?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Update delta compression message to be less misleading
[git/jrn.git] / git-svn.perl
blobbc3ba064e4343fd84535e38ecfd7ac691a644248
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 $_repository
8 $_q $_authors %users/;
9 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10 $VERSION = '@@GIT_VERSION@@';
11
12 # From which subdir have we been invoked?
13 my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15 } || '';
16
17 my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
18 $ENV{GIT_DIR} ||= '.git';
19 $Git::SVN::default_repo_id = 'svn';
20 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
21 $Git::SVN::Ra::_log_window_size = 100;
22
23 $Git::SVN::Log::TZ = $ENV{TZ};
24 $ENV{TZ} = 'UTC';
25 $| = 1; # unbuffer STDOUT
26
27 sub fatal (@) { print STDERR "@_\n"; exit 1 }
28 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
29 require SVN::Ra;
30 require SVN::Delta;
31 if ($SVN::Core::VERSION lt '1.1.0') {
32 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
33 }
34 push @Git::SVN::Ra::ISA, 'SVN::Ra';
35 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
36 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
37 use Carp qw/croak/;
38 use Digest::MD5;
39 use IO::File qw//;
40 use File::Basename qw/dirname basename/;
41 use File::Path qw/mkpath/;
42 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
43 use IPC::Open3;
44 use Git;
45
46 BEGIN {
47 # import functions from Git into our packages, en masse
48 no strict 'refs';
49 foreach (qw/command command_oneline command_noisy command_output_pipe
50 command_input_pipe command_close_pipe
51 command_bidi_pipe command_close_bidi_pipe/) {
52 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
53 Git::SVN::Migration Git::SVN::Log Git::SVN),
54 __PACKAGE__) {
55 *{"${package}::$_"} = \&{"Git::$_"};
56 }
57 }
58 }
59
60 my ($SVN);
61
62 $sha1 = qr/[a-f\d]{40}/;
63 $sha1_short = qr/[a-f\d]{4,40}/;
64 my ($_stdin, $_help, $_edit,
65 $_message, $_file,
66 $_template, $_shared,
67 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
68 $_merge, $_strategy, $_dry_run, $_local,
69 $_prefix, $_no_checkout, $_url, $_verbose,
70 $_git_format, $_commit_url, $_tag);
71 $Git::SVN::_follow_parent = 1;
72 $_q ||= 0;
73 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
74 'config-dir=s' => \$Git::SVN::Ra::config_dir,
75 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
76 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
77 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
78 'authors-file|A=s' => \$_authors,
79 'repack:i' => \$Git::SVN::_repack,
80 'noMetadata' => \$Git::SVN::_no_metadata,
81 'useSvmProps' => \$Git::SVN::_use_svm_props,
82 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
83 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
84 'no-checkout' => \$_no_checkout,
85 'quiet|q+' => \$_q,
86 'repack-flags|repack-args|repack-opts=s' =>
87 \$Git::SVN::_repack_flags,
88 'use-log-author' => \$Git::SVN::_use_log_author,
89 'add-author-from' => \$Git::SVN::_add_author_from,
90 'localtime' => \$Git::SVN::_localtime,
91 %remote_opts );
92
93 my ($_trunk, $_tags, $_branches, $_stdlayout);
94 my %icv;
95 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
96 'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
97 'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
98 'stdlayout|s' => \$_stdlayout,
99 'minimize-url|m' => \$Git::SVN::_minimize_url,
100 'no-metadata' => sub { $icv{noMetadata} = 1 },
101 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
102 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
103 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
104 %remote_opts );
105 my %cmt_opts = ( 'edit|e' => \$_edit,
106 'rmdir' => \$SVN::Git::Editor::_rmdir,
107 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
108 'l=i' => \$SVN::Git::Editor::_rename_limit,
109 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
110 );
111
112 my %cmd = (
113 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
114 { 'revision|r=s' => \$_revision,
115 'fetch-all|all' => \$_fetch_all,
116 'parent|p' => \$_fetch_parent,
117 %fc_opts } ],
118 clone => [ \&cmd_clone, "Initialize and fetch revisions",
119 { 'revision|r=s' => \$_revision,
120 %fc_opts, %init_opts } ],
121 init => [ \&cmd_init, "Initialize a repo for tracking" .
122 " (requires URL argument)",
123 \%init_opts ],
124 'multi-init' => [ \&cmd_multi_init,
125 "Deprecated alias for ".
126 "'$0 init -T<trunk> -b<branches> -t<tags>'",
127 \%init_opts ],
128 dcommit => [ \&cmd_dcommit,
129 'Commit several diffs to merge with upstream',
130 { 'merge|m|M' => \$_merge,
131 'strategy|s=s' => \$_strategy,
132 'verbose|v' => \$_verbose,
133 'dry-run|n' => \$_dry_run,
134 'fetch-all|all' => \$_fetch_all,
135 'commit-url=s' => \$_commit_url,
136 'revision|r=i' => \$_revision,
137 'no-rebase' => \$_no_rebase,
138 %cmt_opts, %fc_opts } ],
139 branch => [ \&cmd_branch,
140 'Create a branch in the SVN repository',
141 { 'message|m=s' => \$_message,
142 'dry-run|n' => \$_dry_run,
143 'tag|t' => \$_tag } ],
144 tag => [ sub { $_tag = 1; cmd_branch(@_) },
145 'Create a tag in the SVN repository',
146 { 'message|m=s' => \$_message,
147 'dry-run|n' => \$_dry_run } ],
148 'set-tree' => [ \&cmd_set_tree,
149 "Set an SVN repository to a git tree-ish",
150 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
151 'create-ignore' => [ \&cmd_create_ignore,
152 'Create a .gitignore per svn:ignore',
153 { 'revision|r=i' => \$_revision
154 } ],
155 'propget' => [ \&cmd_propget,
156 'Print the value of a property on a file or directory',
157 { 'revision|r=i' => \$_revision } ],
158 'proplist' => [ \&cmd_proplist,
159 'List all properties of a file or directory',
160 { 'revision|r=i' => \$_revision } ],
161 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
162 { 'revision|r=i' => \$_revision
163 } ],
164 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
165 { 'revision|r=i' => \$_revision
166 } ],
167 'multi-fetch' => [ \&cmd_multi_fetch,
168 "Deprecated alias for $0 fetch --all",
169 { 'revision|r=s' => \$_revision, %fc_opts } ],
170 'migrate' => [ sub { },
171 # no-op, we automatically run this anyways,
172 'Migrate configuration/metadata/layout from
173 previous versions of git-svn',
174 { 'minimize' => \$Git::SVN::Migration::_minimize,
175 %remote_opts } ],
176 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
177 { 'limit=i' => \$Git::SVN::Log::limit,
178 'revision|r=s' => \$_revision,
179 'verbose|v' => \$Git::SVN::Log::verbose,
180 'incremental' => \$Git::SVN::Log::incremental,
181 'oneline' => \$Git::SVN::Log::oneline,
182 'show-commit' => \$Git::SVN::Log::show_commit,
183 'non-recursive' => \$Git::SVN::Log::non_recursive,
184 'authors-file|A=s' => \$_authors,
185 'color' => \$Git::SVN::Log::color,
186 'pager=s' => \$Git::SVN::Log::pager
187 } ],
188 'find-rev' => [ \&cmd_find_rev,
189 "Translate between SVN revision numbers and tree-ish",
190 {} ],
191 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
192 { 'merge|m|M' => \$_merge,
193 'verbose|v' => \$_verbose,
194 'strategy|s=s' => \$_strategy,
195 'local|l' => \$_local,
196 'fetch-all|all' => \$_fetch_all,
197 'dry-run|n' => \$_dry_run,
198 %fc_opts } ],
199 'commit-diff' => [ \&cmd_commit_diff,
200 'Commit a diff between two trees',
201 { 'message|m=s' => \$_message,
202 'file|F=s' => \$_file,
203 'revision|r=s' => \$_revision,
204 %cmt_opts } ],
205 'info' => [ \&cmd_info,
206 "Show info about the latest SVN revision
207 on the current branch",
208 { 'url' => \$_url, } ],
209 'blame' => [ \&Git::SVN::Log::cmd_blame,
210 "Show what revision and author last modified each line of a file",
211 { 'git-format' => \$_git_format } ],
212 );
213
214 my $cmd;
215 for (my $i = 0; $i < @ARGV; $i++) {
216 if (defined $cmd{$ARGV[$i]}) {
217 $cmd = $ARGV[$i];
218 splice @ARGV, $i, 1;
219 last;
220 }
221 };
222
223 # make sure we're always running at the top-level working directory
224 unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
225 unless (-d $ENV{GIT_DIR}) {
226 if ($git_dir_user_set) {
227 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
228 "but it is not a directory\n";
229 }
230 my $git_dir = delete $ENV{GIT_DIR};
231 my $cdup = undef;
232 git_cmd_try {
233 $cdup = command_oneline(qw/rev-parse --show-cdup/);
234 $git_dir = '.' unless ($cdup);
235 chomp $cdup if ($cdup);
236 $cdup = "." unless ($cdup && length $cdup);
237 } "Already at toplevel, but $git_dir not found\n";
238 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
239 unless (-d $git_dir) {
240 die "$git_dir still not found after going to ",
241 "'$cdup'\n";
242 }
243 $ENV{GIT_DIR} = $git_dir;
244 }
245 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
246 }
247
248 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
249
250 read_repo_config(\%opts);
251 if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
252 Getopt::Long::Configure('pass_through');
253 }
254 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
255 'minimize-connections' => \$Git::SVN::Migration::_minimize,
256 'id|i=s' => \$Git::SVN::default_ref_id,
257 'svn-remote|remote|R=s' => sub {
258 $Git::SVN::no_reuse_existing = 1;
259 $Git::SVN::default_repo_id = $_[1] });
260 exit 1 if (!$rv && $cmd && $cmd ne 'log');
261
262 usage(0) if $_help;
263 version() if $_version;
264 usage(1) unless defined $cmd;
265 load_authors() if $_authors;
266
267 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
268 Git::SVN::Migration::migration_check();
269 }
270 Git::SVN::init_vars();
271 eval {
272 Git::SVN::verify_remotes_sanity();
273 $cmd{$cmd}->[0]->(@ARGV);
274 };
275 fatal $@ if $@;
276 post_fetch_checkout();
277 exit 0;
278
279 ####################### primary functions ######################
280 sub usage {
281 my $exit = shift || 0;
282 my $fd = $exit ? \*STDERR : \*STDOUT;
283 print $fd <<"";
284 git-svn - bidirectional operations between a single Subversion tree and git
285 Usage: git svn <command> [options] [arguments]\n
286
287 print $fd "Available commands:\n" unless $cmd;
288
289 foreach (sort keys %cmd) {
290 next if $cmd && $cmd ne $_;
291 next if /^multi-/; # don't show deprecated commands
292 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
293 foreach (sort keys %{$cmd{$_}->[2]}) {
294 # mixed-case options are for .git/config only
295 next if /[A-Z]/ && /^[a-z]+$/i;
296 # prints out arguments as they should be passed:
297 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
298 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
299 "--$_" : "-$_" }
300 split /\|/,$_)," $x\n";
301 }
302 }
303 print $fd <<"";
304 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
305 arbitrary identifier if you're tracking multiple SVN branches/repositories in
306 one git repository and want to keep them separate. See git-svn(1) for more
307 information.
