?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-svn: replace .rev_db with a more space-efficient .rev_map format
[git/dscho.git] / git-svn.perl
blobb45c7e3266176336a4d8c6deeaa051c6fd0e9be6
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 # From which subdir have we been invoked?
13 my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15 } || '';
16
17 my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
18 $ENV{GIT_DIR} ||= '.git';
19 $Git::SVN::default_repo_id = 'svn';
20 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
21 $Git::SVN::Ra::_log_window_size = 100;
22
23 $Git::SVN::Log::TZ = $ENV{TZ};
24 $ENV{TZ} = 'UTC';
25 $| = 1; # unbuffer STDOUT
26
27 sub fatal (@) { print STDERR "@_\n"; exit 1 }
28 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
29 require SVN::Ra;
30 require SVN::Delta;
31 if ($SVN::Core::VERSION lt '1.1.0') {
32 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
33 }
34 push @Git::SVN::Ra::ISA, 'SVN::Ra';
35 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
36 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
37 use Carp qw/croak/;
38 use Digest::MD5;
39 use IO::File qw//;
40 use File::Basename qw/dirname basename/;
41 use File::Path qw/mkpath/;
42 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
43 use IPC::Open3;
44 use Git;
45
46 BEGIN {
47 # import functions from Git into our packages, en masse
48 no strict 'refs';
49 foreach (qw/command command_oneline command_noisy command_output_pipe
50 command_input_pipe command_close_pipe/) {
51 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
52 Git::SVN::Migration Git::SVN::Log Git::SVN),
53 __PACKAGE__) {
54 *{"${package}::$_"} = \&{"Git::$_"};
55 }
56 }
57 }
58
59 my ($SVN);
60
61 $sha1 = qr/[a-f\d]{40}/;
62 $sha1_short = qr/[a-f\d]{4,40}/;
63 my ($_stdin, $_help, $_edit,
64 $_message, $_file,
65 $_template, $_shared,
66 $_version, $_fetch_all, $_no_rebase,
67 $_merge, $_strategy, $_dry_run, $_local,
68 $_prefix, $_no_checkout, $_url, $_verbose);
69 $Git::SVN::_follow_parent = 1;
70 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
71 'config-dir=s' => \$Git::SVN::Ra::config_dir,
72 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
73 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
74 'authors-file|A=s' => \$_authors,
75 'repack:i' => \$Git::SVN::_repack,
76 'noMetadata' => \$Git::SVN::_no_metadata,
77 'useSvmProps' => \$Git::SVN::_use_svm_props,
78 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
79 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
80 'no-checkout' => \$_no_checkout,
81 'quiet|q' => \$_q,
82 'repack-flags|repack-args|repack-opts=s' =>
83 \$Git::SVN::_repack_flags,
84 'use-log-author' => \$Git::SVN::_use_log_author,
85 %remote_opts );
86
87 my ($_trunk, $_tags, $_branches, $_stdlayout);
88 my %icv;
89 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
90 'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
91 'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
92 'stdlayout|s' => \$_stdlayout,
93 'minimize-url|m' => \$Git::SVN::_minimize_url,
94 'no-metadata' => sub { $icv{noMetadata} = 1 },
95 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
96 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
97 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
98 %remote_opts );
99 my %cmt_opts = ( 'edit|e' => \$_edit,
100 'rmdir' => \$SVN::Git::Editor::_rmdir,
101 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
102 'l=i' => \$SVN::Git::Editor::_rename_limit,
103 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
104 );
105
106 my %cmd = (
107 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
108 { 'revision|r=s' => \$_revision,
109 'fetch-all|all' => \$_fetch_all,
110 %fc_opts } ],
111 clone => [ \&cmd_clone, "Initialize and fetch revisions",
112 { 'revision|r=s' => \$_revision,
113 %fc_opts, %init_opts } ],
114 init => [ \&cmd_init, "Initialize a repo for tracking" .
115 " (requires URL argument)",
116 \%init_opts ],
117 'multi-init' => [ \&cmd_multi_init,
118 "Deprecated alias for ".
119 "'$0 init -T<trunk> -b<branches> -t<tags>'",
120 \%init_opts ],
121 dcommit => [ \&cmd_dcommit,
122 'Commit several diffs to merge with upstream',
123 { 'merge|m|M' => \$_merge,
124 'strategy|s=s' => \$_strategy,
125 'verbose|v' => \$_verbose,
126 'dry-run|n' => \$_dry_run,
127 'fetch-all|all' => \$_fetch_all,
128 'no-rebase' => \$_no_rebase,
129 %cmt_opts, %fc_opts } ],
130 'set-tree' => [ \&cmd_set_tree,
131 "Set an SVN repository to a git tree-ish",
132 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
133 'create-ignore' => [ \&cmd_create_ignore,
134 'Create a .gitignore per svn:ignore',
135 { 'revision|r=i' => \$_revision
136 } ],
137 'propget' => [ \&cmd_propget,
138 'Print the value of a property on a file or directory',
139 { 'revision|r=i' => \$_revision } ],
140 'proplist' => [ \&cmd_proplist,
141 'List all properties of a file or directory',
142 { 'revision|r=i' => \$_revision } ],
143 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
144 { 'revision|r=i' => \$_revision
145 } ],
146 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
147 { 'revision|r=i' => \$_revision
148 } ],
149 'multi-fetch' => [ \&cmd_multi_fetch,
150 "Deprecated alias for $0 fetch --all",
151 { 'revision|r=s' => \$_revision, %fc_opts } ],
152 'migrate' => [ sub { },
153 # no-op, we automatically run this anyways,
154 'Migrate configuration/metadata/layout from
155 previous versions of git-svn',
156 { 'minimize' => \$Git::SVN::Migration::_minimize,
157 %remote_opts } ],
158 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
159 { 'limit=i' => \$Git::SVN::Log::limit,
160 'revision|r=s' => \$_revision,
161 'verbose|v' => \$Git::SVN::Log::verbose,
162 'incremental' => \$Git::SVN::Log::incremental,
163 'oneline' => \$Git::SVN::Log::oneline,
164 'show-commit' => \$Git::SVN::Log::show_commit,
165 'non-recursive' => \$Git::SVN::Log::non_recursive,
166 'authors-file|A=s' => \$_authors,
167 'color' => \$Git::SVN::Log::color,
168 'pager=s' => \$Git::SVN::Log::pager
169 } ],
170 'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
171 {} ],
172 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
173 { 'merge|m|M' => \$_merge,
174 'verbose|v' => \$_verbose,
175 'strategy|s=s' => \$_strategy,
176 'local|l' => \$_local,
177 'fetch-all|all' => \$_fetch_all,
178 %fc_opts } ],
179 'commit-diff' => [ \&cmd_commit_diff,
180 'Commit a diff between two trees',
181 { 'message|m=s' => \$_message,
182 'file|F=s' => \$_file,
183 'revision|r=s' => \$_revision,
184 %cmt_opts } ],
185 'info' => [ \&cmd_info,
186 "Show info about the latest SVN revision
187 on the current branch",
188 { 'url' => \$_url, } ],
189 );
190
191 my $cmd;
192 for (my $i = 0; $i < @ARGV; $i++) {
193 if (defined $cmd{$ARGV[$i]}) {
194 $cmd = $ARGV[$i];
195 splice @ARGV, $i, 1;
196 last;
197 }
198 };
199
200 # make sure we're always running
201 unless ($cmd =~ /(?:clone|init|multi-init)$/) {
202 unless (-d $ENV{GIT_DIR}) {
203 if ($git_dir_user_set) {
204 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
205 "but it is not a directory\n";
206 }
207 my $git_dir = delete $ENV{GIT_DIR};
208 chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
209 unless (length $cdup) {
210 die "Already at toplevel, but $git_dir ",
211 "not found '$cdup'\n";
212 }
213 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
214 unless (-d $git_dir) {
215 die "$git_dir still not found after going to ",
216 "'$cdup'\n";
217 }
218 $ENV{GIT_DIR} = $git_dir;
219 }
220 }
221
222 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
223
224 read_repo_config(\%opts);
225 Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
226 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
227 'minimize-connections' => \$Git::SVN::Migration::_minimize,
228 'id|i=s' => \$Git::SVN::default_ref_id,
229 'svn-remote|remote|R=s' => sub {
230 $Git::SVN::no_reuse_existing = 1;
231 $Git::SVN::default_repo_id = $_[1] });
232 exit 1 if (!$rv && $cmd && $cmd ne 'log');
233
234 usage(0) if $_help;
235 version() if $_version;
236 usage(1) unless defined $cmd;
237 load_authors() if $_authors;
238
239 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
240 Git::SVN::Migration::migration_check();
241 }
242 Git::SVN::init_vars();
243 eval {
244 Git::SVN::verify_remotes_sanity();
245 $cmd{$cmd}->[0]->(@ARGV);
246 };
247 fatal $@ if $@;
248 post_fetch_checkout();
249 exit 0;
250
251 ####################### primary functions ######################
252 sub usage {
253 my $exit = shift || 0;
254 my $fd = $exit ? \*STDERR : \*STDOUT;
255 print $fd <<"";
256 git-svn - bidirectional operations between a single Subversion tree and git
257 Usage: $0 <command> [options] [arguments]\n
258
259 print $fd "Available commands:\n" unless $cmd;
260
261 foreach (sort keys %cmd) {
262 next if $cmd && $cmd ne $_;
263 next if /^multi-/; # don't show deprecated commands
264 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
265 foreach (sort keys %{$cmd{$_}->[2]}) {
266 # mixed-case options are for .git/config only
267 next if /[A-Z]/ && /^[a-z]+$/i;
268 # prints out arguments as they should be passed:
269 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
270 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
271 "--$_" : "-$_" }
272 split /\|/,$_)," $x\n";
273 }
274 }
275 print $fd <<"";
276 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
277 arbitrary identifier if you're tracking multiple SVN branches/repositories in
278 one git repository and want to keep them separate. See git-svn(1) for more
279 information.
280
281 exit $exit;
282 }
283
284 sub version {
285 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
286 exit 0;
287 }
288
289 sub do_git_init_db {
290 unless (-d $ENV{GIT_DIR}) {
291 my @init_db = ('init');
292 push @init_db, "--template=$_template" if defined $_template;
293 if (defined $_shared) {
294 if ($_shared =~ /[a-z]/) {
295 push @init_db, "--shared=$_shared";
296 } else {
297 push @init_db, "--shared";
298 }
299 }
300 command_noisy(@init_db);
301 }
302 my $set;
303 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
304 foreach my $i (keys %icv) {
305 die "'$set' and '$i' cannot both be set\n" if $set;
306 next unless defined $icv{$i};
307 command_noisy('config', "$pfx.$i", $icv{$i});
308 $set = $i;
309 }
310 }
311
312 sub init_subdir {
313 my $repo_path = shift or return;
314 mkpath([$repo_path]) unless -d $repo_path;
315 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
316 $ENV{GIT_DIR} = '.git';
317 }
318
319 sub cmd_clone {
320 my ($url, $path) = @_;
321 if (!defined $path &&
322 (defined $_trunk || defined $_branches || defined $_tags ||
323 defined $_stdlayout) &&
324 $url !~ m#^[a-z\+]+://#) {
325 $path = $url;
326 }
327 $path = basename($url) if !defined $path || !length $path;
328 cmd_init($url, $path);
329 Git::SVN::fetch_all($Git::SVN::default_repo_id);
330 }
331
332 sub cmd_init {
333 if (defined $_stdlayout) {
334 $_trunk = 'trunk' if (!defined $_trunk);
335 $_tags = 'tags' if (!defined $_tags);
336 $_branches = 'branches' if (!defined $_branches);
337 }
338 if (defined $_trunk || defined $_branches || defined $_tags) {
339 return cmd_multi_init(@_);
340 }
341 my $url = shift or die "SVN repository location required ",
342 "as a command-line argument\n";
343 init_subdir(@_);
344 do_git_init_db();
345
346 Git::SVN->init($url);
347 }
348
349 sub cmd_fetch {
350 if (grep /^\d+=./, @_) {
351 die "'<rev>=<commit>' fetch arguments are ",
352 "no longer supported.\n";
353 }
354 my ($remote) = @_;
355 if (@_ > 1) {
356 die "Usage: $0 fetch [--all] [svn-remote]\n";
357 }
358 $remote ||= $Git::SVN::default_repo_id;
359 if ($_fetch_all) {
360 cmd_multi_fetch();
361 } else {
362 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
363 }
364 }
365
366 sub cmd_set_tree {
367 my (@commits) = @_;
368 if ($_stdin || !@commits) {
369 print "Reading from stdin...\n";
370 @commits = ();
371 while (<STDIN>) {
372 if (/\b($sha1_short)\b/o) {
373 unshift @commits, $1;
374 }
375 }
376 }
377 my @revs;
378 foreach my $c (@commits) {
379 my @tmp = command('rev-parse',$c);
380 if (scalar @tmp == 1) {
381 push @revs, $tmp[0];
382 } elsif (scalar @tmp > 1) {
383 push @revs, reverse(command('rev-list',@tmp));
384 } else {
385 fatal "Failed to rev-parse $c";
386 }
387 }
388 my $gs = Git::SVN->new;
389 my ($r_last, $cmt_last) = $gs->last_rev_commit;
390 $gs->fetch;
391 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
392 fatal "There are new revisions that were fetched ",
393 "and need to be merged (or acknowledged) ",
394 "before committing.\nlast rev: $r_last\n",
395 " current: $gs->{last_rev}";
396 }
397 $gs->set_tree($_) foreach @revs;
398 print "Done committing ",scalar @revs," revisions to SVN\n";
399 }
400
401 sub cmd_dcommit {
402 my $head = shift;
403 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
404 'Cannot dcommit with a dirty index. Commit your changes first, '
405 . "or stash them with `git stash'.\n";
406 $head ||= 'HEAD';
407 my @refs;
408 my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
409 print "Committing to $url ...\n";
410 unless ($gs) {
411 die "Unable to determine upstream SVN information from ",
412 "$head history\n";
413 }
414 my $last_rev;
415 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
416 if ($_no_rebase && scalar(@$linear_refs) > 1) {
417 warn "Attempting to commit more than one change while ",
418 "--no-rebase is enabled.\n",
419 "If these changes depend on each other, re-running ",
420 "without --no-rebase will be required."
421 }
422 while (1) {
423 my $d = shift @$linear_refs or last;
424 unless (defined $last_rev) {
425 (undef, $last_rev, undef) = cmt_metadata("$d~1");
426 unless (defined $last_rev) {
427 fatal "Unable to extract revision information ",
428 "from commit $d~1";
429 }
430 }
431 if ($_dry_run) {
432 print "diff-tree $d~1 $d\n";
433 } else {
434 my $cmt_rev;
435 my %ed_opts = ( r => $last_rev,
436 log => get_commit_entry($d)->{log},
437 ra => Git::SVN::Ra->new($gs->full_url),
438 config => SVN::Core::config_get_config(
439 $Git::SVN::Ra::config_dir
440 ),
441 tree_a => "$d~1",
442 tree_b => $d,
443 editor_cb => sub {
444 print "Committed r$_[0]\n";
445 $cmt_rev = $_[0];
446 },
447 svn_path => '');
448 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
449 print "No changes\n$d~1 == $d\n";
450 } elsif ($parents->{$d} && @{$parents->{$d}}) {
451 $gs->{inject_parents_dcommit}->{$cmt_rev} =
452 $parents->{$d};
453 }
454 $_fetch_all ? $gs->fetch_all : $gs->fetch;
455 next if $_no_rebase;
456
457 # we always want to rebase against the current HEAD,
458 # not any head that was passed to us
459 my @diff = command('diff-tree', $d,
460 $gs->refname, '--');
461 my @finish;
462 if (@diff) {
463 @finish = rebase_cmd();
464 print STDERR "W: $d and ", $gs->refname,
465 " differ, using @finish:\n",
466 join("\n", @diff), "\n";
467 } else {
468 print "No changes between current HEAD and ",
469 $gs->refname,
470 "\nResetting to the latest ",
471 $gs->refname, "\n";
472 @finish = qw/reset --mixed/;
473 }
474 command_noisy(@finish, $gs->refname);
475 if (@diff) {
476 @refs = ();
477 my ($url_, $rev_, $uuid_, $gs_) =
478 working_head_info($head, \@refs);
479 my ($linear_refs_, $parents_) =
480 linearize_history($gs_, \@refs);
481 if (scalar(@$linear_refs) !=
482 scalar(@$linear_refs_)) {
483 fatal "# of revisions changed ",
484 "\nbefore:\n",
485 join("\n", @$linear_refs),
486 "\n\nafter:\n",
487 join("\n", @$linear_refs_), "\n",
488 'If you are attempting to commit ',
489 "merges, try running:\n\t",
490 'git rebase --interactive',
491 '--preserve-merges ',
492 $gs->refname,
493 "\nBefore dcommitting";
494 }
495 if ($url_ ne $url) {
496 fatal "URL mismatch after rebase: ",
497 "$url_ != $url";
498 }
499 if ($uuid_ ne $uuid) {
500 fatal "uuid mismatch after rebase: ",
501 "$uuid_ != $uuid";
502 }
503 # remap parents
504 my (%p, @l, $i);
505 for ($i = 0; $i < scalar @$linear_refs; $i++) {
506 my $new = $linear_refs_->[$i] or next;
507 $p{$new} =
508 $parents->{$linear_refs->[$i]};
509 push @l, $new;
510 }
511 $parents = \%p;
512 $linear_refs = \@l;
513 }
514 $last_rev = $cmt_rev;
515 }
516 }
517 }
518
519 sub cmd_find_rev {
520 my $revision_or_hash = shift;
521 my $result;
522 if ($revision_or_hash =~ /^r\d+$/) {
523 my $head = shift;
524 $head ||= 'HEAD';
525 my @refs;
526 my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
527 unless ($gs) {
528 die "Unable to determine upstream SVN information from ",
529 "$head history\n";
530 }
531 my $desired_revision = substr($revision_or_hash, 1);
532 $result = $gs->rev_map_get($desired_revision);
533 } else {
534 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
535 $result = $rev;
536 }
537 print "$result\n" if $result;
538 }
539
540 sub cmd_rebase {
541 command_noisy(qw/update-index --refresh/);
542 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
543 unless ($gs) {
544 die "Unable to determine upstream SVN information from ",
545 "working tree history\n";
546 }
547 if (command(qw/diff-index HEAD --/)) {
548 print STDERR "Cannot rebase with uncommited changes:\n";
549 command_noisy('status');
550 exit 1;
551 }
552 unless ($_local) {
553 # rebase will checkout for us, so no need to do it explicitly
554 $_no_checkout = 'true';
555 $_fetch_all ? $gs->fetch_all : $gs->fetch;
556 }
557 command_noisy(rebase_cmd(), $gs->refname);
558 }
559
560 sub cmd_show_ignore {
561 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
562 $gs ||= Git::SVN->new;
563 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
564 $gs->prop_walk($gs->{path}, $r, sub {
565 my ($gs, $path, $props) = @_;
566 print STDOUT "\n# $path\n";
567 my $s = $props->{'svn:ignore'} or return;
568 $s =~ s/[\r\n]+/\n/g;
569 chomp $s;
570 $s =~ s#^#$path#gm;
571 print STDOUT "$s\n";
572 });
573 }
574
575 sub cmd_show_externals {
576 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
577 $gs ||= Git::SVN->new;
578 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
579 $gs->prop_walk($gs->{path}, $r, sub {
580 my ($gs, $path, $props) = @_;
581 print STDOUT "\n# $path\n";
582 my $s = $props->{'svn:externals'} or return;
583 $s =~ s/[\r\n]+/\n/g;
584 chomp $s;
585 $s =~ s#^#$path#gm;
586 print STDOUT "$s\n";
587 });
588 }
589
590 sub cmd_create_ignore {
591 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
592 $gs ||= Git::SVN->new;
593 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
594 $gs->prop_walk($gs->{path}, $r, sub {
595 my ($gs, $path, $props) = @_;
596 # $path is of the form /path/to/dir/
597 my $ignore = '.' . $path . '.gitignore';
598 my $s = $props->{'svn:ignore'} or return;
599 open(GITIGNORE, '>', $ignore)
600 or fatal("Failed to open `$ignore' for writing: $!");
601 $s =~ s/[\r\n]+/\n/g;
602 chomp $s;
603 # Prefix all patterns so that the ignore doesn't apply
604 # to sub-directories.
