?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-svn: add a show-externals command.
[git/dkf.git] / git-svn.perl
blob8dbaf2f33c577749f8267279462e0d5493847120
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 $_fetch_all ? $gs->fetch_all : $gs->fetch;
554 }
555 command_noisy(rebase_cmd(), $gs->refname);
556 }
557
558 sub cmd_show_ignore {
559 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
560 $gs ||= Git::SVN->new;
561 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
562 $gs->prop_walk($gs->{path}, $r, sub {
563 my ($gs, $path, $props) = @_;
564 print STDOUT "\n# $path\n";
565 my $s = $props->{'svn:ignore'} or return;
566 $s =~ s/[\r\n]+/\n/g;
567 chomp $s;
568 $s =~ s#^#$path#gm;
569 print STDOUT "$s\n";
570 });
571 }
572
573 sub cmd_show_externals {
574 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
575 $gs ||= Git::SVN->new;
576 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
577 $gs->prop_walk($gs->{path}, $r, sub {
578 my ($gs, $path, $props) = @_;
579 print STDOUT "\n# $path\n";
580 my $s = $props->{'svn:externals'} or return;
581 $s =~ s/[\r\n]+/\n/g;
582 chomp $s;
583 $s =~ s#^#$path#gm;
584 print STDOUT "$s\n";
585 });
586 }
587
588 sub cmd_create_ignore {
589 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
590 $gs ||= Git::SVN->new;
591 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
592 $gs->prop_walk($gs->{path}, $r, sub {
593 my ($gs, $path, $props) = @_;
594 # $path is of the form /path/to/dir/
595 my $ignore = '.' . $path . '.gitignore';
596 my $s = $props->{'svn:ignore'} or return;
597 open(GITIGNORE, '>', $ignore)
598 or fatal("Failed to open `$ignore' for writing: $!");
599 $s =~ s/[\r\n]+/\n/g;
600 chomp $s;
601 # Prefix all patterns so that the ignore doesn't apply
602 # to sub-directories.
603 $s =~ s#^#/#gm;
604 print GITIGNORE "$s\n";
605 close(GITIGNORE)
606 or fatal("Failed to close `$ignore': $!");
607 command_noisy('add', $ignore);
608 });
609 }
610
611 sub canonicalize_path {
612 my ($path) = @_;
613 my $dot_slash_added = 0;
614 if (substr($path, 0, 1) ne "/") {
615 $path = "./" . $path;
616 $dot_slash_added = 1;
617 }
618 # File::Spec->canonpath doesn't collapse x/../y into y (for a
619 # good reason), so let's do this manually.
620 $path =~ s#/+#/#g;
621 $path =~ s#/\.(?:/|$)#/#g;
622 $path =~ s#/[^/]+/\.\.##g;
623 $path =~ s#/$##g;
624 $path =~ s#^\./## if $dot_slash_added;
625 return $path;
626 }
627
628 # get_svnprops(PATH)
629 # ------------------
630 # Helper for cmd_propget and cmd_proplist below.
631 sub get_svnprops {
632 my $path = shift;
633 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
634 $gs ||= Git::SVN->new;
635
636 # prefix THE PATH by the sub-directory from which the user
637 # invoked us.
638 $path = $cmd_dir_prefix . $path;
639 fatal("No such file or directory: $path") unless -e $path;
640 my $is_dir = -d $path ? 1 : 0;
641 $path = $gs->{path} . '/' . $path;
642
643 # canonicalize the path (otherwise libsvn will abort or fail to
644 # find the file)
645 $path = canonicalize_path($path);
646
647 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
648 my $props;
649 if ($is_dir) {
650 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
651 }
652 else {
653 (undef, $props) = $gs->ra->get_file($path, $r, undef);
654 }
655 return $props;
656 }
657
658 # cmd_propget (PROP, PATH)
659 # ------------------------
660 # Print the SVN property PROP for PATH.
661 sub cmd_propget {
662 my ($prop, $path) = @_;
663 $path = '.' if not defined $path;
664 usage(1) if not defined $prop;
665 my $props = get_svnprops($path);
666 if (not defined $props->{$prop}) {
667 fatal("`$path' does not have a `$prop' SVN property.");
668 }
669 print $props->{$prop} . "\n";
670 }
671
672 # cmd_proplist (PATH)
673 # -------------------
674 # Print the list of SVN properties for PATH.
675 sub cmd_proplist {
676 my $path = shift;
677 $path = '.' if not defined $path;
678 my $props = get_svnprops($path);
679 print "Properties on '$path':\n";
680 foreach (sort keys %{$props}) {
681 print " $_\n";
682 }
683 }
684
685 sub cmd_multi_init {
686 my $url = shift;
687 unless (defined $_trunk || defined $_branches || defined $_tags) {
688 usage(1);
689 }
690
691 # there are currently some bugs that prevent multi-init/multi-fetch
692 # setups from working well without this.
693 $Git::SVN::_minimize_url = 1;
694
695 $_prefix = '' unless defined $_prefix;
696 if (defined $url) {
697 $url =~ s#/+$##;
698 init_subdir(@_);
699 }
700 do_git_init_db();
701 if (defined $_trunk) {
702 my $trunk_ref = $_prefix . 'trunk';
703 # try both old-style and new-style lookups:
704 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
705 unless ($gs_trunk) {
706 my ($trunk_url, $trunk_path) =
707 complete_svn_url($url, $_trunk);
708 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
709 undef, $trunk_ref);
710 }
711 }
712 return unless defined $_branches || defined $_tags;
713 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
714 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
715 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
716 }
717
718 sub cmd_multi_fetch {
719 my $remotes = Git::SVN::read_all_remotes();
720 foreach my $repo_id (sort keys %$remotes) {
721 if ($remotes->{$repo_id}->{url}) {
722 Git::SVN::fetch_all($repo_id, $remotes);
723 }
724 }
725 }
726
727 # this command is special because it requires no metadata
728 sub cmd_commit_diff {
729 my ($ta, $tb, $url) = @_;
730 my $usage = "Usage: $0 commit-diff -r<revision> ".
731 "<tree-ish> <tree-ish> [<URL>]";
732 fatal($usage) if (!defined $ta || !defined $tb);
733 my $svn_path;
734 if (!defined $url) {
735 my $gs = eval { Git::SVN->new };
736 if (!$gs) {
737 fatal("Needed URL or usable git-svn --id in ",
738 "the command-line\n", $usage);
739 }
740 $url = $gs->{url};
741 $svn_path = $gs->{path};
742 }
743 unless (defined $_revision) {
744 fatal("-r|--revision is a required argument\n", $usage);
745 }
746 if (defined $_message && defined $_file) {
747 fatal("Both --message/-m and --file/-F specified ",
748 "for the commit message.\n",
749 "I have no idea what you mean");
750 }
751 if (defined $_file) {
752 $_message = file_to_s($_file);
753 } else {
754 $_message ||= get_commit_entry($tb)->{log};
755 }
756 my $ra ||= Git::SVN::Ra->new($url);
757 $svn_path ||= $ra->{svn_path};
758 my $r = $_revision;
759 if ($r eq 'HEAD') {
760 $r = $ra->get_latest_revnum;
761 } elsif ($r !~ /^\d+$/) {
762 die "revision argument: $r not understood by git-svn\n";
763 }
764 my %ed_opts = ( r => $r,
765 log => $_message,
766 ra => $ra,
767 tree_a => $ta,
768 tree_b => $tb,
769 editor_cb => sub { print "Committed r$_[0]\n" },
770 svn_path => $svn_path );
771 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
772 print "No changes\n$ta == $tb\n";
773 }
774 }
775
776 sub cmd_info {
777 my $path = canonicalize_path(shift or ".");
778 unless (scalar(@_) == 0) {
779 die "Too many arguments specified\n";
780 }
781
782 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
783
784 if (!$file_type && !$diff_status) {
785 print STDERR "$path: (Not a versioned resource)\n\n";
786 return;
787 }
788
789 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
790 unless ($gs) {
791 die "Unable to determine upstream SVN information from ",
792 "working tree history\n";
793 }
794 my $full_url = $url . ($path eq "." ? "" : "/$path");
795
796 if ($_url) {
797 print $full_url, "\n";
798 return;
799 }
800
801 my $result = "Path: $path\n";
802 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
803 $result .= "URL: " . $full_url . "\n";
804
805 eval {
806 my $repos_root = $gs->repos_root;
807 Git::SVN::remove_username($repos_root);
808 $result .= "Repository Root: $repos_root\n";
809 };
810 if ($@) {
811 $result .= "Repository Root: (offline)\n";
812 }
813 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
814 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
815
816 $result .= "Node Kind: " .
817 ($file_type eq "dir" ? "directory" : "file") . "\n";
818
819 my $schedule = $diff_status eq "A"
820 ? "add"
821 : ($diff_status eq "D" ? "delete" : "normal");
822 $result .= "Schedule: $schedule\n";
823
824 if ($diff_status eq "A") {
825 print $result, "\n";
826 return;
827 }
828
829 my ($lc_author, $lc_rev, $lc_date_utc);
830 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $path);
831 my $log = command_output_pipe(@args);
832 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
833 while (<$log>) {
834 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
835 $lc_author = $1;
836 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
837 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
838 (undef, $lc_rev, undef) = ::extract_metadata($1);
839 }
840 }
841 close $log;
842
843 Git::SVN::Log::set_local_timezone();
844
845 $result .= "Last Changed Author: $lc_author\n";
846 $result .= "Last Changed Rev: $lc_rev\n";
847 $result .= "Last Changed Date: " .
848 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
849
850 if ($file_type ne "dir") {
851 my $text_last_updated_date =
852 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
853 $result .=
854 "Text Last Updated: " .
855 Git::SVN::Log::format_svn_date($text_last_updated_date) .
856 "\n";
857 my $checksum;
858 if ($diff_status eq "D") {
859 my ($fh, $ctx) =
860 command_output_pipe(qw(cat-file blob), "HEAD:$path");
861 if ($file_type eq "link") {
862 my $file_name = <$fh>;
863 $checksum = md5sum("link $file_name");
864 } else {
865 $checksum = md5sum($fh);
866 }
867 command_close_pipe($fh, $ctx);
868 } elsif ($file_type eq "link") {
869 my $file_name =
870 command(qw(cat-file blob), "HEAD:$path");
871 $checksum =
872 md5sum("link " . $file_name);
873 } else {
874 open FILE, "<", $path or die $!;
875 $checksum = md5sum(\*FILE);
876 close FILE or die $!;
877 }
878 $result .= "Checksum: " . $checksum . "\n";
879 }
880
881 print $result, "\n";
882 }
883
884 ########################### utility functions #########################
885
886 sub rebase_cmd {
887 my @cmd = qw/rebase/;
888 push @cmd, '-v' if $_verbose;
889 push @cmd, qw/--merge/ if $_merge;
890 push @cmd, "--strategy=$_strategy" if $_strategy;
891 @cmd;
892 }
893
894 sub post_fetch_checkout {
895 return if $_no_checkout;
896 my $gs = $Git::SVN::_head or return;
897 return if verify_ref('refs/heads/master^0');
898
899 my $valid_head = verify_ref('HEAD^0');
900 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
901 return if ($valid_head || !verify_ref('HEAD^0'));
902
903 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
904 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
905 return if -f $index;
906
907 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
908 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
909 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
910 print STDERR "Checked out HEAD:\n ",
911 $gs->full_url, " r", $gs->last_rev, "\n";
912 }
913
914 sub complete_svn_url {
915 my ($url, $path) = @_;
916 $path =~ s#/+$##;
917 if ($path !~ m#^[a-z\+]+://#) {
918 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
919 fatal("E: '$path' is not a complete URL ",
920 "and a separate URL is not specified");
921 }
922 return ($url, $path);
923 }
924 return ($path, '');
925 }
926
927 sub complete_url_ls_init {
928 my ($ra, $repo_path, $switch, $pfx) = @_;
929 unless ($repo_path) {
930 print STDERR "W: $switch not specified\n";
931 return;
932 }
933 $repo_path =~ s#/+$##;
934 if ($repo_path =~ m#^[a-z\+]+://#) {
935 $ra = Git::SVN::Ra->new($repo_path);
936 $repo_path = '';
937 } else {
938 $repo_path =~ s#^/+##;
939 unless ($ra) {
940 fatal("E: '$repo_path' is not a complete URL ",
941 "and a separate URL is not specified");
942 }
943 }
944 my $url = $ra->{url};
945 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
946 my $k = "svn-remote.$gs->{repo_id}.url";
947 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
948 if ($orig_url && ($orig_url ne $gs->{url})) {
949 die "$k already set: $orig_url\n",
950 "wanted to set to: $gs->{url}\n";
951 }
952 command_oneline('config', $k, $gs->{url}) unless $orig_url;
953 my $remote_path = "$ra->{svn_path}/$repo_path/*";
954 $remote_path =~ s#/+#/#g;
955 $remote_path =~ s#^/##g;
956 my ($n) = ($switch =~ /^--(\w+)/);
957 if (length $pfx && $pfx !~ m#/$#) {
958 die "--prefix='$pfx' must have a trailing slash '/'\n";
959 }
960 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
961 "$remote_path:refs/remotes/$pfx*");
962 }
963
964 sub verify_ref {
965 my ($ref) = @_;
966 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
967 { STDERR => 0 }); };
968 }
969
970 sub get_tree_from_treeish {
971 my ($treeish) = @_;
972 # $treeish can be a symbolic ref, too:
973 my $type = command_oneline(qw/cat-file -t/, $treeish);
974 my $expected;
975 while ($type eq 'tag') {
976 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
977 }
978 if ($type eq 'commit') {
979 $expected = (grep /^tree /, command(qw/cat-file commit/,
980 $treeish))[0];
981 ($expected) = ($expected =~ /^tree ($sha1)$/o);
982 die "Unable to get tree from $treeish\n" unless $expected;
983 } elsif ($type eq 'tree') {
984 $expected = $treeish;
985 } else {
986 die "$treeish is a $type, expected tree, tag or commit\n";
987 }
988 return $expected;
989 }
990
991 sub get_commit_entry {
992 my ($treeish) = shift;
993 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
994 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
995 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
996 open my $log_fh, '>', $commit_editmsg or croak $!;
997
998 my $type = command_oneline(qw/cat-file -t/, $treeish);
999 if ($type eq 'commit' || $type eq 'tag') {
1000 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1001 $type, $treeish);
1002 my $in_msg = 0;
1003 while (<$msg_fh>) {
1004 if (!$in_msg) {
1005 $in_msg = 1 if (/^\s*$/);
1006 } elsif (/^git-svn-id: /) {
1007 # skip this for now, we regenerate the
1008 # correct one on re-fetch anyways
1009 # TODO: set *:merge properties or like...
1010 } else {
1011 print $log_fh $_ or croak $!;
1012 }
1013 }
1014 command_close_pipe($msg_fh, $ctx);
1015 }
1016 close $log_fh or croak $!;
1017
1018 if ($_edit || ($type eq 'tree')) {
1019 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1020 # TODO: strip out spaces, comments, like git-commit.sh
1021 system($editor, $commit_editmsg);
1022 }
1023 rename $commit_editmsg, $commit_msg or croak $!;
1024 open $log_fh, '<', $commit_msg or croak $!;
1025 { local $/; chomp($log_entry{log} = <$log_fh>); }
1026 close $log_fh or croak $!;
1027 unlink $commit_msg;
1028 \%log_entry;
1029 }
1030
1031 sub s_to_file {
1032 my ($str, $file, $mode) = @_;
1033 open my $fd,'>',$file or croak $!;
1034 print $fd $str,"\n" or croak $!;
1035 close $fd or croak $!;
1036 chmod ($mode &~ umask, $file) if (defined $mode);
1037 }
1038
1039 sub file_to_s {
1040 my $file = shift;
1041 open my $fd,'<',$file or croak "$!: file: $file\n";
1042 local $/;
1043 my $ret = <$fd>;
1044 close $fd or croak $!;
1045 $ret =~ s/\s*$//s;
1046 return $ret;
1047 }
1048
1049 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1050 sub load_authors {
1051 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1052 my $log = $cmd eq 'log';
1053 while (<$authors>) {
1054 chomp;
1055 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1056 my ($user, $name, $email) = ($1, $2, $3);
1057 if ($log) {
1058 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1059 } else {
1060 $users{$user} = [$name, $email];
1061 }
1062 }
1063 close $authors or croak $!;
1064 }
1065
1066 # convert GetOpt::Long specs for use by git-config
1067 sub read_repo_config {
1068 return unless -d $ENV{GIT_DIR};
1069 my $opts = shift;
1070 my @config_only;
1071 foreach my $o (keys %$opts) {
1072 # if we have mixedCase and a long option-only, then
1073 # it's a config-only variable that we don't need for
1074 # the command-line.