308
309 exit $exit;
310 }
311
312 sub version {
313 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
314 exit 0;
315 }
316
317 sub do_git_init_db {
318 unless (-d $ENV{GIT_DIR}) {
319 my @init_db = ('init');
320 push @init_db, "--template=$_template" if defined $_template;
321 if (defined $_shared) {
322 if ($_shared =~ /[a-z]/) {
323 push @init_db, "--shared=$_shared";
324 } else {
325 push @init_db, "--shared";
326 }
327 }
328 command_noisy(@init_db);
329 $_repository = Git->repository(Repository => ".git");
330 }
331 my $set;
332 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
333 foreach my $i (keys %icv) {
334 die "'$set' and '$i' cannot both be set\n" if $set;
335 next unless defined $icv{$i};
336 command_noisy('config', "$pfx.$i", $icv{$i});
337 $set = $i;
338 }
339 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
340 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
341 if defined $$ignore_regex;
342 }
343
344 sub init_subdir {
345 my $repo_path = shift or return;
346 mkpath([$repo_path]) unless -d $repo_path;
347 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
348 $ENV{GIT_DIR} = '.git';
349 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
350 }
351
352 sub cmd_clone {
353 my ($url, $path) = @_;
354 if (!defined $path &&
355 (defined $_trunk || defined $_branches || defined $_tags ||
356 defined $_stdlayout) &&
357 $url !~ m#^[a-z\+]+://#) {
358 $path = $url;
359 }
360 $path = basename($url) if !defined $path || !length $path;
361 cmd_init($url, $path);
362 Git::SVN::fetch_all($Git::SVN::default_repo_id);
363 }
364
365 sub cmd_init {
366 if (defined $_stdlayout) {
367 $_trunk = 'trunk' if (!defined $_trunk);
368 $_tags = 'tags' if (!defined $_tags);
369 $_branches = 'branches' if (!defined $_branches);
370 }
371 if (defined $_trunk || defined $_branches || defined $_tags) {
372 return cmd_multi_init(@_);
373 }
374 my $url = shift or die "SVN repository location required ",
375 "as a command-line argument\n";
376 init_subdir(@_);
377 do_git_init_db();
378
379 Git::SVN->init($url);
380 }
381
382 sub cmd_fetch {
383 if (grep /^\d+=./, @_) {
384 die "'<rev>=<commit>' fetch arguments are ",
385 "no longer supported.\n";
386 }
387 my ($remote) = @_;
388 if (@_ > 1) {
389 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
390 }
391 if ($_fetch_parent) {
392 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
393 unless ($gs) {
394 die "Unable to determine upstream SVN information from ",
395 "working tree history\n";
396 }
397 # just fetch, don't checkout.
398 $_no_checkout = 'true';
399 $_fetch_all ? $gs->fetch_all : $gs->fetch;
400 } elsif ($_fetch_all) {
401 cmd_multi_fetch();
402 } else {
403 $remote ||= $Git::SVN::default_repo_id;
404 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
405 }
406 }
407
408 sub cmd_set_tree {
409 my (@commits) = @_;
410 if ($_stdin || !@commits) {
411 print "Reading from stdin...\n";
412 @commits = ();
413 while (<STDIN>) {
414 if (/\b($sha1_short)\b/o) {
415 unshift @commits, $1;
416 }
417 }
418 }
419 my @revs;
420 foreach my $c (@commits) {
421 my @tmp = command('rev-parse',$c);
422 if (scalar @tmp == 1) {
423 push @revs, $tmp[0];
424 } elsif (scalar @tmp > 1) {
425 push @revs, reverse(command('rev-list',@tmp));
426 } else {
427 fatal "Failed to rev-parse $c";
428 }
429 }
430 my $gs = Git::SVN->new;
431 my ($r_last, $cmt_last) = $gs->last_rev_commit;
432 $gs->fetch;
433 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
434 fatal "There are new revisions that were fetched ",
435 "and need to be merged (or acknowledged) ",
436 "before committing.\nlast rev: $r_last\n",
437 " current: $gs->{last_rev}";
438 }
439 $gs->set_tree($_) foreach @revs;
440 print "Done committing ",scalar @revs," revisions to SVN\n";
441 unlink $gs->{index};
442 }
443
444 sub cmd_dcommit {
445 my $head = shift;
446 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
447 'Cannot dcommit with a dirty index. Commit your changes first, '
448 . "or stash them with `git stash'.\n";
449 $head ||= 'HEAD';
450 my @refs;
451 my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
452 unless ($gs) {
453 die "Unable to determine upstream SVN information from ",
454 "$head history.\nPerhaps the repository is empty.";
455 }
456
457 if (defined $_commit_url) {
458 $url = $_commit_url;
459 } else {
460 $url = eval { command_oneline('config', '--get',
461 "svn-remote.$gs->{repo_id}.commiturl") };
462 if (!$url) {
463 $url = $gs->full_url
464 }
465 }
466
467 my $last_rev = $_revision if defined $_revision;
468 if ($url) {
469 print "Committing to $url ...\n";
470 }
471 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
472 if ($_no_rebase && scalar(@$linear_refs) > 1) {
473 warn "Attempting to commit more than one change while ",
474 "--no-rebase is enabled.\n",
475 "If these changes depend on each other, re-running ",
476 "without --no-rebase may be required."
477 }
478 my $expect_url = $url;
479 Git::SVN::remove_username($expect_url);
480 while (1) {
481 my $d = shift @$linear_refs or last;
482 unless (defined $last_rev) {
483 (undef, $last_rev, undef) = cmt_metadata("$d~1");
484 unless (defined $last_rev) {
485 fatal "Unable to extract revision information ",
486 "from commit $d~1";
487 }
488 }
489 if ($_dry_run) {
490 print "diff-tree $d~1 $d\n";
491 } else {
492 my $cmt_rev;
493 my %ed_opts = ( r => $last_rev,
494 log => get_commit_entry($d)->{log},
495 ra => Git::SVN::Ra->new($url),
496 config => SVN::Core::config_get_config(
497 $Git::SVN::Ra::config_dir
498 ),
499 tree_a => "$d~1",
500 tree_b => $d,
501 editor_cb => sub {
502 print "Committed r$_[0]\n";
503 $cmt_rev = $_[0];
504 },
505 svn_path => '');
506 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
507 print "No changes\n$d~1 == $d\n";
508 } elsif ($parents->{$d} && @{$parents->{$d}}) {
509 $gs->{inject_parents_dcommit}->{$cmt_rev} =
510 $parents->{$d};
511 }
512 $_fetch_all ? $gs->fetch_all : $gs->fetch;
513 $last_rev = $cmt_rev;
514 next if $_no_rebase;
515
516 # we always want to rebase against the current HEAD,
517 # not any head that was passed to us
518 my @diff = command('diff-tree', $d,
519 $gs->refname, '--');
520 my @finish;
521 if (@diff) {
522 @finish = rebase_cmd();
523 print STDERR "W: $d and ", $gs->refname,
524 " differ, using @finish:\n",
525 join("\n", @diff), "\n";
526 } else {
527 print "No changes between current HEAD and ",
528 $gs->refname,
529 "\nResetting to the latest ",
530 $gs->refname, "\n";
531 @finish = qw/reset --mixed/;
532 }
533 command_noisy(@finish, $gs->refname);
534 if (@diff) {
535 @refs = ();
536 my ($url_, $rev_, $uuid_, $gs_) =
537 working_head_info($head, \@refs);
538 my ($linear_refs_, $parents_) =
539 linearize_history($gs_, \@refs);
540 if (scalar(@$linear_refs) !=
541 scalar(@$linear_refs_)) {
542 fatal "# of revisions changed ",
543 "\nbefore:\n",
544 join("\n", @$linear_refs),
545 "\n\nafter:\n",
546 join("\n", @$linear_refs_), "\n",
547 'If you are attempting to commit ',
548 "merges, try running:\n\t",
549 'git rebase --interactive',
550 '--preserve-merges ',
551 $gs->refname,
552 "\nBefore dcommitting";
553 }
554 if ($url_ ne $expect_url) {
555 fatal "URL mismatch after rebase: ",
556 "$url_ != $expect_url";
557 }
558 if ($uuid_ ne $uuid) {
559 fatal "uuid mismatch after rebase: ",
560 "$uuid_ != $uuid";
561 }
562 # remap parents
563 my (%p, @l, $i);
564 for ($i = 0; $i < scalar @$linear_refs; $i++) {
565 my $new = $linear_refs_->[$i] or next;
566 $p{$new} =
567 $parents->{$linear_refs->[$i]};
568 push @l, $new;
569 }
570 $parents = \%p;
571 $linear_refs = \@l;
572 }
573 }
574 }
575 unlink $gs->{index};
576 }
577
578 sub cmd_branch {
579 my ($branch_name, $head) = @_;
580
581 unless (defined $branch_name && length $branch_name) {
582 die(($_tag ? "tag" : "branch") . " name required\n");
583 }
584 $head ||= 'HEAD';
585
586 my ($src, $rev, undef, $gs) = working_head_info($head);
587
588 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
589 my $glob = $remote->{ $_tag ? 'tags' : 'branches' };
590 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
591 my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
592
593 my $ctx = SVN::Client->new(
594 auth => Git::SVN::Ra::_auth_providers(),
595 log_msg => sub {
596 ${ $_[0] } = defined $_message
597 ? $_message
598 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
599 . $branch_name;
600 },
601 );
602
603 eval {
604 $ctx->ls($dst, 'HEAD', 0);
605 } and die "branch ${branch_name} already exists\n";
606
607 print "Copying ${src} at r${rev} to ${dst}...\n";
608 $ctx->copy($src, $rev, $dst)
609 unless $_dry_run;
610
611 $gs->fetch_all;
612 }
613
614 sub cmd_find_rev {
615 my $revision_or_hash = shift or die "SVN or git revision required ",
616 "as a command-line argument\n";
617 my $result;
618 if ($revision_or_hash =~ /^r\d+$/) {
619 my $head = shift;
620 $head ||= 'HEAD';
621 my @refs;
622 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
623 unless ($gs) {
624 die "Unable to determine upstream SVN information from ",
625 "$head history\n";
626 }
627 my $desired_revision = substr($revision_or_hash, 1);
628 $result = $gs->rev_map_get($desired_revision, $uuid);
629 } else {
630 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
631 $result = $rev;
632 }
633 print "$result\n" if $result;
634 }
635
636 sub cmd_rebase {
637 command_noisy(qw/update-index --refresh/);
638 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
639 unless ($gs) {
640 die "Unable to determine upstream SVN information from ",
641 "working tree history\n";
642 }
643 if ($_dry_run) {
644 print "Remote Branch: " . $gs->refname . "\n";
645 print "SVN URL: " . $url . "\n";
646 return;
647 }
648 if (command(qw/diff-index HEAD --/)) {
649 print STDERR "Cannot rebase with uncommited changes:\n";
650 command_noisy('status');
651 exit 1;
652 }
653 unless ($_local) {
654 # rebase will checkout for us, so no need to do it explicitly
655 $_no_checkout = 'true';
656 $_fetch_all ? $gs->fetch_all : $gs->fetch;
657 }
658 command_noisy(rebase_cmd(), $gs->refname);
659 }
660
661 sub cmd_show_ignore {
662 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
663 $gs ||= Git::SVN->new;
664 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
665 $gs->prop_walk($gs->{path}, $r, sub {
666 my ($gs, $path, $props) = @_;
667 print STDOUT "\n# $path\n";
668 my $s = $props->{'svn:ignore'} or return;
669 $s =~ s/[\r\n]+/\n/g;
670 chomp $s;
671 $s =~ s#^#$path#gm;
672 print STDOUT "$s\n";
673 });
674 }
675
676 sub cmd_show_externals {
677 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
678 $gs ||= Git::SVN->new;
679 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
680 $gs->prop_walk($gs->{path}, $r, sub {
681 my ($gs, $path, $props) = @_;
682 print STDOUT "\n# $path\n";
683 my $s = $props->{'svn:externals'} or return;
684 $s =~ s/[\r\n]+/\n/g;
685 chomp $s;
686 $s =~ s#^#$path#gm;
687 print STDOUT "$s\n";
688 });
689 }
690
691 sub cmd_create_ignore {
692 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
693 $gs ||= Git::SVN->new;
694 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
695 $gs->prop_walk($gs->{path}, $r, sub {
696 my ($gs, $path, $props) = @_;
697 # $path is of the form /path/to/dir/
698 $path = '.' . $path;
699 # SVN can have attributes on empty directories,
700 # which git won't track
701 mkpath([$path]) unless -d $path;
702 my $ignore = $path . '.gitignore';
703 my $s = $props->{'svn:ignore'} or return;
704 open(GITIGNORE, '>', $ignore)
705 or fatal("Failed to open `$ignore' for writing: $!");
706 $s =~ s/[\r\n]+/\n/g;
707 chomp $s;
708 # Prefix all patterns so that the ignore doesn't apply
709 # to sub-directories.