605 $s =~ s#^#/#gm;
606 print GITIGNORE "$s\n";
607 close(GITIGNORE)
608 or fatal("Failed to close `$ignore': $!");
609 command_noisy('add', $ignore);
610 });
611 }
612
613 sub canonicalize_path {
614 my ($path) = @_;
615 my $dot_slash_added = 0;
616 if (substr($path, 0, 1) ne "/") {
617 $path = "./" . $path;
618 $dot_slash_added = 1;
619 }
620 # File::Spec->canonpath doesn't collapse x/../y into y (for a
621 # good reason), so let's do this manually.
622 $path =~ s#/+#/#g;
623 $path =~ s#/\.(?:/|$)#/#g;
624 $path =~ s#/[^/]+/\.\.##g;
625 $path =~ s#/$##g;
626 $path =~ s#^\./## if $dot_slash_added;
627 return $path;
628 }
629
630 # get_svnprops(PATH)
631 # ------------------
632 # Helper for cmd_propget and cmd_proplist below.
633 sub get_svnprops {
634 my $path = shift;
635 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
636 $gs ||= Git::SVN->new;
637
638 # prefix THE PATH by the sub-directory from which the user
639 # invoked us.
640 $path = $cmd_dir_prefix . $path;
641 fatal("No such file or directory: $path") unless -e $path;
642 my $is_dir = -d $path ? 1 : 0;
643 $path = $gs->{path} . '/' . $path;
644
645 # canonicalize the path (otherwise libsvn will abort or fail to
646 # find the file)
647 $path = canonicalize_path($path);
648
649 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
650 my $props;
651 if ($is_dir) {
652 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
653 }
654 else {
655 (undef, $props) = $gs->ra->get_file($path, $r, undef);
656 }
657 return $props;
658 }
659
660 # cmd_propget (PROP, PATH)
661 # ------------------------
662 # Print the SVN property PROP for PATH.
663 sub cmd_propget {
664 my ($prop, $path) = @_;
665 $path = '.' if not defined $path;
666 usage(1) if not defined $prop;
667 my $props = get_svnprops($path);
668 if (not defined $props->{$prop}) {
669 fatal("`$path' does not have a `$prop' SVN property.");
670 }
671 print $props->{$prop} . "\n";
672 }
673
674 # cmd_proplist (PATH)
675 # -------------------
676 # Print the list of SVN properties for PATH.
677 sub cmd_proplist {
678 my $path = shift;
679 $path = '.' if not defined $path;
680 my $props = get_svnprops($path);
681 print "Properties on '$path':\n";
682 foreach (sort keys %{$props}) {
683 print " $_\n";
684 }
685 }
686
687 sub cmd_multi_init {
688 my $url = shift;
689 unless (defined $_trunk || defined $_branches || defined $_tags) {
690 usage(1);
691 }
692
693 # there are currently some bugs that prevent multi-init/multi-fetch
694 # setups from working well without this.
695 $Git::SVN::_minimize_url = 1;
696
697 $_prefix = '' unless defined $_prefix;
698 if (defined $url) {
699 $url =~ s#/+$##;
700 init_subdir(@_);
701 }
702 do_git_init_db();
703 if (defined $_trunk) {
704 my $trunk_ref = $_prefix . 'trunk';
705 # try both old-style and new-style lookups:
706 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
707 unless ($gs_trunk) {
708 my ($trunk_url, $trunk_path) =
709 complete_svn_url($url, $_trunk);
710 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
711 undef, $trunk_ref);
712 }
713 }
714 return unless defined $_branches || defined $_tags;
715 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
716 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
717 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
718 }
719
720 sub cmd_multi_fetch {
721 my $remotes = Git::SVN::read_all_remotes();
722 foreach my $repo_id (sort keys %$remotes) {
723 if ($remotes->{$repo_id}->{url}) {
724 Git::SVN::fetch_all($repo_id, $remotes);
725 }
726 }
727 }
728
729 # this command is special because it requires no metadata
730 sub cmd_commit_diff {
731 my ($ta, $tb, $url) = @_;
732 my $usage = "Usage: $0 commit-diff -r<revision> ".
733 "<tree-ish> <tree-ish> [<URL>]";
734 fatal($usage) if (!defined $ta || !defined $tb);
735 my $svn_path;
736 if (!defined $url) {
737 my $gs = eval { Git::SVN->new };
738 if (!$gs) {
739 fatal("Needed URL or usable git-svn --id in ",
740 "the command-line\n", $usage);
741 }
742 $url = $gs->{url};
743 $svn_path = $gs->{path};
744 }
745 unless (defined $_revision) {
746 fatal("-r|--revision is a required argument\n", $usage);
747 }
748 if (defined $_message && defined $_file) {
749 fatal("Both --message/-m and --file/-F specified ",
750 "for the commit message.\n",
751 "I have no idea what you mean");
752 }
753 if (defined $_file) {
754 $_message = file_to_s($_file);
755 } else {
756 $_message ||= get_commit_entry($tb)->{log};
757 }
758 my $ra ||= Git::SVN::Ra->new($url);
759 $svn_path ||= $ra->{svn_path};
760 my $r = $_revision;
761 if ($r eq 'HEAD') {
762 $r = $ra->get_latest_revnum;
763 } elsif ($r !~ /^\d+$/) {
764 die "revision argument: $r not understood by git-svn\n";
765 }
766 my %ed_opts = ( r => $r,
767 log => $_message,
768 ra => $ra,
769 tree_a => $ta,
770 tree_b => $tb,
771 editor_cb => sub { print "Committed r$_[0]\n" },
772 svn_path => $svn_path );
773 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
774 print "No changes\n$ta == $tb\n";
775 }
776 }
777
778 sub cmd_info {
779 my $path = canonicalize_path(shift or ".");
780 unless (scalar(@_) == 0) {
781 die "Too many arguments specified\n";
782 }
783
784 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
785
786 if (!$file_type && !$diff_status) {
787 print STDERR "$path: (Not a versioned resource)\n\n";
788 return;
789 }
790
791 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
792 unless ($gs) {
793 die "Unable to determine upstream SVN information from ",
794 "working tree history\n";
795 }
796 my $full_url = $url . ($path eq "." ? "" : "/$path");
797
798 if ($_url) {
799 print $full_url, "\n";
800 return;
801 }
802
803 my $result = "Path: $path\n";
804 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
805 $result .= "URL: " . $full_url . "\n";
806
807 eval {
808 my $repos_root = $gs->repos_root;
809 Git::SVN::remove_username($repos_root);
810 $result .= "Repository Root: $repos_root\n";
811 };
812 if ($@) {
813 $result .= "Repository Root: (offline)\n";
814 }
815 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
816 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
817
818 $result .= "Node Kind: " .
819 ($file_type eq "dir" ? "directory" : "file") . "\n";
820
821 my $schedule = $diff_status eq "A"
822 ? "add"
823 : ($diff_status eq "D" ? "delete" : "normal");
824 $result .= "Schedule: $schedule\n";
825
826 if ($diff_status eq "A") {
827 print $result, "\n";
828 return;
829 }
830
831 my ($lc_author, $lc_rev, $lc_date_utc);
832 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $path);
833 my $log = command_output_pipe(@args);
834 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
835 while (<$log>) {
836 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
837 $lc_author = $1;
838 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
839 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
840 (undef, $lc_rev, undef) = ::extract_metadata($1);
841 }
842 }
843 close $log;
844
845 Git::SVN::Log::set_local_timezone();
846
847 $result .= "Last Changed Author: $lc_author\n";
848 $result .= "Last Changed Rev: $lc_rev\n";
849 $result .= "Last Changed Date: " .
850 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
851
852 if ($file_type ne "dir") {
853 my $text_last_updated_date =
854 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
855 $result .=
856 "Text Last Updated: " .
857 Git::SVN::Log::format_svn_date($text_last_updated_date) .
858 "\n";
859 my $checksum;
860 if ($diff_status eq "D") {
861 my ($fh, $ctx) =
862 command_output_pipe(qw(cat-file blob), "HEAD:$path");
863 if ($file_type eq "link") {
864 my $file_name = <$fh>;
865 $checksum = md5sum("link $file_name");
866 } else {
867 $checksum = md5sum($fh);
868 }
869 command_close_pipe($fh, $ctx);
870 } elsif ($file_type eq "link") {
871 my $file_name =
872 command(qw(cat-file blob), "HEAD:$path");
873 $checksum =
874 md5sum("link " . $file_name);
875 } else {
876 open FILE, "<", $path or die $!;
877 $checksum = md5sum(\*FILE);
878 close FILE or die $!;
879 }
880 $result .= "Checksum: " . $checksum . "\n";
881 }
882
883 print $result, "\n";
884 }
885
886 ########################### utility functions #########################
887
888 sub rebase_cmd {
889 my @cmd = qw/rebase/;
890 push @cmd, '-v' if $_verbose;
891 push @cmd, qw/--merge/ if $_merge;
892 push @cmd, "--strategy=$_strategy" if $_strategy;
893 @cmd;
894 }
895
896 sub post_fetch_checkout {
897 return if $_no_checkout;
898 my $gs = $Git::SVN::_head or return;
899 return if verify_ref('refs/heads/master^0');
900
901 my $valid_head = verify_ref('HEAD^0');
902 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
903 return if ($valid_head || !verify_ref('HEAD^0'));
904
905 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
906 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
907 return if -f $index;
908
909 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
910 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
911 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
912 print STDERR "Checked out HEAD:\n ",
913 $gs->full_url, " r", $gs->last_rev, "\n";
914 }
915
916 sub complete_svn_url {
917 my ($url, $path) = @_;
918 $path =~ s#/+$##;
919 if ($path !~ m#^[a-z\+]+://#) {
920 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
921 fatal("E: '$path' is not a complete URL ",
922 "and a separate URL is not specified");
923 }
924 return ($url, $path);
925 }
926 return ($path, '');
927 }
928
929 sub complete_url_ls_init {
930 my ($ra, $repo_path, $switch, $pfx) = @_;
931 unless ($repo_path) {
932 print STDERR "W: $switch not specified\n";
933 return;
934 }
935 $repo_path =~ s#/+$##;
936 if ($repo_path =~ m#^[a-z\+]+://#) {
937 $ra = Git::SVN::Ra->new($repo_path);
938 $repo_path = '';
939 } else {
940 $repo_path =~ s#^/+##;
941 unless ($ra) {
942 fatal("E: '$repo_path' is not a complete URL ",
943 "and a separate URL is not specified");
944 }
945 }
946 my $url = $ra->{url};
947 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
948 my $k = "svn-remote.$gs->{repo_id}.url";
949 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
950 if ($orig_url && ($orig_url ne $gs->{url})) {
951 die "$k already set: $orig_url\n",
952 "wanted to set to: $gs->{url}\n";
953 }
954 command_oneline('config', $k, $gs->{url}) unless $orig_url;
955 my $remote_path = "$ra->{svn_path}/$repo_path/*";
956 $remote_path =~ s#/+#/#g;
957 $remote_path =~ s#^/##g;
958 my ($n) = ($switch =~ /^--(\w+)/);
959 if (length $pfx && $pfx !~ m#/$#) {
960 die "--prefix='$pfx' must have a trailing slash '/'\n";
961 }
962 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
963 "$remote_path:refs/remotes/$pfx*");
964 }
965
966 sub verify_ref {
967 my ($ref) = @_;
968 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
969 { STDERR => 0 }); };
970 }
971
972 sub get_tree_from_treeish {
973 my ($treeish) = @_;
974 # $treeish can be a symbolic ref, too:
975 my $type = command_oneline(qw/cat-file -t/, $treeish);
976 my $expected;
977 while ($type eq 'tag') {
978 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
979 }
980 if ($type eq 'commit') {
981 $expected = (grep /^tree /, command(qw/cat-file commit/,
982 $treeish))[0];
983 ($expected) = ($expected =~ /^tree ($sha1)$/o);
984 die "Unable to get tree from $treeish\n" unless $expected;
985 } elsif ($type eq 'tree') {
986 $expected = $treeish;
987 } else {
988 die "$treeish is a $type, expected tree, tag or commit\n";
989 }
990 return $expected;
991 }
992
993 sub get_commit_entry {
994 my ($treeish) = shift;
995 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
996 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
997 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
998 open my $log_fh, '>', $commit_editmsg or croak $!;
999
1000 my $type = command_oneline(qw/cat-file -t/, $treeish);
1001 if ($type eq 'commit' || $type eq 'tag') {
1002 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1003 $type, $treeish);
1004 my $in_msg = 0;
1005 while (<$msg_fh>) {
1006 if (!$in_msg) {
1007 $in_msg = 1 if (/^\s*$/);
1008 } elsif (/^git-svn-id: /) {
1009 # skip this for now, we regenerate the
1010 # correct one on re-fetch anyways
1011 # TODO: set *:merge properties or like...
1012 } else {
1013 print $log_fh $_ or croak $!;
1014 }
1015 }
1016 command_close_pipe($msg_fh, $ctx);
1017 }
1018 close $log_fh or croak $!;
1019
1020 if ($_edit || ($type eq 'tree')) {
1021 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1022 # TODO: strip out spaces, comments, like git-commit.sh
1023 system($editor, $commit_editmsg);
1024 }
1025 rename $commit_editmsg, $commit_msg or croak $!;
1026 open $log_fh, '<', $commit_msg or croak $!;
1027 { local $/; chomp($log_entry{log} = <$log_fh>); }
1028 close $log_fh or croak $!;
1029 unlink $commit_msg;
1030 \%log_entry;
1031 }
1032
1033 sub s_to_file {
1034 my ($str, $file, $mode) = @_;
1035 open my $fd,'>',$file or croak $!;
1036 print $fd $str,"\n" or croak $!;
1037 close $fd or croak $!;
1038 chmod ($mode &~ umask, $file) if (defined $mode);
1039 }
1040
1041 sub file_to_s {
1042 my $file = shift;
1043 open my $fd,'<',$file or croak "$!: file: $file\n";
1044 local $/;
1045 my $ret = <$fd>;
1046 close $fd or croak $!;
1047 $ret =~ s/\s*$//s;
1048 return $ret;
1049 }
1050
1051 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1052 sub load_authors {
1053 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1054 my $log = $cmd eq 'log';
1055 while (<$authors>) {
1056 chomp;
1057 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1058 my ($user, $name, $email) = ($1, $2, $3);
1059 if ($log) {
1060 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1061 } else {
1062 $users{$user} = [$name, $email];
1063 }
1064 }
1065 close $authors or croak $!;
1066 }
1067
1068 # convert GetOpt::Long specs for use by git-config
1069 sub read_repo_config {
1070 return unless -d $ENV{GIT_DIR};
1071 my $opts = shift;
1072 my @config_only;
1073 foreach my $o (keys %$opts) {
1074 # if we have mixedCase and a long option-only, then
1075 # it's a config-only variable that we don't need for
1076 # the command-line.
