!;
1346 my $pos = tell $fh;
1347 if ($pos < $offset) {
1348 print $fh (('0' x 40),"\n") x (($offset - $pos) / 41)
1349 or croak $!;
1350 }
1351 seek $fh, $offset, 0 or croak $!;
1352 print $fh $commit,"\n" or croak $!;
1353 close $fh or croak $!;
1354 }
1355
1356 sub rev_db_get {
1357 my ($self, $rev) = @_;
1358 my $ret;
1359 my $offset = $rev * 41;
1360 open my $fh, '<', $self->{db_path} or croak $!;
1361 if (seek $fh, $offset, 0) {
1362 $ret = readline $fh;
1363 if (defined $ret) {
1364 chomp $ret;
1365 $ret = undef if ($ret =~ /^0{40}$/);
1366 }
1367 }
1368 close $fh or croak $!;
1369 $ret;
1370 }
1371
1372 sub find_rev_before {
1373 my ($self, $rev, $eq_ok) = @_;
1374 --$rev unless $eq_ok;
1375 while ($rev > 0) {
1376 if (my $c = $self->rev_db_get($rev)) {
1377 return ($rev, $c);
1378 }
1379 --$rev;
1380 }
1381 return (undef, undef);
1382 }
1383
1384 sub _new {
1385 my ($class, $repo_id, $ref_id, $path) = @_;
1386 unless (defined $repo_id && length $repo_id) {
1387 $repo_id = $Git::SVN::default_repo_id;
1388 }
1389 unless (defined $ref_id && length $ref_id) {
1390 $_[2] = $ref_id = $Git::SVN::default_ref_id;
1391 }
1392 $_[1] = $repo_id = sanitize_remote_name($repo_id);
1393 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1394 $_[3] = $path = '' unless (defined $path);
1395 mkpath([$dir]);
1396 unless (-f "$dir/.rev_db") {
1397 open my $fh, '>>', "$dir/.rev_db" or croak $!;
1398 close $fh or croak $!;
1399 }
1400 bless { ref_id => $ref_id, dir => $dir, index => "$dir/index",
1401 path => $path,
1402 db_path => "$dir/.rev_db", repo_id => $repo_id }, $class;
1403 }
1404
1405 sub uri_encode {
1406 my ($f) = @_;
1407 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1408 $f
1409 }
1410
1411 package Git::SVN::Prompt;
1412 use strict;
1413 use warnings;
1414 require SVN::Core;
1415 use vars qw/$_no_auth_cache $_username/;
1416
1417 sub simple {
1418 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1419 $may_save = undef if $_no_auth_cache;
1420 $default_username = $_username if defined $_username;
1421 if (defined $default_username && length $default_username) {
1422 if (defined $realm && length $realm) {
1423 print STDERR "Authentication realm: $realm\n";
1424 STDERR->flush;
1425 }
1426 $cred->username($default_username);
1427 } else {
1428 username($cred, $realm, $may_save, $pool);
1429 }
1430 $cred->password(_read_password("Password for '" .
1431 $cred->username . "': ", $realm));
1432 $cred->may_save($may_save);
1433 $SVN::_Core::SVN_NO_ERROR;
1434 }
1435
1436 sub ssl_server_trust {
1437 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1438 $may_save = undef if $_no_auth_cache;
1439 print STDERR "Error validating server certificate for '$realm':\n";
1440 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1441 print STDERR " - The certificate is not issued by a trusted ",
1442 "authority. Use the\n",
1443 " fingerprint to validate the certificate manually!\n";
1444 }
1445 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1446 print STDERR " - The certificate hostname does not match.\n";
1447 }
1448 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1449 print STDERR " - The certificate is not yet valid.\n";
1450 }
1451 if ($failures & $SVN::Auth::SSL::EXPIRED) {
1452 print STDERR " - The certificate has expired.\n";
1453 }
1454 if ($failures & $SVN::Auth::SSL::OTHER) {
1455 print STDERR " - The certificate has an unknown error.\n";
1456 }
1457 printf STDERR
1458 "Certificate information:\n".
1459 " - Hostname: %s\n".
1460 " - Valid: from %s until %s\n".
1461 " - Issuer: %s\n".
1462 " - Fingerprint: %s\n",
1463 map $cert_info->$_, qw(hostname valid_from valid_until
1464 issuer_dname fingerprint);
1465 my $choice;
1466 prompt:
1467 print STDERR $may_save ?
1468 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1469 "(R)eject or accept (t)emporarily? ";
1470 STDERR->flush;
1471 $choice = lc(substr(<STDIN> || 'R', 0, 1));
1472 if ($choice =~ /^t$/i) {
1473 $cred->may_save(undef);
1474 } elsif ($choice =~ /^r$/i) {
1475 return -1;
1476 } elsif ($may_save && $choice =~ /^p$/i) {
1477 $cred->may_save($may_save);
1478 } else {
1479 goto prompt;
1480 }
1481 $cred->accepted_failures($failures);
1482 $SVN::_Core::SVN_NO_ERROR;
1483 }
1484
1485 sub ssl_client_cert {
1486 my ($cred, $realm, $may_save, $pool) = @_;
1487 $may_save = undef if $_no_auth_cache;
1488 print STDERR "Client certificate filename: ";
1489 STDERR->flush;
1490 chomp(my $filename = <STDIN>);
1491 $cred->cert_file($filename);
1492 $cred->may_save($may_save);
1493 $SVN::_Core::SVN_NO_ERROR;
1494 }
1495
1496 sub ssl_client_cert_pw {
1497 my ($cred, $realm, $may_save, $pool) = @_;
1498 $may_save = undef if $_no_auth_cache;
1499 $cred->password(_read_password("Password: ", $realm));
1500 $cred->may_save($may_save);
1501 $SVN::_Core::SVN_NO_ERROR;
1502 }
1503
1504 sub username {
1505 my ($cred, $realm, $may_save, $pool) = @_;
1506 $may_save = undef if $_no_auth_cache;
1507 if (defined $realm && length $realm) {
1508 print STDERR "Authentication realm: $realm\n";
1509 }
1510 my $username;
1511 if (defined $_username) {
1512 $username = $_username;
1513 } else {
1514 print STDERR "Username: ";
1515 STDERR->flush;
1516 chomp($username = <STDIN>);
1517 }
1518 $cred->username($username);
1519 $cred->may_save($may_save);
1520 $SVN::_Core::SVN_NO_ERROR;
1521 }
1522
1523 sub _read_password {
1524 my ($prompt, $realm) = @_;
1525 print STDERR $prompt;
1526 STDERR->flush;
1527 require Term::ReadKey;
1528 Term::ReadKey::ReadMode('noecho');
1529 my $password = '';
1530 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1531 last if $key =~ /[\012\015]/; # \n\r
1532 $password .= $key;
1533 }
1534 Term::ReadKey::ReadMode('restore');
1535 print STDERR "\n";
1536 STDERR->flush;
1537 $password;
1538 }
1539
1540 package main;
1541
1542 {
1543 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1544 $SVN::Node::dir.$SVN::Node::unknown.
1545 $SVN::Node::none.$SVN::Node::file.
1546 $SVN::Node::dir.$SVN::Node::unknown.