1075 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1076 my $v = $opts->{$o};
1077 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1078 $key =~ s/-//g;
1079 my $arg = 'git-config';
1080 $arg .= ' --int' if ($o =~ /[:=]i$/);
1081 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1082 if (ref $v eq 'ARRAY') {
1083 chomp(my @tmp = `$arg --get-all svn.$key`);
1084 @$v = @tmp if @tmp;
1085 } else {
1086 chomp(my $tmp = `$arg --get svn.$key`);
1087 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1088 $$v = $tmp;
1089 }
1090 }
1091 }
1092 delete @$opts{@config_only} if @config_only;
1093 }
1094
1095 sub extract_metadata {
1096 my $id = shift or return (undef, undef, undef);
1097 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1098 \s([a-f\d\-]+)$/x);
1099 if (!defined $rev || !$uuid || !$url) {
1100 # some of the original repositories I made had
1101 # identifiers like this:
1102 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1103 }
1104 return ($url, $rev, $uuid);
1105 }
1106
1107 sub cmt_metadata {
1108 return extract_metadata((grep(/^git-svn-id: /,
1109 command(qw/cat-file commit/, shift)))[-1]);
1110 }
1111
1112 sub working_head_info {
1113 my ($head, $refs) = @_;
1114 my @args = ('log', '--no-color', '--first-parent');
1115 my ($fh, $ctx) = command_output_pipe(@args, $head);
1116 my $hash;
1117 my %max;
1118 while (<$fh>) {
1119 if ( m{^commit ($::sha1)$} ) {
1120 unshift @$refs, $hash if $hash and $refs;
1121 $hash = $1;
1122 next;
1123 }
1124 next unless s{^\s*(git-svn-id:)}{$1};
1125 my ($url, $rev, $uuid) = extract_metadata($_);
1126 if (defined $url && defined $rev) {
1127 next if $max{$url} and $max{$url} < $rev;
1128 if (my $gs = Git::SVN->find_by_url($url)) {
1129 my $c = $gs->rev_db_get($rev);
1130 if ($c && $c eq $hash) {
1131 close $fh; # break the pipe
1132 return ($url, $rev, $uuid, $gs);
1133 } else {
1134 $max{$url} ||= $gs->rev_db_max;
1135 }
1136 }
1137 }
1138 }
1139 command_close_pipe($fh, $ctx);
1140 (undef, undef, undef, undef);
1141 }
1142
1143 sub read_commit_parents {
1144 my ($parents, $c) = @_;
1145 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1146 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1147 @{$parents->{$c}} = split(/ /, $p);
1148 }
1149
1150 sub linearize_history {
1151 my ($gs, $refs) = @_;
1152 my %parents;
1153 foreach my $c (@$refs) {
1154 read_commit_parents(\%parents, $c);
1155 }
1156
1157 my @linear_refs;
1158 my %skip = ();
1159 my $last_svn_commit = $gs->last_commit;
1160 foreach my $c (reverse @$refs) {
1161 next if $c eq $last_svn_commit;
1162 last if $skip{$c};
1163
1164 unshift @linear_refs, $c;
1165 $skip{$c} = 1;
1166
1167 # we only want the first parent to diff against for linear
1168 # history, we save the rest to inject when we finalize the
1169 # svn commit
1170 my $fp_a = verify_ref("$c~1");
1171 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1172 if (!$fp_a || !$fp_b) {
1173 die "Commit $c\n",
1174 "has no parent commit, and therefore ",
1175 "nothing to diff against.\n",
1176 "You should be working from a repository ",
1177 "originally created by git-svn\n";
1178 }
1179 if ($fp_a ne $fp_b) {
1180 die "$c~1 = $fp_a, however parsing commit $c ",
1181 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1182 }
1183
1184 foreach my $p (@{$parents{$c}}) {
1185 $skip{$p} = 1;
1186 }
1187 }
1188 (\@linear_refs, \%parents);
1189 }
1190
1191 sub find_file_type_and_diff_status {
1192 my ($path) = @_;
1193 return ('dir', '') if $path eq '.';
1194
1195 my $diff_output =
1196 command_oneline(qw(diff --cached --name-status --), $path) || "";
1197 my $diff_status = (split(' ', $diff_output))[0] || "";
1198
1199 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1200
1201 return (undef, undef) if !$diff_status && !$ls_tree;
1202
1203 if ($diff_status eq "A") {
1204 return ("link", $diff_status) if -l $path;
1205 return ("dir", $diff_status) if -d $path;
1206 return ("file", $diff_status);
1207 }
1208
1209 my $mode = (split(' ', $ls_tree))[0] || "";
1210
1211 return ("link", $diff_status) if $mode eq "120000";
1212 return ("dir", $diff_status) if $mode eq "040000";
1213 return ("file", $diff_status);
1214 }
1215
1216 sub md5sum {
1217 my $arg = shift;
1218 my $ref = ref $arg;
1219 my $md5 = Digest::MD5->new();
1220 if ($ref eq 'GLOB' || $ref eq 'IO::File') {
1221 $md5->addfile($arg) or croak $!;
1222 } elsif ($ref eq 'SCALAR') {
1223 $md5->add($$arg) or croak $!;
1224 } elsif (!$ref) {
1225 $md5->add($arg) or croak $!;
1226 } else {
1227 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1228 }
1229 return $md5->hexdigest();
1230 }
1231
1232 package Git::SVN;
1233 use strict;
1234 use warnings;
1235 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1236 $_repack $_repack_flags $_use_svm_props $_head
1237 $_use_svnsync_props $no_reuse_existing $_minimize_url
1238 $_use_log_author/;
1239 use Carp qw/croak/;
1240 use File::Path qw/mkpath/;
1241 use File::Copy qw/copy/;
1242 use IPC::Open3;
1243
1244 my $_repack_nr;
1245 # properties that we do not log:
1246 my %SKIP_PROP;
1247 BEGIN {
1248 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1249 svn:special svn:executable
1250 svn:entry:committed-rev
1251 svn:entry:last-author
1252 svn:entry:uuid
1253 svn:entry:committed-date/;
1254
1255 # some options are read globally, but can be overridden locally
1256 # per [svn-remote "..."] section. Command-line options will *NOT*
1257 # override options set in an [svn-remote "..."] section
1258 no strict 'refs';
1259 for my $option (qw/follow_parent no_metadata use_svm_props
1260 use_svnsync_props/) {
1261 my $key = $option;
1262 $key =~ tr/_//d;
1263 my $prop = "-$option";
1264 *$option = sub {
1265 my ($self) = @_;
1266 return $self->{$prop} if exists $self->{$prop};
1267 my $k = "svn-remote.$self->{repo_id}.$key";
1268 eval { command_oneline(qw/config --get/, $k) };
1269 if ($@) {
1270 $self->{$prop} = ${"Git::SVN::_$option"};
1271 } else {
1272 my $v = command_oneline(qw/config --bool/,$k);
1273 $self->{$prop} = $v eq 'false' ? 0 : 1;
1274 }
1275 return $self->{$prop};
1276 }
1277 }
1278 }
1279
1280 my %LOCKFILES;
1281 END { unlink keys %LOCKFILES if %LOCKFILES }
1282
1283 sub resolve_local_globs {
1284 my ($url, $fetch, $glob_spec) = @_;
1285 return unless defined $glob_spec;
1286 my $ref = $glob_spec->{ref};
1287 my $path = $glob_spec->{path};
1288 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1289 next unless m#^refs/remotes/$ref->{regex}$#;
1290 my $p = $1;
1291 my $pathname = desanitize_refname($path->full_path($p));
1292 my $refname = desanitize_refname($ref->full_path($p));
1293 if (my $existing = $fetch->{$pathname}) {
1294 if ($existing ne $refname) {
1295 die "Refspec conflict:\n",
1296 "existing: refs/remotes/$existing\n",
1297 " globbed: refs/remotes/$refname\n";
1298 }
1299 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1300 $u =~ s!^\Q$url\E(/|$)!! or die
1301 "refs/remotes/$refname: '$url' not found in '$u'\n";
1302 if ($pathname ne $u) {
1303 warn "W: Refspec glob conflict ",
1304 "(ref: refs/remotes/$refname):\n",
1305 "expected path: $pathname\n",
1306 " real path: $u\n",
1307 "Continuing ahead with $u\n";
1308 next;
1309 }
1310 } else {
1311 $fetch->{$pathname} = $refname;
1312 }
1313 }
1314 }
1315
1316 sub parse_revision_argument {
1317 my ($base, $head) = @_;
1318 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1319 return ($base, $head);
1320 }
1321 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1322 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1323 return ($head, $head) if ($::_revision eq 'HEAD');
1324 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1325 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1326 die "revision argument: $::_revision not understood by git-svn\n";
1327 }
1328
1329 sub fetch_all {
1330 my ($repo_id, $remotes) = @_;
1331 if (ref $repo_id) {
1332 my $gs = $repo_id;
1333 $repo_id = undef;
1334 $repo_id = $gs->{repo_id};
1335 }
1336 $remotes ||= read_all_remotes();
1337 my $remote = $remotes->{$repo_id} or
1338 die "[svn-remote \"$repo_id\"] unknown\n";
1339 my $fetch = $remote->{fetch};
1340 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1341 my (@gs, @globs);
1342 my $ra = Git::SVN::Ra->new($url);
1343 my $uuid = $ra->get_uuid;
1344 my $head = $ra->get_latest_revnum;
1345 my $base = defined $fetch ? $head : 0;
1346
1347 # read the max revs for wildcard expansion (branches/*, tags/*)
1348 foreach my $t (qw/branches tags/) {
1349 defined $remote->{$t} or next;
1350 push @globs, $remote->{$t};
1351 my $max_rev = eval { tmp_config(qw/--int --get/,
1352 "svn-remote.$repo_id.${t}-maxRev") };
1353 if (defined $max_rev && ($max_rev < $base)) {
1354 $base = $max_rev;
1355 } elsif (!defined $max_rev) {
1356 $base = 0;
1357 }
1358 }
1359
1360 if ($fetch) {
1361 foreach my $p (sort keys %$fetch) {
1362 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1363 my $lr = $gs->rev_db_max;
1364 if (defined $lr) {
1365 $base = $lr if ($lr < $base);
1366 }
1367 push @gs, $gs;
1368 }
1369 }
1370
1371 ($base, $head) = parse_revision_argument($base, $head);
1372 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1373 }
1374
1375 sub read_all_remotes {
1376 my $r = {};
1377 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1378 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
1379 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1380 $local_ref =~ s{^/}{};
1381 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1382 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1383 $r->{$1}->{url} = $2;
1384 } elsif (m!^(.+)\.(branches|tags)=
1385 (.*):refs/remotes/(.+)\s*$/!x) {
1386 my ($p, $g) = ($3, $4);
1387 my $rs = $r->{$1}->{$2} = {
1388 t => $2,
1389 remote => $1,
1390 path => Git::SVN::GlobSpec->new($p),
1391 ref => Git::SVN::GlobSpec->new($g) };
1392 if (length($rs->{ref}->{right}) != 0) {
1393 die "The '*' glob character must be the last ",
1394 "character of '$g'\n";
1395 }
1396 }
1397 }
1398 $r;
1399 }
1400
1401 sub init_vars {
1402 if (defined $_repack) {
1403 $_repack = 1000 if ($_repack <= 0);
1404 $_repack_nr = $_repack;
1405 $_repack_flags ||= '-d';
1406 }
1407 }
1408
1409 sub verify_remotes_sanity {
1410 return unless -d $ENV{GIT_DIR};
1411 my %seen;
1412 foreach (command(qw/config -l/)) {
1413 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1414 if ($seen{$1}) {
1415 die "Remote ref refs/remote/$1 is tracked by",
1416 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1417 "Please resolve this ambiguity in ",
1418 "your git configuration file before ",
1419 "continuing\n";
1420 }
1421 $seen{$1} = $_;
1422 }
1423 }
1424 }
1425
1426 # we allow more chars than remotes2config.sh...
1427 sub sanitize_remote_name {
1428 my ($name) = @_;
1429 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1430 $name;
1431 }
1432
1433 sub find_existing_remote {
1434 my ($url, $remotes) = @_;
1435 return undef if $no_reuse_existing;
1436 my $existing;
1437 foreach my $repo_id (keys %$remotes) {
1438 my $u = $remotes->{$repo_id}->{url} or next;
1439 next if $u ne $url;
1440 $existing = $repo_id;
1441 last;
1442 }
1443 $existing;
1444 }
1445
1446 sub init_remote_config {
1447 my ($self, $url, $no_write) = @_;
1448 $url =~ s!/+$!!; # strip trailing slash
1449 my $r = read_all_remotes();
1450 my $existing = find_existing_remote($url, $r);
1451 if ($existing) {
1452 unless ($no_write) {
1453 print STDERR "Using existing ",
1454 "[svn-remote \"$existing\"]\n";
1455 }
1456 $self->{repo_id} = $existing;
1457 } elsif ($_minimize_url) {
1458 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1459 $existing = find_existing_remote($min_url, $r);
1460 if ($existing) {
1461 unless ($no_write) {
1462 print STDERR "Using existing ",
1463 "[svn-remote \"$existing\"]\n";
1464 }
1465 $self->{repo_id} = $existing;
1466 }
1467 if ($min_url ne $url) {
1468 unless ($no_write) {
1469 print STDERR "Using higher level of URL: ",
1470 "$url => $min_url\n";
1471 }
1472 my $old_path = $self->{path};
1473 $self->{path} = $url;
1474 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1475 if (length $old_path) {
1476 $self->{path} .= "/$old_path";
1477 }
1478 $url = $min_url;
1479 }
1480 }
1481 my $orig_url;
1482 if (!$existing) {
1483 # verify that we aren't overwriting anything:
1484 $orig_url = eval {
1485 command_oneline('config', '--get',
1486 "svn-remote.$self->{repo_id}.url")
1487 };
1488 if ($orig_url && ($orig_url ne $url)) {
1489 die "svn-remote.$self->{repo_id}.url already set: ",
1490 "$orig_url\nwanted to set to: $url\n";
1491 }
1492 }
1493 my ($xrepo_id, $xpath) = find_ref($self->refname);
1494 if (defined $xpath) {
1495 die "svn-remote.$xrepo_id.fetch already set to track ",
1496 "$xpath:refs/remotes/", $self->refname, "\n";
1497 }
1498 unless ($no_write) {
1499 command_noisy('config',
1500 "svn-remote.$self->{repo_id}.url", $url);
1501 $self->{path} =~ s{^/}{};
1502 command_noisy('config', '--add',
1503 "svn-remote.$self->{repo_id}.fetch",
1504 "$self->{path}:".$self->refname);
1505 }
1506 $self->{url} = $url;
1507 }
1508
1509 sub find_by_url { # repos_root and, path are optional
1510 my ($class, $full_url, $repos_root, $path) = @_;
1511
1512 return undef unless defined $full_url;
1513 remove_username($full_url);
1514 remove_username($repos_root) if defined $repos_root;
1515 my $remotes = read_all_remotes();
1516 if (defined $full_url && defined $repos_root && !defined $path) {
1517 $path = $full_url;
1518 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1519 }
1520 foreach my $repo_id (keys %$remotes) {
1521 my $u = $remotes->{$repo_id}->{url} or next;
1522 remove_username($u);
1523 next if defined $repos_root && $repos_root ne $u;
1524
1525 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1526 foreach (qw/branches tags/) {
1527 resolve_local_globs($u, $fetch,
1528 $remotes->{$repo_id}->{$_});
1529 }
1530 my $p = $path;
1531 unless (defined $p) {
1532 $p = $full_url;
1533 $p =~ s#^\Q$u\E(?:/|$)## or next;
1534 }
1535 foreach my $f (keys %$fetch) {
1536 next if $f ne $p;
1537 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1538 }
1539 }
1540 undef;
1541 }
1542
1543 sub init {
1544 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1545 my $self = _new($class, $repo_id, $ref_id, $path);
1546 if (defined $url) {
1547 $self->init_remote_config($url, $no_write);
1548 }
1549 $self;
1550 }
1551
1552 sub find_ref {
1553 my ($ref_id) = @_;
1554 foreach (command(qw/config -l/)) {
1555 next unless m!^svn-remote\.(.+)\.fetch=
1556 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1557 my ($repo_id, $path, $ref) = ($1, $2, $3);
1558 if ($ref eq $ref_id) {
1559 $path = '' if ($path =~ m#^\./?#);
1560 return ($repo_id, $path);
1561 }
1562 }
1563 (undef, undef, undef);
1564 }
1565
1566 sub new {
1567 my ($class, $ref_id, $repo_id, $path) = @_;
1568 if (defined $ref_id && !defined $repo_id && !defined $path) {
1569 ($repo_id, $path) = find_ref($ref_id);
1570 if (!defined $repo_id) {
1571 die "Could not find a \"svn-remote.*.fetch\" key ",
1572 "in the repository configuration matching: ",
1573 "refs/remotes/$ref_id\n";
1574 }
1575 }
1576 my $self = _new($class, $repo_id, $ref_id, $path);
1577 if (!defined $self->{path} || !length $self->{path}) {
1578 my $fetch = command_oneline('config', '--get',
1579 "svn-remote.$repo_id.fetch",
1580 ":refs/remotes/$ref_id\$") or
1581 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1582 "\":refs/remotes/$ref_id\$\" in config\n";
1583 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1584 }
1585 $self->{url} = command_oneline('config', '--get',
1586 "svn-remote.$repo_id.url") or
1587 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1588 $self->rebuild;
1589 $self;
1590 }
1591
1592 sub refname {
1593 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1594
1595 # It cannot end with a slash /, we'll throw up on this because
1596 # SVN can't have directories with a slash in their name, either:
1597 if ($refname =~ m{/$}) {
1598 die "ref: '$refname' ends with a trailing slash, this is ",
1599 "not permitted by git nor Subversion\n";
1600 }
1601
1602 # It cannot have ASCII control character space, tilde ~, caret ^,
1603 # colon :, question-mark ?, asterisk *, space, or open bracket [
1604 # anywhere.