1077 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1078 my $v = $opts->{$o};
1079 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1080 $key =~ s/-//g;
1081 my $arg = 'git-config';
1082 $arg .= ' --int' if ($o =~ /[:=]i$/);
1083 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1084 if (ref $v eq 'ARRAY') {
1085 chomp(my @tmp = `$arg --get-all svn.$key`);
1086 @$v = @tmp if @tmp;
1087 } else {
1088 chomp(my $tmp = `$arg --get svn.$key`);
1089 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1090 $$v = $tmp;
1091 }
1092 }
1093 }
1094 delete @$opts{@config_only} if @config_only;
1095 }
1096
1097 sub extract_metadata {
1098 my $id = shift or return (undef, undef, undef);
1099 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1100 \s([a-f\d\-]+)$/x);
1101 if (!defined $rev || !$uuid || !$url) {
1102 # some of the original repositories I made had
1103 # identifiers like this:
1104 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1105 }
1106 return ($url, $rev, $uuid);
1107 }
1108
1109 sub cmt_metadata {
1110 return extract_metadata((grep(/^git-svn-id: /,
1111 command(qw/cat-file commit/, shift)))[-1]);
1112 }
1113
1114 sub working_head_info {
1115 my ($head, $refs) = @_;
1116 my @args = ('log', '--no-color', '--first-parent');
1117 my ($fh, $ctx) = command_output_pipe(@args, $head);
1118 my $hash;
1119 my %max;
1120 while (<$fh>) {
1121 if ( m{^commit ($::sha1)$} ) {
1122 unshift @$refs, $hash if $hash and $refs;
1123 $hash = $1;
1124 next;
1125 }
1126 next unless s{^\s*(git-svn-id:)}{$1};
1127 my ($url, $rev, $uuid) = extract_metadata($_);
1128 if (defined $url && defined $rev) {
1129 next if $max{$url} and $max{$url} < $rev;
1130 if (my $gs = Git::SVN->find_by_url($url)) {
1131 my $c = $gs->rev_map_get($rev);
1132 if ($c && $c eq $hash) {
1133 close $fh; # break the pipe
1134 return ($url, $rev, $uuid, $gs);
1135 } else {
1136 $max{$url} ||= $gs->rev_map_max;
1137 }
1138 }
1139 }
1140 }
1141 command_close_pipe($fh, $ctx);
1142 (undef, undef, undef, undef);
1143 }
1144
1145 sub read_commit_parents {
1146 my ($parents, $c) = @_;
1147 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1148 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1149 @{$parents->{$c}} = split(/ /, $p);
1150 }
1151
1152 sub linearize_history {
1153 my ($gs, $refs) = @_;
1154 my %parents;
1155 foreach my $c (@$refs) {
1156 read_commit_parents(\%parents, $c);
1157 }
1158
1159 my @linear_refs;
1160 my %skip = ();
1161 my $last_svn_commit = $gs->last_commit;
1162 foreach my $c (reverse @$refs) {
1163 next if $c eq $last_svn_commit;
1164 last if $skip{$c};
1165
1166 unshift @linear_refs, $c;
1167 $skip{$c} = 1;
1168
1169 # we only want the first parent to diff against for linear
1170 # history, we save the rest to inject when we finalize the
1171 # svn commit
1172 my $fp_a = verify_ref("$c~1");
1173 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1174 if (!$fp_a || !$fp_b) {
1175 die "Commit $c\n",
1176 "has no parent commit, and therefore ",
1177 "nothing to diff against.\n",
1178 "You should be working from a repository ",
1179 "originally created by git-svn\n";
1180 }
1181 if ($fp_a ne $fp_b) {
1182 die "$c~1 = $fp_a, however parsing commit $c ",
1183 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1184 }
1185
1186 foreach my $p (@{$parents{$c}}) {
1187 $skip{$p} = 1;
1188 }
1189 }
1190 (\@linear_refs, \%parents);
1191 }
1192
1193 sub find_file_type_and_diff_status {
1194 my ($path) = @_;
1195 return ('dir', '') if $path eq '.';
1196
1197 my $diff_output =
1198 command_oneline(qw(diff --cached --name-status --), $path) || "";
1199 my $diff_status = (split(' ', $diff_output))[0] || "";
1200
1201 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1202
1203 return (undef, undef) if !$diff_status && !$ls_tree;
1204
1205 if ($diff_status eq "A") {
1206 return ("link", $diff_status) if -l $path;
1207 return ("dir", $diff_status) if -d $path;
1208 return ("file", $diff_status);
1209 }
1210
1211 my $mode = (split(' ', $ls_tree))[0] || "";
1212
1213 return ("link", $diff_status) if $mode eq "120000";
1214 return ("dir", $diff_status) if $mode eq "040000";
1215 return ("file", $diff_status);
1216 }
1217
1218 sub md5sum {
1219 my $arg = shift;
1220 my $ref = ref $arg;
1221 my $md5 = Digest::MD5->new();
1222 if ($ref eq 'GLOB' || $ref eq 'IO::File') {
1223 $md5->addfile($arg) or croak $!;
1224 } elsif ($ref eq 'SCALAR') {
1225 $md5->add($$arg) or croak $!;
1226 } elsif (!$ref) {
1227 $md5->add($arg) or croak $!;
1228 } else {
1229 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1230 }
1231 return $md5->hexdigest();
1232 }
1233
1234 package Git::SVN;
1235 use strict;
1236 use warnings;
1237 use Fcntl qw/:DEFAULT :seek/;
1238 use constant rev_map_fmt => 'NH40';
1239 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1240 $_repack $_repack_flags $_use_svm_props $_head
1241 $_use_svnsync_props $no_reuse_existing $_minimize_url
1242 $_use_log_author/;
1243 use Carp qw/croak/;
1244 use File::Path qw/mkpath/;
1245 use File::Copy qw/copy/;
1246 use IPC::Open3;
1247
1248 my $_repack_nr;
1249 # properties that we do not log:
1250 my %SKIP_PROP;
1251 BEGIN {
1252 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1253 svn:special svn:executable
1254 svn:entry:committed-rev
1255 svn:entry:last-author
1256 svn:entry:uuid
1257 svn:entry:committed-date/;
1258
1259 # some options are read globally, but can be overridden locally
1260 # per [svn-remote "..."] section. Command-line options will *NOT*
1261 # override options set in an [svn-remote "..."] section
1262 no strict 'refs';
1263 for my $option (qw/follow_parent no_metadata use_svm_props
1264 use_svnsync_props/) {
1265 my $key = $option;
1266 $key =~ tr/_//d;
1267 my $prop = "-$option";
1268 *$option = sub {
1269 my ($self) = @_;
1270 return $self->{$prop} if exists $self->{$prop};
1271 my $k = "svn-remote.$self->{repo_id}.$key";
1272 eval { command_oneline(qw/config --get/, $k) };
1273 if ($@) {
1274 $self->{$prop} = ${"Git::SVN::_$option"};
1275 } else {
1276 my $v = command_oneline(qw/config --bool/,$k);
1277 $self->{$prop} = $v eq 'false' ? 0 : 1;
1278 }
1279 return $self->{$prop};
1280 }
1281 }
1282 }
1283
1284 my %LOCKFILES;
1285 END { unlink keys %LOCKFILES if %LOCKFILES }
1286
1287 sub resolve_local_globs {
1288 my ($url, $fetch, $glob_spec) = @_;
1289 return unless defined $glob_spec;
1290 my $ref = $glob_spec->{ref};
1291 my $path = $glob_spec->{path};
1292 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1293 next unless m#^refs/remotes/$ref->{regex}$#;
1294 my $p = $1;
1295 my $pathname = desanitize_refname($path->full_path($p));
1296 my $refname = desanitize_refname($ref->full_path($p));
1297 if (my $existing = $fetch->{$pathname}) {
1298 if ($existing ne $refname) {
1299 die "Refspec conflict:\n",
1300 "existing: refs/remotes/$existing\n",
1301 " globbed: refs/remotes/$refname\n";
1302 }
1303 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1304 $u =~ s!^\Q$url\E(/|$)!! or die
1305 "refs/remotes/$refname: '$url' not found in '$u'\n";
1306 if ($pathname ne $u) {
1307 warn "W: Refspec glob conflict ",
1308 "(ref: refs/remotes/$refname):\n",
1309 "expected path: $pathname\n",
1310 " real path: $u\n",
1311 "Continuing ahead with $u\n";
1312 next;
1313 }
1314 } else {
1315 $fetch->{$pathname} = $refname;
1316 }
1317 }
1318 }
1319
1320 sub parse_revision_argument {
1321 my ($base, $head) = @_;
1322 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1323 return ($base, $head);
1324 }
1325 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1326 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1327 return ($head, $head) if ($::_revision eq 'HEAD');
1328 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1329 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1330 die "revision argument: $::_revision not understood by git-svn\n";
1331 }
1332
1333 sub fetch_all {
1334 my ($repo_id, $remotes) = @_;
1335 if (ref $repo_id) {
1336 my $gs = $repo_id;
1337 $repo_id = undef;
1338 $repo_id = $gs->{repo_id};
1339 }
1340 $remotes ||= read_all_remotes();
1341 my $remote = $remotes->{$repo_id} or
1342 die "[svn-remote \"$repo_id\"] unknown\n";
1343 my $fetch = $remote->{fetch};
1344 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1345 my (@gs, @globs);
1346 my $ra = Git::SVN::Ra->new($url);
1347 my $uuid = $ra->get_uuid;
1348 my $head = $ra->get_latest_revnum;
1349 my $base = defined $fetch ? $head : 0;
1350
1351 # read the max revs for wildcard expansion (branches/*, tags/*)
1352 foreach my $t (qw/branches tags/) {
1353 defined $remote->{$t} or next;
1354 push @globs, $remote->{$t};
1355 my $max_rev = eval { tmp_config(qw/--int --get/,
1356 "svn-remote.$repo_id.${t}-maxRev") };
1357 if (defined $max_rev && ($max_rev < $base)) {
1358 $base = $max_rev;
1359 } elsif (!defined $max_rev) {
1360 $base = 0;
1361 }
1362 }
1363
1364 if ($fetch) {
1365 foreach my $p (sort keys %$fetch) {
1366 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1367 my $lr = $gs->rev_map_max;
1368 if (defined $lr) {
1369 $base = $lr if ($lr < $base);
1370 }
1371 push @gs, $gs;
1372 }
1373 }
1374
1375 ($base, $head) = parse_revision_argument($base, $head);
1376 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1377 }
1378
1379 sub read_all_remotes {
1380 my $r = {};
1381 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1382 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
1383 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1384 $local_ref =~ s{^/}{};
1385 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1386 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1387 $r->{$1}->{url} = $2;
1388 } elsif (m!^(.+)\.(branches|tags)=
1389 (.*):refs/remotes/(.+)\s*$/!x) {
1390 my ($p, $g) = ($3, $4);
1391 my $rs = $r->{$1}->{$2} = {
1392 t => $2,
1393 remote => $1,
1394 path => Git::SVN::GlobSpec->new($p),
1395 ref => Git::SVN::GlobSpec->new($g) };
1396 if (length($rs->{ref}->{right}) != 0) {
1397 die "The '*' glob character must be the last ",
1398 "character of '$g'\n";
1399 }
1400 }
1401 }
1402 $r;
1403 }
1404
1405 sub init_vars {
1406 if (defined $_repack) {
1407 $_repack = 1000 if ($_repack <= 0);
1408 $_repack_nr = $_repack;
1409 $_repack_flags ||= '-d';
1410 }
1411 }
1412
1413 sub verify_remotes_sanity {
1414 return unless -d $ENV{GIT_DIR};
1415 my %seen;
1416 foreach (command(qw/config -l/)) {
1417 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1418 if ($seen{$1}) {
1419 die "Remote ref refs/remote/$1 is tracked by",
1420 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1421 "Please resolve this ambiguity in ",
1422 "your git configuration file before ",
1423 "continuing\n";
1424 }
1425 $seen{$1} = $_;
1426 }
1427 }
1428 }
1429
1430 # we allow more chars than remotes2config.sh...
1431 sub sanitize_remote_name {
1432 my ($name) = @_;
1433 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1434 $name;
1435 }
1436
1437 sub find_existing_remote {
1438 my ($url, $remotes) = @_;
1439 return undef if $no_reuse_existing;
1440 my $existing;
1441 foreach my $repo_id (keys %$remotes) {
1442 my $u = $remotes->{$repo_id}->{url} or next;
1443 next if $u ne $url;
1444 $existing = $repo_id;
1445 last;
1446 }
1447 $existing;
1448 }
1449
1450 sub init_remote_config {
1451 my ($self, $url, $no_write) = @_;
1452 $url =~ s!/+$!!; # strip trailing slash
1453 my $r = read_all_remotes();
1454 my $existing = find_existing_remote($url, $r);
1455 if ($existing) {
1456 unless ($no_write) {
1457 print STDERR "Using existing ",
1458 "[svn-remote \"$existing\"]\n";
1459 }
1460 $self->{repo_id} = $existing;
1461 } elsif ($_minimize_url) {
1462 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1463 $existing = find_existing_remote($min_url, $r);
1464 if ($existing) {
1465 unless ($no_write) {
1466 print STDERR "Using existing ",
1467 "[svn-remote \"$existing\"]\n";
1468 }
1469 $self->{repo_id} = $existing;
1470 }
1471 if ($min_url ne $url) {
1472 unless ($no_write) {
1473 print STDERR "Using higher level of URL: ",
1474 "$url => $min_url\n";
1475 }
1476 my $old_path = $self->{path};
1477 $self->{path} = $url;
1478 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1479 if (length $old_path) {
1480 $self->{path} .= "/$old_path";
1481 }
1482 $url = $min_url;
1483 }
1484 }
1485 my $orig_url;
1486 if (!$existing) {
1487 # verify that we aren't overwriting anything:
1488 $orig_url = eval {
1489 command_oneline('config', '--get',
1490 "svn-remote.$self->{repo_id}.url")
1491 };
1492 if ($orig_url && ($orig_url ne $url)) {
1493 die "svn-remote.$self->{repo_id}.url already set: ",
1494 "$orig_url\nwanted to set to: $url\n";
1495 }
1496 }
1497 my ($xrepo_id, $xpath) = find_ref($self->refname);
1498 if (defined $xpath) {
1499 die "svn-remote.$xrepo_id.fetch already set to track ",
1500 "$xpath:refs/remotes/", $self->refname, "\n";
1501 }
1502 unless ($no_write) {
1503 command_noisy('config',
1504 "svn-remote.$self->{repo_id}.url", $url);
1505 $self->{path} =~ s{^/}{};
1506 command_noisy('config', '--add',
1507 "svn-remote.$self->{repo_id}.fetch",
1508 "$self->{path}:".$self->refname);
1509 }
1510 $self->{url} = $url;
1511 }
1512
1513 sub find_by_url { # repos_root and, path are optional
1514 my ($class, $full_url, $repos_root, $path) = @_;
1515
1516 return undef unless defined $full_url;
1517 remove_username($full_url);
1518 remove_username($repos_root) if defined $repos_root;
1519 my $remotes = read_all_remotes();
1520 if (defined $full_url && defined $repos_root && !defined $path) {
1521 $path = $full_url;
1522 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1523 }
1524 foreach my $repo_id (keys %$remotes) {
1525 my $u = $remotes->{$repo_id}->{url} or next;
1526 remove_username($u);
1527 next if defined $repos_root && $repos_root ne $u;
1528
1529 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1530 foreach (qw/branches tags/) {
1531 resolve_local_globs($u, $fetch,
1532 $remotes->{$repo_id}->{$_});
1533 }
1534 my $p = $path;
1535 unless (defined $p) {
1536 $p = $full_url;
1537 $p =~ s#^\Q$u\E(?:/|$)## or next;
1538 }
1539 foreach my $f (keys %$fetch) {
1540 next if $f ne $p;
1541 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1542 }
1543 }
1544 undef;
1545 }
1546
1547 sub init {
1548 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1549 my $self = _new($class, $repo_id, $ref_id, $path);
1550 if (defined $url) {
1551 $self->init_remote_config($url, $no_write);
1552 }
1553 $self;
1554 }
1555
1556 sub find_ref {
1557 my ($ref_id) = @_;
1558 foreach (command(qw/config -l/)) {
1559 next unless m!^svn-remote\.(.+)\.fetch=
1560 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1561 my ($repo_id, $path, $ref) = ($1, $2, $3);
1562 if ($ref eq $ref_id) {
1563 $path = '' if ($path =~ m#^\./?#);
1564 return ($repo_id, $path);
1565 }
1566 }
1567 (undef, undef, undef);
1568 }
1569
1570 sub new {
1571 my ($class, $ref_id, $repo_id, $path) = @_;
1572 if (defined $ref_id && !defined $repo_id && !defined $path) {
1573 ($repo_id, $path) = find_ref($ref_id);
1574 if (!defined $repo_id) {
1575 die "Could not find a \"svn-remote.*.fetch\" key ",
1576 "in the repository configuration matching: ",
1577 "refs/remotes/$ref_id\n";
1578 }
1579 }
1580 my $self = _new($class, $repo_id, $ref_id, $path);
1581 if (!defined $self->{path} || !length $self->{path}) {
1582 my $fetch = command_oneline('config', '--get',
1583 "svn-remote.$repo_id.fetch",
1584 ":refs/remotes/$ref_id\$") or
1585 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1586 "\":refs/remotes/$ref_id\$\" in config\n";
1587 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1588 }
1589 $self->{url} = command_oneline('config', '--get',
1590 "svn-remote.$repo_id.url") or
1591 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1592 $self->rebuild;
1593 $self;
1594 }
1595
1596 sub refname {
1597 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1598
1599 # It cannot end with a slash /, we'll throw up on this because
1600 # SVN can't have directories with a slash in their name, either:
1601 if ($refname =~ m{/$}) {
1602 die "ref: '$refname' ends with a trailing slash, this is ",
1603 "not permitted by git nor Subversion\n";
1604 }
1605
1606 # It cannot have ASCII control character space, tilde ~, caret ^,
1607 # colon :, question-mark ?, asterisk *, space, or open bracket [
1608 # anywhere.
1609 #
1610 # Additionally, % must be escaped because it is used for escaping
1611 # and we want our escaped refname to be reversible
1612 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1613
1614 # no slash-separated component can begin with a dot .
1615 # /.* becomes /%2E*
1616 $refname =~ s{/\.}{/%2E}g;
1617
1618 # It cannot have two consecutive dots .. anywhere
1619 # .. becomes %2E%2E
1620 $refname =~ s{\.\.}{%2E%2E}g;
1621
1622 return $refname;
1623 }
1624
1625 sub desanitize_refname {
1626 my ($refname) = @_;
1627 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1628 return $refname;
1629 }
1630
1631 sub svm_uuid {
1632 my ($self) = @_;
1633 return $self->{svm}->{uuid} if $self->svm;
1634 $self->ra;
1635 unless ($self->{svm}) {
1636 die "SVM UUID not cached, and reading remotely failed\n";
1637 }
1638 $self->{svm}->{uuid};
1639 }
1640
1641 sub svm {
1642 my ($self) = @_;
1643 return $self->{svm} if $self->{svm};
1644 my $svm;
1645 # see if we have it in our config, first:
1646 eval {
1647 my $section = "svn-remote.$self->{repo_id}";
1648 $svm = {
1649 source => tmp_config('--get', "$section.svm-source"),
1650 uuid => tmp_config('--get', "$section.svm-uuid"),
1651 replace => tmp_config('--get', "$section.svm-replace"),
1652 }
1653 };
1654 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1655 $self->{svm} = $svm;
1656 }
1657 $self->{svm};
1658 }
1659
1660 sub _set_svm_vars {
1661 my ($self, $ra) = @_;
1662 return $ra if $self->svm;
1663
1664 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1665 "(svm:source, svm:uuid) ",
1666 "from the following URLs:\n" );
1667 sub read_svm_props {
1668 my ($self, $ra, $path, $r) = @_;
1669 my $props = ($ra->get_dir($path, $r))[2];
1670 my $src = $props->{'svm:source'};
1671 my $uuid = $props->{'svm:uuid'};
1672 return undef if (!$src || !$uuid);
1673
1674 chomp($src, $uuid);
1675
1676 $uuid =~ m{^[0-9a-f\-]{30,}$}
1677 or die "doesn't look right - svm:uuid is '$uuid'\n";
1678
1679 # the '!' is used to mark the repos_root!/relative/path
1680 $src =~ s{/?!/?}{/};
1681 $src =~ s{/+$}{}; # no trailing slashes please
1682 # username is of no interest
1683 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1684
1685 my $replace = $ra->{url};
1686 $replace .= "/$path" if length $path;
1687
1688 my $section = "svn-remote.$self->{repo_id}";
1689 tmp_config("$section.svm-source", $src);
1690 tmp_config("$section.svm-replace", $replace);
1691 tmp_config("$section.svm-uuid", $uuid);
1692 $self->{svm} = {
1693 source => $src,
1694 uuid => $uuid,
1695 replace => $replace
1696 };
1697 }
1698
1699 my $r = $ra->get_latest_revnum;
1700 my $path = $self->{path};
1701 my %tried;
1702 while (length $path) {
1703 unless ($tried{"$self->{url}/$path"}) {
1704 return $ra if $self->read_svm_props($ra, $path, $r);
1705 $tried{"$self->{url}/$path"} = 1;
1706 }
1707 $path =~ s#/?[^/]+$##;
1708 }
1709 die "Path: '$path' should be ''\n" if $path ne '';
1710 return $ra if $self->read_svm_props($ra, $path, $r);
1711 $tried{"$self->{url}/$path"} = 1;
1712
1713 if ($ra->{repos_root} eq $self->{url}) {
1714 die @err, (map { " $_\n" } keys %tried), "\n";
1715 }
1716
1717 # nope, make sure we're connected to the repository root:
1718 my $ok;
1719 my @tried_b;
1720 $path = $ra->{svn_path};
1721 $ra = Git::SVN::Ra->new($ra->{repos_root});
1722 while (length $path) {
1723 unless ($tried{"$ra->{url}/$path"}) {
1724 $ok = $self->read_svm_props($ra, $path, $r);
1725 last if $ok;
1726 $tried{"$ra->{url}/$path"} = 1;
1727 }
1728 $path =~ s#/?[^/]+$##;
1729 }
1730 die "Path: '$path' should be ''\n" if $path ne '';
1731 $ok ||= $self->read_svm_props($ra, $path, $r);
1732 $tried{"$ra->{url}/$path"} = 1;
1733 if (!$ok) {
1734 die @err, (map { " $_\n" } keys %tried), "\n";
1735 }
1736 Git::SVN::Ra->new($self->{url});
1737 }
1738
1739 sub svnsync {
1740 my ($self) = @_;
1741 return $self->{svnsync} if $self->{svnsync};
1742
1743 if ($self->no_metadata) {
1744 die "Can't have both 'noMetadata' and ",
1745 "'useSvnsyncProps' options set!\n";
1746 }
1747 if ($self->rewrite_root) {
1748 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1749 "options set!\n";
1750 }
1751
1752 my $svnsync;
1753 # see if we have it in our config, first:
1754 eval {
1755 my $section = "svn-remote.$self->{repo_id}";
1756 $svnsync = {
1757 url => tmp_config('--get', "$section.svnsync-url"),
1758 uuid => tmp_config('--get', "$section.svnsync-uuid"),
1759 }
1760 };
1761 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1762 return $self->{svnsync} = $svnsync;
1763 }
1764
1765 my $err = "useSvnsyncProps set, but failed to read " .
1766 "svnsync property: svn:sync-from-";
1767 my $rp = $self->ra->rev_proplist(0);
1768
1769 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1770 $url =~ m{^[a-z\+]+://} or
1771 die "doesn't look right - svn:sync-from-url is '$url'\n";
1772
1773 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1774 $uuid =~ m{^[0-9a-f\-]{30,}$} or
1775 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1776
1777 my $section = "svn-remote.$self->{repo_id}";
1778 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1779 tmp_config('--add', "$section.svnsync-url", $url);
1780 return $self->{svnsync} = { url => $url, uuid => $uuid };
1781 }
1782
1783 # this allows us to memoize our SVN::Ra UUID locally and avoid a
1784 # remote lookup (useful for 'git svn log').