1547 $SVN::Auth::SSL::CNMISMATCH.
1548 $SVN::Auth::SSL::NOTYETVALID.
1549 $SVN::Auth::SSL::EXPIRED.
1550 $SVN::Auth::SSL::UNKNOWNCA.
1551 $SVN::Auth::SSL::OTHER;
1552 }
1553
1554 package SVN::Git::Fetcher;
1555 use vars qw/@ISA/;
1556 use strict;
1557 use warnings;
1558 use Carp qw/croak/;
1559 use IO::File qw//;
1560 use Digest::MD5;
1561
1562 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
1563 sub new {
1564 my ($class, $git_svn) = @_;
1565 my $self = SVN::Delta::Editor->new;
1566 bless $self, $class;
1567 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
1568 $self->{empty} = {};
1569 $self->{dir_prop} = {};
1570 $self->{file_prop} = {};
1571 $self->{absent_dir} = {};
1572 $self->{absent_file} = {};
1573 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
1574 $self;
1575 }
1576
1577 sub set_path_strip {
1578 my ($self, $path) = @_;
1579 $self->{path_strip} = qr/^\Q$path\E\/?/;
1580 }
1581
1582 sub open_root {
1583 { path => '' };
1584 }
1585
1586 sub open_directory {
1587 my ($self, $path, $pb, $rev) = @_;
1588 { path => $path };
1589 }
1590
1591 sub git_path {
1592 my ($self, $path) = @_;
1593 if ($self->{path_strip}) {
1594 $path =~ s!$self->{path_strip}!! or
1595 die "Failed to strip path '$path' ($self->{path_strip})\n";
1596 }
1597 $path;
1598 }
1599
1600 sub delete_entry {
1601 my ($self, $path, $rev, $pb) = @_;
1602
1603 my $gpath = $self->git_path($path);
1604 return undef if ($gpath eq '');
1605
1606 # remove entire directories.
1607 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
1608 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
1609 -r --name-only -z/,
1610 $self->{c}, '--', $gpath);
1611 local $/ = "\0";
1612 while (<$ls>) {
1613 chomp;
1614 $self->{gii}->remove($_);
1615 print "\tD\t$_\n" unless $self->{q};
1616 }
1617 print "\tD\t$gpath/\n" unless $self->{q};
1618 command_close_pipe($ls, $ctx);
1619 $self->{empty}->{$path} = 0
1620 } else {
1621 $self->{gii}->remove($gpath);
1622 print "\tD\t$gpath\n" unless $self->{q};
1623 }
1624 undef;
1625 }
1626
1627 sub open_file {
1628 my ($self, $path, $pb, $rev) = @_;
1629 my $gpath = $self->git_path($path);
1630 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
1631 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
1632 unless (defined $mode && defined $blob) {
1633 die "$path was not found in commit $self->{c} (r$rev)\n";
1634 }
1635 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
1636 pool => SVN::Pool->new, action => 'M' };
1637 }
1638
1639 sub add_file {
1640 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
1641 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1642 delete $self->{empty}->{$dir};
1643 { path => $path, mode_a => 100644, mode_b => 100644,
1644 pool => SVN::Pool->new, action => 'A' };
1645 }
1646
1647 sub add_directory {
1648 my ($self, $path, $cp_path, $cp_rev) = @_;
1649 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1650 delete $self->{empty}->{$dir};
1651 $self->{empty}->{$path} = 1;
1652 { path => $path };
1653 }
1654
1655 sub change_dir_prop {
1656 my ($self, $db, $prop, $value) = @_;
1657 $self->{dir_prop}->{$db->{path}} ||= {};
1658 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
1659 undef;
1660 }
1661
1662 sub absent_directory {
1663 my ($self, $path, $pb) = @_;
1664 $self->{absent_dir}->{$pb->{path}} ||= [];
1665 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
1666 undef;
1667 }
1668
1669 sub absent_file {
1670 my ($self, $path, $pb) = @_;
1671 $self->{absent_file}->{$pb->{path}} ||= [];
1672 push @{$self->{absent_file}->{$pb->{path}}}, $path;
1673 undef;
1674 }
1675
1676 sub change_file_prop {
1677 my ($self, $fb, $prop, $value) = @_;
1678 if ($prop eq 'svn:executable') {
1679 if ($fb->{mode_b} != 120000) {
1680 $fb->{mode_b} = defined $value ? 100755 : 100644;
1681 }
1682 } elsif ($prop eq 'svn:special') {
1683 $fb->{mode_b} = defined $value ? 120000 : 100644;
1684 } else {
1685 $self->{file_prop}->{$fb->{path}} ||= {};
1686 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
1687 }
1688 undef;
1689 }
1690
1691 sub apply_textdelta {
1692 my ($self, $fb, $exp) = @_;
1693 my $fh = IO::File->new_tmpfile;
1694 $fh->autoflush(1);
1695 # $fh gets auto-closed() by SVN::TxDelta::apply(),
1696 # (but $base does not,) so dup() it for reading in close_file
1697 open my $dup, '<&', $fh or croak $!;
1698 my $base = IO::File->new_tmpfile;
1699 $base->autoflush(1);
1700 if ($fb->{blob}) {
1701 defined (my $pid = fork) or croak $!;
1702 if (!$pid) {
1703 open STDOUT, '>&', $base or croak $!;
1704 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
1705 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
1706 }
1707 waitpid $pid, 0;
1708 croak $? if $?;
1709
1710 if (defined $exp) {
1711 seek $base, 0, 0 or croak $!;
1712 my $md5 = Digest::MD5->new;
1713 $md5->addfile($base);
1714 my $got = $md5->hexdigest;
1715 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
1716 "expected: $exp\n",
1717 " got: $got\n" if ($got ne $exp);
1718 }
1719 }
1720 seek $base, 0, 0 or croak $!;
1721 $fb->{fh} = $dup;
1722 $fb->{base} = $base;
1723 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
1724 }
1725
1726 sub close_file {
1727 my ($self, $fb, $exp) = @_;
1728 my $hash;
1729 my $path = $self->git_path($fb->{path});
1730 if (my $fh = $fb->{fh}) {
1731 seek($fh, 0, 0) or croak $!;
1732 my $md5 = Digest::MD5->new;
1733 $md5->addfile($fh);
1734 my $got = $md5->hexdigest;
1735 die "Checksum mismatch: $path\n",
1736 "expected: $exp\n got: $got\n" if ($got ne $exp);
1737 seek($fh, 0, 0) or croak $!;
1738 if ($fb->{mode_b} == 120000) {
1739 read($fh, my $buf, 5) == 5 or croak $!;
1740 $buf eq 'link ' or die "$path has mode 120000",
1741 "but is not a link\n";
1742 }
1743 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
1744 if (!$pid) {
1745 open STDIN, '<&', $fh or croak $!;
1746 exec qw/git-hash-object -w --stdin/ or croak $!;
1747 }
1748 chomp($hash = do { local $/; <$out> });
1749 close $out or croak $!;
1750 close $fh or croak $!;
1751 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
1752 close $fb->{base} or croak $!;
1753 } else {
1754 $hash = $fb->{blob} or die "no blob information\n";
1755 }
1756 $fb->{pool}->clear;
1757 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
1758 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
1759 undef;
1760 }
1761
1762 sub abort_edit {
1763 my $self = shift;
1764 $self->{nr} = $self->{gii}->{nr};
1765 delete $self->{gii};
1766 $self->SUPER::abort_edit(@_);
1767 }
1768
1769 sub close_edit {
1770 my $self = shift;
1771 $self->{git_commit_ok} = 1;
1772 $self->{nr} = $self->{gii}->{nr};
1773 delete $self->{gii};
1774 $self->SUPER::close_edit(@_);
1775 }
1776
1777 package SVN::Git::Editor;
1778 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
1779 use strict;
1780 use warnings;
1781 use Carp qw/croak/;
1782 use IO::File;
1783 use Digest::MD5;
1784
1785 sub new {
1786 my ($class, $opts) = @_;
1787 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
1788 die "$_ required!\n" unless (defined $opts->{$_});
1789 }
1790
1791 my $pool = SVN::Pool->new;
1792 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
1793 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
1794 $opts->{r}, $mods);
1795
1796 # $opts->{ra} functions should not be used after this:
1797 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
1798 $opts->{editor_cb}, $pool);
1799 my $self = SVN::Delta::Editor->new(@ce, $pool);
1800 bless $self, $class;
1801 foreach (qw/svn_path r tree_a tree_b/) {
1802 $self->{$_} = $opts->{$_};
1803 }
1804 $self->{url} = $opts->{ra}->{url};
1805 $self->{mods} = $mods;
1806 $self->{types} = $types;
1807 $self->{pool} = $pool;
1808 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
1809 $self->{rm} = { };
1810 $self->{path_prefix} = length $self->{svn_path} ?