710 $s =~ s#^#/#gm;
711 print GITIGNORE "$s\n";
712 close(GITIGNORE)
713 or fatal("Failed to close `$ignore': $!");
714 command_noisy('add', '-f', $ignore);
715 });
716 }
717
718 sub canonicalize_path {
719 my ($path) = @_;
720 my $dot_slash_added = 0;
721 if (substr($path, 0, 1) ne "/") {
722 $path = "./" . $path;
723 $dot_slash_added = 1;
724 }
725 # File::Spec->canonpath doesn't collapse x/../y into y (for a
726 # good reason), so let's do this manually.
727 $path =~ s#/+#/#g;
728 $path =~ s#/\.(?:/|$)#/#g;
729 $path =~ s#/[^/]+/\.\.##g;
730 $path =~ s#/$##g;
731 $path =~ s#^\./## if $dot_slash_added;
732 $path =~ s#^/##;
733 $path =~ s#^\.$##;
734 return $path;
735 }
736
737 # get_svnprops(PATH)
738 # ------------------
739 # Helper for cmd_propget and cmd_proplist below.
740 sub get_svnprops {
741 my $path = shift;
742 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
743 $gs ||= Git::SVN->new;
744
745 # prefix THE PATH by the sub-directory from which the user
746 # invoked us.
747 $path = $cmd_dir_prefix . $path;
748 fatal("No such file or directory: $path") unless -e $path;
749 my $is_dir = -d $path ? 1 : 0;
750 $path = $gs->{path} . '/' . $path;
751
752 # canonicalize the path (otherwise libsvn will abort or fail to
753 # find the file)
754 $path = canonicalize_path($path);
755
756 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
757 my $props;
758 if ($is_dir) {
759 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
760 }
761 else {
762 (undef, $props) = $gs->ra->get_file($path, $r, undef);
763 }
764 return $props;
765 }
766
767 # cmd_propget (PROP, PATH)
768 # ------------------------
769 # Print the SVN property PROP for PATH.
770 sub cmd_propget {
771 my ($prop, $path) = @_;
772 $path = '.' if not defined $path;
773 usage(1) if not defined $prop;
774 my $props = get_svnprops($path);
775 if (not defined $props->{$prop}) {
776 fatal("`$path' does not have a `$prop' SVN property.");
777 }
778 print $props->{$prop} . "\n";
779 }
780
781 # cmd_proplist (PATH)
782 # -------------------
783 # Print the list of SVN properties for PATH.
784 sub cmd_proplist {
785 my $path = shift;
786 $path = '.' if not defined $path;
787 my $props = get_svnprops($path);
788 print "Properties on '$path':\n";
789 foreach (sort keys %{$props}) {
790 print " $_\n";
791 }
792 }
793
794 sub cmd_multi_init {
795 my $url = shift;
796 unless (defined $_trunk || defined $_branches || defined $_tags) {
797 usage(1);
798 }
799
800 # there are currently some bugs that prevent multi-init/multi-fetch
801 # setups from working well without this.
802 $Git::SVN::_minimize_url = 1;
803
804 $_prefix = '' unless defined $_prefix;
805 if (defined $url) {
806 $url =~ s#/+$##;
807 init_subdir(@_);
808 }
809 do_git_init_db();
810 if (defined $_trunk) {
811 my $trunk_ref = $_prefix . 'trunk';
812 # try both old-style and new-style lookups:
813 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
814 unless ($gs_trunk) {
815 my ($trunk_url, $trunk_path) =
816 complete_svn_url($url, $_trunk);
817 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
818 undef, $trunk_ref);
819 }
820 }
821 return unless defined $_branches || defined $_tags;
822 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
823 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
824 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
825 }
826
827 sub cmd_multi_fetch {
828 my $remotes = Git::SVN::read_all_remotes();
829 foreach my $repo_id (sort keys %$remotes) {
830 if ($remotes->{$repo_id}->{url}) {
831 Git::SVN::fetch_all($repo_id, $remotes);
832 }
833 }
834 }
835
836 # this command is special because it requires no metadata
837 sub cmd_commit_diff {
838 my ($ta, $tb, $url) = @_;
839 my $usage = "Usage: $0 commit-diff -r<revision> ".
840 "<tree-ish> <tree-ish> [<URL>]";
841 fatal($usage) if (!defined $ta || !defined $tb);
842 my $svn_path = '';
843 if (!defined $url) {
844 my $gs = eval { Git::SVN->new };
845 if (!$gs) {
846 fatal("Needed URL or usable git-svn --id in ",
847 "the command-line\n", $usage);
848 }
849 $url = $gs->{url};
850 $svn_path = $gs->{path};
851 }
852 unless (defined $_revision) {
853 fatal("-r|--revision is a required argument\n", $usage);
854 }
855 if (defined $_message && defined $_file) {
856 fatal("Both --message/-m and --file/-F specified ",
857 "for the commit message.\n",
858 "I have no idea what you mean");
859 }
860 if (defined $_file) {
861 $_message = file_to_s($_file);
862 } else {
863 $_message ||= get_commit_entry($tb)->{log};
864 }
865 my $ra ||= Git::SVN::Ra->new($url);
866 my $r = $_revision;
867 if ($r eq 'HEAD') {
868 $r = $ra->get_latest_revnum;
869 } elsif ($r !~ /^\d+$/) {
870 die "revision argument: $r not understood by git-svn\n";
871 }
872 my %ed_opts = ( r => $r,
873 log => $_message,
874 ra => $ra,
875 tree_a => $ta,
876 tree_b => $tb,
877 editor_cb => sub { print "Committed r$_[0]\n" },
878 svn_path => $svn_path );
879 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
880 print "No changes\n$ta == $tb\n";
881 }
882 }
883
884 sub escape_uri_only {
885 my ($uri) = @_;
886 my @tmp;
887 foreach (split m{/}, $uri) {
888 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
889 push @tmp, $_;
890 }
891 join('/', @tmp);
892 }
893
894 sub escape_url {
895 my ($url) = @_;
896 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
897 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
898 $url = "$scheme://$domain$uri";
899 }
900 $url;
901 }
902
903 sub cmd_info {
904 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
905 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
906 if (exists $_[1]) {
907 die "Too many arguments specified\n";
908 }
909
910 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
911
912 if (!$file_type && !$diff_status) {
913 print STDERR "svn: '$path' is not under version control\n";
914 exit 1;
915 }
916
917 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
918 unless ($gs) {
919 die "Unable to determine upstream SVN information from ",
920 "working tree history\n";
921 }
922
923 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
924 $path = "." if $path eq "";
925
926 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
927
928 if ($_url) {
929 print escape_url($full_url), "\n";
930 return;
931 }
932
933 my $result = "Path: $path\n";
934 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
935 $result .= "URL: " . escape_url($full_url) . "\n";
936
937 eval {
938 my $repos_root = $gs->repos_root;
939 Git::SVN::remove_username($repos_root);
940 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
941 };
942 if ($@) {
943 $result .= "Repository Root: (offline)\n";
944 }
945 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
946 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
947 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
948
949 $result .= "Node Kind: " .
950 ($file_type eq "dir" ? "directory" : "file") . "\n";
951
952 my $schedule = $diff_status eq "A"
953 ? "add"
954 : ($diff_status eq "D" ? "delete" : "normal");
955 $result .= "Schedule: $schedule\n";
956
957 if ($diff_status eq "A") {
958 print $result, "\n";
959 return;
960 }
961
962 my ($lc_author, $lc_rev, $lc_date_utc);
963 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
964 my $log = command_output_pipe(@args);
965 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
966 while (<$log>) {
967 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
968 $lc_author = $1;
969 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
970 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
971 (undef, $lc_rev, undef) = ::extract_metadata($1);
972 }
973 }
974 close $log;
975
976 Git::SVN::Log::set_local_timezone();
977
978 $result .= "Last Changed Author: $lc_author\n";
979 $result .= "Last Changed Rev: $lc_rev\n";
980 $result .= "Last Changed Date: " .
981 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
982
983 if ($file_type ne "dir") {
984 my $text_last_updated_date =
985 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
986 $result .=
987 "Text Last Updated: " .