1605 #
1606 # Additionally, % must be escaped because it is used for escaping
1607 # and we want our escaped refname to be reversible
1608 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1609
1610 # no slash-separated component can begin with a dot .
1611 # /.* becomes /%2E*
1612 $refname =~ s{/\.}{/%2E}g;
1613
1614 # It cannot have two consecutive dots .. anywhere
1615 # .. becomes %2E%2E
1616 $refname =~ s{\.\.}{%2E%2E}g;
1617
1618 return $refname;
1619 }
1620
1621 sub desanitize_refname {
1622 my ($refname) = @_;
1623 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1624 return $refname;
1625 }
1626
1627 sub svm_uuid {
1628 my ($self) = @_;
1629 return $self->{svm}->{uuid} if $self->svm;
1630 $self->ra;
1631 unless ($self->{svm}) {
1632 die "SVM UUID not cached, and reading remotely failed\n";
1633 }
1634 $self->{svm}->{uuid};
1635 }
1636
1637 sub svm {
1638 my ($self) = @_;
1639 return $self->{svm} if $self->{svm};
1640 my $svm;
1641 # see if we have it in our config, first:
1642 eval {
1643 my $section = "svn-remote.$self->{repo_id}";
1644 $svm = {
1645 source => tmp_config('--get', "$section.svm-source"),
1646 uuid => tmp_config('--get', "$section.svm-uuid"),
1647 replace => tmp_config('--get', "$section.svm-replace"),
1648 }
1649 };
1650 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1651 $self->{svm} = $svm;
1652 }
1653 $self->{svm};
1654 }
1655
1656 sub _set_svm_vars {
1657 my ($self, $ra) = @_;
1658 return $ra if $self->svm;
1659
1660 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1661 "(svm:source, svm:uuid) ",
1662 "from the following URLs:\n" );
1663 sub read_svm_props {
1664 my ($self, $ra, $path, $r) = @_;
1665 my $props = ($ra->get_dir($path, $r))[2];
1666 my $src = $props->{'svm:source'};
1667 my $uuid = $props->{'svm:uuid'};
1668 return undef if (!$src || !$uuid);
1669
1670 chomp($src, $uuid);
1671
1672 $uuid =~ m{^[0-9a-f\-]{30,}$}
1673 or die "doesn't look right - svm:uuid is '$uuid'\n";
1674
1675 # the '!' is used to mark the repos_root!/relative/path
1676 $src =~ s{/?!/?}{/};
1677 $src =~ s{/+$}{}; # no trailing slashes please
1678 # username is of no interest
1679 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1680
1681 my $replace = $ra->{url};
1682 $replace .= "/$path" if length $path;
1683
1684 my $section = "svn-remote.$self->{repo_id}";
1685 tmp_config("$section.svm-source", $src);
1686 tmp_config("$section.svm-replace", $replace);
1687 tmp_config("$section.svm-uuid", $uuid);
1688 $self->{svm} = {
1689 source => $src,
1690 uuid => $uuid,
1691 replace => $replace
1692 };
1693 }
1694
1695 my $r = $ra->get_latest_revnum;
1696 my $path = $self->{path};
1697 my %tried;
1698 while (length $path) {
1699 unless ($tried{"$self->{url}/$path"}) {
1700 return $ra if $self->read_svm_props($ra, $path, $r);
1701 $tried{"$self->{url}/$path"} = 1;
1702 }
1703 $path =~ s#/?[^/]+$##;
1704 }
1705 die "Path: '$path' should be ''\n" if $path ne '';
1706 return $ra if $self->read_svm_props($ra, $path, $r);
1707 $tried{"$self->{url}/$path"} = 1;
1708
1709 if ($ra->{repos_root} eq $self->{url}) {
1710 die @err, (map { " $_\n" } keys %tried), "\n";
1711 }
1712
1713 # nope, make sure we're connected to the repository root:
1714 my $ok;
1715 my @tried_b;
1716 $path = $ra->{svn_path};
1717 $ra = Git::SVN::Ra->new($ra->{repos_root});
1718 while (length $path) {
1719 unless ($tried{"$ra->{url}/$path"}) {
1720 $ok = $self->read_svm_props($ra, $path, $r);
1721 last if $ok;
1722 $tried{"$ra->{url}/$path"} = 1;
1723 }
1724 $path =~ s#/?[^/]+$##;
1725 }
1726 die "Path: '$path' should be ''\n" if $path ne '';
1727 $ok ||= $self->read_svm_props($ra, $path, $r);
1728 $tried{"$ra->{url}/$path"} = 1;
1729 if (!$ok) {
1730 die @err, (map { " $_\n" } keys %tried), "\n";
1731 }
1732 Git::SVN::Ra->new($self->{url});
1733 }
1734
1735 sub svnsync {
1736 my ($self) = @_;
1737 return $self->{svnsync} if $self->{svnsync};
1738
1739 if ($self->no_metadata) {
1740 die "Can't have both 'noMetadata' and ",
1741 "'useSvnsyncProps' options set!\n";
1742 }
1743 if ($self->rewrite_root) {
1744 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1745 "options set!\n";
1746 }
1747
1748 my $svnsync;
1749 # see if we have it in our config, first:
1750 eval {
1751 my $section = "svn-remote.$self->{repo_id}";
1752 $svnsync = {
1753 url => tmp_config('--get', "$section.svnsync-url"),
1754 uuid => tmp_config('--get', "$section.svnsync-uuid"),
1755 }
1756 };
1757 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1758 return $self->{svnsync} = $svnsync;
1759 }
1760
1761 my $err = "useSvnsyncProps set, but failed to read " .
1762 "svnsync property: svn:sync-from-";
1763 my $rp = $self->ra->rev_proplist(0);
1764
1765 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1766 $url =~ m{^[a-z\+]+://} or
1767 die "doesn't look right - svn:sync-from-url is '$url'\n";
1768
1769 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1770 $uuid =~ m{^[0-9a-f\-]{30,}$} or
1771 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1772
1773 my $section = "svn-remote.$self->{repo_id}";
1774 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1775 tmp_config('--add', "$section.svnsync-url", $url);
1776 return $self->{svnsync} = { url => $url, uuid => $uuid };
1777 }
1778
1779 # this allows us to memoize our SVN::Ra UUID locally and avoid a
1780 # remote lookup (useful for 'git svn log').
1781 sub ra_uuid {
1782 my ($self) = @_;
1783 unless ($self->{ra_uuid}) {
1784 my $key = "svn-remote.$self->{repo_id}.uuid";
1785 my $uuid = eval { tmp_config('--get', $key) };
1786 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1787 $self->{ra_uuid} = $uuid;
1788 } else {
1789 die "ra_uuid called without URL\n" unless $self->{url};
1790 $self->{ra_uuid} = $self->ra->get_uuid;
1791 tmp_config('--add', $key, $self->{ra_uuid});
1792 }
1793 }
1794 $self->{ra_uuid};
1795 }
1796
1797 sub _set_repos_root {
1798 my ($self, $repos_root) = @_;
1799 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1800 $repos_root ||= $self->ra->{repos_root};
1801 tmp_config($k, $repos_root);
1802 $repos_root;
1803 }
1804
1805 sub repos_root {
1806 my ($self) = @_;
1807 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1808 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1809 }
1810
1811 sub ra {
1812 my ($self) = shift;
1813 my $ra = Git::SVN::Ra->new($self->{url});
1814 $self->_set_repos_root($ra->{repos_root});
1815 if ($self->use_svm_props && !$self->{svm}) {
1816 if ($self->no_metadata) {
1817 die "Can't have both 'noMetadata' and ",
1818 "'useSvmProps' options set!\n";
1819 } elsif ($self->use_svnsync_props) {
1820 die "Can't have both 'useSvnsyncProps' and ",
1821 "'useSvmProps' options set!\n";
1822 }
1823 $ra = $self->_set_svm_vars($ra);
1824 $self->{-want_revprops} = 1;
1825 }
1826 $ra;
1827 }
1828
1829 sub rel_path {
1830 my ($self) = @_;
1831 my $repos_root = $self->ra->{repos_root};
1832 return $self->{path} if ($self->{url} eq $repos_root);
1833 my $url = $self->{url} .
1834 (length $self->{path} ? "/$self->{path}" : $self->{path});
1835 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1836 $url;
1837 }
1838
1839 # prop_walk(PATH, REV, SUB)
1840 # -------------------------
1841 # Recursively traverse PATH at revision REV and invoke SUB for each
1842 # directory that contains a SVN property. SUB will be invoked as
1843 # follows: &SUB(gs, path, props); where `gs' is this instance of
1844 # Git::SVN, `path' the path to the directory where the properties
1845 # `props' were found. The `path' will be relative to point of checkout,
1846 # that is, if url://repo/trunk is the current Git branch, and that
1847 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
1848 # as `path' (note the trailing `/').
1849 sub prop_walk {
1850 my ($self, $path, $rev, $sub) = @_;
1851
1852 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1853 $path =~ s#^/*#/#g;
1854 my $p = $path;
1855 # Strip the irrelevant part of the path.
1856 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1857 # Ensure the path is terminated by a `/'.
1858 $p =~ s#/*$#/#;
1859
1860 # The properties contain all the internal SVN stuff nobody
1861 # (usually) cares about.
1862 my $interesting_props = 0;
1863 foreach (keys %{$props}) {
1864 # If it doesn't start with `svn:', it must be a
1865 # user-defined property.
1866 ++$interesting_props and next if $_ !~ /^svn:/;
1867 # FIXME: Fragile, if SVN adds new public properties,
1868 # this needs to be updated.
1869 ++$interesting_props if /^svn:(?:ignore|keywords|executable
1870 |eol-style|mime-type
1871 |externals|needs-lock)$/x;
1872 }
1873 &$sub($self, $p, $props) if $interesting_props;
1874
1875 foreach (sort keys %$dirent) {
1876 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1877 $self->prop_walk($path . '/' . $_, $rev, $sub);
1878 }
1879 }
1880
1881 sub last_rev { ($_[0]->last_rev_commit)[0] }
1882 sub last_commit { ($_[0]->last_rev_commit)[1] }
1883
1884 # returns the newest SVN revision number and newest commit SHA1
1885 sub last_rev_commit {
1886 my ($self) = @_;
1887 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1888 return ($self->{last_rev}, $self->{last_commit});
1889 }
1890 my $c = ::verify_ref($self->refname.'^0');
1891 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1892 my $rev = (::cmt_metadata($c))[1];
1893 if (defined $rev) {
1894 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1895 return ($rev, $c);
1896 }
1897 }
1898 my $db_path = $self->db_path;
1899 unless (-e $db_path) {
1900 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1901 return (undef, undef);
1902 }
1903 my $offset = -41; # from tail
1904 my $rl;
1905 open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1906 sysseek($fh, $offset, 2); # don't care for errors
1907 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1908 chomp $rl;
1909 while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1910 $offset -= 41;
1911 sysseek($fh, $offset, 2); # don't care for errors
1912 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1913 chomp $rl;
1914 }
1915 if ($c && $c ne $rl) {
1916 die "$db_path and ", $self->refname,
1917 " inconsistent!:\n$c != $rl\n";
1918 }
1919 my $rev = sysseek($fh, 0, 1) or croak $!;
1920 $rev = ($rev - 41) / 41;
1921 close $fh or croak $!;
1922 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1923 return ($rev, $c);
1924 }
1925
1926 sub get_fetch_range {
1927 my ($self, $min, $max) = @_;
1928 $max ||= $self->ra->get_latest_revnum;
1929 $min ||= $self->rev_db_max;
1930 (++$min, $max);
1931 }
1932
1933 sub tmp_config {
1934 my (@args) = @_;
1935 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1936 my $config = "$ENV{GIT_DIR}/svn/.metadata";
1937 if (! -f $config && -f $old_def_config) {
1938 rename $old_def_config, $config or
1939 die "Failed rename $old_def_config => $config: $!\n";
1940 }
1941 my $old_config = $ENV{GIT_CONFIG};
1942 $ENV{GIT_CONFIG} = $config;
1943 $@ = undef;
1944 my @ret = eval {
1945 unless (-f $config) {
1946 mkfile($config);
1947 open my $fh, '>', $config or
1948 die "Can't open $config: $!\n";
1949 print $fh "; This file is used internally by ",
1950 "git-svn\n" or die
1951 "Couldn't write to $config: $!\n";
1952 print $fh "; You should not have to edit it\n" or
1953 die "Couldn't write to $config: $!\n";
1954 close $fh or die "Couldn't close $config: $!\n";
1955 }
1956 command('config', @args);
1957 };
1958 my $err = $@;
1959 if (defined $old_config) {
1960 $ENV{GIT_CONFIG} = $old_config;
1961 } else {
1962 delete $ENV{GIT_CONFIG};
1963 }
1964 die $err if $err;
1965 wantarray ? @ret : $ret[0];
1966 }
1967
1968 sub tmp_index_do {
1969 my ($self, $sub) = @_;
1970 my $old_index = $ENV{GIT_INDEX_FILE};
1971 $ENV{GIT_INDEX_FILE} = $self->{index};
1972 $@ = undef;
1973 my @ret = eval {
1974 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1975 mkpath([$dir]) unless -d $dir;
1976 &$sub;
1977 };
1978 my $err = $@;
1979 if (defined $old_index) {
1980 $ENV{GIT_INDEX_FILE} = $old_index;
1981 } else {
1982 delete $ENV{GIT_INDEX_FILE};
1983 }
1984 die $err if $err;
1985 wantarray ? @ret : $ret[0];
1986 }
1987
1988 sub assert_index_clean {
1989 my ($self, $treeish) = @_;
1990
1991 $self->tmp_index_do(sub {
1992 command_noisy('read-tree', $treeish) unless -e $self->{index};
1993 my $x = command_oneline('write-tree');
1994 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1995 /^tree ($::sha1)/mo);
1996 return if $y eq $x;
1997
1998 warn "Index mismatch: $y != $x\nrereading $treeish\n";
1999 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2000 command_noisy('read-tree', $treeish);
2001 $x = command_oneline('write-tree');
2002 if ($y ne $x) {
2003 ::fatal "trees ($treeish) $y != $x\n",
2004 "Something is seriously wrong...";
2005 }
2006 });
2007 }
2008
2009 sub get_commit_parents {
2010 my ($self, $log_entry) = @_;
2011 my (%seen, @ret, @tmp);
2012 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2013 if (my $ip = $self->{inject_parents}) {
2014 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2015 push @tmp, $commit;
2016 }
2017 }
2018 if (my $cur = ::verify_ref($self->refname.'^0')) {
2019 push @tmp, $cur;
2020 }
2021 if (my $ipd = $self->{inject_parents_dcommit}) {
2022 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2023 push @tmp, @$commit;
2024 }
2025 }
2026 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2027 while (my $p = shift @tmp) {
2028 next if $seen{$p};
2029 $seen{$p} = 1;
2030 push @ret, $p;
2031 # MAXPARENT is defined to 16 in commit-tree.c:
2032 last if @ret >= 16;
2033 }
2034 if (@tmp) {
2035 die "r$log_entry->{revision}: No room for parents:\n\t",
2036 join("\n\t", @tmp), "\n";
2037 }
2038 @ret;
2039 }
2040
2041 sub rewrite_root {
2042 my ($self) = @_;
2043 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2044 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2045 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2046 if ($rwr) {
2047 $rwr =~ s#/+$##;
2048 if ($rwr !~ m#^[a-z\+]+://#) {
2049 die "$rwr is not a valid URL (key: $k)\n";
2050 }
2051 }
2052 $self->{-rewrite_root} = $rwr;
2053 }
2054
2055 sub metadata_url {
2056 my ($self) = @_;
2057 ($self->rewrite_root || $self->{url}) .
2058 (length $self->{path} ? '/' . $self->{path} : '');
2059 }
2060
2061 sub full_url {
2062 my ($self) = @_;
2063 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2064 }
2065
2066 sub do_git_commit {
2067 my ($self, $log_entry) = @_;
2068 my $lr = $self->last_rev;
2069 if (defined $lr && $lr >= $log_entry->{revision}) {
2070 die "Last fetched revision of ", $self->refname,
2071 " was r$lr, but we are about to fetch: ",
2072 "r$log_entry->{revision}!\n";
2073 }
2074 if (my $c = $self->rev_db_get($log_entry->{revision})) {
2075 croak "$log_entry->{revision} = $c already exists! ",
2076 "Why are we refetching it?\n";
2077 }
2078 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2079 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2080 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2081
2082 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2083 ? $log_entry->{commit_name}
2084 : $log_entry->{name};
2085 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2086 ? $log_entry->{commit_email}
2087 : $log_entry->{email};
2088
2089 my $tree = $log_entry->{tree};
2090 if (!defined $tree) {
2091 $tree = $self->tmp_index_do(sub {
2092 command_oneline('write-tree') });
2093 }
2094 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2095
2096 my @exec = ('git-commit-tree', $tree);
2097 foreach ($self->get_commit_parents($log_entry)) {
2098 push @exec, '-p', $_;
2099 }
2100 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2101 or croak $!;
2102 print $msg_fh $log_entry->{log} or croak $!;
2103 unless ($self->no_metadata) {
2104 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2105 or croak $!;
2106 }
2107 $msg_fh->flush == 0 or croak $!;
2108 close $msg_fh or croak $!;
2109 chomp(my $commit = do { local $/; <$out_fh> });
2110 close $out_fh or croak $!;
2111 waitpid $pid, 0;
2112 croak $? if $?;
2113 if ($commit !~ /^$::sha1$/o) {
2114 die "Failed to commit, invalid sha1: $commit\n";
2115 }
2116
2117 $self->rev_db_set($log_entry->{revision}, $commit, 1);
2118
2119 $self->{last_rev} = $log_entry->{revision};
2120 $self->{last_commit} = $commit;
2121 print "r$log_entry->{revision}";
2122 if (defined $log_entry->{svm_revision}) {
2123 print " (\@$log_entry->{svm_revision})";
2124 $self->rev_db_set($log_entry->{svm_revision}, $commit,
2125 0, $self->svm_uuid);
2126 }
2127 print " = $commit ($self->{ref_id})\n";
2128 if (defined $_repack && (--$_repack_nr == 0)) {
2129 $_repack_nr = $_repack;
2130 # repack doesn't use any arguments with spaces in them, does it?
