?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'jc/clean-fix'
[git/peff.git] / git-svn.perl
blob9f884eb2132c76b86475b97256e0ed565f84bb2e
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 = $gs->rev_db_get($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_db_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_db_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 $md5->hexdigest();
1232 }
1233
1234 package Git::SVN;
1235 use strict;
1236 use warnings;
1237 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1238 $_repack $_repack_flags $_use_svm_props $_head
1239 $_use_svnsync_props $no_reuse_existing $_minimize_url
1240 $_use_log_author/;
1241 use Carp qw/croak/;
1242 use File::Path qw/mkpath/;
1243 use File::Copy qw/copy/;
1244 use IPC::Open3;
1245
1246 my $_repack_nr;
1247 # properties that we do not log:
1248 my %SKIP_PROP;
1249 BEGIN {
1250 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1251 svn:special svn:executable
1252 svn:entry:committed-rev
1253 svn:entry:last-author
1254 svn:entry:uuid
1255 svn:entry:committed-date/;
1256
1257 # some options are read globally, but can be overridden locally
1258 # per [svn-remote "..."] section. Command-line options will *NOT*
1259 # override options set in an [svn-remote "..."] section
1260 no strict 'refs';
1261 for my $option (qw/follow_parent no_metadata use_svm_props
1262 use_svnsync_props/) {
1263 my $key = $option;
1264 $key =~ tr/_//d;
1265 my $prop = "-$option";
1266 *$option = sub {
1267 my ($self) = @_;
1268 return $self->{$prop} if exists $self->{$prop};
1269 my $k = "svn-remote.$self->{repo_id}.$key";
1270 eval { command_oneline(qw/config --get/, $k) };
1271 if ($@) {
1272 $self->{$prop} = ${"Git::SVN::_$option"};
1273 } else {
1274 my $v = command_oneline(qw/config --bool/,$k);
1275 $self->{$prop} = $v eq 'false' ? 0 : 1;
1276 }
1277 return $self->{$prop};
1278 }
1279 }
1280 }
1281
1282 my %LOCKFILES;
1283 END { unlink keys %LOCKFILES if %LOCKFILES }
1284
1285 sub resolve_local_globs {
1286 my ($url, $fetch, $glob_spec) = @_;
1287 return unless defined $glob_spec;
1288 my $ref = $glob_spec->{ref};
1289 my $path = $glob_spec->{path};
1290 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1291 next unless m#^refs/remotes/$ref->{regex}$#;
1292 my $p = $1;
1293 my $pathname = desanitize_refname($path->full_path($p));
1294 my $refname = desanitize_refname($ref->full_path($p));
1295 if (my $existing = $fetch->{$pathname}) {
1296 if ($existing ne $refname) {
1297 die "Refspec conflict:\n",
1298 "existing: refs/remotes/$existing\n",
1299 " globbed: refs/remotes/$refname\n";
1300 }
1301 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1302 $u =~ s!^\Q$url\E(/|$)!! or die
1303 "refs/remotes/$refname: '$url' not found in '$u'\n";
1304 if ($pathname ne $u) {
1305 warn "W: Refspec glob conflict ",
1306 "(ref: refs/remotes/$refname):\n",
1307 "expected path: $pathname\n",
1308 " real path: $u\n",
1309 "Continuing ahead with $u\n";
1310 next;
1311 }
1312 } else {
1313 $fetch->{$pathname} = $refname;
1314 }
1315 }
1316 }
1317
1318 sub parse_revision_argument {
1319 my ($base, $head) = @_;
1320 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1321 return ($base, $head);
1322 }
1323 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1324 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1325 return ($head, $head) if ($::_revision eq 'HEAD');
1326 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1327 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1328 die "revision argument: $::_revision not understood by git-svn\n";
1329 }
1330
1331 sub fetch_all {
1332 my ($repo_id, $remotes) = @_;
1333 if (ref $repo_id) {
1334 my $gs = $repo_id;
1335 $repo_id = undef;
1336 $repo_id = $gs->{repo_id};
1337 }
1338 $remotes ||= read_all_remotes();
1339 my $remote = $remotes->{$repo_id} or
1340 die "[svn-remote \"$repo_id\"] unknown\n";
1341 my $fetch = $remote->{fetch};
1342 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1343 my (@gs, @globs);
1344 my $ra = Git::SVN::Ra->new($url);
1345 my $uuid = $ra->get_uuid;
1346 my $head = $ra->get_latest_revnum;
1347 my $base = defined $fetch ? $head : 0;
1348
1349 # read the max revs for wildcard expansion (branches/*, tags/*)
1350 foreach my $t (qw/branches tags/) {
1351 defined $remote->{$t} or next;
1352 push @globs, $remote->{$t};
1353 my $max_rev = eval { tmp_config(qw/--int --get/,
1354 "svn-remote.$repo_id.${t}-maxRev") };
1355 if (defined $max_rev && ($max_rev < $base)) {
1356 $base = $max_rev;
1357 } elsif (!defined $max_rev) {
1358 $base = 0;
1359 }
1360 }
1361
1362 if ($fetch) {
1363 foreach my $p (sort keys %$fetch) {
1364 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1365 my $lr = $gs->rev_db_max;
1366 if (defined $lr) {
1367 $base = $lr if ($lr < $base);
1368 }
1369 push @gs, $gs;
1370 }
1371 }
1372
1373 ($base, $head) = parse_revision_argument($base, $head);
1374 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1375 }
1376
1377 sub read_all_remotes {
1378 my $r = {};
1379 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1380 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
1381 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1382 $local_ref =~ s{^/}{};
1383 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1384 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1385 $r->{$1}->{url} = $2;
1386 } elsif (m!^(.+)\.(branches|tags)=
1387 (.*):refs/remotes/(.+)\s*$/!x) {
1388 my ($p, $g) = ($3, $4);
1389 my $rs = $r->{$1}->{$2} = {
1390 t => $2,
1391 remote => $1,
1392 path => Git::SVN::GlobSpec->new($p),
1393 ref => Git::SVN::GlobSpec->new($g) };
1394 if (length($rs->{ref}->{right}) != 0) {
1395 die "The '*' glob character must be the last ",
1396 "character of '$g'\n";
1397 }
1398 }
1399 }
1400 $r;
1401 }
1402
1403 sub init_vars {
1404 if (defined $_repack) {
1405 $_repack = 1000 if ($_repack <= 0);
1406 $_repack_nr = $_repack;
1407 $_repack_flags ||= '-d';
1408 }
1409 }
1410
1411 sub verify_remotes_sanity {
1412 return unless -d $ENV{GIT_DIR};
1413 my %seen;
1414 foreach (command(qw/config -l/)) {
1415 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1416 if ($seen{$1}) {
1417 die "Remote ref refs/remote/$1 is tracked by",
1418 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1419 "Please resolve this ambiguity in ",
1420 "your git configuration file before ",
1421 "continuing\n";
1422 }
1423 $seen{$1} = $_;
1424 }
1425 }
1426 }
1427
1428 # we allow more chars than remotes2config.sh...
1429 sub sanitize_remote_name {
1430 my ($name) = @_;
1431 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1432 $name;
1433 }
1434
1435 sub find_existing_remote {
1436 my ($url, $remotes) = @_;
1437 return undef if $no_reuse_existing;
1438 my $existing;
1439 foreach my $repo_id (keys %$remotes) {
1440 my $u = $remotes->{$repo_id}->{url} or next;
1441 next if $u ne $url;
1442 $existing = $repo_id;
1443 last;
1444 }
1445 $existing;
1446 }
1447
1448 sub init_remote_config {
1449 my ($self, $url, $no_write) = @_;
1450 $url =~ s!/+$!!; # strip trailing slash
1451 my $r = read_all_remotes();
1452 my $existing = find_existing_remote($url, $r);
1453 if ($existing) {
1454 unless ($no_write) {
1455 print STDERR "Using existing ",
1456 "[svn-remote \"$existing\"]\n";
1457 }
1458 $self->{repo_id} = $existing;
1459 } elsif ($_minimize_url) {
1460 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1461 $existing = find_existing_remote($min_url, $r);
1462 if ($existing) {
1463 unless ($no_write) {
1464 print STDERR "Using existing ",
1465 "[svn-remote \"$existing\"]\n";
1466 }
1467 $self->{repo_id} = $existing;
1468 }
1469 if ($min_url ne $url) {
1470 unless ($no_write) {
1471 print STDERR "Using higher level of URL: ",
1472 "$url => $min_url\n";
1473 }
1474 my $old_path = $self->{path};
1475 $self->{path} = $url;
1476 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1477 if (length $old_path) {
1478 $self->{path} .= "/$old_path";
1479 }
1480 $url = $min_url;
1481 }
1482 }
1483 my $orig_url;
1484 if (!$existing) {
1485 # verify that we aren't overwriting anything:
1486 $orig_url = eval {
1487 command_oneline('config', '--get',
1488 "svn-remote.$self->{repo_id}.url")
1489 };
1490 if ($orig_url && ($orig_url ne $url)) {
1491 die "svn-remote.$self->{repo_id}.url already set: ",
1492 "$orig_url\nwanted to set to: $url\n";
1493 }
1494 }
1495 my ($xrepo_id, $xpath) = find_ref($self->refname);
1496 if (defined $xpath) {
1497 die "svn-remote.$xrepo_id.fetch already set to track ",
1498 "$xpath:refs/remotes/", $self->refname, "\n";
1499 }
1500 unless ($no_write) {
1501 command_noisy('config',
1502 "svn-remote.$self->{repo_id}.url", $url);
1503 $self->{path} =~ s{^/}{};
1504 command_noisy('config', '--add',
1505 "svn-remote.$self->{repo_id}.fetch",
1506 "$self->{path}:".$self->refname);
1507 }
1508 $self->{url} = $url;
1509 }
1510
1511 sub find_by_url { # repos_root and, path are optional
1512 my ($class, $full_url, $repos_root, $path) = @_;
1513
1514 return undef unless defined $full_url;
1515 remove_username($full_url);
1516 remove_username($repos_root) if defined $repos_root;
1517 my $remotes = read_all_remotes();
1518 if (defined $full_url && defined $repos_root && !defined $path) {
1519 $path = $full_url;
1520 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1521 }
1522 foreach my $repo_id (keys %$remotes) {
1523 my $u = $remotes->{$repo_id}->{url} or next;
1524 remove_username($u);
1525 next if defined $repos_root && $repos_root ne $u;
1526
1527 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1528 foreach (qw/branches tags/) {
1529 resolve_local_globs($u, $fetch,
1530 $remotes->{$repo_id}->{$_});
1531 }
1532 my $p = $path;
1533 unless (defined $p) {
1534 $p = $full_url;
1535 $p =~ s#^\Q$u\E(?:/|$)## or next;
1536 }
1537 foreach my $f (keys %$fetch) {
1538 next if $f ne $p;
1539 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1540 }
1541 }
1542 undef;
1543 }
1544
1545 sub init {
1546 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1547 my $self = _new($class, $repo_id, $ref_id, $path);
1548 if (defined $url) {
1549 $self->init_remote_config($url, $no_write);
1550 }
1551 $self;
1552 }
1553
1554 sub find_ref {
1555 my ($ref_id) = @_;
1556 foreach (command(qw/config -l/)) {
1557 next unless m!^svn-remote\.(.+)\.fetch=
1558 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1559 my ($repo_id, $path, $ref) = ($1, $2, $3);
1560 if ($ref eq $ref_id) {
1561 $path = '' if ($path =~ m#^\./?#);
1562 return ($repo_id, $path);
1563 }
1564 }
1565 (undef, undef, undef);
1566 }
1567
1568 sub new {
1569 my ($class, $ref_id, $repo_id, $path) = @_;
1570 if (defined $ref_id && !defined $repo_id && !defined $path) {
1571 ($repo_id, $path) = find_ref($ref_id);
1572 if (!defined $repo_id) {
1573 die "Could not find a \"svn-remote.*.fetch\" key ",
1574 "in the repository configuration matching: ",
1575 "refs/remotes/$ref_id\n";
1576 }
1577 }
1578 my $self = _new($class, $repo_id, $ref_id, $path);
1579 if (!defined $self->{path} || !length $self->{path}) {
1580 my $fetch = command_oneline('config', '--get',
1581 "svn-remote.$repo_id.fetch",
1582 ":refs/remotes/$ref_id\$") or
1583 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1584 "\":refs/remotes/$ref_id\$\" in config\n";
1585 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1586 }
1587 $self->{url} = command_oneline('config', '--get',
1588 "svn-remote.$repo_id.url") or
1589 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1590 $self->rebuild;
1591 $self;
1592 }
1593
1594 sub refname {
1595 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1596
1597 # It cannot end with a slash /, we'll throw up on this because
1598 # SVN can't have directories with a slash in their name, either:
1599 if ($refname =~ m{/$}) {
1600 die "ref: '$refname' ends with a trailing slash, this is ",
1601 "not permitted by git nor Subversion\n";
1602 }
1603
1604 # It cannot have ASCII control character space, tilde ~, caret ^,
1605 # colon :, question-mark ?, asterisk *, space, or open bracket [
1606 # anywhere.
1607 #
1608 # Additionally, % must be escaped because it is used for escaping
1609 # and we want our escaped refname to be reversible
1610 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1611
1612 # no slash-separated component can begin with a dot .