1811 "$self->{svn_path}/" : '';
1812 return $self;
1813 }
1814
1815 sub generate_diff {
1816 my ($tree_a, $tree_b) = @_;
1817 my @diff_tree = qw(diff-tree -z -r);
1818 if ($_cp_similarity) {
1819 push @diff_tree, "-C$_cp_similarity";
1820 } else {
1821 push @diff_tree, '-C';
1822 }
1823 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1824 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
1825 push @diff_tree, $tree_a, $tree_b;
1826 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
1827 local $/ = "\0";
1828 my $state = 'meta';
1829 my @mods;
1830 while (<$diff_fh>) {
1831 chomp $_; # this gets rid of the trailing "\0"
1832 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1833 $::sha1\s($::sha1)\s
1834 ([MTCRAD])\d*$/xo) {
1835 push @mods, { mode_a => $1, mode_b => $2,
1836 sha1_b => $3, chg => $4 };
1837 if ($4 =~ /^(?:C|R)$/) {
1838 $state = 'file_a';
1839 } else {
1840 $state = 'file_b';
1841 }
1842 } elsif ($state eq 'file_a') {
1843 my $x = $mods[$#mods] or croak "Empty array\n";
1844 if ($x->{chg} !~ /^(?:C|R)$/) {
1845 croak "Error parsing $_, $x->{chg}\n";
1846 }
1847 $x->{file_a} = $_;
1848 $state = 'file_b';
1849 } elsif ($state eq 'file_b') {
1850 my $x = $mods[$#mods] or croak "Empty array\n";
1851 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1852 croak "Error parsing $_, $x->{chg}\n";
1853 }
1854 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1855 croak "Error parsing $_, $x->{chg}\n";
1856 }
1857 $x->{file_b} = $_;
1858 $state = 'meta';
1859 } else {
1860 croak "Error parsing $_\n";
1861 }
1862 }
1863 command_close_pipe($diff_fh, $ctx);
1864 \@mods;
1865 }
1866
1867 sub check_diff_paths {
1868 my ($ra, $pfx, $rev, $mods) = @_;
1869 my %types;
1870 $pfx .= '/' if length $pfx;
1871
1872 sub type_diff_paths {
1873 my ($ra, $types, $path, $rev) = @_;
1874 my @p = split m#/+#, $path;
1875 my $c = shift @p;
1876 unless (defined $types->{$c}) {
1877 $types->{$c} = $ra->check_path($c, $rev);
1878 }
1879 while (@p) {
1880 $c .= '/' . shift @p;
1881 next if defined $types->{$c};
1882 $types->{$c} = $ra->check_path($c, $rev);
1883 }
1884 }
1885
1886 foreach my $m (@$mods) {
1887 foreach my $f (qw/file_a file_b/) {
1888 next unless defined $m->{$f};
1889 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
1890 if (length $pfx.$dir && ! defined $types{$dir}) {
1891 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
1892 }
1893 }
1894 }
1895 \%types;
1896 }
1897
1898 sub split_path {
1899 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
1900 }
1901
1902 sub repo_path {
1903 my ($self, $path) = @_;
1904 $self->{path_prefix}.(defined $path ? $path : '');
1905 }
1906
1907 sub url_path {
1908 my ($self, $path) = @_;
1909 $self->{url} . '/' . $self->repo_path($path);
1910 }
1911
1912 sub rmdirs {
1913 my ($self) = @_;
1914 my $rm = $self->{rm};
1915 delete $rm->{''}; # we never delete the url we're tracking
1916 return unless %$rm;
1917
1918 foreach (keys %$rm) {
1919 my @d = split m#/#, $_;
1920 my $c = shift @d;
1921 $rm->{$c} = 1;
1922 while (@d) {
1923 $c .= '/' . shift @d;
1924 $rm->{$c} = 1;
1925 }
1926 }
1927 delete $rm->{$self->{svn_path}};
1928 delete $rm->{''}; # we never delete the url we're tracking
1929 return unless %$rm;
1930
1931 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
1932 $self->{tree_b});
1933 local $/ = "\0";
1934 while (<$fh>) {
1935 chomp;
1936 my @dn = split m#/#, $_;
1937 while (pop @dn) {
1938 delete $rm->{join '/', @dn};
1939 }
1940 unless (%$rm) {
1941 close $fh;
1942 return;
1943 }
1944 }
1945 command_close_pipe($fh, $ctx);
1946
1947 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
1948 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
1949 $self->close_directory($bat->{$d}, $p);
1950 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
1951 print "\tD+\t$d/\n" unless $::_q;
1952 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
1953 delete $bat->{$d};
1954 }
1955 }
1956
1957 sub open_or_add_dir {
1958 my ($self, $full_path, $baton) = @_;
1959 my $t = $self->{types}->{$full_path};
1960 if (!defined $t) {
1961 die "$full_path not known in r$self->{r} or we have a bug!\n";
1962 }
1963 if ($t == $SVN::Node::none) {
1964 return $self->add_directory($full_path, $baton,
1965 undef, -1, $self->{pool});
1966 } elsif ($t == $SVN::Node::dir) {
1967 return $self->open_directory($full_path, $baton,
1968 $self->{r}, $self->{pool});
1969 }
1970 print STDERR "$full_path already exists in repository at ",
1971 "r$self->{r} and it is not a directory (",
1972 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
1973 exit 1;
1974 }
1975
1976 sub ensure_path {
1977 my ($self, $path) = @_;
1978 my $bat = $self->{bat};
1979 my $repo_path = $self->repo_path($path);
1980 return $bat->{''} unless (length $repo_path);
1981 my @p = split m#/+#, $repo_path;
1982 my $c = shift @p;
1983 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
1984 while (@p) {
1985 my $c0 = $c;
1986 $c .= '/' . shift @p;
1987 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
1988 }
1989 return $bat->{$c};
1990 }
1991
1992 sub A {
1993 my ($self, $m) = @_;
1994 my ($dir, $file) = split_path($m->{file_b});
1995 my $pbat = $self->ensure_path($dir);
1996 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
1997 undef, -1);
1998 print "\tA\t$m->{file_b}\n" unless $::_q;
1999 $self->chg_file($fbat, $m);
2000 $self->close_file($fbat,undef,$self->{pool});
2001 }
2002
2003 sub C {
2004 my ($self, $m) = @_;
2005 my ($dir, $file) = split_path($m->{file_b});
2006 my $pbat = $self->ensure_path($dir);
2007 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2008 $self->url_path($m->{file_a}), $self->{r});
2009 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2010 $self->chg_file($fbat, $m);
2011 $self->close_file($fbat,undef,$self->{pool});
2012 }
2013
2014 sub delete_entry {
2015 my ($self, $path, $pbat) = @_;
2016 my $rpath = $self->repo_path($path);
2017 my ($dir, $file) = split_path($rpath);
2018 $self->{rm}->{$dir} = 1;
2019 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2020 }
2021
2022 sub R {
2023 my ($self, $m) = @_;
2024 my ($dir, $file) = split_path($m->{file_b});
2025 my $pbat = $self->ensure_path($dir);
2026 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2027 $self->url_path($m->{file_a}), $self->{r});
2028 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2029 $self->chg_file($fbat, $m);
2030 $self->close_file($fbat,undef,$self->{pool});
2031
2032 ($dir, $file) = split_path($m->{file_a});