2131 print "Running git repack $_repack_flags ...\n";
2132 command_noisy('repack', split(/\s+/, $_repack_flags));
2133 print "Done repacking\n";
2134 }
2135 return $commit;
2136 }
2137
2138 sub match_paths {
2139 my ($self, $paths, $r) = @_;
2140 return 1 if $self->{path} eq '';
2141 if (my $path = $paths->{"/$self->{path}"}) {
2142 return ($path->{action} eq 'D') ? 0 : 1;
2143 }
2144 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2145 if (grep /$self->{path_regex}/, keys %$paths) {
2146 return 1;
2147 }
2148 my $c = '';
2149 foreach (split m#/#, $self->{path}) {
2150 $c .= "/$_";
2151 next unless ($paths->{$c} &&
2152 ($paths->{$c}->{action} =~ /^[AR]$/));
2153 if ($self->ra->check_path($self->{path}, $r) ==
2154 $SVN::Node::dir) {
2155 return 1;
2156 }
2157 }
2158 return 0;
2159 }
2160
2161 sub find_parent_branch {
2162 my ($self, $paths, $rev) = @_;
2163 return undef unless $self->follow_parent;
2164 unless (defined $paths) {
2165 my $err_handler = $SVN::Error::handler;
2166 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2167 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2168 $paths =
2169 Git::SVN::Ra::dup_changed_paths($_[0]) });
2170 $SVN::Error::handler = $err_handler;
2171 }
2172 return undef unless defined $paths;
2173
2174 # look for a parent from another branch:
2175 my @b_path_components = split m#/#, $self->rel_path;
2176 my @a_path_components;
2177 my $i;
2178 while (@b_path_components) {
2179 $i = $paths->{'/'.join('/', @b_path_components)};
2180 last if $i && defined $i->{copyfrom_path};
2181 unshift(@a_path_components, pop(@b_path_components));
2182 }
2183 return undef unless defined $i && defined $i->{copyfrom_path};
2184 my $branch_from = $i->{copyfrom_path};
2185 if (@a_path_components) {
2186 print STDERR "branch_from: $branch_from => ";
2187 $branch_from .= '/'.join('/', @a_path_components);
2188 print STDERR $branch_from, "\n";
2189 }
2190 my $r = $i->{copyfrom_rev};
2191 my $repos_root = $self->ra->{repos_root};
2192 my $url = $self->ra->{url};
2193 my $new_url = $repos_root . $branch_from;
2194 print STDERR "Found possible branch point: ",
2195 "$new_url => ", $self->full_url, ", $r\n";
2196 $branch_from =~ s#^/##;
2197 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2198 unless ($gs) {
2199 my $ref_id = $self->{ref_id};
2200 $ref_id =~ s/\@\d+$//;
2201 $ref_id .= "\@$r";
2202 # just grow a tail if we're not unique enough :x
2203 $ref_id .= '-' while find_ref($ref_id);
2204 print STDERR "Initializing parent: $ref_id\n";
2205 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
2206 }
2207 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2208 if (!defined $r0 || !defined $parent) {
2209 my ($base, $head) = parse_revision_argument(0, $r);
2210 if ($base <= $r) {
2211 $gs->fetch($base, $r);
2212 }
2213 ($r0, $parent) = $gs->last_rev_commit;
2214 }
2215 if (defined $r0 && defined $parent) {
2216 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2217 my $ed;
2218 if ($self->ra->can_do_switch) {
2219 $self->assert_index_clean($parent);
2220 print STDERR "Following parent with do_switch\n";
2221 # do_switch works with svn/trunk >= r22312, but that
2222 # is not included with SVN 1.4.3 (the latest version
2223 # at the moment), so we can't rely on it
2224 $self->{last_commit} = $parent;
2225 $ed = SVN::Git::Fetcher->new($self);
2226 $gs->ra->gs_do_switch($r0, $rev, $gs,
2227 $self->full_url, $ed)
2228 or die "SVN connection failed somewhere...\n";
2229 } elsif ($self->ra->trees_match($new_url, $r0,
2230 $self->full_url, $rev)) {
2231 print STDERR "Trees match:\n",
2232 " $new_url\@$r0\n",
2233 " ${\$self->full_url}\@$rev\n",
2234 "Following parent with no changes\n";
2235 $self->tmp_index_do(sub {
2236 command_noisy('read-tree', $parent);
2237 });
2238 $self->{last_commit} = $parent;
2239 } else {
2240 print STDERR "Following parent with do_update\n";
2241 $ed = SVN::Git::Fetcher->new($self);
2242 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2243 or die "SVN connection failed somewhere...\n";
2244 }
2245 print STDERR "Successfully followed parent\n";
2246 return $self->make_log_entry($rev, [$parent], $ed);
2247 }
2248 return undef;
2249 }
2250
2251 sub do_fetch {
2252 my ($self, $paths, $rev) = @_;
2253 my $ed;
2254 my ($last_rev, @parents);
2255 if (my $lc = $self->last_commit) {
2256 # we can have a branch that was deleted, then re-added
2257 # under the same name but copied from another path, in
2258 # which case we'll have multiple parents (we don't
2259 # want to break the original ref, nor lose copypath info):
2260 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2261 push @{$log_entry->{parents}}, $lc;
2262 return $log_entry;
2263 }
2264 $ed = SVN::Git::Fetcher->new($self);
2265 $last_rev = $self->{last_rev};
2266 $ed->{c} = $lc;
2267 @parents = ($lc);
2268 } else {
2269 $last_rev = $rev;
2270 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2271 return $log_entry;
2272 }
2273 $ed = SVN::Git::Fetcher->new($self);
2274 }
2275 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2276 die "SVN connection failed somewhere...\n";
2277 }
2278 $self->make_log_entry($rev, \@parents, $ed);
2279 }
2280
2281 sub get_untracked {
2282 my ($self, $ed) = @_;
2283 my @out;
2284 my $h = $ed->{empty};
2285 foreach (sort keys %$h) {
2286 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2287 push @out, " $act: " . uri_encode($_);
2288 warn "W: $act: $_\n";
2289 }
2290 foreach my $t (qw/dir_prop file_prop/) {
2291 $h = $ed->{$t} or next;
2292 foreach my $path (sort keys %$h) {
2293 my $ppath = $path eq '' ? '.' : $path;
2294 foreach my $prop (sort keys %{$h->{$path}}) {
2295 next if $SKIP_PROP{$prop};
2296 my $v = $h->{$path}->{$prop};
2297 my $t_ppath_prop = "$t: " .
2298 uri_encode($ppath) . ' ' .
2299 uri_encode($prop);
2300 if (defined $v) {
2301 push @out, " +$t_ppath_prop " .
2302 uri_encode($v);
2303 } else {
2304 push @out, " -$t_ppath_prop";
2305 }
2306 }
2307 }
2308 }
2309 foreach my $t (qw/absent_file absent_directory/) {
2310 $h = $ed->{$t} or next;
2311 foreach my $parent (sort keys %$h) {
2312 foreach my $path (sort @{$h->{$parent}}) {
2313 push @out, " $t: " .
2314 uri_encode("$parent/$path");
2315 warn "W: $t: $parent/$path ",
2316 "Insufficient permissions?\n";
2317 }
2318 }
2319 }
2320 \@out;
2321 }
2322
2323 sub parse_svn_date {
2324 my $date = shift || return '+0000 1970-01-01 00:00:00';
2325 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2326 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2327 croak "Unable to parse date: $date\n";
2328 "+0000 $Y-$m-$d $H:$M:$S";
2329 }
2330
2331 sub check_author {
2332 my ($author) = @_;
2333 if (!defined $author || length $author == 0) {
2334 $author = '(no author)';
2335 }
2336 if (defined $::_authors && ! defined $::users{$author}) {
2337 die "Author: $author not defined in $::_authors file\n";
2338 }
2339 $author;
2340 }
2341
2342 sub make_log_entry {
2343 my ($self, $rev, $parents, $ed) = @_;
2344 my $untracked = $self->get_untracked($ed);
2345
2346 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2347 print $un "r$rev\n" or croak $!;
2348 print $un $_, "\n" foreach @$untracked;
2349 my %log_entry = ( parents => $parents || [], revision => $rev,
2350 log => '');
2351
2352 my $headrev;
2353 my $logged = delete $self->{logged_rev_props};
2354 if (!$logged || $self->{-want_revprops}) {
2355 my $rp = $self->ra->rev_proplist($rev);
2356 foreach (sort keys %$rp) {
2357 my $v = $rp->{$_};
2358 if (/^svn:(author|date|log)$/) {
2359 $log_entry{$1} = $v;
2360 } elsif ($_ eq 'svm:headrev') {
2361 $headrev = $v;
2362 } else {
2363 print $un " rev_prop: ", uri_encode($_), ' ',
2364 uri_encode($v), "\n";
2365 }
2366 }
2367 } else {
2368 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2369 }
2370 close $un or croak $!;
2371
2372 $log_entry{date} = parse_svn_date($log_entry{date});
2373 $log_entry{log} .= "\n";
2374 my $author = $log_entry{author} = check_author($log_entry{author});
2375 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2376 : ($author, undef);
2377
2378 my ($commit_name, $commit_email) = ($name, $email);
2379 if ($_use_log_author) {
2380 if ($log_entry{log} =~ /From:\s+(.*?)\s+<(.*)>\s*\n/) {
2381 ($name, $email) = ($1, $2);
2382 } elsif ($log_entry{log} =~
2383 /Signed-off-by:\s+(.*?)\s+<(.*)>\s*\n/) {
2384 ($name, $email) = ($1, $2);
2385 }
2386 }
2387 if (defined $headrev && $self->use_svm_props) {
2388 if ($self->rewrite_root) {
2389 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2390 "options set!\n";
2391 }
2392 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2393 # we don't want "SVM: initializing mirror for junk" ...
2394 return undef if $r == 0;
2395 my $svm = $self->svm;
2396 if ($uuid ne $svm->{uuid}) {
2397 die "UUID mismatch on SVM path:\n",
2398 "expected: $svm->{uuid}\n",
2399 " got: $uuid\n";
2400 }
2401 my $full_url = $self->full_url;
2402 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2403 die "Failed to replace '$svm->{replace}' with ",
2404 "'$svm->{source}' in $full_url\n";
2405 # throw away username for storing in records
2406 remove_username($full_url);
2407 $log_entry{metadata} = "$full_url\@$r $uuid";
2408 $log_entry{svm_revision} = $r;
2409 $email ||= "$author\@$uuid";
2410 $commit_email ||= "$author\@$uuid";
2411 } elsif ($self->use_svnsync_props) {
2412 my $full_url = $self->svnsync->{url};
2413 $full_url .= "/$self->{path}" if length $self->{path};
2414 remove_username($full_url);
2415 my $uuid = $self->svnsync->{uuid};
2416 $log_entry{metadata} = "$full_url\@$rev $uuid";
2417 $email ||= "$author\@$uuid";
2418 $commit_email ||= "$author\@$uuid";
2419 } else {
2420 my $url = $self->metadata_url;
2421 remove_username($url);
2422 $log_entry{metadata} = "$url\@$rev " .
2423 $self->ra->get_uuid;
2424 $email ||= "$author\@" . $self->ra->get_uuid;
2425 $commit_email ||= "$author\@" . $self->ra->get_uuid;
2426 }
2427 $log_entry{name} = $name;
2428 $log_entry{email} = $email;
2429 $log_entry{commit_name} = $commit_name;
2430 $log_entry{commit_email} = $commit_email;
2431 \%log_entry;
2432 }
2433
2434 sub fetch {
2435 my ($self, $min_rev, $max_rev, @parents) = @_;
2436 my ($last_rev, $last_commit) = $self->last_rev_commit;
2437 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2438 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2439 }
2440
2441 sub set_tree_cb {
2442 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2443 $self->{inject_parents} = { $rev => $tree };
2444 $self->fetch(undef, undef);
2445 }
2446
2447 sub set_tree {
2448 my ($self, $tree) = (shift, shift);
2449 my $log_entry = ::get_commit_entry($tree);
2450 unless ($self->{last_rev}) {
2451 fatal("Must have an existing revision to commit");
2452 }
2453 my %ed_opts = ( r => $self->{last_rev},
2454 log => $log_entry->{log},
2455 ra => $self->ra,
2456 tree_a => $self->{last_commit},
2457 tree_b => $tree,
2458 editor_cb => sub {
2459 $self->set_tree_cb($log_entry, $tree, @_) },
2460 svn_path => $self->{path} );
2461 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2462 print "No changes\nr$self->{last_rev} = $tree\n";
2463 }
2464 }
2465
2466 sub rebuild {
2467 my ($self) = @_;
2468 my $db_path = $self->db_path;
2469 return if (-e $db_path && ! -z $db_path);
2470 return unless ::verify_ref($self->refname.'^0');
2471 if (-f $self->{db_root}) {
2472 rename $self->{db_root}, $db_path or die
2473 "rename $self->{db_root} => $db_path failed: $!\n";
2474 my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
2475 symlink $base, $self->{db_root} or die
2476 "symlink $base => $self->{db_root} failed: $!\n";
2477 return;
2478 }
2479 print "Rebuilding $db_path ...\n";
2480 my ($log, $ctx) = command_output_pipe("log", '--no-color', $self->refname);
2481 my $latest;
2482 my $full_url = $self->full_url;
2483 remove_username($full_url);
2484 my $svn_uuid;
2485 my $c;
2486 while (<$log>) {
2487 if ( m{^commit ($::sha1)$} ) {
2488 $c = $1;
2489 next;
2490 }
2491 next unless s{^\s*(git-svn-id:)}{$1};
2492 my ($url, $rev, $uuid) = ::extract_metadata($_);
2493 remove_username($url);
2494
2495 # ignore merges (from set-tree)
2496 next if (!defined $rev || !$uuid);
2497
2498 # if we merged or otherwise started elsewhere, this is
2499 # how we break out of it
2500 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
2501 ($full_url && $url && ($url ne $full_url))) {
2502 next;
2503 }
2504 $latest ||= $rev;
2505 $svn_uuid ||= $uuid;
2506
2507 $self->rev_db_set($rev, $c);
2508 print "r$rev = $c\n";
2509 }
2510 command_close_pipe($log, $ctx);
2511 print "Done rebuilding $db_path\n";
2512 }
2513
2514 # rev_db:
2515 # Tie::File seems to be prone to offset errors if revisions get sparse,
2516 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
2517 # one of my favorite modules is out :< Next up would be one of the DBM
2518 # modules, but I'm not sure which is most portable... So I'll just
2519 # go with something that's plain-text, but still capable of
2520 # being randomly accessed. So here's my ultra-simple fixed-width
2521 # database. All records are 40 characters + "\n", so it's easy to seek
2522 # to a revision: (41 * rev) is the byte offset.
2523 # A record of 40 0s denotes an empty revision.
2524 # And yes, it's still pretty fast (faster than Tie::File).
