?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-svn: add support for pulling author from From: and Signed-off-by:
[git/jnareb-git.git] / git-svn.perl
blob61f0c1a58aa9df44dbc15e69686770704e41de3a
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 IO::File qw//;
39 use File::Basename qw/dirname basename/;
40 use File::Path qw/mkpath/;
41 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
42 use IPC::Open3;
43 use Git;
44
45 BEGIN {
46 # import functions from Git into our packages, en masse
47 no strict 'refs';
48 foreach (qw/command command_oneline command_noisy command_output_pipe
49 command_input_pipe command_close_pipe/) {
50 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
51 Git::SVN::Migration Git::SVN::Log Git::SVN
52 Git::SVN::Util),
53 __PACKAGE__) {
54 *{"${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 'multi-fetch' => [ \&cmd_multi_fetch,
147 "Deprecated alias for $0 fetch --all",
148 { 'revision|r=s' => \$_revision, %fc_opts } ],
149 'migrate' => [ sub { },
150 # no-op, we automatically run this anyways,
151 'Migrate configuration/metadata/layout from
152 previous versions of git-svn',
153 { 'minimize' => \$Git::SVN::Migration::_minimize,
154 %remote_opts } ],
155 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
156 { 'limit=i' => \$Git::SVN::Log::limit,
157 'revision|r=s' => \$_revision,
158 'verbose|v' => \$Git::SVN::Log::verbose,
159 'incremental' => \$Git::SVN::Log::incremental,
160 'oneline' => \$Git::SVN::Log::oneline,
161 'show-commit' => \$Git::SVN::Log::show_commit,
162 'non-recursive' => \$Git::SVN::Log::non_recursive,
163 'authors-file|A=s' => \$_authors,
164 'color' => \$Git::SVN::Log::color,
165 'pager=s' => \$Git::SVN::Log::pager
166 } ],
167 'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
168 {} ],
169 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
170 { 'merge|m|M' => \$_merge,
171 'verbose|v' => \$_verbose,
172 'strategy|s=s' => \$_strategy,
173 'local|l' => \$_local,
174 'fetch-all|all' => \$_fetch_all,
175 %fc_opts } ],
176 'commit-diff' => [ \&cmd_commit_diff,
177 'Commit a diff between two trees',
178 { 'message|m=s' => \$_message,
179 'file|F=s' => \$_file,
180 'revision|r=s' => \$_revision,
181 %cmt_opts } ],
182 'info' => [ \&cmd_info,
183 "Show info about the latest SVN revision
184 on the current branch",
185 { 'url' => \$_url, } ],
186 );
187
188 my $cmd;
189 for (my $i = 0; $i < @ARGV; $i++) {
190 if (defined $cmd{$ARGV[$i]}) {
191 $cmd = $ARGV[$i];
192 splice @ARGV, $i, 1;
193 last;
194 }
195 };
196
197 # make sure we're always running
198 unless ($cmd =~ /(?:clone|init|multi-init)$/) {
199 unless (-d $ENV{GIT_DIR}) {
200 if ($git_dir_user_set) {
201 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
202 "but it is not a directory\n";
203 }
204 my $git_dir = delete $ENV{GIT_DIR};
205 chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
206 unless (length $cdup) {
207 die "Already at toplevel, but $git_dir ",
208 "not found '$cdup'\n";
209 }
210 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
211 unless (-d $git_dir) {
212 die "$git_dir still not found after going to ",
213 "'$cdup'\n";
214 }
215 $ENV{GIT_DIR} = $git_dir;
216 }
217 }
218
219 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
220
221 read_repo_config(\%opts);
222 Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
223 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
224 'minimize-connections' => \$Git::SVN::Migration::_minimize,
225 'id|i=s' => \$Git::SVN::default_ref_id,
226 'svn-remote|remote|R=s' => sub {
227 $Git::SVN::no_reuse_existing = 1;
228 $Git::SVN::default_repo_id = $_[1] });
229 exit 1 if (!$rv && $cmd && $cmd ne 'log');
230
231 usage(0) if $_help;
232 version() if $_version;
233 usage(1) unless defined $cmd;
234 load_authors() if $_authors;
235
236 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
237 Git::SVN::Migration::migration_check();
238 }
239 Git::SVN::init_vars();
240 eval {
241 Git::SVN::verify_remotes_sanity();
242 $cmd{$cmd}->[0]->(@ARGV);
243 };
244 fatal $@ if $@;
245 post_fetch_checkout();
246 exit 0;
247
248 ####################### primary functions ######################
249 sub usage {
250 my $exit = shift || 0;
251 my $fd = $exit ? \*STDERR : \*STDOUT;
252 print $fd <<"";
253 git-svn - bidirectional operations between a single Subversion tree and git
254 Usage: $0 <command> [options] [arguments]\n
255
256 print $fd "Available commands:\n" unless $cmd;
257
258 foreach (sort keys %cmd) {
259 next if $cmd && $cmd ne $_;
260 next if /^multi-/; # don't show deprecated commands
261 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
262 foreach (sort keys %{$cmd{$_}->[2]}) {
263 # mixed-case options are for .git/config only
264 next if /[A-Z]/ && /^[a-z]+$/i;
265 # prints out arguments as they should be passed:
266 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
267 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
268 "--$_" : "-$_" }
269 split /\|/,$_)," $x\n";
270 }
271 }
272 print $fd <<"";
273 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
274 arbitrary identifier if you're tracking multiple SVN branches/repositories in
275 one git repository and want to keep them separate. See git-svn(1) for more
276 information.
277
278 exit $exit;
279 }
280
281 sub version {
282 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
283 exit 0;
284 }
285
286 sub do_git_init_db {
287 unless (-d $ENV{GIT_DIR}) {
288 my @init_db = ('init');
289 push @init_db, "--template=$_template" if defined $_template;
290 if (defined $_shared) {
291 if ($_shared =~ /[a-z]/) {
292 push @init_db, "--shared=$_shared";
293 } else {
294 push @init_db, "--shared";
295 }
296 }
297 command_noisy(@init_db);
298 }
299 my $set;
300 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
301 foreach my $i (keys %icv) {
302 die "'$set' and '$i' cannot both be set\n" if $set;
303 next unless defined $icv{$i};
304 command_noisy('config', "$pfx.$i", $icv{$i});
305 $set = $i;
306 }
307 }
308
309 sub init_subdir {
310 my $repo_path = shift or return;
311 mkpath([$repo_path]) unless -d $repo_path;
312 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
313 $ENV{GIT_DIR} = '.git';
314 }
315
316 sub cmd_clone {
317 my ($url, $path) = @_;
318 if (!defined $path &&
319 (defined $_trunk || defined $_branches || defined $_tags ||
320 defined $_stdlayout) &&
321 $url !~ m#^[a-z\+]+://#) {
322 $path = $url;
323 }
324 $path = basename($url) if !defined $path || !length $path;
325 cmd_init($url, $path);
326 Git::SVN::fetch_all($Git::SVN::default_repo_id);
327 }
328
329 sub cmd_init {
330 if (defined $_stdlayout) {
331 $_trunk = 'trunk' if (!defined $_trunk);
332 $_tags = 'tags' if (!defined $_tags);
333 $_branches = 'branches' if (!defined $_branches);
334 }
335 if (defined $_trunk || defined $_branches || defined $_tags) {
336 return cmd_multi_init(@_);
337 }
338 my $url = shift or die "SVN repository location required ",
339 "as a command-line argument\n";
340 init_subdir(@_);
341 do_git_init_db();
342
343 Git::SVN->init($url);
344 }
345
346 sub cmd_fetch {
347 if (grep /^\d+=./, @_) {
348 die "'<rev>=<commit>' fetch arguments are ",
349 "no longer supported.\n";
350 }
351 my ($remote) = @_;
352 if (@_ > 1) {
353 die "Usage: $0 fetch [--all] [svn-remote]\n";
354 }
355 $remote ||= $Git::SVN::default_repo_id;
356 if ($_fetch_all) {
357 cmd_multi_fetch();
358 } else {
359 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
360 }
361 }
362
363 sub cmd_set_tree {
364 my (@commits) = @_;
365 if ($_stdin || !@commits) {
366 print "Reading from stdin...\n";
367 @commits = ();
368 while (<STDIN>) {
369 if (/\b($sha1_short)\b/o) {
370 unshift @commits, $1;
371 }
372 }
373 }
374 my @revs;
375 foreach my $c (@commits) {
376 my @tmp = command('rev-parse',$c);
377 if (scalar @tmp == 1) {
378 push @revs, $tmp[0];
379 } elsif (scalar @tmp > 1) {
380 push @revs, reverse(command('rev-list',@tmp));
381 } else {
382 fatal "Failed to rev-parse $c";
383 }
384 }
385 my $gs = Git::SVN->new;
386 my ($r_last, $cmt_last) = $gs->last_rev_commit;
387 $gs->fetch;
388 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
389 fatal "There are new revisions that were fetched ",
390 "and need to be merged (or acknowledged) ",
391 "before committing.\nlast rev: $r_last\n",
392 " current: $gs->{last_rev}";
393 }
394 $gs->set_tree($_) foreach @revs;
395 print "Done committing ",scalar @revs," revisions to SVN\n";
396 }
397
398 sub cmd_dcommit {
399 my $head = shift;
400 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
401 'Cannot dcommit with a dirty index. Commit your changes first, '
402 . "or stash them with `git stash'.\n";
403 $head ||= 'HEAD';
404 my @refs;
405 my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
406 print "Committing to $url ...\n";
407 unless ($gs) {
408 die "Unable to determine upstream SVN information from ",
409 "$head history\n";
410 }
411 my $last_rev;
412 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
413 if ($_no_rebase && scalar(@$linear_refs) > 1) {
414 warn "Attempting to commit more than one change while ",
415 "--no-rebase is enabled.\n",
416 "If these changes depend on each other, re-running ",
417 "without --no-rebase will be required."
418 }
419 while (1) {
420 my $d = shift @$linear_refs or last;
421 unless (defined $last_rev) {
422 (undef, $last_rev, undef) = cmt_metadata("$d~1");
423 unless (defined $last_rev) {
424 fatal "Unable to extract revision information ",
425 "from commit $d~1";
426 }
427 }
428 if ($_dry_run) {
429 print "diff-tree $d~1 $d\n";
430 } else {
431 my $cmt_rev;
432 my %ed_opts = ( r => $last_rev,
433 log => get_commit_entry($d)->{log},
434 ra => Git::SVN::Ra->new($gs->full_url),
435 config => SVN::Core::config_get_config(
436 $Git::SVN::Ra::config_dir
437 ),
438 tree_a => "$d~1",
439 tree_b => $d,
440 editor_cb => sub {
441 print "Committed r$_[0]\n";
442 $cmt_rev = $_[0];
443 },
444 svn_path => '');
445 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
446 print "No changes\n$d~1 == $d\n";
447 } elsif ($parents->{$d} && @{$parents->{$d}}) {
448 $gs->{inject_parents_dcommit}->{$cmt_rev} =
449 $parents->{$d};
450 }
451 $_fetch_all ? $gs->fetch_all : $gs->fetch;
452 next if $_no_rebase;
453
454 # we always want to rebase against the current HEAD,
455 # not any head that was passed to us
456 my @diff = command('diff-tree', $d,
457 $gs->refname, '--');
458 my @finish;
459 if (@diff) {
460 @finish = rebase_cmd();
461 print STDERR "W: $d and ", $gs->refname,
462 " differ, using @finish:\n",
463 join("\n", @diff), "\n";
464 } else {
465 print "No changes between current HEAD and ",
466 $gs->refname,
467 "\nResetting to the latest ",
468 $gs->refname, "\n";
469 @finish = qw/reset --mixed/;
470 }
471 command_noisy(@finish, $gs->refname);
472 if (@diff) {
473 @refs = ();
474 my ($url_, $rev_, $uuid_, $gs_) =
475 working_head_info($head, \@refs);
476 my ($linear_refs_, $parents_) =
477 linearize_history($gs_, \@refs);
478 if (scalar(@$linear_refs) !=
479 scalar(@$linear_refs_)) {
480 fatal "# of revisions changed ",
481 "\nbefore:\n",
482 join("\n", @$linear_refs),
483 "\n\nafter:\n",
484 join("\n", @$linear_refs_), "\n",
485 'If you are attempting to commit ',
486 "merges, try running:\n\t",
487 'git rebase --interactive',
488 '--preserve-merges ',
489 $gs->refname,
490 "\nBefore dcommitting";
491 }
492 if ($url_ ne $url) {
493 fatal "URL mismatch after rebase: ",
494 "$url_ != $url";
495 }
496 if ($uuid_ ne $uuid) {
497 fatal "uuid mismatch after rebase: ",
498 "$uuid_ != $uuid";
499 }
500 # remap parents
501 my (%p, @l, $i);
502 for ($i = 0; $i < scalar @$linear_refs; $i++) {
503 my $new = $linear_refs_->[$i] or next;
504 $p{$new} =
505 $parents->{$linear_refs->[$i]};
506 push @l, $new;
507 }
508 $parents = \%p;
509 $linear_refs = \@l;
510 }
511 $last_rev = $cmt_rev;
512 }
513 }
514 }
515
516 sub cmd_find_rev {
517 my $revision_or_hash = shift;
518 my $result;
519 if ($revision_or_hash =~ /^r\d+$/) {
520 my $head = shift;
521 $head ||= 'HEAD';
522 my @refs;
523 my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
524 unless ($gs) {
525 die "Unable to determine upstream SVN information from ",
526 "$head history\n";
527 }
528 my $desired_revision = substr($revision_or_hash, 1);
529 $result = $gs->rev_db_get($desired_revision);
530 } else {
531 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
532 $result = $rev;
533 }
534 print "$result\n" if $result;
535 }
536
537 sub cmd_rebase {
538 command_noisy(qw/update-index --refresh/);
539 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
540 unless ($gs) {
541 die "Unable to determine upstream SVN information from ",
542 "working tree history\n";
543 }
544 if (command(qw/diff-index HEAD --/)) {
545 print STDERR "Cannot rebase with uncommited changes:\n";
546 command_noisy('status');
547 exit 1;
548 }
549 unless ($_local) {
550 $_fetch_all ? $gs->fetch_all : $gs->fetch;
551 }
552 command_noisy(rebase_cmd(), $gs->refname);
553 }
554
555 sub cmd_show_ignore {
556 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
557 $gs ||= Git::SVN->new;
558 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
559 $gs->prop_walk($gs->{path}, $r, sub {
560 my ($gs, $path, $props) = @_;
561 print STDOUT "\n# $path\n";
562 my $s = $props->{'svn:ignore'} or return;
563 $s =~ s/[\r\n]+/\n/g;
564 chomp $s;
565 $s =~ s#^#$path#gm;
566 print STDOUT "$s\n";
567 });
568 }
569
570 sub cmd_create_ignore {
571 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
572 $gs ||= Git::SVN->new;
573 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
574 $gs->prop_walk($gs->{path}, $r, sub {
575 my ($gs, $path, $props) = @_;
576 # $path is of the form /path/to/dir/
577 my $ignore = '.' . $path . '.gitignore';
578 my $s = $props->{'svn:ignore'} or return;
579 open(GITIGNORE, '>', $ignore)
580 or fatal("Failed to open `$ignore' for writing: $!");
581 $s =~ s/[\r\n]+/\n/g;
582 chomp $s;
583 # Prefix all patterns so that the ignore doesn't apply
584 # to sub-directories.
585 $s =~ s#^#/#gm;
586 print GITIGNORE "$s\n";
587 close(GITIGNORE)
588 or fatal("Failed to close `$ignore': $!");
589 command_noisy('add', $ignore);
590 });
591 }
592
593 sub canonicalize_path {
594 my ($path) = @_;
595 my $dot_slash_added = 0;
596 if (substr($path, 0, 1) ne "/") {
597 $path = "./" . $path;
598 $dot_slash_added = 1;
599 }
600 # File::Spec->canonpath doesn't collapse x/../y into y (for a
601 # good reason), so let's do this manually.
602 $path =~ s#/+#/#g;
603 $path =~ s#/\.(?:/|$)#/#g;
604 $path =~ s#/[^/]+/\.\.##g;
605 $path =~ s#/$##g;
606 $path =~ s#^\./## if $dot_slash_added;
607 return $path;
608 }
609
610 # get_svnprops(PATH)
611 # ------------------
612 # Helper for cmd_propget and cmd_proplist below.
613 sub get_svnprops {
614 my $path = shift;
615 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
616 $gs ||= Git::SVN->new;
617
618 # prefix THE PATH by the sub-directory from which the user
619 # invoked us.
620 $path = $cmd_dir_prefix . $path;
621 fatal("No such file or directory: $path") unless -e $path;
622 my $is_dir = -d $path ? 1 : 0;
623 $path = $gs->{path} . '/' . $path;
624
625 # canonicalize the path (otherwise libsvn will abort or fail to
626 # find the file)
627 $path = canonicalize_path($path);
628
629 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
630 my $props;
631 if ($is_dir) {
632 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
633 }
634 else {
635 (undef, $props) = $gs->ra->get_file($path, $r, undef);
636 }
637 return $props;
638 }
639
640 # cmd_propget (PROP, PATH)
641 # ------------------------
642 # Print the SVN property PROP for PATH.
643 sub cmd_propget {
644 my ($prop, $path) = @_;
645 $path = '.' if not defined $path;
646 usage(1) if not defined $prop;
647 my $props = get_svnprops($path);
648 if (not defined $props->{$prop}) {
649 fatal("`$path' does not have a `$prop' SVN property.");
650 }
651 print $props->{$prop} . "\n";
652 }
653
654 # cmd_proplist (PATH)
655 # -------------------
656 # Print the list of SVN properties for PATH.
657 sub cmd_proplist {
658 my $path = shift;
659 $path = '.' if not defined $path;
660 my $props = get_svnprops($path);
661 print "Properties on '$path':\n";
662 foreach (sort keys %{$props}) {
663 print " $_\n";
664 }
665 }
666
667 sub cmd_multi_init {
668 my $url = shift;
669 unless (defined $_trunk || defined $_branches || defined $_tags) {
670 usage(1);
671 }
672
673 # there are currently some bugs that prevent multi-init/multi-fetch
674 # setups from working well without this.
675 $Git::SVN::_minimize_url = 1;
676
677 $_prefix = '' unless defined $_prefix;
678 if (defined $url) {
679 $url =~ s#/+$##;
680 init_subdir(@_);
681 }
682 do_git_init_db();
683 if (defined $_trunk) {
684 my $trunk_ref = $_prefix . 'trunk';
685 # try both old-style and new-style lookups:
686 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
687 unless ($gs_trunk) {
688 my ($trunk_url, $trunk_path) =
689 complete_svn_url($url, $_trunk);
690 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
691 undef, $trunk_ref);
692 }
693 }
694 return unless defined $_branches || defined $_tags;
695 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
696 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
697 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
698 }
699
700 sub cmd_multi_fetch {
701 my $remotes = Git::SVN::read_all_remotes();
702 foreach my $repo_id (sort keys %$remotes) {
703 if ($remotes->{$repo_id}->{url}) {
704 Git::SVN::fetch_all($repo_id, $remotes);
705 }
706 }
707 }
708
709 # this command is special because it requires no metadata
710 sub cmd_commit_diff {
711 my ($ta, $tb, $url) = @_;
712 my $usage = "Usage: $0 commit-diff -r<revision> ".