1613 # /.* becomes /%2E*
1614 $refname =~ s{/\.}{/%2E}g;
1615
1616 # It cannot have two consecutive dots .. anywhere
1617 # .. becomes %2E%2E
1618 $refname =~ s{\.\.}{%2E%2E}g;
1619
1620 return $refname;
1621 }
1622
1623 sub desanitize_refname {
1624 my ($refname) = @_;
1625 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1626 return $refname;
1627 }
1628
1629 sub svm_uuid {
1630 my ($self) = @_;
1631 return $self->{svm}->{uuid} if $self->svm;
1632 $self->ra;
1633 unless ($self->{svm}) {
1634 die "SVM UUID not cached, and reading remotely failed\n";
1635 }
1636 $self->{svm}->{uuid};
1637 }
1638
1639 sub svm {
1640 my ($self) = @_;
1641 return $self->{svm} if $self->{svm};
1642 my $svm;
1643 # see if we have it in our config, first:
1644 eval {
1645 my $section = "svn-remote.$self->{repo_id}";
1646 $svm = {
1647 source => tmp_config('--get', "$section.svm-source"),
1648 uuid => tmp_config('--get', "$section.svm-uuid"),
1649 replace => tmp_config('--get', "$section.svm-replace"),
1650 }
1651 };
1652 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1653 $self->{svm} = $svm;
1654 }
1655 $self->{svm};
1656 }
1657
1658 sub _set_svm_vars {
1659 my ($self, $ra) = @_;
1660 return $ra if $self->svm;
1661
1662 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1663 "(svm:source, svm:uuid) ",
1664 "from the following URLs:\n" );
1665 sub read_svm_props {
1666 my ($self, $ra, $path, $r) = @_;
1667 my $props = ($ra->get_dir($path, $r))[2];
1668 my $src = $props->{'svm:source'};
1669 my $uuid = $props->{'svm:uuid'};
1670 return undef if (!$src || !$uuid);
1671
1672 chomp($src, $uuid);
1673
1674 $uuid =~ m{^[0-9a-f\-]{30,}$}
1675 or die "doesn't look right - svm:uuid is '$uuid'\n";
1676
1677 # the '!' is used to mark the repos_root!/relative/path
1678 $src =~ s{/?!/?}{/};
1679 $src =~ s{/+$}{}; # no trailing slashes please
1680 # username is of no interest
1681 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1682
1683 my $replace = $ra->{url};
1684 $replace .= "/$path" if length $path;
1685
1686 my $section = "svn-remote.$self->{repo_id}";
1687 tmp_config("$section.svm-source", $src);
1688 tmp_config("$section.svm-replace", $replace);
1689 tmp_config("$section.svm-uuid", $uuid);
1690 $self->{svm} = {
1691 source => $src,
1692 uuid => $uuid,
1693 replace => $replace
1694 };
1695 }
1696
1697 my $r = $ra->get_latest_revnum;
1698 my $path = $self->{path};
1699 my %tried;
1700 while (length $path) {
1701 unless ($tried{"$self->{url}/$path"}) {
1702 return $ra if $self->read_svm_props($ra, $path, $r);
1703 $tried{"$self->{url}/$path"} = 1;
1704 }
1705 $path =~ s#/?[^/]+$##;
1706 }
1707 die "Path: '$path' should be ''\n" if $path ne '';
1708 return $ra if $self->read_svm_props($ra, $path, $r);
1709 $tried{"$self->{url}/$path"} = 1;
1710
1711 if ($ra->{repos_root} eq $self->{url}) {
1712 die @err, (map { " $_\n" } keys %tried), "\n";
1713 }
1714
1715 # nope, make sure we're connected to the repository root:
1716 my $ok;
1717 my @tried_b;
1718 $path = $ra->{svn_path};
1719 $ra = Git::SVN::Ra->new($ra->{repos_root});
1720 while (length $path) {
1721 unless ($tried{"$ra->{url}/$path"}) {
1722 $ok = $self->read_svm_props($ra, $path, $r);
1723 last if $ok;
1724 $tried{"$ra->{url}/$path"} = 1;
1725 }
1726 $path =~ s#/?[^/]+$##;
1727 }
1728 die "Path: '$path' should be ''\n" if $path ne '';
1729 $ok ||= $self->read_svm_props($ra, $path, $r);
1730 $tried{"$ra->{url}/$path"} = 1;
1731 if (!$ok) {
1732 die @err, (map { " $_\n" } keys %tried), "\n";
1733 }
1734 Git::SVN::Ra->new($self->{url});
1735 }
1736
1737 sub svnsync {
1738 my ($self) = @_;
1739 return $self->{svnsync} if $self->{svnsync};
1740
1741 if ($self->no_metadata) {
1742 die "Can't have both 'noMetadata' and ",
1743 "'useSvnsyncProps' options set!\n";
1744 }
1745 if ($self->rewrite_root) {
1746 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1747 "options set!\n";
1748 }
1749
1750 my $svnsync;
1751 # see if we have it in our config, first:
1752 eval {
1753 my $section = "svn-remote.$self->{repo_id}";
1754 $svnsync = {
1755 url => tmp_config('--get', "$section.svnsync-url"),
1756 uuid => tmp_config('--get', "$section.svnsync-uuid"),
1757 }
1758 };
1759 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1760 return $self->{svnsync} = $svnsync;
1761 }
1762
1763 my $err = "useSvnsyncProps set, but failed to read " .
1764 "svnsync property: svn:sync-from-";
1765 my $rp = $self->ra->rev_proplist(0);
1766
1767 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1768 $url =~ m{^[a-z\+]+://} or
1769 die "doesn't look right - svn:sync-from-url is '$url'\n";
1770
1771 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1772 $uuid =~ m{^[0-9a-f\-]{30,}$} or
1773 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1774
1775 my $section = "svn-remote.$self->{repo_id}";
1776 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1777 tmp_config('--add', "$section.svnsync-url", $url);
1778 return $self->{svnsync} = { url => $url, uuid => $uuid };
1779 }
1780
1781 # this allows us to memoize our SVN::Ra UUID locally and avoid a
1782 # remote lookup (useful for 'git svn log').
1783 sub ra_uuid {
1784 my ($self) = @_;
1785 unless ($self->{ra_uuid}) {
1786 my $key = "svn-remote.$self->{repo_id}.uuid";
1787 my $uuid = eval { tmp_config('--get', $key) };
1788 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1789 $self->{ra_uuid} = $uuid;
1790 } else {
1791 die "ra_uuid called without URL\n" unless $self->{url};
1792 $self->{ra_uuid} = $self->ra->get_uuid;
1793 tmp_config('--add', $key, $self->{ra_uuid});
1794 }
1795 }
1796 $self->{ra_uuid};
1797 }
1798
1799 sub _set_repos_root {
1800 my ($self, $repos_root) = @_;
1801 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1802 $repos_root ||= $self->ra->{repos_root};
1803 tmp_config($k, $repos_root);
1804 $repos_root;
1805 }
1806
1807 sub repos_root {
1808 my ($self) = @_;
1809 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1810 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1811 }
1812
1813 sub ra {
1814 my ($self) = shift;
1815 my $ra = Git::SVN::Ra->new($self->{url});
1816 $self->_set_repos_root($ra->{repos_root});
1817 if ($self->use_svm_props && !$self->{svm}) {
1818 if ($self->no_metadata) {
1819 die "Can't have both 'noMetadata' and ",
1820 "'useSvmProps' options set!\n";
1821 } elsif ($self->use_svnsync_props) {
1822 die "Can't have both 'useSvnsyncProps' and ",
1823 "'useSvmProps' options set!\n";
1824 }
1825 $ra = $self->_set_svm_vars($ra);
1826 $self->{-want_revprops} = 1;
1827 }
1828 $ra;
1829 }
1830
1831 sub rel_path {
1832 my ($self) = @_;
1833 my $repos_root = $self->ra->{repos_root};
1834 return $self->{path} if ($self->{url} eq $repos_root);
1835 my $url = $self->{url} .
1836 (length $self->{path} ? "/$self->{path}" : $self->{path});
1837 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1838 $url;
1839 }
1840
1841 # prop_walk(PATH, REV, SUB)
1842 # -------------------------
1843 # Recursively traverse PATH at revision REV and invoke SUB for each
1844 # directory that contains a SVN property. SUB will be invoked as
1845 # follows: &SUB(gs, path, props); where `gs' is this instance of
1846 # Git::SVN, `path' the path to the directory where the properties
1847 # `props' were found. The `path' will be relative to point of checkout,
1848 # that is, if url://repo/trunk is the current Git branch, and that
1849 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
1850 # as `path' (note the trailing `/').
1851 sub prop_walk {
1852 my ($self, $path, $rev, $sub) = @_;
1853
1854 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1855 $path =~ s#^/*#/#g;
1856 my $p = $path;
1857 # Strip the irrelevant part of the path.
1858 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1859 # Ensure the path is terminated by a `/'.
1860 $p =~ s#/*$#/#;
1861
1862 # The properties contain all the internal SVN stuff nobody
1863 # (usually) cares about.
1864 my $interesting_props = 0;
1865 foreach (keys %{$props}) {
1866 # If it doesn't start with `svn:', it must be a
1867 # user-defined property.
1868 ++$interesting_props and next if $_ !~ /^svn:/;
1869 # FIXME: Fragile, if SVN adds new public properties,
1870 # this needs to be updated.
1871 ++$interesting_props if /^svn:(?:ignore|keywords|executable
1872 |eol-style|mime-type
1873 |externals|needs-lock)$/x;
1874 }
1875 &$sub($self, $p, $props) if $interesting_props;
1876
1877 foreach (sort keys %$dirent) {
1878 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1879 $self->prop_walk($path . '/' . $_, $rev, $sub);
1880 }
1881 }
1882
1883 sub last_rev { ($_[0]->last_rev_commit)[0] }
1884 sub last_commit { ($_[0]->last_rev_commit)[1] }
1885
1886 # returns the newest SVN revision number and newest commit SHA1
1887 sub last_rev_commit {
1888 my ($self) = @_;
1889 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1890 return ($self->{last_rev}, $self->{last_commit});
1891 }
1892 my $c = ::verify_ref($self->refname.'^0');
1893 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1894 my $rev = (::cmt_metadata($c))[1];
1895 if (defined $rev) {
1896 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1897 return ($rev, $c);
1898 }
1899 }
1900 my $db_path = $self->db_path;
1901 unless (-e $db_path) {
1902 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1903 return (undef, undef);
1904 }
1905 my $offset = -41; # from tail
1906 my $rl;
1907 open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1908 sysseek($fh, $offset, 2); # don't care for errors
1909 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1910 chomp $rl;
1911 while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1912 $offset -= 41;
1913 sysseek($fh, $offset, 2); # don't care for errors
1914 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1915 chomp $rl;
1916 }
1917 if ($c && $c ne $rl) {
1918 die "$db_path and ", $self->refname,
1919 " inconsistent!:\n$c != $rl\n";
1920 }
1921 my $rev = sysseek($fh, 0, 1) or croak $!;
1922 $rev = ($rev - 41) / 41;
1923 close $fh or croak $!;
1924 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1925 return ($rev, $c);
1926 }
1927
1928 sub get_fetch_range {
1929 my ($self, $min, $max) = @_;
1930 $max ||= $self->ra->get_latest_revnum;
1931 $min ||= $self->rev_db_max;
1932 (++$min, $max);
1933 }
1934
1935 sub tmp_config {
1936 my (@args) = @_;
1937 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1938 my $config = "$ENV{GIT_DIR}/svn/.metadata";
1939 if (! -f $config && -f $old_def_config) {
1940 rename $old_def_config, $config or
1941 die "Failed rename $old_def_config => $config: $!\n";
1942 }
1943 my $old_config = $ENV{GIT_CONFIG};
1944 $ENV{GIT_CONFIG} = $config;
1945 $@ = undef;
1946 my @ret = eval {
1947 unless (-f $config) {
1948 mkfile($config);
1949 open my $fh, '>', $config or
1950 die "Can't open $config: $!\n";
1951 print $fh "; This file is used internally by ",
1952 "git-svn\n" or die
1953 "Couldn't write to $config: $!\n";
1954 print $fh "; You should not have to edit it\n" or
1955 die "Couldn't write to $config: $!\n";
1956 close $fh or die "Couldn't close $config: $!\n";
1957 }
1958 command('config', @args);
1959 };
1960 my $err = $@;
1961 if (defined $old_config) {
1962 $ENV{GIT_CONFIG} = $old_config;
1963 } else {
1964 delete $ENV{GIT_CONFIG};
1965 }
1966 die $err if $err;
1967 wantarray ? @ret : $ret[0];
1968 }
1969
1970 sub tmp_index_do {
1971 my ($self, $sub) = @_;
1972 my $old_index = $ENV{GIT_INDEX_FILE};
1973 $ENV{GIT_INDEX_FILE} = $self->{index};
1974 $@ = undef;
1975 my @ret = eval {
1976 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1977 mkpath([$dir]) unless -d $dir;
1978 &$sub;
1979 };
1980 my $err = $@;
1981 if (defined $old_index) {
1982 $ENV{GIT_INDEX_FILE} = $old_index;
1983 } else {
1984 delete $ENV{GIT_INDEX_FILE};
1985 }
1986 die $err if $err;
1987 wantarray ? @ret : $ret[0];
1988 }
1989
1990 sub assert_index_clean {
1991 my ($self, $treeish) = @_;
1992
1993 $self->tmp_index_do(sub {
1994 command_noisy('read-tree', $treeish) unless -e $self->{index};
1995 my $x = command_oneline('write-tree');
1996 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1997 /^tree ($::sha1)/mo);
1998 return if $y eq $x;
1999
2000 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2001 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2002 command_noisy('read-tree', $treeish);
2003 $x = command_oneline('write-tree');
2004 if ($y ne $x) {
2005 ::fatal "trees ($treeish) $y != $x\n",
2006 "Something is seriously wrong...";
2007 }
2008 });
2009 }
2010
2011 sub get_commit_parents {
2012 my ($self, $log_entry) = @_;
2013 my (%seen, @ret, @tmp);
2014 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2015 if (my $ip = $self->{inject_parents}) {
2016 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2017 push @tmp, $commit;
2018 }
2019 }
2020 if (my $cur = ::verify_ref($self->refname.'^0')) {
2021 push @tmp, $cur;
2022 }
2023 if (my $ipd = $self->{inject_parents_dcommit}) {
2024 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2025 push @tmp, @$commit;
2026 }
2027 }
2028 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2029 while (my $p = shift @tmp) {
2030 next if $seen{$p};
2031 $seen{$p} = 1;
2032 push @ret, $p;
2033 # MAXPARENT is defined to 16 in commit-tree.c:
2034 last if @ret >= 16;
2035 }
2036 if (@tmp) {
2037 die "r$log_entry->{revision}: No room for parents:\n\t",
2038 join("\n\t", @tmp), "\n";
2039 }
2040 @ret;
2041 }
2042
2043 sub rewrite_root {
2044 my ($self) = @_;
2045 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2046 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2047 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2048 if ($rwr) {
2049 $rwr =~ s#/+$##;
2050 if ($rwr !~ m#^[a-z\+]+://#) {
2051 die "$rwr is not a valid URL (key: $k)\n";
2052 }
2053 }
2054 $self->{-rewrite_root} = $rwr;
2055 }
2056
2057 sub metadata_url {
2058 my ($self) = @_;
2059 ($self->rewrite_root || $self->{url}) .