2033 $pbat = $self->ensure_path($dir);
2034 $self->delete_entry($m->{file_a}, $pbat);
2035 }
2036
2037 sub M {
2038 my ($self, $m) = @_;
2039 my ($dir, $file) = split_path($m->{file_b});
2040 my $pbat = $self->ensure_path($dir);
2041 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2042 $pbat,$self->{r},$self->{pool});
2043 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2044 $self->chg_file($fbat, $m);
2045 $self->close_file($fbat,undef,$self->{pool});
2046 }
2047
2048 sub T { shift->M(@_) }
2049
2050 sub change_file_prop {
2051 my ($self, $fbat, $pname, $pval) = @_;
2052 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2053 }
2054
2055 sub chg_file {
2056 my ($self, $fbat, $m) = @_;
2057 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2058 $self->change_file_prop($fbat,'svn:executable','*');
2059 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2060 $self->change_file_prop($fbat,'svn:executable',undef);
2061 }
2062 my $fh = IO::File->new_tmpfile or croak $!;
2063 if ($m->{mode_b} =~ /^120/) {
2064 print $fh 'link ' or croak $!;
2065 $self->change_file_prop($fbat,'svn:special','*');
2066 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2067 $self->change_file_prop($fbat,'svn:special',undef);
2068 }
2069 defined(my $pid = fork) or croak $!;
2070 if (!$pid) {
2071 open STDOUT, '>&', $fh or croak $!;
2072 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2073 }
2074 waitpid $pid, 0;
2075 croak $? if $?;
2076 $fh->flush == 0 or croak $!;
2077 seek $fh, 0, 0 or croak $!;
2078
2079 my $md5 = Digest::MD5->new;
2080 $md5->addfile($fh) or croak $!;
2081 seek $fh, 0, 0 or croak $!;
2082
2083 my $exp = $md5->hexdigest;
2084 my $pool = SVN::Pool->new;
2085 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2086 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2087 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2088 $pool->clear;
2089
2090 close $fh or croak $!;
2091 }
2092
2093 sub D {
2094 my ($self, $m) = @_;
2095 my ($dir, $file) = split_path($m->{file_b});
2096 my $pbat = $self->ensure_path($dir);
2097 print "\tD\t$m->{file_b}\n" unless $::_q;
2098 $self->delete_entry($m->{file_b}, $pbat);
2099 }
2100
2101 sub close_edit {
2102 my ($self) = @_;
2103 my ($p,$bat) = ($self->{pool}, $self->{bat});
2104 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2105 $self->close_directory($bat->{$_}, $p);
2106 }
2107 $self->SUPER::close_edit($p);
2108 $p->clear;
2109 }
2110
2111 sub abort_edit {
2112 my ($self) = @_;
2113 $self->SUPER::abort_edit($self->{pool});
2114 }
2115
2116 sub DESTROY {
2117 my $self = shift;
2118 $self->SUPER::DESTROY(@_);
2119 $self->{pool}->clear;
2120 }
2121
2122 # this drives the editor
2123 sub apply_diff {
2124 my ($self) = @_;
2125 my $mods = $self->{mods};
2126 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2127 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2128 my $f = $m->{chg};
2129 if (defined $o{$f}) {
2130 $self->$f($m);
2131 } else {
2132 fatal("Invalid change type: $f\n");
2133 }
2134 }
2135 $self->rmdirs if $_rmdir;
2136 if (@$mods == 0) {
2137 $self->abort_edit;
2138 } else {
2139 $self->close_edit;
2140 }
2141 return scalar @$mods;
2142 }
2143
2144 package Git::SVN::Ra;
2145 use vars qw/@ISA $config_dir/;
2146 use strict;
2147 use warnings;
2148 my ($can_do_switch);
2149 my $RA;
2150
2151 BEGIN {
2152 # enforce temporary pool usage for some simple functions
2153 my $e;
2154 foreach (qw/get_latest_revnum rev_proplist get_file
2155 check_path get_dir get_uuid get_repos_root/) {
2156 $e .= "sub $_ {
2157 my \$self = shift;
2158 my \$pool = SVN::Pool->new;
2159 my \@ret = \$self->SUPER::$_(\@_,\$pool);
2160 \$pool->clear;
2161 wantarray ? \@ret : \$ret[0]; }\n";
2162 }
2163 eval $e;
2164 }
2165
2166 sub new {
2167 my ($class, $url) = @_;
2168 $url =~ s!/+$!!;
2169 return $RA if ($RA && $RA->{url} eq $url);
2170
2171 SVN::_Core::svn_config_ensure($config_dir, undef);
2172 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2173 SVN::Client::get_simple_provider(),
2174 SVN::Client::get_ssl_server_trust_file_provider(),
2175 SVN::Client::get_simple_prompt_provider(
2176 \&Git::SVN::Prompt::simple, 2),
2177 SVN::Client::get_ssl_client_cert_prompt_provider(
2178 \&Git::SVN::Prompt::ssl_client_cert, 2),
2179 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2180 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2181 SVN::Client::get_username_provider(),
2182 SVN::Client::get_ssl_server_trust_prompt_provider(
2183 \&Git::SVN::Prompt::ssl_server_trust),
2184 SVN::Client::get_username_prompt_provider(
2185 \&Git::SVN::Prompt::username, 2),
2186 ]);
2187 my $config = SVN::Core::config_get_config($config_dir);
2188 my $self = SVN::Ra->new(url => $url, auth => $baton,
2189 config => $config,
2190 pool => SVN::Pool->new,
2191 auth_provider_callbacks => $callbacks);
2192 $self->{svn_path} = $url;
2193 $self->{repos_root} = $self->get_repos_root;
2194 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E/*##;
2195 $RA = bless $self, $class;
2196 }
2197
2198 sub DESTROY {
2199 # do not call the real DESTROY since we store ourselves in $RA
2200 }
2201
2202 sub get_log {
2203 my ($self, @args) = @_;
2204 my $pool = SVN::Pool->new;
2205 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2206 my $ret = $self->SUPER::get_log(@args, $pool);
2207 $pool->clear;
2208 $ret;
2209 }
2210
2211 sub get_commit_editor {
2212 my ($self, $log, $cb, $pool) = @_;
2213 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2214 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2215 }
2216
2217 sub uuid {
2218 my ($self) = @_;
2219 $self->{uuid} ||= $self->get_uuid;
2220 }
2221
2222 sub gs_do_update {
2223 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
2224 my $new = ($rev_a == $rev_b);
2225 my $path = $gs->{path};
2226
2227 my $ta = $self->check_path($path, $rev_a);
2228 my $tb = $new ? $ta : $self->check_path($path, $rev_b);
2229 return 1 if ($tb != $SVN::Node::dir && $ta != $SVN::Node::dir);
2230 if ($ta == $SVN::Node::none) {
2231 $rev_a = $rev_b;
2232 $new = 1;
2233 }
2234
2235 my $pool = SVN::Pool->new;
2236 $editor->set_path_strip($path);
2237 my (@pc) = split m#/#, $path;
2238 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
2239 1, $editor, $pool);
2240 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2241
2242 # Since we can't rely on svn_ra_reparent being available, we'll
2243 # just have to do some magic with set_path to make it so
2244 # we only want a partial path.