988 Git::SVN::Log::format_svn_date($text_last_updated_date) .
989 "\n";
990 my $checksum;
991 if ($diff_status eq "D") {
992 my ($fh, $ctx) =
993 command_output_pipe(qw(cat-file blob), "HEAD:$path");
994 if ($file_type eq "link") {
995 my $file_name = <$fh>;
996 $checksum = md5sum("link $file_name");
997 } else {
998 $checksum = md5sum($fh);
999 }
1000 command_close_pipe($fh, $ctx);
1001 } elsif ($file_type eq "link") {
1002 my $file_name =
1003 command(qw(cat-file blob), "HEAD:$path");
1004 $checksum =
1005 md5sum("link " . $file_name);
1006 } else {
1007 open FILE, "<", $path or die $!;
1008 $checksum = md5sum(\*FILE);
1009 close FILE or die $!;
1010 }
1011 $result .= "Checksum: " . $checksum . "\n";
1012 }
1013
1014 print $result, "\n";
1015 }
1016
1017 ########################### utility functions #########################
1018
1019 sub rebase_cmd {
1020 my @cmd = qw/rebase/;
1021 push @cmd, '-v' if $_verbose;
1022 push @cmd, qw/--merge/ if $_merge;
1023 push @cmd, "--strategy=$_strategy" if $_strategy;
1024 @cmd;
1025 }
1026
1027 sub post_fetch_checkout {
1028 return if $_no_checkout;
1029 my $gs = $Git::SVN::_head or return;
1030 return if verify_ref('refs/heads/master^0');
1031
1032 my $valid_head = verify_ref('HEAD^0');
1033 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1034 return if ($valid_head || !verify_ref('HEAD^0'));
1035
1036 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1037 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1038 return if -f $index;
1039
1040 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1041 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1042 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1043 print STDERR "Checked out HEAD:\n ",
1044 $gs->full_url, " r", $gs->last_rev, "\n";
1045 }
1046
1047 sub complete_svn_url {
1048 my ($url, $path) = @_;
1049 $path =~ s#/+$##;
1050 if ($path !~ m#^[a-z\+]+://#) {
1051 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1052 fatal("E: '$path' is not a complete URL ",
1053 "and a separate URL is not specified");
1054 }
1055 return ($url, $path);
1056 }
1057 return ($path, '');
1058 }
1059
1060 sub complete_url_ls_init {
1061 my ($ra, $repo_path, $switch, $pfx) = @_;
1062 unless ($repo_path) {
1063 print STDERR "W: $switch not specified\n";
1064 return;
1065 }
1066 $repo_path =~ s#/+$##;
1067 if ($repo_path =~ m#^[a-z\+]+://#) {
1068 $ra = Git::SVN::Ra->new($repo_path);
1069 $repo_path = '';
1070 } else {
1071 $repo_path =~ s#^/+##;
1072 unless ($ra) {
1073 fatal("E: '$repo_path' is not a complete URL ",
1074 "and a separate URL is not specified");
1075 }
1076 }
1077 my $url = $ra->{url};
1078 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1079 my $k = "svn-remote.$gs->{repo_id}.url";
1080 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1081 if ($orig_url && ($orig_url ne $gs->{url})) {
1082 die "$k already set: $orig_url\n",
1083 "wanted to set to: $gs->{url}\n";
1084 }
1085 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1086 my $remote_path = "$ra->{svn_path}/$repo_path";
1087 $remote_path =~ s#/+#/#g;
1088 $remote_path =~ s#^/##g;
1089 $remote_path .= "/*" if $remote_path !~ /\*/;
1090 my ($n) = ($switch =~ /^--(\w+)/);
1091 if (length $pfx && $pfx !~ m#/$#) {
1092 die "--prefix='$pfx' must have a trailing slash '/'\n";
1093 }
1094 command_noisy('config',
1095 "svn-remote.$gs->{repo_id}.$n",
1096 "$remote_path:refs/remotes/$pfx*" .
1097 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1098 }
1099
1100 sub verify_ref {
1101 my ($ref) = @_;
1102 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1103 { STDERR => 0 }); };
1104 }
1105
1106 sub get_tree_from_treeish {
1107 my ($treeish) = @_;
1108 # $treeish can be a symbolic ref, too:
1109 my $type = command_oneline(qw/cat-file -t/, $treeish);
1110 my $expected;
1111 while ($type eq 'tag') {
1112 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1113 }
1114 if ($type eq 'commit') {
1115 $expected = (grep /^tree /, command(qw/cat-file commit/,
1116 $treeish))[0];
1117 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1118 die "Unable to get tree from $treeish\n" unless $expected;
1119 } elsif ($type eq 'tree') {
1120 $expected = $treeish;
1121 } else {
1122 die "$treeish is a $type, expected tree, tag or commit\n";
1123 }
1124 return $expected;
1125 }
1126
1127 sub get_commit_entry {
1128 my ($treeish) = shift;
1129 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1130 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1131 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1132 open my $log_fh, '>', $commit_editmsg or croak $!;
1133
1134 my $type = command_oneline(qw/cat-file -t/, $treeish);
1135 if ($type eq 'commit' || $type eq 'tag') {
1136 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1137 $type, $treeish);
1138 my $in_msg = 0;
1139 my $author;
1140 my $saw_from = 0;
1141 my $msgbuf = "";
1142 while (<$msg_fh>) {
1143 if (!$in_msg) {
1144 $in_msg = 1 if (/^\s*$/);
1145 $author = $1 if (/^author (.*>)/);
1146 } elsif (/^git-svn-id: /) {
1147 # skip this for now, we regenerate the
1148 # correct one on re-fetch anyways
1149 # TODO: set *:merge properties or like...
1150 } else {
1151 if (/^From:/ || /^Signed-off-by:/) {
1152 $saw_from = 1;
1153 }
1154 $msgbuf .= $_;
1155 }
1156 }
1157 $msgbuf =~ s/\s+$//s;
1158 if ($Git::SVN::_add_author_from && defined($author)
1159 && !$saw_from) {
1160 $msgbuf .= "\n\nFrom: $author";
1161 }
1162 print $log_fh $msgbuf or croak $!;
1163 command_close_pipe($msg_fh, $ctx);
1164 }
1165 close $log_fh or croak $!;
1166
1167 if ($_edit || ($type eq 'tree')) {
1168 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1169 # TODO: strip out spaces, comments, like git-commit.sh
1170 system($editor, $commit_editmsg);
1171 }
1172 rename $commit_editmsg, $commit_msg or croak $!;
1173 {
1174 # SVN requires messages to be UTF-8 when entering the repo
1175 local $/;
1176 open $log_fh, '<', $commit_msg or croak $!;
1177 binmode $log_fh;
1178 chomp($log_entry{log} = <$log_fh>);
1179
1180 if (my $enc = Git::config('i18n.commitencoding')) {
1181 require Encode;
1182 Encode::from_to($log_entry{log}, $enc, 'UTF-8');
1183 }
1184 close $log_fh or croak $!;
1185 }
1186 unlink $commit_msg;
1187 \%log_entry;
1188 }
1189
1190 sub s_to_file {
1191 my ($str, $file, $mode) = @_;
1192 open my $fd,'>',$file or croak $!;
1193 print $fd $str,"\n" or croak $!;
1194 close $fd or croak $!;
1195 chmod ($mode &~ umask, $file) if (defined $mode);
1196 }
1197
1198 sub file_to_s {
1199 my $file = shift;
1200 open my $fd,'<',$file or croak "$!: file: $file\n";
1201 local $/;
1202 my $ret = <$fd>;
1203 close $fd or croak $!;
1204 $ret =~ s/\s*$//s;
1205 return $ret;
1206 }
1207
1208 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1209 sub load_authors {
1210 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1211 my $log = $cmd eq 'log';
1212 while (<$authors>) {
1213 chomp;
1214 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1215 my ($user, $name, $email) = ($1, $2, $3);
1216 if ($log) {
1217 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1218 } else {
1219 $users{$user} = [$name, $email];
1220 }
1221 }
1222 close $authors or croak $!;
1223 }
1224
1225 # convert GetOpt::Long specs for use by git-config
1226 sub read_repo_config {
1227 return unless -d $ENV{GIT_DIR};
1228 my $opts = shift;
1229 my @config_only;
1230 foreach my $o (keys %$opts) {
1231 # if we have mixedCase and a long option-only, then
1232 # it's a config-only variable that we don't need for
1233 # the command-line.
1234 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1235 my $v = $opts->{$o};
1236 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1237 $key =~ s/-//g;
1238 my $arg = 'git config';
1239 $arg .= ' --int' if ($o =~ /[:=]i$/);
1240 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1241 if (ref $v eq 'ARRAY') {
1242 chomp(my @tmp = `$arg --get-all svn.$key`);
1243 @$v = @tmp if @tmp;
1244 } else {
1245 chomp(my $tmp = `$arg --get svn.$key`);
1246 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1247 $$v = $tmp;
1248 }
1249 }
1250 }
1251 delete @$opts{@config_only} if @config_only;
1252 }
1253
1254 sub extract_metadata {
1255 my $id = shift or return (undef, undef, undef);
1256 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1257 \s([a-f\d\-]+)$/x);
1258 if (!defined $rev || !$uuid || !$url) {
1259 # some of the original repositories I made had
1260 # identifiers like this:
1261 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1262 }
1263 return ($url, $rev, $uuid);
1264 }
1265
1266 sub cmt_metadata {
1267 return extract_metadata((grep(/^git-svn-id: /,
1268 command(qw/cat-file commit/, shift)))[-1]);
1269 }
1270
1271 sub cmt_sha2rev_batch {
1272 my %s2r;
1273 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1274 my $list = shift;
1275
1276 foreach my $sha (@{$list}) {
1277 my $first = 1;
1278 my $size = 0;
1279 print $out $sha, "\n";
1280
1281 while (my $line = <$in>) {
1282 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1283 last;
1284 } elsif ($first &&
1285 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1286 $first = 0;
1287 $size = $1;
1288 next;
1289 } elsif ($line =~ /^(git-svn-id: )/) {
1290 my (undef, $rev, undef) =
1291 extract_metadata($line);
1292 $s2r{$sha} = $rev;
1293 }
1294
1295 $size -= length($line);
1296 last if ($size == 0);
1297 }
1298 }
1299
1300 command_close_bidi_pipe($pid, $in, $out, $ctx);
1301
1302 return \%s2r;
1303 }
1304
1305 sub working_head_info {
1306 my ($head, $refs) = @_;
1307 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1308 my ($fh, $ctx) = command_output_pipe(@args, $head);
1309 my $hash;
1310 my %max;
1311 while (<$fh>) {
1312 if ( m{^commit ($::sha1)$} ) {
1313 unshift @$refs, $hash if $hash and $refs;
1314 $hash = $1;
1315 next;
1316 }
1317 next unless s{^\s*(git-svn-id:)}{$1};
1318 my ($url, $rev, $uuid) = extract_metadata($_);
1319 if (defined $url && defined $rev) {
1320 next if $max{$url} and $max{$url} < $rev;
1321 if (my $gs = Git::SVN->find_by_url($url)) {
1322 my $c = $gs->rev_map_get($rev, $uuid);
1323 if ($c && $c eq $hash) {
1324 close $fh; # break the pipe
1325 return ($url, $rev, $uuid, $gs);
1326 } else {
1327 $max{$url} ||= $gs->rev_map_max;
1328 }
1329 }
1330 }
1331 }
1332 command_close_pipe($fh, $ctx);
1333 (undef, undef, undef, undef);
1334 }
1335
1336 sub read_commit_parents {
1337 my ($parents, $c) = @_;
1338 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1339 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1340 @{$parents->{$c}} = split(/ /, $p);