713 "<tree-ish> <tree-ish> [<URL>]";
714 fatal($usage) if (!defined $ta || !defined $tb);
715 my $svn_path;
716 if (!defined $url) {
717 my $gs = eval { Git::SVN->new };
718 if (!$gs) {
719 fatal("Needed URL or usable git-svn --id in ",
720 "the command-line\n", $usage);
721 }
722 $url = $gs->{url};
723 $svn_path = $gs->{path};
724 }
725 unless (defined $_revision) {
726 fatal("-r|--revision is a required argument\n", $usage);
727 }
728 if (defined $_message && defined $_file) {
729 fatal("Both --message/-m and --file/-F specified ",
730 "for the commit message.\n",
731 "I have no idea what you mean");
732 }
733 if (defined $_file) {
734 $_message = file_to_s($_file);
735 } else {
736 $_message ||= get_commit_entry($tb)->{log};
737 }
738 my $ra ||= Git::SVN::Ra->new($url);
739 $svn_path ||= $ra->{svn_path};
740 my $r = $_revision;
741 if ($r eq 'HEAD') {
742 $r = $ra->get_latest_revnum;
743 } elsif ($r !~ /^\d+$/) {
744 die "revision argument: $r not understood by git-svn\n";
745 }
746 my %ed_opts = ( r => $r,
747 log => $_message,
748 ra => $ra,
749 tree_a => $ta,
750 tree_b => $tb,
751 editor_cb => sub { print "Committed r$_[0]\n" },
752 svn_path => $svn_path );
753 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
754 print "No changes\n$ta == $tb\n";
755 }
756 }
757
758 sub cmd_info {
759 my $path = canonicalize_path(shift or ".");
760 unless (scalar(@_) == 0) {
761 die "Too many arguments specified\n";
762 }
763
764 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
765
766 if (!$file_type && !$diff_status) {
767 print STDERR "$path: (Not a versioned resource)\n\n";
768 return;
769 }
770
771 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
772 unless ($gs) {
773 die "Unable to determine upstream SVN information from ",
774 "working tree history\n";
775 }
776 my $full_url = $url . ($path eq "." ? "" : "/$path");
777
778 if ($_url) {
779 print $full_url, "\n";
780 return;
781 }
782
783 my $result = "Path: $path\n";
784 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
785 $result .= "URL: " . $full_url . "\n";
786
787 eval {
788 my $repos_root = $gs->repos_root;
789 Git::SVN::remove_username($repos_root);
790 $result .= "Repository Root: $repos_root\n";
791 };
792 if ($@) {
793 $result .= "Repository Root: (offline)\n";
794 }
795 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
796 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
797
798 $result .= "Node Kind: " .
799 ($file_type eq "dir" ? "directory" : "file") . "\n";
800
801 my $schedule = $diff_status eq "A"
802 ? "add"
803 : ($diff_status eq "D" ? "delete" : "normal");
804 $result .= "Schedule: $schedule\n";
805
806 if ($diff_status eq "A") {
807 print $result, "\n";
808 return;
809 }
810
811 my ($lc_author, $lc_rev, $lc_date_utc);
812 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $path);
813 my $log = command_output_pipe(@args);
814 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
815 while (<$log>) {
816 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
817 $lc_author = $1;
818 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
819 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
820 (undef, $lc_rev, undef) = ::extract_metadata($1);
821 }
822 }
823 close $log;
824
825 Git::SVN::Log::set_local_timezone();
826
827 $result .= "Last Changed Author: $lc_author\n";
828 $result .= "Last Changed Rev: $lc_rev\n";
829 $result .= "Last Changed Date: " .
830 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
831
832 if ($file_type ne "dir") {
833 my $text_last_updated_date =
834 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
835 $result .=
836 "Text Last Updated: " .
837 Git::SVN::Log::format_svn_date($text_last_updated_date) .
838 "\n";
839 my $checksum;
840 if ($diff_status eq "D") {
841 my ($fh, $ctx) =
842 command_output_pipe(qw(cat-file blob), "HEAD:$path");
843 if ($file_type eq "link") {
844 my $file_name = <$fh>;
845 $checksum = Git::SVN::Util::md5sum("link $file_name");
846 } else {
847 $checksum = Git::SVN::Util::md5sum($fh);
848 }
849 command_close_pipe($fh, $ctx);
850 } elsif ($file_type eq "link") {
851 my $file_name =
852 command(qw(cat-file blob), "HEAD:$path");
853 $checksum =
854 Git::SVN::Util::md5sum("link " . $file_name);
855 } else {
856 open FILE, "<", $path or die $!;
857 $checksum = Git::SVN::Util::md5sum(\*FILE);
858 close FILE or die $!;
859 }
860 $result .= "Checksum: " . $checksum . "\n";
861 }
862
863 print $result, "\n";
864 }
865
866 ########################### utility functions #########################
867
868 sub rebase_cmd {
869 my @cmd = qw/rebase/;
870 push @cmd, '-v' if $_verbose;
871 push @cmd, qw/--merge/ if $_merge;
872 push @cmd, "--strategy=$_strategy" if $_strategy;
873 @cmd;
874 }
875
876 sub post_fetch_checkout {
877 return if $_no_checkout;
878 my $gs = $Git::SVN::_head or return;
879 return if verify_ref('refs/heads/master^0');
880
881 my $valid_head = verify_ref('HEAD^0');
882 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
883 return if ($valid_head || !verify_ref('HEAD^0'));
884
885 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
886 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
887 return if -f $index;
888
889 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
890 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
891 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
892 print STDERR "Checked out HEAD:\n ",
893 $gs->full_url, " r", $gs->last_rev, "\n";
894 }
895
896 sub complete_svn_url {
897 my ($url, $path) = @_;
898 $path =~ s#/+$##;
899 if ($path !~ m#^[a-z\+]+://#) {
900 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
901 fatal("E: '$path' is not a complete URL ",
902 "and a separate URL is not specified");
903 }
904 return ($url, $path);
905 }
906 return ($path, '');
907 }
908
909 sub complete_url_ls_init {
910 my ($ra, $repo_path, $switch, $pfx) = @_;
911 unless ($repo_path) {
912 print STDERR "W: $switch not specified\n";
913 return;
914 }
915 $repo_path =~ s#/+$##;
916 if ($repo_path =~ m#^[a-z\+]+://#) {
917 $ra = Git::SVN::Ra->new($repo_path);
918 $repo_path = '';
919 } else {
920 $repo_path =~ s#^/+##;
921 unless ($ra) {
922 fatal("E: '$repo_path' is not a complete URL ",
923 "and a separate URL is not specified");
924 }
925 }
926 my $url = $ra->{url};
927 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
928 my $k = "svn-remote.$gs->{repo_id}.url";
929 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
930 if ($orig_url && ($orig_url ne $gs->{url})) {
931 die "$k already set: $orig_url\n",
932 "wanted to set to: $gs->{url}\n";
933 }
934 command_oneline('config', $k, $gs->{url}) unless $orig_url;
935 my $remote_path = "$ra->{svn_path}/$repo_path/*";
936 $remote_path =~ s#/+#/#g;
937 $remote_path =~ s#^/##g;
938 my ($n) = ($switch =~ /^--(\w+)/);
939 if (length $pfx && $pfx !~ m#/$#) {
940 die "--prefix='$pfx' must have a trailing slash '/'\n";
941 }
942 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
943 "$remote_path:refs/remotes/$pfx*");
944 }
945
946 sub verify_ref {
947 my ($ref) = @_;
948 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
949 { STDERR => 0 }); };
950 }
951
952 sub get_tree_from_treeish {
953 my ($treeish) = @_;
954 # $treeish can be a symbolic ref, too:
955 my $type = command_oneline(qw/cat-file -t/, $treeish);
956 my $expected;
957 while ($type eq 'tag') {
958 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
959 }
960 if ($type eq 'commit') {
961 $expected = (grep /^tree /, command(qw/cat-file commit/,
962 $treeish))[0];
963 ($expected) = ($expected =~ /^tree ($sha1)$/o);
964 die "Unable to get tree from $treeish\n" unless $expected;
965 } elsif ($type eq 'tree') {
966 $expected = $treeish;
967 } else {
968 die "$treeish is a $type, expected tree, tag or commit\n";
969 }
970 return $expected;
971 }
972
973 sub get_commit_entry {
974 my ($treeish) = shift;
975 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
976 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
977 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
978 open my $log_fh, '>', $commit_editmsg or croak $!;
979
980 my $type = command_oneline(qw/cat-file -t/, $treeish);
981 if ($type eq 'commit' || $type eq 'tag') {
982 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
983 $type, $treeish);
984 my $in_msg = 0;
985 while (<$msg_fh>) {
986 if (!$in_msg) {
987 $in_msg = 1 if (/^\s*$/);
988 } elsif (/^git-svn-id: /) {
989 # skip this for now, we regenerate the
990 # correct one on re-fetch anyways
991 # TODO: set *:merge properties or like...
992 } else {
993 print $log_fh $_ or croak $!;
994 }
995 }
996 command_close_pipe($msg_fh, $ctx);
997 }
998 close $log_fh or croak $!;
999
1000 if ($_edit || ($type eq 'tree')) {
1001 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1002 # TODO: strip out spaces, comments, like git-commit.sh
1003 system($editor, $commit_editmsg);
1004 }
1005 rename $commit_editmsg, $commit_msg or croak $!;
1006 open $log_fh, '<', $commit_msg or croak $!;
1007 { local $/; chomp($log_entry{log} = <$log_fh>); }
1008 close $log_fh or croak $!;
1009 unlink $commit_msg;
1010 \%log_entry;
1011 }
1012
1013 sub s_to_file {
1014 my ($str, $file, $mode) = @_;
1015 open my $fd,'>',$file or croak $!;
1016 print $fd $str,"\n" or croak $!;
1017 close $fd or croak $!;
1018 chmod ($mode &~ umask, $file) if (defined $mode);
1019 }
1020
1021 sub file_to_s {
1022 my $file = shift;
1023 open my $fd,'<',$file or croak "$!: file: $file\n";
1024 local $/;
1025 my $ret = <$fd>;
1026 close $fd or croak $!;
1027 $ret =~ s/\s*$//s;
1028 return $ret;
1029 }
1030
1031 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1032 sub load_authors {
1033 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1034 my $log = $cmd eq 'log';
1035 while (<$authors>) {
1036 chomp;
1037 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1038 my ($user, $name, $email) = ($1, $2, $3);
1039 if ($log) {
1040 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1041 } else {
1042 $users{$user} = [$name, $email];
1043 }
1044 }
1045 close $authors or croak $!;
1046 }
1047
1048 # convert GetOpt::Long specs for use by git-config
1049 sub read_repo_config {
1050 return unless -d $ENV{GIT_DIR};
1051 my $opts = shift;
1052 my @config_only;
1053 foreach my $o (keys %$opts) {
1054 # if we have mixedCase and a long option-only, then
1055 # it's a config-only variable that we don't need for
1056 # the command-line.
1057 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1058 my $v = $opts->{$o};
1059 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1060 $key =~ s/-//g;
1061 my $arg = 'git-config';
1062 $arg .= ' --int' if ($o =~ /[:=]i$/);
1063 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1064 if (ref $v eq 'ARRAY') {
1065 chomp(my @tmp = `$arg --get-all svn.$key`);
1066 @$v = @tmp if @tmp;
1067 } else {
1068 chomp(my $tmp = `$arg --get svn.$key`);
1069 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1070 $$v = $tmp;
1071 }
1072 }
1073 }
1074 delete @$opts{@config_only} if @config_only;
1075 }
1076
1077 sub extract_metadata {
1078 my $id = shift or return (undef, undef, undef);
1079 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1080 \s([a-f\d\-]+)$/x);
1081 if (!defined $rev || !$uuid || !$url) {
1082 # some of the original repositories I made had
1083 # identifiers like this:
1084 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1085 }
1086 return ($url, $rev, $uuid);
1087 }
1088
1089 sub cmt_metadata {
1090 return extract_metadata((grep(/^git-svn-id: /,
1091 command(qw/cat-file commit/, shift)))[-1]);
1092 }
1093
1094 sub working_head_info {
1095 my ($head, $refs) = @_;
1096 my @args = ('log', '--no-color', '--first-parent');
1097 my ($fh, $ctx) = command_output_pipe(@args, $head);
1098 my $hash;
1099 my %max;
1100 while (<$fh>) {
1101 if ( m{^commit ($::sha1)$} ) {
1102 unshift @$refs, $hash if $hash and $refs;
1103 $hash = $1;
1104 next;
1105 }
1106 next unless s{^\s*(git-svn-id:)}{$1};
1107 my ($url, $rev, $uuid) = extract_metadata($_);
1108 if (defined $url && defined $rev) {
1109 next if $max{$url} and $max{$url} < $rev;
1110 if (my $gs = Git::SVN->find_by_url($url)) {
1111 my $c = $gs->rev_db_get($rev);
1112 if ($c && $c eq $hash) {
1113 close $fh; # break the pipe
1114 return ($url, $rev, $uuid, $gs);
1115 } else {
1116 $max{$url} ||= $gs->rev_db_max;
1117 }
1118 }
1119 }
1120 }
1121 command_close_pipe($fh, $ctx);
1122 (undef, undef, undef, undef);
1123 }
1124
1125 sub read_commit_parents {
1126 my ($parents, $c) = @_;
1127 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1128 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1129 @{$parents->{$c}} = split(/ /, $p);
1130 }
1131
1132 sub linearize_history {
1133 my ($gs, $refs) = @_;
1134 my %parents;
1135 foreach my $c (@$refs) {
1136 read_commit_parents(\%parents, $c);
1137 }
1138
1139 my @linear_refs;
1140 my %skip = ();
1141 my $last_svn_commit = $gs->last_commit;
1142 foreach my $c (reverse @$refs) {
1143 next if $c eq $last_svn_commit;
1144 last if $skip{$c};
1145
1146 unshift @linear_refs, $c;
1147 $skip{$c} = 1;
1148
1149 # we only want the first parent to diff against for linear
1150 # history, we save the rest to inject when we finalize the
1151 # svn commit
1152 my $fp_a = verify_ref("$c~1");
1153 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1154 if (!$fp_a || !$fp_b) {
1155 die "Commit $c\n",
1156 "has no parent commit, and therefore ",
1157 "nothing to diff against.\n",
1158 "You should be working from a repository ",
1159 "originally created by git-svn\n";
1160 }
1161 if ($fp_a ne $fp_b) {
1162 die "$c~1 = $fp_a, however parsing commit $c ",
1163 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1164 }
1165
1166 foreach my $p (@{$parents{$c}}) {
1167 $skip{$p} = 1;
1168 }
1169 }
1170 (\@linear_refs, \%parents);
1171 }
1172
1173 sub find_file_type_and_diff_status {
1174 my ($path) = @_;
1175 return ('dir', '') if $path eq '.';
1176
1177 my $diff_output =
1178 command_oneline(qw(diff --cached --name-status --), $path) || "";
1179 my $diff_status = (split(' ', $diff_output))[0] || "";
1180
1181 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1182
1183 return (undef, undef) if !$diff_status && !$ls_tree;
1184
1185 if ($diff_status eq "A") {
1186 return ("link", $diff_status) if -l $path;
1187 return ("dir", $diff_status) if -d $path;
1188 return ("file", $diff_status);
1189 }
1190
1191 my $mode = (split(' ', $ls_tree))[0] || "";
1192
1193 return ("link", $diff_status) if $mode eq "120000";
1194 return ("dir", $diff_status) if $mode eq "040000";
1195 return ("file", $diff_status);
1196 }
1197
1198 package Git::SVN::Util;
1199 use strict;
1200 use warnings;
1201 use Digest::MD5;
1202
1203 sub md5sum {
1204 my $arg = shift;
1205 my $ref = ref $arg;
1206 my $md5 = Digest::MD5->new();
1207 if ($ref eq 'GLOB' || $ref eq 'IO::File') {
1208 $md5->addfile($arg) or croak $!;
1209 } elsif ($ref eq 'SCALAR') {
1210 $md5->add($$arg) or croak $!;
1211 } elsif (!$ref) {
1212 $md5->add($arg) or croak $!;
1213 } else {
1214 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1215 }
1216 return $md5->hexdigest();
1217 }
1218
1219 package Git::SVN;
1220 use strict;
1221 use warnings;
1222 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1223 $_repack $_repack_flags $_use_svm_props $_head
1224 $_use_svnsync_props $no_reuse_existing $_minimize_url
1225 $_use_log_author/;
1226 use Carp qw/croak/;
1227 use File::Path qw/mkpath/;
1228 use File::Copy qw/copy/;
1229 use IPC::Open3;
1230
1231 my $_repack_nr;
1232 # properties that we do not log:
1233 my %SKIP_PROP;
1234 BEGIN {
1235 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1236 svn:special svn:executable
1237 svn:entry:committed-rev
1238 svn:entry:last-author
1239 svn:entry:uuid
1240 svn:entry:committed-date/;
1241
1242 # some options are read globally, but can be overridden locally
1243 # per [svn-remote "..."] section. Command-line options will *NOT*
1244 # override options set in an [svn-remote "..."] section
1245 no strict 'refs';
1246 for my $option (qw/follow_parent no_metadata use_svm_props
1247 use_svnsync_props/) {
1248 my $key = $option;
1249 $key =~ tr/_//d;
1250 my $prop = "-$option";
1251 *$option = sub {
1252 my ($self) = @_;
1253 return $self->{$prop} if exists $self->{$prop};
1254 my $k = "svn-remote.$self->{repo_id}.$key";
1255 eval { command_oneline(qw/config --get/, $k) };
1256 if ($@) {
1257 $self->{$prop} = ${"Git::SVN::_$option"};
1258 } else {
1259 my $v = command_oneline(qw/config --bool/,$k);
1260 $self->{$prop} = $v eq 'false' ? 0 : 1;
1261 }
1262 return $self->{$prop};
1263 }
1264 }
1265 }
1266
1267 my %LOCKFILES;
1268 END { unlink keys %LOCKFILES if %LOCKFILES }
1269
1270 sub resolve_local_globs {
1271 my ($url, $fetch, $glob_spec) = @_;
1272 return unless defined $glob_spec;
1273 my $ref = $glob_spec->{ref};
1274 my $path = $glob_spec->{path};
1275 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1276 next unless m#^refs/remotes/$ref->{regex}$#;
1277 my $p = $1;
1278 my $pathname = desanitize_refname($path->full_path($p));
1279 my $refname = desanitize_refname($ref->full_path($p));
1280 if (my $existing = $fetch->{$pathname}) {
1281 if ($existing ne $refname) {
1282 die "Refspec conflict:\n",
1283 "existing: refs/remotes/$existing\n",
1284 " globbed: refs/remotes/$refname\n";
1285 }
1286 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1287 $u =~ s!^\Q$url\E(/|$)!! or die
1288 "refs/remotes/$refname: '$url' not found in '$u'\n";
1289 if ($pathname ne $u) {
1290 warn "W: Refspec glob conflict ",
1291 "(ref: refs/remotes/$refname):\n",
1292 "expected path: $pathname\n",
1293 " real path: $u\n",
1294 "Continuing ahead with $u\n";
1295 next;
1296 }
1297 } else {
1298 $fetch->{$pathname} = $refname;
1299 }
1300 }
1301 }
1302
1303 sub parse_revision_argument {
1304 my ($base, $head) = @_;
1305 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1306 return ($base, $head);
1307 }
1308 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1309 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1310 return ($head, $head) if ($::_revision eq 'HEAD');
1311 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1312 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1313 die "revision argument: $::_revision not understood by git-svn\n";
1314 }
1315
1316 sub fetch_all {
1317 my ($repo_id, $remotes) = @_;
1318 if (ref $repo_id) {
1319 my $gs = $repo_id;
1320 $repo_id = undef;
1321 $repo_id = $gs->{repo_id};
1322 }
1323 $remotes ||= read_all_remotes();
1324 my $remote = $remotes->{$repo_id} or
1325 die "[svn-remote \"$repo_id\"] unknown\n";
1326 my $fetch = $remote->{fetch};
1327 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1328 my (@gs, @globs);
1329 my $ra = Git::SVN::Ra->new($url);
1330 my $uuid = $ra->get_uuid;
1331 my $head = $ra->get_latest_revnum;
1332 my $base = defined $fetch ? $head : 0;
1333
1334 # read the max revs for wildcard expansion (branches/*, tags/*)
1335 foreach my $t (qw/branches tags/) {
1336 defined $remote->{$t} or next;
1337 push @globs, $remote->{$t};
1338 my $max_rev = eval { tmp_config(qw/--int --get/,
1339 "svn-remote.$repo_id.${t}-maxRev") };
1340 if (defined $max_rev && ($max_rev < $base)) {
1341 $base = $max_rev;
1342 } elsif (!defined $max_rev) {
1343 $base = 0;
1344 }
1345 }
1346
1347 if ($fetch) {
1348 foreach my $p (sort keys %$fetch) {
1349 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1350 my $lr = $gs->rev_db_max;
1351 if (defined $lr) {
1352 $base = $lr if ($lr < $base);
1353 }
1354 push @gs, $gs;
1355 }
1356 }
1357
1358 ($base, $head) = parse_revision_argument($base, $head);
1359 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1360 }
1361
1362 sub read_all_remotes {
1363 my $r = {};
1364 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1365 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
1366 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1367 $local_ref =~ s{^/}{};
1368 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1369 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1370 $r->{$1}->{url} = $2;
1371 } elsif (m!^(.+)\.(branches|tags)=
1372 (.*):refs/remotes/(.+)\s*$/!x) {
1373 my ($p, $g) = ($3, $4);
1374 my $rs = $r->{$1}->{$2} = {
1375 t => $2,
1376 remote => $1,
1377 path => Git::SVN::GlobSpec->new($p),
1378 ref => Git::SVN::GlobSpec->new($g) };
1379 if (length($rs->{ref}->{right}) != 0) {
1380 die "The '*' glob character must be the last ",
1381 "character of '$g'\n";
1382 }
1383 }
1384 }
1385 $r;
1386 }
1387
1388 sub init_vars {
1389 if (defined $_repack) {
1390 $_repack = 1000 if ($_repack <= 0);
1391 $_repack_nr = $_repack;
1392 $_repack_flags ||= '-d';
1393 }
1394 }
1395
1396 sub verify_remotes_sanity {
1397 return unless -d $ENV{GIT_DIR};
1398 my %seen;
1399 foreach (command(qw/config -l/)) {
1400 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1401 if ($seen{$1}) {
1402 die "Remote ref refs/remote/$1 is tracked by",
1403 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1404 "Please resolve this ambiguity in ",
1405 "your git configuration file before ",
1406 "continuing\n";
1407 }
1408 $seen{$1} = $_;
1409 }
1410 }
1411 }
1412
1413 # we allow more chars than remotes2config.sh...