2245 my $sp = '';
2246 my $final = join('/', @pc);
2247 while (@pc) {
2248 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
2249 $sp .= '/' if length $sp;
2250 $sp .= shift @pc;
2251 }
2252 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
2253
2254 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
2255
2256 $reporter->finish_report($pool);
2257 $pool->clear;
2258 $editor->{git_commit_ok};
2259 }
2260
2261 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
2262 # svn_ra_reparent didn't work before 1.4)
2263 sub gs_do_switch {
2264 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
2265 my $path = $gs->{path};
2266 my $pool = SVN::Pool->new;
2267
2268 my $full_url = $self->{url};
2269 my $old_url = $full_url;
2270 $full_url .= "/$path" if length $path;
2271 my ($ra, $reparented);
2272 if ($old_url ne $full_url) {
2273 if ($old_url !~ m#^svn(\+ssh)?://#) {
2274 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
2275 $pool);
2276 $self->{url} = $full_url;
2277 $reparented = 1;
2278 } else {
2279 $ra = Git::SVN::Ra->new($full_url);
2280 }
2281 }
2282 $ra ||= $self;
2283 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
2284 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2285 $reporter->set_path('', $rev_a, 0, @lock, $pool);
2286 $reporter->finish_report($pool);
2287
2288 if ($reparented) {
2289 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
2290 $self->{url} = $old_url;
2291 }
2292
2293 $pool->clear;
2294 $editor->{git_commit_ok};
2295 }
2296
2297 sub gs_fetch_loop_common {
2298 my ($self, $base, $head, @gs) = @_;
2299 my $inc = 1000;
2300 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
2301 foreach my $gs (@gs) {
2302 if (my $last_commit = $gs->last_commit) {
2303 $gs->assert_index_clean($last_commit);
2304 }
2305 }
2306 while (1) {
2307 my %revs;
2308 my $err;
2309 my $err_handler = $SVN::Error::handler;
2310 $SVN::Error::handler = sub {
2311 ($err) = @_;
2312 skip_unknown_revs($err);
2313 };
2314 foreach my $gs (@gs) {
2315 $self->get_log([$gs->{path}], $min, $max, 0, 1, 1, sub
2316 { my ($paths, $rev) = @_;
2317 push @{$revs{$rev}},
2318 [ $gs,
2319 dup_changed_paths($paths) ] });
2320
2321 next unless ($err && $max >= $head);
2322
2323 print STDERR "Path '$gs->{path}' ",
2324 "was probably deleted:\n",
2325 $err->expanded_message,
2326 "\nWill attempt to follow ",
2327 "revisions r$min .. r$max ",
2328 "committed before the deletion\n";
2329 my $hi = $max;
2330 while (--$hi >= $min) {
2331 my $ok;
2332 $self->get_log([$gs->{path}], $min, $hi,
2333 0, 1, 1, sub {
2334 my ($paths, $rev) = @_;
2335 $ok = $rev;
2336 push @{$revs{$rev}}, [ $gs,
2337 dup_changed_paths($_[0])]});
2338 if ($ok) {
2339 print STDERR "r$min .. r$ok OK\n";
2340 last;
2341 }
2342 }
2343 }
2344 $SVN::Error::handler = $err_handler;
2345 foreach my $r (sort {$a <=> $b} keys %revs) {
2346 foreach (@{$revs{$r}}) {
2347 my ($gs, $paths) = @$_;
2348 my $lr = $gs->last_rev;
2349 next if defined $lr && $lr >= $r;
2350 next if defined $gs->rev_db_get($r);
2351 if (my $log_entry = $gs->do_fetch($paths, $r)) {
2352 $gs->do_git_commit($log_entry);
2353 }
2354 }
2355 }
2356 last if $max >= $head;
2357 $min = $max + 1;
2358 $max += $inc;
2359 $max = $head if ($max > $head);
2360 }
2361 }
2362
2363 sub minimize_url {
2364 my ($self) = @_;
2365 return $self->{url} if ($self->{url} eq $self->{repos_root});
2366 my $url = $self->{repos_root};
2367 my @components = split(m!/!, $self->{svn_path});
2368 my $c = '';
2369 do {
2370 $url .= "/$c" if length $c;
2371 eval { (ref $self)->new($url)->get_latest_revnum };
2372 } while ($@ && ($c = shift @components));
2373 $url;
2374 }
2375
2376 sub can_do_switch {
2377 my $self = shift;
2378 unless (defined $can_do_switch) {
2379 my $pool = SVN::Pool->new;
2380 my $rep = eval {
2381 $self->do_switch(1, '', 0, $self->{url},
2382 SVN::Delta::Editor->new, $pool);
2383 };
2384 if ($@) {
2385 $can_do_switch = 0;
2386 } else {
2387 $rep->abort_report($pool);
2388 $can_do_switch = 1;
2389 }
2390 $pool->clear;
2391 }
2392 $can_do_switch;
2393 }
2394
2395 sub skip_unknown_revs {
2396 my ($err) = @_;
2397 my $errno = $err->apr_err();
2398 # Maybe the branch we're tracking didn't
2399 # exist when the repo started, so it's
2400 # not an error if it doesn't, just continue
2401 #
2402 # Wonderfully consistent library, eh?
2403 # 160013 - svn:// and file://
2404 # 175002 - http(s)://
2405 # 175007 - http(s):// (this repo required authorization, too...)