1785 sub ra_uuid {
1786 my ($self) = @_;
1787 unless ($self->{ra_uuid}) {
1788 my $key = "svn-remote.$self->{repo_id}.uuid";
1789 my $uuid = eval { tmp_config('--get', $key) };
1790 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1791 $self->{ra_uuid} = $uuid;
1792 } else {
1793 die "ra_uuid called without URL\n" unless $self->{url};
1794 $self->{ra_uuid} = $self->ra->get_uuid;
1795 tmp_config('--add', $key, $self->{ra_uuid});
1796 }
1797 }
1798 $self->{ra_uuid};
1799 }
1800
1801 sub _set_repos_root {
1802 my ($self, $repos_root) = @_;
1803 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1804 $repos_root ||= $self->ra->{repos_root};
1805 tmp_config($k, $repos_root);
1806 $repos_root;
1807 }
1808
1809 sub repos_root {
1810 my ($self) = @_;
1811 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1812 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1813 }
1814
1815 sub ra {
1816 my ($self) = shift;
1817 my $ra = Git::SVN::Ra->new($self->{url});
1818 $self->_set_repos_root($ra->{repos_root});
1819 if ($self->use_svm_props && !$self->{svm}) {
1820 if ($self->no_metadata) {
1821 die "Can't have both 'noMetadata' and ",
1822 "'useSvmProps' options set!\n";
1823 } elsif ($self->use_svnsync_props) {
1824 die "Can't have both 'useSvnsyncProps' and ",
1825 "'useSvmProps' options set!\n";
1826 }
1827 $ra = $self->_set_svm_vars($ra);
1828 $self->{-want_revprops} = 1;
1829 }
1830 $ra;
1831 }
1832
1833 sub rel_path {
1834 my ($self) = @_;
1835 my $repos_root = $self->ra->{repos_root};
1836 return $self->{path} if ($self->{url} eq $repos_root);
1837 my $url = $self->{url} .
1838 (length $self->{path} ? "/$self->{path}" : $self->{path});
1839 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1840 $url;
1841 }
1842
1843 # prop_walk(PATH, REV, SUB)
1844 # -------------------------
1845 # Recursively traverse PATH at revision REV and invoke SUB for each
1846 # directory that contains a SVN property. SUB will be invoked as
1847 # follows: &SUB(gs, path, props); where `gs' is this instance of
1848 # Git::SVN, `path' the path to the directory where the properties
1849 # `props' were found. The `path' will be relative to point of checkout,
1850 # that is, if url://repo/trunk is the current Git branch, and that
1851 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
1852 # as `path' (note the trailing `/').
1853 sub prop_walk {
1854 my ($self, $path, $rev, $sub) = @_;
1855
1856 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1857 $path =~ s#^/*#/#g;
1858 my $p = $path;
1859 # Strip the irrelevant part of the path.
1860 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1861 # Ensure the path is terminated by a `/'.
1862 $p =~ s#/*$#/#;
1863
1864 # The properties contain all the internal SVN stuff nobody
1865 # (usually) cares about.
1866 my $interesting_props = 0;
1867 foreach (keys %{$props}) {
1868 # If it doesn't start with `svn:', it must be a
1869 # user-defined property.
1870 ++$interesting_props and next if $_ !~ /^svn:/;
1871 # FIXME: Fragile, if SVN adds new public properties,
1872 # this needs to be updated.
1873 ++$interesting_props if /^svn:(?:ignore|keywords|executable
1874 |eol-style|mime-type
1875 |externals|needs-lock)$/x;
1876 }
1877 &$sub($self, $p, $props) if $interesting_props;
1878
1879 foreach (sort keys %$dirent) {
1880 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1881 $self->prop_walk($path . '/' . $_, $rev, $sub);
1882 }
1883 }
1884
1885 sub last_rev { ($_[0]->last_rev_commit)[0] }
1886 sub last_commit { ($_[0]->last_rev_commit)[1] }
1887
1888 # returns the newest SVN revision number and newest commit SHA1
1889 sub last_rev_commit {
1890 my ($self) = @_;
1891 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1892 return ($self->{last_rev}, $self->{last_commit});
1893 }
1894 my $c = ::verify_ref($self->refname.'^0');
1895 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1896 my $rev = (::cmt_metadata($c))[1];
1897 if (defined $rev) {
1898 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1899 return ($rev, $c);
1900 }
1901 }
1902 my $map_path = $self->map_path;
1903 unless (-e $map_path) {
1904 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1905 return (undef, undef);
1906 }
1907 my ($rev, $commit) = $self->rev_map_max;
1908 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
1909 return ($rev, $commit);
1910 }
1911
1912 sub get_fetch_range {
1913 my ($self, $min, $max) = @_;
1914 $max ||= $self->ra->get_latest_revnum;
1915 $min ||= $self->rev_map_max;
1916 (++$min, $max);
1917 }
1918
1919 sub tmp_config {
1920 my (@args) = @_;
1921 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1922 my $config = "$ENV{GIT_DIR}/svn/.metadata";
1923 if (! -f $config && -f $old_def_config) {
1924 rename $old_def_config, $config or
1925 die "Failed rename $old_def_config => $config: $!\n";
1926 }
1927 my $old_config = $ENV{GIT_CONFIG};
1928 $ENV{GIT_CONFIG} = $config;
1929 $@ = undef;
1930 my @ret = eval {
1931 unless (-f $config) {
1932 mkfile($config);
1933 open my $fh, '>', $config or
1934 die "Can't open $config: $!\n";
1935 print $fh "; This file is used internally by ",
1936 "git-svn\n" or die
1937 "Couldn't write to $config: $!\n";
1938 print $fh "; You should not have to edit it\n" or
1939 die "Couldn't write to $config: $!\n";
1940 close $fh or die "Couldn't close $config: $!\n";
1941 }
1942 command('config', @args);
1943 };
1944 my $err = $@;
1945 if (defined $old_config) {
1946 $ENV{GIT_CONFIG} = $old_config;
1947 } else {
1948 delete $ENV{GIT_CONFIG};
1949 }
1950 die $err if $err;
1951 wantarray ? @ret : $ret[0];
1952 }
1953
1954 sub tmp_index_do {
1955 my ($self, $sub) = @_;
1956 my $old_index = $ENV{GIT_INDEX_FILE};
1957 $ENV{GIT_INDEX_FILE} = $self->{index};
1958 $@ = undef;
1959 my @ret = eval {
1960 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1961 mkpath([$dir]) unless -d $dir;
1962 &$sub;
1963 };
1964 my $err = $@;
1965 if (defined $old_index) {
1966 $ENV{GIT_INDEX_FILE} = $old_index;
1967 } else {
1968 delete $ENV{GIT_INDEX_FILE};
1969 }
1970 die $err if $err;
1971 wantarray ? @ret : $ret[0];
1972 }
1973
1974 sub assert_index_clean {
1975 my ($self, $treeish) = @_;
1976
1977 $self->tmp_index_do(sub {
1978 command_noisy('read-tree', $treeish) unless -e $self->{index};
1979 my $x = command_oneline('write-tree');
1980 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1981 /^tree ($::sha1)/mo);
1982 return if $y eq $x;
1983
1984 warn "Index mismatch: $y != $x\nrereading $treeish\n";
1985 unlink $self->{index} or die "unlink $self->{index}: $!\n";
1986 command_noisy('read-tree', $treeish);
1987 $x = command_oneline('write-tree');
1988 if ($y ne $x) {
1989 ::fatal "trees ($treeish) $y != $x\n",
1990 "Something is seriously wrong...";
1991 }
1992 });
1993 }
1994
1995 sub get_commit_parents {
1996 my ($self, $log_entry) = @_;
1997 my (%seen, @ret, @tmp);
1998 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1999 if (my $ip = $self->{inject_parents}) {
2000 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2001 push @tmp, $commit;
2002 }
2003 }
2004 if (my $cur = ::verify_ref($self->refname.'^0')) {
2005 push @tmp, $cur;
2006 }
2007 if (my $ipd = $self->{inject_parents_dcommit}) {
2008 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2009 push @tmp, @$commit;
2010 }
2011 }
2012 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2013 while (my $p = shift @tmp) {
2014 next if $seen{$p};
2015 $seen{$p} = 1;
2016 push @ret, $p;
2017 # MAXPARENT is defined to 16 in commit-tree.c:
2018 last if @ret >= 16;
2019 }
2020 if (@tmp) {
2021 die "r$log_entry->{revision}: No room for parents:\n\t",
2022 join("\n\t", @tmp), "\n";
2023 }
2024 @ret;
2025 }
2026
2027 sub rewrite_root {
2028 my ($self) = @_;
2029 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2030 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2031 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2032 if ($rwr) {
2033 $rwr =~ s#/+$##;
2034 if ($rwr !~ m#^[a-z\+]+://#) {
2035 die "$rwr is not a valid URL (key: $k)\n";
2036 }
2037 }
2038 $self->{-rewrite_root} = $rwr;
2039 }
2040
2041 sub metadata_url {
2042 my ($self) = @_;
2043 ($self->rewrite_root || $self->{url}) .
2044 (length $self->{path} ? '/' . $self->{path} : '');
2045 }
2046
2047 sub full_url {
2048 my ($self) = @_;
2049 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2050 }
2051
2052 sub do_git_commit {
2053 my ($self, $log_entry) = @_;
2054 my $lr = $self->last_rev;
2055 if (defined $lr && $lr >= $log_entry->{revision}) {
2056 die "Last fetched revision of ", $self->refname,
2057 " was r$lr, but we are about to fetch: ",
2058 "r$log_entry->{revision}!\n";
2059 }
2060 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2061 croak "$log_entry->{revision} = $c already exists! ",
2062 "Why are we refetching it?\n";
2063 }
2064 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2065 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2066 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2067
2068 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2069 ? $log_entry->{commit_name}
2070 : $log_entry->{name};
2071 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2072 ? $log_entry->{commit_email}
2073 : $log_entry->{email};
2074
2075 my $tree = $log_entry->{tree};
2076 if (!defined $tree) {
2077 $tree = $self->tmp_index_do(sub {
2078 command_oneline('write-tree') });
2079 }
2080 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2081
2082 my @exec = ('git-commit-tree', $tree);
2083 foreach ($self->get_commit_parents($log_entry)) {
2084 push @exec, '-p', $_;
2085 }
2086 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2087 or croak $!;
2088 print $msg_fh $log_entry->{log} or croak $!;
2089 unless ($self->no_metadata) {
2090 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2091 or croak $!;
2092 }
2093 $msg_fh->flush == 0 or croak $!;
2094 close $msg_fh or croak $!;
2095 chomp(my $commit = do { local $/; <$out_fh> });
2096 close $out_fh or croak $!;
2097 waitpid $pid, 0;
2098 croak $? if $?;
2099 if ($commit !~ /^$::sha1$/o) {
2100 die "Failed to commit, invalid sha1: $commit\n";
2101 }
2102
2103 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2104
2105 $self->{last_rev} = $log_entry->{revision};
2106 $self->{last_commit} = $commit;
2107 print "r$log_entry->{revision}";
2108 if (defined $log_entry->{svm_revision}) {
2109 print " (\@$log_entry->{svm_revision})";
2110 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2111 0, $self->svm_uuid);
2112 }
2113 print " = $commit ($self->{ref_id})\n";
2114 if (defined $_repack && (--$_repack_nr == 0)) {
2115 $_repack_nr = $_repack;
2116 # repack doesn't use any arguments with spaces in them, does it?
2117 print "Running git repack $_repack_flags ...\n";
2118 command_noisy('repack', split(/\s+/, $_repack_flags));
2119 print "Done repacking\n";
2120 }
2121 return $commit;
2122 }
2123
2124 sub match_paths {
2125 my ($self, $paths, $r) = @_;
2126 return 1 if $self->{path} eq '';
2127 if (my $path = $paths->{"/$self->{path}"}) {
2128 return ($path->{action} eq 'D') ? 0 : 1;
2129 }
2130 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2131 if (grep /$self->{path_regex}/, keys %$paths) {
2132 return 1;
2133 }
2134 my $c = '';
2135 foreach (split m#/#, $self->{path}) {
2136 $c .= "/$_";
2137 next unless ($paths->{$c} &&
2138 ($paths->{$c}->{action} =~ /^[AR]$/));
2139 if ($self->ra->check_path($self->{path}, $r) ==
2140 $SVN::Node::dir) {
2141 return 1;
2142 }
2143 }
2144 return 0;
2145 }
2146
2147 sub find_parent_branch {
2148 my ($self, $paths, $rev) = @_;
2149 return undef unless $self->follow_parent;
2150 unless (defined $paths) {
2151 my $err_handler = $SVN::Error::handler;
2152 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2153 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2154 $paths =
2155 Git::SVN::Ra::dup_changed_paths($_[0]) });
2156 $SVN::Error::handler = $err_handler;
2157 }
2158 return undef unless defined $paths;
2159
2160 # look for a parent from another branch:
2161 my @b_path_components = split m#/#, $self->rel_path;
2162 my @a_path_components;
2163 my $i;
2164 while (@b_path_components) {
2165 $i = $paths->{'/'.join('/', @b_path_components)};
2166 last if $i && defined $i->{copyfrom_path};
2167 unshift(@a_path_components, pop(@b_path_components));
2168 }
2169 return undef unless defined $i && defined $i->{copyfrom_path};
2170 my $branch_from = $i->{copyfrom_path};
2171 if (@a_path_components) {
2172 print STDERR "branch_from: $branch_from => ";
2173 $branch_from .= '/'.join('/', @a_path_components);
2174 print STDERR $branch_from, "\n";
2175 }
2176 my $r = $i->{copyfrom_rev};
2177 my $repos_root = $self->ra->{repos_root};
2178 my $url = $self->ra->{url};
2179 my $new_url = $repos_root . $branch_from;
2180 print STDERR "Found possible branch point: ",
2181 "$new_url => ", $self->full_url, ", $r\n";
2182 $branch_from =~ s#^/##;
2183 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2184 unless ($gs) {
2185 my $ref_id = $self->{ref_id};
2186 $ref_id =~ s/\@\d+$//;
2187 $ref_id .= "\@$r";
2188 # just grow a tail if we're not unique enough :x
2189 $ref_id .= '-' while find_ref($ref_id);
2190 print STDERR "Initializing parent: $ref_id\n";
2191 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
2192 }
2193 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2194 if (!defined $r0 || !defined $parent) {
2195 my ($base, $head) = parse_revision_argument(0, $r);
2196 if ($base <= $r) {
2197 $gs->fetch($base, $r);
2198 }
2199 ($r0, $parent) = $gs->last_rev_commit;
2200 }
2201 if (defined $r0 && defined $parent) {
2202 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2203 my $ed;
2204 if ($self->ra->can_do_switch) {
2205 $self->assert_index_clean($parent);
2206 print STDERR "Following parent with do_switch\n";
2207 # do_switch works with svn/trunk >= r22312, but that
2208 # is not included with SVN 1.4.3 (the latest version
2209 # at the moment), so we can't rely on it
2210 $self->{last_commit} = $parent;
2211 $ed = SVN::Git::Fetcher->new($self);
2212 $gs->ra->gs_do_switch($r0, $rev, $gs,
2213 $self->full_url, $ed)
2214 or die "SVN connection failed somewhere...\n";
2215 } elsif ($self->ra->trees_match($new_url, $r0,
2216 $self->full_url, $rev)) {
2217 print STDERR "Trees match:\n",
2218 " $new_url\@$r0\n",
2219 " ${\$self->full_url}\@$rev\n",
2220 "Following parent with no changes\n";
2221 $self->tmp_index_do(sub {
2222 command_noisy('read-tree', $parent);
2223 });
2224 $self->{last_commit} = $parent;
2225 } else {
2226 print STDERR "Following parent with do_update\n";
2227 $ed = SVN::Git::Fetcher->new($self);
2228 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2229 or die "SVN connection failed somewhere...\n";
2230 }
2231 print STDERR "Successfully followed parent\n";
2232 return $self->make_log_entry($rev, [$parent], $ed);
2233 }
2234 return undef;
2235 }
2236
2237 sub do_fetch {
2238 my ($self, $paths, $rev) = @_;
2239 my $ed;
2240 my ($last_rev, @parents);
2241 if (my $lc = $self->last_commit) {
2242 # we can have a branch that was deleted, then re-added
2243 # under the same name but copied from another path, in
2244 # which case we'll have multiple parents (we don't
2245 # want to break the original ref, nor lose copypath info):
2246 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2247 push @{$log_entry->{parents}}, $lc;
2248 return $log_entry;
2249 }
2250 $ed = SVN::Git::Fetcher->new($self);
2251 $last_rev = $self->{last_rev};
2252 $ed->{c} = $lc;
2253 @parents = ($lc);
2254 } else {
2255 $last_rev = $rev;
2256 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2257 return $log_entry;
2258 }
2259 $ed = SVN::Git::Fetcher->new($self);
2260 }
2261 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2262 die "SVN connection failed somewhere...\n";
2263 }
2264 $self->make_log_entry($rev, \@parents, $ed);
2265 }
2266
2267 sub get_untracked {
2268 my ($self, $ed) = @_;
2269 my @out;
2270 my $h = $ed->{empty};
2271 foreach (sort keys %$h) {
2272 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2273 push @out, " $act: " . uri_encode($_);
2274 warn "W: $act: $_\n";
2275 }
2276 foreach my $t (qw/dir_prop file_prop/) {
2277 $h = $ed->{$t} or next;
2278 foreach my $path (sort keys %$h) {
2279 my $ppath = $path eq '' ? '.' : $path;
2280 foreach my $prop (sort keys %{$h->{$path}}) {
2281 next if $SKIP_PROP{$prop};
2282 my $v = $h->{$path}->{$prop};
2283 my $t_ppath_prop = "$t: " .
2284 uri_encode($ppath) . ' ' .
2285 uri_encode($prop);
2286 if (defined $v) {
2287 push @out, " +$t_ppath_prop " .
2288 uri_encode($v);
2289 } else {
2290 push @out, " -$t_ppath_prop";
2291 }
2292 }
2293 }
2294 }
2295 foreach my $t (qw/absent_file absent_directory/) {
2296 $h = $ed->{$t} or next;
2297 foreach my $parent (sort keys %$h) {
2298 foreach my $path (sort @{$h->{$parent}}) {
2299 push @out, " $t: " .
2300 uri_encode("$parent/$path");
2301 warn "W: $t: $parent/$path ",
2302 "Insufficient permissions?\n";
2303 }
2304 }
2305 }
2306 \@out;
2307 }
2308
2309 sub parse_svn_date {
2310 my $date = shift || return '+0000 1970-01-01 00:00:00';
2311 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2312 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2313 croak "Unable to parse date: $date\n";
2314 "+0000 $Y-$m-$d $H:$M:$S";
2315 }
2316
2317 sub check_author {
2318 my ($author) = @_;
2319 if (!defined $author || length $author == 0) {
2320 $author = '(no author)';
2321 }
2322 if (defined $::_authors && ! defined $::users{$author}) {
2323 die "Author: $author not defined in $::_authors file\n";
2324 }
2325 $author;
2326 }
2327
2328 sub make_log_entry {
2329 my ($self, $rev, $parents, $ed) = @_;
2330 my $untracked = $self->get_untracked($ed);
2331
2332 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2333 print $un "r$rev\n" or croak $!;
2334 print $un $_, "\n" foreach @$untracked;
2335 my %log_entry = ( parents => $parents || [], revision => $rev,
2336 log => '');
2337
2338 my $headrev;
2339 my $logged = delete $self->{logged_rev_props};
2340 if (!$logged || $self->{-want_revprops}) {
2341 my $rp = $self->ra->rev_proplist($rev);
2342 foreach (sort keys %$rp) {
2343 my $v = $rp->{$_};
2344 if (/^svn:(author|date|log)$/) {
2345 $log_entry{$1} = $v;
2346 } elsif ($_ eq 'svm:headrev') {
2347 $headrev = $v;
2348 } else {
2349 print $un " rev_prop: ", uri_encode($_), ' ',
2350 uri_encode($v), "\n";
2351 }
2352 }
2353 } else {
2354 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2355 }
2356 close $un or croak $!;
2357
2358 $log_entry{date} = parse_svn_date($log_entry{date});
2359 $log_entry{log} .= "\n";
2360 my $author = $log_entry{author} = check_author($log_entry{author});
2361 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2362 : ($author, undef);
2363
2364 my ($commit_name, $commit_email) = ($name, $email);
2365 if ($_use_log_author) {
2366 if ($log_entry{log} =~ /From:\s+(.*?)\s+<(.*)>\s*\n/) {
2367 ($name, $email) = ($1, $2);
2368 } elsif ($log_entry{log} =~
2369 /Signed-off-by:\s+(.*?)\s+<(.*)>\s*\n/) {
2370 ($name, $email) = ($1, $2);
2371 }
2372 }
2373 if (defined $headrev && $self->use_svm_props) {
2374 if ($self->rewrite_root) {
2375 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2376 "options set!\n";
2377 }
2378 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2379 # we don't want "SVM: initializing mirror for junk" ...