1341 }
1342
1343 sub linearize_history {
1344 my ($gs, $refs) = @_;
1345 my %parents;
1346 foreach my $c (@$refs) {
1347 read_commit_parents(\%parents, $c);
1348 }
1349
1350 my @linear_refs;
1351 my %skip = ();
1352 my $last_svn_commit = $gs->last_commit;
1353 foreach my $c (reverse @$refs) {
1354 next if $c eq $last_svn_commit;
1355 last if $skip{$c};
1356
1357 unshift @linear_refs, $c;
1358 $skip{$c} = 1;
1359
1360 # we only want the first parent to diff against for linear
1361 # history, we save the rest to inject when we finalize the
1362 # svn commit
1363 my $fp_a = verify_ref("$c~1");
1364 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1365 if (!$fp_a || !$fp_b) {
1366 die "Commit $c\n",
1367 "has no parent commit, and therefore ",
1368 "nothing to diff against.\n",
1369 "You should be working from a repository ",
1370 "originally created by git-svn\n";
1371 }
1372 if ($fp_a ne $fp_b) {
1373 die "$c~1 = $fp_a, however parsing commit $c ",
1374 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1375 }
1376
1377 foreach my $p (@{$parents{$c}}) {
1378 $skip{$p} = 1;
1379 }
1380 }
1381 (\@linear_refs, \%parents);
1382 }
1383
1384 sub find_file_type_and_diff_status {
1385 my ($path) = @_;
1386 return ('dir', '') if $path eq '';
1387
1388 my $diff_output =
1389 command_oneline(qw(diff --cached --name-status --), $path) || "";
1390 my $diff_status = (split(' ', $diff_output))[0] || "";
1391
1392 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1393
1394 return (undef, undef) if !$diff_status && !$ls_tree;
1395
1396 if ($diff_status eq "A") {
1397 return ("link", $diff_status) if -l $path;
1398 return ("dir", $diff_status) if -d $path;
1399 return ("file", $diff_status);
1400 }
1401
1402 my $mode = (split(' ', $ls_tree))[0] || "";
1403
1404 return ("link", $diff_status) if $mode eq "120000";
1405 return ("dir", $diff_status) if $mode eq "040000";
1406 return ("file", $diff_status);
1407 }
1408
1409 sub md5sum {
1410 my $arg = shift;
1411 my $ref = ref $arg;
1412 my $md5 = Digest::MD5->new();
1413 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1414 $md5->addfile($arg) or croak $!;
1415 } elsif ($ref eq 'SCALAR') {
1416 $md5->add($$arg) or croak $!;
1417 } elsif (!$ref) {
1418 $md5->add($arg) or croak $!;
1419 } else {
1420 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1421 }
1422 return $md5->hexdigest();
1423 }
1424
1425 package Git::SVN;
1426 use strict;
1427 use warnings;
1428 use Fcntl qw/:DEFAULT :seek/;
1429 use constant rev_map_fmt => 'NH40';
1430 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1431 $_repack $_repack_flags $_use_svm_props $_head
1432 $_use_svnsync_props $no_reuse_existing $_minimize_url
1433 $_use_log_author $_add_author_from $_localtime/;
1434 use Carp qw/croak/;
1435 use File::Path qw/mkpath/;
1436 use File::Copy qw/copy/;
1437 use IPC::Open3;
1438
1439 my ($_gc_nr, $_gc_period);
1440
1441 # properties that we do not log:
1442 my %SKIP_PROP;
1443 BEGIN {
1444 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1445 svn:special svn:executable
1446 svn:entry:committed-rev
1447 svn:entry:last-author
1448 svn:entry:uuid
1449 svn:entry:committed-date/;
1450
1451 # some options are read globally, but can be overridden locally
1452 # per [svn-remote "..."] section. Command-line options will *NOT*
1453 # override options set in an [svn-remote "..."] section
1454 no strict 'refs';
1455 for my $option (qw/follow_parent no_metadata use_svm_props
1456 use_svnsync_props/) {
1457 my $key = $option;
1458 $key =~ tr/_//d;
1459 my $prop = "-$option";
1460 *$option = sub {
1461 my ($self) = @_;
1462 return $self->{$prop} if exists $self->{$prop};
1463 my $k = "svn-remote.$self->{repo_id}.$key";
1464 eval { command_oneline(qw/config --get/, $k) };
1465 if ($@) {
1466 $self->{$prop} = ${"Git::SVN::_$option"};
1467 } else {
1468 my $v = command_oneline(qw/config --bool/,$k);
1469 $self->{$prop} = $v eq 'false' ? 0 : 1;
1470 }
1471 return $self->{$prop};
1472 }
1473 }
1474 }
1475
1476
1477 my (%LOCKFILES, %INDEX_FILES);
1478 END {
1479 unlink keys %LOCKFILES if %LOCKFILES;
1480 unlink keys %INDEX_FILES if %INDEX_FILES;
1481 }
1482
1483 sub resolve_local_globs {
1484 my ($url, $fetch, $glob_spec) = @_;
1485 return unless defined $glob_spec;
1486 my $ref = $glob_spec->{ref};
1487 my $path = $glob_spec->{path};
1488 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1489 next unless m#^refs/remotes/$ref->{regex}$#;
1490 my $p = $1;
1491 my $pathname = desanitize_refname($path->full_path($p));
1492 my $refname = desanitize_refname($ref->full_path($p));
1493 if (my $existing = $fetch->{$pathname}) {
1494 if ($existing ne $refname) {
1495 die "Refspec conflict:\n",
1496 "existing: refs/remotes/$existing\n",
1497 " globbed: refs/remotes/$refname\n";
1498 }
1499 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1500 $u =~ s!^\Q$url\E(/|$)!! or die
1501 "refs/remotes/$refname: '$url' not found in '$u'\n";
1502 if ($pathname ne $u) {
1503 warn "W: Refspec glob conflict ",
1504 "(ref: refs/remotes/$refname):\n",
1505 "expected path: $pathname\n",
1506 " real path: $u\n",
1507 "Continuing ahead with $u\n";
1508 next;
1509 }
1510 } else {
1511 $fetch->{$pathname} = $refname;
1512 }
1513 }
1514 }
1515
1516 sub parse_revision_argument {
1517 my ($base, $head) = @_;
1518 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1519 return ($base, $head);
1520 }
1521 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1522 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1523 return ($head, $head) if ($::_revision eq 'HEAD');
1524 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1525 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1526 die "revision argument: $::_revision not understood by git-svn\n";
1527 }
1528
1529 sub fetch_all {
1530 my ($repo_id, $remotes) = @_;
1531 if (ref $repo_id) {
1532 my $gs = $repo_id;
1533 $repo_id = undef;
1534 $repo_id = $gs->{repo_id};
1535 }
1536 $remotes ||= read_all_remotes();
1537 my $remote = $remotes->{$repo_id} or
1538 die "[svn-remote \"$repo_id\"] unknown\n";
1539 my $fetch = $remote->{fetch};
1540 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1541 my (@gs, @globs);
1542 my $ra = Git::SVN::Ra->new($url);
1543 my $uuid = $ra->get_uuid;
1544 my $head = $ra->get_latest_revnum;
1545 my $base = defined $fetch ? $head : 0;
1546
1547 # read the max revs for wildcard expansion (branches/*, tags/*)
1548 foreach my $t (qw/branches tags/) {
1549 defined $remote->{$t} or next;
1550 push @globs, $remote->{$t};
1551 my $max_rev = eval { tmp_config(qw/--int --get/,
1552 "svn-remote.$repo_id.${t}-maxRev") };
1553 if (defined $max_rev && ($max_rev < $base)) {
1554 $base = $max_rev;
1555 } elsif (!defined $max_rev) {
1556 $base = 0;
1557 }
1558 }
1559
1560 if ($fetch) {
1561 foreach my $p (sort keys %$fetch) {
1562 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1563 my $lr = $gs->rev_map_max;
1564 if (defined $lr) {
1565 $base = $lr if ($lr < $base);
1566 }
1567 push @gs, $gs;
1568 }
1569 }
1570
1571 ($base, $head) = parse_revision_argument($base, $head);
1572 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1573 }
1574
1575 sub read_all_remotes {
1576 my $r = {};
1577 my $use_svm_props = eval { command_oneline(qw/config --bool
1578 svn.useSvmProps/) };
1579 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1580 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1581 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1582 my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1583 die("svn-remote.$remote: remote ref '$_remote_ref' "
1584 . "must start with 'refs/remotes/'\n")
1585 unless $_remote_ref =~ m{^refs/remotes/(.+)};
1586 my $remote_ref = $1;
1587 $local_ref =~ s{^/}{};
1588 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1589 $r->{$remote}->{svm} = {} if $use_svm_props;
1590 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1591 $r->{$1}->{svm} = {};
1592 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1593 $r->{$1}->{url} = $2;
1594 } elsif (m!^(.+)\.(branches|tags)=
1595 (.*):refs/remotes/(.+)\s*$/!x) {
1596 my ($p, $g) = ($3, $4);
1597 my $rs = $r->{$1}->{$2} = {
1598 t => $2,
1599 remote => $1,
1600 path => Git::SVN::GlobSpec->new($p),
1601 ref => Git::SVN::GlobSpec->new($g) };
1602 if (length($rs->{ref}->{right}) != 0) {
1603 die "The '*' glob character must be the last ",
1604 "character of '$g'\n";
1605 }
1606 }
1607 }
1608
1609 map {
1610 if (defined $r->{$_}->{svm}) {
1611 my $svm;
1612 eval {
1613 my $section = "svn-remote.$_";
1614 $svm = {
1615 source => tmp_config('--get',
1616 "$section.svm-source"),
1617 replace => tmp_config('--get',
1618 "$section.svm-replace"),
1619 }
1620 };
1621 $r->{$_}->{svm} = $svm;
1622 }
1623 } keys %$r;
1624
1625 $r;
1626 }
1627
1628 sub init_vars {
1629 $_gc_nr = $_gc_period = 1000;
1630 if (defined $_repack || defined $_repack_flags) {
1631 warn "Repack options are obsolete; they have no effect.\n";
1632 }
1633 }
1634
1635 sub verify_remotes_sanity {
1636 return unless -d $ENV{GIT_DIR};
1637 my %seen;
1638 foreach (command(qw/config -l/)) {
1639 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1640 if ($seen{$1}) {
1641 die "Remote ref refs/remote/$1 is tracked by",
1642 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1643 "Please resolve this ambiguity in ",
1644 "your git configuration file before ",
1645 "continuing\n";
1646 }
1647 $seen{$1} = $_;
1648 }
1649 }
1650 }
1651
1652 sub find_existing_remote {
1653 my ($url, $remotes) = @_;
1654 return undef if $no_reuse_existing;
1655 my $existing;
1656 foreach my $repo_id (keys %$remotes) {
1657 my $u = $remotes->{$repo_id}->{url} or next;
1658 next if $u ne $url;
1659 $existing = $repo_id;
1660 last;
1661 }
1662 $existing;
1663 }
1664
1665 sub init_remote_config {
1666 my ($self, $url, $no_write) = @_;
1667 $url =~ s!/+$!!; # strip trailing slash
1668 my $r = read_all_remotes();
1669 my $existing = find_existing_remote($url, $r);
1670 if ($existing) {
1671 unless ($no_write) {
1672 print STDERR "Using existing ",
1673 "[svn-remote \"$existing\"]\n";
1674 }
1675 $self->{repo_id} = $existing;
1676 } elsif ($_minimize_url) {
1677 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1678 $existing = find_existing_remote($min_url, $r);
1679 if ($existing) {
1680 unless ($no_write) {
1681 print STDERR "Using existing ",
1682 "[svn-remote \"$existing\"]\n";
1683 }
1684 $self->{repo_id} = $existing;
1685 }
1686 if ($min_url ne $url) {