1414 sub sanitize_remote_name {
1415 my ($name) = @_;
1416 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1417 $name;
1418 }
1419
1420 sub find_existing_remote {
1421 my ($url, $remotes) = @_;
1422 return undef if $no_reuse_existing;
1423 my $existing;
1424 foreach my $repo_id (keys %$remotes) {
1425 my $u = $remotes->{$repo_id}->{url} or next;
1426 next if $u ne $url;
1427 $existing = $repo_id;
1428 last;
1429 }
1430 $existing;
1431 }
1432
1433 sub init_remote_config {
1434 my ($self, $url, $no_write) = @_;
1435 $url =~ s!/+$!!; # strip trailing slash
1436 my $r = read_all_remotes();
1437 my $existing = find_existing_remote($url, $r);
1438 if ($existing) {
1439 unless ($no_write) {
1440 print STDERR "Using existing ",
1441 "[svn-remote \"$existing\"]\n";
1442 }
1443 $self->{repo_id} = $existing;
1444 } elsif ($_minimize_url) {
1445 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1446 $existing = find_existing_remote($min_url, $r);
1447 if ($existing) {
1448 unless ($no_write) {
1449 print STDERR "Using existing ",
1450 "[svn-remote \"$existing\"]\n";
1451 }
1452 $self->{repo_id} = $existing;
1453 }
1454 if ($min_url ne $url) {
1455 unless ($no_write) {
1456 print STDERR "Using higher level of URL: ",
1457 "$url => $min_url\n";
1458 }
1459 my $old_path = $self->{path};
1460 $self->{path} = $url;
1461 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1462 if (length $old_path) {
1463 $self->{path} .= "/$old_path";
1464 }
1465 $url = $min_url;
1466 }
1467 }
1468 my $orig_url;
1469 if (!$existing) {
1470 # verify that we aren't overwriting anything:
1471 $orig_url = eval {
1472 command_oneline('config', '--get',
1473 "svn-remote.$self->{repo_id}.url")
1474 };
1475 if ($orig_url && ($orig_url ne $url)) {
1476 die "svn-remote.$self->{repo_id}.url already set: ",
1477 "$orig_url\nwanted to set to: $url\n";
1478 }
1479 }
1480 my ($xrepo_id, $xpath) = find_ref($self->refname);
1481 if (defined $xpath) {
1482 die "svn-remote.$xrepo_id.fetch already set to track ",
1483 "$xpath:refs/remotes/", $self->refname, "\n";
1484 }
1485 unless ($no_write) {
1486 command_noisy('config',
1487 "svn-remote.$self->{repo_id}.url", $url);
1488 $self->{path} =~ s{^/}{};
1489 command_noisy('config', '--add',
1490 "svn-remote.$self->{repo_id}.fetch",
1491 "$self->{path}:".$self->refname);
1492 }
1493 $self->{url} = $url;
1494 }
1495
1496 sub find_by_url { # repos_root and, path are optional
1497 my ($class, $full_url, $repos_root, $path) = @_;
1498
1499 return undef unless defined $full_url;
1500 remove_username($full_url);
1501 remove_username($repos_root) if defined $repos_root;
1502 my $remotes = read_all_remotes();
1503 if (defined $full_url && defined $repos_root && !defined $path) {
1504 $path = $full_url;
1505 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1506 }
1507 foreach my $repo_id (keys %$remotes) {
1508 my $u = $remotes->{$repo_id}->{url} or next;
1509 remove_username($u);
1510 next if defined $repos_root && $repos_root ne $u;
1511
1512 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1513 foreach (qw/branches tags/) {
1514 resolve_local_globs($u, $fetch,
1515 $remotes->{$repo_id}->{$_});
1516 }
1517 my $p = $path;
1518 unless (defined $p) {
1519 $p = $full_url;
1520 $p =~ s#^\Q$u\E(?:/|$)## or next;
1521 }
1522 foreach my $f (keys %$fetch) {
1523 next if $f ne $p;
1524 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1525 }
1526 }
1527 undef;
1528 }
1529
1530 sub init {
1531 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1532 my $self = _new($class, $repo_id, $ref_id, $path);
1533 if (defined $url) {
1534 $self->init_remote_config($url, $no_write);
1535 }
1536 $self;
1537 }
1538
1539 sub find_ref {
1540 my ($ref_id) = @_;
1541 foreach (command(qw/config -l/)) {
1542 next unless m!^svn-remote\.(.+)\.fetch=
1543 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1544 my ($repo_id, $path, $ref) = ($1, $2, $3);
1545 if ($ref eq $ref_id) {
1546 $path = '' if ($path =~ m#^\./?#);
1547 return ($repo_id, $path);
1548 }
1549 }
1550 (undef, undef, undef);
1551 }
1552
1553 sub new {
1554 my ($class, $ref_id, $repo_id, $path) = @_;
1555 if (defined $ref_id && !defined $repo_id && !defined $path) {
1556 ($repo_id, $path) = find_ref($ref_id);
1557 if (!defined $repo_id) {
1558 die "Could not find a \"svn-remote.*.fetch\" key ",
1559 "in the repository configuration matching: ",
1560 "refs/remotes/$ref_id\n";
1561 }
1562 }
1563 my $self = _new($class, $repo_id, $ref_id, $path);
1564 if (!defined $self->{path} || !length $self->{path}) {
1565 my $fetch = command_oneline('config', '--get',
1566 "svn-remote.$repo_id.fetch",
1567 ":refs/remotes/$ref_id\$") or
1568 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1569 "\":refs/remotes/$ref_id\$\" in config\n";
1570 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1571 }
1572 $self->{url} = command_oneline('config', '--get',
1573 "svn-remote.$repo_id.url") or
1574 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1575 $self->rebuild;
1576 $self;
1577 }
1578
1579 sub refname {
1580 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1581
1582 # It cannot end with a slash /, we'll throw up on this because
1583 # SVN can't have directories with a slash in their name, either:
1584 if ($refname =~ m{/$}) {
1585 die "ref: '$refname' ends with a trailing slash, this is ",
1586 "not permitted by git nor Subversion\n";
1587 }
1588
1589 # It cannot have ASCII control character space, tilde ~, caret ^,
1590 # colon :, question-mark ?, asterisk *, space, or open bracket [
1591 # anywhere.
1592 #
1593 # Additionally, % must be escaped because it is used for escaping
1594 # and we want our escaped refname to be reversible
1595 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1596
1597 # no slash-separated component can begin with a dot .
1598 # /.* becomes /%2E*
1599 $refname =~ s{/\.}{/%2E}g;
1600
1601 # It cannot have two consecutive dots .. anywhere
1602 # .. becomes %2E%2E
1603 $refname =~ s{\.\.}{%2E%2E}g;
1604
1605 return $refname;
1606 }
1607
1608 sub desanitize_refname {
1609 my ($refname) = @_;
1610 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1611 return $refname;
1612 }
1613
1614 sub svm_uuid {
1615 my ($self) = @_;
1616 return $self->{svm}->{uuid} if $self->svm;
1617 $self->ra;
1618 unless ($self->{svm}) {
1619 die "SVM UUID not cached, and reading remotely failed\n";
1620 }
1621 $self->{svm}->{uuid};
1622 }
1623
1624 sub svm {
1625 my ($self) = @_;
1626 return $self->{svm} if $self->{svm};
1627 my $svm;
1628 # see if we have it in our config, first:
1629 eval {
1630 my $section = "svn-remote.$self->{repo_id}";
1631 $svm = {
1632 source => tmp_config('--get', "$section.svm-source"),
1633 uuid => tmp_config('--get', "$section.svm-uuid"),
1634 replace => tmp_config('--get', "$section.svm-replace"),
1635 }
1636 };
1637 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1638 $self->{svm} = $svm;
1639 }
1640 $self->{svm};
1641 }
1642
1643 sub _set_svm_vars {
1644 my ($self, $ra) = @_;
1645 return $ra if $self->svm;
1646
1647 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1648 "(svm:source, svm:uuid) ",
1649 "from the following URLs:\n" );
1650 sub read_svm_props {
1651 my ($self, $ra, $path, $r) = @_;
1652 my $props = ($ra->get_dir($path, $r))[2];
1653 my $src = $props->{'svm:source'};
1654 my $uuid = $props->{'svm:uuid'};
1655 return undef if (!$src || !$uuid);
1656
1657 chomp($src, $uuid);
1658
1659 $uuid =~ m{^[0-9a-f\-]{30,}$}
1660 or die "doesn't look right - svm:uuid is '$uuid'\n";
1661
1662 # the '!' is used to mark the repos_root!/relative/path
1663 $src =~ s{/?!/?}{/};
1664 $src =~ s{/+$}{}; # no trailing slashes please
1665 # username is of no interest
1666 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1667
1668 my $replace = $ra->{url};
1669 $replace .= "/$path" if length $path;
1670
1671 my $section = "svn-remote.$self->{repo_id}";
1672 tmp_config("$section.svm-source", $src);
1673 tmp_config("$section.svm-replace", $replace);
1674 tmp_config("$section.svm-uuid", $uuid);
1675 $self->{svm} = {
1676 source => $src,
1677 uuid => $uuid,
1678 replace => $replace
1679 };
1680 }
1681
1682 my $r = $ra->get_latest_revnum;
1683 my $path = $self->{path};
1684 my %tried;
1685 while (length $path) {
1686 unless ($tried{"$self->{url}/$path"}) {
1687 return $ra if $self->read_svm_props($ra, $path, $r);
1688 $tried{"$self->{url}/$path"} = 1;
1689 }
1690 $path =~ s#/?[^/]+$##;
1691 }
1692 die "Path: '$path' should be ''\n" if $path ne '';
1693 return $ra if $self->read_svm_props($ra, $path, $r);
1694 $tried{"$self->{url}/$path"} = 1;
1695
1696 if ($ra->{repos_root} eq $self->{url}) {
1697 die @err, (map { " $_\n" } keys %tried), "\n";
1698 }
1699
1700 # nope, make sure we're connected to the repository root:
1701 my $ok;
1702 my @tried_b;
1703 $path = $ra->{svn_path};
1704 $ra = Git::SVN::Ra->new($ra->{repos_root});
1705 while (length $path) {
1706 unless ($tried{"$ra->{url}/$path"}) {
1707 $ok = $self->read_svm_props($ra, $path, $r);
1708 last if $ok;
1709 $tried{"$ra->{url}/$path"} = 1;
1710 }
1711 $path =~ s#/?[^/]+$##;
1712 }
1713 die "Path: '$path' should be ''\n" if $path ne '';
1714 $ok ||= $self->read_svm_props($ra, $path, $r);
1715 $tried{"$ra->{url}/$path"} = 1;
1716 if (!$ok) {
1717 die @err, (map { " $_\n" } keys %tried), "\n";
1718 }
1719 Git::SVN::Ra->new($self->{url});
1720 }
1721
1722 sub svnsync {
1723 my ($self) = @_;
1724 return $self->{svnsync} if $self->{svnsync};
1725
1726 if ($self->no_metadata) {
1727 die "Can't have both 'noMetadata' and ",
1728 "'useSvnsyncProps' options set!\n";
1729 }
1730 if ($self->rewrite_root) {
1731 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1732 "options set!\n";
1733 }
1734
1735 my $svnsync;
1736 # see if we have it in our config, first:
1737 eval {
1738 my $section = "svn-remote.$self->{repo_id}";
1739 $svnsync = {
1740 url => tmp_config('--get', "$section.svnsync-url"),
1741 uuid => tmp_config('--get', "$section.svnsync-uuid"),
1742 }
1743 };
1744 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1745 return $self->{svnsync} = $svnsync;
1746 }
1747
1748 my $err = "useSvnsyncProps set, but failed to read " .
1749 "svnsync property: svn:sync-from-";
1750 my $rp = $self->ra->rev_proplist(0);
1751
1752 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1753 $url =~ m{^[a-z\+]+://} or
1754 die "doesn't look right - svn:sync-from-url is '$url'\n";
1755
1756 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1757 $uuid =~ m{^[0-9a-f\-]{30,}$} or
1758 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1759
1760 my $section = "svn-remote.$self->{repo_id}";
1761 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1762 tmp_config('--add', "$section.svnsync-url", $url);
1763 return $self->{svnsync} = { url => $url, uuid => $uuid };
1764 }
1765
1766 # this allows us to memoize our SVN::Ra UUID locally and avoid a
1767 # remote lookup (useful for 'git svn log').
1768 sub ra_uuid {
1769 my ($self) = @_;
1770 unless ($self->{ra_uuid}) {
1771 my $key = "svn-remote.$self->{repo_id}.uuid";
1772 my $uuid = eval { tmp_config('--get', $key) };
1773 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1774 $self->{ra_uuid} = $uuid;
1775 } else {
1776 die "ra_uuid called without URL\n" unless $self->{url};
1777 $self->{ra_uuid} = $self->ra->get_uuid;
1778 tmp_config('--add', $key, $self->{ra_uuid});
1779 }
1780 }
1781 $self->{ra_uuid};
1782 }
1783
1784 sub _set_repos_root {
1785 my ($self, $repos_root) = @_;
1786 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1787 $repos_root ||= $self->ra->{repos_root};
1788 tmp_config($k, $repos_root);
1789 $repos_root;
1790 }
1791
1792 sub repos_root {
1793 my ($self) = @_;
1794 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1795 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1796 }
1797
1798 sub ra {
1799 my ($self) = shift;
1800 my $ra = Git::SVN::Ra->new($self->{url});
1801 $self->_set_repos_root($ra->{repos_root});
1802 if ($self->use_svm_props && !$self->{svm}) {
1803 if ($self->no_metadata) {
1804 die "Can't have both 'noMetadata' and ",
1805 "'useSvmProps' options set!\n";
1806 } elsif ($self->use_svnsync_props) {
1807 die "Can't have both 'useSvnsyncProps' and ",
1808 "'useSvmProps' options set!\n";
1809 }
1810 $ra = $self->_set_svm_vars($ra);
1811 $self->{-want_revprops} = 1;
1812 }
1813 $ra;
1814 }
1815
1816 sub rel_path {
1817 my ($self) = @_;
1818 my $repos_root = $self->ra->{repos_root};
1819 return $self->{path} if ($self->{url} eq $repos_root);
1820 my $url = $self->{url} .
1821 (length $self->{path} ? "/$self->{path}" : $self->{path});
1822 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1823 $url;
1824 }
1825
1826 # prop_walk(PATH, REV, SUB)
1827 # -------------------------
1828 # Recursively traverse PATH at revision REV and invoke SUB for each
1829 # directory that contains a SVN property. SUB will be invoked as
1830 # follows: &SUB(gs, path, props); where `gs' is this instance of
1831 # Git::SVN, `path' the path to the directory where the properties
1832 # `props' were found. The `path' will be relative to point of checkout,
1833 # that is, if url://repo/trunk is the current Git branch, and that
1834 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
1835 # as `path' (note the trailing `/').