2525 # These files are disposable unless noMetadata or useSvmProps is set
2526
2527 sub _rev_db_set {
2528 my ($fh, $rev, $commit) = @_;
2529 my $offset = $rev * 41;
2530 # assume that append is the common case:
2531 seek $fh, 0, 2 or croak $!;
2532 my $pos = tell $fh;
2533 if ($pos < $offset) {
2534 for (1 .. (($offset - $pos) / 41)) {
2535 print $fh (('0' x 40),"\n") or croak $!;
2536 }
2537 }
2538 seek $fh, $offset, 0 or croak $!;
2539 print $fh $commit,"\n" or croak $!;
2540 }
2541
2542 sub mkfile {
2543 my ($path) = @_;
2544 unless (-e $path) {
2545 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2546 mkpath([$dir]) unless -d $dir;
2547 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2548 close $fh or die "Couldn't close (create) $path: $!\n";
2549 }
2550 }
2551
2552 sub rev_db_set {
2553 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2554 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2555 my $db = $self->db_path($uuid);
2556 my $db_lock = "$db.lock";
2557 my $sig;
2558 if ($update_ref) {
2559 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2560 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2561 }
2562 mkfile($db);
2563
2564 $LOCKFILES{$db_lock} = 1;
2565 my $sync;
2566 # both of these options make our .rev_db file very, very important
2567 # and we can't afford to lose it because rebuild() won't work
2568 if ($self->use_svm_props || $self->no_metadata) {
2569 $sync = 1;
2570 copy($db, $db_lock) or die "rev_db_set(@_): ",
2571 "Failed to copy: ",
2572 "$db => $db_lock ($!)\n";
2573 } else {
2574 rename $db, $db_lock or die "rev_db_set(@_): ",
2575 "Failed to rename: ",
2576 "$db => $db_lock ($!)\n";
2577 }
2578 open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
2579 _rev_db_set($fh, $rev, $commit);
2580 if ($sync) {
2581 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2582 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2583 }
2584 close $fh or croak $!;
2585 if ($update_ref) {
2586 $_head = $self;
2587 command_noisy('update-ref', '-m', "r$rev",
2588 $self->refname, $commit);
2589 }
2590 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
2591 "$db_lock => $db ($!)\n";
2592 delete $LOCKFILES{$db_lock};
2593 if ($update_ref) {
2594 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2595 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2596 kill $sig, $$ if defined $sig;
2597 }
2598 }
2599
2600 sub rev_db_max {
2601 my ($self) = @_;
2602 $self->rebuild;
2603 my $db_path = $self->db_path;
2604 my @stat = stat $db_path or return 0;
2605 ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
2606 my $max = $stat[7] / 41;
2607 (($max > 0) ? $max - 1 : 0);
2608 }
2609
2610 sub rev_db_get {
2611 my ($self, $rev, $uuid) = @_;
2612 my $ret;
2613 my $offset = $rev * 41;
2614 my $db_path = $self->db_path($uuid);
2615 return undef unless -e $db_path;
2616 open my $fh, '<', $db_path or croak $!;
2617 if (sysseek($fh, $offset, 0) == $offset) {
2618 my $read = sysread($fh, $ret, 40);
2619 $ret = undef if ($read != 40 || $ret eq ('0'x40));
2620 }
2621 close $fh or croak $!;
2622 $ret;
2623 }
2624
2625 # Finds the first svn revision that exists on (if $eq_ok is true) or
2626 # before $rev for the current branch. It will not search any lower
2627 # than $min_rev. Returns the git commit hash and svn revision number
2628 # if found, else (undef, undef).
2629 sub find_rev_before {
2630 my ($self, $rev, $eq_ok, $min_rev) = @_;
2631 --$rev unless $eq_ok;
2632 $min_rev ||= 1;
2633 while ($rev >= $min_rev) {
2634 if (my $c = $self->rev_db_get($rev)) {
2635 return ($rev, $c);
2636 }
2637 --$rev;
2638 }
2639 return (undef, undef);
2640 }
2641
2642 # Finds the first svn revision that exists on (if $eq_ok is true) or
2643 # after $rev for the current branch. It will not search any higher
2644 # than $max_rev. Returns the git commit hash and svn revision number
2645 # if found, else (undef, undef).
2646 sub find_rev_after {
2647 my ($self, $rev, $eq_ok, $max_rev) = @_;
2648 ++$rev unless $eq_ok;
2649 $max_rev ||= $self->rev_db_max();
2650 while ($rev <= $max_rev) {
2651 if (my $c = $self->rev_db_get($rev)) {
2652 return ($rev, $c);
2653 }
2654 ++$rev;
2655 }
2656 return (undef, undef);
2657 }
2658
2659 sub _new {
2660 my ($class, $repo_id, $ref_id, $path) = @_;
2661 unless (defined $repo_id && length $repo_id) {
2662 $repo_id = $Git::SVN::default_repo_id;
2663 }
2664 unless (defined $ref_id && length $ref_id) {
2665 $_[2] = $ref_id = $Git::SVN::default_ref_id;
2666 }
2667 $_[1] = $repo_id = sanitize_remote_name($repo_id);
2668 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2669 $_[3] = $path = '' unless (defined $path);
2670 mkpath(["$ENV{GIT_DIR}/svn"]);
2671 bless {
2672 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2673 path => $path, config => "$ENV{GIT_DIR}/svn/config",
2674 db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2675 }
2676
2677 sub db_path {
2678 my ($self, $uuid) = @_;
2679 $uuid ||= $self->ra_uuid;
2680 "$self->{db_root}.$uuid";
2681 }
2682
2683 sub uri_encode {
2684 my ($f) = @_;
2685 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2686 $f
2687 }
2688
2689 sub remove_username {
2690 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2691 }
2692
2693 package Git::SVN::Prompt;
2694 use strict;
2695 use warnings;
2696 require SVN::Core;
2697 use vars qw/$_no_auth_cache $_username/;
2698
2699 sub simple {
2700 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2701 $may_save = undef if $_no_auth_cache;
2702 $default_username = $_username if defined $_username;
2703 if (defined $default_username && length $default_username) {
2704 if (defined $realm && length $realm) {
2705 print STDERR "Authentication realm: $realm\n";
2706 STDERR->flush;
2707 }
2708 $cred->username($default_username);
2709 } else {
2710 username($cred, $realm, $may_save, $pool);
2711 }
2712 $cred->password(_read_password("Password for '" .
2713 $cred->username . "': ", $realm));
2714 $cred->may_save($may_save);
2715 $SVN::_Core::SVN_NO_ERROR;
2716 }
2717
2718 sub ssl_server_trust {
2719 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2720 $may_save = undef if $_no_auth_cache;
2721 print STDERR "Error validating server certificate for '$realm':\n";
2722 {
2723 no warnings 'once';
2724 # All variables SVN::Auth::SSL::* are used only once,
2725 # so we're shutting up Perl warnings about this.
2726 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2727 print STDERR " - The certificate is not issued ",
2728 "by a trusted authority. Use the\n",
2729 " fingerprint to validate ",
2730 "the certificate manually!\n";
2731 }
2732 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2733 print STDERR " - The certificate hostname ",
2734 "does not match.\n";
2735 }
2736 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2737 print STDERR " - The certificate is not yet valid.\n";
2738 }
2739 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2740 print STDERR " - The certificate has expired.\n";
2741 }
2742 if ($failures & $SVN::Auth::SSL::OTHER) {
2743 print STDERR " - The certificate has ",
2744 "an unknown error.\n";
2745 }
2746 } # no warnings 'once'
2747 printf STDERR
2748 "Certificate information:\n".
2749 " - Hostname: %s\n".
2750 " - Valid: from %s until %s\n".
2751 " - Issuer: %s\n".
2752 " - Fingerprint: %s\n",
2753 map $cert_info->$_, qw(hostname valid_from valid_until
2754 issuer_dname fingerprint);
2755 my $choice;
2756 prompt:
2757 print STDERR $may_save ?
2758 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2759 "(R)eject or accept (t)emporarily? ";
2760 STDERR->flush;
2761 $choice = lc(substr(<STDIN> || 'R', 0, 1));
2762 if ($choice =~ /^t$/i) {
2763 $cred->may_save(undef);
2764 } elsif ($choice =~ /^r$/i) {
2765 return -1;
2766 } elsif ($may_save && $choice =~ /^p$/i) {
2767 $cred->may_save($may_save);
2768 } else {
2769 goto prompt;
2770 }
2771 $cred->accepted_failures($failures);
2772 $SVN::_Core::SVN_NO_ERROR;
2773 }
2774
2775 sub ssl_client_cert {
2776 my ($cred, $realm, $may_save, $pool) = @_;
2777 $may_save = undef if $_no_auth_cache;
2778 print STDERR "Client certificate filename: ";
2779 STDERR->flush;
2780 chomp(my $filename = <STDIN>);
2781 $cred->cert_file($filename);
2782 $cred->may_save($may_save);
2783 $SVN::_Core::SVN_NO_ERROR;
2784 }
2785
2786 sub ssl_client_cert_pw {
2787 my ($cred, $realm, $may_save, $pool) = @_;
2788 $may_save = undef if $_no_auth_cache;
2789 $cred->password(_read_password("Password: ", $realm));
2790 $cred->may_save($may_save);
2791 $SVN::_Core::SVN_NO_ERROR;
2792 }
2793
2794 sub username {
2795 my ($cred, $realm, $may_save, $pool) = @_;
2796 $may_save = undef if $_no_auth_cache;
2797 if (defined $realm && length $realm) {
2798 print STDERR "Authentication realm: $realm\n";
2799 }
2800 my $username;
2801 if (defined $_username) {
2802 $username = $_username;
2803 } else {
2804 print STDERR "Username: ";
2805 STDERR->flush;
2806 chomp($username = <STDIN>);
2807 }
2808 $cred->username($username);
2809 $cred->may_save($may_save);
2810 $SVN::_Core::SVN_NO_ERROR;
2811 }
2812
2813 sub _read_password {
2814 my ($prompt, $realm) = @_;
2815 print STDERR $prompt;
2816 STDERR->flush;
2817 require Term::ReadKey;
2818 Term::ReadKey::ReadMode('noecho');
2819 my $password = '';
2820 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2821 last if $key =~ /[\012\015]/; # \n\r
2822 $password .= $key;
2823 }
2824 Term::ReadKey::ReadMode('restore');
2825 print STDERR "\n";
2826 STDERR->flush;
2827 $password;
2828 }
2829
2830 package SVN::Git::Fetcher;
2831 use vars qw/@ISA/;
2832 use strict;
2833 use warnings;
2834 use Carp qw/croak/;
2835 use IO::File qw//;
2836
2837 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
2838 sub new {
2839 my ($class, $git_svn) = @_;
2840 my $self = SVN::Delta::Editor->new;
2841 bless $self, $class;
2842 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2843 $self->{empty} = {};
2844 $self->{dir_prop} = {};
2845 $self->{file_prop} = {};
2846 $self->{absent_dir} = {};
2847 $self->{absent_file} = {};
2848 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2849 $self;
2850 }
2851
2852 sub set_path_strip {
2853 my ($self, $path) = @_;
2854 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2855 }
2856
2857 sub open_root {
2858 { path => '' };
2859 }
2860
2861 sub open_directory {
2862 my ($self, $path, $pb, $rev) = @_;
2863 { path => $path };
2864 }
2865
2866 sub git_path {
2867 my ($self, $path) = @_;
2868 if ($self->{path_strip}) {
2869 $path =~ s!$self->{path_strip}!! or
2870 die "Failed to strip path '$path' ($self->{path_strip})\n";
2871 }
2872 $path;
2873 }
2874
2875 sub delete_entry {
2876 my ($self, $path, $rev, $pb) = @_;
2877
2878 my $gpath = $self->git_path($path);
2879 return undef if ($gpath eq '');
2880
2881 # remove entire directories.
2882 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2883 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2884 -r --name-only -z/,
2885 $self->{c}, '--', $gpath);
2886 local $/ = "\0";
2887 while (<$ls>) {
2888 chomp;
2889 $self->{gii}->remove($_);
2890 print "\tD\t$_\n" unless $::_q;
2891 }
2892 print "\tD\t$gpath/\n" unless $::_q;
2893 command_close_pipe($ls, $ctx);
2894 $self->{empty}->{$path} = 0
2895 } else {
2896 $self->{gii}->remove($gpath);
2897 print "\tD\t$gpath\n" unless $::_q;
2898 }
2899 undef;
2900 }
2901
2902 sub open_file {
2903 my ($self, $path, $pb, $rev) = @_;
2904 my $gpath = $self->git_path($path);
2905 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2906 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2907 unless (defined $mode && defined $blob) {
2908 die "$path was not found in commit $self->{c} (r$rev)\n";
2909 }
2910 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2911 pool => SVN::Pool->new, action => 'M' };
2912 }
2913
2914 sub add_file {
2915 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2916 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2917 delete $self->{empty}->{$dir};
2918 { path => $path, mode_a => 100644, mode_b => 100644,
2919 pool => SVN::Pool->new, action => 'A' };
2920 }
2921
2922 sub add_directory {
2923 my ($self, $path, $cp_path, $cp_rev) = @_;
2924 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2925 delete $self->{empty}->{$dir};
2926 $self->{empty}->{$path} = 1;
2927 { path => $path };
2928 }
2929
2930 sub change_dir_prop {
2931 my ($self, $db, $prop, $value) = @_;
2932 $self->{dir_prop}->{$db->{path}} ||= {};
2933 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2934 undef;
2935 }
2936
2937 sub absent_directory {
2938 my ($self, $path, $pb) = @_;
2939 $self->{absent_dir}->{$pb->{path}} ||= [];
2940 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2941 undef;
2942 }
2943
2944 sub absent_file {
2945 my ($self, $path, $pb) = @_;
2946 $self->{absent_file}->{$pb->{path}} ||= [];
2947 push @{$self->{absent_file}->{$pb->{path}}}, $path;
2948 undef;
2949 }
2950
2951 sub change_file_prop {
2952 my ($self, $fb, $prop, $value) = @_;
2953 if ($prop eq 'svn:executable') {
2954 if ($fb->{mode_b} != 120000) {
2955 $fb->{mode_b} = defined $value ? 100755 : 100644;
2956 }
2957 } elsif ($prop eq 'svn:special') {
2958 $fb->{mode_b} = defined $value ? 120000 : 100644;
2959 } else {
2960 $self->{file_prop}->{$fb->{path}} ||= {};
2961 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2962 }
2963 undef;
2964 }
2965
2966 sub apply_textdelta {
2967 my ($self, $fb, $exp) = @_;
2968 my $fh = IO::File->new_tmpfile;
2969 $fh->autoflush(1);
2970 # $fh gets auto-closed() by SVN::TxDelta::apply(),
2971 # (but $base does not,) so dup() it for reading in close_file
2972 open my $dup, '<&', $fh or croak $!;
2973 my $base = IO::File->new_tmpfile;
2974 $base->autoflush(1);
2975 if ($fb->{blob}) {
2976 defined (my $pid = fork) or croak $!;
2977 if (!$pid) {
2978 open STDOUT, '>&', $base or croak $!;
2979 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2980 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2981 }
2982 waitpid $pid, 0;
2983 croak $? if $?;
2984
2985 if (defined $exp) {
2986 seek $base, 0, 0 or croak $!;
2987 my $got = ::md5sum($base);
2988 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2989 "expected: $exp\n",
2990 " got: $got\n" if ($got ne $exp);
2991 }
2992 }
2993 seek $base, 0, 0 or croak $!;
2994 $fb->{fh} = $dup;
2995 $fb->{base} = $base;
2996 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2997 }
2998
2999 sub close_file {
3000 my ($self, $fb, $exp) = @_;
3001 my $hash;
3002 my $path = $self->git_path($fb->{path});
3003 if (my $fh = $fb->{fh}) {
3004 if (defined $exp) {
3005 seek($fh, 0, 0) or croak $!;
3006 my $got = ::md5sum($fh);
3007 if ($got ne $exp) {
3008 die "Checksum mismatch: $path\n",
3009 "expected: $exp\n got: $got\n";
3010 }
3011 }
3012 sysseek($fh, 0, 0) or croak $!;
3013 if ($fb->{mode_b} == 120000) {
3014 sysread($fh, my $buf, 5) == 5 or croak $!;
3015 $buf eq 'link ' or die "$path has mode 120000",
3016 "but is not a link\n";
3017 }
3018 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
3019 if (!$pid) {
3020 open STDIN, '<&', $fh or croak $!;
3021 exec qw/git-hash-object -w --stdin/ or croak $!;
3022 }
3023 chomp($hash = do { local $/; <$out> });
3024 close $out or croak $!;
3025 close $fh or croak $!;
3026 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3027 close $fb->{base} or croak $!;
3028 } else {
3029 $hash = $fb->{blob} or die "no blob information\n";
3030 }
3031 $fb->{pool}->clear;
3032 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3033 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3034 undef;
3035 }
3036
3037 sub abort_edit {
3038 my $self = shift;
3039 $self->{nr} = $self->{gii}->{nr};
3040 delete $self->{gii};
3041 $self->SUPER::abort_edit(@_);
3042 }
3043
3044 sub close_edit {
3045 my $self = shift;
3046 $self->{git_commit_ok} = 1;
3047 $self->{nr} = $self->{gii}->{nr};
3048 delete $self->{gii};
3049 $self->SUPER::close_edit(@_);
3050 }
3051
3052 package SVN::Git::Editor;
3053 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3054 use strict;
3055 use warnings;
3056 use Carp qw/croak/;
3057 use IO::File;
3058
3059 sub new {
3060 my ($class, $opts) = @_;
3061 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3062 die "$_ required!\n" unless (defined $opts->{$_});
3063 }
3064
3065 my $pool = SVN::Pool->new;
3066 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3067 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3068 $opts->{r}, $mods);
3069
3070 # $opts->{ra} functions should not be used after this:
3071 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3072 $opts->{editor_cb}, $pool);
3073 my $self = SVN::Delta::Editor->new(@ce, $pool);
3074 bless $self, $class;
3075 foreach (qw/svn_path r tree_a tree_b/) {
3076 $self->{$_} = $opts->{$_};
3077 }
3078 $self->{url} = $opts->{ra}->{url};
3079 $self->{mods} = $mods;
3080 $self->{types} = $types;
3081 $self->{pool} = $pool;
3082 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3083 $self->{rm} = { };
3084 $self->{path_prefix} = length $self->{svn_path} ?