1687 unless ($no_write) {
1688 print STDERR "Using higher level of URL: ",
1689 "$url => $min_url\n";
1690 }
1691 my $old_path = $self->{path};
1692 $self->{path} = $url;
1693 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1694 if (length $old_path) {
1695 $self->{path} .= "/$old_path";
1696 }
1697 $url = $min_url;
1698 }
1699 }
1700 my $orig_url;
1701 if (!$existing) {
1702 # verify that we aren't overwriting anything:
1703 $orig_url = eval {
1704 command_oneline('config', '--get',
1705 "svn-remote.$self->{repo_id}.url")
1706 };
1707 if ($orig_url && ($orig_url ne $url)) {
1708 die "svn-remote.$self->{repo_id}.url already set: ",
1709 "$orig_url\nwanted to set to: $url\n";
1710 }
1711 }
1712 my ($xrepo_id, $xpath) = find_ref($self->refname);
1713 if (defined $xpath) {
1714 die "svn-remote.$xrepo_id.fetch already set to track ",
1715 "$xpath:refs/remotes/", $self->refname, "\n";
1716 }
1717 unless ($no_write) {
1718 command_noisy('config',
1719 "svn-remote.$self->{repo_id}.url", $url);
1720 $self->{path} =~ s{^/}{};
1721 command_noisy('config', '--add',
1722 "svn-remote.$self->{repo_id}.fetch",
1723 "$self->{path}:".$self->refname);
1724 }
1725 $self->{url} = $url;
1726 }
1727
1728 sub find_by_url { # repos_root and, path are optional
1729 my ($class, $full_url, $repos_root, $path) = @_;
1730
1731 return undef unless defined $full_url;
1732 remove_username($full_url);
1733 remove_username($repos_root) if defined $repos_root;
1734 my $remotes = read_all_remotes();
1735 if (defined $full_url && defined $repos_root && !defined $path) {
1736 $path = $full_url;
1737 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1738 }
1739 foreach my $repo_id (keys %$remotes) {
1740 my $u = $remotes->{$repo_id}->{url} or next;
1741 remove_username($u);
1742 next if defined $repos_root && $repos_root ne $u;
1743
1744 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1745 foreach (qw/branches tags/) {
1746 resolve_local_globs($u, $fetch,
1747 $remotes->{$repo_id}->{$_});
1748 }
1749 my $p = $path;
1750 my $rwr = rewrite_root({repo_id => $repo_id});
1751 my $svm = $remotes->{$repo_id}->{svm}
1752 if defined $remotes->{$repo_id}->{svm};
1753 unless (defined $p) {
1754 $p = $full_url;
1755 my $z = $u;
1756 my $prefix = '';
1757 if ($rwr) {
1758 $z = $rwr;
1759 remove_username($z);
1760 } elsif (defined $svm) {
1761 $z = $svm->{source};
1762 $prefix = $svm->{replace};
1763 $prefix =~ s#^\Q$u\E(?:/|$)##;
1764 $prefix =~ s#/$##;
1765 }
1766 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1767 }
1768 foreach my $f (keys %$fetch) {
1769 next if $f ne $p;
1770 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1771 }
1772 }
1773 undef;
1774 }
1775
1776 sub init {
1777 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1778 my $self = _new($class, $repo_id, $ref_id, $path);
1779 if (defined $url) {
1780 $self->init_remote_config($url, $no_write);
1781 }
1782 $self;
1783 }
1784
1785 sub find_ref {
1786 my ($ref_id) = @_;
1787 foreach (command(qw/config -l/)) {
1788 next unless m!^svn-remote\.(.+)\.fetch=
1789 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1790 my ($repo_id, $path, $ref) = ($1, $2, $3);
1791 if ($ref eq $ref_id) {
1792 $path = '' if ($path =~ m#^\./?#);
1793 return ($repo_id, $path);
1794 }
1795 }
1796 (undef, undef, undef);
1797 }
1798
1799 sub new {
1800 my ($class, $ref_id, $repo_id, $path) = @_;
1801 if (defined $ref_id && !defined $repo_id && !defined $path) {
1802 ($repo_id, $path) = find_ref($ref_id);
1803 if (!defined $repo_id) {
1804 die "Could not find a \"svn-remote.*.fetch\" key ",
1805 "in the repository configuration matching: ",
1806 "refs/remotes/$ref_id\n";
1807 }
1808 }
1809 my $self = _new($class, $repo_id, $ref_id, $path);
1810 if (!defined $self->{path} || !length $self->{path}) {
1811 my $fetch = command_oneline('config', '--get',
1812 "svn-remote.$repo_id.fetch",
1813 ":refs/remotes/$ref_id\$") or
1814 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1815 "\":refs/remotes/$ref_id\$\" in config\n";
1816 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1817 }
1818 $self->{url} = command_oneline('config', '--get',
1819 "svn-remote.$repo_id.url") or
1820 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1821 $self->rebuild;
1822 $self;
1823 }
1824
1825 sub refname {
1826 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1827
1828 # It cannot end with a slash /, we'll throw up on this because
1829 # SVN can't have directories with a slash in their name, either:
1830 if ($refname =~ m{/$}) {
1831 die "ref: '$refname' ends with a trailing slash, this is ",
1832 "not permitted by git nor Subversion\n";
1833 }
1834
1835 # It cannot have ASCII control character space, tilde ~, caret ^,
1836 # colon :, question-mark ?, asterisk *, space, or open bracket [
1837 # anywhere.
1838 #
1839 # Additionally, % must be escaped because it is used for escaping
1840 # and we want our escaped refname to be reversible
1841 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1842
1843 # no slash-separated component can begin with a dot .
1844 # /.* becomes /%2E*
1845 $refname =~ s{/\.}{/%2E}g;
1846
1847 # It cannot have two consecutive dots .. anywhere
1848 # .. becomes %2E%2E
1849 $refname =~ s{\.\.}{%2E%2E}g;
1850
1851 return $refname;
1852 }
1853
1854 sub desanitize_refname {
1855 my ($refname) = @_;
1856 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1857 return $refname;
1858 }
1859
1860 sub svm_uuid {
1861 my ($self) = @_;
1862 return $self->{svm}->{uuid} if $self->svm;
1863 $self->ra;
1864 unless ($self->{svm}) {
1865 die "SVM UUID not cached, and reading remotely failed\n";
1866 }
1867 $self->{svm}->{uuid};
1868 }
1869
1870 sub svm {
1871 my ($self) = @_;
1872 return $self->{svm} if $self->{svm};
1873 my $svm;
1874 # see if we have it in our config, first:
1875 eval {
1876 my $section = "svn-remote.$self->{repo_id}";
1877 $svm = {
1878 source => tmp_config('--get', "$section.svm-source"),
1879 uuid => tmp_config('--get', "$section.svm-uuid"),
1880 replace => tmp_config('--get', "$section.svm-replace"),
1881 }
1882 };
1883 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1884 $self->{svm} = $svm;
1885 }
1886 $self->{svm};
1887 }
1888
1889 sub _set_svm_vars {
1890 my ($self, $ra) = @_;
1891 return $ra if $self->svm;
1892
1893 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1894 "(svm:source, svm:uuid) ",
1895 "from the following URLs:\n" );
1896 sub read_svm_props {
1897 my ($self, $ra, $path, $r) = @_;
1898 my $props = ($ra->get_dir($path, $r))[2];
1899 my $src = $props->{'svm:source'};
1900 my $uuid = $props->{'svm:uuid'};
1901 return undef if (!$src || !$uuid);
1902
1903 chomp($src, $uuid);
1904
1905 $uuid =~ m{^[0-9a-f\-]{30,}$}
1906 or die "doesn't look right - svm:uuid is '$uuid'\n";
1907
1908 # the '!' is used to mark the repos_root!/relative/path
1909 $src =~ s{/?!/?}{/};
1910 $src =~ s{/+$}{}; # no trailing slashes please
1911 # username is of no interest
1912 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1913
1914 my $replace = $ra->{url};
1915 $replace .= "/$path" if length $path;
1916
1917 my $section = "svn-remote.$self->{repo_id}";
1918 tmp_config("$section.svm-source", $src);
1919 tmp_config("$section.svm-replace", $replace);
1920 tmp_config("$section.svm-uuid", $uuid);
1921 $self->{svm} = {
1922 source => $src,
1923 uuid => $uuid,
1924 replace => $replace
1925 };
1926 }
1927
1928 my $r = $ra->get_latest_revnum;
1929 my $path = $self->{path};
1930 my %tried;
1931 while (length $path) {
1932 unless ($tried{"$self->{url}/$path"}) {
1933 return $ra if $self->read_svm_props($ra, $path, $r);
1934 $tried{"$self->{url}/$path"} = 1;
1935 }
1936 $path =~ s#/?[^/]+$##;
1937 }
1938 die "Path: '$path' should be ''\n" if $path ne '';
1939 return $ra if $self->read_svm_props($ra, $path, $r);
1940 $tried{"$self->{url}/$path"} = 1;
1941
1942 if ($ra->{repos_root} eq $self->{url}) {
1943 die @err, (map { " $_\n" } keys %tried), "\n";
1944 }
1945
1946 # nope, make sure we're connected to the repository root:
1947 my $ok;
1948 my @tried_b;
1949 $path = $ra->{svn_path};
1950 $ra = Git::SVN::Ra->new($ra->{repos_root});
1951 while (length $path) {
1952 unless ($tried{"$ra->{url}/$path"}) {
1953 $ok = $self->read_svm_props($ra, $path, $r);
1954 last if $ok;
1955 $tried{"$ra->{url}/$path"} = 1;
1956 }
1957 $path =~ s#/?[^/]+$##;
1958 }
1959 die "Path: '$path' should be ''\n" if $path ne '';
1960 $ok ||= $self->read_svm_props($ra, $path, $r);
1961 $tried{"$ra->{url}/$path"} = 1;
1962 if (!$ok) {
1963 die @err, (map { " $_\n" } keys %tried), "\n";
1964 }
1965 Git::SVN::Ra->new($self->{url});
1966 }
1967
1968 sub svnsync {
1969 my ($self) = @_;
1970 return $self->{svnsync} if $self->{svnsync};
1971
1972 if ($self->no_metadata) {
1973 die "Can't have both 'noMetadata' and ",
1974 "'useSvnsyncProps' options set!\n";
1975 }
1976 if ($self->rewrite_root) {
1977 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1978 "options set!\n";
1979 }
1980
1981 my $svnsync;
1982 # see if we have it in our config, first:
1983 eval {
1984 my $section = "svn-remote.$self->{repo_id}";
1985
1986 my $url = tmp_config('--get', "$section.svnsync-url");
1987 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1988 die "doesn't look right - svn:sync-from-url is '$url'\n";
1989
1990 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
1991 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1992 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1993
1994 $svnsync = { url => $url, uuid => $uuid }
1995 };
1996 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1997 return $self->{svnsync} = $svnsync;
1998 }
1999
2000 my $err = "useSvnsyncProps set, but failed to read " .
2001 "svnsync property: svn:sync-from-";
2002 my $rp = $self->ra->rev_proplist(0);
2003
2004 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2005 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2006 die "doesn't look right - svn:sync-from-url is '$url'\n";
2007
2008 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2009 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
2010 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2011
2012 my $section = "svn-remote.$self->{repo_id}";
2013 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2014 tmp_config('--add', "$section.svnsync-url", $url);
2015 return $self->{svnsync} = { url => $url, uuid => $uuid };
2016 }
2017
2018 # this allows us to memoize our SVN::Ra UUID locally and avoid a
2019 # remote lookup (useful for 'git svn log').