1836 sub prop_walk {
1837 my ($self, $path, $rev, $sub) = @_;
1838
1839 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1840 $path =~ s#^/*#/#g;
1841 my $p = $path;
1842 # Strip the irrelevant part of the path.
1843 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1844 # Ensure the path is terminated by a `/'.
1845 $p =~ s#/*$#/#;
1846
1847 # The properties contain all the internal SVN stuff nobody
1848 # (usually) cares about.
1849 my $interesting_props = 0;
1850 foreach (keys %{$props}) {
1851 # If it doesn't start with `svn:', it must be a
1852 # user-defined property.
1853 ++$interesting_props and next if $_ !~ /^svn:/;
1854 # FIXME: Fragile, if SVN adds new public properties,
1855 # this needs to be updated.
1856 ++$interesting_props if /^svn:(?:ignore|keywords|executable
1857 |eol-style|mime-type
1858 |externals|needs-lock)$/x;
1859 }
1860 &$sub($self, $p, $props) if $interesting_props;
1861
1862 foreach (sort keys %$dirent) {
1863 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1864 $self->prop_walk($path . '/' . $_, $rev, $sub);
1865 }
1866 }
1867
1868 sub last_rev { ($_[0]->last_rev_commit)[0] }
1869 sub last_commit { ($_[0]->last_rev_commit)[1] }
1870
1871 # returns the newest SVN revision number and newest commit SHA1
1872 sub last_rev_commit {
1873 my ($self) = @_;
1874 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1875 return ($self->{last_rev}, $self->{last_commit});
1876 }
1877 my $c = ::verify_ref($self->refname.'^0');
1878 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1879 my $rev = (::cmt_metadata($c))[1];
1880 if (defined $rev) {
1881 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1882 return ($rev, $c);
1883 }
1884 }
1885 my $db_path = $self->db_path;
1886 unless (-e $db_path) {
1887 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1888 return (undef, undef);
1889 }
1890 my $offset = -41; # from tail
1891 my $rl;
1892 open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1893 sysseek($fh, $offset, 2); # don't care for errors
1894 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1895 chomp $rl;
1896 while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1897 $offset -= 41;
1898 sysseek($fh, $offset, 2); # don't care for errors
1899 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1900 chomp $rl;
1901 }
1902 if ($c && $c ne $rl) {
1903 die "$db_path and ", $self->refname,
1904 " inconsistent!:\n$c != $rl\n";
1905 }
1906 my $rev = sysseek($fh, 0, 1) or croak $!;
1907 $rev = ($rev - 41) / 41;
1908 close $fh or croak $!;
1909 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1910 return ($rev, $c);
1911 }
1912
1913 sub get_fetch_range {
1914 my ($self, $min, $max) = @_;
1915 $max ||= $self->ra->get_latest_revnum;
1916 $min ||= $self->rev_db_max;
1917 (++$min, $max);
1918 }
1919
1920 sub tmp_config {
1921 my (@args) = @_;
1922 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1923 my $config = "$ENV{GIT_DIR}/svn/.metadata";
1924 if (! -f $config && -f $old_def_config) {
1925 rename $old_def_config, $config or
1926 die "Failed rename $old_def_config => $config: $!\n";
1927 }
1928 my $old_config = $ENV{GIT_CONFIG};
1929 $ENV{GIT_CONFIG} = $config;
1930 $@ = undef;
1931 my @ret = eval {
1932 unless (-f $config) {
1933 mkfile($config);
1934 open my $fh, '>', $config or
1935 die "Can't open $config: $!\n";
1936 print $fh "; This file is used internally by ",
1937 "git-svn\n" or die
1938 "Couldn't write to $config: $!\n";
1939 print $fh "; You should not have to edit it\n" or
1940 die "Couldn't write to $config: $!\n";
1941 close $fh or die "Couldn't close $config: $!\n";
1942 }
1943 command('config', @args);
1944 };
1945 my $err = $@;
1946 if (defined $old_config) {
1947 $ENV{GIT_CONFIG} = $old_config;
1948 } else {
1949 delete $ENV{GIT_CONFIG};
1950 }
1951 die $err if $err;
1952 wantarray ? @ret : $ret[0];
1953 }
1954
1955 sub tmp_index_do {
1956 my ($self, $sub) = @_;
1957 my $old_index = $ENV{GIT_INDEX_FILE};
1958 $ENV{GIT_INDEX_FILE} = $self->{index};
1959 $@ = undef;
1960 my @ret = eval {
1961 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1962 mkpath([$dir]) unless -d $dir;
1963 &$sub;
1964 };
1965 my $err = $@;
1966 if (defined $old_index) {
1967 $ENV{GIT_INDEX_FILE} = $old_index;
1968 } else {
1969 delete $ENV{GIT_INDEX_FILE};
1970 }
1971 die $err if $err;
1972 wantarray ? @ret : $ret[0];
1973 }
1974
1975 sub assert_index_clean {
1976 my ($self, $treeish) = @_;
1977
1978 $self->tmp_index_do(sub {
1979 command_noisy('read-tree', $treeish) unless -e $self->{index};
1980 my $x = command_oneline('write-tree');
1981 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1982 /^tree ($::sha1)/mo);
1983 return if $y eq $x;
1984
1985 warn "Index mismatch: $y != $x\nrereading $treeish\n";
1986 unlink $self->{index} or die "unlink $self->{index}: $!\n";
1987 command_noisy('read-tree', $treeish);
1988 $x = command_oneline('write-tree');
1989 if ($y ne $x) {
1990 ::fatal "trees ($treeish) $y != $x\n",
1991 "Something is seriously wrong...";
1992 }
1993 });
1994 }
1995
1996 sub get_commit_parents {
1997 my ($self, $log_entry) = @_;
1998 my (%seen, @ret, @tmp);
1999 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2000 if (my $ip = $self->{inject_parents}) {
2001 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2002 push @tmp, $commit;
2003 }
2004 }
2005 if (my $cur = ::verify_ref($self->refname.'^0')) {
2006 push @tmp, $cur;
2007 }
2008 if (my $ipd = $self->{inject_parents_dcommit}) {
2009 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2010 push @tmp, @$commit;
2011 }
2012 }
2013 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2014 while (my $p = shift @tmp) {
2015 next if $seen{$p};
2016 $seen{$p} = 1;
2017 push @ret, $p;
2018 # MAXPARENT is defined to 16 in commit-tree.c:
2019 last if @ret >= 16;
2020 }
2021 if (@tmp) {
2022 die "r$log_entry->{revision}: No room for parents:\n\t",
2023 join("\n\t", @tmp), "\n";
2024 }
2025 @ret;
2026 }
2027
2028 sub rewrite_root {
2029 my ($self) = @_;
2030 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2031 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2032 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2033 if ($rwr) {
2034 $rwr =~ s#/+$##;
2035 if ($rwr !~ m#^[a-z\+]+://#) {
2036 die "$rwr is not a valid URL (key: $k)\n";
2037 }
2038 }
2039 $self->{-rewrite_root} = $rwr;
2040 }
2041
2042 sub metadata_url {
2043 my ($self) = @_;
2044 ($self->rewrite_root || $self->{url}) .
2045 (length $self->{path} ? '/' . $self->{path} : '');
2046 }
2047
2048 sub full_url {
2049 my ($self) = @_;
2050 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2051 }
2052
2053 sub do_git_commit {
2054 my ($self, $log_entry) = @_;
2055 my $lr = $self->last_rev;
2056 if (defined $lr && $lr >= $log_entry->{revision}) {
2057 die "Last fetched revision of ", $self->refname,
2058 " was r$lr, but we are about to fetch: ",
2059 "r$log_entry->{revision}!\n";
2060 }
2061 if (my $c = $self->rev_db_get($log_entry->{revision})) {
2062 croak "$log_entry->{revision} = $c already exists! ",
2063 "Why are we refetching it?\n";
2064 }
2065 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2066 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2067 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2068
2069 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2070 ? $log_entry->{commit_name}
2071 : $log_entry->{name};
2072 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2073 ? $log_entry->{commit_email}
2074 : $log_entry->{email};
2075
2076 my $tree = $log_entry->{tree};
2077 if (!defined $tree) {
2078 $tree = $self->tmp_index_do(sub {
2079 command_oneline('write-tree') });
2080 }
2081 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2082
2083 my @exec = ('git-commit-tree', $tree);
2084 foreach ($self->get_commit_parents($log_entry)) {
2085 push @exec, '-p', $_;
2086 }
2087 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2088 or croak $!;
2089 print $msg_fh $log_entry->{log} or croak $!;
2090 unless ($self->no_metadata) {
2091 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2092 or croak $!;
2093 }
2094 $msg_fh->flush == 0 or croak $!;
2095 close $msg_fh or croak $!;
2096 chomp(my $commit = do { local $/; <$out_fh> });
2097 close $out_fh or croak $!;
2098 waitpid $pid, 0;
2099 croak $? if $?;
2100 if ($commit !~ /^$::sha1$/o) {
2101 die "Failed to commit, invalid sha1: $commit\n";
2102 }
2103
2104 $self->rev_db_set($log_entry->{revision}, $commit, 1);
2105
2106 $self->{last_rev} = $log_entry->{revision};
2107 $self->{last_commit} = $commit;
2108 print "r$log_entry->{revision}";
2109 if (defined $log_entry->{svm_revision}) {
2110 print " (\@$log_entry->{svm_revision})";
2111 $self->rev_db_set($log_entry->{svm_revision}, $commit,
2112 0, $self->svm_uuid);
2113 }
2114 print " = $commit ($self->{ref_id})\n";
2115 if (defined $_repack && (--$_repack_nr == 0)) {
2116 $_repack_nr = $_repack;
2117 # repack doesn't use any arguments with spaces in them, does it?
2118 print "Running git repack $_repack_flags ...\n";
2119 command_noisy('repack', split(/\s+/, $_repack_flags));
2120 print "Done repacking\n";
2121 }
2122 return $commit;
2123 }
2124
2125 sub match_paths {
2126 my ($self, $paths, $r) = @_;
2127 return 1 if $self->{path} eq '';
2128 if (my $path = $paths->{"/$self->{path}"}) {
2129 return ($path->{action} eq 'D') ? 0 : 1;
2130 }
2131 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2132 if (grep /$self->{path_regex}/, keys %$paths) {
2133 return 1;
2134 }
2135 my $c = '';
2136 foreach (split m#/#, $self->{path}) {
2137 $c .= "/$_";
2138 next unless ($paths->{$c} &&
2139 ($paths->{$c}->{action} =~ /^[AR]$/));
2140 if ($self->ra->check_path($self->{path}, $r) ==
2141 $SVN::Node::dir) {
2142 return 1;
2143 }
2144 }
2145 return 0;
2146 }
2147
2148 sub find_parent_branch {
2149 my ($self, $paths, $rev) = @_;
2150 return undef unless $self->follow_parent;
2151 unless (defined $paths) {
2152 my $err_handler = $SVN::Error::handler;
2153 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2154 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2155 $paths =
2156 Git::SVN::Ra::dup_changed_paths($_[0]) });
2157 $SVN::Error::handler = $err_handler;
2158 }
2159 return undef unless defined $paths;
2160
2161 # look for a parent from another branch:
2162 my @b_path_components = split m#/#, $self->rel_path;
2163 my @a_path_components;
2164 my $i;
2165 while (@b_path_components) {
2166 $i = $paths->{'/'.join('/', @b_path_components)};
2167 last if $i && defined $i->{copyfrom_path};
2168 unshift(@a_path_components, pop(@b_path_components));
2169 }
2170 return undef unless defined $i && defined $i->{copyfrom_path};
2171 my $branch_from = $i->{copyfrom_path};
2172 if (@a_path_components) {
2173 print STDERR "branch_from: $branch_from => ";
2174 $branch_from .= '/'.join('/', @a_path_components);
2175 print STDERR $branch_from, "\n";
2176 }
2177 my $r = $i->{copyfrom_rev};
2178 my $repos_root = $self->ra->{repos_root};
2179 my $url = $self->ra->{url};
2180 my $new_url = $repos_root . $branch_from;
2181 print STDERR "Found possible branch point: ",
2182 "$new_url => ", $self->full_url, ", $r\n";
2183 $branch_from =~ s#^/##;
2184 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2185 unless ($gs) {
2186 my $ref_id = $self->{ref_id};
2187 $ref_id =~ s/\@\d+$//;
2188 $ref_id .= "\@$r";
2189 # just grow a tail if we're not unique enough :x
2190 $ref_id .= '-' while find_ref($ref_id);
2191 print STDERR "Initializing parent: $ref_id\n";
2192 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
2193 }
2194 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2195 if (!defined $r0 || !defined $parent) {
2196 my ($base, $head) = parse_revision_argument(0, $r);
2197 if ($base <= $r) {
2198 $gs->fetch($base, $r);
2199 }
2200 ($r0, $parent) = $gs->last_rev_commit;
2201 }
2202 if (defined $r0 && defined $parent) {
2203 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2204 my $ed;
2205 if ($self->ra->can_do_switch) {
2206 $self->assert_index_clean($parent);
2207 print STDERR "Following parent with do_switch\n";
2208 # do_switch works with svn/trunk >= r22312, but that
2209 # is not included with SVN 1.4.3 (the latest version
2210 # at the moment), so we can't rely on it
2211 $self->{last_commit} = $parent;
2212 $ed = SVN::Git::Fetcher->new($self);
2213 $gs->ra->gs_do_switch($r0, $rev, $gs,
2214 $self->full_url, $ed)
2215 or die "SVN connection failed somewhere...\n";
2216 } elsif ($self->ra->trees_match($new_url, $r0,
2217 $self->full_url, $rev)) {
2218 print STDERR "Trees match:\n",
2219 " $new_url\@$r0\n",
2220 " ${\$self->full_url}\@$rev\n",
2221 "Following parent with no changes\n";
2222 $self->tmp_index_do(sub {
2223 command_noisy('read-tree', $parent);
2224 });
2225 $self->{last_commit} = $parent;
2226 } else {
2227 print STDERR "Following parent with do_update\n";
2228 $ed = SVN::Git::Fetcher->new($self);
2229 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2230 or die "SVN connection failed somewhere...\n";
2231 }
2232 print STDERR "Successfully followed parent\n";
2233 return $self->make_log_entry($rev, [$parent], $ed);
2234 }
2235 return undef;
2236 }
2237
2238 sub do_fetch {
2239 my ($self, $paths, $rev) = @_;
2240 my $ed;
2241 my ($last_rev, @parents);
2242 if (my $lc = $self->last_commit) {
2243 # we can have a branch that was deleted, then re-added
2244 # under the same name but copied from another path, in
2245 # which case we'll have multiple parents (we don't
2246 # want to break the original ref, nor lose copypath info):
2247 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2248 push @{$log_entry->{parents}}, $lc;
2249 return $log_entry;
2250 }
2251 $ed = SVN::Git::Fetcher->new($self);
2252 $last_rev = $self->{last_rev};
2253 $ed->{c} = $lc;
2254 @parents = ($lc);
2255 } else {
2256 $last_rev = $rev;
2257 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2258 return $log_entry;
2259 }
2260 $ed = SVN::Git::Fetcher->new($self);
2261 }
2262 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2263 die "SVN connection failed somewhere...\n";
2264 }
2265 $self->make_log_entry($rev, \@parents, $ed);
2266 }
2267
2268 sub get_untracked {
2269 my ($self, $ed) = @_;
2270 my @out;
2271 my $h = $ed->{empty};
2272 foreach (sort keys %$h) {
2273 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2274 push @out, " $act: " . uri_encode($_);
2275 warn "W: $act: $_\n";
2276 }
2277 foreach my $t (qw/dir_prop file_prop/) {
2278 $h = $ed->{$t} or next;
2279 foreach my $path (sort keys %$h) {
2280 my $ppath = $path eq '' ? '.' : $path;
2281 foreach my $prop (sort keys %{$h->{$path}}) {
2282 next if $SKIP_PROP{$prop};
2283 my $v = $h->{$path}->{$prop};
2284 my $t_ppath_prop = "$t: " .
2285 uri_encode($ppath) . ' ' .
2286 uri_encode($prop);
2287 if (defined $v) {
2288 push @out, " +$t_ppath_prop " .
2289 uri_encode($v);
2290 } else {
2291 push @out, " -$t_ppath_prop";
2292 }
2293 }
2294 }
2295 }
2296 foreach my $t (qw/absent_file absent_directory/) {
2297 $h = $ed->{$t} or next;
2298 foreach my $parent (sort keys %$h) {
2299 foreach my $path (sort @{$h->{$parent}}) {
2300 push @out, " $t: " .
2301 uri_encode("$parent/$path");
2302 warn "W: $t: $parent/$path ",
2303 "Insufficient permissions?\n";
2304 }
2305 }
2306 }
2307 \@out;
2308 }
2309
2310 sub parse_svn_date {
2311 my $date = shift || return '+0000 1970-01-01 00:00:00';
2312 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2313 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2314 croak "Unable to parse date: $date\n";
2315 "+0000 $Y-$m-$d $H:$M:$S";
2316 }
2317
2318 sub check_author {
2319 my ($author) = @_;
2320 if (!defined $author || length $author == 0) {
2321 $author = '(no author)';
2322 }
2323 if (defined $::_authors && ! defined $::users{$author}) {
2324 die "Author: $author not defined in $::_authors file\n";
2325 }
2326 $author;
2327 }
2328
2329 sub make_log_entry {
2330 my ($self, $rev, $parents, $ed) = @_;
2331 my $untracked = $self->get_untracked($ed);
2332
2333 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2334 print $un "r$rev\n" or croak $!;
2335 print $un $_, "\n" foreach @$untracked;
2336 my %log_entry = ( parents => $parents || [], revision => $rev,
2337 log => '');
2338
2339 my $headrev;
2340 my $logged = delete $self->{logged_rev_props};
2341 if (!$logged || $self->{-want_revprops}) {
2342 my $rp = $self->ra->rev_proplist($rev);
2343 foreach (sort keys %$rp) {
2344 my $v = $rp->{$_};
2345 if (/^svn:(author|date|log)$/) {
2346 $log_entry{$1} = $v;
2347 } elsif ($_ eq 'svm:headrev') {
2348 $headrev = $v;
2349 } else {
2350 print $un " rev_prop: ", uri_encode($_), ' ',
2351 uri_encode($v), "\n";
2352 }
2353 }
2354 } else {
2355 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2356 }
2357 close $un or croak $!;
2358
2359 $log_entry{date} = parse_svn_date($log_entry{date});
2360 $log_entry{log} .= "\n";
2361 my $author = $log_entry{author} = check_author($log_entry{author});
2362 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2363 : ($author, undef);
2364
2365 my ($commit_name, $commit_email) = ($name, $email);
2366 if ($_use_log_author) {
2367 if ($log_entry{log} =~ /From:\s+(.*?)\s+<(.*)>\s*\n/) {
2368 ($name, $email) = ($1, $2);
2369 } elsif ($log_entry{log} =~
2370 /Signed-off-by:\s+(.*?)\s+<(.*)>\s*\n/) {
2371 ($name, $email) = ($1, $2);
2372 }
2373 }
2374 if (defined $headrev && $self->use_svm_props) {
2375 if ($self->rewrite_root) {
2376 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2377 "options set!\n";
2378 }
2379 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2380 # we don't want "SVM: initializing mirror for junk" ...