2406 # More codes may be discovered later...
2407 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
2408 return;
2409 }
2410 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2411 }
2412
2413 # svn_log_changed_path_t objects passed to get_log are likely to be
2414 # overwritten even if only the refs are copied to an external variable,
2415 # so we should dup the structures in their entirety. Using an externally
2416 # passed pool (instead of our temporary and quickly cleared pool in
2417 # Git::SVN::Ra) does not help matters at all...
2418 sub dup_changed_paths {
2419 my ($paths) = @_;
2420 return undef unless $paths;
2421 my %ret;
2422 foreach my $p (keys %$paths) {
2423 my $i = $paths->{$p};
2424 my %s = map { $_ => $i->$_ }
2425 qw/copyfrom_path copyfrom_rev action/;
2426 $ret{$p} = \%s;
2427 }
2428 \%ret;
2429 }
2430
2431 package Git::SVN::Log;
2432 use strict;
2433 use warnings;
2434 use POSIX qw/strftime/;
2435 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2436 %rusers $show_commit $incremental/;
2437 my $l_fmt;
2438
2439 sub cmt_showable {
2440 my ($c) = @_;
2441 return 1 if defined $c->{r};
2442 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2443 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2444 my @log = command(qw/cat-file commit/, $c->{c});
2445 shift @log while ($log[0] ne "\n");
2446 shift @log;
2447 @{$c->{l}} = grep !/^git-svn-id: /, @log;
2448
2449 (undef, $c->{r}, undef) = ::extract_metadata(
2450 (grep(/^git-svn-id: /, @log))[-1]);
2451 }
2452 return defined $c->{r};
2453 }
2454
2455 sub log_use_color {
2456 return 1 if $color;
2457 my ($dc, $dcvar);
2458 $dcvar = 'color.diff';
2459 $dc = `git-config --get $dcvar`;
2460 if ($dc eq '') {
2461 # nothing at all; fallback to "diff.color"
2462 $dcvar = 'diff.color';
2463 $dc = `git-config --get $dcvar`;
2464 }
2465 chomp($dc);
2466 if ($dc eq 'auto') {
2467 my $pc;
2468 $pc = `git-config --get color.pager`;
2469 if ($pc eq '') {
2470 # does not have it -- fallback to pager.color
2471 $pc = `git-config --bool --get pager.color`;
2472 }
2473 else {
2474 $pc = `git-config --bool --get color.pager`;
2475 if ($?) {
2476 $pc = 'false';
2477 }
2478 }
2479 chomp($pc);
2480 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2481 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2482 }
2483 return 0;
2484 }
2485 return 0 if $dc eq 'never';
2486 return 1 if $dc eq 'always';
2487 chomp($dc = `git-config --bool --get $dcvar`);
2488 return ($dc eq 'true');
2489 }
2490
2491 sub git_svn_log_cmd {
2492 my ($r_min, $r_max) = @_;
2493 my $gs = Git::SVN->_new;
2494 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2495 $gs->refname);
2496 push @cmd, '-r' unless $non_recursive;
2497 push @cmd, qw/--raw --name-status/ if $verbose;
2498 push @cmd, '--color' if log_use_color();
2499 return @cmd unless defined $r_max;
2500 if ($r_max == $r_min) {
2501 push @cmd, '--max-count=1';
2502 if (my $c = $gs->rev_db_get($r_max)) {
2503 push @cmd, $c;
2504 }
2505 } else {
2506 my ($c_min, $c_max);
2507 $c_max = $gs->rev_db_get($r_max);
2508 $c_min = $gs->rev_db_get($r_min);
2509 if (defined $c_min && defined $c_max) {
2510 if ($r_max > $r_max) {
2511 push @cmd, "$c_min..$c_max";
2512 } else {
2513 push @cmd, "$c_max..$c_min";
2514 }
2515 } elsif ($r_max > $r_min) {
2516 push @cmd, $c_max;
2517 } else {
2518 push @cmd, $c_min;
2519 }
2520 }
2521 return @cmd;
2522 }
2523
2524 # adapted from pager.c
2525 sub config_pager {
2526 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2527 if (!defined $pager) {
2528 $pager = 'less';
2529 } elsif (length $pager == 0 || $pager eq 'cat') {
2530 $pager = undef;
2531 }
2532 }
2533
2534 sub run_pager {
2535 return unless -t *STDOUT;
2536 pipe my $rfd, my $wfd or return;
2537 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2538 if (!$pid) {
2539 open STDOUT, '>&', $wfd or
2540 ::fatal "Can't redirect to stdout: $!\n";
2541 return;
2542 }
2543 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2544 $ENV{LESS} ||= 'FRSX';
2545 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2546 }
2547
2548 sub tz_to_s_offset {
2549 my ($tz) = @_;
2550 $tz =~ s/(\d\d)$//;
2551 return ($1 * 60) + ($tz * 3600);
2552 }
2553
2554 sub get_author_info {
2555 my ($dest, $author, $t, $tz) = @_;
2556 $author =~ s/(?:^\s*|\s*$)//g;
2557 $dest->{a_raw} = $author;
2558 my $au;
2559 if ($::_authors) {
2560 $au = $rusers{$author} || undef;
2561 }
2562 if (!$au) {
2563 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2564 }
2565 $dest->{t} = $t;
2566 $dest->{tz} = $tz;
2567 $dest->{a} = $au;
2568 # Date::Parse isn't in the standard Perl distro :(
2569 if ($tz =~ s/^\+//) {
2570 $t += tz_to_s_offset($tz);
2571 } elsif ($tz =~ s/^\-//) {
2572 $t -= tz_to_s_offset($tz);
2573 }
2574 $dest->{t_utc} = $t;
2575 }
2576
2577 sub process_commit {
2578 my ($c, $r_min, $r_max, $defer) = @_;
2579 if (defined $r_min && defined $r_max) {
2580 if ($r_min == $c->{r} && $r_min == $r_max) {
2581 show_commit($c);
2582 return 0;
2583 }
2584 return 1 if $r_min == $r_max;
2585 if ($r_min < $r_max) {
2586 # we need to reverse the print order
2587 return 0 if (defined $limit && --$limit < 0);
2588 push @$defer, $c;
2589 return 1;
2590 }
2591 if ($r_min != $r_max) {
2592 return 1 if ($r_min < $c->{r});
2593 return 1 if ($r_max > $c->{r});
2594 }
2595 }
2596 return 0 if (defined $limit && --$limit < 0);
2597 show_commit($c);
2598 return 1;
2599 }
2600
2601 sub show_commit {
2602 my $c = shift;
2603 if ($oneline) {
2604 my $x = "\n";
2605 if (my $l = $c->{l}) {
2606 while ($l->[0] =~ /^\s*$/) { shift @$l }
2607 $x = $l->[0];
2608 }
2609 $l_fmt ||= 'A' . length($c->{r});
2610 print 'r',pack($l_fmt, $c->{r}),' | ';
2611 print "$c->{c} | " if $show_commit;
2612 print $x;
2613 } else {
2614 show_commit_normal($c);