2380 return undef if $r == 0;
2381 my $svm = $self->svm;
2382 if ($uuid ne $svm->{uuid}) {
2383 die "UUID mismatch on SVM path:\n",
2384 "expected: $svm->{uuid}\n",
2385 " got: $uuid\n";
2386 }
2387 my $full_url = $self->full_url;
2388 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2389 die "Failed to replace '$svm->{replace}' with ",
2390 "'$svm->{source}' in $full_url\n";
2391 # throw away username for storing in records
2392 remove_username($full_url);
2393 $log_entry{metadata} = "$full_url\@$r $uuid";
2394 $log_entry{svm_revision} = $r;
2395 $email ||= "$author\@$uuid";
2396 $commit_email ||= "$author\@$uuid";
2397 } elsif ($self->use_svnsync_props) {
2398 my $full_url = $self->svnsync->{url};
2399 $full_url .= "/$self->{path}" if length $self->{path};
2400 remove_username($full_url);
2401 my $uuid = $self->svnsync->{uuid};
2402 $log_entry{metadata} = "$full_url\@$rev $uuid";
2403 $email ||= "$author\@$uuid";
2404 $commit_email ||= "$author\@$uuid";
2405 } else {
2406 my $url = $self->metadata_url;
2407 remove_username($url);
2408 $log_entry{metadata} = "$url\@$rev " .
2409 $self->ra->get_uuid;
2410 $email ||= "$author\@" . $self->ra->get_uuid;
2411 $commit_email ||= "$author\@" . $self->ra->get_uuid;
2412 }
2413 $log_entry{name} = $name;
2414 $log_entry{email} = $email;
2415 $log_entry{commit_name} = $commit_name;
2416 $log_entry{commit_email} = $commit_email;
2417 \%log_entry;
2418 }
2419
2420 sub fetch {
2421 my ($self, $min_rev, $max_rev, @parents) = @_;
2422 my ($last_rev, $last_commit) = $self->last_rev_commit;
2423 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2424 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2425 }
2426
2427 sub set_tree_cb {
2428 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2429 $self->{inject_parents} = { $rev => $tree };
2430 $self->fetch(undef, undef);
2431 }
2432
2433 sub set_tree {
2434 my ($self, $tree) = (shift, shift);
2435 my $log_entry = ::get_commit_entry($tree);
2436 unless ($self->{last_rev}) {
2437 fatal("Must have an existing revision to commit");
2438 }
2439 my %ed_opts = ( r => $self->{last_rev},
2440 log => $log_entry->{log},
2441 ra => $self->ra,
2442 tree_a => $self->{last_commit},
2443 tree_b => $tree,
2444 editor_cb => sub {
2445 $self->set_tree_cb($log_entry, $tree, @_) },
2446 svn_path => $self->{path} );
2447 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2448 print "No changes\nr$self->{last_rev} = $tree\n";
2449 }
2450 }
2451
2452 sub rebuild_from_rev_db {
2453 my ($self, $path) = @_;
2454 my $r = -1;
2455 open my $fh, '<', $path or croak "open: $!";
2456 while (<$fh>) {
2457 length($_) == 41 or croak "inconsistent size in ($_) != 41";
2458 chomp($_);
2459 ++$r;
2460 next if $_ eq ('0' x 40);
2461 $self->rev_map_set($r, $_);
2462 print "r$r = $_\n";
2463 }
2464 close $fh or croak "close: $!";
2465 unlink $path or croak "unlink: $!";
2466 }
2467
2468 sub rebuild {
2469 my ($self) = @_;
2470 my $map_path = $self->map_path;
2471 return if (-e $map_path && ! -z $map_path);
2472 return unless ::verify_ref($self->refname.'^0');
2473 if ($self->use_svm_props || $self->no_metadata) {
2474 my $rev_db = $self->rev_db_path;
2475 $self->rebuild_from_rev_db($rev_db);
2476 if ($self->use_svm_props) {
2477 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2478 $self->rebuild_from_rev_db($svm_rev_db);
2479 }
2480 $self->unlink_rev_db_symlink;
2481 return;
2482 }
2483 print "Rebuilding $map_path ...\n";
2484 my ($log, $ctx) =
2485 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2486 $self->refname, '--');
2487 my $full_url = $self->full_url;
2488 remove_username($full_url);
2489 my $svn_uuid = $self->ra_uuid;
2490 my $c;
2491 while (<$log>) {
2492 if ( m{^commit ($::sha1)$} ) {
2493 $c = $1;
2494 next;
2495 }
2496 next unless s{^\s*(git-svn-id:)}{$1};
2497 my ($url, $rev, $uuid) = ::extract_metadata($_);
2498 remove_username($url);
2499
2500 # ignore merges (from set-tree)
2501 next if (!defined $rev || !$uuid);
2502
2503 # if we merged or otherwise started elsewhere, this is
2504 # how we break out of it
2505 if (($uuid ne $svn_uuid) ||
2506 ($full_url && $url && ($url ne $full_url))) {
2507 next;
2508 }
2509
2510 $self->rev_map_set($rev, $c);
2511 print "r$rev = $c\n";
2512 }
2513 command_close_pipe($log, $ctx);
2514 print "Done rebuilding $map_path\n";
2515 my $rev_db_path = $self->rev_db_path;
2516 if (-f $self->rev_db_path) {
2517 unlink $self->rev_db_path or croak "unlink: $!";
2518 }
2519 $self->unlink_rev_db_symlink;
2520 }
2521
2522 # rev_map:
2523 # Tie::File seems to be prone to offset errors if revisions get sparse,
2524 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
2525 # one of my favorite modules is out :< Next up would be one of the DBM
2526 # modules, but I'm not sure which is most portable...
2527 #
2528 # This is the replacement for the rev_db format, which was too big
2529 # and inefficient for large repositories with a lot of sparse history
2530 # (mainly tags)
2531 #
2532 # The format is this:
2533 # - 24 bytes for every record,
2534 # * 4 bytes for the integer representing an SVN revision number
2535 # * 20 bytes representing the sha1 of a git commit
2536 # - No empty padding records like the old format
2537 # - new records are written append-only since SVN revision numbers
2538 # increase monotonically
2539 # - lookups on SVN revision number are done via a binary search
2540 # - Piping the file to xxd(1) -c24 is a good way of dumping it for
2541 # viewing or editing, should the need ever arise.
2542 #
2543 # These files are disposable unless noMetadata or useSvmProps is set
2544
2545 sub _rev_map_set {
2546 my ($fh, $rev, $commit) = @_;
2547
2548 my $size = (stat($fh))[7];
2549 ($size % 24) == 0 or croak "inconsistent size: $size";
2550
2551 if ($size > 0) {
2552 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2553 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2554 $read == 24 or croak "read only $read bytes (!= 24)";
2555 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2556 if ($last_rev >= $rev) {
2557 croak "last_rev is higher!: $last_rev >= $rev";
2558 }
2559 }
2560 sysseek($fh, 0, SEEK_END) or croak "seek: $!";
2561 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2562 croak "write: $!";
2563 }
2564
2565 sub mkfile {
2566 my ($path) = @_;
2567 unless (-e $path) {
2568 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2569 mkpath([$dir]) unless -d $dir;
2570 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2571 close $fh or die "Couldn't close (create) $path: $!\n";
2572 }
2573 }
2574
2575 sub rev_map_set {
2576 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2577 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2578 my $db = $self->map_path($uuid);
2579 my $db_lock = "$db.lock";
2580 my $sig;
2581 if ($update_ref) {
2582 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2583 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2584 }
2585 mkfile($db);
2586
2587 $LOCKFILES{$db_lock} = 1;
2588 my $sync;
2589 # both of these options make our .rev_db file very, very important
2590 # and we can't afford to lose it because rebuild() won't work
2591 if ($self->use_svm_props || $self->no_metadata) {
2592 $sync = 1;
2593 copy($db, $db_lock) or die "rev_map_set(@_): ",
2594 "Failed to copy: ",
2595 "$db => $db_lock ($!)\n";
2596 } else {
2597 rename $db, $db_lock or die "rev_map_set(@_): ",
2598 "Failed to rename: ",
2599 "$db => $db_lock ($!)\n";
2600 }
2601
2602 sysopen(my $fh, $db_lock, O_RDWR | O_APPEND | O_CREAT)
2603 or croak "Couldn't open $db_lock: $!\n";
2604 _rev_map_set($fh, $rev, $commit);
2605 if ($sync) {
2606 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2607 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2608 }
2609 close $fh or croak $!;
2610 if ($update_ref) {
2611 $_head = $self;
2612 command_noisy('update-ref', '-m', "r$rev",
2613 $self->refname, $commit);
2614 }
2615 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2616 "$db_lock => $db ($!)\n";
2617 delete $LOCKFILES{$db_lock};
2618 if ($update_ref) {
2619 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2620 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2621 kill $sig, $$ if defined $sig;
2622 }
2623 }
2624
2625 sub rev_map_max {
2626 my ($self) = @_;
2627 $self->rebuild;
2628 my $map_path = $self->map_path;
2629 stat $map_path or return wantarray ? (0, undef) : 0;
2630 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2631 my $size = (stat($fh))[7];
2632 ($size % 24) == 0 or croak "inconsistent size: $size";
2633
2634 if ($size == 0) {
2635 close $fh or croak "close: $!";
2636 return wantarray ? (0, undef) : 0;
2637 }
2638
2639 sysseek($fh, -24, SEEK_END);
2640 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2641 close $fh or croak "close: $!";
2642 my ($r, $c) = unpack(rev_map_fmt, $buf);
2643 wantarray ? ($r, $c) : $r;
2644 }
2645
2646 sub rev_map_get {
2647 my ($self, $rev, $uuid) = @_;
2648 my $map_path = $self->map_path($uuid);
2649 return undef unless -e $map_path;
2650
2651 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2652 my $size = (stat($fh))[7];
2653 ($size % 24) == 0 or croak "inconsistent size: $size";
2654
2655 if ($size == 0) {
2656 close $fh or croak "close: $fh";
2657 return undef;
2658 }
2659
2660 my ($l, $u) = (0, $size - 24);
2661 my ($r, $c, $buf);
2662
2663 while ($l <= $u) {
2664 my $i = int(($l/24 + $u/24) / 2) * 24;
2665 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2666 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2667 my ($r, $c) = unpack('NH40', $buf);
2668
2669 if ($r < $rev) {
2670 $l = $i + 24;
2671 } elsif ($r > $rev) {
2672 $u = $i - 24;
2673 } else { # $r == $rev
2674 close($fh) or croak "close: $!";
2675 return $c;
2676 }
2677 }
2678 close($fh) or croak "close: $!";
2679 undef;
2680 }
2681
2682 # Finds the first svn revision that exists on (if $eq_ok is true) or
2683 # before $rev for the current branch. It will not search any lower
2684 # than $min_rev. Returns the git commit hash and svn revision number
2685 # if found, else (undef, undef).
2686 sub find_rev_before {
2687 my ($self, $rev, $eq_ok, $min_rev) = @_;
2688 --$rev unless $eq_ok;
2689 $min_rev ||= 1;
2690 while ($rev >= $min_rev) {
2691 if (my $c = $self->rev_map_get($rev)) {
2692 return ($rev, $c);
2693 }
2694 --$rev;
2695 }
2696 return (undef, undef);
2697 }
2698
2699 # Finds the first svn revision that exists on (if $eq_ok is true) or
2700 # after $rev for the current branch. It will not search any higher
2701 # than $max_rev. Returns the git commit hash and svn revision number
2702 # if found, else (undef, undef).
2703 sub find_rev_after {
2704 my ($self, $rev, $eq_ok, $max_rev) = @_;
2705 ++$rev unless $eq_ok;
2706 $max_rev ||= $self->rev_map_max;
2707 while ($rev <= $max_rev) {
2708 if (my $c = $self->rev_map_get($rev)) {
2709 return ($rev, $c);
2710 }
2711 ++$rev;
2712 }
2713 return (undef, undef);
2714 }
2715
2716 sub _new {
2717 my ($class, $repo_id, $ref_id, $path) = @_;
2718 unless (defined $repo_id && length $repo_id) {
2719 $repo_id = $Git::SVN::default_repo_id;
2720 }
2721 unless (defined $ref_id && length $ref_id) {
2722 $_[2] = $ref_id = $Git::SVN::default_ref_id;
2723 }
2724 $_[1] = $repo_id = sanitize_remote_name($repo_id);
2725 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2726 $_[3] = $path = '' unless (defined $path);
2727 mkpath(["$ENV{GIT_DIR}/svn"]);
2728 bless {
2729 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2730 path => $path, config => "$ENV{GIT_DIR}/svn/config",
2731 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
2732 }
2733
2734 # for read-only access of old .rev_db formats
2735 sub unlink_rev_db_symlink {
2736 my ($self) = @_;
2737 my $link = $self->rev_db_path;
2738 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2739 if (-l $link) {
2740 unlink $link or croak "unlink: $link failed!";
2741 }
2742 }
2743
2744 sub rev_db_path {
2745 my ($self, $uuid) = @_;
2746 my $db_path = $self->map_path($uuid);
2747 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2748 or croak "map_path: $db_path does not contain '/.rev_map.' !";
2749 $db_path;
2750 }
2751
2752 # the new replacement for .rev_db
2753 sub map_path {
2754 my ($self, $uuid) = @_;
2755 $uuid ||= $self->ra_uuid;
2756 "$self->{map_root}.$uuid";
2757 }
2758
2759 sub uri_encode {
2760 my ($f) = @_;
2761 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2762 $f
2763 }
2764
2765 sub remove_username {
2766 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2767 }
2768
2769 package Git::SVN::Prompt;
2770 use strict;
2771 use warnings;
2772 require SVN::Core;
2773 use vars qw/$_no_auth_cache $_username/;
2774
2775 sub simple {
2776 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2777 $may_save = undef if $_no_auth_cache;
2778 $default_username = $_username if defined $_username;
2779 if (defined $default_username && length $default_username) {
2780 if (defined $realm && length $realm) {
2781 print STDERR "Authentication realm: $realm\n";
2782 STDERR->flush;
2783 }
2784 $cred->username($default_username);
2785 } else {
2786 username($cred, $realm, $may_save, $pool);
2787 }
2788 $cred->password(_read_password("Password for '" .
2789 $cred->username . "': ", $realm));
2790 $cred->may_save($may_save);
2791 $SVN::_Core::SVN_NO_ERROR;
2792 }
2793
2794 sub ssl_server_trust {
2795 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2796 $may_save = undef if $_no_auth_cache;
2797 print STDERR "Error validating server certificate for '$realm':\n";
2798 {
2799 no warnings 'once';
2800 # All variables SVN::Auth::SSL::* are used only once,
2801 # so we're shutting up Perl warnings about this.
2802 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2803 print STDERR " - The certificate is not issued ",
2804 "by a trusted authority. Use the\n",
2805 " fingerprint to validate ",
2806 "the certificate manually!\n";
2807 }
2808 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2809 print STDERR " - The certificate hostname ",
2810 "does not match.\n";
2811 }
2812 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2813 print STDERR " - The certificate is not yet valid.\n";
2814 }
2815 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2816 print STDERR " - The certificate has expired.\n";
2817 }
2818 if ($failures & $SVN::Auth::SSL::OTHER) {
2819 print STDERR " - The certificate has ",
2820 "an unknown error.\n";
2821 }
2822 } # no warnings 'once'
2823 printf STDERR
2824 "Certificate information:\n".
2825 " - Hostname: %s\n".
2826 " - Valid: from %s until %s\n".
2827 " - Issuer: %s\n".
2828 " - Fingerprint: %s\n",
2829 map $cert_info->$_, qw(hostname valid_from valid_until
2830 issuer_dname fingerprint);
2831 my $choice;
2832 prompt:
2833 print STDERR $may_save ?
2834 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2835 "(R)eject or accept (t)emporarily? ";
2836 STDERR->flush;
2837 $choice = lc(substr(<STDIN> || 'R', 0, 1));
2838 if ($choice =~ /^t$/i) {
2839 $cred->may_save(undef);
2840 } elsif ($choice =~ /^r$/i) {
2841 return -1;
2842 } elsif ($may_save && $choice =~ /^p$/i) {
2843 $cred->may_save($may_save);
2844 } else {
2845 goto prompt;
2846 }
2847 $cred->accepted_failures($failures);
2848 $SVN::_Core::SVN_NO_ERROR;
2849 }
2850
2851 sub ssl_client_cert {
2852 my ($cred, $realm, $may_save, $pool) = @_;
2853 $may_save = undef if $_no_auth_cache;
2854 print STDERR "Client certificate filename: ";
2855 STDERR->flush;
2856 chomp(my $filename = <STDIN>);
2857 $cred->cert_file($filename);
2858 $cred->may_save($may_save);
2859 $SVN::_Core::SVN_NO_ERROR;
2860 }
2861
2862 sub ssl_client_cert_pw {
2863 my ($cred, $realm, $may_save, $pool) = @_;
2864 $may_save = undef if $_no_auth_cache;
2865 $cred->password(_read_password("Password: ", $realm));
2866 $cred->may_save($may_save);
2867 $SVN::_Core::SVN_NO_ERROR;
2868 }
2869
2870 sub username {
2871 my ($cred, $realm, $may_save, $pool) = @_;
2872 $may_save = undef if $_no_auth_cache;
2873 if (defined $realm && length $realm) {
2874 print STDERR "Authentication realm: $realm\n";
2875 }
2876 my $username;
2877 if (defined $_username) {
2878 $username = $_username;
2879 } else {
2880 print STDERR "Username: ";
2881 STDERR->flush;
2882 chomp($username = <STDIN>);
2883 }
2884 $cred->username($username);
2885 $cred->may_save($may_save);
2886 $SVN::_Core::SVN_NO_ERROR;
2887 }
2888
2889 sub _read_password {
2890 my ($prompt, $realm) = @_;
2891 print STDERR $prompt;
2892 STDERR->flush;
2893 require Term::ReadKey;
2894 Term::ReadKey::ReadMode('noecho');
2895 my $password = '';
2896 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2897 last if $key =~ /[\012\015]/; # \n\r
2898 $password .= $key;
2899 }
2900 Term::ReadKey::ReadMode('restore');
2901 print STDERR "\n";
2902 STDERR->flush;
2903 $password;
2904 }
2905
2906 package SVN::Git::Fetcher;
2907 use vars qw/@ISA/;
2908 use strict;
2909 use warnings;
2910 use Carp qw/croak/;
2911 use IO::File qw//;
2912
2913 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
2914 sub new {
2915 my ($class, $git_svn) = @_;
2916 my $self = SVN::Delta::Editor->new;
2917 bless $self, $class;
2918 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2919 $self->{empty} = {};
2920 $self->{dir_prop} = {};
2921 $self->{file_prop} = {};
2922 $self->{absent_dir} = {};
2923 $self->{absent_file} = {};
2924 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2925 $self;
2926 }
2927
2928 sub set_path_strip {
2929 my ($self, $path) = @_;
2930 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2931 }
2932
2933 sub open_root {
2934 { path => '' };
2935 }
2936
2937 sub open_directory {
2938 my ($self, $path, $pb, $rev) = @_;
2939 { path => $path };
2940 }
2941
2942 sub git_path {
2943 my ($self, $path) = @_;
2944 if ($self->{path_strip}) {
2945 $path =~ s!$self->{path_strip}!! or
2946 die "Failed to strip path '$path' ($self->{path_strip})\n";
2947 }
2948 $path;
2949 }
2950
2951 sub delete_entry {
2952 my ($self, $path, $rev, $pb) = @_;
2953
2954 my $gpath = $self->git_path($path);
2955 return undef if ($gpath eq '');
2956
2957 # remove entire directories.