2381 return undef if $r == 0;
2382 my $svm = $self->svm;
2383 if ($uuid ne $svm->{uuid}) {
2384 die "UUID mismatch on SVM path:\n",
2385 "expected: $svm->{uuid}\n",
2386 " got: $uuid\n";
2387 }
2388 my $full_url = $self->full_url;
2389 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2390 die "Failed to replace '$svm->{replace}' with ",
2391 "'$svm->{source}' in $full_url\n";
2392 # throw away username for storing in records
2393 remove_username($full_url);
2394 $log_entry{metadata} = "$full_url\@$r $uuid";
2395 $log_entry{svm_revision} = $r;
2396 $email ||= "$author\@$uuid";
2397 $commit_email ||= "$author\@$uuid";
2398 } elsif ($self->use_svnsync_props) {
2399 my $full_url = $self->svnsync->{url};
2400 $full_url .= "/$self->{path}" if length $self->{path};
2401 remove_username($full_url);
2402 my $uuid = $self->svnsync->{uuid};
2403 $log_entry{metadata} = "$full_url\@$rev $uuid";
2404 $email ||= "$author\@$uuid";
2405 $commit_email ||= "$author\@$uuid";
2406 } else {
2407 my $url = $self->metadata_url;
2408 remove_username($url);
2409 $log_entry{metadata} = "$url\@$rev " .
2410 $self->ra->get_uuid;
2411 $email ||= "$author\@" . $self->ra->get_uuid;
2412 $commit_email ||= "$author\@" . $self->ra->get_uuid;
2413 }
2414 $log_entry{name} = $name;
2415 $log_entry{email} = $email;
2416 $log_entry{commit_name} = $commit_name;
2417 $log_entry{commit_email} = $commit_email;
2418 \%log_entry;
2419 }
2420
2421 sub fetch {
2422 my ($self, $min_rev, $max_rev, @parents) = @_;
2423 my ($last_rev, $last_commit) = $self->last_rev_commit;
2424 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2425 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2426 }
2427
2428 sub set_tree_cb {
2429 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2430 $self->{inject_parents} = { $rev => $tree };
2431 $self->fetch(undef, undef);
2432 }
2433
2434 sub set_tree {
2435 my ($self, $tree) = (shift, shift);
2436 my $log_entry = ::get_commit_entry($tree);
2437 unless ($self->{last_rev}) {
2438 fatal("Must have an existing revision to commit");
2439 }
2440 my %ed_opts = ( r => $self->{last_rev},
2441 log => $log_entry->{log},
2442 ra => $self->ra,
2443 tree_a => $self->{last_commit},
2444 tree_b => $tree,
2445 editor_cb => sub {
2446 $self->set_tree_cb($log_entry, $tree, @_) },
2447 svn_path => $self->{path} );
2448 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2449 print "No changes\nr$self->{last_rev} = $tree\n";
2450 }
2451 }
2452
2453 sub rebuild {
2454 my ($self) = @_;
2455 my $db_path = $self->db_path;
2456 return if (-e $db_path && ! -z $db_path);
2457 return unless ::verify_ref($self->refname.'^0');
2458 if (-f $self->{db_root}) {
2459 rename $self->{db_root}, $db_path or die
2460 "rename $self->{db_root} => $db_path failed: $!\n";
2461 my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
2462 symlink $base, $self->{db_root} or die
2463 "symlink $base => $self->{db_root} failed: $!\n";
2464 return;
2465 }
2466 print "Rebuilding $db_path ...\n";
2467 my ($log, $ctx) = command_output_pipe("log", '--no-color', $self->refname);
2468 my $latest;
2469 my $full_url = $self->full_url;
2470 remove_username($full_url);
2471 my $svn_uuid;
2472 my $c;
2473 while (<$log>) {
2474 if ( m{^commit ($::sha1)$} ) {
2475 $c = $1;
2476 next;
2477 }
2478 next unless s{^\s*(git-svn-id:)}{$1};
2479 my ($url, $rev, $uuid) = ::extract_metadata($_);
2480 remove_username($url);
2481
2482 # ignore merges (from set-tree)
2483 next if (!defined $rev || !$uuid);
2484
2485 # if we merged or otherwise started elsewhere, this is
2486 # how we break out of it
2487 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
2488 ($full_url && $url && ($url ne $full_url))) {
2489 next;
2490 }
2491 $latest ||= $rev;
2492 $svn_uuid ||= $uuid;
2493
2494 $self->rev_db_set($rev, $c);
2495 print "r$rev = $c\n";
2496 }
2497 command_close_pipe($log, $ctx);
2498 print "Done rebuilding $db_path\n";
2499 }
2500
2501 # rev_db:
2502 # Tie::File seems to be prone to offset errors if revisions get sparse,
2503 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
2504 # one of my favorite modules is out :< Next up would be one of the DBM
2505 # modules, but I'm not sure which is most portable... So I'll just
2506 # go with something that's plain-text, but still capable of
2507 # being randomly accessed. So here's my ultra-simple fixed-width
2508 # database. All records are 40 characters + "\n", so it's easy to seek
2509 # to a revision: (41 * rev) is the byte offset.
2510 # A record of 40 0s denotes an empty revision.
2511 # And yes, it's still pretty fast (faster than Tie::File).
2512 # These files are disposable unless noMetadata or useSvmProps is set
2513
2514 sub _rev_db_set {
2515 my ($fh, $rev, $commit) = @_;
2516 my $offset = $rev * 41;
2517 # assume that append is the common case:
2518 seek $fh, 0, 2 or croak $!;
2519 my $pos = tell $fh;
2520 if ($pos < $offset) {
2521 for (1 .. (($offset - $pos) / 41)) {
2522 print $fh (('0' x 40),"\n") or croak $!;
2523 }
2524 }
2525 seek $fh, $offset, 0 or croak $!;
2526 print $fh $commit,"\n" or croak $!;
2527 }
2528
2529 sub mkfile {
2530 my ($path) = @_;
2531 unless (-e $path) {
2532 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2533 mkpath([$dir]) unless -d $dir;
2534 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2535 close $fh or die "Couldn't close (create) $path: $!\n";
2536 }
2537 }
2538
2539 sub rev_db_set {
2540 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2541 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2542 my $db = $self->db_path($uuid);
2543 my $db_lock = "$db.lock";
2544 my $sig;
2545 if ($update_ref) {
2546 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2547 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2548 }
2549 mkfile($db);
2550
2551 $LOCKFILES{$db_lock} = 1;
2552 my $sync;
2553 # both of these options make our .rev_db file very, very important
2554 # and we can't afford to lose it because rebuild() won't work
2555 if ($self->use_svm_props || $self->no_metadata) {
2556 $sync = 1;
2557 copy($db, $db_lock) or die "rev_db_set(@_): ",
2558 "Failed to copy: ",
2559 "$db => $db_lock ($!)\n";
2560 } else {
2561 rename $db, $db_lock or die "rev_db_set(@_): ",
2562 "Failed to rename: ",
2563 "$db => $db_lock ($!)\n";
2564 }
2565 open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
2566 _rev_db_set($fh, $rev, $commit);
2567 if ($sync) {
2568 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2569 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2570 }
2571 close $fh or croak $!;
2572 if ($update_ref) {
2573 $_head = $self;
2574 command_noisy('update-ref', '-m', "r$rev",
2575 $self->refname, $commit);
2576 }
2577 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
2578 "$db_lock => $db ($!)\n";
2579 delete $LOCKFILES{$db_lock};
2580 if ($update_ref) {
2581 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2582 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2583 kill $sig, $$ if defined $sig;
2584 }
2585 }
2586
2587 sub rev_db_max {
2588 my ($self) = @_;
2589 $self->rebuild;
2590 my $db_path = $self->db_path;
2591 my @stat = stat $db_path or return 0;
2592 ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
2593 my $max = $stat[7] / 41;
2594 (($max > 0) ? $max - 1 : 0);
2595 }
2596
2597 sub rev_db_get {
2598 my ($self, $rev, $uuid) = @_;
2599 my $ret;
2600 my $offset = $rev * 41;
2601 my $db_path = $self->db_path($uuid);
2602 return undef unless -e $db_path;
2603 open my $fh, '<', $db_path or croak $!;
2604 if (sysseek($fh, $offset, 0) == $offset) {
2605 my $read = sysread($fh, $ret, 40);
2606 $ret = undef if ($read != 40 || $ret eq ('0'x40));
2607 }
2608 close $fh or croak $!;
2609 $ret;
2610 }
2611
2612 # Finds the first svn revision that exists on (if $eq_ok is true) or
2613 # before $rev for the current branch. It will not search any lower
2614 # than $min_rev. Returns the git commit hash and svn revision number
2615 # if found, else (undef, undef).
2616 sub find_rev_before {
2617 my ($self, $rev, $eq_ok, $min_rev) = @_;
2618 --$rev unless $eq_ok;
2619 $min_rev ||= 1;
2620 while ($rev >= $min_rev) {
2621 if (my $c = $self->rev_db_get($rev)) {
2622 return ($rev, $c);
2623 }
2624 --$rev;
2625 }
2626 return (undef, undef);
2627 }
2628
2629 # Finds the first svn revision that exists on (if $eq_ok is true) or
2630 # after $rev for the current branch. It will not search any higher
2631 # than $max_rev. Returns the git commit hash and svn revision number
2632 # if found, else (undef, undef).
2633 sub find_rev_after {
2634 my ($self, $rev, $eq_ok, $max_rev) = @_;
2635 ++$rev unless $eq_ok;
2636 $max_rev ||= $self->rev_db_max();
2637 while ($rev <= $max_rev) {
2638 if (my $c = $self->rev_db_get($rev)) {
2639 return ($rev, $c);
2640 }
2641 ++$rev;
2642 }
2643 return (undef, undef);
2644 }
2645
2646 sub _new {
2647 my ($class, $repo_id, $ref_id, $path) = @_;
2648 unless (defined $repo_id && length $repo_id) {
2649 $repo_id = $Git::SVN::default_repo_id;
2650 }
2651 unless (defined $ref_id && length $ref_id) {
2652 $_[2] = $ref_id = $Git::SVN::default_ref_id;
2653 }
2654 $_[1] = $repo_id = sanitize_remote_name($repo_id);
2655 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2656 $_[3] = $path = '' unless (defined $path);
2657 mkpath(["$ENV{GIT_DIR}/svn"]);
2658 bless {
2659 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2660 path => $path, config => "$ENV{GIT_DIR}/svn/config",
2661 db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2662 }
2663
2664 sub db_path {
2665 my ($self, $uuid) = @_;
2666 $uuid ||= $self->ra_uuid;
2667 "$self->{db_root}.$uuid";
2668 }
2669
2670 sub uri_encode {
2671 my ($f) = @_;
2672 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2673 $f
2674 }
2675
2676 sub remove_username {
2677 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2678 }
2679
2680 package Git::SVN::Prompt;
2681 use strict;
2682 use warnings;
2683 require SVN::Core;
2684 use vars qw/$_no_auth_cache $_username/;
2685
2686 sub simple {
2687 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2688 $may_save = undef if $_no_auth_cache;
2689 $default_username = $_username if defined $_username;
2690 if (defined $default_username && length $default_username) {
2691 if (defined $realm && length $realm) {
2692 print STDERR "Authentication realm: $realm\n";
2693 STDERR->flush;
2694 }
2695 $cred->username($default_username);
2696 } else {
2697 username($cred, $realm, $may_save, $pool);
2698 }
2699 $cred->password(_read_password("Password for '" .
2700 $cred->username . "': ", $realm));
2701 $cred->may_save($may_save);
2702 $SVN::_Core::SVN_NO_ERROR;
2703 }
2704
2705 sub ssl_server_trust {
2706 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2707 $may_save = undef if $_no_auth_cache;
2708 print STDERR "Error validating server certificate for '$realm':\n";
2709 {
2710 no warnings 'once';
2711 # All variables SVN::Auth::SSL::* are used only once,
2712 # so we're shutting up Perl warnings about this.
2713 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2714 print STDERR " - The certificate is not issued ",
2715 "by a trusted authority. Use the\n",
2716 " fingerprint to validate ",
2717 "the certificate manually!\n";
2718 }
2719 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2720 print STDERR " - The certificate hostname ",
2721 "does not match.\n";
2722 }
2723 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2724 print STDERR " - The certificate is not yet valid.\n";
2725 }
2726 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2727 print STDERR " - The certificate has expired.\n";
2728 }
2729 if ($failures & $SVN::Auth::SSL::OTHER) {
2730 print STDERR " - The certificate has ",
2731 "an unknown error.\n";
2732 }
2733 } # no warnings 'once'
2734 printf STDERR
2735 "Certificate information:\n".
2736 " - Hostname: %s\n".
2737 " - Valid: from %s until %s\n".
2738 " - Issuer: %s\n".
2739 " - Fingerprint: %s\n",
2740 map $cert_info->$_, qw(hostname valid_from valid_until
2741 issuer_dname fingerprint);
2742 my $choice;
2743 prompt:
2744 print STDERR $may_save ?
2745 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2746 "(R)eject or accept (t)emporarily? ";
2747 STDERR->flush;
2748 $choice = lc(substr(<STDIN> || 'R', 0, 1));
2749 if ($choice =~ /^t$/i) {
2750 $cred->may_save(undef);
2751 } elsif ($choice =~ /^r$/i) {
2752 return -1;
2753 } elsif ($may_save && $choice =~ /^p$/i) {
2754 $cred->may_save($may_save);
2755 } else {
2756 goto prompt;
2757 }
2758 $cred->accepted_failures($failures);
2759 $SVN::_Core::SVN_NO_ERROR;
2760 }
2761
2762 sub ssl_client_cert {
2763 my ($cred, $realm, $may_save, $pool) = @_;
2764 $may_save = undef if $_no_auth_cache;
2765 print STDERR "Client certificate filename: ";
2766 STDERR->flush;
2767 chomp(my $filename = <STDIN>);
2768 $cred->cert_file($filename);
2769 $cred->may_save($may_save);
2770 $SVN::_Core::SVN_NO_ERROR;
2771 }
2772
2773 sub ssl_client_cert_pw {
2774 my ($cred, $realm, $may_save, $pool) = @_;
2775 $may_save = undef if $_no_auth_cache;
2776 $cred->password(_read_password("Password: ", $realm));
2777 $cred->may_save($may_save);
2778 $SVN::_Core::SVN_NO_ERROR;
2779 }
2780
2781 sub username {
2782 my ($cred, $realm, $may_save, $pool) = @_;
2783 $may_save = undef if $_no_auth_cache;
2784 if (defined $realm && length $realm) {
2785 print STDERR "Authentication realm: $realm\n";
2786 }
2787 my $username;
2788 if (defined $_username) {
2789 $username = $_username;
2790 } else {
2791 print STDERR "Username: ";
2792 STDERR->flush;
2793 chomp($username = <STDIN>);
2794 }
2795 $cred->username($username);
2796 $cred->may_save($may_save);
2797 $SVN::_Core::SVN_NO_ERROR;
2798 }
2799
2800 sub _read_password {
2801 my ($prompt, $realm) = @_;
2802 print STDERR $prompt;
2803 STDERR->flush;
2804 require Term::ReadKey;
2805 Term::ReadKey::ReadMode('noecho');
2806 my $password = '';
2807 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2808 last if $key =~ /[\012\015]/; # \n\r
2809 $password .= $key;
2810 }
2811 Term::ReadKey::ReadMode('restore');
2812 print STDERR "\n";
2813 STDERR->flush;
2814 $password;
2815 }
2816
2817 package SVN::Git::Fetcher;
2818 use vars qw/@ISA/;
2819 use strict;
2820 use warnings;
2821 use Carp qw/croak/;
2822 use IO::File qw//;
2823
2824 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
2825 sub new {
2826 my ($class, $git_svn) = @_;
2827 my $self = SVN::Delta::Editor->new;
2828 bless $self, $class;
2829 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2830 $self->{empty} = {};
2831 $self->{dir_prop} = {};
2832 $self->{file_prop} = {};
2833 $self->{absent_dir} = {};
2834 $self->{absent_file} = {};
2835 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2836 $self;
2837 }
2838
2839 sub set_path_strip {
2840 my ($self, $path) = @_;
2841 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2842 }
2843
2844 sub open_root {
2845 { path => '' };
2846 }
2847
2848 sub open_directory {
2849 my ($self, $path, $pb, $rev) = @_;
2850 { path => $path };
2851 }
2852
2853 sub git_path {
2854 my ($self, $path) = @_;
2855 if ($self->{path_strip}) {
2856 $path =~ s!$self->{path_strip}!! or
2857 die "Failed to strip path '$path' ($self->{path_strip})\n";
2858 }
2859 $path;
2860 }
2861
2862 sub delete_entry {
2863 my ($self, $path, $rev, $pb) = @_;
2864
2865 my $gpath = $self->git_path($path);
2866 return undef if ($gpath eq '');
2867
2868 # remove entire directories.