2060 (length $self->{path} ? '/' . $self->{path} : '');
2061 }
2062
2063 sub full_url {
2064 my ($self) = @_;
2065 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2066 }
2067
2068 sub do_git_commit {
2069 my ($self, $log_entry) = @_;
2070 my $lr = $self->last_rev;
2071 if (defined $lr && $lr >= $log_entry->{revision}) {
2072 die "Last fetched revision of ", $self->refname,
2073 " was r$lr, but we are about to fetch: ",
2074 "r$log_entry->{revision}!\n";
2075 }
2076 if (my $c = $self->rev_db_get($log_entry->{revision})) {
2077 croak "$log_entry->{revision} = $c already exists! ",
2078 "Why are we refetching it?\n";
2079 }
2080 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2081 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2082 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2083
2084 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2085 ? $log_entry->{commit_name}
2086 : $log_entry->{name};
2087 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2088 ? $log_entry->{commit_email}
2089 : $log_entry->{email};
2090
2091 my $tree = $log_entry->{tree};
2092 if (!defined $tree) {
2093 $tree = $self->tmp_index_do(sub {
2094 command_oneline('write-tree') });
2095 }
2096 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2097
2098 my @exec = ('git-commit-tree', $tree);
2099 foreach ($self->get_commit_parents($log_entry)) {
2100 push @exec, '-p', $_;
2101 }
2102 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2103 or croak $!;
2104 print $msg_fh $log_entry->{log} or croak $!;
2105 unless ($self->no_metadata) {
2106 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2107 or croak $!;
2108 }
2109 $msg_fh->flush == 0 or croak $!;
2110 close $msg_fh or croak $!;
2111 chomp(my $commit = do { local $/; <$out_fh> });
2112 close $out_fh or croak $!;
2113 waitpid $pid, 0;
2114 croak $? if $?;
2115 if ($commit !~ /^$::sha1$/o) {
2116 die "Failed to commit, invalid sha1: $commit\n";
2117 }
2118
2119 $self->rev_db_set($log_entry->{revision}, $commit, 1);
2120
2121 $self->{last_rev} = $log_entry->{revision};
2122 $self->{last_commit} = $commit;
2123 print "r$log_entry->{revision}";
2124 if (defined $log_entry->{svm_revision}) {
2125 print " (\@$log_entry->{svm_revision})";
2126 $self->rev_db_set($log_entry->{svm_revision}, $commit,
2127 0, $self->svm_uuid);
2128 }
2129 print " = $commit ($self->{ref_id})\n";
2130 if (defined $_repack && (--$_repack_nr == 0)) {
2131 $_repack_nr = $_repack;
2132 # repack doesn't use any arguments with spaces in them, does it?
2133 print "Running git repack $_repack_flags ...\n";
2134 command_noisy('repack', split(/\s+/, $_repack_flags));
2135 print "Done repacking\n";
2136 }
2137 return $commit;
2138 }
2139
2140 sub match_paths {
2141 my ($self, $paths, $r) = @_;
2142 return 1 if $self->{path} eq '';
2143 if (my $path = $paths->{"/$self->{path}"}) {
2144 return ($path->{action} eq 'D') ? 0 : 1;
2145 }
2146 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2147 if (grep /$self->{path_regex}/, keys %$paths) {
2148 return 1;
2149 }
2150 my $c = '';
2151 foreach (split m#/#, $self->{path}) {
2152 $c .= "/$_";
2153 next unless ($paths->{$c} &&
2154 ($paths->{$c}->{action} =~ /^[AR]$/));
2155 if ($self->ra->check_path($self->{path}, $r) ==
2156 $SVN::Node::dir) {
2157 return 1;
2158 }
2159 }
2160 return 0;
2161 }
2162
2163 sub find_parent_branch {
2164 my ($self, $paths, $rev) = @_;
2165 return undef unless $self->follow_parent;
2166 unless (defined $paths) {
2167 my $err_handler = $SVN::Error::handler;
2168 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2169 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2170 $paths =
2171 Git::SVN::Ra::dup_changed_paths($_[0]) });
2172 $SVN::Error::handler = $err_handler;
2173 }
2174 return undef unless defined $paths;
2175
2176 # look for a parent from another branch:
2177 my @b_path_components = split m#/#, $self->rel_path;
2178 my @a_path_components;
2179 my $i;
2180 while (@b_path_components) {
2181 $i = $paths->{'/'.join('/', @b_path_components)};
2182 last if $i && defined $i->{copyfrom_path};
2183 unshift(@a_path_components, pop(@b_path_components));
2184 }
2185 return undef unless defined $i && defined $i->{copyfrom_path};
2186 my $branch_from = $i->{copyfrom_path};
2187 if (@a_path_components) {
2188 print STDERR "branch_from: $branch_from => ";
2189 $branch_from .= '/'.join('/', @a_path_components);
2190 print STDERR $branch_from, "\n";
2191 }
2192 my $r = $i->{copyfrom_rev};
2193 my $repos_root = $self->ra->{repos_root};
2194 my $url = $self->ra->{url};
2195 my $new_url = $repos_root . $branch_from;
2196 print STDERR "Found possible branch point: ",
2197 "$new_url => ", $self->full_url, ", $r\n";
2198 $branch_from =~ s#^/##;
2199 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2200 unless ($gs) {
2201 my $ref_id = $self->{ref_id};
2202 $ref_id =~ s/\@\d+$//;
2203 $ref_id .= "\@$r";
2204 # just grow a tail if we're not unique enough :x
2205 $ref_id .= '-' while find_ref($ref_id);
2206 print STDERR "Initializing parent: $ref_id\n";
2207 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
2208 }
2209 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2210 if (!defined $r0 || !defined $parent) {
2211 my ($base, $head) = parse_revision_argument(0, $r);
2212 if ($base <= $r) {
2213 $gs->fetch($base, $r);
2214 }
2215 ($r0, $parent) = $gs->last_rev_commit;
2216 }
2217 if (defined $r0 && defined $parent) {
2218 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2219 my $ed;
2220 if ($self->ra->can_do_switch) {
2221 $self->assert_index_clean($parent);
2222 print STDERR "Following parent with do_switch\n";
2223 # do_switch works with svn/trunk >= r22312, but that
2224 # is not included with SVN 1.4.3 (the latest version
2225 # at the moment), so we can't rely on it
2226 $self->{last_commit} = $parent;
2227 $ed = SVN::Git::Fetcher->new($self);
2228 $gs->ra->gs_do_switch($r0, $rev, $gs,
2229 $self->full_url, $ed)
2230 or die "SVN connection failed somewhere...\n";
2231 } elsif ($self->ra->trees_match($new_url, $r0,
2232 $self->full_url, $rev)) {
2233 print STDERR "Trees match:\n",
2234 " $new_url\@$r0\n",
2235 " ${\$self->full_url}\@$rev\n",
2236 "Following parent with no changes\n";
2237 $self->tmp_index_do(sub {
2238 command_noisy('read-tree', $parent);
2239 });
2240 $self->{last_commit} = $parent;
2241 } else {
2242 print STDERR "Following parent with do_update\n";
2243 $ed = SVN::Git::Fetcher->new($self);
2244 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2245 or die "SVN connection failed somewhere...\n";
2246 }
2247 print STDERR "Successfully followed parent\n";
2248 return $self->make_log_entry($rev, [$parent], $ed);
2249 }
2250 return undef;
2251 }
2252
2253 sub do_fetch {
2254 my ($self, $paths, $rev) = @_;
2255 my $ed;
2256 my ($last_rev, @parents);
2257 if (my $lc = $self->last_commit) {
2258 # we can have a branch that was deleted, then re-added
2259 # under the same name but copied from another path, in
2260 # which case we'll have multiple parents (we don't
2261 # want to break the original ref, nor lose copypath info):
2262 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2263 push @{$log_entry->{parents}}, $lc;
2264 return $log_entry;
2265 }
2266 $ed = SVN::Git::Fetcher->new($self);
2267 $last_rev = $self->{last_rev};
2268 $ed->{c} = $lc;
2269 @parents = ($lc);
2270 } else {
2271 $last_rev = $rev;
2272 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2273 return $log_entry;
2274 }
2275 $ed = SVN::Git::Fetcher->new($self);
2276 }
2277 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2278 die "SVN connection failed somewhere...\n";
2279 }
2280 $self->make_log_entry($rev, \@parents, $ed);
2281 }
2282
2283 sub get_untracked {
2284 my ($self, $ed) = @_;
2285 my @out;
2286 my $h = $ed->{empty};
2287 foreach (sort keys %$h) {
2288 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2289 push @out, " $act: " . uri_encode($_);
2290 warn "W: $act: $_\n";
2291 }
2292 foreach my $t (qw/dir_prop file_prop/) {
2293 $h = $ed->{$t} or next;
2294 foreach my $path (sort keys %$h) {
2295 my $ppath = $path eq '' ? '.' : $path;
2296 foreach my $prop (sort keys %{$h->{$path}}) {
2297 next if $SKIP_PROP{$prop};
2298 my $v = $h->{$path}->{$prop};
2299 my $t_ppath_prop = "$t: " .
2300 uri_encode($ppath) . ' ' .
2301 uri_encode($prop);
2302 if (defined $v) {
2303 push @out, " +$t_ppath_prop " .
2304 uri_encode($v);
2305 } else {
2306 push @out, " -$t_ppath_prop";
2307 }
2308 }
2309 }
2310 }
2311 foreach my $t (qw/absent_file absent_directory/) {
2312 $h = $ed->{$t} or next;
2313 foreach my $parent (sort keys %$h) {
2314 foreach my $path (sort @{$h->{$parent}}) {
2315 push @out, " $t: " .
2316 uri_encode("$parent/$path");
2317 warn "W: $t: $parent/$path ",
2318 "Insufficient permissions?\n";
2319 }
2320 }
2321 }
2322 \@out;
2323 }
2324
2325 sub parse_svn_date {
2326 my $date = shift || return '+0000 1970-01-01 00:00:00';
2327 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2328 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2329 croak "Unable to parse date: $date\n";
2330 "+0000 $Y-$m-$d $H:$M:$S";
2331 }
2332
2333 sub check_author {
2334 my ($author) = @_;
2335 if (!defined $author || length $author == 0) {
2336 $author = '(no author)';
2337 }
2338 if (defined $::_authors && ! defined $::users{$author}) {
2339 die "Author: $author not defined in $::_authors file\n";
2340 }
2341 $author;
2342 }
2343
2344 sub make_log_entry {
2345 my ($self, $rev, $parents, $ed) = @_;
2346 my $untracked = $self->get_untracked($ed);
2347
2348 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2349 print $un "r$rev\n" or croak $!;
2350 print $un $_, "\n" foreach @$untracked;
2351 my %log_entry = ( parents => $parents || [], revision => $rev,
2352 log => '');
2353
2354 my $headrev;
2355 my $logged = delete $self->{logged_rev_props};
2356 if (!$logged || $self->{-want_revprops}) {
2357 my $rp = $self->ra->rev_proplist($rev);
2358 foreach (sort keys %$rp) {
2359 my $v = $rp->{$_};
2360 if (/^svn:(author|date|log)$/) {
2361 $log_entry{$1} = $v;
2362 } elsif ($_ eq 'svm:headrev') {
2363 $headrev = $v;
2364 } else {
2365 print $un " rev_prop: ", uri_encode($_), ' ',
2366 uri_encode($v), "\n";
2367 }
2368 }
2369 } else {
2370 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2371 }
2372 close $un or croak $!;
2373
2374 $log_entry{date} = parse_svn_date($log_entry{date});
2375 $log_entry{log} .= "\n";
2376 my $author = $log_entry{author} = check_author($log_entry{author});
2377 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2378 : ($author, undef);
2379
2380 my ($commit_name, $commit_email) = ($name, $email);
2381 if ($_use_log_author) {
2382 if ($log_entry{log} =~ /From:\s+(.*?)\s+<(.*)>\s*\n/) {
2383 ($name, $email) = ($1, $2);
2384 } elsif ($log_entry{log} =~
2385 /Signed-off-by:\s+(.*?)\s+<(.*)>\s*\n/) {
2386 ($name, $email) = ($1, $2);
2387 }
2388 }
2389 if (defined $headrev && $self->use_svm_props) {
2390 if ($self->rewrite_root) {
2391 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2392 "options set!\n";
2393 }
2394 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2395 # we don't want "SVM: initializing mirror for junk" ...
2396 return undef if $r == 0;
2397 my $svm = $self->svm;
2398 if ($uuid ne $svm->{uuid}) {
2399 die "UUID mismatch on SVM path:\n",
2400 "expected: $svm->{uuid}\n",
2401 " got: $uuid\n";
2402 }
2403 my $full_url = $self->full_url;
2404 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2405 die "Failed to replace '$svm->{replace}' with ",
2406 "'$svm->{source}' in $full_url\n";
2407 # throw away username for storing in records
2408 remove_username($full_url);
2409 $log_entry{metadata} = "$full_url\@$r $uuid";
2410 $log_entry{svm_revision} = $r;
2411 $email ||= "$author\@$uuid";
2412 $commit_email ||= "$author\@$uuid";
2413 } elsif ($self->use_svnsync_props) {
2414 my $full_url = $self->svnsync->{url};
2415 $full_url .= "/$self->{path}" if length $self->{path};
2416 remove_username($full_url);
2417 my $uuid = $self->svnsync->{uuid};
2418 $log_entry{metadata} = "$full_url\@$rev $uuid";
2419 $email ||= "$author\@$uuid";
2420 $commit_email ||= "$author\@$uuid";
2421 } else {
2422 my $url = $self->metadata_url;
2423 remove_username($url);
2424 $log_entry{metadata} = "$url\@$rev " .