2958 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2959 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2960 -r --name-only -z/,
2961 $self->{c}, '--', $gpath);
2962 local $/ = "\0";
2963 while (<$ls>) {
2964 chomp;
2965 $self->{gii}->remove($_);
2966 print "\tD\t$_\n" unless $::_q;
2967 }
2968 print "\tD\t$gpath/\n" unless $::_q;
2969 command_close_pipe($ls, $ctx);
2970 $self->{empty}->{$path} = 0
2971 } else {
2972 $self->{gii}->remove($gpath);
2973 print "\tD\t$gpath\n" unless $::_q;
2974 }
2975 undef;
2976 }
2977
2978 sub open_file {
2979 my ($self, $path, $pb, $rev) = @_;
2980 my $gpath = $self->git_path($path);
2981 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2982 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2983 unless (defined $mode && defined $blob) {
2984 die "$path was not found in commit $self->{c} (r$rev)\n";
2985 }
2986 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2987 pool => SVN::Pool->new, action => 'M' };
2988 }
2989
2990 sub add_file {
2991 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2992 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2993 delete $self->{empty}->{$dir};
2994 { path => $path, mode_a => 100644, mode_b => 100644,
2995 pool => SVN::Pool->new, action => 'A' };
2996 }
2997
2998 sub add_directory {
2999 my ($self, $path, $cp_path, $cp_rev) = @_;
3000 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3001 delete $self->{empty}->{$dir};
3002 $self->{empty}->{$path} = 1;
3003 { path => $path };
3004 }
3005
3006 sub change_dir_prop {
3007 my ($self, $db, $prop, $value) = @_;
3008 $self->{dir_prop}->{$db->{path}} ||= {};
3009 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3010 undef;
3011 }
3012
3013 sub absent_directory {
3014 my ($self, $path, $pb) = @_;
3015 $self->{absent_dir}->{$pb->{path}} ||= [];
3016 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3017 undef;
3018 }
3019
3020 sub absent_file {
3021 my ($self, $path, $pb) = @_;
3022 $self->{absent_file}->{$pb->{path}} ||= [];
3023 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3024 undef;
3025 }
3026
3027 sub change_file_prop {
3028 my ($self, $fb, $prop, $value) = @_;
3029 if ($prop eq 'svn:executable') {
3030 if ($fb->{mode_b} != 120000) {
3031 $fb->{mode_b} = defined $value ? 100755 : 100644;
3032 }
3033 } elsif ($prop eq 'svn:special') {
3034 $fb->{mode_b} = defined $value ? 120000 : 100644;
3035 } else {
3036 $self->{file_prop}->{$fb->{path}} ||= {};
3037 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3038 }
3039 undef;
3040 }
3041
3042 sub apply_textdelta {
3043 my ($self, $fb, $exp) = @_;
3044 my $fh = IO::File->new_tmpfile;
3045 $fh->autoflush(1);
3046 # $fh gets auto-closed() by SVN::TxDelta::apply(),
3047 # (but $base does not,) so dup() it for reading in close_file
3048 open my $dup, '<&', $fh or croak $!;
3049 my $base = IO::File->new_tmpfile;
3050 $base->autoflush(1);
3051 if ($fb->{blob}) {
3052 defined (my $pid = fork) or croak $!;
3053 if (!$pid) {
3054 open STDOUT, '>&', $base or croak $!;
3055 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
3056 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
3057 }
3058 waitpid $pid, 0;
3059 croak $? if $?;
3060
3061 if (defined $exp) {
3062 seek $base, 0, 0 or croak $!;
3063 my $got = ::md5sum($base);
3064 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3065 "expected: $exp\n",
3066 " got: $got\n" if ($got ne $exp);
3067 }
3068 }
3069 seek $base, 0, 0 or croak $!;
3070 $fb->{fh} = $dup;
3071 $fb->{base} = $base;
3072 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
3073 }
3074
3075 sub close_file {
3076 my ($self, $fb, $exp) = @_;
3077 my $hash;
3078 my $path = $self->git_path($fb->{path});
3079 if (my $fh = $fb->{fh}) {
3080 if (defined $exp) {
3081 seek($fh, 0, 0) or croak $!;
3082 my $got = ::md5sum($fh);
3083 if ($got ne $exp) {
3084 die "Checksum mismatch: $path\n",
3085 "expected: $exp\n got: $got\n";
3086 }
3087 }
3088 sysseek($fh, 0, 0) or croak $!;
3089 if ($fb->{mode_b} == 120000) {
3090 sysread($fh, my $buf, 5) == 5 or croak $!;
3091 $buf eq 'link ' or die "$path has mode 120000",
3092 "but is not a link\n";
3093 }
3094 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
3095 if (!$pid) {
3096 open STDIN, '<&', $fh or croak $!;
3097 exec qw/git-hash-object -w --stdin/ or croak $!;
3098 }
3099 chomp($hash = do { local $/; <$out> });
3100 close $out or croak $!;
3101 close $fh or croak $!;
3102 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3103 close $fb->{base} or croak $!;
3104 } else {
3105 $hash = $fb->{blob} or die "no blob information\n";
3106 }
3107 $fb->{pool}->clear;
3108 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3109 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3110 undef;
3111 }
3112
3113 sub abort_edit {
3114 my $self = shift;
3115 $self->{nr} = $self->{gii}->{nr};
3116 delete $self->{gii};
3117 $self->SUPER::abort_edit(@_);
3118 }
3119
3120 sub close_edit {
3121 my $self = shift;
3122 $self->{git_commit_ok} = 1;
3123 $self->{nr} = $self->{gii}->{nr};
3124 delete $self->{gii};
3125 $self->SUPER::close_edit(@_);
3126 }
3127
3128 package SVN::Git::Editor;
3129 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3130 use strict;
3131 use warnings;
3132 use Carp qw/croak/;
3133 use IO::File;
3134
3135 sub new {
3136 my ($class, $opts) = @_;
3137 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3138 die "$_ required!\n" unless (defined $opts->{$_});
3139 }
3140
3141 my $pool = SVN::Pool->new;
3142 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3143 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3144 $opts->{r}, $mods);
3145
3146 # $opts->{ra} functions should not be used after this:
3147 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3148 $opts->{editor_cb}, $pool);
3149 my $self = SVN::Delta::Editor->new(@ce, $pool);
3150 bless $self, $class;
3151 foreach (qw/svn_path r tree_a tree_b/) {
3152 $self->{$_} = $opts->{$_};
3153 }
3154 $self->{url} = $opts->{ra}->{url};
3155 $self->{mods} = $mods;
3156 $self->{types} = $types;
3157 $self->{pool} = $pool;
3158 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3159 $self->{rm} = { };
3160 $self->{path_prefix} = length $self->{svn_path} ?
3161 "$self->{svn_path}/" : '';
3162 return $self;
3163 }
3164
3165 sub generate_diff {
3166 my ($tree_a, $tree_b) = @_;
3167 my @diff_tree = qw(diff-tree -z -r);
3168 if ($_cp_similarity) {
3169 push @diff_tree, "-C$_cp_similarity";
3170 } else {
3171 push @diff_tree, '-C';
3172 }
3173 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3174 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3175 push @diff_tree, $tree_a, $tree_b;
3176 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3177 local $/ = "\0";
3178 my $state = 'meta';
3179 my @mods;
3180 while (<$diff_fh>) {
3181 chomp $_; # this gets rid of the trailing "\0"
3182 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3183 $::sha1\s($::sha1)\s
3184 ([MTCRAD])\d*$/xo) {
3185 push @mods, { mode_a => $1, mode_b => $2,
3186 sha1_b => $3, chg => $4 };
3187 if ($4 =~ /^(?:C|R)$/) {
3188 $state = 'file_a';
3189 } else {
3190 $state = 'file_b';
3191 }
3192 } elsif ($state eq 'file_a') {
3193 my $x = $mods[$#mods] or croak "Empty array\n";
3194 if ($x->{chg} !~ /^(?:C|R)$/) {
3195 croak "Error parsing $_, $x->{chg}\n";
3196 }
3197 $x->{file_a} = $_;
3198 $state = 'file_b';
3199 } elsif ($state eq 'file_b') {
3200 my $x = $mods[$#mods] or croak "Empty array\n";
3201 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3202 croak "Error parsing $_, $x->{chg}\n";
3203 }
3204 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3205 croak "Error parsing $_, $x->{chg}\n";
3206 }
3207 $x->{file_b} = $_;
3208 $state = 'meta';
3209 } else {
3210 croak "Error parsing $_\n";
3211 }
3212 }
3213 command_close_pipe($diff_fh, $ctx);
3214 \@mods;
3215 }
3216
3217 sub check_diff_paths {
3218 my ($ra, $pfx, $rev, $mods) = @_;
3219 my %types;
3220 $pfx .= '/' if length $pfx;
3221
3222 sub type_diff_paths {
3223 my ($ra, $types, $path, $rev) = @_;
3224 my @p = split m#/+#, $path;
3225 my $c = shift @p;
3226 unless (defined $types->{$c}) {
3227 $types->{$c} = $ra->check_path($c, $rev);
3228 }
3229 while (@p) {
3230 $c .= '/' . shift @p;
3231 next if defined $types->{$c};
3232 $types->{$c} = $ra->check_path($c, $rev);
3233 }
3234 }
3235
3236 foreach my $m (@$mods) {
3237 foreach my $f (qw/file_a file_b/) {
3238 next unless defined $m->{$f};
3239 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3240 if (length $pfx.$dir && ! defined $types{$dir}) {
3241 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3242 }
3243 }
3244 }
3245 \%types;
3246 }
3247
3248 sub split_path {
3249 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3250 }
3251
3252 sub repo_path {
3253 my ($self, $path) = @_;
3254 $self->{path_prefix}.(defined $path ? $path : '');
3255 }
3256
3257 sub url_path {
3258 my ($self, $path) = @_;
3259 if ($self->{url} =~ m#^https?://#) {
3260 $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3261 }
3262 $self->{url} . '/' . $self->repo_path($path);
3263 }
3264
3265 sub rmdirs {
3266 my ($self) = @_;
3267 my $rm = $self->{rm};
3268 delete $rm->{''}; # we never delete the url we're tracking
3269 return unless %$rm;
3270
3271 foreach (keys %$rm) {
3272 my @d = split m#/#, $_;
3273 my $c = shift @d;
3274 $rm->{$c} = 1;
3275 while (@d) {
3276 $c .= '/' . shift @d;
3277 $rm->{$c} = 1;
3278 }
3279 }
3280 delete $rm->{$self->{svn_path}};
3281 delete $rm->{''}; # we never delete the url we're tracking
3282 return unless %$rm;
3283
3284 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3285 $self->{tree_b});
3286 local $/ = "\0";
3287 while (<$fh>) {
3288 chomp;
3289 my @dn = split m#/#, $_;
3290 while (pop @dn) {
3291 delete $rm->{join '/', @dn};
3292 }
3293 unless (%$rm) {
3294 close $fh;
3295 return;
3296 }
3297 }
3298 command_close_pipe($fh, $ctx);
3299
3300 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3301 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3302 $self->close_directory($bat->{$d}, $p);
3303 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3304 print "\tD+\t$d/\n" unless $::_q;
3305 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3306 delete $bat->{$d};
3307 }
3308 }
3309
3310 sub open_or_add_dir {
3311 my ($self, $full_path, $baton) = @_;
3312 my $t = $self->{types}->{$full_path};
3313 if (!defined $t) {
3314 die "$full_path not known in r$self->{r} or we have a bug!\n";
3315 }
3316 {
3317 no warnings 'once';
3318 # SVN::Node::none and SVN::Node::file are used only once,
3319 # so we're shutting up Perl's warnings about them.
3320 if ($t == $SVN::Node::none) {
3321 return $self->add_directory($full_path, $baton,
3322 undef, -1, $self->{pool});
3323 } elsif ($t == $SVN::Node::dir) {
3324 return $self->open_directory($full_path, $baton,
3325 $self->{r}, $self->{pool});
3326 } # no warnings 'once'
3327 print STDERR "$full_path already exists in repository at ",
3328 "r$self->{r} and it is not a directory (",
3329 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3330 } # no warnings 'once'
3331 exit 1;
3332 }
3333
3334 sub ensure_path {
3335 my ($self, $path) = @_;
3336 my $bat = $self->{bat};
3337 my $repo_path = $self->repo_path($path);
3338 return $bat->{''} unless (length $repo_path);
3339 my @p = split m#/+#, $repo_path;
3340 my $c = shift @p;
3341 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3342 while (@p) {
3343 my $c0 = $c;
3344 $c .= '/' . shift @p;
3345 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3346 }
3347 return $bat->{$c};
3348 }
3349
3350 sub A {
3351 my ($self, $m) = @_;
3352 my ($dir, $file) = split_path($m->{file_b});
3353 my $pbat = $self->ensure_path($dir);
3354 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3355 undef, -1);
3356 print "\tA\t$m->{file_b}\n" unless $::_q;
3357 $self->chg_file($fbat, $m);
3358 $self->close_file($fbat,undef,$self->{pool});
3359 }
3360
3361 sub C {
3362 my ($self, $m) = @_;
3363 my ($dir, $file) = split_path($m->{file_b});
3364 my $pbat = $self->ensure_path($dir);
3365 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3366 $self->url_path($m->{file_a}), $self->{r});
3367 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3368 $self->chg_file($fbat, $m);
3369 $self->close_file($fbat,undef,$self->{pool});
3370 }
3371
3372 sub delete_entry {
3373 my ($self, $path, $pbat) = @_;
3374 my $rpath = $self->repo_path($path);
3375 my ($dir, $file) = split_path($rpath);
3376 $self->{rm}->{$dir} = 1;
3377 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3378 }
3379
3380 sub R {
3381 my ($self, $m) = @_;
3382 my ($dir, $file) = split_path($m->{file_b});
3383 my $pbat = $self->ensure_path($dir);
3384 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3385 $self->url_path($m->{file_a}), $self->{r});
3386 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3387 $self->chg_file($fbat, $m);
3388 $self->close_file($fbat,undef,$self->{pool});
3389
3390 ($dir, $file) = split_path($m->{file_a});
3391 $pbat = $self->ensure_path($dir);
3392 $self->delete_entry($m->{file_a}, $pbat);
3393 }
3394
3395 sub M {
3396 my ($self, $m) = @_;
3397 my ($dir, $file) = split_path($m->{file_b});
3398 my $pbat = $self->ensure_path($dir);
3399 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3400 $pbat,$self->{r},$self->{pool});
3401 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3402 $self->chg_file($fbat, $m);
3403 $self->close_file($fbat,undef,$self->{pool});
3404 }
3405
3406 sub T { shift->M(@_) }
3407
3408 sub change_file_prop {
3409 my ($self, $fbat, $pname, $pval) = @_;
3410 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3411 }
3412
3413 sub chg_file {
3414 my ($self, $fbat, $m) = @_;
3415 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3416 $self->change_file_prop($fbat,'svn:executable','*');
3417 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3418 $self->change_file_prop($fbat,'svn:executable',undef);
3419 }
3420 my $fh = IO::File->new_tmpfile or croak $!;
3421 if ($m->{mode_b} =~ /^120/) {
3422 print $fh 'link ' or croak $!;
3423 $self->change_file_prop($fbat,'svn:special','*');
3424 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3425 $self->change_file_prop($fbat,'svn:special',undef);
3426 }
3427 defined(my $pid = fork) or croak $!;
3428 if (!$pid) {
3429 open STDOUT, '>&', $fh or croak $!;
3430 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3431 }
3432 waitpid $pid, 0;
3433 croak $? if $?;
3434 $fh->flush == 0 or croak $!;
3435 seek $fh, 0, 0 or croak $!;
3436
3437 my $exp = ::md5sum($fh);
3438 seek $fh, 0, 0 or croak $!;
3439
3440 my $pool = SVN::Pool->new;
3441 my $atd = $self->apply_textdelta($fbat, undef, $pool);
3442 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3443 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3444 $pool->clear;
3445
3446 close $fh or croak $!;
3447 }
3448
3449 sub D {
3450 my ($self, $m) = @_;
3451 my ($dir, $file) = split_path($m->{file_b});
3452 my $pbat = $self->ensure_path($dir);
3453 print "\tD\t$m->{file_b}\n" unless $::_q;
3454 $self->delete_entry($m->{file_b}, $pbat);
3455 }
3456
3457 sub close_edit {
3458 my ($self) = @_;
3459 my ($p,$bat) = ($self->{pool}, $self->{bat});
3460 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3461 next if $_ eq '';
3462 $self->close_directory($bat->{$_}, $p);
3463 }
3464 $self->close_directory($bat->{''}, $p);
3465 $self->SUPER::close_edit($p);
3466 $p->clear;
3467 }
3468
3469 sub abort_edit {
3470 my ($self) = @_;
3471 $self->SUPER::abort_edit($self->{pool});
3472 }
3473
3474 sub DESTROY {
3475 my $self = shift;
3476 $self->SUPER::DESTROY(@_);
3477 $self->{pool}->clear;
3478 }
3479
3480 # this drives the editor
3481 sub apply_diff {
3482 my ($self) = @_;
3483 my $mods = $self->{mods};
3484 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3485 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3486 my $f = $m->{chg};
3487 if (defined $o{$f}) {
3488 $self->$f($m);
3489 } else {
3490 fatal("Invalid change type: $f");
3491 }
3492 }
3493 $self->rmdirs if $_rmdir;
3494 if (@$mods == 0) {
3495 $self->abort_edit;
3496 } else {
3497 $self->close_edit;
3498 }
3499 return scalar @$mods;
3500 }
3501
3502 package Git::SVN::Ra;
3503 use vars qw/@ISA $config_dir $_log_window_size/;
3504 use strict;
3505 use warnings;
3506 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3507
3508 BEGIN {
3509 # enforce temporary pool usage for some simple functions
3510 no strict 'refs';
3511 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3512 my $SUPER = "SUPER::$f";
3513 *$f = sub {
3514 my $self = shift;
3515 my $pool = SVN::Pool->new;
3516 my @ret = $self->$SUPER(@_,$pool);
3517 $pool->clear;
3518 wantarray ? @ret : $ret[0];
3519 };
3520 }
3521 }
3522
3523 sub _auth_providers () {
3524 [
3525 SVN::Client::get_simple_provider(),
3526 SVN::Client::get_ssl_server_trust_file_provider(),
3527 SVN::Client::get_simple_prompt_provider(
3528 \&Git::SVN::Prompt::simple, 2),
3529 SVN::Client::get_ssl_client_cert_file_provider(),
3530 SVN::Client::get_ssl_client_cert_prompt_provider(
3531 \&Git::SVN::Prompt::ssl_client_cert, 2),
3532 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3533 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3534 SVN::Client::get_username_provider(),
3535 SVN::Client::get_ssl_server_trust_prompt_provider(
3536 \&Git::SVN::Prompt::ssl_server_trust),
3537 SVN::Client::get_username_prompt_provider(
3538 \&Git::SVN::Prompt::username, 2)
3539 ]
3540 }
3541
3542 sub escape_uri_only {
3543 my ($uri) = @_;
3544 my @tmp;
3545 foreach (split m{/}, $uri) {
3546 s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
3547 push @tmp, $_;
3548 }
3549 join('/', @tmp);
3550 }
3551
3552 sub escape_url {
3553 my ($url) = @_;
3554 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3555 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3556 $url = "$scheme://$domain$uri";
3557 }
3558 $url;
3559 }
3560
3561 sub new {
3562 my ($class, $url) = @_;
3563 $url =~ s!/+$!!;
3564 return $RA if ($RA && $RA->{url} eq $url);
3565
3566 SVN::_Core::svn_config_ensure($config_dir, undef);
3567 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3568 my $config = SVN::Core::config_get_config($config_dir);
3569 $RA = undef;
3570 my $dont_store_passwords = 1;
3571 my $conf_t = ${$config}{'config'};
3572 {
3573 no warnings 'once';
3574 # The usage of $SVN::_Core::SVN_CONFIG_* variables
3575 # produces warnings that variables are used only once.