2869 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2870 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2871 -r --name-only -z/,
2872 $self->{c}, '--', $gpath);
2873 local $/ = "\0";
2874 while (<$ls>) {
2875 chomp;
2876 $self->{gii}->remove($_);
2877 print "\tD\t$_\n" unless $::_q;
2878 }
2879 print "\tD\t$gpath/\n" unless $::_q;
2880 command_close_pipe($ls, $ctx);
2881 $self->{empty}->{$path} = 0
2882 } else {
2883 $self->{gii}->remove($gpath);
2884 print "\tD\t$gpath\n" unless $::_q;
2885 }
2886 undef;
2887 }
2888
2889 sub open_file {
2890 my ($self, $path, $pb, $rev) = @_;
2891 my $gpath = $self->git_path($path);
2892 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2893 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2894 unless (defined $mode && defined $blob) {
2895 die "$path was not found in commit $self->{c} (r$rev)\n";
2896 }
2897 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2898 pool => SVN::Pool->new, action => 'M' };
2899 }
2900
2901 sub add_file {
2902 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2903 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2904 delete $self->{empty}->{$dir};
2905 { path => $path, mode_a => 100644, mode_b => 100644,
2906 pool => SVN::Pool->new, action => 'A' };
2907 }
2908
2909 sub add_directory {
2910 my ($self, $path, $cp_path, $cp_rev) = @_;
2911 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2912 delete $self->{empty}->{$dir};
2913 $self->{empty}->{$path} = 1;
2914 { path => $path };
2915 }
2916
2917 sub change_dir_prop {
2918 my ($self, $db, $prop, $value) = @_;
2919 $self->{dir_prop}->{$db->{path}} ||= {};
2920 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2921 undef;
2922 }
2923
2924 sub absent_directory {
2925 my ($self, $path, $pb) = @_;
2926 $self->{absent_dir}->{$pb->{path}} ||= [];
2927 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2928 undef;
2929 }
2930
2931 sub absent_file {
2932 my ($self, $path, $pb) = @_;
2933 $self->{absent_file}->{$pb->{path}} ||= [];
2934 push @{$self->{absent_file}->{$pb->{path}}}, $path;
2935 undef;
2936 }
2937
2938 sub change_file_prop {
2939 my ($self, $fb, $prop, $value) = @_;
2940 if ($prop eq 'svn:executable') {
2941 if ($fb->{mode_b} != 120000) {
2942 $fb->{mode_b} = defined $value ? 100755 : 100644;
2943 }
2944 } elsif ($prop eq 'svn:special') {
2945 $fb->{mode_b} = defined $value ? 120000 : 100644;
2946 } else {
2947 $self->{file_prop}->{$fb->{path}} ||= {};
2948 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2949 }
2950 undef;
2951 }
2952
2953 sub apply_textdelta {
2954 my ($self, $fb, $exp) = @_;
2955 my $fh = IO::File->new_tmpfile;
2956 $fh->autoflush(1);
2957 # $fh gets auto-closed() by SVN::TxDelta::apply(),
2958 # (but $base does not,) so dup() it for reading in close_file
2959 open my $dup, '<&', $fh or croak $!;
2960 my $base = IO::File->new_tmpfile;
2961 $base->autoflush(1);
2962 if ($fb->{blob}) {
2963 defined (my $pid = fork) or croak $!;
2964 if (!$pid) {
2965 open STDOUT, '>&', $base or croak $!;
2966 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2967 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2968 }
2969 waitpid $pid, 0;
2970 croak $? if $?;
2971
2972 if (defined $exp) {
2973 seek $base, 0, 0 or croak $!;
2974 my $got = Git::SVN::Util::md5sum($base);
2975 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2976 "expected: $exp\n",
2977 " got: $got\n" if ($got ne $exp);
2978 }
2979 }
2980 seek $base, 0, 0 or croak $!;
2981 $fb->{fh} = $dup;
2982 $fb->{base} = $base;
2983 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2984 }
2985
2986 sub close_file {
2987 my ($self, $fb, $exp) = @_;
2988 my $hash;
2989 my $path = $self->git_path($fb->{path});
2990 if (my $fh = $fb->{fh}) {
2991 if (defined $exp) {
2992 seek($fh, 0, 0) or croak $!;
2993 my $got = Git::SVN::Util::md5sum($fh);
2994 if ($got ne $exp) {
2995 die "Checksum mismatch: $path\n",
2996 "expected: $exp\n got: $got\n";
2997 }
2998 }
2999 sysseek($fh, 0, 0) or croak $!;
3000 if ($fb->{mode_b} == 120000) {
3001 sysread($fh, my $buf, 5) == 5 or croak $!;
3002 $buf eq 'link ' or die "$path has mode 120000",
3003 "but is not a link\n";
3004 }
3005 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
3006 if (!$pid) {
3007 open STDIN, '<&', $fh or croak $!;
3008 exec qw/git-hash-object -w --stdin/ or croak $!;
3009 }
3010 chomp($hash = do { local $/; <$out> });
3011 close $out or croak $!;
3012 close $fh or croak $!;
3013 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3014 close $fb->{base} or croak $!;
3015 } else {
3016 $hash = $fb->{blob} or die "no blob information\n";
3017 }
3018 $fb->{pool}->clear;
3019 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3020 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3021 undef;
3022 }
3023
3024 sub abort_edit {
3025 my $self = shift;
3026 $self->{nr} = $self->{gii}->{nr};
3027 delete $self->{gii};
3028 $self->SUPER::abort_edit(@_);
3029 }
3030
3031 sub close_edit {
3032 my $self = shift;
3033 $self->{git_commit_ok} = 1;
3034 $self->{nr} = $self->{gii}->{nr};
3035 delete $self->{gii};
3036 $self->SUPER::close_edit(@_);
3037 }
3038
3039 package SVN::Git::Editor;
3040 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3041 use strict;
3042 use warnings;
3043 use Carp qw/croak/;
3044 use IO::File;
3045
3046 sub new {
3047 my ($class, $opts) = @_;
3048 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3049 die "$_ required!\n" unless (defined $opts->{$_});
3050 }
3051
3052 my $pool = SVN::Pool->new;
3053 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3054 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3055 $opts->{r}, $mods);
3056
3057 # $opts->{ra} functions should not be used after this:
3058 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3059 $opts->{editor_cb}, $pool);
3060 my $self = SVN::Delta::Editor->new(@ce, $pool);
3061 bless $self, $class;
3062 foreach (qw/svn_path r tree_a tree_b/) {
3063 $self->{$_} = $opts->{$_};
3064 }
3065 $self->{url} = $opts->{ra}->{url};
3066 $self->{mods} = $mods;
3067 $self->{types} = $types;
3068 $self->{pool} = $pool;
3069 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3070 $self->{rm} = { };
3071 $self->{path_prefix} = length $self->{svn_path} ?
3072 "$self->{svn_path}/" : '';
3073 return $self;
3074 }
3075
3076 sub generate_diff {
3077 my ($tree_a, $tree_b) = @_;
3078 my @diff_tree = qw(diff-tree -z -r);
3079 if ($_cp_similarity) {
3080 push @diff_tree, "-C$_cp_similarity";
3081 } else {
3082 push @diff_tree, '-C';
3083 }
3084 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3085 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3086 push @diff_tree, $tree_a, $tree_b;
3087 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3088 local $/ = "\0";
3089 my $state = 'meta';
3090 my @mods;
3091 while (<$diff_fh>) {
3092 chomp $_; # this gets rid of the trailing "\0"
3093 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3094 $::sha1\s($::sha1)\s
3095 ([MTCRAD])\d*$/xo) {
3096 push @mods, { mode_a => $1, mode_b => $2,
3097 sha1_b => $3, chg => $4 };
3098 if ($4 =~ /^(?:C|R)$/) {
3099 $state = 'file_a';
3100 } else {
3101 $state = 'file_b';
3102 }
3103 } elsif ($state eq 'file_a') {
3104 my $x = $mods[$#mods] or croak "Empty array\n";
3105 if ($x->{chg} !~ /^(?:C|R)$/) {
3106 croak "Error parsing $_, $x->{chg}\n";
3107 }
3108 $x->{file_a} = $_;
3109 $state = 'file_b';
3110 } elsif ($state eq 'file_b') {
3111 my $x = $mods[$#mods] or croak "Empty array\n";
3112 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3113 croak "Error parsing $_, $x->{chg}\n";
3114 }
3115 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3116 croak "Error parsing $_, $x->{chg}\n";
3117 }
3118 $x->{file_b} = $_;
3119 $state = 'meta';
3120 } else {
3121 croak "Error parsing $_\n";
3122 }
3123 }
3124 command_close_pipe($diff_fh, $ctx);
3125 \@mods;
3126 }
3127
3128 sub check_diff_paths {
3129 my ($ra, $pfx, $rev, $mods) = @_;
3130 my %types;
3131 $pfx .= '/' if length $pfx;
3132
3133 sub type_diff_paths {
3134 my ($ra, $types, $path, $rev) = @_;
3135 my @p = split m#/+#, $path;
3136 my $c = shift @p;
3137 unless (defined $types->{$c}) {
3138 $types->{$c} = $ra->check_path($c, $rev);
3139 }
3140 while (@p) {
3141 $c .= '/' . shift @p;
3142 next if defined $types->{$c};
3143 $types->{$c} = $ra->check_path($c, $rev);
3144 }
3145 }
3146
3147 foreach my $m (@$mods) {
3148 foreach my $f (qw/file_a file_b/) {
3149 next unless defined $m->{$f};
3150 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3151 if (length $pfx.$dir && ! defined $types{$dir}) {
3152 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3153 }
3154 }
3155 }
3156 \%types;
3157 }
3158
3159 sub split_path {
3160 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3161 }
3162
3163 sub repo_path {
3164 my ($self, $path) = @_;
3165 $self->{path_prefix}.(defined $path ? $path : '');
3166 }
3167
3168 sub url_path {
3169 my ($self, $path) = @_;
3170 if ($self->{url} =~ m#^https?://#) {
3171 $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3172 }
3173 $self->{url} . '/' . $self->repo_path($path);
3174 }
3175
3176 sub rmdirs {
3177 my ($self) = @_;
3178 my $rm = $self->{rm};
3179 delete $rm->{''}; # we never delete the url we're tracking
3180 return unless %$rm;
3181
3182 foreach (keys %$rm) {
3183 my @d = split m#/#, $_;
3184 my $c = shift @d;
3185 $rm->{$c} = 1;
3186 while (@d) {
3187 $c .= '/' . shift @d;
3188 $rm->{$c} = 1;
3189 }
3190 }
3191 delete $rm->{$self->{svn_path}};
3192 delete $rm->{''}; # we never delete the url we're tracking
3193 return unless %$rm;
3194
3195 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3196 $self->{tree_b});
3197 local $/ = "\0";
3198 while (<$fh>) {
3199 chomp;
3200 my @dn = split m#/#, $_;
3201 while (pop @dn) {
3202 delete $rm->{join '/', @dn};
3203 }
3204 unless (%$rm) {
3205 close $fh;
3206 return;
3207 }
3208 }
3209 command_close_pipe($fh, $ctx);
3210
3211 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3212 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3213 $self->close_directory($bat->{$d}, $p);
3214 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3215 print "\tD+\t$d/\n" unless $::_q;
3216 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3217 delete $bat->{$d};
3218 }
3219 }
3220
3221 sub open_or_add_dir {
3222 my ($self, $full_path, $baton) = @_;
3223 my $t = $self->{types}->{$full_path};
3224 if (!defined $t) {
3225 die "$full_path not known in r$self->{r} or we have a bug!\n";
3226 }
3227 {
3228 no warnings 'once';
3229 # SVN::Node::none and SVN::Node::file are used only once,
3230 # so we're shutting up Perl's warnings about them.
3231 if ($t == $SVN::Node::none) {
3232 return $self->add_directory($full_path, $baton,
3233 undef, -1, $self->{pool});
3234 } elsif ($t == $SVN::Node::dir) {
3235 return $self->open_directory($full_path, $baton,
3236 $self->{r}, $self->{pool});
3237 } # no warnings 'once'
3238 print STDERR "$full_path already exists in repository at ",
3239 "r$self->{r} and it is not a directory (",
3240 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3241 } # no warnings 'once'
3242 exit 1;
3243 }
3244
3245 sub ensure_path {
3246 my ($self, $path) = @_;
3247 my $bat = $self->{bat};
3248 my $repo_path = $self->repo_path($path);
3249 return $bat->{''} unless (length $repo_path);
3250 my @p = split m#/+#, $repo_path;
3251 my $c = shift @p;
3252 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3253 while (@p) {
3254 my $c0 = $c;
3255 $c .= '/' . shift @p;
3256 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3257 }
3258 return $bat->{$c};
3259 }
3260
3261 sub A {
3262 my ($self, $m) = @_;
3263 my ($dir, $file) = split_path($m->{file_b});
3264 my $pbat = $self->ensure_path($dir);
3265 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3266 undef, -1);
3267 print "\tA\t$m->{file_b}\n" unless $::_q;
3268 $self->chg_file($fbat, $m);
3269 $self->close_file($fbat,undef,$self->{pool});
3270 }
3271
3272 sub C {
3273 my ($self, $m) = @_;
3274 my ($dir, $file) = split_path($m->{file_b});
3275 my $pbat = $self->ensure_path($dir);
3276 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3277 $self->url_path($m->{file_a}), $self->{r});
3278 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3279 $self->chg_file($fbat, $m);
3280 $self->close_file($fbat,undef,$self->{pool});
3281 }
3282
3283 sub delete_entry {
3284 my ($self, $path, $pbat) = @_;
3285 my $rpath = $self->repo_path($path);
3286 my ($dir, $file) = split_path($rpath);
3287 $self->{rm}->{$dir} = 1;
3288 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3289 }
3290
3291 sub R {
3292 my ($self, $m) = @_;
3293 my ($dir, $file) = split_path($m->{file_b});
3294 my $pbat = $self->ensure_path($dir);
3295 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3296 $self->url_path($m->{file_a}), $self->{r});
3297 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3298 $self->chg_file($fbat, $m);
3299 $self->close_file($fbat,undef,$self->{pool});
3300
3301 ($dir, $file) = split_path($m->{file_a});
3302 $pbat = $self->ensure_path($dir);
3303 $self->delete_entry($m->{file_a}, $pbat);
3304 }
3305
3306 sub M {
3307 my ($self, $m) = @_;
3308 my ($dir, $file) = split_path($m->{file_b});
3309 my $pbat = $self->ensure_path($dir);
3310 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3311 $pbat,$self->{r},$self->{pool});
3312 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3313 $self->chg_file($fbat, $m);
3314 $self->close_file($fbat,undef,$self->{pool});
3315 }
3316
3317 sub T { shift->M(@_) }
3318
3319 sub change_file_prop {
3320 my ($self, $fbat, $pname, $pval) = @_;
3321 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3322 }
3323
3324 sub chg_file {
3325 my ($self, $fbat, $m) = @_;
3326 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3327 $self->change_file_prop($fbat,'svn:executable','*');
3328 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3329 $self->change_file_prop($fbat,'svn:executable',undef);
3330 }
3331 my $fh = IO::File->new_tmpfile or croak $!;
3332 if ($m->{mode_b} =~ /^120/) {
3333 print $fh 'link ' or croak $!;
3334 $self->change_file_prop($fbat,'svn:special','*');
3335 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3336 $self->change_file_prop($fbat,'svn:special',undef);
3337 }
3338 defined(my $pid = fork) or croak $!;
3339 if (!$pid) {
3340 open STDOUT, '>&', $fh or croak $!;
3341 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3342 }
3343 waitpid $pid, 0;
3344 croak $? if $?;
3345 $fh->flush == 0 or croak $!;
3346 seek $fh, 0, 0 or croak $!;
3347
3348 my $exp = Git::SVN::Util::md5sum($fh);
3349 seek $fh, 0, 0 or croak $!;
3350
3351 my $pool = SVN::Pool->new;
3352 my $atd = $self->apply_textdelta($fbat, undef, $pool);
3353 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3354 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3355 $pool->clear;
3356
3357 close $fh or croak $!;
3358 }
3359
3360 sub D {
3361 my ($self, $m) = @_;
3362 my ($dir, $file) = split_path($m->{file_b});
3363 my $pbat = $self->ensure_path($dir);
3364 print "\tD\t$m->{file_b}\n" unless $::_q;
3365 $self->delete_entry($m->{file_b}, $pbat);
3366 }
3367
3368 sub close_edit {
3369 my ($self) = @_;
3370 my ($p,$bat) = ($self->{pool}, $self->{bat});
3371 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3372 next if $_ eq '';
3373 $self->close_directory($bat->{$_}, $p);
3374 }
3375 $self->close_directory($bat->{''}, $p);
3376 $self->SUPER::close_edit($p);
3377 $p->clear;
3378 }
3379
3380 sub abort_edit {
3381 my ($self) = @_;
3382 $self->SUPER::abort_edit($self->{pool});
3383 }
3384
3385 sub DESTROY {
3386 my $self = shift;
3387 $self->SUPER::DESTROY(@_);
3388 $self->{pool}->clear;
3389 }
3390
3391 # this drives the editor
3392 sub apply_diff {
3393 my ($self) = @_;
3394 my $mods = $self->{mods};
3395 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3396 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3397 my $f = $m->{chg};
3398 if (defined $o{$f}) {
3399 $self->$f($m);
3400 } else {
3401 fatal("Invalid change type: $f");
3402 }
3403 }
3404 $self->rmdirs if $_rmdir;
3405 if (@$mods == 0) {
3406 $self->abort_edit;
3407 } else {
3408 $self->close_edit;
3409 }
3410 return scalar @$mods;
3411 }
3412
3413 package Git::SVN::Ra;
3414 use vars qw/@ISA $config_dir $_log_window_size/;
3415 use strict;
3416 use warnings;
3417 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3418
3419 BEGIN {
3420 # enforce temporary pool usage for some simple functions
3421 no strict 'refs';
3422 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3423 my $SUPER = "SUPER::$f";
3424 *$f = sub {
3425 my $self = shift;
3426 my $pool = SVN::Pool->new;
3427 my @ret = $self->$SUPER(@_,$pool);
3428 $pool->clear;
3429 wantarray ? @ret : $ret[0];
3430 };
3431 }
3432 }
3433
3434 sub _auth_providers () {
3435 [
3436 SVN::Client::get_simple_provider(),
3437 SVN::Client::get_ssl_server_trust_file_provider(),
3438 SVN::Client::get_simple_prompt_provider(
3439 \&Git::SVN::Prompt::simple, 2),
3440 SVN::Client::get_ssl_client_cert_file_provider(),
3441 SVN::Client::get_ssl_client_cert_prompt_provider(
3442 \&Git::SVN::Prompt::ssl_client_cert, 2),
3443 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3444 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3445 SVN::Client::get_username_provider(),
3446 SVN::Client::get_ssl_server_trust_prompt_provider(
3447 \&Git::SVN::Prompt::ssl_server_trust),
3448 SVN::Client::get_username_prompt_provider(
3449 \&Git::SVN::Prompt::username, 2)
3450 ]
3451 }
3452
3453 sub escape_uri_only {
3454 my ($uri) = @_;
3455 my @tmp;
3456 foreach (split m{/}, $uri) {
3457 s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
3458 push @tmp, $_;
3459 }
3460 join('/', @tmp);
3461 }
3462
3463 sub escape_url {
3464 my ($url) = @_;
3465 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3466 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3467 $url = "$scheme://$domain$uri";
3468 }
3469 $url;
3470 }
3471
3472 sub new {
3473 my ($class, $url) = @_;
3474 $url =~ s!/+$!!;
3475 return $RA if ($RA && $RA->{url} eq $url);
3476
3477 SVN::_Core::svn_config_ensure($config_dir, undef);
3478 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3479 my $config = SVN::Core::config_get_config($config_dir);
3480 $RA = undef;
3481 my $dont_store_passwords = 1;
3482 my $conf_t = ${$config}{'config'};
3483 {
3484 no warnings 'once';
3485 # The usage of $SVN::_Core::SVN_CONFIG_* variables
3486 # produces warnings that variables are used only once.
3487 # I had not found the better way to shut them up, so
3488 # the warnings of type 'once' are disabled in this block.