2425 $self->ra->get_uuid;
2426 $email ||= "$author\@" . $self->ra->get_uuid;
2427 $commit_email ||= "$author\@" . $self->ra->get_uuid;
2428 }
2429 $log_entry{name} = $name;
2430 $log_entry{email} = $email;
2431 $log_entry{commit_name} = $commit_name;
2432 $log_entry{commit_email} = $commit_email;
2433 \%log_entry;
2434 }
2435
2436 sub fetch {
2437 my ($self, $min_rev, $max_rev, @parents) = @_;
2438 my ($last_rev, $last_commit) = $self->last_rev_commit;
2439 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2440 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2441 }
2442
2443 sub set_tree_cb {
2444 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2445 $self->{inject_parents} = { $rev => $tree };
2446 $self->fetch(undef, undef);
2447 }
2448
2449 sub set_tree {
2450 my ($self, $tree) = (shift, shift);
2451 my $log_entry = ::get_commit_entry($tree);
2452 unless ($self->{last_rev}) {
2453 fatal("Must have an existing revision to commit");
2454 }
2455 my %ed_opts = ( r => $self->{last_rev},
2456 log => $log_entry->{log},
2457 ra => $self->ra,
2458 tree_a => $self->{last_commit},
2459 tree_b => $tree,
2460 editor_cb => sub {
2461 $self->set_tree_cb($log_entry, $tree, @_) },
2462 svn_path => $self->{path} );
2463 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2464 print "No changes\nr$self->{last_rev} = $tree\n";
2465 }
2466 }
2467
2468 sub rebuild {
2469 my ($self) = @_;
2470 my $db_path = $self->db_path;
2471 return if (-e $db_path && ! -z $db_path);
2472 return unless ::verify_ref($self->refname.'^0');
2473 if (-f $self->{db_root}) {
2474 rename $self->{db_root}, $db_path or die
2475 "rename $self->{db_root} => $db_path failed: $!\n";
2476 my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
2477 symlink $base, $self->{db_root} or die
2478 "symlink $base => $self->{db_root} failed: $!\n";
2479 return;
2480 }
2481 print "Rebuilding $db_path ...\n";
2482 my ($log, $ctx) = command_output_pipe("log", '--no-color', $self->refname);
2483 my $latest;
2484 my $full_url = $self->full_url;
2485 remove_username($full_url);
2486 my $svn_uuid;
2487 my $c;
2488 while (<$log>) {
2489 if ( m{^commit ($::sha1)$} ) {
2490 $c = $1;
2491 next;
2492 }
2493 next unless s{^\s*(git-svn-id:)}{$1};
2494 my ($url, $rev, $uuid) = ::extract_metadata($_);
2495 remove_username($url);
2496
2497 # ignore merges (from set-tree)
2498 next if (!defined $rev || !$uuid);
2499
2500 # if we merged or otherwise started elsewhere, this is
2501 # how we break out of it
2502 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
2503 ($full_url && $url && ($url ne $full_url))) {
2504 next;
2505 }
2506 $latest ||= $rev;
2507 $svn_uuid ||= $uuid;
2508
2509 $self->rev_db_set($rev, $c);
2510 print "r$rev = $c\n";
2511 }
2512 command_close_pipe($log, $ctx);
2513 print "Done rebuilding $db_path\n";
2514 }
2515
2516 # rev_db:
2517 # Tie::File seems to be prone to offset errors if revisions get sparse,
2518 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
2519 # one of my favorite modules is out :< Next up would be one of the DBM
2520 # modules, but I'm not sure which is most portable... So I'll just
2521 # go with something that's plain-text, but still capable of
2522 # being randomly accessed. So here's my ultra-simple fixed-width
2523 # database. All records are 40 characters + "\n", so it's easy to seek
2524 # to a revision: (41 * rev) is the byte offset.
2525 # A record of 40 0s denotes an empty revision.
2526 # And yes, it's still pretty fast (faster than Tie::File).
2527 # These files are disposable unless noMetadata or useSvmProps is set
2528
2529 sub _rev_db_set {
2530 my ($fh, $rev, $commit) = @_;
2531 my $offset = $rev * 41;
2532 # assume that append is the common case:
2533 seek $fh, 0, 2 or croak $!;
2534 my $pos = tell $fh;
2535 if ($pos < $offset) {
2536 for (1 .. (($offset - $pos) / 41)) {
2537 print $fh (('0' x 40),"\n") or croak $!;
2538 }
2539 }
2540 seek $fh, $offset, 0 or croak $!;
2541 print $fh $commit,"\n" or croak $!;
2542 }
2543
2544 sub mkfile {
2545 my ($path) = @_;
2546 unless (-e $path) {
2547 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2548 mkpath([$dir]) unless -d $dir;
2549 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2550 close $fh or die "Couldn't close (create) $path: $!\n";
2551 }
2552 }
2553
2554 sub rev_db_set {
2555 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2556 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2557 my $db = $self->db_path($uuid);
2558 my $db_lock = "$db.lock";
2559 my $sig;
2560 if ($update_ref) {
2561 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2562 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2563 }
2564 mkfile($db);
2565
2566 $LOCKFILES{$db_lock} = 1;
2567 my $sync;
2568 # both of these options make our .rev_db file very, very important
2569 # and we can't afford to lose it because rebuild() won't work
2570 if ($self->use_svm_props || $self->no_metadata) {
2571 $sync = 1;
2572 copy($db, $db_lock) or die "rev_db_set(@_): ",
2573 "Failed to copy: ",
2574 "$db => $db_lock ($!)\n";
2575 } else {
2576 rename $db, $db_lock or die "rev_db_set(@_): ",
2577 "Failed to rename: ",
2578 "$db => $db_lock ($!)\n";
2579 }
2580 open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
2581 _rev_db_set($fh, $rev, $commit);
2582 if ($sync) {
2583 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2584 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2585 }
2586 close $fh or croak $!;
2587 if ($update_ref) {
2588 $_head = $self;
2589 command_noisy('update-ref', '-m', "r$rev",
2590 $self->refname, $commit);
2591 }
2592 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
2593 "$db_lock => $db ($!)\n";
2594 delete $LOCKFILES{$db_lock};
2595 if ($update_ref) {
2596 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2597 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2598 kill $sig, $$ if defined $sig;
2599 }
2600 }
2601
2602 sub rev_db_max {
2603 my ($self) = @_;
2604 $self->rebuild;
2605 my $db_path = $self->db_path;
2606 my @stat = stat $db_path or return 0;
2607 ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
2608 my $max = $stat[7] / 41;
2609 (($max > 0) ? $max - 1 : 0);
2610 }
2611
2612 sub rev_db_get {
2613 my ($self, $rev, $uuid) = @_;
2614 my $ret;
2615 my $offset = $rev * 41;
2616 my $db_path = $self->db_path($uuid);
2617 return undef unless -e $db_path;
2618 open my $fh, '<', $db_path or croak $!;
2619 if (sysseek($fh, $offset, 0) == $offset) {
2620 my $read = sysread($fh, $ret, 40);
2621 $ret = undef if ($read != 40 || $ret eq ('0'x40));
2622 }
2623 close $fh or croak $!;
2624 $ret;
2625 }
2626
2627 # Finds the first svn revision that exists on (if $eq_ok is true) or
2628 # before $rev for the current branch. It will not search any lower
2629 # than $min_rev. Returns the git commit hash and svn revision number
2630 # if found, else (undef, undef).
2631 sub find_rev_before {
2632 my ($self, $rev, $eq_ok, $min_rev) = @_;
2633 --$rev unless $eq_ok;
2634 $min_rev ||= 1;
2635 while ($rev >= $min_rev) {
2636 if (my $c = $self->rev_db_get($rev)) {
2637 return ($rev, $c);
2638 }
2639 --$rev;
2640 }
2641 return (undef, undef);
2642 }
2643
2644 # Finds the first svn revision that exists on (if $eq_ok is true) or
2645 # after $rev for the current branch. It will not search any higher
2646 # than $max_rev. Returns the git commit hash and svn revision number
2647 # if found, else (undef, undef).
2648 sub find_rev_after {
2649 my ($self, $rev, $eq_ok, $max_rev) = @_;
2650 ++$rev unless $eq_ok;
2651 $max_rev ||= $self->rev_db_max();
2652 while ($rev <= $max_rev) {
2653 if (my $c = $self->rev_db_get($rev)) {
2654 return ($rev, $c);
2655 }
2656 ++$rev;
2657 }
2658 return (undef, undef);
2659 }
2660
2661 sub _new {
2662 my ($class, $repo_id, $ref_id, $path) = @_;
2663 unless (defined $repo_id && length $repo_id) {
2664 $repo_id = $Git::SVN::default_repo_id;
2665 }
2666 unless (defined $ref_id && length $ref_id) {
2667 $_[2] = $ref_id = $Git::SVN::default_ref_id;
2668 }
2669 $_[1] = $repo_id = sanitize_remote_name($repo_id);
2670 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2671 $_[3] = $path = '' unless (defined $path);
2672 mkpath(["$ENV{GIT_DIR}/svn"]);
2673 bless {
2674 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2675 path => $path, config => "$ENV{GIT_DIR}/svn/config",
2676 db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2677 }
2678
2679 sub db_path {
2680 my ($self, $uuid) = @_;
2681 $uuid ||= $self->ra_uuid;
2682 "$self->{db_root}.$uuid";
2683 }
2684
2685 sub uri_encode {
2686 my ($f) = @_;
2687 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2688 $f
2689 }
2690
2691 sub remove_username {
2692 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2693 }
2694
2695 package Git::SVN::Prompt;
2696 use strict;
2697 use warnings;
2698 require SVN::Core;
2699 use vars qw/$_no_auth_cache $_username/;
2700
2701 sub simple {
2702 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2703 $may_save = undef if $_no_auth_cache;
2704 $default_username = $_username if defined $_username;
2705 if (defined $default_username && length $default_username) {
2706 if (defined $realm && length $realm) {
2707 print STDERR "Authentication realm: $realm\n";
2708 STDERR->flush;
2709 }
2710 $cred->username($default_username);
2711 } else {
2712 username($cred, $realm, $may_save, $pool);
2713 }
2714 $cred->password(_read_password("Password for '" .
2715 $cred->username . "': ", $realm));
2716 $cred->may_save($may_save);
2717 $SVN::_Core::SVN_NO_ERROR;
2718 }
2719
2720 sub ssl_server_trust {
2721 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2722 $may_save = undef if $_no_auth_cache;
2723 print STDERR "Error validating server certificate for '$realm':\n";
2724 {
2725 no warnings 'once';
2726 # All variables SVN::Auth::SSL::* are used only once,
2727 # so we're shutting up Perl warnings about this.
2728 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2729 print STDERR " - The certificate is not issued ",
2730 "by a trusted authority. Use the\n",
2731 " fingerprint to validate ",
2732 "the certificate manually!\n";
2733 }
2734 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2735 print STDERR " - The certificate hostname ",
2736 "does not match.\n";
2737 }
2738 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2739 print STDERR " - The certificate is not yet valid.\n";
2740 }
2741 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2742 print STDERR " - The certificate has expired.\n";
2743 }
2744 if ($failures & $SVN::Auth::SSL::OTHER) {
2745 print STDERR " - The certificate has ",
2746 "an unknown error.\n";
2747 }
2748 } # no warnings 'once'
2749 printf STDERR
2750 "Certificate information:\n".
2751 " - Hostname: %s\n".
2752 " - Valid: from %s until %s\n".
2753 " - Issuer: %s\n".
2754 " - Fingerprint: %s\n",
2755 map $cert_info->$_, qw(hostname valid_from valid_until
2756 issuer_dname fingerprint);
2757 my $choice;
2758 prompt:
2759 print STDERR $may_save ?
2760 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2761 "(R)eject or accept (t)emporarily? ";
2762 STDERR->flush;
2763 $choice = lc(substr(<STDIN> || 'R', 0, 1));
2764 if ($choice =~ /^t$/i) {
2765 $cred->may_save(undef);
2766 } elsif ($choice =~ /^r$/i) {
2767 return -1;
2768 } elsif ($may_save && $choice =~ /^p$/i) {
2769 $cred->may_save($may_save);
2770 } else {
2771 goto prompt;
2772 }
2773 $cred->accepted_failures($failures);
2774 $SVN::_Core::SVN_NO_ERROR;
2775 }
2776
2777 sub ssl_client_cert {
2778 my ($cred, $realm, $may_save, $pool) = @_;
2779 $may_save = undef if $_no_auth_cache;
2780 print STDERR "Client certificate filename: ";
2781 STDERR->flush;
2782 chomp(my $filename = <STDIN>);
2783 $cred->cert_file($filename);
2784 $cred->may_save($may_save);
2785 $SVN::_Core::SVN_NO_ERROR;
2786 }
2787
2788 sub ssl_client_cert_pw {
2789 my ($cred, $realm, $may_save, $pool) = @_;
2790 $may_save = undef if $_no_auth_cache;
2791 $cred->password(_read_password("Password: ", $realm));
2792 $cred->may_save($may_save);
2793 $SVN::_Core::SVN_NO_ERROR;
2794 }
2795
2796 sub username {
2797 my ($cred, $realm, $may_save, $pool) = @_;
2798 $may_save = undef if $_no_auth_cache;
2799 if (defined $realm && length $realm) {
2800 print STDERR "Authentication realm: $realm\n";
2801 }
2802 my $username;
2803 if (defined $_username) {
2804 $username = $_username;
2805 } else {
2806 print STDERR "Username: ";
2807 STDERR->flush;
2808 chomp($username = <STDIN>);
2809 }
2810 $cred->username($username);
2811 $cred->may_save($may_save);
2812 $SVN::_Core::SVN_NO_ERROR;
2813 }
2814
2815 sub _read_password {
2816 my ($prompt, $realm) = @_;
2817 print STDERR $prompt;
2818 STDERR->flush;
2819 require Term::ReadKey;
2820 Term::ReadKey::ReadMode('noecho');
2821 my $password = '';
2822 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2823 last if $key =~ /[\012\015]/; # \n\r
2824 $password .= $key;
2825 }
2826 Term::ReadKey::ReadMode('restore');
2827 print STDERR "\n";
2828 STDERR->flush;
2829 $password;
2830 }
2831
2832 package SVN::Git::Fetcher;
2833 use vars qw/@ISA/;
2834 use strict;
2835 use warnings;
2836 use Carp qw/croak/;
2837 use IO::File qw//;
2838
2839 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
2840 sub new {
2841 my ($class, $git_svn) = @_;
2842 my $self = SVN::Delta::Editor->new;
2843 bless $self, $class;
2844 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2845 $self->{empty} = {};
2846 $self->{dir_prop} = {};
2847 $self->{file_prop} = {};
2848 $self->{absent_dir} = {};
2849 $self->{absent_file} = {};
2850 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2851 $self;
2852 }
2853
2854 sub set_path_strip {
2855 my ($self, $path) = @_;
2856 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2857 }
2858
2859 sub open_root {
2860 { path => '' };
2861 }
2862
2863 sub open_directory {
2864 my ($self, $path, $pb, $rev) = @_;
2865 { path => $path };
2866 }
2867
2868 sub git_path {
2869 my ($self, $path) = @_;
2870 if ($self->{path_strip}) {
2871 $path =~ s!$self->{path_strip}!! or
2872 die "Failed to strip path '$path' ($self->{path_strip})\n";
2873 }
2874 $path;
2875 }
2876
2877 sub delete_entry {
2878 my ($self, $path, $rev, $pb) = @_;
2879
2880 my $gpath = $self->git_path($path);
2881 return undef if ($gpath eq '');
2882
2883 # remove entire directories.