3576 # I had not found the better way to shut them up, so
3577 # the warnings of type 'once' are disabled in this block.
3578 if (SVN::_Core::svn_config_get_bool($conf_t,
3579 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3580 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3581 1) == 0) {
3582 SVN::_Core::svn_auth_set_parameter($baton,
3583 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3584 bless (\$dont_store_passwords, "_p_void"));
3585 }
3586 if (SVN::_Core::svn_config_get_bool($conf_t,
3587 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3588 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3589 1) == 0) {
3590 $Git::SVN::Prompt::_no_auth_cache = 1;
3591 }
3592 } # no warnings 'once'
3593 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3594 config => $config,
3595 pool => SVN::Pool->new,
3596 auth_provider_callbacks => $callbacks);
3597 $self->{url} = $url;
3598 $self->{svn_path} = $url;
3599 $self->{repos_root} = $self->get_repos_root;
3600 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3601 $self->{cache} = { check_path => { r => 0, data => {} },
3602 get_dir => { r => 0, data => {} } };
3603 $RA = bless $self, $class;
3604 }
3605
3606 sub check_path {
3607 my ($self, $path, $r) = @_;
3608 my $cache = $self->{cache}->{check_path};
3609 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3610 return $cache->{data}->{$path};
3611 }
3612 my $pool = SVN::Pool->new;
3613 my $t = $self->SUPER::check_path($path, $r, $pool);
3614 $pool->clear;
3615 if ($r != $cache->{r}) {
3616 %{$cache->{data}} = ();
3617 $cache->{r} = $r;
3618 }
3619 $cache->{data}->{$path} = $t;
3620 }
3621
3622 sub get_dir {
3623 my ($self, $dir, $r) = @_;
3624 my $cache = $self->{cache}->{get_dir};
3625 if ($r == $cache->{r}) {
3626 if (my $x = $cache->{data}->{$dir}) {
3627 return wantarray ? @$x : $x->[0];
3628 }
3629 }
3630 my $pool = SVN::Pool->new;
3631 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3632 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3633 $pool->clear;
3634 if ($r != $cache->{r}) {
3635 %{$cache->{data}} = ();
3636 $cache->{r} = $r;
3637 }
3638 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3639 wantarray ? (\%dirents, $r, $props) : \%dirents;
3640 }
3641
3642 sub DESTROY {
3643 # do not call the real DESTROY since we store ourselves in $RA
3644 }
3645
3646 sub get_log {
3647 my ($self, @args) = @_;
3648 my $pool = SVN::Pool->new;
3649 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3650 my $ret = $self->SUPER::get_log(@args, $pool);
3651 $pool->clear;
3652 $ret;
3653 }
3654
3655 sub trees_match {
3656 my ($self, $url1, $rev1, $url2, $rev2) = @_;
3657 my $ctx = SVN::Client->new(auth => _auth_providers);
3658 my $out = IO::File->new_tmpfile;
3659
3660 # older SVN (1.1.x) doesn't take $pool as the last parameter for
3661 # $ctx->diff(), so we'll create a default one
3662 my $pool = SVN::Pool->new_default_sub;
3663
3664 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3665 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3666 $out->flush;
3667 my $ret = (($out->stat)[7] == 0);
3668 close $out or croak $!;
3669
3670 $ret;
3671 }
3672
3673 sub get_commit_editor {
3674 my ($self, $log, $cb, $pool) = @_;
3675 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3676 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3677 }
3678
3679 sub gs_do_update {
3680 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3681 my $new = ($rev_a == $rev_b);
3682 my $path = $gs->{path};
3683
3684 if ($new && -e $gs->{index}) {
3685 unlink $gs->{index} or die
3686 "Couldn't unlink index: $gs->{index}: $!\n";
3687 }
3688 my $pool = SVN::Pool->new;
3689 $editor->set_path_strip($path);
3690 my (@pc) = split m#/#, $path;
3691 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3692 1, $editor, $pool);
3693 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3694
3695 # Since we can't rely on svn_ra_reparent being available, we'll
3696 # just have to do some magic with set_path to make it so
3697 # we only want a partial path.
3698 my $sp = '';
3699 my $final = join('/', @pc);
3700 while (@pc) {
3701 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3702 $sp .= '/' if length $sp;
3703 $sp .= shift @pc;
3704 }
3705 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3706
3707 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3708
3709 $reporter->finish_report($pool);
3710 $pool->clear;
3711 $editor->{git_commit_ok};
3712 }
3713
3714 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3715 # svn_ra_reparent didn't work before 1.4)
3716 sub gs_do_switch {
3717 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3718 my $path = $gs->{path};
3719 my $pool = SVN::Pool->new;
3720
3721 my $full_url = $self->{url};
3722 my $old_url = $full_url;
3723 $full_url .= '/' . escape_uri_only($path) if length $path;
3724 my ($ra, $reparented);
3725 if ($old_url ne $full_url) {
3726 if ($old_url !~ m#^svn(\+ssh)?://#) {
3727 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3728 $pool);
3729 $self->{url} = $full_url;
3730 $reparented = 1;
3731 } else {
3732 $_[0] = undef;
3733 $self = undef;
3734 $RA = undef;
3735 $ra = Git::SVN::Ra->new($full_url);
3736 $ra_invalid = 1;
3737 }
3738 }
3739 $ra ||= $self;
3740 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3741 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3742 $reporter->set_path('', $rev_a, 0, @lock, $pool);
3743 $reporter->finish_report($pool);
3744
3745 if ($reparented) {
3746 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3747 $self->{url} = $old_url;
3748 }
3749
3750 $pool->clear;
3751 $editor->{git_commit_ok};
3752 }
3753
3754 sub longest_common_path {
3755 my ($gsv, $globs) = @_;
3756 my %common;
3757 my $common_max = scalar @$gsv;
3758
3759 foreach my $gs (@$gsv) {
3760 my @tmp = split m#/#, $gs->{path};
3761 my $p = '';
3762 foreach (@tmp) {
3763 $p .= length($p) ? "/$_" : $_;
3764 $common{$p} ||= 0;
3765 $common{$p}++;
3766 }
3767 }
3768 $globs ||= [];
3769 $common_max += scalar @$globs;
3770 foreach my $glob (@$globs) {
3771 my @tmp = split m#/#, $glob->{path}->{left};
3772 my $p = '';
3773 foreach (@tmp) {
3774 $p .= length($p) ? "/$_" : $_;
3775 $common{$p} ||= 0;
3776 $common{$p}++;
3777 }
3778 }
3779
3780 my $longest_path = '';
3781 foreach (sort {length $b <=> length $a} keys %common) {
3782 if ($common{$_} == $common_max) {
3783 $longest_path = $_;
3784 last;
3785 }
3786 }
3787 $longest_path;
3788 }
3789
3790 sub gs_fetch_loop_common {
3791 my ($self, $base, $head, $gsv, $globs) = @_;
3792 return if ($base > $head);
3793 my $inc = $_log_window_size;
3794 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3795 my $longest_path = longest_common_path($gsv, $globs);
3796 my $ra_url = $self->{url};
3797 while (1) {
3798 my %revs;
3799 my $err;
3800 my $err_handler = $SVN::Error::handler;
3801 $SVN::Error::handler = sub {
3802 ($err) = @_;
3803 skip_unknown_revs($err);
3804 };
3805 sub _cb {
3806 my ($paths, $r, $author, $date, $log) = @_;
3807 [ dup_changed_paths($paths),
3808 { author => $author, date => $date, log => $log } ];
3809 }
3810 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3811 sub { $revs{$_[1]} = _cb(@_) });
3812 if ($err && $max >= $head) {
3813 print STDERR "Path '$longest_path' ",
3814 "was probably deleted:\n",
3815 $err->expanded_message,
3816 "\nWill attempt to follow ",
3817 "revisions r$min .. r$max ",
3818 "committed before the deletion\n";
3819 my $hi = $max;
3820 while (--$hi >= $min) {
3821 my $ok;
3822 $self->get_log([$longest_path], $min, $hi,
3823 0, 1, 1, sub {
3824 $ok ||= $_[1];
3825 $revs{$_[1]} = _cb(@_) });
3826 if ($ok) {
3827 print STDERR "r$min .. r$ok OK\n";
3828 last;
3829 }
3830 }
3831 }
3832 $SVN::Error::handler = $err_handler;
3833
3834 my %exists = map { $_->{path} => $_ } @$gsv;
3835 foreach my $r (sort {$a <=> $b} keys %revs) {
3836 my ($paths, $logged) = @{$revs{$r}};
3837
3838 foreach my $gs ($self->match_globs(\%exists, $paths,
3839 $globs, $r)) {
3840 if ($gs->rev_map_max >= $r) {
3841 next;
3842 }
3843 next unless $gs->match_paths($paths, $r);
3844 $gs->{logged_rev_props} = $logged;
3845 if (my $last_commit = $gs->last_commit) {
3846 $gs->assert_index_clean($last_commit);
3847 }
3848 my $log_entry = $gs->do_fetch($paths, $r);
3849 if ($log_entry) {
3850 $gs->do_git_commit($log_entry);
3851 }
3852 }
3853 foreach my $g (@$globs) {
3854 my $k = "svn-remote.$g->{remote}." .
3855 "$g->{t}-maxRev";
3856 Git::SVN::tmp_config($k, $r);
3857 }
3858 if ($ra_invalid) {
3859 $_[0] = undef;
3860 $self = undef;
3861 $RA = undef;
3862 $self = Git::SVN::Ra->new($ra_url);
3863 $ra_invalid = undef;
3864 }
3865 }
3866 foreach my $g (@$globs) {
3867 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3868 Git::SVN::tmp_config($k, $max);
3869 }
3870 last if $max >= $head;
3871 $min = $max + 1;
3872 $max += $inc;
3873 $max = $head if ($max > $head);
3874 }
3875 }
3876
3877 sub match_globs {
3878 my ($self, $exists, $paths, $globs, $r) = @_;
3879
3880 sub get_dir_check {
3881 my ($self, $exists, $g, $r) = @_;
3882 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3883 return unless scalar @x == 3;
3884 my $dirents = $x[0];
3885 foreach my $de (keys %$dirents) {
3886 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
3887 my $p = $g->{path}->full_path($de);
3888 next if $exists->{$p};
3889 next if (length $g->{path}->{right} &&
3890 ($self->check_path($p, $r) !=
3891 $SVN::Node::dir));
3892 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3893 $g->{ref}->full_path($de), 1);
3894 }
3895 }
3896 foreach my $g (@$globs) {
3897 if (my $path = $paths->{"/$g->{path}->{left}"}) {
3898 if ($path->{action} =~ /^[AR]$/) {
3899 get_dir_check($self, $exists, $g, $r);
3900 }
3901 }
3902 foreach (keys %$paths) {
3903 if (/$g->{path}->{left_regex}/ &&
3904 !/$g->{path}->{regex}/) {
3905 next if $paths->{$_}->{action} !~ /^[AR]$/;
3906 get_dir_check($self, $exists, $g, $r);
3907 }
3908 next unless /$g->{path}->{regex}/;
3909 my $p = $1;
3910 my $pathname = $g->{path}->full_path($p);
3911 next if $exists->{$pathname};
3912 next if ($self->check_path($pathname, $r) !=
3913 $SVN::Node::dir);
3914 $exists->{$pathname} = Git::SVN->init(
3915 $self->{url}, $pathname, undef,
3916 $g->{ref}->full_path($p), 1);
3917 }
3918 my $c = '';
3919 foreach (split m#/#, $g->{path}->{left}) {
3920 $c .= "/$_";
3921 next unless ($paths->{$c} &&
3922 ($paths->{$c}->{action} =~ /^[AR]$/));
3923 get_dir_check($self, $exists, $g, $r);
3924 }
3925 }
3926 values %$exists;
3927 }
3928
3929 sub minimize_url {
3930 my ($self) = @_;
3931 return $self->{url} if ($self->{url} eq $self->{repos_root});
3932 my $url = $self->{repos_root};
3933 my @components = split(m!/!, $self->{svn_path});
3934 my $c = '';
3935 do {
3936 $url .= "/$c" if length $c;
3937 eval { (ref $self)->new($url)->get_latest_revnum };
3938 } while ($@ && ($c = shift @components));
3939 $url;
3940 }
3941
3942 sub can_do_switch {
3943 my $self = shift;
3944 unless (defined $can_do_switch) {
3945 my $pool = SVN::Pool->new;
3946 my $rep = eval {
3947 $self->do_switch(1, '', 0, $self->{url},
3948 SVN::Delta::Editor->new, $pool);
3949 };
3950 if ($@) {
3951 $can_do_switch = 0;
3952 } else {
3953 $rep->abort_report($pool);
3954 $can_do_switch = 1;
3955 }
3956 $pool->clear;
3957 }
3958 $can_do_switch;
3959 }
3960
3961 sub skip_unknown_revs {
3962 my ($err) = @_;
3963 my $errno = $err->apr_err();
3964 # Maybe the branch we're tracking didn't
3965 # exist when the repo started, so it's
3966 # not an error if it doesn't, just continue
3967 #
3968 # Wonderfully consistent library, eh?
3969 # 160013 - svn:// and file://
3970 # 175002 - http(s)://
3971 # 175007 - http(s):// (this repo required authorization, too...)
3972 # More codes may be discovered later...
3973 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3974 my $err_key = $err->expanded_message;
3975 # revision numbers change every time, filter them out
3976 $err_key =~ s/\d+/\0/g;
3977 $err_key = "$errno\0$err_key";
3978 unless ($ignored_err{$err_key}) {
3979 warn "W: Ignoring error from SVN, path probably ",
3980 "does not exist: ($errno): ",
3981 $err->expanded_message,"\n";
3982 $ignored_err{$err_key} = 1;
3983 }
3984 return;
3985 }
3986 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3987 }
3988
3989 # svn_log_changed_path_t objects passed to get_log are likely to be
3990 # overwritten even if only the refs are copied to an external variable,
3991 # so we should dup the structures in their entirety. Using an externally
3992 # passed pool (instead of our temporary and quickly cleared pool in
3993 # Git::SVN::Ra) does not help matters at all...
3994 sub dup_changed_paths {
3995 my ($paths) = @_;
3996 return undef unless $paths;
3997 my %ret;
3998 foreach my $p (keys %$paths) {
3999 my $i = $paths->{$p};
4000 my %s = map { $_ => $i->$_ }
4001 qw/copyfrom_path copyfrom_rev action/;
4002 $ret{$p} = \%s;
4003 }
4004 \%ret;
4005 }
4006
4007 package Git::SVN::Log;
4008 use strict;
4009 use warnings;
4010 use POSIX qw/strftime/;
4011 use constant commit_log_separator => ('-' x 72) . "\n";
4012 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4013 %rusers $show_commit $incremental/;
4014 my $l_fmt;
4015
4016 sub cmt_showable {
4017 my ($c) = @_;
4018 return 1 if defined $c->{r};
4019
4020 # big commit message got truncated by the 16k pretty buffer in rev-list
4021 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4022 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4023 @{$c->{l}} = ();
4024 my @log = command(qw/cat-file commit/, $c->{c});
4025
4026 # shift off the headers
4027 shift @log while ($log[0] ne '');
4028 shift @log;
4029
4030 # TODO: make $c->{l} not have a trailing newline in the future
4031 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4032
4033 (undef, $c->{r}, undef) = ::extract_metadata(
4034 (grep(/^git-svn-id: /, @log))[-1]);
4035 }
4036 return defined $c->{r};
4037 }
4038
4039 sub log_use_color {
4040 return 1 if $color;
4041 my ($dc, $dcvar);
4042 $dcvar = 'color.diff';
4043 $dc = `git-config --get $dcvar`;
4044 if ($dc eq '') {
4045 # nothing at all; fallback to "diff.color"
4046 $dcvar = 'diff.color';
4047 $dc = `git-config --get $dcvar`;
4048 }
4049 chomp($dc);
4050 if ($dc eq 'auto') {
4051 my $pc;
4052 $pc = `git-config --get color.pager`;
4053 if ($pc eq '') {
4054 # does not have it -- fallback to pager.color
4055 $pc = `git-config --bool --get pager.color`;
4056 }
4057 else {
4058 $pc = `git-config --bool --get color.pager`;
4059 if ($?) {
4060 $pc = 'false';
4061 }
4062 }
4063 chomp($pc);
4064 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
4065 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
4066 }
4067 return 0;
4068 }
4069 return 0 if $dc eq 'never';
4070 return 1 if $dc eq 'always';
4071 chomp($dc = `git-config --bool --get $dcvar`);
4072 return ($dc eq 'true');
4073 }
4074
4075 sub git_svn_log_cmd {
4076 my ($r_min, $r_max, @args) = @_;
4077 my $head = 'HEAD';
4078 my (@files, @log_opts);
4079 foreach my $x (@args) {
4080 if ($x eq '--' || @files) {
4081 push @files, $x;
4082 } else {
4083 if (::verify_ref("$x^0")) {
4084 $head = $x;
4085 } else {
4086 push @log_opts, $x;
4087 }
4088 }
4089 }
4090
4091 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4092 $gs ||= Git::SVN->_new;
4093 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4094 $gs->refname);
4095 push @cmd, '-r' unless $non_recursive;
4096 push @cmd, qw/--raw --name-status/ if $verbose;
4097 push @cmd, '--color' if log_use_color();
4098 push @cmd, @log_opts;
4099 if (defined $r_max && $r_max == $r_min) {
4100 push @cmd, '--max-count=1';
4101 if (my $c = $gs->rev_map_get($r_max)) {
4102 push @cmd, $c;
4103 }
4104 } elsif (defined $r_max) {
4105 if ($r_max < $r_min) {
4106 ($r_min, $r_max) = ($r_max, $r_min);
4107 }
4108 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4109 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4110 # If there are no commits in the range, both $c_max and $c_min
4111 # will be undefined. If there is at least 1 commit in the
4112 # range, both will be defined.