3489 if (SVN::_Core::svn_config_get_bool($conf_t,
3490 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3491 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3492 1) == 0) {
3493 SVN::_Core::svn_auth_set_parameter($baton,
3494 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3495 bless (\$dont_store_passwords, "_p_void"));
3496 }
3497 if (SVN::_Core::svn_config_get_bool($conf_t,
3498 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3499 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3500 1) == 0) {
3501 $Git::SVN::Prompt::_no_auth_cache = 1;
3502 }
3503 } # no warnings 'once'
3504 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3505 config => $config,
3506 pool => SVN::Pool->new,
3507 auth_provider_callbacks => $callbacks);
3508 $self->{url} = $url;
3509 $self->{svn_path} = $url;
3510 $self->{repos_root} = $self->get_repos_root;
3511 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3512 $self->{cache} = { check_path => { r => 0, data => {} },
3513 get_dir => { r => 0, data => {} } };
3514 $RA = bless $self, $class;
3515 }
3516
3517 sub check_path {
3518 my ($self, $path, $r) = @_;
3519 my $cache = $self->{cache}->{check_path};
3520 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3521 return $cache->{data}->{$path};
3522 }
3523 my $pool = SVN::Pool->new;
3524 my $t = $self->SUPER::check_path($path, $r, $pool);
3525 $pool->clear;
3526 if ($r != $cache->{r}) {
3527 %{$cache->{data}} = ();
3528 $cache->{r} = $r;
3529 }
3530 $cache->{data}->{$path} = $t;
3531 }
3532
3533 sub get_dir {
3534 my ($self, $dir, $r) = @_;
3535 my $cache = $self->{cache}->{get_dir};
3536 if ($r == $cache->{r}) {
3537 if (my $x = $cache->{data}->{$dir}) {
3538 return wantarray ? @$x : $x->[0];
3539 }
3540 }
3541 my $pool = SVN::Pool->new;
3542 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3543 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3544 $pool->clear;
3545 if ($r != $cache->{r}) {
3546 %{$cache->{data}} = ();
3547 $cache->{r} = $r;
3548 }
3549 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3550 wantarray ? (\%dirents, $r, $props) : \%dirents;
3551 }
3552
3553 sub DESTROY {
3554 # do not call the real DESTROY since we store ourselves in $RA
3555 }
3556
3557 sub get_log {
3558 my ($self, @args) = @_;
3559 my $pool = SVN::Pool->new;
3560 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3561 my $ret = $self->SUPER::get_log(@args, $pool);
3562 $pool->clear;
3563 $ret;
3564 }
3565
3566 sub trees_match {
3567 my ($self, $url1, $rev1, $url2, $rev2) = @_;
3568 my $ctx = SVN::Client->new(auth => _auth_providers);
3569 my $out = IO::File->new_tmpfile;
3570
3571 # older SVN (1.1.x) doesn't take $pool as the last parameter for
3572 # $ctx->diff(), so we'll create a default one
3573 my $pool = SVN::Pool->new_default_sub;
3574
3575 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3576 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3577 $out->flush;
3578 my $ret = (($out->stat)[7] == 0);
3579 close $out or croak $!;
3580
3581 $ret;
3582 }
3583
3584 sub get_commit_editor {
3585 my ($self, $log, $cb, $pool) = @_;
3586 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3587 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3588 }
3589
3590 sub gs_do_update {
3591 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3592 my $new = ($rev_a == $rev_b);
3593 my $path = $gs->{path};
3594
3595 if ($new && -e $gs->{index}) {
3596 unlink $gs->{index} or die
3597 "Couldn't unlink index: $gs->{index}: $!\n";
3598 }
3599 my $pool = SVN::Pool->new;
3600 $editor->set_path_strip($path);
3601 my (@pc) = split m#/#, $path;
3602 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3603 1, $editor, $pool);
3604 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3605
3606 # Since we can't rely on svn_ra_reparent being available, we'll
3607 # just have to do some magic with set_path to make it so
3608 # we only want a partial path.
3609 my $sp = '';
3610 my $final = join('/', @pc);
3611 while (@pc) {
3612 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3613 $sp .= '/' if length $sp;
3614 $sp .= shift @pc;
3615 }
3616 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3617
3618 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3619
3620 $reporter->finish_report($pool);
3621 $pool->clear;
3622 $editor->{git_commit_ok};
3623 }
3624
3625 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3626 # svn_ra_reparent didn't work before 1.4)
3627 sub gs_do_switch {
3628 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3629 my $path = $gs->{path};
3630 my $pool = SVN::Pool->new;
3631
3632 my $full_url = $self->{url};
3633 my $old_url = $full_url;
3634 $full_url .= '/' . escape_uri_only($path) if length $path;
3635 my ($ra, $reparented);
3636 if ($old_url ne $full_url) {
3637 if ($old_url !~ m#^svn(\+ssh)?://#) {
3638 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3639 $pool);
3640 $self->{url} = $full_url;
3641 $reparented = 1;
3642 } else {
3643 $_[0] = undef;
3644 $self = undef;
3645 $RA = undef;
3646 $ra = Git::SVN::Ra->new($full_url);
3647 $ra_invalid = 1;
3648 }
3649 }
3650 $ra ||= $self;
3651 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3652 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3653 $reporter->set_path('', $rev_a, 0, @lock, $pool);
3654 $reporter->finish_report($pool);
3655
3656 if ($reparented) {
3657 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3658 $self->{url} = $old_url;
3659 }
3660
3661 $pool->clear;
3662 $editor->{git_commit_ok};
3663 }
3664
3665 sub longest_common_path {
3666 my ($gsv, $globs) = @_;
3667 my %common;
3668 my $common_max = scalar @$gsv;
3669
3670 foreach my $gs (@$gsv) {
3671 my @tmp = split m#/#, $gs->{path};
3672 my $p = '';
3673 foreach (@tmp) {
3674 $p .= length($p) ? "/$_" : $_;
3675 $common{$p} ||= 0;
3676 $common{$p}++;
3677 }
3678 }
3679 $globs ||= [];
3680 $common_max += scalar @$globs;
3681 foreach my $glob (@$globs) {
3682 my @tmp = split m#/#, $glob->{path}->{left};
3683 my $p = '';
3684 foreach (@tmp) {
3685 $p .= length($p) ? "/$_" : $_;
3686 $common{$p} ||= 0;
3687 $common{$p}++;
3688 }
3689 }
3690
3691 my $longest_path = '';
3692 foreach (sort {length $b <=> length $a} keys %common) {
3693 if ($common{$_} == $common_max) {
3694 $longest_path = $_;
3695 last;
3696 }
3697 }
3698 $longest_path;
3699 }
3700
3701 sub gs_fetch_loop_common {
3702 my ($self, $base, $head, $gsv, $globs) = @_;
3703 return if ($base > $head);
3704 my $inc = $_log_window_size;
3705 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3706 my $longest_path = longest_common_path($gsv, $globs);
3707 my $ra_url = $self->{url};
3708 while (1) {
3709 my %revs;
3710 my $err;
3711 my $err_handler = $SVN::Error::handler;
3712 $SVN::Error::handler = sub {
3713 ($err) = @_;
3714 skip_unknown_revs($err);
3715 };
3716 sub _cb {
3717 my ($paths, $r, $author, $date, $log) = @_;
3718 [ dup_changed_paths($paths),
3719 { author => $author, date => $date, log => $log } ];
3720 }
3721 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3722 sub { $revs{$_[1]} = _cb(@_) });
3723 if ($err && $max >= $head) {
3724 print STDERR "Path '$longest_path' ",
3725 "was probably deleted:\n",
3726 $err->expanded_message,
3727 "\nWill attempt to follow ",
3728 "revisions r$min .. r$max ",
3729 "committed before the deletion\n";
3730 my $hi = $max;
3731 while (--$hi >= $min) {
3732 my $ok;
3733 $self->get_log([$longest_path], $min, $hi,
3734 0, 1, 1, sub {
3735 $ok ||= $_[1];
3736 $revs{$_[1]} = _cb(@_) });
3737 if ($ok) {
3738 print STDERR "r$min .. r$ok OK\n";
3739 last;
3740 }
3741 }
3742 }
3743 $SVN::Error::handler = $err_handler;
3744
3745 my %exists = map { $_->{path} => $_ } @$gsv;
3746 foreach my $r (sort {$a <=> $b} keys %revs) {
3747 my ($paths, $logged) = @{$revs{$r}};
3748
3749 foreach my $gs ($self->match_globs(\%exists, $paths,
3750 $globs, $r)) {
3751 if ($gs->rev_db_max >= $r) {
3752 next;
3753 }
3754 next unless $gs->match_paths($paths, $r);
3755 $gs->{logged_rev_props} = $logged;
3756 if (my $last_commit = $gs->last_commit) {
3757 $gs->assert_index_clean($last_commit);
3758 }
3759 my $log_entry = $gs->do_fetch($paths, $r);
3760 if ($log_entry) {
3761 $gs->do_git_commit($log_entry);
3762 }
3763 }
3764 foreach my $g (@$globs) {
3765 my $k = "svn-remote.$g->{remote}." .
3766 "$g->{t}-maxRev";
3767 Git::SVN::tmp_config($k, $r);
3768 }
3769 if ($ra_invalid) {
3770 $_[0] = undef;
3771 $self = undef;
3772 $RA = undef;
3773 $self = Git::SVN::Ra->new($ra_url);
3774 $ra_invalid = undef;
3775 }
3776 }
3777 # pre-fill the .rev_db since it'll eventually get filled in
3778 # with '0' x40 if something new gets committed
3779 foreach my $gs (@$gsv) {
3780 next if defined $gs->rev_db_get($max);
3781 $gs->rev_db_set($max, 0 x40);
3782 }
3783 foreach my $g (@$globs) {
3784 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3785 Git::SVN::tmp_config($k, $max);
3786 }
3787 last if $max >= $head;
3788 $min = $max + 1;
3789 $max += $inc;
3790 $max = $head if ($max > $head);
3791 }
3792 }
3793
3794 sub match_globs {
3795 my ($self, $exists, $paths, $globs, $r) = @_;
3796
3797 sub get_dir_check {
3798 my ($self, $exists, $g, $r) = @_;
3799 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3800 return unless scalar @x == 3;
3801 my $dirents = $x[0];
3802 foreach my $de (keys %$dirents) {
3803 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
3804 my $p = $g->{path}->full_path($de);
3805 next if $exists->{$p};
3806 next if (length $g->{path}->{right} &&
3807 ($self->check_path($p, $r) !=
3808 $SVN::Node::dir));
3809 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3810 $g->{ref}->full_path($de), 1);
3811 }
3812 }
3813 foreach my $g (@$globs) {
3814 if (my $path = $paths->{"/$g->{path}->{left}"}) {
3815 if ($path->{action} =~ /^[AR]$/) {
3816 get_dir_check($self, $exists, $g, $r);
3817 }
3818 }
3819 foreach (keys %$paths) {
3820 if (/$g->{path}->{left_regex}/ &&
3821 !/$g->{path}->{regex}/) {
3822 next if $paths->{$_}->{action} !~ /^[AR]$/;
3823 get_dir_check($self, $exists, $g, $r);
3824 }
3825 next unless /$g->{path}->{regex}/;
3826 my $p = $1;
3827 my $pathname = $g->{path}->full_path($p);
3828 next if $exists->{$pathname};
3829 next if ($self->check_path($pathname, $r) !=
3830 $SVN::Node::dir);
3831 $exists->{$pathname} = Git::SVN->init(
3832 $self->{url}, $pathname, undef,
3833 $g->{ref}->full_path($p), 1);
3834 }
3835 my $c = '';
3836 foreach (split m#/#, $g->{path}->{left}) {
3837 $c .= "/$_";
3838 next unless ($paths->{$c} &&
3839 ($paths->{$c}->{action} =~ /^[AR]$/));
3840 get_dir_check($self, $exists, $g, $r);
3841 }
3842 }
3843 values %$exists;
3844 }
3845
3846 sub minimize_url {
3847 my ($self) = @_;
3848 return $self->{url} if ($self->{url} eq $self->{repos_root});
3849 my $url = $self->{repos_root};
3850 my @components = split(m!/!, $self->{svn_path});
3851 my $c = '';
3852 do {
3853 $url .= "/$c" if length $c;
3854 eval { (ref $self)->new($url)->get_latest_revnum };
3855 } while ($@ && ($c = shift @components));
3856 $url;
3857 }
3858
3859 sub can_do_switch {
3860 my $self = shift;
3861 unless (defined $can_do_switch) {
3862 my $pool = SVN::Pool->new;
3863 my $rep = eval {
3864 $self->do_switch(1, '', 0, $self->{url},
3865 SVN::Delta::Editor->new, $pool);
3866 };
3867 if ($@) {
3868 $can_do_switch = 0;
3869 } else {
3870 $rep->abort_report($pool);
3871 $can_do_switch = 1;
3872 }
3873 $pool->clear;
3874 }
3875 $can_do_switch;
3876 }
3877
3878 sub skip_unknown_revs {
3879 my ($err) = @_;
3880 my $errno = $err->apr_err();
3881 # Maybe the branch we're tracking didn't
3882 # exist when the repo started, so it's
3883 # not an error if it doesn't, just continue
3884 #
3885 # Wonderfully consistent library, eh?
3886 # 160013 - svn:// and file://
3887 # 175002 - http(s)://
3888 # 175007 - http(s):// (this repo required authorization, too...)
3889 # More codes may be discovered later...
3890 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3891 my $err_key = $err->expanded_message;
3892 # revision numbers change every time, filter them out
3893 $err_key =~ s/\d+/\0/g;
3894 $err_key = "$errno\0$err_key";
3895 unless ($ignored_err{$err_key}) {
3896 warn "W: Ignoring error from SVN, path probably ",
3897 "does not exist: ($errno): ",
3898 $err->expanded_message,"\n";
3899 $ignored_err{$err_key} = 1;
3900 }
3901 return;
3902 }
3903 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3904 }
3905
3906 # svn_log_changed_path_t objects passed to get_log are likely to be
3907 # overwritten even if only the refs are copied to an external variable,
3908 # so we should dup the structures in their entirety. Using an externally
3909 # passed pool (instead of our temporary and quickly cleared pool in
3910 # Git::SVN::Ra) does not help matters at all...
3911 sub dup_changed_paths {
3912 my ($paths) = @_;
3913 return undef unless $paths;
3914 my %ret;
3915 foreach my $p (keys %$paths) {
3916 my $i = $paths->{$p};
3917 my %s = map { $_ => $i->$_ }
3918 qw/copyfrom_path copyfrom_rev action/;
3919 $ret{$p} = \%s;
3920 }
3921 \%ret;
3922 }
3923
3924 package Git::SVN::Log;
3925 use strict;
3926 use warnings;
3927 use POSIX qw/strftime/;
3928 use constant commit_log_separator => ('-' x 72) . "\n";
3929 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3930 %rusers $show_commit $incremental/;
3931 my $l_fmt;
3932
3933 sub cmt_showable {
3934 my ($c) = @_;
3935 return 1 if defined $c->{r};
3936
3937 # big commit message got truncated by the 16k pretty buffer in rev-list
3938 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3939 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3940 @{$c->{l}} = ();
3941 my @log = command(qw/cat-file commit/, $c->{c});
3942
3943 # shift off the headers
3944 shift @log while ($log[0] ne '');
3945 shift @log;
3946
3947 # TODO: make $c->{l} not have a trailing newline in the future
3948 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
3949
3950 (undef, $c->{r}, undef) = ::extract_metadata(
3951 (grep(/^git-svn-id: /, @log))[-1]);
3952 }
3953 return defined $c->{r};
3954 }
3955
3956 sub log_use_color {
3957 return 1 if $color;
3958 my ($dc, $dcvar);
3959 $dcvar = 'color.diff';
3960 $dc = `git-config --get $dcvar`;
3961 if ($dc eq '') {
3962 # nothing at all; fallback to "diff.color"
3963 $dcvar = 'diff.color';
3964 $dc = `git-config --get $dcvar`;
3965 }
3966 chomp($dc);
3967 if ($dc eq 'auto') {
3968 my $pc;
3969 $pc = `git-config --get color.pager`;
3970 if ($pc eq '') {
3971 # does not have it -- fallback to pager.color
3972 $pc = `git-config --bool --get pager.color`;
3973 }
3974 else {
3975 $pc = `git-config --bool --get color.pager`;
3976 if ($?) {
3977 $pc = 'false';
3978 }
3979 }
3980 chomp($pc);
3981 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3982 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3983 }
3984 return 0;
3985 }
3986 return 0 if $dc eq 'never';
3987 return 1 if $dc eq 'always';
3988 chomp($dc = `git-config --bool --get $dcvar`);
3989 return ($dc eq 'true');
3990 }
3991
3992 sub git_svn_log_cmd {
3993 my ($r_min, $r_max, @args) = @_;
3994 my $head = 'HEAD';
3995 my (@files, @log_opts);
3996 foreach my $x (@args) {
3997 if ($x eq '--' || @files) {
3998 push @files, $x;
3999 } else {
4000 if (::verify_ref("$x^0")) {
4001 $head = $x;
4002 } else {
4003 push @log_opts, $x;
4004 }
4005 }
4006 }
4007
4008 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4009 $gs ||= Git::SVN->_new;
4010 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4011 $gs->refname);
4012 push @cmd, '-r' unless $non_recursive;
4013 push @cmd, qw/--raw --name-status/ if $verbose;
4014 push @cmd, '--color' if log_use_color();
4015 push @cmd, @log_opts;
4016 if (defined $r_max && $r_max == $r_min) {
4017 push @cmd, '--max-count=1';
4018 if (my $c = $gs->rev_db_get($r_max)) {
4019 push @cmd, $c;
4020 }
4021 } elsif (defined $r_max) {
4022 if ($r_max < $r_min) {
4023 ($r_min, $r_max) = ($r_max, $r_min);
4024 }
4025 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4026 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4027 # If there are no commits in the range, both $c_max and $c_min
4028 # will be undefined. If there is at least 1 commit in the
4029 # range, both will be defined.