2884 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2885 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2886 -r --name-only -z/,
2887 $self->{c}, '--', $gpath);
2888 local $/ = "\0";
2889 while (<$ls>) {
2890 chomp;
2891 $self->{gii}->remove($_);
2892 print "\tD\t$_\n" unless $::_q;
2893 }
2894 print "\tD\t$gpath/\n" unless $::_q;
2895 command_close_pipe($ls, $ctx);
2896 $self->{empty}->{$path} = 0
2897 } else {
2898 $self->{gii}->remove($gpath);
2899 print "\tD\t$gpath\n" unless $::_q;
2900 }
2901 undef;
2902 }
2903
2904 sub open_file {
2905 my ($self, $path, $pb, $rev) = @_;
2906 my $gpath = $self->git_path($path);
2907 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2908 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2909 unless (defined $mode && defined $blob) {
2910 die "$path was not found in commit $self->{c} (r$rev)\n";
2911 }
2912 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2913 pool => SVN::Pool->new, action => 'M' };
2914 }
2915
2916 sub add_file {
2917 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2918 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2919 delete $self->{empty}->{$dir};
2920 { path => $path, mode_a => 100644, mode_b => 100644,
2921 pool => SVN::Pool->new, action => 'A' };
2922 }
2923
2924 sub add_directory {
2925 my ($self, $path, $cp_path, $cp_rev) = @_;
2926 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2927 delete $self->{empty}->{$dir};
2928 $self->{empty}->{$path} = 1;
2929 { path => $path };
2930 }
2931
2932 sub change_dir_prop {
2933 my ($self, $db, $prop, $value) = @_;
2934 $self->{dir_prop}->{$db->{path}} ||= {};
2935 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2936 undef;
2937 }
2938
2939 sub absent_directory {
2940 my ($self, $path, $pb) = @_;
2941 $self->{absent_dir}->{$pb->{path}} ||= [];
2942 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2943 undef;
2944 }
2945
2946 sub absent_file {
2947 my ($self, $path, $pb) = @_;
2948 $self->{absent_file}->{$pb->{path}} ||= [];
2949 push @{$self->{absent_file}->{$pb->{path}}}, $path;
2950 undef;
2951 }
2952
2953 sub change_file_prop {
2954 my ($self, $fb, $prop, $value) = @_;
2955 if ($prop eq 'svn:executable') {
2956 if ($fb->{mode_b} != 120000) {
2957 $fb->{mode_b} = defined $value ? 100755 : 100644;
2958 }
2959 } elsif ($prop eq 'svn:special') {
2960 $fb->{mode_b} = defined $value ? 120000 : 100644;
2961 } else {
2962 $self->{file_prop}->{$fb->{path}} ||= {};
2963 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2964 }
2965 undef;
2966 }
2967
2968 sub apply_textdelta {
2969 my ($self, $fb, $exp) = @_;
2970 my $fh = IO::File->new_tmpfile;
2971 $fh->autoflush(1);
2972 # $fh gets auto-closed() by SVN::TxDelta::apply(),
2973 # (but $base does not,) so dup() it for reading in close_file
2974 open my $dup, '<&', $fh or croak $!;
2975 my $base = IO::File->new_tmpfile;
2976 $base->autoflush(1);
2977 if ($fb->{blob}) {
2978 defined (my $pid = fork) or croak $!;
2979 if (!$pid) {
2980 open STDOUT, '>&', $base or croak $!;
2981 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2982 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2983 }
2984 waitpid $pid, 0;
2985 croak $? if $?;
2986
2987 if (defined $exp) {
2988 seek $base, 0, 0 or croak $!;
2989 my $got = ::md5sum($base);
2990 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2991 "expected: $exp\n",
2992 " got: $got\n" if ($got ne $exp);
2993 }
2994 }
2995 seek $base, 0, 0 or croak $!;
2996 $fb->{fh} = $dup;
2997 $fb->{base} = $base;
2998 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2999 }
3000
3001 sub close_file {
3002 my ($self, $fb, $exp) = @_;
3003 my $hash;
3004 my $path = $self->git_path($fb->{path});
3005 if (my $fh = $fb->{fh}) {
3006 if (defined $exp) {
3007 seek($fh, 0, 0) or croak $!;
3008 my $got = ::md5sum($fh);
3009 if ($got ne $exp) {
3010 die "Checksum mismatch: $path\n",
3011 "expected: $exp\n got: $got\n";
3012 }
3013 }
3014 sysseek($fh, 0, 0) or croak $!;
3015 if ($fb->{mode_b} == 120000) {
3016 sysread($fh, my $buf, 5) == 5 or croak $!;
3017 $buf eq 'link ' or die "$path has mode 120000",
3018 "but is not a link\n";
3019 }
3020 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
3021 if (!$pid) {
3022 open STDIN, '<&', $fh or croak $!;
3023 exec qw/git-hash-object -w --stdin/ or croak $!;
3024 }
3025 chomp($hash = do { local $/; <$out> });
3026 close $out or croak $!;
3027 close $fh or croak $!;
3028 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3029 close $fb->{base} or croak $!;
3030 } else {
3031 $hash = $fb->{blob} or die "no blob information\n";
3032 }
3033 $fb->{pool}->clear;
3034 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3035 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3036 undef;
3037 }
3038
3039 sub abort_edit {
3040 my $self = shift;
3041 $self->{nr} = $self->{gii}->{nr};
3042 delete $self->{gii};
3043 $self->SUPER::abort_edit(@_);
3044 }
3045
3046 sub close_edit {
3047 my $self = shift;
3048 $self->{git_commit_ok} = 1;
3049 $self->{nr} = $self->{gii}->{nr};
3050 delete $self->{gii};
3051 $self->SUPER::close_edit(@_);
3052 }
3053
3054 package SVN::Git::Editor;
3055 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3056 use strict;
3057 use warnings;
3058 use Carp qw/croak/;
3059 use IO::File;
3060
3061 sub new {
3062 my ($class, $opts) = @_;
3063 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3064 die "$_ required!\n" unless (defined $opts->{$_});
3065 }
3066
3067 my $pool = SVN::Pool->new;
3068 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3069 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3070 $opts->{r}, $mods);
3071
3072 # $opts->{ra} functions should not be used after this:
3073 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3074 $opts->{editor_cb}, $pool);
3075 my $self = SVN::Delta::Editor->new(@ce, $pool);
3076 bless $self, $class;
3077 foreach (qw/svn_path r tree_a tree_b/) {
3078 $self->{$_} = $opts->{$_};
3079 }
3080 $self->{url} = $opts->{ra}->{url};
3081 $self->{mods} = $mods;
3082 $self->{types} = $types;
3083 $self->{pool} = $pool;
3084 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3085 $self->{rm} = { };
3086 $self->{path_prefix} = length $self->{svn_path} ?
3087 "$self->{svn_path}/" : '';
3088 return $self;
3089 }
3090
3091 sub generate_diff {
3092 my ($tree_a, $tree_b) = @_;
3093 my @diff_tree = qw(diff-tree -z -r);
3094 if ($_cp_similarity) {
3095 push @diff_tree, "-C$_cp_similarity";
3096 } else {
3097 push @diff_tree, '-C';
3098 }
3099 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3100 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3101 push @diff_tree, $tree_a, $tree_b;
3102 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3103 local $/ = "\0";
3104 my $state = 'meta';
3105 my @mods;
3106 while (<$diff_fh>) {
3107 chomp $_; # this gets rid of the trailing "\0"
3108 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3109 $::sha1\s($::sha1)\s
3110 ([MTCRAD])\d*$/xo) {
3111 push @mods, { mode_a => $1, mode_b => $2,
3112 sha1_b => $3, chg => $4 };
3113 if ($4 =~ /^(?:C|R)$/) {
3114 $state = 'file_a';
3115 } else {
3116 $state = 'file_b';
3117 }
3118 } elsif ($state eq 'file_a') {
3119 my $x = $mods[$#mods] or croak "Empty array\n";
3120 if ($x->{chg} !~ /^(?:C|R)$/) {
3121 croak "Error parsing $_, $x->{chg}\n";
3122 }
3123 $x->{file_a} = $_;
3124 $state = 'file_b';
3125 } elsif ($state eq 'file_b') {
3126 my $x = $mods[$#mods] or croak "Empty array\n";
3127 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3128 croak "Error parsing $_, $x->{chg}\n";
3129 }
3130 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3131 croak "Error parsing $_, $x->{chg}\n";
3132 }
3133 $x->{file_b} = $_;
3134 $state = 'meta';
3135 } else {
3136 croak "Error parsing $_\n";
3137 }
3138 }
3139 command_close_pipe($diff_fh, $ctx);
3140 \@mods;
3141 }
3142
3143 sub check_diff_paths {
3144 my ($ra, $pfx, $rev, $mods) = @_;
3145 my %types;
3146 $pfx .= '/' if length $pfx;
3147
3148 sub type_diff_paths {
3149 my ($ra, $types, $path, $rev) = @_;
3150 my @p = split m#/+#, $path;
3151 my $c = shift @p;
3152 unless (defined $types->{$c}) {
3153 $types->{$c} = $ra->check_path($c, $rev);
3154 }
3155 while (@p) {
3156 $c .= '/' . shift @p;
3157 next if defined $types->{$c};
3158 $types->{$c} = $ra->check_path($c, $rev);
3159 }
3160 }
3161
3162 foreach my $m (@$mods) {
3163 foreach my $f (qw/file_a file_b/) {
3164 next unless defined $m->{$f};
3165 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3166 if (length $pfx.$dir && ! defined $types{$dir}) {
3167 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3168 }
3169 }
3170 }
3171 \%types;
3172 }
3173
3174 sub split_path {
3175 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3176 }
3177
3178 sub repo_path {
3179 my ($self, $path) = @_;
3180 $self->{path_prefix}.(defined $path ? $path : '');
3181 }
3182
3183 sub url_path {
3184 my ($self, $path) = @_;
3185 if ($self->{url} =~ m#^https?://#) {
3186 $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3187 }
3188 $self->{url} . '/' . $self->repo_path($path);
3189 }
3190
3191 sub rmdirs {
3192 my ($self) = @_;
3193 my $rm = $self->{rm};
3194 delete $rm->{''}; # we never delete the url we're tracking
3195 return unless %$rm;
3196
3197 foreach (keys %$rm) {
3198 my @d = split m#/#, $_;
3199 my $c = shift @d;
3200 $rm->{$c} = 1;
3201 while (@d) {
3202 $c .= '/' . shift @d;
3203 $rm->{$c} = 1;
3204 }
3205 }
3206 delete $rm->{$self->{svn_path}};
3207 delete $rm->{''}; # we never delete the url we're tracking
3208 return unless %$rm;
3209
3210 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3211 $self->{tree_b});
3212 local $/ = "\0";
3213 while (<$fh>) {
3214 chomp;
3215 my @dn = split m#/#, $_;
3216 while (pop @dn) {
3217 delete $rm->{join '/', @dn};
3218 }
3219 unless (%$rm) {
3220 close $fh;
3221 return;
3222 }
3223 }
3224 command_close_pipe($fh, $ctx);
3225
3226 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3227 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3228 $self->close_directory($bat->{$d}, $p);
3229 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3230 print "\tD+\t$d/\n" unless $::_q;
3231 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3232 delete $bat->{$d};
3233 }
3234 }
3235
3236 sub open_or_add_dir {
3237 my ($self, $full_path, $baton) = @_;
3238 my $t = $self->{types}->{$full_path};
3239 if (!defined $t) {
3240 die "$full_path not known in r$self->{r} or we have a bug!\n";
3241 }
3242 {
3243 no warnings 'once';
3244 # SVN::Node::none and SVN::Node::file are used only once,
3245 # so we're shutting up Perl's warnings about them.