4113 return () if !defined $c_min || !defined $c_max;
4114 if ($c_min eq $c_max) {
4115 push @cmd, '--max-count=1', $c_min;
4116 } else {
4117 push @cmd, '--boundary', "$c_min..$c_max";
4118 }
4119 }
4120 return (@cmd, @files);
4121 }
4122
4123 # adapted from pager.c
4124 sub config_pager {
4125 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4126 if (!defined $pager) {
4127 $pager = 'less';
4128 } elsif (length $pager == 0 || $pager eq 'cat') {
4129 $pager = undef;
4130 }
4131 }
4132
4133 sub run_pager {
4134 return unless -t *STDOUT && defined $pager;
4135 pipe my $rfd, my $wfd or return;
4136 defined(my $pid = fork) or ::fatal "Can't fork: $!";
4137 if (!$pid) {
4138 open STDOUT, '>&', $wfd or
4139 ::fatal "Can't redirect to stdout: $!";
4140 return;
4141 }
4142 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4143 $ENV{LESS} ||= 'FRSX';
4144 exec $pager or ::fatal "Can't run pager: $! ($pager)";
4145 }
4146
4147 sub format_svn_date {
4148 return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4149 }
4150
4151 sub parse_git_date {
4152 my ($t, $tz) = @_;
4153 # Date::Parse isn't in the standard Perl distro :(
4154 if ($tz =~ s/^\+//) {
4155 $t += tz_to_s_offset($tz);
4156 } elsif ($tz =~ s/^\-//) {
4157 $t -= tz_to_s_offset($tz);
4158 }
4159 return $t;
4160 }
4161
4162 sub set_local_timezone {
4163 if (defined $TZ) {
4164 $ENV{TZ} = $TZ;
4165 } else {
4166 delete $ENV{TZ};
4167 }
4168 }
4169
4170 sub tz_to_s_offset {
4171 my ($tz) = @_;
4172 $tz =~ s/(\d\d)$//;
4173 return ($1 * 60) + ($tz * 3600);
4174 }
4175
4176 sub get_author_info {
4177 my ($dest, $author, $t, $tz) = @_;
4178 $author =~ s/(?:^\s*|\s*$)//g;
4179 $dest->{a_raw} = $author;
4180 my $au;
4181 if ($::_authors) {
4182 $au = $rusers{$author} || undef;
4183 }
4184 if (!$au) {
4185 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4186 }
4187 $dest->{t} = $t;
4188 $dest->{tz} = $tz;
4189 $dest->{a} = $au;
4190 $dest->{t_utc} = parse_git_date($t, $tz);
4191 }
4192
4193 sub process_commit {
4194 my ($c, $r_min, $r_max, $defer) = @_;
4195 if (defined $r_min && defined $r_max) {
4196 if ($r_min == $c->{r} && $r_min == $r_max) {
4197 show_commit($c);
4198 return 0;
4199 }
4200 return 1 if $r_min == $r_max;
4201 if ($r_min < $r_max) {
4202 # we need to reverse the print order
4203 return 0 if (defined $limit && --$limit < 0);
4204 push @$defer, $c;
4205 return 1;
4206 }
4207 if ($r_min != $r_max) {
4208 return 1 if ($r_min < $c->{r});
4209 return 1 if ($r_max > $c->{r});
4210 }
4211 }
4212 return 0 if (defined $limit && --$limit < 0);
4213 show_commit($c);
4214 return 1;
4215 }
4216
4217 sub show_commit {
4218 my $c = shift;
4219 if ($oneline) {
4220 my $x = "\n";
4221 if (my $l = $c->{l}) {
4222 while ($l->[0] =~ /^\s*$/) { shift @$l }
4223 $x = $l->[0];
4224 }
4225 $l_fmt ||= 'A' . length($c->{r});
4226 print 'r',pack($l_fmt, $c->{r}),' | ';
4227 print "$c->{c} | " if $show_commit;
4228 print $x;
4229 } else {
4230 show_commit_normal($c);
4231 }
4232 }
4233
4234 sub show_commit_changed_paths {
4235 my ($c) = @_;
4236 return unless $c->{changed};
4237 print "Changed paths:\n", @{$c->{changed}};
4238 }
4239
4240 sub show_commit_normal {
4241 my ($c) = @_;
4242 print commit_log_separator, "r$c->{r} | ";
4243 print "$c->{c} | " if $show_commit;
4244 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4245 my $nr_line = 0;
4246
4247 if (my $l = $c->{l}) {
4248 while ($l->[$#$l] eq "\n" && $#$l > 0
4249 && $l->[($#$l - 1)] eq "\n") {
4250 pop @$l;
4251 }
4252 $nr_line = scalar @$l;
4253 if (!$nr_line) {
4254 print "1 line\n\n\n";
4255 } else {
4256 if ($nr_line == 1) {
4257 $nr_line = '1 line';
4258 } else {
4259 $nr_line .= ' lines';
4260 }
4261 print $nr_line, "\n";
4262 show_commit_changed_paths($c);
4263 print "\n";
4264 print $_ foreach @$l;
4265 }
4266 } else {
4267 print "1 line\n";
4268 show_commit_changed_paths($c);
4269 print "\n";
4270
4271 }
4272 foreach my $x (qw/raw stat diff/) {
4273 if ($c->{$x}) {
4274 print "\n";
4275 print $_ foreach @{$c->{$x}}
4276 }
4277 }
4278 }
4279
4280 sub cmd_show_log {
4281 my (@args) = @_;
4282 my ($r_min, $r_max);
4283 my $r_last = -1; # prevent dupes
4284 set_local_timezone();
4285 if (defined $::_revision) {
4286 if ($::_revision =~ /^(\d+):(\d+)$/) {
4287 ($r_min, $r_max) = ($1, $2);
4288 } elsif ($::_revision =~ /^\d+$/) {
4289 $r_min = $r_max = $::_revision;
4290 } else {
4291 ::fatal "-r$::_revision is not supported, use ",
4292 "standard 'git log' arguments instead";
4293 }
4294 }
4295
4296 config_pager();
4297 @args = git_svn_log_cmd($r_min, $r_max, @args);
4298 if (!@args) {
4299 print commit_log_separator unless $incremental || $oneline;
4300 return;
4301 }
4302 my $log = command_output_pipe(@args);
4303 run_pager();
4304 my (@k, $c, $d, $stat);
4305 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4306 while (<$log>) {
4307 if (/^${esc_color}commit -?($::sha1_short)/o) {
4308 my $cmt = $1;
4309 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4310 $r_last = $c->{r};
4311 process_commit($c, $r_min, $r_max, \@k) or
4312 goto out;
4313 }
4314 $d = undef;
4315 $c = { c => $cmt };
4316 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4317 get_author_info($c, $1, $2, $3);
4318 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4319 # ignore
4320 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4321 push @{$c->{raw}}, $_;
4322 } elsif (/^${esc_color}[ACRMDT]\t/) {
4323 # we could add $SVN->{svn_path} here, but that requires
4324 # remote access at the moment (repo_path_split)...
4325 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
4326 push @{$c->{changed}}, $_;
4327 } elsif (/^${esc_color}diff /o) {
4328 $d = 1;
4329 push @{$c->{diff}}, $_;
4330 } elsif ($d) {
4331 push @{$c->{diff}}, $_;
4332 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4333 $esc_color*[\+\-]*$esc_color$/x) {
4334 $stat = 1;
4335 push @{$c->{stat}}, $_;
4336 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4337 push @{$c->{stat}}, $_;
4338 $stat = undef;
4339 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
4340 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4341 } elsif (s/^${esc_color} //o) {
4342 push @{$c->{l}}, $_;
4343 }
4344 }
4345 if ($c && defined $c->{r} && $c->{r} != $r_last) {
4346 $r_last = $c->{r};
4347 process_commit($c, $r_min, $r_max, \@k);
4348 }
4349 if (@k) {
4350 ($r_min, $r_max) = ($r_max, $r_min);
4351 process_commit($_, $r_min, $r_max) foreach reverse @k;
4352 }
4353 out:
4354 close $log;
4355 print commit_log_separator unless $incremental || $oneline;
4356 }
4357
4358 package Git::SVN::Migration;
4359 # these version numbers do NOT correspond to actual version numbers
4360 # of git nor git-svn. They are just relative.
4361 #
4362 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4363 #
4364 # v1 layout: .git/$id/info/url, refs/remotes/$id
4365 #
4366 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4367 #
4368 # v3 layout: .git/svn/$id, refs/remotes/$id
4369 # - info/url may remain for backwards compatibility
4370 # - this is what we migrate up to this layout automatically,
4371 # - this will be used by git svn init on single branches
4372 # v3.1 layout (auto migrated):
4373 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4374 # for backwards compatibility
4375 #
4376 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4377 # - this is only created for newly multi-init-ed
4378 # repositories. Similar in spirit to the
4379 # --use-separate-remotes option in git-clone (now default)
4380 # - we do not automatically migrate to this (following
4381 # the example set by core git)
4382 #
4383 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
4384 # - newer, more-efficient format that uses 24-bytes per record
4385 # with no filler space.
4386 # - use xxd -c24 < .rev_map.$UUID to view and debug
4387 # - This is a one-way migration, repositories updated to the
4388 # new format will not be able to use old git-svn without
4389 # rebuilding the .rev_db. Rebuilding the rev_db is not
4390 # possible if noMetadata or useSvmProps are set; but should
4391 # be no problem for users that use the (sensible) defaults.
4392 use strict;
4393 use warnings;
4394 use Carp qw/croak/;
4395 use File::Path qw/mkpath/;
4396 use File::Basename qw/dirname basename/;
4397 use vars qw/$_minimize/;
4398
4399 sub migrate_from_v0 {
4400 my $git_dir = $ENV{GIT_DIR};
4401 return undef unless -d $git_dir;
4402 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4403 my $migrated = 0;
4404 while (<$fh>) {
4405 chomp;
4406 my ($id, $orig_ref) = ($_, $_);
4407 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4408 next unless -f "$git_dir/$id/info/url";
4409 my $new_ref = "refs/remotes/$id";
4410 if (::verify_ref("$new_ref^0")) {
4411 print STDERR "W: $orig_ref is probably an old ",
4412 "branch used by an ancient version of ",
4413 "git-svn.\n",
4414 "However, $new_ref also exists.\n",
4415 "We will not be able ",
4416 "to use this branch until this ",
4417 "ambiguity is resolved.\n";
4418 next;
4419 }
4420 print STDERR "Migrating from v0 layout...\n" if !$migrated;
4421 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4422 command_noisy('update-ref', $new_ref, $orig_ref);
4423 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4424 $migrated++;
4425 }
4426 command_close_pipe($fh, $ctx);
4427 print STDERR "Done migrating from v0 layout...\n" if $migrated;
4428 $migrated;
4429 }
4430
4431 sub migrate_from_v1 {
4432 my $git_dir = $ENV{GIT_DIR};
4433 my $migrated = 0;
4434 return $migrated unless -d $git_dir;
4435 my $svn_dir = "$git_dir/svn";
4436
4437 # just in case somebody used 'svn' as their $id at some point...
4438 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4439
4440 print STDERR "Migrating from a git-svn v1 layout...\n";
4441 mkpath([$svn_dir]);
4442 print STDERR "Data from a previous version of git-svn exists, but\n\t",
4443 "$svn_dir\n\t(required for this version ",
4444 "($::VERSION) of git-svn) does not. exist\n";
4445 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4446 while (<$fh>) {
4447 my $x = $_;
4448 next unless $x =~ s#^refs/remotes/##;
4449 chomp $x;
4450 next unless -f "$git_dir/$x/info/url";
4451 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4452 next unless $u;
4453 my $dn = dirname("$git_dir/svn/$x");
4454 mkpath([$dn]) unless -d $dn;
4455 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4456 mkpath(["$git_dir/svn/svn"]);
4457 print STDERR " - $git_dir/$x/info => ",
4458 "$git_dir/svn/$x/info\n";
4459 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4460 croak "$!: $x";
4461 # don't worry too much about these, they probably
4462 # don't exist with repos this old (save for index,
4463 # and we can easily regenerate that)
4464 foreach my $f (qw/unhandled.log index .rev_db/) {
4465 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4466 }
4467 } else {
4468 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4469 rename "$git_dir/$x", "$git_dir/svn/$x" or
4470 croak "$!: $x";
4471 }
4472 $migrated++;
4473 }
4474 command_close_pipe($fh, $ctx);
4475 print STDERR "Done migrating from a git-svn v1 layout\n";
4476 $migrated;
4477 }
4478
4479 sub read_old_urls {
4480 my ($l_map, $pfx, $path) = @_;
4481 my @dir;
4482 foreach (<$path/*>) {
4483 if (-r "$_/info/url") {
4484 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4485 my $ref_id = $pfx . basename $_;
4486 my $url = ::file_to_s("$_/info/url");
4487 $l_map->{$ref_id} = $url;
4488 } elsif (-d $_) {
4489 push @dir, $_;
4490 }
4491 }
4492 foreach (@dir) {
4493 my $x = $_;
4494 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4495 read_old_urls($l_map, $x, $_);
4496 }
4497 }
4498
4499 sub migrate_from_v2 {
4500 my @cfg = command(qw/config -l/);
4501 return if grep /^svn-remote\..+\.url=/, @cfg;
4502 my %l_map;
4503 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4504 my $migrated = 0;
4505
4506 foreach my $ref_id (sort keys %l_map) {
4507 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4508 if ($@) {
4509 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4510 }
4511 $migrated++;
4512 }
4513 $migrated;
4514 }
4515
4516 sub minimize_connections {
4517 my $r = Git::SVN::read_all_remotes();
4518 my $new_urls = {};
4519 my $root_repos = {};
4520 foreach my $repo_id (keys %$r) {
4521 my $url = $r->{$repo_id}->{url} or next;
4522 my $fetch = $r->{$repo_id}->{fetch} or next;
4523 my $ra = Git::SVN::Ra->new($url);
4524
4525 # skip existing cases where we already connect to the root
4526 if (($ra->{url} eq $ra->{repos_root}) ||
4527 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
4528 $repo_id)) {
4529 $root_repos->{$ra->{url}} = $repo_id;
4530 next;
4531 }
4532
4533 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4534 my $root_path = $ra->{url};
4535 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4536 foreach my $path (keys %$fetch) {
4537 my $ref_id = $fetch->{$path};
4538 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4539
4540 # make sure we can read when connecting to
4541 # a higher level of a repository
4542 my ($last_rev, undef) = $gs->last_rev_commit;
4543 if (!defined $last_rev) {
4544 $last_rev = eval {
4545 $root_ra->get_latest_revnum;
4546 };
4547 next if $@;
4548 }
4549 my $new = $root_path;
4550 $new .= length $path ? "/$path" : '';
4551 eval {
4552 $root_ra->get_log([$new], $last_rev, $last_rev,
4553 0, 0, 1, sub { });
4554 };
4555 next if $@;
4556 $new_urls->{$ra->{repos_root}}->{$new} =
4557 { ref_id => $ref_id,
4558 old_repo_id => $repo_id,
4559 old_path => $path };
4560 }
4561 }
4562
4563 my @emptied;
4564 foreach my $url (keys %$new_urls) {
4565 # see if we can re-use an existing [svn-remote "repo_id"]
4566 # instead of creating a(n ugly) new section:
4567 my $repo_id = $root_repos->{$url} ||
4568 Git::SVN::sanitize_remote_name($url);
4569
4570 my $fetch = $new_urls->{$url};
4571 foreach my $path (keys %$fetch) {
4572 my $x = $fetch->{$path};
4573 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4574 my $pfx = "svn-remote.$x->{old_repo_id}";
4575
4576 my $old_fetch = quotemeta("$x->{old_path}:".
4577 "refs/remotes/$x->{ref_id}");
4578 command_noisy(qw/config --unset/,
4579 "$pfx.fetch", '^'. $old_fetch . '$');
4580 delete $r->{$x->{old_repo_id}}->
4581 {fetch}->{$x->{old_path}};
4582 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4583 command_noisy(qw/config --unset/,
4584 "$pfx.url");
4585 push @emptied, $x->{old_repo_id}
4586 }
4587 }
4588 }
4589 if (@emptied) {
4590 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4591 "$ENV{GIT_DIR}/config";
4592 print STDERR <<EOF;
4593 The following [svn-remote] sections in your config file ($file) are empty
4594 and can be safely removed:
4595 EOF
4596 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4597 }
4598 }
4599
4600 sub migration_check {
4601 migrate_from_v0();
4602 migrate_from_v1();
4603 migrate_from_v2();
4604 minimize_connections() if $_minimize;
4605 }
4606
4607 package Git::IndexInfo;
4608 use strict;
4609 use warnings;
4610 use Git qw/command_input_pipe command_close_pipe/;
4611
4612 sub new {
4613 my ($class) = @_;
4614 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4615 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4616 }
4617
4618 sub remove {
4619 my ($self, $path) = @_;
4620 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4621 return ++$self->{nr};
4622 }
4623 undef;
4624 }
4625
4626 sub update {
4627 my ($self, $mode, $hash, $path) = @_;
4628 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4629 return ++$self->{nr};
4630 }
4631 undef;
4632 }
4633
4634 sub DESTROY {
4635 my ($self) = @_;
4636 command_close_pipe($self->{gui}, $self->{ctx});
4637 }
4638
4639 package Git::SVN::GlobSpec;
4640 use strict;
4641 use warnings;
4642
4643 sub new {
4644 my ($class, $glob) = @_;
4645 my $re = $glob;
4646 $re =~ s!/+$!!g; # no need for trailing slashes
4647 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
4648 my ($left, $right) = ($1, $2);
4649 if ($nr > 1) {
4650 die "Only one '*' wildcard expansion ",
4651 "is supported (got $nr): '$glob'\n";
4652 } elsif ($nr == 0) {
4653 die "One '*' is needed for glob: '$glob'\n";
4654 }
4655 $re = quotemeta($left) . $re . quotemeta($right);
4656 if (length $left && !($left =~ s!/+$!!g)) {
4657 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4658 }
4659 if (length $right && !($right =~ s!^/+!!g)) {
4660 die "Missing leading '/' on right side of: '$glob' ($right)\n";
4661 }
4662 my $left_re = qr/^\/\Q$left\E(\/|$)/;
4663 bless { left => $left, right => $right, left_regex => $left_re,
4664 regex => qr/$re/, glob => $glob }, $class;
4665 }
4666
4667 sub full_path {
4668 my ($self, $path) = @_;
4669 return (length $self->{left} ? "$self->{left}/" : '') .
4670 $path . (length $self->{right} ? "/$self->{right}" : '');
4671 }
4672
4673 __END__
4674
4675 Data structures:
4676
4677
4678 $remotes = { # returned by read_all_remotes()
4679 'svn' => {
4680 # svn-remote.svn.url=https://svn.musicpd.org
4681 url => 'https://svn.musicpd.org',
4682 # svn-remote.svn.fetch=mpd/trunk:trunk
4683 fetch => {
4684 'mpd/trunk' => 'trunk',
4685 },
4686 # svn-remote.svn.tags=mpd/tags/*:tags/*
4687 tags => {
4688 path => {
4689 left => 'mpd/tags',
4690 right => '',
4691 regex => qr!mpd/tags/([^/]+)$!,
4692 glob => 'tags/*',
4693 },
4694 ref => {
4695 left => 'tags',
4696 right => '',
4697 regex => qr!tags/([^/]+)$!,
4698 glob => 'tags/*',
4699 },
4700 }
4701 }
4702 };
4703
4704 $log_entry hashref as returned by libsvn_log_entry()
4705 {
4706 log => 'whitespace-formatted log entry
4707 ', # trailing newline is preserved
4708 revision => '8', # integer
4709 date => '2004-02-24T17:01:44.108345Z', # commit date
4710 author => 'committer name'
4711 };
4712
4713
4714 # this is generated by generate_diff();
4715 @mods = array of diff-index line hashes, each element represents one line
4716 of diff-index output
4717
4718 diff-index line ($m hash)
4719 {
4720 mode_a => first column of diff-index output, no leading ':',
4721 mode_b => second column of diff-index output,
4722 sha1_b => sha1sum of the final blob,
4723 chg => change type [MCRADT],
4724 file_a => original file name of a file (iff chg is 'C' or 'R')
4725 file_b => new/current file name of a file (any chg)
4726 }
4727 ;
4728
4729 # retval of read_url_paths{,_all}();
4730 $l_map = {
4731 # repository root url
4732 'https://svn.musicpd.org' => {
4733 # repository path # GIT_SVN_ID
4734 'mpd/trunk' => 'trunk',
4735 'mpd/tags/0.11.5' => 'tags/0.11.5',
4736 },
4737 }
4738
4739 Notes:
4740 I don't trust the each() function on unless I created %hash myself
4741 because the internal iterator may not have started at base.
Dscho's miscellaneous git branches