3085 "$self->{svn_path}/" : '';
3086 return $self;
3087 }
3088
3089 sub generate_diff {
3090 my ($tree_a, $tree_b) = @_;
3091 my @diff_tree = qw(diff-tree -z -r);
3092 if ($_cp_similarity) {
3093 push @diff_tree, "-C$_cp_similarity";
3094 } else {
3095 push @diff_tree, '-C';
3096 }
3097 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3098 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3099 push @diff_tree, $tree_a, $tree_b;
3100 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3101 local $/ = "\0";
3102 my $state = 'meta';
3103 my @mods;
3104 while (<$diff_fh>) {
3105 chomp $_; # this gets rid of the trailing "\0"
3106 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3107 $::sha1\s($::sha1)\s
3108 ([MTCRAD])\d*$/xo) {
3109 push @mods, { mode_a => $1, mode_b => $2,
3110 sha1_b => $3, chg => $4 };
3111 if ($4 =~ /^(?:C|R)$/) {
3112 $state = 'file_a';
3113 } else {
3114 $state = 'file_b';
3115 }
3116 } elsif ($state eq 'file_a') {
3117 my $x = $mods[$#mods] or croak "Empty array\n";
3118 if ($x->{chg} !~ /^(?:C|R)$/) {
3119 croak "Error parsing $_, $x->{chg}\n";
3120 }
3121 $x->{file_a} = $_;
3122 $state = 'file_b';
3123 } elsif ($state eq 'file_b') {
3124 my $x = $mods[$#mods] or croak "Empty array\n";
3125 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3126 croak "Error parsing $_, $x->{chg}\n";
3127 }
3128 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3129 croak "Error parsing $_, $x->{chg}\n";
3130 }
3131 $x->{file_b} = $_;
3132 $state = 'meta';
3133 } else {
3134 croak "Error parsing $_\n";
3135 }
3136 }
3137 command_close_pipe($diff_fh, $ctx);
3138 \@mods;
3139 }
3140
3141 sub check_diff_paths {
3142 my ($ra, $pfx, $rev, $mods) = @_;
3143 my %types;
3144 $pfx .= '/' if length $pfx;
3145
3146 sub type_diff_paths {
3147 my ($ra, $types, $path, $rev) = @_;
3148 my @p = split m#/+#, $path;
3149 my $c = shift @p;
3150 unless (defined $types->{$c}) {
3151 $types->{$c} = $ra->check_path($c, $rev);
3152 }
3153 while (@p) {
3154 $c .= '/' . shift @p;
3155 next if defined $types->{$c};
3156 $types->{$c} = $ra->check_path($c, $rev);
3157 }
3158 }
3159
3160 foreach my $m (@$mods) {
3161 foreach my $f (qw/file_a file_b/) {
3162 next unless defined $m->{$f};
3163 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3164 if (length $pfx.$dir && ! defined $types{$dir}) {
3165 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3166 }
3167 }
3168 }
3169 \%types;
3170 }
3171
3172 sub split_path {
3173 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3174 }
3175
3176 sub repo_path {
3177 my ($self, $path) = @_;
3178 $self->{path_prefix}.(defined $path ? $path : '');
3179 }
3180
3181 sub url_path {
3182 my ($self, $path) = @_;
3183 if ($self->{url} =~ m#^https?://#) {
3184 $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3185 }
3186 $self->{url} . '/' . $self->repo_path($path);
3187 }
3188
3189 sub rmdirs {
3190 my ($self) = @_;
3191 my $rm = $self->{rm};
3192 delete $rm->{''}; # we never delete the url we're tracking
3193 return unless %$rm;
3194
3195 foreach (keys %$rm) {
3196 my @d = split m#/#, $_;
3197 my $c = shift @d;
3198 $rm->{$c} = 1;
3199 while (@d) {
3200 $c .= '/' . shift @d;
3201 $rm->{$c} = 1;
3202 }
3203 }
3204 delete $rm->{$self->{svn_path}};
3205 delete $rm->{''}; # we never delete the url we're tracking
3206 return unless %$rm;
3207
3208 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3209 $self->{tree_b});
3210 local $/ = "\0";
3211 while (<$fh>) {
3212 chomp;
3213 my @dn = split m#/#, $_;
3214 while (pop @dn) {
3215 delete $rm->{join '/', @dn};
3216 }
3217 unless (%$rm) {
3218 close $fh;
3219 return;
3220 }
3221 }
3222 command_close_pipe($fh, $ctx);
3223
3224 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3225 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3226 $self->close_directory($bat->{$d}, $p);
3227 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3228 print "\tD+\t$d/\n" unless $::_q;
3229 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3230 delete $bat->{$d};
3231 }
3232 }
3233
3234 sub open_or_add_dir {
3235 my ($self, $full_path, $baton) = @_;
3236 my $t = $self->{types}->{$full_path};
3237 if (!defined $t) {
3238 die "$full_path not known in r$self->{r} or we have a bug!\n";
3239 }
3240 {
3241 no warnings 'once';
3242 # SVN::Node::none and SVN::Node::file are used only once,
3243 # so we're shutting up Perl's warnings about them.
3244 if ($t == $SVN::Node::none) {
3245 return $self->add_directory($full_path, $baton,
3246 undef, -1, $self->{pool});
3247 } elsif ($t == $SVN::Node::dir) {
3248 return $self->open_directory($full_path, $baton,
3249 $self->{r}, $self->{pool});
3250 } # no warnings 'once'
3251 print STDERR "$full_path already exists in repository at ",
3252 "r$self->{r} and it is not a directory (",
3253 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3254 } # no warnings 'once'
3255 exit 1;
3256 }
3257
3258 sub ensure_path {
3259 my ($self, $path) = @_;
3260 my $bat = $self->{bat};
3261 my $repo_path = $self->repo_path($path);
3262 return $bat->{''} unless (length $repo_path);
3263 my @p = split m#/+#, $repo_path;
3264 my $c = shift @p;
3265 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3266 while (@p) {
3267 my $c0 = $c;
3268 $c .= '/' . shift @p;
3269 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3270 }
3271 return $bat->{$c};
3272 }
3273
3274 sub A {
3275 my ($self, $m) = @_;
3276 my ($dir, $file) = split_path($m->{file_b});
3277 my $pbat = $self->ensure_path($dir);
3278 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3279 undef, -1);
3280 print "\tA\t$m->{file_b}\n" unless $::_q;
3281 $self->chg_file($fbat, $m);
3282 $self->close_file($fbat,undef,$self->{pool});
3283 }
3284
3285 sub C {
3286 my ($self, $m) = @_;
3287 my ($dir, $file) = split_path($m->{file_b});
3288 my $pbat = $self->ensure_path($dir);
3289 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3290 $self->url_path($m->{file_a}), $self->{r});
3291 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3292 $self->chg_file($fbat, $m);
3293 $self->close_file($fbat,undef,$self->{pool});
3294 }
3295
3296 sub delete_entry {
3297 my ($self, $path, $pbat) = @_;
3298 my $rpath = $self->repo_path($path);
3299 my ($dir, $file) = split_path($rpath);
3300 $self->{rm}->{$dir} = 1;
3301 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3302 }
3303
3304 sub R {
3305 my ($self, $m) = @_;
3306 my ($dir, $file) = split_path($m->{file_b});
3307 my $pbat = $self->ensure_path($dir);
3308 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3309 $self->url_path($m->{file_a}), $self->{r});
3310 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3311 $self->chg_file($fbat, $m);
3312 $self->close_file($fbat,undef,$self->{pool});
3313
3314 ($dir, $file) = split_path($m->{file_a});
3315 $pbat = $self->ensure_path($dir);
3316 $self->delete_entry($m->{file_a}, $pbat);
3317 }
3318
3319 sub M {
3320 my ($self, $m) = @_;
3321 my ($dir, $file) = split_path($m->{file_b});
3322 my $pbat = $self->ensure_path($dir);
3323 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3324 $pbat,$self->{r},$self->{pool});
3325 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3326 $self->chg_file($fbat, $m);
3327 $self->close_file($fbat,undef,$self->{pool});
3328 }
3329
3330 sub T { shift->M(@_) }
3331
3332 sub change_file_prop {
3333 my ($self, $fbat, $pname, $pval) = @_;
3334 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3335 }
3336
3337 sub chg_file {
3338 my ($self, $fbat, $m) = @_;
3339 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3340 $self->change_file_prop($fbat,'svn:executable','*');
3341 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3342 $self->change_file_prop($fbat,'svn:executable',undef);
3343 }
3344 my $fh = IO::File->new_tmpfile or croak $!;
3345 if ($m->{mode_b} =~ /^120/) {
3346 print $fh 'link ' or croak $!;
3347 $self->change_file_prop($fbat,'svn:special','*');
3348 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3349 $self->change_file_prop($fbat,'svn:special',undef);
3350 }
3351 defined(my $pid = fork) or croak $!;
3352 if (!$pid) {
3353 open STDOUT, '>&', $fh or croak $!;
3354 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3355 }
3356 waitpid $pid, 0;
3357 croak $? if $?;
3358 $fh->flush == 0 or croak $!;
3359 seek $fh, 0, 0 or croak $!;
3360
3361 my $exp = ::md5sum($fh);
3362 seek $fh, 0, 0 or croak $!;
3363
3364 my $pool = SVN::Pool->new;
3365 my $atd = $self->apply_textdelta($fbat, undef, $pool);
3366 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3367 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3368 $pool->clear;
3369
3370 close $fh or croak $!;
3371 }
3372
3373 sub D {
3374 my ($self, $m) = @_;
3375 my ($dir, $file) = split_path($m->{file_b});
3376 my $pbat = $self->ensure_path($dir);
3377 print "\tD\t$m->{file_b}\n" unless $::_q;
3378 $self->delete_entry($m->{file_b}, $pbat);
3379 }
3380
3381 sub close_edit {
3382 my ($self) = @_;
3383 my ($p,$bat) = ($self->{pool}, $self->{bat});
3384 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3385 next if $_ eq '';
3386 $self->close_directory($bat->{$_}, $p);
3387 }
3388 $self->close_directory($bat->{''}, $p);
3389 $self->SUPER::close_edit($p);
3390 $p->clear;
3391 }
3392
3393 sub abort_edit {
3394 my ($self) = @_;
3395 $self->SUPER::abort_edit($self->{pool});
3396 }
3397
3398 sub DESTROY {
3399 my $self = shift;
3400 $self->SUPER::DESTROY(@_);
3401 $self->{pool}->clear;
3402 }
3403
3404 # this drives the editor
3405 sub apply_diff {
3406 my ($self) = @_;
3407 my $mods = $self->{mods};
3408 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3409 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3410 my $f = $m->{chg};
3411 if (defined $o{$f}) {
3412 $self->$f($m);
3413 } else {
3414 fatal("Invalid change type: $f");
3415 }
3416 }
3417 $self->rmdirs if $_rmdir;
3418 if (@$mods == 0) {
3419 $self->abort_edit;
3420 } else {
3421 $self->close_edit;
3422 }
3423 return scalar @$mods;
3424 }
3425
3426 package Git::SVN::Ra;
3427 use vars qw/@ISA $config_dir $_log_window_size/;
3428 use strict;
3429 use warnings;
3430 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3431
3432 BEGIN {
3433 # enforce temporary pool usage for some simple functions
3434 no strict 'refs';
3435 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3436 my $SUPER = "SUPER::$f";
3437 *$f = sub {
3438 my $self = shift;
3439 my $pool = SVN::Pool->new;
3440 my @ret = $self->$SUPER(@_,$pool);
3441 $pool->clear;
3442 wantarray ? @ret : $ret[0];
3443 };
3444 }
3445 }
3446
3447 sub _auth_providers () {
3448 [
3449 SVN::Client::get_simple_provider(),
3450 SVN::Client::get_ssl_server_trust_file_provider(),
3451 SVN::Client::get_simple_prompt_provider(
3452 \&Git::SVN::Prompt::simple, 2),
3453 SVN::Client::get_ssl_client_cert_file_provider(),
3454 SVN::Client::get_ssl_client_cert_prompt_provider(
3455 \&Git::SVN::Prompt::ssl_client_cert, 2),
3456 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3457 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3458 SVN::Client::get_username_provider(),
3459 SVN::Client::get_ssl_server_trust_prompt_provider(
3460 \&Git::SVN::Prompt::ssl_server_trust),
3461 SVN::Client::get_username_prompt_provider(
3462 \&Git::SVN::Prompt::username, 2)
3463 ]
3464 }
3465
3466 sub escape_uri_only {
3467 my ($uri) = @_;
3468 my @tmp;
3469 foreach (split m{/}, $uri) {
3470 s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
3471 push @tmp, $_;
3472 }
3473 join('/', @tmp);
3474 }
3475
3476 sub escape_url {
3477 my ($url) = @_;
3478 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3479 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3480 $url = "$scheme://$domain$uri";
3481 }
3482 $url;
3483 }
3484
3485 sub new {
3486 my ($class, $url) = @_;
3487 $url =~ s!/+$!!;
3488 return $RA if ($RA && $RA->{url} eq $url);
3489
3490 SVN::_Core::svn_config_ensure($config_dir, undef);
3491 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3492 my $config = SVN::Core::config_get_config($config_dir);
3493 $RA = undef;
3494 my $dont_store_passwords = 1;
3495 my $conf_t = ${$config}{'config'};
3496 {
3497 no warnings 'once';
3498 # The usage of $SVN::_Core::SVN_CONFIG_* variables
3499 # produces warnings that variables are used only once.
3500 # I had not found the better way to shut them up, so
3501 # the warnings of type 'once' are disabled in this block.
3502 if (SVN::_Core::svn_config_get_bool($conf_t,
3503 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3504 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3505 1) == 0) {
3506 SVN::_Core::svn_auth_set_parameter($baton,
3507 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3508 bless (\$dont_store_passwords, "_p_void"));
3509 }
3510 if (SVN::_Core::svn_config_get_bool($conf_t,
3511 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3512 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3513 1) == 0) {
3514 $Git::SVN::Prompt::_no_auth_cache = 1;
3515 }
3516 } # no warnings 'once'
3517 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3518 config => $config,
3519 pool => SVN::Pool->new,
3520 auth_provider_callbacks => $callbacks);
3521 $self->{url} = $url;
3522 $self->{svn_path} = $url;
3523 $self->{repos_root} = $self->get_repos_root;
3524 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3525 $self->{cache} = { check_path => { r => 0, data => {} },
3526 get_dir => { r => 0, data => {} } };
3527 $RA = bless $self, $class;
3528 }
3529
3530 sub check_path {
3531 my ($self, $path, $r) = @_;
3532 my $cache = $self->{cache}->{check_path};
3533 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3534 return $cache->{data}->{$path};
3535 }
3536 my $pool = SVN::Pool->new;
3537 my $t = $self->SUPER::check_path($path, $r, $pool);
3538 $pool->clear;
3539 if ($r != $cache->{r}) {
3540 %{$cache->{data}} = ();
3541 $cache->{r} = $r;
3542 }
3543 $cache->{data}->{$path} = $t;
3544 }
3545
3546 sub get_dir {
3547 my ($self, $dir, $r) = @_;
3548 my $cache = $self->{cache}->{get_dir};
3549 if ($r == $cache->{r}) {
3550 if (my $x = $cache->{data}->{$dir}) {
3551 return wantarray ? @$x : $x->[0];
3552 }
3553 }
3554 my $pool = SVN::Pool->new;
3555 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3556 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3557 $pool->clear;
3558 if ($r != $cache->{r}) {
3559 %{$cache->{data}} = ();
3560 $cache->{r} = $r;
3561 }
3562 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3563 wantarray ? (\%dirents, $r, $props) : \%dirents;
3564 }
3565
3566 sub DESTROY {
3567 # do not call the real DESTROY since we store ourselves in $RA
3568 }
3569
3570 sub get_log {
3571 my ($self, @args) = @_;
3572 my $pool = SVN::Pool->new;
3573 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3574 my $ret = $self->SUPER::get_log(@args, $pool);
3575 $pool->clear;
3576 $ret;
3577 }
3578
3579 sub trees_match {
3580 my ($self, $url1, $rev1, $url2, $rev2) = @_;
3581 my $ctx = SVN::Client->new(auth => _auth_providers);
3582 my $out = IO::File->new_tmpfile;
3583
3584 # older SVN (1.1.x) doesn't take $pool as the last parameter for
3585 # $ctx->diff(), so we'll create a default one
3586 my $pool = SVN::Pool->new_default_sub;
3587
3588 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3589 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3590 $out->flush;
3591 my $ret = (($out->stat)[7] == 0);
3592 close $out or croak $!;
3593
3594 $ret;
3595 }
3596
3597 sub get_commit_editor {
3598 my ($self, $log, $cb, $pool) = @_;
3599 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3600 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3601 }
3602
3603 sub gs_do_update {
3604 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3605 my $new = ($rev_a == $rev_b);
3606 my $path = $gs->{path};
3607
3608 if ($new && -e $gs->{index}) {
3609 unlink $gs->{index} or die
3610 "Couldn't unlink index: $gs->{index}: $!\n";
3611 }
3612 my $pool = SVN::Pool->new;
3613 $editor->set_path_strip($path);
3614 my (@pc) = split m#/#, $path;
3615 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3616 1, $editor, $pool);
3617 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3618
3619 # Since we can't rely on svn_ra_reparent being available, we'll
3620 # just have to do some magic with set_path to make it so
3621 # we only want a partial path.