3246 if ($t == $SVN::Node::none) {
3247 return $self->add_directory($full_path, $baton,
3248 undef, -1, $self->{pool});
3249 } elsif ($t == $SVN::Node::dir) {
3250 return $self->open_directory($full_path, $baton,
3251 $self->{r}, $self->{pool});
3252 } # no warnings 'once'
3253 print STDERR "$full_path already exists in repository at ",
3254 "r$self->{r} and it is not a directory (",
3255 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3256 } # no warnings 'once'
3257 exit 1;
3258 }
3259
3260 sub ensure_path {
3261 my ($self, $path) = @_;
3262 my $bat = $self->{bat};
3263 my $repo_path = $self->repo_path($path);
3264 return $bat->{''} unless (length $repo_path);
3265 my @p = split m#/+#, $repo_path;
3266 my $c = shift @p;
3267 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3268 while (@p) {
3269 my $c0 = $c;
3270 $c .= '/' . shift @p;
3271 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3272 }
3273 return $bat->{$c};
3274 }
3275
3276 sub A {
3277 my ($self, $m) = @_;
3278 my ($dir, $file) = split_path($m->{file_b});
3279 my $pbat = $self->ensure_path($dir);
3280 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3281 undef, -1);
3282 print "\tA\t$m->{file_b}\n" unless $::_q;
3283 $self->chg_file($fbat, $m);
3284 $self->close_file($fbat,undef,$self->{pool});
3285 }
3286
3287 sub C {
3288 my ($self, $m) = @_;
3289 my ($dir, $file) = split_path($m->{file_b});
3290 my $pbat = $self->ensure_path($dir);
3291 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3292 $self->url_path($m->{file_a}), $self->{r});
3293 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3294 $self->chg_file($fbat, $m);
3295 $self->close_file($fbat,undef,$self->{pool});
3296 }
3297
3298 sub delete_entry {
3299 my ($self, $path, $pbat) = @_;
3300 my $rpath = $self->repo_path($path);
3301 my ($dir, $file) = split_path($rpath);
3302 $self->{rm}->{$dir} = 1;
3303 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3304 }
3305
3306 sub R {
3307 my ($self, $m) = @_;
3308 my ($dir, $file) = split_path($m->{file_b});
3309 my $pbat = $self->ensure_path($dir);
3310 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3311 $self->url_path($m->{file_a}), $self->{r});
3312 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3313 $self->chg_file($fbat, $m);
3314 $self->close_file($fbat,undef,$self->{pool});
3315
3316 ($dir, $file) = split_path($m->{file_a});
3317 $pbat = $self->ensure_path($dir);
3318 $self->delete_entry($m->{file_a}, $pbat);
3319 }
3320
3321 sub M {
3322 my ($self, $m) = @_;
3323 my ($dir, $file) = split_path($m->{file_b});
3324 my $pbat = $self->ensure_path($dir);
3325 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3326 $pbat,$self->{r},$self->{pool});
3327 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3328 $self->chg_file($fbat, $m);
3329 $self->close_file($fbat,undef,$self->{pool});
3330 }
3331
3332 sub T { shift->M(@_) }
3333
3334 sub change_file_prop {
3335 my ($self, $fbat, $pname, $pval) = @_;
3336 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3337 }
3338
3339 sub chg_file {
3340 my ($self, $fbat, $m) = @_;
3341 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3342 $self->change_file_prop($fbat,'svn:executable','*');
3343 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3344 $self->change_file_prop($fbat,'svn:executable',undef);
3345 }
3346 my $fh = IO::File->new_tmpfile or croak $!;
3347 if ($m->{mode_b} =~ /^120/) {
3348 print $fh 'link ' or croak $!;
3349 $self->change_file_prop($fbat,'svn:special','*');
3350 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3351 $self->change_file_prop($fbat,'svn:special',undef);
3352 }
3353 defined(my $pid = fork) or croak $!;
3354 if (!$pid) {
3355 open STDOUT, '>&', $fh or croak $!;
3356 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3357 }
3358 waitpid $pid, 0;
3359 croak $? if $?;
3360 $fh->flush == 0 or croak $!;
3361 seek $fh, 0, 0 or croak $!;
3362
3363 my $exp = ::md5sum($fh);
3364 seek $fh, 0, 0 or croak $!;
3365
3366 my $pool = SVN::Pool->new;
3367 my $atd = $self->apply_textdelta($fbat, undef, $pool);
3368 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3369 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3370 $pool->clear;
3371
3372 close $fh or croak $!;
3373 }
3374
3375 sub D {
3376 my ($self, $m) = @_;
3377 my ($dir, $file) = split_path($m->{file_b});
3378 my $pbat = $self->ensure_path($dir);
3379 print "\tD\t$m->{file_b}\n" unless $::_q;
3380 $self->delete_entry($m->{file_b}, $pbat);
3381 }
3382
3383 sub close_edit {
3384 my ($self) = @_;
3385 my ($p,$bat) = ($self->{pool}, $self->{bat});
3386 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3387 next if $_ eq '';
3388 $self->close_directory($bat->{$_}, $p);
3389 }
3390 $self->close_directory($bat->{''}, $p);
3391 $self->SUPER::close_edit($p);
3392 $p->clear;
3393 }
3394
3395 sub abort_edit {
3396 my ($self) = @_;
3397 $self->SUPER::abort_edit($self->{pool});
3398 }
3399
3400 sub DESTROY {
3401 my $self = shift;
3402 $self->SUPER::DESTROY(@_);
3403 $self->{pool}->clear;
3404 }
3405
3406 # this drives the editor
3407 sub apply_diff {
3408 my ($self) = @_;
3409 my $mods = $self->{mods};
3410 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3411 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3412 my $f = $m->{chg};
3413 if (defined $o{$f}) {
3414 $self->$f($m);
3415 } else {
3416 fatal("Invalid change type: $f");
3417 }
3418 }
3419 $self->rmdirs if $_rmdir;
3420 if (@$mods == 0) {
3421 $self->abort_edit;
3422 } else {
3423 $self->close_edit;
3424 }
3425 return scalar @$mods;
3426 }
3427
3428 package Git::SVN::Ra;
3429 use vars qw/@ISA $config_dir $_log_window_size/;
3430 use strict;
3431 use warnings;
3432 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3433
3434 BEGIN {
3435 # enforce temporary pool usage for some simple functions
3436 no strict 'refs';
3437 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3438 my $SUPER = "SUPER::$f";
3439 *$f = sub {
3440 my $self = shift;
3441 my $pool = SVN::Pool->new;
3442 my @ret = $self->$SUPER(@_,$pool);
3443 $pool->clear;
3444 wantarray ? @ret : $ret[0];
3445 };
3446 }
3447 }
3448
3449 sub _auth_providers () {
3450 [
3451 SVN::Client::get_simple_provider(),
3452 SVN::Client::get_ssl_server_trust_file_provider(),
3453 SVN::Client::get_simple_prompt_provider(
3454 \&Git::SVN::Prompt::simple, 2),
3455 SVN::Client::get_ssl_client_cert_file_provider(),
3456 SVN::Client::get_ssl_client_cert_prompt_provider(
3457 \&Git::SVN::Prompt::ssl_client_cert, 2),
3458 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3459 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3460 SVN::Client::get_username_provider(),
3461 SVN::Client::get_ssl_server_trust_prompt_provider(
3462 \&Git::SVN::Prompt::ssl_server_trust),
3463 SVN::Client::get_username_prompt_provider(
3464 \&Git::SVN::Prompt::username, 2)
3465 ]
3466 }
3467
3468 sub escape_uri_only {
3469 my ($uri) = @_;
3470 my @tmp;
3471 foreach (split m{/}, $uri) {
3472 s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
3473 push @tmp, $_;
3474 }
3475 join('/', @tmp);
3476 }
3477
3478 sub escape_url {
3479 my ($url) = @_;
3480 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3481 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3482 $url = "$scheme://$domain$uri";
3483 }
3484 $url;
3485 }
3486
3487 sub new {
3488 my ($class, $url) = @_;
3489 $url =~ s!/+$!!;
3490 return $RA if ($RA && $RA->{url} eq $url);
3491
3492 SVN::_Core::svn_config_ensure($config_dir, undef);
3493 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3494 my $config = SVN::Core::config_get_config($config_dir);
3495 $RA = undef;
3496 my $dont_store_passwords = 1;
3497 my $conf_t = ${$config}{'config'};
3498 {
3499 no warnings 'once';
3500 # The usage of $SVN::_Core::SVN_CONFIG_* variables
3501 # produces warnings that variables are used only once.
3502 # I had not found the better way to shut them up, so
3503 # the warnings of type 'once' are disabled in this block.
3504 if (SVN::_Core::svn_config_get_bool($conf_t,
3505 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3506 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3507 1) == 0) {
3508 SVN::_Core::svn_auth_set_parameter($baton,
3509 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3510 bless (\$dont_store_passwords, "_p_void"));
3511 }
3512 if (SVN::_Core::svn_config_get_bool($conf_t,
3513 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3514 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3515 1) == 0) {
3516 $Git::SVN::Prompt::_no_auth_cache = 1;
3517 }
3518 } # no warnings 'once'
3519 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3520 config => $config,
3521 pool => SVN::Pool->new,
3522 auth_provider_callbacks => $callbacks);
3523 $self->{url} = $url;
3524 $self->{svn_path} = $url;
3525 $self->{repos_root} = $self->get_repos_root;
3526 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3527 $self->{cache} = { check_path => { r => 0, data => {} },
3528 get_dir => { r => 0, data => {} } };
3529 $RA = bless $self, $class;
3530 }
3531
3532 sub check_path {
3533 my ($self, $path, $r) = @_;
3534 my $cache = $self->{cache}->{check_path};
3535 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3536 return $cache->{data}->{$path};
3537 }
3538 my $pool = SVN::Pool->new;
3539 my $t = $self->SUPER::check_path($path, $r, $pool);
3540 $pool->clear;
3541 if ($r != $cache->{r}) {
3542 %{$cache->{data}} = ();
3543 $cache->{r} = $r;
3544 }
3545 $cache->{data}->{$path} = $t;
3546 }
3547
3548 sub get_dir {
3549 my ($self, $dir, $r) = @_;
3550 my $cache = $self->{cache}->{get_dir};
3551 if ($r == $cache->{r}) {
3552 if (my $x = $cache->{data}->{$dir}) {
3553 return wantarray ? @$x : $x->[0];
3554 }
3555 }
3556 my $pool = SVN::Pool->new;
3557 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3558 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3559 $pool->clear;
3560 if ($r != $cache->{r}) {
3561 %{$cache->{data}} = ();
3562 $cache->{r} = $r;
3563 }
3564 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3565 wantarray ? (\%dirents, $r, $props) : \%dirents;
3566 }
3567
3568 sub DESTROY {
3569 # do not call the real DESTROY since we store ourselves in $RA
3570 }
3571
3572 sub get_log {
3573 my ($self, @args) = @_;
3574 my $pool = SVN::Pool->new;
3575 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3576 my $ret = $self->SUPER::get_log(@args, $pool);
3577 $pool->clear;
3578 $ret;
3579 }
3580
3581 sub trees_match {
3582 my ($self, $url1, $rev1, $url2, $rev2) = @_;
3583 my $ctx = SVN::Client->new(auth => _auth_providers);
3584 my $out = IO::File->new_tmpfile;
3585
3586 # older SVN (1.1.x) doesn't take $pool as the last parameter for
3587 # $ctx->diff(), so we'll create a default one
3588 my $pool = SVN::Pool->new_default_sub;
3589
3590 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3591 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3592 $out->flush;
3593 my $ret = (($out->stat)[7] == 0);
3594 close $out or croak $!;
3595
3596 $ret;
3597 }
3598
3599 sub get_commit_editor {
3600 my ($self, $log, $cb, $pool) = @_;
3601 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3602 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3603 }
3604
3605 sub gs_do_update {
3606 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3607 my $new = ($rev_a == $rev_b);
3608 my $path = $gs->{path};
3609
3610 if ($new && -e $gs->{index}) {
3611 unlink $gs->{index} or die
3612 "Couldn't unlink index: $gs->{index}: $!\n";
3613 }
3614 my $pool = SVN::Pool->new;
3615 $editor->set_path_strip($path);
3616 my (@pc) = split m#/#, $path;
3617 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3618 1, $editor, $pool);
3619 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3620
3621 # Since we can't rely on svn_ra_reparent being available, we'll
3622 # just have to do some magic with set_path to make it so
3623 # we only want a partial path.
3624 my $sp = '';
3625 my $final = join('/', @pc);
3626 while (@pc) {
3627 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3628 $sp .= '/' if length $sp;
3629 $sp .= shift @pc;
3630 }
3631 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3632
3633 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3634
3635 $reporter->finish_report($pool);
3636 $pool->clear;
3637 $editor->{git_commit_ok};
3638 }
3639
3640 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3641 # svn_ra_reparent didn't work before 1.4)
3642 sub gs_do_switch {
3643 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3644 my $path = $gs->{path};
3645 my $pool = SVN::Pool->new;
3646
3647 my $full_url = $self->{url};
3648 my $old_url = $full_url;
3649 $full_url .= '/' . escape_uri_only($path) if length $path;
3650 my ($ra, $reparented);
3651 if ($old_url ne $full_url) {
3652 if ($old_url !~ m#^svn(\+ssh)?://#) {
3653 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3654 $pool);
3655 $self->{url} = $full_url;
3656 $reparented = 1;
3657 } else {
3658 $_[0] = undef;
3659 $self = undef;
3660 $RA = undef;
3661 $ra = Git::SVN::Ra->new($full_url);
3662 $ra_invalid = 1;
3663 }
3664 }
3665 $ra ||= $self;
3666 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3667 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3668 $reporter->set_path('', $rev_a, 0, @lock, $pool);
3669 $reporter->finish_report($pool);
3670
3671 if ($reparented) {
3672 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3673 $self->{url} = $old_url;
3674 }
3675
3676 $pool->clear;
3677 $editor->{git_commit_ok};
3678 }
3679
3680 sub longest_common_path {
3681 my ($gsv, $globs) = @_;
3682 my %common;
3683 my $common_max = scalar @$gsv;
3684
3685 foreach my $gs (@$gsv) {
3686 my @tmp = split m#/#, $gs->{path};
3687 my $p = '';
3688 foreach (@tmp) {
3689 $p .= length($p) ? "/$_" : $_;
3690 $common{$p} ||= 0;
3691 $common{$p}++;
3692 }
3693 }
3694 $globs ||= [];
3695 $common_max += scalar @$globs;
3696 foreach my $glob (@$globs) {
3697 my @tmp = split m#/#, $glob->{path}->{left};
3698 my $p = '';
3699 foreach (@tmp) {
3700 $p .= length($p) ? "/$_" : $_;
3701 $common{$p} ||= 0;
3702 $common{$p}++;
3703 }
3704 }
3705
3706 my $longest_path = '';
3707 foreach (sort {length $b <=> length $a} keys %common) {
3708 if ($common{$_} == $common_max) {
3709 $longest_path = $_;
3710 last;
3711 }
3712 }
3713 $longest_path;
3714 }
3715
3716 sub gs_fetch_loop_common {
3717 my ($self, $base, $head, $gsv, $globs) = @_;
3718 return if ($base > $head);
3719 my $inc = $_log_window_size;
3720 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3721 my $longest_path = longest_common_path($gsv, $globs);
3722 my $ra_url = $self->{url};
3723 while (1) {
3724 my %revs;
3725 my $err;
3726 my $err_handler = $SVN::Error::handler;
3727 $SVN::Error::handler = sub {
3728 ($err) = @_;
3729 skip_unknown_revs($err);
3730 };
3731 sub _cb {
3732 my ($paths, $r, $author, $date, $log) = @_;
3733 [ dup_changed_paths($paths),
3734 { author => $author, date => $date, log => $log } ];
3735 }
3736 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3737 sub { $revs{$_[1]} = _cb(@_) });
3738 if ($err && $max >= $head) {
3739 print STDERR "Path '$longest_path' ",
3740 "was probably deleted:\n",
3741 $err->expanded_message,
3742 "\nWill attempt to follow ",
3743 "revisions r$min .. r$max ",
3744 "committed before the deletion\n";
3745 my $hi = $max;
3746 while (--$hi >= $min) {
3747 my $ok;
3748 $self->get_log([$longest_path], $min, $hi,
3749 0, 1, 1, sub {
3750 $ok ||= $_[1];
3751 $revs{$_[1]} = _cb(@_) });
3752 if ($ok) {
3753 print STDERR "r$min .. r$ok OK\n";
3754 last;
3755 }
3756 }
3757 }
3758 $SVN::Error::handler = $err_handler;
3759
3760 my %exists = map { $_->{path} => $_ } @$gsv;
3761 foreach my $r (sort {$a <=> $b} keys %revs) {
3762 my ($paths, $logged) = @{$revs{$r}};
3763
3764 foreach my $gs ($self->match_globs(\%exists, $paths,
3765 $globs, $r)) {
3766 if ($gs->rev_db_max >= $r) {
3767 next;
3768 }
3769 next unless $gs->match_paths($paths, $r);
3770 $gs->{logged_rev_props} = $logged;
3771 if (my $last_commit = $gs->last_commit) {
3772 $gs->assert_index_clean($last_commit);
3773 }
3774 my $log_entry = $gs->do_fetch($paths, $r);
3775 if ($log_entry) {
3776 $gs->do_git_commit($log_entry);
3777 }
3778 }
3779 foreach my $g (@$globs) {
3780 my $k = "svn-remote.$g->{remote}." .