4030 return () if !defined $c_min || !defined $c_max;
4031 if ($c_min eq $c_max) {
4032 push @cmd, '--max-count=1', $c_min;
4033 } else {
4034 push @cmd, '--boundary', "$c_min..$c_max";
4035 }
4036 }
4037 return (@cmd, @files);
4038 }
4039
4040 # adapted from pager.c
4041 sub config_pager {
4042 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4043 if (!defined $pager) {
4044 $pager = 'less';
4045 } elsif (length $pager == 0 || $pager eq 'cat') {
4046 $pager = undef;
4047 }
4048 }
4049
4050 sub run_pager {
4051 return unless -t *STDOUT && defined $pager;
4052 pipe my $rfd, my $wfd or return;
4053 defined(my $pid = fork) or ::fatal "Can't fork: $!";
4054 if (!$pid) {
4055 open STDOUT, '>&', $wfd or
4056 ::fatal "Can't redirect to stdout: $!";
4057 return;
4058 }
4059 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4060 $ENV{LESS} ||= 'FRSX';
4061 exec $pager or ::fatal "Can't run pager: $! ($pager)";
4062 }
4063
4064 sub format_svn_date {
4065 return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4066 }
4067
4068 sub parse_git_date {
4069 my ($t, $tz) = @_;
4070 # Date::Parse isn't in the standard Perl distro :(
4071 if ($tz =~ s/^\+//) {
4072 $t += tz_to_s_offset($tz);
4073 } elsif ($tz =~ s/^\-//) {
4074 $t -= tz_to_s_offset($tz);
4075 }
4076 return $t;
4077 }
4078
4079 sub set_local_timezone {
4080 if (defined $TZ) {
4081 $ENV{TZ} = $TZ;
4082 } else {
4083 delete $ENV{TZ};
4084 }
4085 }
4086
4087 sub tz_to_s_offset {
4088 my ($tz) = @_;
4089 $tz =~ s/(\d\d)$//;
4090 return ($1 * 60) + ($tz * 3600);
4091 }
4092
4093 sub get_author_info {
4094 my ($dest, $author, $t, $tz) = @_;
4095 $author =~ s/(?:^\s*|\s*$)//g;
4096 $dest->{a_raw} = $author;
4097 my $au;
4098 if ($::_authors) {
4099 $au = $rusers{$author} || undef;
4100 }
4101 if (!$au) {
4102 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4103 }
4104 $dest->{t} = $t;
4105 $dest->{tz} = $tz;
4106 $dest->{a} = $au;
4107 $dest->{t_utc} = parse_git_date($t, $tz);
4108 }
4109
4110 sub process_commit {
4111 my ($c, $r_min, $r_max, $defer) = @_;
4112 if (defined $r_min && defined $r_max) {
4113 if ($r_min == $c->{r} && $r_min == $r_max) {
4114 show_commit($c);
4115 return 0;
4116 }
4117 return 1 if $r_min == $r_max;
4118 if ($r_min < $r_max) {
4119 # we need to reverse the print order
4120 return 0 if (defined $limit && --$limit < 0);
4121 push @$defer, $c;
4122 return 1;
4123 }
4124 if ($r_min != $r_max) {
4125 return 1 if ($r_min < $c->{r});
4126 return 1 if ($r_max > $c->{r});
4127 }
4128 }
4129 return 0 if (defined $limit && --$limit < 0);
4130 show_commit($c);
4131 return 1;
4132 }
4133
4134 sub show_commit {
4135 my $c = shift;
4136 if ($oneline) {
4137 my $x = "\n";
4138 if (my $l = $c->{l}) {
4139 while ($l->[0] =~ /^\s*$/) { shift @$l }
4140 $x = $l->[0];
4141 }
4142 $l_fmt ||= 'A' . length($c->{r});
4143 print 'r',pack($l_fmt, $c->{r}),' | ';
4144 print "$c->{c} | " if $show_commit;
4145 print $x;
4146 } else {
4147 show_commit_normal($c);
4148 }
4149 }
4150
4151 sub show_commit_changed_paths {
4152 my ($c) = @_;
4153 return unless $c->{changed};
4154 print "Changed paths:\n", @{$c->{changed}};
4155 }
4156
4157 sub show_commit_normal {
4158 my ($c) = @_;
4159 print commit_log_separator, "r$c->{r} | ";
4160 print "$c->{c} | " if $show_commit;
4161 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4162 my $nr_line = 0;
4163
4164 if (my $l = $c->{l}) {
4165 while ($l->[$#$l] eq "\n" && $#$l > 0
4166 && $l->[($#$l - 1)] eq "\n") {
4167 pop @$l;
4168 }
4169 $nr_line = scalar @$l;
4170 if (!$nr_line) {
4171 print "1 line\n\n\n";
4172 } else {
4173 if ($nr_line == 1) {
4174 $nr_line = '1 line';
4175 } else {
4176 $nr_line .= ' lines';
4177 }
4178 print $nr_line, "\n";
4179 show_commit_changed_paths($c);
4180 print "\n";
4181 print $_ foreach @$l;
4182 }
4183 } else {
4184 print "1 line\n";
4185 show_commit_changed_paths($c);
4186 print "\n";
4187
4188 }
4189 foreach my $x (qw/raw stat diff/) {
4190 if ($c->{$x}) {
4191 print "\n";
4192 print $_ foreach @{$c->{$x}}
4193 }
4194 }
4195 }
4196
4197 sub cmd_show_log {
4198 my (@args) = @_;
4199 my ($r_min, $r_max);
4200 my $r_last = -1; # prevent dupes
4201 set_local_timezone();
4202 if (defined $::_revision) {
4203 if ($::_revision =~ /^(\d+):(\d+)$/) {
4204 ($r_min, $r_max) = ($1, $2);
4205 } elsif ($::_revision =~ /^\d+$/) {
4206 $r_min = $r_max = $::_revision;
4207 } else {
4208 ::fatal "-r$::_revision is not supported, use ",
4209 "standard 'git log' arguments instead";
4210 }
4211 }
4212
4213 config_pager();
4214 @args = git_svn_log_cmd($r_min, $r_max, @args);
4215 if (!@args) {
4216 print commit_log_separator unless $incremental || $oneline;
4217 return;
4218 }
4219 my $log = command_output_pipe(@args);
4220 run_pager();
4221 my (@k, $c, $d, $stat);
4222 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4223 while (<$log>) {
4224 if (/^${esc_color}commit -?($::sha1_short)/o) {
4225 my $cmt = $1;
4226 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4227 $r_last = $c->{r};
4228 process_commit($c, $r_min, $r_max, \@k) or
4229 goto out;
4230 }
4231 $d = undef;
4232 $c = { c => $cmt };
4233 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4234 get_author_info($c, $1, $2, $3);
4235 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4236 # ignore
4237 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4238 push @{$c->{raw}}, $_;
4239 } elsif (/^${esc_color}[ACRMDT]\t/) {
4240 # we could add $SVN->{svn_path} here, but that requires
4241 # remote access at the moment (repo_path_split)...
4242 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
4243 push @{$c->{changed}}, $_;
4244 } elsif (/^${esc_color}diff /o) {
4245 $d = 1;
4246 push @{$c->{diff}}, $_;
4247 } elsif ($d) {
4248 push @{$c->{diff}}, $_;
4249 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4250 $esc_color*[\+\-]*$esc_color$/x) {
4251 $stat = 1;
4252 push @{$c->{stat}}, $_;
4253 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4254 push @{$c->{stat}}, $_;
4255 $stat = undef;
4256 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
4257 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4258 } elsif (s/^${esc_color} //o) {
4259 push @{$c->{l}}, $_;
4260 }
4261 }
4262 if ($c && defined $c->{r} && $c->{r} != $r_last) {
4263 $r_last = $c->{r};
4264 process_commit($c, $r_min, $r_max, \@k);
4265 }
4266 if (@k) {
4267 ($r_min, $r_max) = ($r_max, $r_min);
4268 process_commit($_, $r_min, $r_max) foreach reverse @k;
4269 }
4270 out:
4271 close $log;
4272 print commit_log_separator unless $incremental || $oneline;
4273 }
4274
4275 package Git::SVN::Migration;
4276 # these version numbers do NOT correspond to actual version numbers
4277 # of git nor git-svn. They are just relative.
4278 #
4279 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4280 #
4281 # v1 layout: .git/$id/info/url, refs/remotes/$id
4282 #
4283 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4284 #
4285 # v3 layout: .git/svn/$id, refs/remotes/$id
4286 # - info/url may remain for backwards compatibility
4287 # - this is what we migrate up to this layout automatically,
4288 # - this will be used by git svn init on single branches
4289 # v3.1 layout (auto migrated):
4290 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4291 # for backwards compatibility
4292 #
4293 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4294 # - this is only created for newly multi-init-ed
4295 # repositories. Similar in spirit to the
4296 # --use-separate-remotes option in git-clone (now default)
4297 # - we do not automatically migrate to this (following
4298 # the example set by core git)
4299 use strict;
4300 use warnings;
4301 use Carp qw/croak/;
4302 use File::Path qw/mkpath/;
4303 use File::Basename qw/dirname basename/;
4304 use vars qw/$_minimize/;
4305
4306 sub migrate_from_v0 {
4307 my $git_dir = $ENV{GIT_DIR};
4308 return undef unless -d $git_dir;
4309 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4310 my $migrated = 0;
4311 while (<$fh>) {
4312 chomp;
4313 my ($id, $orig_ref) = ($_, $_);
4314 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4315 next unless -f "$git_dir/$id/info/url";
4316 my $new_ref = "refs/remotes/$id";
4317 if (::verify_ref("$new_ref^0")) {
4318 print STDERR "W: $orig_ref is probably an old ",
4319 "branch used by an ancient version of ",
4320 "git-svn.\n",
4321 "However, $new_ref also exists.\n",
4322 "We will not be able ",
4323 "to use this branch until this ",
4324 "ambiguity is resolved.\n";
4325 next;
4326 }
4327 print STDERR "Migrating from v0 layout...\n" if !$migrated;
4328 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4329 command_noisy('update-ref', $new_ref, $orig_ref);
4330 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4331 $migrated++;
4332 }
4333 command_close_pipe($fh, $ctx);
4334 print STDERR "Done migrating from v0 layout...\n" if $migrated;
4335 $migrated;
4336 }
4337
4338 sub migrate_from_v1 {
4339 my $git_dir = $ENV{GIT_DIR};
4340 my $migrated = 0;
4341 return $migrated unless -d $git_dir;
4342 my $svn_dir = "$git_dir/svn";
4343
4344 # just in case somebody used 'svn' as their $id at some point...
4345 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4346
4347 print STDERR "Migrating from a git-svn v1 layout...\n";
4348 mkpath([$svn_dir]);
4349 print STDERR "Data from a previous version of git-svn exists, but\n\t",
4350 "$svn_dir\n\t(required for this version ",
4351 "($::VERSION) of git-svn) does not. exist\n";
4352 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4353 while (<$fh>) {
4354 my $x = $_;
4355 next unless $x =~ s#^refs/remotes/##;
4356 chomp $x;
4357 next unless -f "$git_dir/$x/info/url";
4358 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4359 next unless $u;
4360 my $dn = dirname("$git_dir/svn/$x");
4361 mkpath([$dn]) unless -d $dn;
4362 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4363 mkpath(["$git_dir/svn/svn"]);
4364 print STDERR " - $git_dir/$x/info => ",
4365 "$git_dir/svn/$x/info\n";
4366 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4367 croak "$!: $x";
4368 # don't worry too much about these, they probably
4369 # don't exist with repos this old (save for index,
4370 # and we can easily regenerate that)
4371 foreach my $f (qw/unhandled.log index .rev_db/) {
4372 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4373 }
4374 } else {
4375 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4376 rename "$git_dir/$x", "$git_dir/svn/$x" or
4377 croak "$!: $x";
4378 }
4379 $migrated++;
4380 }
4381 command_close_pipe($fh, $ctx);
4382 print STDERR "Done migrating from a git-svn v1 layout\n";
4383 $migrated;
4384 }
4385
4386 sub read_old_urls {
4387 my ($l_map, $pfx, $path) = @_;
4388 my @dir;
4389 foreach (<$path/*>) {
4390 if (-r "$_/info/url") {
4391 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4392 my $ref_id = $pfx . basename $_;
4393 my $url = ::file_to_s("$_/info/url");
4394 $l_map->{$ref_id} = $url;
4395 } elsif (-d $_) {
4396 push @dir, $_;
4397 }
4398 }
4399 foreach (@dir) {
4400 my $x = $_;
4401 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4402 read_old_urls($l_map, $x, $_);
4403 }
4404 }
4405
4406 sub migrate_from_v2 {
4407 my @cfg = command(qw/config -l/);
4408 return if grep /^svn-remote\..+\.url=/, @cfg;
4409 my %l_map;
4410 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4411 my $migrated = 0;
4412
4413 foreach my $ref_id (sort keys %l_map) {
4414 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4415 if ($@) {
4416 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4417 }
4418 $migrated++;
4419 }
4420 $migrated;
4421 }
4422
4423 sub minimize_connections {
4424 my $r = Git::SVN::read_all_remotes();
4425 my $new_urls = {};
4426 my $root_repos = {};
4427 foreach my $repo_id (keys %$r) {
4428 my $url = $r->{$repo_id}->{url} or next;
4429 my $fetch = $r->{$repo_id}->{fetch} or next;
4430 my $ra = Git::SVN::Ra->new($url);
4431
4432 # skip existing cases where we already connect to the root
4433 if (($ra->{url} eq $ra->{repos_root}) ||
4434 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
4435 $repo_id)) {
4436 $root_repos->{$ra->{url}} = $repo_id;
4437 next;
4438 }
4439
4440 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4441 my $root_path = $ra->{url};
4442 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4443 foreach my $path (keys %$fetch) {
4444 my $ref_id = $fetch->{$path};
4445 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4446
4447 # make sure we can read when connecting to
4448 # a higher level of a repository
4449 my ($last_rev, undef) = $gs->last_rev_commit;
4450 if (!defined $last_rev) {
4451 $last_rev = eval {
4452 $root_ra->get_latest_revnum;
4453 };
4454 next if $@;
4455 }
4456 my $new = $root_path;
4457 $new .= length $path ? "/$path" : '';
4458 eval {
4459 $root_ra->get_log([$new], $last_rev, $last_rev,
4460 0, 0, 1, sub { });
4461 };
4462 next if $@;
4463 $new_urls->{$ra->{repos_root}}->{$new} =
4464 { ref_id => $ref_id,
4465 old_repo_id => $repo_id,
4466 old_path => $path };
4467 }
4468 }
4469
4470 my @emptied;
4471 foreach my $url (keys %$new_urls) {
4472 # see if we can re-use an existing [svn-remote "repo_id"]
4473 # instead of creating a(n ugly) new section:
4474 my $repo_id = $root_repos->{$url} ||
4475 Git::SVN::sanitize_remote_name($url);
4476
4477 my $fetch = $new_urls->{$url};
4478 foreach my $path (keys %$fetch) {
4479 my $x = $fetch->{$path};
4480 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4481 my $pfx = "svn-remote.$x->{old_repo_id}";
4482
4483 my $old_fetch = quotemeta("$x->{old_path}:".
4484 "refs/remotes/$x->{ref_id}");
4485 command_noisy(qw/config --unset/,
4486 "$pfx.fetch", '^'. $old_fetch . '$');
4487 delete $r->{$x->{old_repo_id}}->
4488 {fetch}->{$x->{old_path}};
4489 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4490 command_noisy(qw/config --unset/,
4491 "$pfx.url");
4492 push @emptied, $x->{old_repo_id}
4493 }
4494 }
4495 }
4496 if (@emptied) {
4497 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4498 "$ENV{GIT_DIR}/config";
4499 print STDERR <<EOF;
4500 The following [svn-remote] sections in your config file ($file) are empty
4501 and can be safely removed:
4502 EOF
4503 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4504 }
4505 }
4506
4507 sub migration_check {
4508 migrate_from_v0();
4509 migrate_from_v1();
4510 migrate_from_v2();
4511 minimize_connections() if $_minimize;
4512 }
4513
4514 package Git::IndexInfo;
4515 use strict;
4516 use warnings;
4517 use Git qw/command_input_pipe command_close_pipe/;
4518
4519 sub new {
4520 my ($class) = @_;
4521 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4522 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4523 }
4524
4525 sub remove {
4526 my ($self, $path) = @_;
4527 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4528 return ++$self->{nr};
4529 }
4530 undef;
4531 }
4532
4533 sub update {
4534 my ($self, $mode, $hash, $path) = @_;
4535 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4536 return ++$self->{nr};
4537 }
4538 undef;
4539 }
4540
4541 sub DESTROY {
4542 my ($self) = @_;
4543 command_close_pipe($self->{gui}, $self->{ctx});
4544 }
4545
4546 package Git::SVN::GlobSpec;
4547 use strict;
4548 use warnings;
4549
4550 sub new {
4551 my ($class, $glob) = @_;
4552 my $re = $glob;
4553 $re =~ s!/+$!!g; # no need for trailing slashes
4554 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
4555 my ($left, $right) = ($1, $2);
4556 if ($nr > 1) {
4557 die "Only one '*' wildcard expansion ",
4558 "is supported (got $nr): '$glob'\n";
4559 } elsif ($nr == 0) {
4560 die "One '*' is needed for glob: '$glob'\n";
4561 }
4562 $re = quotemeta($left) . $re . quotemeta($right);
4563 if (length $left && !($left =~ s!/+$!!g)) {
4564 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4565 }
4566 if (length $right && !($right =~ s!^/+!!g)) {
4567 die "Missing leading '/' on right side of: '$glob' ($right)\n";
4568 }
4569 my $left_re = qr/^\/\Q$left\E(\/|$)/;
4570 bless { left => $left, right => $right, left_regex => $left_re,
4571 regex => qr/$re/, glob => $glob }, $class;
4572 }
4573
4574 sub full_path {
4575 my ($self, $path) = @_;
4576 return (length $self->{left} ? "$self->{left}/" : '') .
4577 $path . (length $self->{right} ? "/$self->{right}" : '');
4578 }
4579
4580 __END__
4581
4582 Data structures:
4583
4584
4585 $remotes = { # returned by read_all_remotes()
4586 'svn' => {
4587 # svn-remote.svn.url=https://svn.musicpd.org
4588 url => 'https://svn.musicpd.org',
4589 # svn-remote.svn.fetch=mpd/trunk:trunk
4590 fetch => {
4591 'mpd/trunk' => 'trunk',
4592 },
4593 # svn-remote.svn.tags=mpd/tags/*:tags/*
4594 tags => {
4595 path => {
4596 left => 'mpd/tags',
4597 right => '',
4598 regex => qr!mpd/tags/([^/]+)$!,
4599 glob => 'tags/*',
4600 },
4601 ref => {
4602 left => 'tags',
4603 right => '',
4604 regex => qr!tags/([^/]+)$!,
4605 glob => 'tags/*',
4606 },
4607 }
4608 }
4609 };
4610
4611 $log_entry hashref as returned by libsvn_log_entry()
4612 {
4613 log => 'whitespace-formatted log entry
4614 ', # trailing newline is preserved
4615 revision => '8', # integer
4616 date => '2004-02-24T17:01:44.108345Z', # commit date
4617 author => 'committer name'
4618 };
4619
4620
4621 # this is generated by generate_diff();
4622 @mods = array of diff-index line hashes, each element represents one line
4623 of diff-index output
4624
4625 diff-index line ($m hash)
4626 {
4627 mode_a => first column of diff-index output, no leading ':',
4628 mode_b => second column of diff-index output,
4629 sha1_b => sha1sum of the final blob,
4630 chg => change type [MCRADT],
4631 file_a => original file name of a file (iff chg is 'C' or 'R')
4632 file_b => new/current file name of a file (any chg)
4633 }
4634 ;
4635
4636 # retval of read_url_paths{,_all}();
4637 $l_map = {
4638 # repository root url
4639 'https://svn.musicpd.org' => {
4640 # repository path # GIT_SVN_ID
4641 'mpd/trunk' => 'trunk',
4642 'mpd/tags/0.11.5' => 'tags/0.11.5',
4643 },
4644 }
4645
4646 Notes:
4647 I don't trust the each() function on unless I created %hash myself
4648 because the internal iterator may not have started at base.
git with Jakub Narębski modifications