2615 }
2616 }
2617
2618 sub show_commit_changed_paths {
2619 my ($c) = @_;
2620 return unless $c->{changed};
2621 print "Changed paths:\n", @{$c->{changed}};
2622 }
2623
2624 sub show_commit_normal {
2625 my ($c) = @_;
2626 print '-' x72, "\nr$c->{r} | ";
2627 print "$c->{c} | " if $show_commit;
2628 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2629 localtime($c->{t_utc})), ' | ';
2630 my $nr_line = 0;
2631
2632 if (my $l = $c->{l}) {
2633 while ($l->[$#$l] eq "\n" && $#$l > 0
2634 && $l->[($#$l - 1)] eq "\n") {
2635 pop @$l;
2636 }
2637 $nr_line = scalar @$l;
2638 if (!$nr_line) {
2639 print "1 line\n\n\n";
2640 } else {
2641 if ($nr_line == 1) {
2642 $nr_line = '1 line';
2643 } else {
2644 $nr_line .= ' lines';
2645 }
2646 print $nr_line, "\n";
2647 show_commit_changed_paths($c);
2648 print "\n";
2649 print $_ foreach @$l;
2650 }
2651 } else {
2652 print "1 line\n";
2653 show_commit_changed_paths($c);
2654 print "\n";
2655
2656 }
2657 foreach my $x (qw/raw diff/) {
2658 if ($c->{$x}) {
2659 print "\n";
2660 print $_ foreach @{$c->{$x}}
2661 }
2662 }
2663 }
2664
2665 sub cmd_show_log {
2666 my (@args) = @_;
2667 my ($r_min, $r_max);
2668 my $r_last = -1; # prevent dupes
2669 if (defined $TZ) {
2670 $ENV{TZ} = $TZ;
2671 } else {
2672 delete $ENV{TZ};
2673 }
2674 if (defined $::_revision) {
2675 if ($::_revision =~ /^(\d+):(\d+)$/) {
2676 ($r_min, $r_max) = ($1, $2);
2677 } elsif ($::_revision =~ /^\d+$/) {
2678 $r_min = $r_max = $::_revision;
2679 } else {
2680 ::fatal "-r$::_revision is not supported, use ",
2681 "standard \'git log\' arguments instead\n";
2682 }
2683 }
2684
2685 config_pager();
2686 @args = (git_svn_log_cmd($r_min, $r_max), @args);
2687 my $log = command_output_pipe(@args);
2688 run_pager();
2689 my (@k, $c, $d);
2690 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
2691 while (<$log>) {
2692 if (/^${esc_color}commit ($::sha1_short)/o) {
2693 my $cmt = $1;
2694 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
2695 $r_last = $c->{r};
2696 process_commit($c, $r_min, $r_max, \@k) or
2697 goto out;
2698 }
2699 $d = undef;
2700 $c = { c => $cmt };
2701 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
2702 get_author_info($c, $1, $2, $3);
2703 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
2704 # ignore
2705 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
2706 push @{$c->{raw}}, $_;
2707 } elsif (/^${esc_color}[ACRMDT]\t/) {
2708 # we could add $SVN->{svn_path} here, but that requires
2709 # remote access at the moment (repo_path_split)...
2710 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
2711 push @{$c->{changed}}, $_;
2712 } elsif (/^${esc_color}diff /o) {
2713 $d = 1;
2714 push @{$c->{diff}}, $_;
2715 } elsif ($d) {
2716 push @{$c->{diff}}, $_;
2717 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
2718 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
2719 } elsif (s/^${esc_color} //o) {
2720 push @{$c->{l}}, $_;
2721 }
2722 }
2723 if ($c && defined $c->{r} && $c->{r} != $r_last) {
2724 $r_last = $c->{r};
2725 process_commit($c, $r_min, $r_max, \@k);
2726 }
2727 if (@k) {
2728 my $swap = $r_max;
2729 $r_max = $r_min;
2730 $r_min = $swap;
2731 process_commit($_, $r_min, $r_max) foreach reverse @k;
2732 }
2733 out:
2734 close $log;
2735 print '-' x72,"\n" unless $incremental || $oneline;
2736 }
2737
2738 package Git::SVN::Migration;
2739 # these version numbers do NOT correspond to actual version numbers
2740 # of git nor git-svn. They are just relative.
2741 #
2742 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
2743 #
2744 # v1 layout: .git/$id/info/url, refs/remotes/$id
2745 #
2746 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
2747 #
2748 # v3 layout: .git/svn/$id, refs/remotes/$id
2749 # - info/url may remain for backwards compatibility
2750 # - this is what we migrate up to this layout automatically,
2751 # - this will be used by git svn init on single branches
2752 #
2753 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
2754 # - this is only created for newly multi-init-ed
2755 # repositories. Similar in spirit to the
2756 # --use-separate-remotes option in git-clone (now default)
2757 # - we do not automatically migrate to this (following
2758 # the example set by core git)
2759 use strict;
2760 use warnings;
2761 use Carp qw/croak/;
2762 use File::Path qw/mkpath/;
2763 use File::Basename qw/dirname basename/;
2764 use vars qw/$_minimize/;
2765
2766 sub migrate_from_v0 {
2767 my $git_dir = $ENV{GIT_DIR};
2768 return undef unless -d $git_dir;
2769 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2770 my $migrated = 0;
2771 while (<$fh>) {
2772 chomp;
2773 my ($id, $orig_ref) = ($_, $_);
2774 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
2775 next unless -f "$git_dir/$id/info/url";
2776 my $new_ref = "refs/remotes/$id";
2777 if (::verify_ref("$new_ref^0")) {
2778 print STDERR "W: $orig_ref is probably an old ",
2779 "branch used by an ancient version of ",
2780 "git-svn.\n",
2781 "However, $new_ref also exists.\n",
2782 "We will not be able ",
2783 "to use this branch until this ",
2784 "ambiguity is resolved.\n";
2785 next;
2786 }
2787 print STDERR "Migrating from v0 layout...\n" if !$migrated;
2788 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
2789 command_noisy('update-ref', $new_ref, $orig_ref);
2790 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
2791 $migrated++;
2792 }
2793 command_close_pipe($fh, $ctx);
2794 print STDERR "Done migrating from v0 layout...\n" if $migrated;
2795 $migrated;
2796 }
2797
2798 sub migrate_from_v1 {
2799 my $git_dir = $ENV{GIT_DIR};
2800 my $migrated = 0;
2801 return $migrated unless -d $git_dir;
2802 my $svn_dir = "$git_dir/svn";
2803
2804 # just in case somebody used 'svn' as their $id at some point...