3781 "$g->{t}-maxRev";
3782 Git::SVN::tmp_config($k, $r);
3783 }
3784 if ($ra_invalid) {
3785 $_[0] = undef;
3786 $self = undef;
3787 $RA = undef;
3788 $self = Git::SVN::Ra->new($ra_url);
3789 $ra_invalid = undef;
3790 }
3791 }
3792 # pre-fill the .rev_db since it'll eventually get filled in
3793 # with '0' x40 if something new gets committed
3794 foreach my $gs (@$gsv) {
3795 next if defined $gs->rev_db_get($max);
3796 $gs->rev_db_set($max, 0 x40);
3797 }
3798 foreach my $g (@$globs) {
3799 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3800 Git::SVN::tmp_config($k, $max);
3801 }
3802 last if $max >= $head;
3803 $min = $max + 1;
3804 $max += $inc;
3805 $max = $head if ($max > $head);
3806 }
3807 }
3808
3809 sub match_globs {
3810 my ($self, $exists, $paths, $globs, $r) = @_;
3811
3812 sub get_dir_check {
3813 my ($self, $exists, $g, $r) = @_;
3814 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3815 return unless scalar @x == 3;
3816 my $dirents = $x[0];
3817 foreach my $de (keys %$dirents) {
3818 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
3819 my $p = $g->{path}->full_path($de);
3820 next if $exists->{$p};
3821 next if (length $g->{path}->{right} &&
3822 ($self->check_path($p, $r) !=
3823 $SVN::Node::dir));
3824 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3825 $g->{ref}->full_path($de), 1);
3826 }
3827 }
3828 foreach my $g (@$globs) {
3829 if (my $path = $paths->{"/$g->{path}->{left}"}) {
3830 if ($path->{action} =~ /^[AR]$/) {
3831 get_dir_check($self, $exists, $g, $r);
3832 }
3833 }
3834 foreach (keys %$paths) {
3835 if (/$g->{path}->{left_regex}/ &&
3836 !/$g->{path}->{regex}/) {
3837 next if $paths->{$_}->{action} !~ /^[AR]$/;
3838 get_dir_check($self, $exists, $g, $r);
3839 }
3840 next unless /$g->{path}->{regex}/;
3841 my $p = $1;
3842 my $pathname = $g->{path}->full_path($p);
3843 next if $exists->{$pathname};
3844 next if ($self->check_path($pathname, $r) !=
3845 $SVN::Node::dir);
3846 $exists->{$pathname} = Git::SVN->init(
3847 $self->{url}, $pathname, undef,
3848 $g->{ref}->full_path($p), 1);
3849 }
3850 my $c = '';
3851 foreach (split m#/#, $g->{path}->{left}) {
3852 $c .= "/$_";
3853 next unless ($paths->{$c} &&
3854 ($paths->{$c}->{action} =~ /^[AR]$/));
3855 get_dir_check($self, $exists, $g, $r);
3856 }
3857 }
3858 values %$exists;
3859 }
3860
3861 sub minimize_url {
3862 my ($self) = @_;
3863 return $self->{url} if ($self->{url} eq $self->{repos_root});
3864 my $url = $self->{repos_root};
3865 my @components = split(m!/!, $self->{svn_path});
3866 my $c = '';
3867 do {
3868 $url .= "/$c" if length $c;
3869 eval { (ref $self)->new($url)->get_latest_revnum };
3870 } while ($@ && ($c = shift @components));
3871 $url;
3872 }
3873
3874 sub can_do_switch {
3875 my $self = shift;
3876 unless (defined $can_do_switch) {
3877 my $pool = SVN::Pool->new;
3878 my $rep = eval {
3879 $self->do_switch(1, '', 0, $self->{url},
3880 SVN::Delta::Editor->new, $pool);
3881 };
3882 if ($@) {
3883 $can_do_switch = 0;
3884 } else {
3885 $rep->abort_report($pool);
3886 $can_do_switch = 1;
3887 }
3888 $pool->clear;
3889 }
3890 $can_do_switch;
3891 }
3892
3893 sub skip_unknown_revs {
3894 my ($err) = @_;
3895 my $errno = $err->apr_err();
3896 # Maybe the branch we're tracking didn't
3897 # exist when the repo started, so it's
3898 # not an error if it doesn't, just continue
3899 #
3900 # Wonderfully consistent library, eh?
3901 # 160013 - svn:// and file://
3902 # 175002 - http(s)://
3903 # 175007 - http(s):// (this repo required authorization, too...)
3904 # More codes may be discovered later...
3905 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3906 my $err_key = $err->expanded_message;
3907 # revision numbers change every time, filter them out
3908 $err_key =~ s/\d+/\0/g;
3909 $err_key = "$errno\0$err_key";
3910 unless ($ignored_err{$err_key}) {
3911 warn "W: Ignoring error from SVN, path probably ",
3912 "does not exist: ($errno): ",
3913 $err->expanded_message,"\n";
3914 $ignored_err{$err_key} = 1;
3915 }
3916 return;
3917 }
3918 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3919 }
3920
3921 # svn_log_changed_path_t objects passed to get_log are likely to be
3922 # overwritten even if only the refs are copied to an external variable,
3923 # so we should dup the structures in their entirety. Using an externally
3924 # passed pool (instead of our temporary and quickly cleared pool in
3925 # Git::SVN::Ra) does not help matters at all...
3926 sub dup_changed_paths {
3927 my ($paths) = @_;
3928 return undef unless $paths;
3929 my %ret;
3930 foreach my $p (keys %$paths) {
3931 my $i = $paths->{$p};
3932 my %s = map { $_ => $i->$_ }
3933 qw/copyfrom_path copyfrom_rev action/;
3934 $ret{$p} = \%s;
3935 }
3936 \%ret;
3937 }
3938
3939 package Git::SVN::Log;
3940 use strict;
3941 use warnings;
3942 use POSIX qw/strftime/;
3943 use constant commit_log_separator => ('-' x 72) . "\n";
3944 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3945 %rusers $show_commit $incremental/;
3946 my $l_fmt;
3947
3948 sub cmt_showable {
3949 my ($c) = @_;
3950 return 1 if defined $c->{r};
3951
3952 # big commit message got truncated by the 16k pretty buffer in rev-list
3953 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3954 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3955 @{$c->{l}} = ();
3956 my @log = command(qw/cat-file commit/, $c->{c});
3957
3958 # shift off the headers
3959 shift @log while ($log[0] ne '');
3960 shift @log;
3961
3962 # TODO: make $c->{l} not have a trailing newline in the future
3963 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
3964
3965 (undef, $c->{r}, undef) = ::extract_metadata(
3966 (grep(/^git-svn-id: /, @log))[-1]);
3967 }
3968 return defined $c->{r};
3969 }
3970
3971 sub log_use_color {
3972 return 1 if $color;
3973 my ($dc, $dcvar);
3974 $dcvar = 'color.diff';
3975 $dc = `git-config --get $dcvar`;
3976 if ($dc eq '') {
3977 # nothing at all; fallback to "diff.color"
3978 $dcvar = 'diff.color';
3979 $dc = `git-config --get $dcvar`;
3980 }
3981 chomp($dc);
3982 if ($dc eq 'auto') {
3983 my $pc;
3984 $pc = `git-config --get color.pager`;
3985 if ($pc eq '') {
3986 # does not have it -- fallback to pager.color
3987 $pc = `git-config --bool --get pager.color`;
3988 }
3989 else {
3990 $pc = `git-config --bool --get color.pager`;
3991 if ($?) {
3992 $pc = 'false';
3993 }
3994 }
3995 chomp($pc);
3996 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3997 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3998 }
3999 return 0;
4000 }
4001 return 0 if $dc eq 'never';
4002 return 1 if $dc eq 'always';
4003 chomp($dc = `git-config --bool --get $dcvar`);
4004 return ($dc eq 'true');
4005 }
4006
4007 sub git_svn_log_cmd {
4008 my ($r_min, $r_max, @args) = @_;
4009 my $head = 'HEAD';
4010 my (@files, @log_opts);
4011 foreach my $x (@args) {
4012 if ($x eq '--' || @files) {
4013 push @files, $x;
4014 } else {
4015 if (::verify_ref("$x^0")) {
4016 $head = $x;
4017 } else {
4018 push @log_opts, $x;
4019 }
4020 }
4021 }
4022
4023 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4024 $gs ||= Git::SVN->_new;
4025 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4026 $gs->refname);
4027 push @cmd, '-r' unless $non_recursive;
4028 push @cmd, qw/--raw --name-status/ if $verbose;
4029 push @cmd, '--color' if log_use_color();
4030 push @cmd, @log_opts;
4031 if (defined $r_max && $r_max == $r_min) {
4032 push @cmd, '--max-count=1';
4033 if (my $c = $gs->rev_db_get($r_max)) {
4034 push @cmd, $c;
4035 }
4036 } elsif (defined $r_max) {
4037 if ($r_max < $r_min) {
4038 ($r_min, $r_max) = ($r_max, $r_min);
4039 }
4040 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4041 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4042 # If there are no commits in the range, both $c_max and $c_min
4043 # will be undefined. If there is at least 1 commit in the
4044 # range, both will be defined.
4045 return () if !defined $c_min || !defined $c_max;
4046 if ($c_min eq $c_max) {
4047 push @cmd, '--max-count=1', $c_min;
4048 } else {
4049 push @cmd, '--boundary', "$c_min..$c_max";
4050 }
4051 }
4052 return (@cmd, @files);
4053 }
4054
4055 # adapted from pager.c
4056 sub config_pager {
4057 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4058 if (!defined $pager) {
4059 $pager = 'less';
4060 } elsif (length $pager == 0 || $pager eq 'cat') {
4061 $pager = undef;
4062 }
4063 }
4064
4065 sub run_pager {
4066 return unless -t *STDOUT && defined $pager;
4067 pipe my $rfd, my $wfd or return;
4068 defined(my $pid = fork) or ::fatal "Can't fork: $!";
4069 if (!$pid) {
4070 open STDOUT, '>&', $wfd or
4071 ::fatal "Can't redirect to stdout: $!";
4072 return;
4073 }
4074 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4075 $ENV{LESS} ||= 'FRSX';
4076 exec $pager or ::fatal "Can't run pager: $! ($pager)";
4077 }
4078
4079 sub format_svn_date {
4080 return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4081 }
4082
4083 sub parse_git_date {
4084 my ($t, $tz) = @_;
4085 # Date::Parse isn't in the standard Perl distro :(
4086 if ($tz =~ s/^\+//) {
4087 $t += tz_to_s_offset($tz);
4088 } elsif ($tz =~ s/^\-//) {
4089 $t -= tz_to_s_offset($tz);
4090 }
4091 return $t;
4092 }
4093
4094 sub set_local_timezone {
4095 if (defined $TZ) {
4096 $ENV{TZ} = $TZ;
4097 } else {
4098 delete $ENV{TZ};
4099 }
4100 }
4101
4102 sub tz_to_s_offset {
4103 my ($tz) = @_;
4104 $tz =~ s/(\d\d)$//;
4105 return ($1 * 60) + ($tz * 3600);
4106 }
4107
4108 sub get_author_info {
4109 my ($dest, $author, $t, $tz) = @_;
4110 $author =~ s/(?:^\s*|\s*$)//g;
4111 $dest->{a_raw} = $author;
4112 my $au;
4113 if ($::_authors) {
4114 $au = $rusers{$author} || undef;
4115 }
4116 if (!$au) {
4117 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4118 }
4119 $dest->{t} = $t;
4120 $dest->{tz} = $tz;
4121 $dest->{a} = $au;
4122 $dest->{t_utc} = parse_git_date($t, $tz);
4123 }
4124
4125 sub process_commit {
4126 my ($c, $r_min, $r_max, $defer) = @_;
4127 if (defined $r_min && defined $r_max) {
4128 if ($r_min == $c->{r} && $r_min == $r_max) {
4129 show_commit($c);
4130 return 0;
4131 }
4132 return 1 if $r_min == $r_max;
4133 if ($r_min < $r_max) {
4134 # we need to reverse the print order
4135 return 0 if (defined $limit && --$limit < 0);
4136 push @$defer, $c;
4137 return 1;
4138 }
4139 if ($r_min != $r_max) {
4140 return 1 if ($r_min < $c->{r});
4141 return 1 if ($r_max > $c->{r});
4142 }
4143 }
4144 return 0 if (defined $limit && --$limit < 0);
4145 show_commit($c);
4146 return 1;
4147 }
4148
4149 sub show_commit {
4150 my $c = shift;
4151 if ($oneline) {
4152 my $x = "\n";
4153 if (my $l = $c->{l}) {
4154 while ($l->[0] =~ /^\s*$/) { shift @$l }
4155 $x = $l->[0];
4156 }
4157 $l_fmt ||= 'A' . length($c->{r});
4158 print 'r',pack($l_fmt, $c->{r}),' | ';
4159 print "$c->{c} | " if $show_commit;
4160 print $x;
4161 } else {
4162 show_commit_normal($c);
4163 }
4164 }
4165
4166 sub show_commit_changed_paths {
4167 my ($c) = @_;
4168 return unless $c->{changed};
4169 print "Changed paths:\n", @{$c->{changed}};
4170 }
4171
4172 sub show_commit_normal {
4173 my ($c) = @_;
4174 print commit_log_separator, "r$c->{r} | ";
4175 print "$c->{c} | " if $show_commit;
4176 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4177 my $nr_line = 0;
4178
4179 if (my $l = $c->{l}) {
4180 while ($l->[$#$l] eq "\n" && $#$l > 0
4181 && $l->[($#$l - 1)] eq "\n") {
4182 pop @$l;
4183 }
4184 $nr_line = scalar @$l;
4185 if (!$nr_line) {
4186 print "1 line\n\n\n";
4187 } else {
4188 if ($nr_line == 1) {
4189 $nr_line = '1 line';
4190 } else {
4191 $nr_line .= ' lines';
4192 }
4193 print $nr_line, "\n";
4194 show_commit_changed_paths($c);
4195 print "\n";
4196 print $_ foreach @$l;
4197 }
4198 } else {
4199 print "1 line\n";
4200 show_commit_changed_paths($c);
4201 print "\n";
4202
4203 }
4204 foreach my $x (qw/raw stat diff/) {
4205 if ($c->{$x}) {
4206 print "\n";
4207 print $_ foreach @{$c->{$x}}
4208 }
4209 }
4210 }
4211
4212 sub cmd_show_log {
4213 my (@args) = @_;
4214 my ($r_min, $r_max);
4215 my $r_last = -1; # prevent dupes
4216 set_local_timezone();
4217 if (defined $::_revision) {
4218 if ($::_revision =~ /^(\d+):(\d+)$/) {
4219 ($r_min, $r_max) = ($1, $2);
4220 } elsif ($::_revision =~ /^\d+$/) {
4221 $r_min = $r_max = $::_revision;
4222 } else {
4223 ::fatal "-r$::_revision is not supported, use ",
4224 "standard 'git log' arguments instead";
4225 }
4226 }
4227
4228 config_pager();
4229 @args = git_svn_log_cmd($r_min, $r_max, @args);
4230 if (!@args) {
4231 print commit_log_separator unless $incremental || $oneline;
4232 return;
4233 }
4234 my $log = command_output_pipe(@args);
4235 run_pager();
4236 my (@k, $c, $d, $stat);
4237 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4238 while (<$log>) {
4239 if (/^${esc_color}commit -?($::sha1_short)/o) {
4240 my $cmt = $1;
4241 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4242 $r_last = $c->{r};
4243 process_commit($c, $r_min, $r_max, \@k) or
4244 goto out;
4245 }
4246 $d = undef;
4247 $c = { c => $cmt };
4248 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4249 get_author_info($c, $1, $2, $3);
4250 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4251 # ignore
4252 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4253 push @{$c->{raw}}, $_;
4254 } elsif (/^${esc_color}[ACRMDT]\t/) {
4255 # we could add $SVN->{svn_path} here, but that requires
4256 # remote access at the moment (repo_path_split)...