2020 sub ra_uuid {
2021 my ($self) = @_;
2022 unless ($self->{ra_uuid}) {
2023 my $key = "svn-remote.$self->{repo_id}.uuid";
2024 my $uuid = eval { tmp_config('--get', $key) };
2025 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
2026 $self->{ra_uuid} = $uuid;
2027 } else {
2028 die "ra_uuid called without URL\n" unless $self->{url};
2029 $self->{ra_uuid} = $self->ra->get_uuid;
2030 tmp_config('--add', $key, $self->{ra_uuid});
2031 }
2032 }
2033 $self->{ra_uuid};
2034 }
2035
2036 sub _set_repos_root {
2037 my ($self, $repos_root) = @_;
2038 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2039 $repos_root ||= $self->ra->{repos_root};
2040 tmp_config($k, $repos_root);
2041 $repos_root;
2042 }
2043
2044 sub repos_root {
2045 my ($self) = @_;
2046 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2047 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2048 }
2049
2050 sub ra {
2051 my ($self) = shift;
2052 my $ra = Git::SVN::Ra->new($self->{url});
2053 $self->_set_repos_root($ra->{repos_root});
2054 if ($self->use_svm_props && !$self->{svm}) {
2055 if ($self->no_metadata) {
2056 die "Can't have both 'noMetadata' and ",
2057 "'useSvmProps' options set!\n";
2058 } elsif ($self->use_svnsync_props) {
2059 die "Can't have both 'useSvnsyncProps' and ",
2060 "'useSvmProps' options set!\n";
2061 }
2062 $ra = $self->_set_svm_vars($ra);
2063 $self->{-want_revprops} = 1;
2064 }
2065 $ra;
2066 }
2067
2068 sub rel_path {
2069 my ($self) = @_;
2070 my $repos_root = $self->ra->{repos_root};
2071 return $self->{path} if ($self->{url} eq $repos_root);
2072 my $url = $self->{url} .
2073 (length $self->{path} ? "/$self->{path}" : $self->{path});
2074 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
2075 $url;
2076 }
2077
2078 # prop_walk(PATH, REV, SUB)
2079 # -------------------------
2080 # Recursively traverse PATH at revision REV and invoke SUB for each
2081 # directory that contains a SVN property. SUB will be invoked as
2082 # follows: &SUB(gs, path, props); where `gs' is this instance of
2083 # Git::SVN, `path' the path to the directory where the properties
2084 # `props' were found. The `path' will be relative to point of checkout,
2085 # that is, if url://repo/trunk is the current Git branch, and that
2086 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
2087 # as `path' (note the trailing `/').
2088 sub prop_walk {
2089 my ($self, $path, $rev, $sub) = @_;
2090
2091 $path =~ s#^/##;
2092 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2093 $path =~ s#^/*#/#g;
2094 my $p = $path;
2095 # Strip the irrelevant part of the path.
2096 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2097 # Ensure the path is terminated by a `/'.
2098 $p =~ s#/*$#/#;
2099
2100 # The properties contain all the internal SVN stuff nobody
2101 # (usually) cares about.
2102 my $interesting_props = 0;
2103 foreach (keys %{$props}) {
2104 # If it doesn't start with `svn:', it must be a
2105 # user-defined property.
2106 ++$interesting_props and next if $_ !~ /^svn:/;
2107 # FIXME: Fragile, if SVN adds new public properties,
2108 # this needs to be updated.
2109 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2110 |eol-style|mime-type
2111 |externals|needs-lock)$/x;
2112 }
2113 &$sub($self, $p, $props) if $interesting_props;
2114
2115 foreach (sort keys %$dirent) {
2116 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2117 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2118 }
2119 }
2120
2121 sub last_rev { ($_[0]->last_rev_commit)[0] }
2122 sub last_commit { ($_[0]->last_rev_commit)[1] }
2123
2124 # returns the newest SVN revision number and newest commit SHA1
2125 sub last_rev_commit {
2126 my ($self) = @_;
2127 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2128 return ($self->{last_rev}, $self->{last_commit});
2129 }
2130 my $c = ::verify_ref($self->refname.'^0');
2131 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2132 my $rev = (::cmt_metadata($c))[1];
2133 if (defined $rev) {
2134 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2135 return ($rev, $c);
2136 }
2137 }
2138 my $map_path = $self->map_path;
2139 unless (-e $map_path) {
2140 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2141 return (undef, undef);
2142 }
2143 my ($rev, $commit) = $self->rev_map_max(1);
2144 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2145 return ($rev, $commit);
2146 }
2147
2148 sub get_fetch_range {
2149 my ($self, $min, $max) = @_;
2150 $max ||= $self->ra->get_latest_revnum;
2151 $min ||= $self->rev_map_max;
2152 (++$min, $max);
2153 }
2154
2155 sub tmp_config {
2156 my (@args) = @_;
2157 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2158 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2159 if (! -f $config && -f $old_def_config) {
2160 rename $old_def_config, $config or
2161 die "Failed rename $old_def_config => $config: $!\n";
2162 }
2163 my $old_config = $ENV{GIT_CONFIG};
2164 $ENV{GIT_CONFIG} = $config;
2165 $@ = undef;
2166 my @ret = eval {
2167 unless (-f $config) {
2168 mkfile($config);
2169 open my $fh, '>', $config or
2170 die "Can't open $config: $!\n";
2171 print $fh "; This file is used internally by ",
2172 "git-svn\n" or die
2173 "Couldn't write to $config: $!\n";
2174 print $fh "; You should not have to edit it\n" or
2175 die "Couldn't write to $config: $!\n";
2176 close $fh or die "Couldn't close $config: $!\n";
2177 }
2178 command('config', @args);
2179 };
2180 my $err = $@;
2181 if (defined $old_config) {
2182 $ENV{GIT_CONFIG} = $old_config;
2183 } else {
2184 delete $ENV{GIT_CONFIG};
2185 }
2186 die $err if $err;
2187 wantarray ? @ret : $ret[0];
2188 }
2189
2190 sub tmp_index_do {
2191 my ($self, $sub) = @_;
2192 my $old_index = $ENV{GIT_INDEX_FILE};
2193 $ENV{GIT_INDEX_FILE} = $self->{index};
2194 $@ = undef;
2195 my @ret = eval {
2196 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2197 mkpath([$dir]) unless -d $dir;
2198 &$sub;
2199 };
2200 my $err = $@;
2201 if (defined $old_index) {
2202 $ENV{GIT_INDEX_FILE} = $old_index;
2203 } else {
2204 delete $ENV{GIT_INDEX_FILE};
2205 }
2206 die $err if $err;
2207 wantarray ? @ret : $ret[0];
2208 }
2209
2210 sub assert_index_clean {
2211 my ($self, $treeish) = @_;
2212
2213 $self->tmp_index_do(sub {
2214 command_noisy('read-tree', $treeish) unless -e $self->{index};
2215 my $x = command_oneline('write-tree');
2216 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2217 /^tree ($::sha1)/mo);
2218 return if $y eq $x;
2219
2220 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2221 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2222 command_noisy('read-tree', $treeish);
2223 $x = command_oneline('write-tree');
2224 if ($y ne $x) {
2225 ::fatal "trees ($treeish) $y != $x\n",
2226 "Something is seriously wrong...";
2227 }
2228 });
2229 }
2230
2231 sub get_commit_parents {
2232 my ($self, $log_entry) = @_;
2233 my (%seen, @ret, @tmp);
2234 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2235 if (my $ip = $self->{inject_parents}) {
2236 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2237 push @tmp, $commit;
2238 }
2239 }
2240 if (my $cur = ::verify_ref($self->refname.'^0')) {
2241 push @tmp, $cur;
2242 }
2243 if (my $ipd = $self->{inject_parents_dcommit}) {
2244 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2245 push @tmp, @$commit;
2246 }
2247 }
2248 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2249 while (my $p = shift @tmp) {
2250 next if $seen{$p};
2251 $seen{$p} = 1;
2252 push @ret, $p;
2253 # MAXPARENT is defined to 16 in commit-tree.c:
2254 last if @ret >= 16;
2255 }
2256 if (@tmp) {
2257 die "r$log_entry->{revision}: No room for parents:\n\t",
2258 join("\n\t", @tmp), "\n";
2259 }
2260 @ret;
2261 }
2262
2263 sub rewrite_root {
2264 my ($self) = @_;
2265 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2266 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2267 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2268 if ($rwr) {
2269 $rwr =~ s#/+$##;
2270 if ($rwr !~ m#^[a-z\+]+://#) {
2271 die "$rwr is not a valid URL (key: $k)\n";
2272 }
2273 }
2274 $self->{-rewrite_root} = $rwr;
2275 }
2276
2277 sub metadata_url {
2278 my ($self) = @_;
2279 ($self->rewrite_root || $self->{url}) .
2280 (length $self->{path} ? '/' . $self->{path} : '');
2281 }
2282
2283 sub full_url {
2284 my ($self) = @_;
2285 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2286 }
2287
2288
2289 sub set_commit_header_env {
2290 my ($log_entry) = @_;
2291 my %env;
2292 foreach my $ned (qw/NAME EMAIL DATE/) {
2293 foreach my $ac (qw/AUTHOR COMMITTER/) {
2294 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2295 }
2296 }
2297
2298 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2299 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2300 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2301
2302 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2303 ? $log_entry->{commit_name}
2304 : $log_entry->{name};
2305 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2306 ? $log_entry->{commit_email}
2307 : $log_entry->{email};
2308 \%env;
2309 }
2310
2311 sub restore_commit_header_env {
2312 my ($env) = @_;
2313 foreach my $ned (qw/NAME EMAIL DATE/) {
2314 foreach my $ac (qw/AUTHOR COMMITTER/) {
2315 my $k = "GIT_${ac}_${ned}";
2316 if (defined $env->{$k}) {
2317 $ENV{$k} = $env->{$k};
2318 } else {
2319 delete $ENV{$k};
2320 }
2321 }
2322 }
2323 }
2324
2325 sub gc {
2326 command_noisy('gc', '--auto');
2327 };
2328
2329 sub do_git_commit {
2330 my ($self, $log_entry) = @_;
2331 my $lr = $self->last_rev;
2332 if (defined $lr && $lr >= $log_entry->{revision}) {
2333 die "Last fetched revision of ", $self->refname,
2334 " was r$lr, but we are about to fetch: ",
2335 "r$log_entry->{revision}!\n";
2336 }
2337 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2338 croak "$log_entry->{revision} = $c already exists! ",
2339 "Why are we refetching it?\n";
2340 }
2341 my $old_env = set_commit_header_env($log_entry);
2342 my $tree = $log_entry->{tree};
2343 if (!defined $tree) {
2344 $tree = $self->tmp_index_do(sub {
2345 command_oneline('write-tree') });
2346 }
2347 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2348
2349 my @exec = ('git', 'commit-tree', $tree);
2350 foreach ($self->get_commit_parents($log_entry)) {
2351 push @exec, '-p', $_;
2352 }
2353 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2354 or croak $!;
2355 binmode $msg_fh;
2356
2357 # we always get UTF-8 from SVN, but we may want our commits in
2358 # a different encoding.