3622 my $sp = '';
3623 my $final = join('/', @pc);
3624 while (@pc) {
3625 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3626 $sp .= '/' if length $sp;
3627 $sp .= shift @pc;
3628 }
3629 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3630
3631 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3632
3633 $reporter->finish_report($pool);
3634 $pool->clear;
3635 $editor->{git_commit_ok};
3636 }
3637
3638 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3639 # svn_ra_reparent didn't work before 1.4)
3640 sub gs_do_switch {
3641 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3642 my $path = $gs->{path};
3643 my $pool = SVN::Pool->new;
3644
3645 my $full_url = $self->{url};
3646 my $old_url = $full_url;
3647 $full_url .= '/' . escape_uri_only($path) if length $path;
3648 my ($ra, $reparented);
3649 if ($old_url ne $full_url) {
3650 if ($old_url !~ m#^svn(\+ssh)?://#) {
3651 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3652 $pool);
3653 $self->{url} = $full_url;
3654 $reparented = 1;
3655 } else {
3656 $_[0] = undef;
3657 $self = undef;
3658 $RA = undef;
3659 $ra = Git::SVN::Ra->new($full_url);
3660 $ra_invalid = 1;
3661 }
3662 }
3663 $ra ||= $self;
3664 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3665 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3666 $reporter->set_path('', $rev_a, 0, @lock, $pool);
3667 $reporter->finish_report($pool);
3668
3669 if ($reparented) {
3670 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3671 $self->{url} = $old_url;
3672 }
3673
3674 $pool->clear;
3675 $editor->{git_commit_ok};
3676 }
3677
3678 sub longest_common_path {
3679 my ($gsv, $globs) = @_;
3680 my %common;
3681 my $common_max = scalar @$gsv;
3682
3683 foreach my $gs (@$gsv) {
3684 my @tmp = split m#/#, $gs->{path};
3685 my $p = '';
3686 foreach (@tmp) {
3687 $p .= length($p) ? "/$_" : $_;
3688 $common{$p} ||= 0;
3689 $common{$p}++;
3690 }
3691 }
3692 $globs ||= [];
3693 $common_max += scalar @$globs;
3694 foreach my $glob (@$globs) {
3695 my @tmp = split m#/#, $glob->{path}->{left};
3696 my $p = '';
3697 foreach (@tmp) {
3698 $p .= length($p) ? "/$_" : $_;
3699 $common{$p} ||= 0;
3700 $common{$p}++;
3701 }
3702 }
3703
3704 my $longest_path = '';
3705 foreach (sort {length $b <=> length $a} keys %common) {
3706 if ($common{$_} == $common_max) {
3707 $longest_path = $_;
3708 last;
3709 }
3710 }
3711 $longest_path;
3712 }
3713
3714 sub gs_fetch_loop_common {
3715 my ($self, $base, $head, $gsv, $globs) = @_;
3716 return if ($base > $head);
3717 my $inc = $_log_window_size;
3718 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3719 my $longest_path = longest_common_path($gsv, $globs);
3720 my $ra_url = $self->{url};
3721 while (1) {
3722 my %revs;
3723 my $err;
3724 my $err_handler = $SVN::Error::handler;
3725 $SVN::Error::handler = sub {
3726 ($err) = @_;
3727 skip_unknown_revs($err);
3728 };
3729 sub _cb {
3730 my ($paths, $r, $author, $date, $log) = @_;
3731 [ dup_changed_paths($paths),
3732 { author => $author, date => $date, log => $log } ];
3733 }
3734 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3735 sub { $revs{$_[1]} = _cb(@_) });
3736 if ($err && $max >= $head) {
3737 print STDERR "Path '$longest_path' ",
3738 "was probably deleted:\n",
3739 $err->expanded_message,
3740 "\nWill attempt to follow ",
3741 "revisions r$min .. r$max ",
3742 "committed before the deletion\n";
3743 my $hi = $max;
3744 while (--$hi >= $min) {
3745 my $ok;
3746 $self->get_log([$longest_path], $min, $hi,
3747 0, 1, 1, sub {
3748 $ok ||= $_[1];
3749 $revs{$_[1]} = _cb(@_) });
3750 if ($ok) {
3751 print STDERR "r$min .. r$ok OK\n";
3752 last;
3753 }
3754 }
3755 }
3756 $SVN::Error::handler = $err_handler;
3757
3758 my %exists = map { $_->{path} => $_ } @$gsv;
3759 foreach my $r (sort {$a <=> $b} keys %revs) {
3760 my ($paths, $logged) = @{$revs{$r}};
3761
3762 foreach my $gs ($self->match_globs(\%exists, $paths,
3763 $globs, $r)) {
3764 if ($gs->rev_db_max >= $r) {
3765 next;
3766 }
3767 next unless $gs->match_paths($paths, $r);
3768 $gs->{logged_rev_props} = $logged;
3769 if (my $last_commit = $gs->last_commit) {
3770 $gs->assert_index_clean($last_commit);
3771 }
3772 my $log_entry = $gs->do_fetch($paths, $r);
3773 if ($log_entry) {
3774 $gs->do_git_commit($log_entry);
3775 }
3776 }
3777 foreach my $g (@$globs) {
3778 my $k = "svn-remote.$g->{remote}." .
3779 "$g->{t}-maxRev";
3780 Git::SVN::tmp_config($k, $r);
3781 }
3782 if ($ra_invalid) {
3783 $_[0] = undef;
3784 $self = undef;
3785 $RA = undef;
3786 $self = Git::SVN::Ra->new($ra_url);
3787 $ra_invalid = undef;
3788 }
3789 }
3790 # pre-fill the .rev_db since it'll eventually get filled in
3791 # with '0' x40 if something new gets committed
3792 foreach my $gs (@$gsv) {
3793 next if defined $gs->rev_db_get($max);
3794 $gs->rev_db_set($max, 0 x40);
3795 }
3796 foreach my $g (@$globs) {
3797 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3798 Git::SVN::tmp_config($k, $max);
3799 }
3800 last if $max >= $head;
3801 $min = $max + 1;
3802 $max += $inc;
3803 $max = $head if ($max > $head);
3804 }
3805 }
3806
3807 sub match_globs {
3808 my ($self, $exists, $paths, $globs, $r) = @_;
3809
3810 sub get_dir_check {
3811 my ($self, $exists, $g, $r) = @_;
3812 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3813 return unless scalar @x == 3;
3814 my $dirents = $x[0];
3815 foreach my $de (keys %$dirents) {
3816 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
3817 my $p = $g->{path}->full_path($de);
3818 next if $exists->{$p};
3819 next if (length $g->{path}->{right} &&
3820 ($self->check_path($p, $r) !=
3821 $SVN::Node::dir));
3822 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3823 $g->{ref}->full_path($de), 1);
3824 }
3825 }
3826 foreach my $g (@$globs) {
3827 if (my $path = $paths->{"/$g->{path}->{left}"}) {
3828 if ($path->{action} =~ /^[AR]$/) {
3829 get_dir_check($self, $exists, $g, $r);
3830 }
3831 }
3832 foreach (keys %$paths) {
3833 if (/$g->{path}->{left_regex}/ &&
3834 !/$g->{path}->{regex}/) {
3835 next if $paths->{$_}->{action} !~ /^[AR]$/;
3836 get_dir_check($self, $exists, $g, $r);
3837 }
3838 next unless /$g->{path}->{regex}/;
3839 my $p = $1;
3840 my $pathname = $g->{path}->full_path($p);
3841 next if $exists->{$pathname};
3842 next if ($self->check_path($pathname, $r) !=
3843 $SVN::Node::dir);
3844 $exists->{$pathname} = Git::SVN->init(
3845 $self->{url}, $pathname, undef,
3846 $g->{ref}->full_path($p), 1);
3847 }
3848 my $c = '';
3849 foreach (split m#/#, $g->{path}->{left}) {
3850 $c .= "/$_";
3851 next unless ($paths->{$c} &&
3852 ($paths->{$c}->{action} =~ /^[AR]$/));
3853 get_dir_check($self, $exists, $g, $r);
3854 }
3855 }
3856 values %$exists;
3857 }
3858
3859 sub minimize_url {
3860 my ($self) = @_;
3861 return $self->{url} if ($self->{url} eq $self->{repos_root});
3862 my $url = $self->{repos_root};
3863 my @components = split(m!/!, $self->{svn_path});
3864 my $c = '';
3865 do {
3866 $url .= "/$c" if length $c;
3867 eval { (ref $self)->new($url)->get_latest_revnum };
3868 } while ($@ && ($c = shift @components));
3869 $url;
3870 }
3871
3872 sub can_do_switch {
3873 my $self = shift;
3874 unless (defined $can_do_switch) {
3875 my $pool = SVN::Pool->new;
3876 my $rep = eval {
3877 $self->do_switch(1, '', 0, $self->{url},
3878 SVN::Delta::Editor->new, $pool);
3879 };
3880 if ($@) {
3881 $can_do_switch = 0;
3882 } else {
3883 $rep->abort_report($pool);
3884 $can_do_switch = 1;
3885 }
3886 $pool->clear;
3887 }
3888 $can_do_switch;
3889 }
3890
3891 sub skip_unknown_revs {
3892 my ($err) = @_;
3893 my $errno = $err->apr_err();
3894 # Maybe the branch we're tracking didn't
3895 # exist when the repo started, so it's
3896 # not an error if it doesn't, just continue
3897 #
3898 # Wonderfully consistent library, eh?
3899 # 160013 - svn:// and file://
3900 # 175002 - http(s)://
3901 # 175007 - http(s):// (this repo required authorization, too...)
3902 # More codes may be discovered later...
3903 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3904 my $err_key = $err->expanded_message;
3905 # revision numbers change every time, filter them out
3906 $err_key =~ s/\d+/\0/g;
3907 $err_key = "$errno\0$err_key";
3908 unless ($ignored_err{$err_key}) {
3909 warn "W: Ignoring error from SVN, path probably ",
3910 "does not exist: ($errno): ",
3911 $err->expanded_message,"\n";
3912 $ignored_err{$err_key} = 1;
3913 }
3914 return;
3915 }
3916 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3917 }
3918
3919 # svn_log_changed_path_t objects passed to get_log are likely to be
3920 # overwritten even if only the refs are copied to an external variable,
3921 # so we should dup the structures in their entirety. Using an externally
3922 # passed pool (instead of our temporary and quickly cleared pool in
3923 # Git::SVN::Ra) does not help matters at all...
3924 sub dup_changed_paths {
3925 my ($paths) = @_;
3926 return undef unless $paths;
3927 my %ret;
3928 foreach my $p (keys %$paths) {
3929 my $i = $paths->{$p};
3930 my %s = map { $_ => $i->$_ }
3931 qw/copyfrom_path copyfrom_rev action/;
3932 $ret{$p} = \%s;
3933 }
3934 \%ret;
3935 }
3936
3937 package Git::SVN::Log;
3938 use strict;
3939 use warnings;
3940 use POSIX qw/strftime/;
3941 use constant commit_log_separator => ('-' x 72) . "\n";
3942 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3943 %rusers $show_commit $incremental/;
3944 my $l_fmt;
3945
3946 sub cmt_showable {
3947 my ($c) = @_;
3948 return 1 if defined $c->{r};
3949
3950 # big commit message got truncated by the 16k pretty buffer in rev-list
3951 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3952 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3953 @{$c->{l}} = ();
3954 my @log = command(qw/cat-file commit/, $c->{c});
3955
3956 # shift off the headers
3957 shift @log while ($log[0] ne '');
3958 shift @log;
3959
3960 # TODO: make $c->{l} not have a trailing newline in the future
3961 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
3962
3963 (undef, $c->{r}, undef) = ::extract_metadata(
3964 (grep(/^git-svn-id: /, @log))[-1]);
3965 }
3966 return defined $c->{r};
3967 }
3968
3969 sub log_use_color {
3970 return 1 if $color;
3971 my ($dc, $dcvar);
3972 $dcvar = 'color.diff';
3973 $dc = `git-config --get $dcvar`;
3974 if ($dc eq '') {
3975 # nothing at all; fallback to "diff.color"
3976 $dcvar = 'diff.color';
3977 $dc = `git-config --get $dcvar`;
3978 }
3979 chomp($dc);
3980 if ($dc eq 'auto') {
3981 my $pc;
3982 $pc = `git-config --get color.pager`;
3983 if ($pc eq '') {
3984 # does not have it -- fallback to pager.color
3985 $pc = `git-config --bool --get pager.color`;
3986 }
3987 else {
3988 $pc = `git-config --bool --get color.pager`;
3989 if ($?) {
3990 $pc = 'false';
3991 }
3992 }
3993 chomp($pc);
3994 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3995 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3996 }
3997 return 0;
3998 }
3999 return 0 if $dc eq 'never';
4000 return 1 if $dc eq 'always';
4001 chomp($dc = `git-config --bool --get $dcvar`);
4002 return ($dc eq 'true');
4003 }
4004
4005 sub git_svn_log_cmd {
4006 my ($r_min, $r_max, @args) = @_;
4007 my $head = 'HEAD';
4008 my (@files, @log_opts);
4009 foreach my $x (@args) {
4010 if ($x eq '--' || @files) {
4011 push @files, $x;
4012 } else {
4013 if (::verify_ref("$x^0")) {
4014 $head = $x;
4015 } else {
4016 push @log_opts, $x;
4017 }
4018 }
4019 }
4020
4021 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4022 $gs ||= Git::SVN->_new;
4023 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4024 $gs->refname);
4025 push @cmd, '-r' unless $non_recursive;
4026 push @cmd, qw/--raw --name-status/ if $verbose;
4027 push @cmd, '--color' if log_use_color();
4028 push @cmd, @log_opts;
4029 if (defined $r_max && $r_max == $r_min) {
4030 push @cmd, '--max-count=1';
4031 if (my $c = $gs->rev_db_get($r_max)) {
4032 push @cmd, $c;
4033 }
4034 } elsif (defined $r_max) {
4035 if ($r_max < $r_min) {
4036 ($r_min, $r_max) = ($r_max, $r_min);
4037 }
4038 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4039 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4040 # If there are no commits in the range, both $c_max and $c_min
4041 # will be undefined. If there is at least 1 commit in the
4042 # range, both will be defined.