4257 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
4258 push @{$c->{changed}}, $_;
4259 } elsif (/^${esc_color}diff /o) {
4260 $d = 1;
4261 push @{$c->{diff}}, $_;
4262 } elsif ($d) {
4263 push @{$c->{diff}}, $_;
4264 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4265 $esc_color*[\+\-]*$esc_color$/x) {
4266 $stat = 1;
4267 push @{$c->{stat}}, $_;
4268 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4269 push @{$c->{stat}}, $_;
4270 $stat = undef;
4271 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
4272 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4273 } elsif (s/^${esc_color} //o) {
4274 push @{$c->{l}}, $_;
4275 }
4276 }
4277 if ($c && defined $c->{r} && $c->{r} != $r_last) {
4278 $r_last = $c->{r};
4279 process_commit($c, $r_min, $r_max, \@k);
4280 }
4281 if (@k) {
4282 ($r_min, $r_max) = ($r_max, $r_min);
4283 process_commit($_, $r_min, $r_max) foreach reverse @k;
4284 }
4285 out:
4286 close $log;
4287 print commit_log_separator unless $incremental || $oneline;
4288 }
4289
4290 package Git::SVN::Migration;
4291 # these version numbers do NOT correspond to actual version numbers
4292 # of git nor git-svn. They are just relative.
4293 #
4294 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4295 #
4296 # v1 layout: .git/$id/info/url, refs/remotes/$id
4297 #
4298 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4299 #
4300 # v3 layout: .git/svn/$id, refs/remotes/$id
4301 # - info/url may remain for backwards compatibility
4302 # - this is what we migrate up to this layout automatically,
4303 # - this will be used by git svn init on single branches
4304 # v3.1 layout (auto migrated):
4305 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4306 # for backwards compatibility
4307 #
4308 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4309 # - this is only created for newly multi-init-ed
4310 # repositories. Similar in spirit to the
4311 # --use-separate-remotes option in git-clone (now default)
4312 # - we do not automatically migrate to this (following
4313 # the example set by core git)
4314 use strict;
4315 use warnings;
4316 use Carp qw/croak/;
4317 use File::Path qw/mkpath/;
4318 use File::Basename qw/dirname basename/;
4319 use vars qw/$_minimize/;
4320
4321 sub migrate_from_v0 {
4322 my $git_dir = $ENV{GIT_DIR};
4323 return undef unless -d $git_dir;
4324 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4325 my $migrated = 0;
4326 while (<$fh>) {
4327 chomp;
4328 my ($id, $orig_ref) = ($_, $_);
4329 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4330 next unless -f "$git_dir/$id/info/url";
4331 my $new_ref = "refs/remotes/$id";
4332 if (::verify_ref("$new_ref^0")) {
4333 print STDERR "W: $orig_ref is probably an old ",
4334 "branch used by an ancient version of ",
4335 "git-svn.\n",
4336 "However, $new_ref also exists.\n",
4337 "We will not be able ",
4338 "to use this branch until this ",
4339 "ambiguity is resolved.\n";
4340 next;
4341 }
4342 print STDERR "Migrating from v0 layout...\n" if !$migrated;
4343 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4344 command_noisy('update-ref', $new_ref, $orig_ref);
4345 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4346 $migrated++;
4347 }
4348 command_close_pipe($fh, $ctx);
4349 print STDERR "Done migrating from v0 layout...\n" if $migrated;
4350 $migrated;
4351 }
4352
4353 sub migrate_from_v1 {
4354 my $git_dir = $ENV{GIT_DIR};
4355 my $migrated = 0;
4356 return $migrated unless -d $git_dir;
4357 my $svn_dir = "$git_dir/svn";
4358
4359 # just in case somebody used 'svn' as their $id at some point...
4360 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4361
4362 print STDERR "Migrating from a git-svn v1 layout...\n";
4363 mkpath([$svn_dir]);
4364 print STDERR "Data from a previous version of git-svn exists, but\n\t",
4365 "$svn_dir\n\t(required for this version ",
4366 "($::VERSION) of git-svn) does not. exist\n";
4367 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4368 while (<$fh>) {
4369 my $x = $_;
4370 next unless $x =~ s#^refs/remotes/##;
4371 chomp $x;
4372 next unless -f "$git_dir/$x/info/url";
4373 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4374 next unless $u;
4375 my $dn = dirname("$git_dir/svn/$x");
4376 mkpath([$dn]) unless -d $dn;
4377 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4378 mkpath(["$git_dir/svn/svn"]);
4379 print STDERR " - $git_dir/$x/info => ",
4380 "$git_dir/svn/$x/info\n";
4381 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4382 croak "$!: $x";
4383 # don't worry too much about these, they probably
4384 # don't exist with repos this old (save for index,
4385 # and we can easily regenerate that)
4386 foreach my $f (qw/unhandled.log index .rev_db/) {
4387 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4388 }
4389 } else {
4390 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4391 rename "$git_dir/$x", "$git_dir/svn/$x" or
4392 croak "$!: $x";
4393 }
4394 $migrated++;
4395 }
4396 command_close_pipe($fh, $ctx);
4397 print STDERR "Done migrating from a git-svn v1 layout\n";
4398 $migrated;
4399 }
4400
4401 sub read_old_urls {
4402 my ($l_map, $pfx, $path) = @_;
4403 my @dir;
4404 foreach (<$path/*>) {
4405 if (-r "$_/info/url") {
4406 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4407 my $ref_id = $pfx . basename $_;
4408 my $url = ::file_to_s("$_/info/url");
4409 $l_map->{$ref_id} = $url;
4410 } elsif (-d $_) {
4411 push @dir, $_;
4412 }
4413 }
4414 foreach (@dir) {
4415 my $x = $_;
4416 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4417 read_old_urls($l_map, $x, $_);
4418 }
4419 }
4420
4421 sub migrate_from_v2 {
4422 my @cfg = command(qw/config -l/);
4423 return if grep /^svn-remote\..+\.url=/, @cfg;
4424 my %l_map;
4425 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4426 my $migrated = 0;
4427
4428 foreach my $ref_id (sort keys %l_map) {
4429 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4430 if ($@) {
4431 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4432 }
4433 $migrated++;
4434 }
4435 $migrated;
4436 }
4437
4438 sub minimize_connections {
4439 my $r = Git::SVN::read_all_remotes();
4440 my $new_urls = {};
4441 my $root_repos = {};
4442 foreach my $repo_id (keys %$r) {
4443 my $url = $r->{$repo_id}->{url} or next;
4444 my $fetch = $r->{$repo_id}->{fetch} or next;
4445 my $ra = Git::SVN::Ra->new($url);
4446
4447 # skip existing cases where we already connect to the root
4448 if (($ra->{url} eq $ra->{repos_root}) ||
4449 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
4450 $repo_id)) {
4451 $root_repos->{$ra->{url}} = $repo_id;
4452 next;
4453 }
4454
4455 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4456 my $root_path = $ra->{url};
4457 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4458 foreach my $path (keys %$fetch) {
4459 my $ref_id = $fetch->{$path};
4460 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4461
4462 # make sure we can read when connecting to
4463 # a higher level of a repository
4464 my ($last_rev, undef) = $gs->last_rev_commit;
4465 if (!defined $last_rev) {
4466 $last_rev = eval {
4467 $root_ra->get_latest_revnum;
4468 };
4469 next if $@;
4470 }
4471 my $new = $root_path;
4472 $new .= length $path ? "/$path" : '';
4473 eval {
4474 $root_ra->get_log([$new], $last_rev, $last_rev,
4475 0, 0, 1, sub { });
4476 };
4477 next if $@;
4478 $new_urls->{$ra->{repos_root}}->{$new} =
4479 { ref_id => $ref_id,
4480 old_repo_id => $repo_id,
4481 old_path => $path };
4482 }
4483 }
4484
4485 my @emptied;
4486 foreach my $url (keys %$new_urls) {
4487 # see if we can re-use an existing [svn-remote "repo_id"]
4488 # instead of creating a(n ugly) new section:
4489 my $repo_id = $root_repos->{$url} ||
4490 Git::SVN::sanitize_remote_name($url);
4491
4492 my $fetch = $new_urls->{$url};
4493 foreach my $path (keys %$fetch) {
4494 my $x = $fetch->{$path};
4495 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4496 my $pfx = "svn-remote.$x->{old_repo_id}";
4497
4498 my $old_fetch = quotemeta("$x->{old_path}:".
4499 "refs/remotes/$x->{ref_id}");
4500 command_noisy(qw/config --unset/,
4501 "$pfx.fetch", '^'. $old_fetch . '$');
4502 delete $r->{$x->{old_repo_id}}->
4503 {fetch}->{$x->{old_path}};
4504 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4505 command_noisy(qw/config --unset/,
4506 "$pfx.url");
4507 push @emptied, $x->{old_repo_id}
4508 }
4509 }
4510 }
4511 if (@emptied) {
4512 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4513 "$ENV{GIT_DIR}/config";
4514 print STDERR <<EOF;
4515 The following [svn-remote] sections in your config file ($file) are empty
4516 and can be safely removed:
4517 EOF
4518 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4519 }
4520 }
4521
4522 sub migration_check {
4523 migrate_from_v0();
4524 migrate_from_v1();
4525 migrate_from_v2();
4526 minimize_connections() if $_minimize;
4527 }
4528
4529 package Git::IndexInfo;
4530 use strict;
4531 use warnings;
4532 use Git qw/command_input_pipe command_close_pipe/;
4533
4534 sub new {
4535 my ($class) = @_;
4536 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4537 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4538 }
4539
4540 sub remove {
4541 my ($self, $path) = @_;
4542 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4543 return ++$self->{nr};
4544 }
4545 undef;
4546 }
4547
4548 sub update {
4549 my ($self, $mode, $hash, $path) = @_;
4550 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4551 return ++$self->{nr};
4552 }
4553 undef;
4554 }
4555
4556 sub DESTROY {
4557 my ($self) = @_;
4558 command_close_pipe($self->{gui}, $self->{ctx});
4559 }
4560
4561 package Git::SVN::GlobSpec;
4562 use strict;
4563 use warnings;
4564
4565 sub new {
4566 my ($class, $glob) = @_;
4567 my $re = $glob;
4568 $re =~ s!/+$!!g; # no need for trailing slashes
4569 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
4570 my ($left, $right) = ($1, $2);
4571 if ($nr > 1) {
4572 die "Only one '*' wildcard expansion ",
4573 "is supported (got $nr): '$glob'\n";
4574 } elsif ($nr == 0) {
4575 die "One '*' is needed for glob: '$glob'\n";
4576 }
4577 $re = quotemeta($left) . $re . quotemeta($right);
4578 if (length $left && !($left =~ s!/+$!!g)) {
4579 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4580 }
4581 if (length $right && !($right =~ s!^/+!!g)) {
4582 die "Missing leading '/' on right side of: '$glob' ($right)\n";
4583 }
4584 my $left_re = qr/^\/\Q$left\E(\/|$)/;
4585 bless { left => $left, right => $right, left_regex => $left_re,
4586 regex => qr/$re/, glob => $glob }, $class;
4587 }
4588
4589 sub full_path {
4590 my ($self, $path) = @_;
4591 return (length $self->{left} ? "$self->{left}/" : '') .
4592 $path . (length $self->{right} ? "/$self->{right}" : '');
4593 }
4594
4595 __END__
4596
4597 Data structures:
4598
4599
4600 $remotes = { # returned by read_all_remotes()
4601 'svn' => {
4602 # svn-remote.svn.url=https://svn.musicpd.org
4603 url => 'https://svn.musicpd.org',
4604 # svn-remote.svn.fetch=mpd/trunk:trunk
4605 fetch => {
4606 'mpd/trunk' => 'trunk',
4607 },
4608 # svn-remote.svn.tags=mpd/tags/*:tags/*
4609 tags => {
4610 path => {
4611 left => 'mpd/tags',
4612 right => '',
4613 regex => qr!mpd/tags/([^/]+)$!,
4614 glob => 'tags/*',
4615 },
4616 ref => {
4617 left => 'tags',
4618 right => '',
4619 regex => qr!tags/([^/]+)$!,
4620 glob => 'tags/*',
4621 },
4622 }
4623 }
4624 };
4625
4626 $log_entry hashref as returned by libsvn_log_entry()
4627 {
4628 log => 'whitespace-formatted log entry
4629 ', # trailing newline is preserved
4630 revision => '8', # integer
4631 date => '2004-02-24T17:01:44.108345Z', # commit date
4632 author => 'committer name'
4633 };
4634
4635
4636 # this is generated by generate_diff();
4637 @mods = array of diff-index line hashes, each element represents one line
4638 of diff-index output
4639
4640 diff-index line ($m hash)
4641 {
4642 mode_a => first column of diff-index output, no leading ':',
4643 mode_b => second column of diff-index output,
4644 sha1_b => sha1sum of the final blob,
4645 chg => change type [MCRADT],
4646 file_a => original file name of a file (iff chg is 'C' or 'R')
4647 file_b => new/current file name of a file (any chg)
4648 }
4649 ;
4650
4651 # retval of read_url_paths{,_all}();
4652 $l_map = {
4653 # repository root url
4654 'https://svn.musicpd.org' => {
4655 # repository path # GIT_SVN_ID
4656 'mpd/trunk' => 'trunk',
4657 'mpd/tags/0.11.5' => 'tags/0.11.5',
4658 },
4659 }
4660
4661 Notes:
4662 I don't trust the each() function on unless I created %hash myself
4663 because the internal iterator may not have started at base.
peff's git enhancements