2805 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
2806
2807 print STDERR "Migrating from a git-svn v1 layout...\n";
2808 mkpath([$svn_dir]);
2809 print STDERR "Data from a previous version of git-svn exists, but\n\t",
2810 "$svn_dir\n\t(required for this version ",
2811 "($::VERSION) of git-svn) does not. exist\n";
2812 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2813 while (<$fh>) {
2814 my $x = $_;
2815 next unless $x =~ s#^refs/remotes/##;
2816 chomp $x;
2817 next unless -f "$git_dir/$x/info/url";
2818 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
2819 next unless $u;
2820 my $dn = dirname("$git_dir/svn/$x");
2821 mkpath([$dn]) unless -d $dn;
2822 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
2823 mkpath(["$git_dir/svn/svn"]);
2824 print STDERR " - $git_dir/$x/info => ",
2825 "$git_dir/svn/$x/info\n";
2826 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
2827 croak "$!: $x";
2828 # don't worry too much about these, they probably
2829 # don't exist with repos this old (save for index,
2830 # and we can easily regenerate that)
2831 foreach my $f (qw/unhandled.log index .rev_db/) {
2832 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
2833 }
2834 } else {
2835 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
2836 rename "$git_dir/$x", "$git_dir/svn/$x" or
2837 croak "$!: $x";
2838 }
2839 $migrated++;
2840 }
2841 command_close_pipe($fh, $ctx);
2842 print STDERR "Done migrating from a git-svn v1 layout\n";
2843 $migrated;
2844 }
2845
2846 sub read_old_urls {
2847 my ($l_map, $pfx, $path) = @_;
2848 my @dir;
2849 foreach (<$path/*>) {
2850 if (-r "$_/info/url") {
2851 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2852 my $ref_id = $pfx . basename $_;
2853 my $url = ::file_to_s("$_/info/url");
2854 $l_map->{$ref_id} = $url;
2855 } elsif (-d $_) {
2856 push @dir, $_;
2857 }
2858 }
2859 foreach (@dir) {
2860 my $x = $_;
2861 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
2862 read_old_urls($l_map, $x, $_);
2863 }
2864 }
2865
2866 sub migrate_from_v2 {
2867 my @cfg = command(qw/config -l/);
2868 return if grep /^svn-remote\..+\.url=/, @cfg;
2869 my %l_map;
2870 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
2871 my $migrated = 0;
2872
2873 foreach my $ref_id (sort keys %l_map) {
2874 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
2875 $migrated++;
2876 }
2877 $migrated;
2878 }
2879
2880 sub minimize_connections {
2881 my $r = Git::SVN::read_all_remotes();
2882 my $new_urls = {};
2883 my $root_repos = {};
2884 foreach my $repo_id (keys %$r) {
2885 my $url = $r->{$repo_id}->{url} or next;
2886 my $fetch = $r->{$repo_id}->{fetch} or next;
2887 my $ra = Git::SVN::Ra->new($url);
2888
2889 # skip existing cases where we already connect to the root
2890 if (($ra->{url} eq $ra->{repos_root}) ||
2891 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
2892 $repo_id)) {
2893 $root_repos->{$ra->{url}} = $repo_id;
2894 next;
2895 }
2896
2897 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
2898 my $root_path = $ra->{url};
2899 $root_path =~ s#^\Q$ra->{repos_root}\E/*##;
2900 foreach my $path (keys %$fetch) {
2901 my $ref_id = $fetch->{$path};
2902 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
2903
2904 # make sure we can read when connecting to
2905 # a higher level of a repository
2906 my ($last_rev, undef) = $gs->last_rev_commit;
2907 if (!defined $last_rev) {
2908 $last_rev = eval {
2909 $root_ra->get_latest_revnum;
2910 };
2911 next if $@;
2912 }
2913 my $new = $root_path;
2914 $new .= length $path ? "/$path" : '';
2915 eval {
2916 $root_ra->get_log([$new], $last_rev, $last_rev,
2917 0, 0, 1, sub { });
2918 };
2919 next if $@;
2920 $new_urls->{$ra->{repos_root}}->{$new} =
2921 { ref_id => $ref_id,
2922 old_repo_id => $repo_id,
2923 old_path => $path };
2924 }
2925 }
2926
2927 my @emptied;
2928 foreach my $url (keys %$new_urls) {
2929 # see if we can re-use an existing [svn-remote "repo_id"]
2930 # instead of creating a(n ugly) new section:
2931 my $repo_id = $root_repos->{$url} ||
2932 Git::SVN::sanitize_remote_name($url);
2933
2934 my $fetch = $new_urls->{$url};
2935 foreach my $path (keys %$fetch) {
2936 my $x = $fetch->{$path};
2937 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
2938 my $pfx = "svn-remote.$x->{old_repo_id}";
2939
2940 my $old_fetch = quotemeta("$x->{old_path}:".
2941 "refs/remotes/$x->{ref_id}");
2942 command_noisy(qw/config --unset/,
2943 "$pfx.fetch", '^'. $old_fetch . '$');
2944 delete $r->{$x->{old_repo_id}}->
2945 {fetch}->{$x->{old_path}};
2946 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
2947 command_noisy(qw/config --unset/,
2948 "$pfx.url");
2949 push @emptied, $x->{old_repo_id}
2950 }
2951 }
2952 }
2953 if (@emptied) {
2954 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
2955 "$ENV{GIT_DIR}/config";
2956 print STDERR <<EOF;
2957 The following [svn-remote] sections in your config file ($file) are empty
2958 and can be safely removed:
2959 EOF
2960 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
2961 }
2962 }
2963
2964 sub migration_check {
2965 migrate_from_v0();
2966 migrate_from_v1();
2967 migrate_from_v2();
2968 minimize_connections() if $_minimize;
2969 }
2970
2971 package Git::IndexInfo;
2972 use strict;
2973 use warnings;
2974 use Git qw/command_input_pipe command_close_pipe/;
2975
2976 sub new {
2977 my ($class) = @_;
2978 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
2979 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
2980 }
2981
2982 sub remove {
2983 my ($self, $path) = @_;
2984 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
2985 return ++$self->{nr};
2986 }
2987 undef;
2988 }
2989
2990 sub update {
2991 my ($self, $mode, $hash, $path) = @_;
2992 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
2993 return ++$self->{nr};
2994 }
2995 undef;
2996 }
2997
2998 sub DESTROY {
2999 my ($self) = @_;
3000 command_close_pipe($self->{gui}, $self->{ctx});
3001 }
3002
3003 __END__
3004
3005 Data structures:
3006
3007 $log_entry hashref as returned by libsvn_log_entry()
3008 {
3009 log => 'whitespace-formatted log entry
3010 ', # trailing newline is preserved
3011 revision => '8', # integer
3012 date => '2004-02-24T17:01:44.108345Z', # commit date
3013 author => 'committer name'
3014 };
3015
3016
3017 # this is generated by generate_diff();
3018 @mods = array of diff-index line hashes, each element represents one line
3019 of diff-index output
3020
3021 diff-index line ($m hash)
3022 {
3023 mode_a => first column of diff-index output, no leading ':',
3024 mode_b => second column of diff-index output,
3025 sha1_b => sha1sum of the final blob,
3026 chg => change type [MCRADT],
3027 file_a => original file name of a file (iff chg is 'C' or 'R')
3028 file_b => new/current file name of a file (any chg)
3029 }
3030 ;
3031
3032 # retval of read_url_paths{,_all}();
3033 $l_map = {
3034 # repository root url
3035 'https://svn.musicpd.org' => {
3036 # repository path # GIT_SVN_ID
3037 'mpd/trunk' => 'trunk',
3038 'mpd/tags/0.11.5' => 'tags/0.11.5',
3039 },
3040 }
3041
3042 Notes:
3043 I don't trust the each() function on unless I created %hash myself
3044 because the internal iterator may not have started at base.
git with Jakub Narębski modifications