4043 return () if !defined $c_min || !defined $c_max;
4044 if ($c_min eq $c_max) {
4045 push @cmd, '--max-count=1', $c_min;
4046 } else {
4047 push @cmd, '--boundary', "$c_min..$c_max";
4048 }
4049 }
4050 return (@cmd, @files);
4051 }
4052
4053 # adapted from pager.c
4054 sub config_pager {
4055 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4056 if (!defined $pager) {
4057 $pager = 'less';
4058 } elsif (length $pager == 0 || $pager eq 'cat') {
4059 $pager = undef;
4060 }
4061 }
4062
4063 sub run_pager {
4064 return unless -t *STDOUT && defined $pager;
4065 pipe my $rfd, my $wfd or return;
4066 defined(my $pid = fork) or ::fatal "Can't fork: $!";
4067 if (!$pid) {
4068 open STDOUT, '>&', $wfd or
4069 ::fatal "Can't redirect to stdout: $!";
4070 return;
4071 }
4072 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4073 $ENV{LESS} ||= 'FRSX';
4074 exec $pager or ::fatal "Can't run pager: $! ($pager)";
4075 }
4076
4077 sub format_svn_date {
4078 return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4079 }
4080
4081 sub parse_git_date {
4082 my ($t, $tz) = @_;
4083 # Date::Parse isn't in the standard Perl distro :(
4084 if ($tz =~ s/^\+//) {
4085 $t += tz_to_s_offset($tz);
4086 } elsif ($tz =~ s/^\-//) {
4087 $t -= tz_to_s_offset($tz);
4088 }
4089 return $t;
4090 }
4091
4092 sub set_local_timezone {
4093 if (defined $TZ) {
4094 $ENV{TZ} = $TZ;
4095 } else {
4096 delete $ENV{TZ};
4097 }
4098 }
4099
4100 sub tz_to_s_offset {
4101 my ($tz) = @_;
4102 $tz =~ s/(\d\d)$//;
4103 return ($1 * 60) + ($tz * 3600);
4104 }
4105
4106 sub get_author_info {
4107 my ($dest, $author, $t, $tz) = @_;
4108 $author =~ s/(?:^\s*|\s*$)//g;
4109 $dest->{a_raw} = $author;
4110 my $au;
4111 if ($::_authors) {
4112 $au = $rusers{$author} || undef;
4113 }
4114 if (!$au) {
4115 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4116 }
4117 $dest->{t} = $t;
4118 $dest->{tz} = $tz;
4119 $dest->{a} = $au;
4120 $dest->{t_utc} = parse_git_date($t, $tz);
4121 }
4122
4123 sub process_commit {
4124 my ($c, $r_min, $r_max, $defer) = @_;
4125 if (defined $r_min && defined $r_max) {
4126 if ($r_min == $c->{r} && $r_min == $r_max) {
4127 show_commit($c);
4128 return 0;
4129 }
4130 return 1 if $r_min == $r_max;
4131 if ($r_min < $r_max) {
4132 # we need to reverse the print order
4133 return 0 if (defined $limit && --$limit < 0);
4134 push @$defer, $c;
4135 return 1;
4136 }
4137 if ($r_min != $r_max) {
4138 return 1 if ($r_min < $c->{r});
4139 return 1 if ($r_max > $c->{r});
4140 }
4141 }
4142 return 0 if (defined $limit && --$limit < 0);
4143 show_commit($c);
4144 return 1;
4145 }
4146
4147 sub show_commit {
4148 my $c = shift;
4149 if ($oneline) {
4150 my $x = "\n";
4151 if (my $l = $c->{l}) {
4152 while ($l->[0] =~ /^\s*$/) { shift @$l }
4153 $x = $l->[0];
4154 }
4155 $l_fmt ||= 'A' . length($c->{r});
4156 print 'r',pack($l_fmt, $c->{r}),' | ';
4157 print "$c->{c} | " if $show_commit;
4158 print $x;
4159 } else {
4160 show_commit_normal($c);
4161 }
4162 }
4163
4164 sub show_commit_changed_paths {
4165 my ($c) = @_;
4166 return unless $c->{changed};
4167 print "Changed paths:\n", @{$c->{changed}};
4168 }
4169
4170 sub show_commit_normal {
4171 my ($c) = @_;
4172 print commit_log_separator, "r$c->{r} | ";
4173 print "$c->{c} | " if $show_commit;
4174 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4175 my $nr_line = 0;
4176
4177 if (my $l = $c->{l}) {
4178 while ($l->[$#$l] eq "\n" && $#$l > 0
4179 && $l->[($#$l - 1)] eq "\n") {
4180 pop @$l;
4181 }
4182 $nr_line = scalar @$l;
4183 if (!$nr_line) {
4184 print "1 line\n\n\n";
4185 } else {
4186 if ($nr_line == 1) {
4187 $nr_line = '1 line';
4188 } else {
4189 $nr_line .= ' lines';
4190 }
4191 print $nr_line, "\n";
4192 show_commit_changed_paths($c);
4193 print "\n";
4194 print $_ foreach @$l;
4195 }
4196 } else {
4197 print "1 line\n";
4198 show_commit_changed_paths($c);
4199 print "\n";
4200
4201 }
4202 foreach my $x (qw/raw stat diff/) {
4203 if ($c->{$x}) {
4204 print "\n";
4205 print $_ foreach @{$c->{$x}}
4206 }
4207 }
4208 }
4209
4210 sub cmd_show_log {
4211 my (@args) = @_;
4212 my ($r_min, $r_max);
4213 my $r_last = -1; # prevent dupes
4214 set_local_timezone();
4215 if (defined $::_revision) {
4216 if ($::_revision =~ /^(\d+):(\d+)$/) {
4217 ($r_min, $r_max) = ($1, $2);
4218 } elsif ($::_revision =~ /^\d+$/) {
4219 $r_min = $r_max = $::_revision;
4220 } else {
4221 ::fatal "-r$::_revision is not supported, use ",
4222 "standard 'git log' arguments instead";
4223 }
4224 }
4225
4226 config_pager();
4227 @args = git_svn_log_cmd($r_min, $r_max, @args);
4228 if (!@args) {
4229 print commit_log_separator unless $incremental || $oneline;
4230 return;
4231 }
4232 my $log = command_output_pipe(@args);
4233 run_pager();
4234 my (@k, $c, $d, $stat);
4235 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4236 while (<$log>) {
4237 if (/^${esc_color}commit -?($::sha1_short)/o) {
4238 my $cmt = $1;
4239 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4240 $r_last = $c->{r};
4241 process_commit($c, $r_min, $r_max, \@k) or
4242 goto out;
4243 }
4244 $d = undef;
4245 $c = { c => $cmt };
4246 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4247 get_author_info($c, $1, $2, $3);
4248 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4249 # ignore
4250 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4251 push @{$c->{raw}}, $_;
4252 } elsif (/^${esc_color}[ACRMDT]\t/) {
4253 # we could add $SVN->{svn_path} here, but that requires
4254 # remote access at the moment (repo_path_split)...
4255 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
4256 push @{$c->{changed}}, $_;
4257 } elsif (/^${esc_color}diff /o) {
4258 $d = 1;
4259 push @{$c->{diff}}, $_;
4260 } elsif ($d) {
4261 push @{$c->{diff}}, $_;
4262 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4263 $esc_color*[\+\-]*$esc_color$/x) {
4264 $stat = 1;
4265 push @{$c->{stat}}, $_;
4266 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4267 push @{$c->{stat}}, $_;
4268 $stat = undef;
4269 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
4270 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4271 } elsif (s/^${esc_color} //o) {
4272 push @{$c->{l}}, $_;
4273 }
4274 }
4275 if ($c && defined $c->{r} && $c->{r} != $r_last) {
4276 $r_last = $c->{r};
4277 process_commit($c, $r_min, $r_max, \@k);
4278 }
4279 if (@k) {
4280 ($r_min, $r_max) = ($r_max, $r_min);
4281 process_commit($_, $r_min, $r_max) foreach reverse @k;
4282 }
4283 out:
4284 close $log;
4285 print commit_log_separator unless $incremental || $oneline;
4286 }
4287
4288 package Git::SVN::Migration;
4289 # these version numbers do NOT correspond to actual version numbers
4290 # of git nor git-svn. They are just relative.
4291 #
4292 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4293 #
4294 # v1 layout: .git/$id/info/url, refs/remotes/$id
4295 #
4296 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4297 #
4298 # v3 layout: .git/svn/$id, refs/remotes/$id
4299 # - info/url may remain for backwards compatibility
4300 # - this is what we migrate up to this layout automatically,
4301 # - this will be used by git svn init on single branches
4302 # v3.1 layout (auto migrated):
4303 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4304 # for backwards compatibility
4305 #
4306 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4307 # - this is only created for newly multi-init-ed
4308 # repositories. Similar in spirit to the
4309 # --use-separate-remotes option in git-clone (now default)
4310 # - we do not automatically migrate to this (following
4311 # the example set by core git)
4312 use strict;
4313 use warnings;
4314 use Carp qw/croak/;
4315 use File::Path qw/mkpath/;
4316 use File::Basename qw/dirname basename/;
4317 use vars qw/$_minimize/;
4318
4319 sub migrate_from_v0 {
4320 my $git_dir = $ENV{GIT_DIR};
4321 return undef unless -d $git_dir;
4322 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4323 my $migrated = 0;
4324 while (<$fh>) {
4325 chomp;
4326 my ($id, $orig_ref) = ($_, $_);
4327 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4328 next unless -f "$git_dir/$id/info/url";
4329 my $new_ref = "refs/remotes/$id";
4330 if (::verify_ref("$new_ref^0")) {
4331 print STDERR "W: $orig_ref is probably an old ",
4332 "branch used by an ancient version of ",
4333 "git-svn.\n",
4334 "However, $new_ref also exists.\n",
4335 "We will not be able ",
4336 "to use this branch until this ",
4337 "ambiguity is resolved.\n";
4338 next;
4339 }
4340 print STDERR "Migrating from v0 layout...\n" if !$migrated;
4341 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4342 command_noisy('update-ref', $new_ref, $orig_ref);
4343 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4344 $migrated++;
4345 }
4346 command_close_pipe($fh, $ctx);
4347 print STDERR "Done migrating from v0 layout...\n" if $migrated;
4348 $migrated;
4349 }
4350
4351 sub migrate_from_v1 {
4352 my $git_dir = $ENV{GIT_DIR};
4353 my $migrated = 0;
4354 return $migrated unless -d $git_dir;
4355 my $svn_dir = "$git_dir/svn";
4356
4357 # just in case somebody used 'svn' as their $id at some point...
4358 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4359
4360 print STDERR "Migrating from a git-svn v1 layout...\n";
4361 mkpath([$svn_dir]);
4362 print STDERR "Data from a previous version of git-svn exists, but\n\t",
4363 "$svn_dir\n\t(required for this version ",
4364 "($::VERSION) of git-svn) does not. exist\n";
4365 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4366 while (<$fh>) {
4367 my $x = $_;
4368 next unless $x =~ s#^refs/remotes/##;
4369 chomp $x;
4370 next unless -f "$git_dir/$x/info/url";
4371 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4372 next unless $u;
4373 my $dn = dirname("$git_dir/svn/$x");
4374 mkpath([$dn]) unless -d $dn;
4375 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4376 mkpath(["$git_dir/svn/svn"]);
4377 print STDERR " - $git_dir/$x/info => ",
4378 "$git_dir/svn/$x/info\n";
4379 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4380 croak "$!: $x";
4381 # don't worry too much about these, they probably
4382 # don't exist with repos this old (save for index,
4383 # and we can easily regenerate that)
4384 foreach my $f (qw/unhandled.log index .rev_db/) {
4385 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4386 }
4387 } else {
4388 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4389 rename "$git_dir/$x", "$git_dir/svn/$x" or
4390 croak "$!: $x";
4391 }
4392 $migrated++;
4393 }
4394 command_close_pipe($fh, $ctx);
4395 print STDERR "Done migrating from a git-svn v1 layout\n";
4396 $migrated;
4397 }
4398
4399 sub read_old_urls {
4400 my ($l_map, $pfx, $path) = @_;
4401 my @dir;
4402 foreach (<$path/*>) {
4403 if (-r "$_/info/url") {
4404 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4405 my $ref_id = $pfx . basename $_;
4406 my $url = ::file_to_s("$_/info/url");
4407 $l_map->{$ref_id} = $url;
4408 } elsif (-d $_) {
4409 push @dir, $_;
4410 }
4411 }
4412 foreach (@dir) {
4413 my $x = $_;
4414 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4415 read_old_urls($l_map, $x, $_);
4416 }
4417 }
4418
4419 sub migrate_from_v2 {
4420 my @cfg = command(qw/config -l/);
4421 return if grep /^svn-remote\..+\.url=/, @cfg;
4422 my %l_map;
4423 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4424 my $migrated = 0;
4425
4426 foreach my $ref_id (sort keys %l_map) {
4427 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4428 if ($@) {
4429 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4430 }
4431 $migrated++;
4432 }
4433 $migrated;
4434 }
4435
4436 sub minimize_connections {
4437 my $r = Git::SVN::read_all_remotes();
4438 my $new_urls = {};
4439 my $root_repos = {};
4440 foreach my $repo_id (keys %$r) {
4441 my $url = $r->{$repo_id}->{url} or next;
4442 my $fetch = $r->{$repo_id}->{fetch} or next;
4443 my $ra = Git::SVN::Ra->new($url);
4444
4445 # skip existing cases where we already connect to the root
4446 if (($ra->{url} eq $ra->{repos_root}) ||
4447 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
4448 $repo_id)) {
4449 $root_repos->{$ra->{url}} = $repo_id;
4450 next;
4451 }
4452
4453 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4454 my $root_path = $ra->{url};
4455 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4456 foreach my $path (keys %$fetch) {
4457 my $ref_id = $fetch->{$path};
4458 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4459
4460 # make sure we can read when connecting to
4461 # a higher level of a repository
4462 my ($last_rev, undef) = $gs->last_rev_commit;
4463 if (!defined $last_rev) {
4464 $last_rev = eval {
4465 $root_ra->get_latest_revnum;
4466 };
4467 next if $@;
4468 }
4469 my $new = $root_path;
4470 $new .= length $path ? "/$path" : '';
4471 eval {
4472 $root_ra->get_log([$new], $last_rev, $last_rev,
4473 0, 0, 1, sub { });
4474 };
4475 next if $@;
4476 $new_urls->{$ra->{repos_root}}->{$new} =
4477 { ref_id => $ref_id,
4478 old_repo_id => $repo_id,
4479 old_path => $path };
4480 }
4481 }
4482
4483 my @emptied;
4484 foreach my $url (keys %$new_urls) {
4485 # see if we can re-use an existing [svn-remote "repo_id"]
4486 # instead of creating a(n ugly) new section:
4487 my $repo_id = $root_repos->{$url} ||
4488 Git::SVN::sanitize_remote_name($url);
4489
4490 my $fetch = $new_urls->{$url};
4491 foreach my $path (keys %$fetch) {
4492 my $x = $fetch->{$path};
4493 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4494 my $pfx = "svn-remote.$x->{old_repo_id}";
4495
4496 my $old_fetch = quotemeta("$x->{old_path}:".
4497 "refs/remotes/$x->{ref_id}");
4498 command_noisy(qw/config --unset/,
4499 "$pfx.fetch", '^'. $old_fetch . '$');
4500 delete $r->{$x->{old_repo_id}}->
4501 {fetch}->{$x->{old_path}};
4502 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4503 command_noisy(qw/config --unset/,
4504 "$pfx.url");
4505 push @emptied, $x->{old_repo_id}
4506 }
4507 }
4508 }
4509 if (@emptied) {
4510 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4511 "$ENV{GIT_DIR}/config";
4512 print STDERR <<EOF;
4513 The following [svn-remote] sections in your config file ($file) are empty
4514 and can be safely removed:
4515 EOF
4516 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4517 }
4518 }
4519
4520 sub migration_check {
4521 migrate_from_v0();
4522 migrate_from_v1();
4523 migrate_from_v2();
4524 minimize_connections() if $_minimize;
4525 }
4526
4527 package Git::IndexInfo;
4528 use strict;
4529 use warnings;
4530 use Git qw/command_input_pipe command_close_pipe/;
4531
4532 sub new {
4533 my ($class) = @_;
4534 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4535 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4536 }
4537
4538 sub remove {
4539 my ($self, $path) = @_;
4540 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4541 return ++$self->{nr};
4542 }
4543 undef;
4544 }
4545
4546 sub update {
4547 my ($self, $mode, $hash, $path) = @_;
4548 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4549 return ++$self->{nr};
4550 }
4551 undef;
4552 }
4553
4554 sub DESTROY {
4555 my ($self) = @_;
4556 command_close_pipe($self->{gui}, $self->{ctx});
4557 }
4558
4559 package Git::SVN::GlobSpec;
4560 use strict;
4561 use warnings;
4562
4563 sub new {
4564 my ($class, $glob) = @_;
4565 my $re = $glob;
4566 $re =~ s!/+$!!g; # no need for trailing slashes
4567 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
4568 my ($left, $right) = ($1, $2);
4569 if ($nr > 1) {
4570 die "Only one '*' wildcard expansion ",
4571 "is supported (got $nr): '$glob'\n";
4572 } elsif ($nr == 0) {
4573 die "One '*' is needed for glob: '$glob'\n";
4574 }
4575 $re = quotemeta($left) . $re . quotemeta($right);
4576 if (length $left && !($left =~ s!/+$!!g)) {
4577 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4578 }
4579 if (length $right && !($right =~ s!^/+!!g)) {
4580 die "Missing leading '/' on right side of: '$glob' ($right)\n";
4581 }
4582 my $left_re = qr/^\/\Q$left\E(\/|$)/;
4583 bless { left => $left, right => $right, left_regex => $left_re,
4584 regex => qr/$re/, glob => $glob }, $class;
4585 }
4586
4587 sub full_path {
4588 my ($self, $path) = @_;
4589 return (length $self->{left} ? "$self->{left}/" : '') .
4590 $path . (length $self->{right} ? "/$self->{right}" : '');
4591 }
4592
4593 __END__
4594
4595 Data structures:
4596
4597
4598 $remotes = { # returned by read_all_remotes()
4599 'svn' => {
4600 # svn-remote.svn.url=https://svn.musicpd.org
4601 url => 'https://svn.musicpd.org',
4602 # svn-remote.svn.fetch=mpd/trunk:trunk
4603 fetch => {
4604 'mpd/trunk' => 'trunk',
4605 },
4606 # svn-remote.svn.tags=mpd/tags/*:tags/*
4607 tags => {
4608 path => {
4609 left => 'mpd/tags',
4610 right => '',
4611 regex => qr!mpd/tags/([^/]+)$!,
4612 glob => 'tags/*',
4613 },
4614 ref => {
4615 left => 'tags',
4616 right => '',
4617 regex => qr!tags/([^/]+)$!,
4618 glob => 'tags/*',
4619 },
4620 }
4621 }
4622 };
4623
4624 $log_entry hashref as returned by libsvn_log_entry()
4625 {
4626 log => 'whitespace-formatted log entry
4627 ', # trailing newline is preserved
4628 revision => '8', # integer
4629 date => '2004-02-24T17:01:44.108345Z', # commit date
4630 author => 'committer name'
4631 };
4632
4633
4634 # this is generated by generate_diff();
4635 @mods = array of diff-index line hashes, each element represents one line
4636 of diff-index output
4637
4638 diff-index line ($m hash)
4639 {
4640 mode_a => first column of diff-index output, no leading ':',
4641 mode_b => second column of diff-index output,
4642 sha1_b => sha1sum of the final blob,
4643 chg => change type [MCRADT],
4644 file_a => original file name of a file (iff chg is 'C' or 'R')
4645 file_b => new/current file name of a file (any chg)
4646 }
4647 ;
4648
4649 # retval of read_url_paths{,_all}();
4650 $l_map = {
4651 # repository root url
4652 'https://svn.musicpd.org' => {
4653 # repository path # GIT_SVN_ID
4654 'mpd/trunk' => 'trunk',
4655 'mpd/tags/0.11.5' => 'tags/0.11.5',
4656 },
4657 }
4658
4659 Notes:
4660 I don't trust the each() function on unless I created %hash myself
4661 because the internal iterator may not have started at base.
knittl's private git fork