2359 if (my $enc = Git::config('i18n.commitencoding')) {
2360 require Encode;
2361 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2362 }
2363 print $msg_fh $log_entry->{log} or croak $!;
2364 restore_commit_header_env($old_env);
2365 unless ($self->no_metadata) {
2366 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2367 or croak $!;
2368 }
2369 $msg_fh->flush == 0 or croak $!;
2370 close $msg_fh or croak $!;
2371 chomp(my $commit = do { local $/; <$out_fh> });
2372 close $out_fh or croak $!;
2373 waitpid $pid, 0;
2374 croak $? if $?;
2375 if ($commit !~ /^$::sha1$/o) {
2376 die "Failed to commit, invalid sha1: $commit\n";
2377 }
2378
2379 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2380
2381 $self->{last_rev} = $log_entry->{revision};
2382 $self->{last_commit} = $commit;
2383 print "r$log_entry->{revision}" unless $::_q > 1;
2384 if (defined $log_entry->{svm_revision}) {
2385 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2386 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2387 0, $self->svm_uuid);
2388 }
2389 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2390 if (--$_gc_nr == 0) {
2391 $_gc_nr = $_gc_period;
2392 gc();
2393 }
2394 return $commit;
2395 }
2396
2397 sub match_paths {
2398 my ($self, $paths, $r) = @_;
2399 return 1 if $self->{path} eq '';
2400 if (my $path = $paths->{"/$self->{path}"}) {
2401 return ($path->{action} eq 'D') ? 0 : 1;
2402 }
2403 my $repos_root = $self->ra->{repos_root};
2404 my $extended_path = $self->{url} . '/' . $self->{path};
2405 $extended_path =~ s#^\Q$repos_root\E(/|$)##;
2406 $self->{path_regex} ||= qr/^\/\Q$extended_path\E\//;
2407 if (grep /$self->{path_regex}/, keys %$paths) {
2408 return 1;
2409 }
2410 my $c = '';
2411 foreach (split m#/#, $self->{path}) {
2412 $c .= "/$_";
2413 next unless ($paths->{$c} &&
2414 ($paths->{$c}->{action} =~ /^[AR]$/));
2415 if ($self->ra->check_path($self->{path}, $r) ==
2416 $SVN::Node::dir) {
2417 return 1;
2418 }
2419 }
2420 return 0;
2421 }
2422
2423 sub find_parent_branch {
2424 my ($self, $paths, $rev) = @_;
2425 return undef unless $self->follow_parent;
2426 unless (defined $paths) {
2427 my $err_handler = $SVN::Error::handler;
2428 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2429 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2430 $paths =
2431 Git::SVN::Ra::dup_changed_paths($_[0]) });
2432 $SVN::Error::handler = $err_handler;
2433 }
2434 return undef unless defined $paths;
2435
2436 # look for a parent from another branch:
2437 my @b_path_components = split m#/#, $self->rel_path;
2438 my @a_path_components;
2439 my $i;
2440 while (@b_path_components) {
2441 $i = $paths->{'/'.join('/', @b_path_components)};
2442 last if $i && defined $i->{copyfrom_path};
2443 unshift(@a_path_components, pop(@b_path_components));
2444 }
2445 return undef unless defined $i && defined $i->{copyfrom_path};
2446 my $branch_from = $i->{copyfrom_path};
2447 if (@a_path_components) {
2448 print STDERR "branch_from: $branch_from => ";
2449 $branch_from .= '/'.join('/', @a_path_components);
2450 print STDERR $branch_from, "\n";
2451 }
2452 my $r = $i->{copyfrom_rev};
2453 my $repos_root = $self->ra->{repos_root};
2454 my $url = $self->ra->{url};
2455 my $new_url = $repos_root . $branch_from;
2456 print STDERR "Found possible branch point: ",
2457 "$new_url => ", $self->full_url, ", $r\n";
2458 $branch_from =~ s#^/##;
2459 my $gs = $self->other_gs($new_url, $url, $repos_root,
2460 $branch_from, $r, $self->{ref_id});
2461 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2462 {
2463 my ($base, $head);
2464 if (!defined $r0 || !defined $parent) {
2465 ($base, $head) = parse_revision_argument(0, $r);
2466 } else {
2467 if ($r0 < $r) {
2468 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2469 0, 1, sub { $base = $_[1] - 1 });
2470 }
2471 }
2472 if (defined $base && $base <= $r) {
2473 $gs->fetch($base, $r);
2474 }
2475 ($r0, $parent) = $gs->find_rev_before($r, 1);
2476 }
2477 if (defined $r0 && defined $parent) {
2478 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2479 my $ed;
2480 if ($self->ra->can_do_switch) {
2481 $self->assert_index_clean($parent);
2482 print STDERR "Following parent with do_switch\n";
2483 # do_switch works with svn/trunk >= r22312, but that
2484 # is not included with SVN 1.4.3 (the latest version
2485 # at the moment), so we can't rely on it
2486 $self->{last_rev} = $r0;
2487 $self->{last_commit} = $parent;
2488 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2489 $gs->ra->gs_do_switch($r0, $rev, $gs,
2490 $self->full_url, $ed)
2491 or die "SVN connection failed somewhere...\n";
2492 } elsif ($self->ra->trees_match($new_url, $r0,
2493 $self->full_url, $rev)) {
2494 print STDERR "Trees match:\n",
2495 " $new_url\@$r0\n",
2496 " ${\$self->full_url}\@$rev\n",
2497 "Following parent with no changes\n";
2498 $self->tmp_index_do(sub {
2499 command_noisy('read-tree', $parent);
2500 });
2501 $self->{last_commit} = $parent;
2502 } else {
2503 print STDERR "Following parent with do_update\n";
2504 $ed = SVN::Git::Fetcher->new($self);
2505 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2506 or die "SVN connection failed somewhere...\n";
2507 }
2508 print STDERR "Successfully followed parent\n";
2509 return $self->make_log_entry($rev, [$parent], $ed);
2510 }
2511 return undef;
2512 }
2513
2514 sub do_fetch {
2515 my ($self, $paths, $rev) = @_;
2516 my $ed;
2517 my ($last_rev, @parents);
2518 if (my $lc = $self->last_commit) {
2519 # we can have a branch that was deleted, then re-added
2520 # under the same name but copied from another path, in
2521 # which case we'll have multiple parents (we don't
2522 # want to break the original ref, nor lose copypath info):
2523 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2524 push @{$log_entry->{parents}}, $lc;
2525 return $log_entry;
2526 }
2527 $ed = SVN::Git::Fetcher->new($self);
2528 $last_rev = $self->{last_rev};
2529 $ed->{c} = $lc;
2530 @parents = ($lc);
2531 } else {
2532 $last_rev = $rev;
2533 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2534 return $log_entry;
2535 }
2536 $ed = SVN::Git::Fetcher->new($self);
2537 }
2538 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2539 die "SVN connection failed somewhere...\n";
2540 }
2541 $self->make_log_entry($rev, \@parents, $ed);
2542 }
2543
2544 sub get_untracked {
2545 my ($self, $ed) = @_;
2546 my @out;
2547 my $h = $ed->{empty};
2548 foreach (sort keys %$h) {
2549 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2550 push @out, " $act: " . uri_encode($_);
2551 warn "W: $act: $_\n";
2552 }
2553 foreach my $t (qw/dir_prop file_prop/) {
2554 $h = $ed->{$t} or next;
2555 foreach my $path (sort keys %$h) {
2556 my $ppath = $path eq '' ? '.' : $path;
2557 foreach my $prop (sort keys %{$h->{$path}}) {
2558 next if $SKIP_PROP{$prop};
2559 my $v = $h->{$path}->{$prop};
2560 my $t_ppath_prop = "$t: " .
2561 uri_encode($ppath) . ' ' .
2562 uri_encode($prop);
2563 if (defined $v) {
2564 push @out, " +$t_ppath_prop " .
2565 uri_encode($v);
2566 } else {
2567 push @out, " -$t_ppath_prop";
2568 }
2569 }
2570 }
2571 }
2572 foreach my $t (qw/absent_file absent_directory/) {
2573 $h = $ed->{$t} or next;
2574 foreach my $parent (sort keys %$h) {
2575 foreach my $path (sort @{$h->{$parent}}) {
2576 push @out, " $t: " .
2577 uri_encode("$parent/$path");
2578 warn "W: $t: $parent/$path ",
2579 "Insufficient permissions?\n";
2580 }
2581 }
2582 }
2583 \@out;
2584 }
2585
2586 # parse_svn_date(DATE)
2587 # --------------------
2588 # Given a date (in UTC) from Subversion, return a string in the format
2589 # "<TZ Offset> <local date/time>" that Git will use.
2590 #
2591 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
2592 # is true we'll convert it to the local timezone instead.
2593 sub parse_svn_date {
2594 my $date = shift || return '+0000 1970-01-01 00:00:00';
2595 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2596 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2597 croak "Unable to parse date: $date\n";
2598 my $parsed_date; # Set next.
2599
2600 if ($Git::SVN::_localtime) {
2601 # Translate the Subversion datetime to an epoch time.
2602 # Begin by switching ourselves to $date's timezone, UTC.
2603 my $old_env_TZ = $ENV{TZ};
2604 $ENV{TZ} = 'UTC';
2605
2606 my $epoch_in_UTC =
2607 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2608
2609 # Determine our local timezone (including DST) at the
2610 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2611 # value of TZ, if any, at the time we were run.
2612 if (defined $Git::SVN::Log::TZ) {
2613 $ENV{TZ} = $Git::SVN::Log::TZ;
2614 } else {
2615 delete $ENV{TZ};
2616 }
2617
2618 my $our_TZ =
2619 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2620
2621 # This converts $epoch_in_UTC into our local timezone.
2622 my ($sec, $min, $hour, $mday, $mon, $year,
2623 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2624
2625 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2626 $our_TZ, $year + 1900, $mon + 1,
2627 $mday, $hour, $min, $sec);
2628
2629 # Reset us to the timezone in effect when we entered
2630 # this routine.
2631 if (defined $old_env_TZ) {
2632 $ENV{TZ} = $old_env_TZ;
2633 } else {
2634 delete $ENV{TZ};
2635 }
2636 } else {
2637 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2638 }
2639
2640 return $parsed_date;
2641 }
2642
2643 sub other_gs {
2644 my ($self, $new_url, $url, $repos_root,
2645 $branch_from, $r, $old_ref_id) = @_;
2646 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2647 unless ($gs) {
2648 my $ref_id = $old_ref_id;
2649 $ref_id =~ s/\@\d+$//;
2650 $ref_id .= "\@$r";
2651 # just grow a tail if we're not unique enough :x
2652 $ref_id .= '-' while find_ref($ref_id);
2653 print STDERR "Initializing parent: $ref_id\n";
2654 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2655 if ($u =~ s#^\Q$url\E(/|$)##) {
2656 $p = $u;
2657 $u = $url;
2658 $repo_id = $self->{repo_id};
2659 }
2660 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2661 }
2662 $gs
2663 }
2664
2665 sub check_author {
2666 my ($author) = @_;
2667 if (!defined $author || length $author == 0) {
2668 $author = '(no author)';
2669 } elsif (defined $::_authors && ! defined $::users{$author}) {
2670 die "Author: $author not defined in $::_authors file\n";
2671 }
2672 $author;
2673 }
2674
2675 sub make_log_entry {
2676 my ($self, $rev, $parents, $ed) = @_;
2677 my $untracked = $self->get_untracked($ed);
2678
2679 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2680 print $un "r$rev\n" or croak $!;
2681 print $un $_, "\n" foreach @$untracked;
2682 my %log_entry = ( parents => $parents || [], revision => $rev,
2683 log => '');
2684
2685 my $headrev;
2686 my $logged = delete $self->{logged_rev_props};
2687 if (!$logged || $self->{-want_revprops}) {
2688 my $rp = $self->ra->rev_proplist($rev);
2689 foreach (sort keys %$rp) {
2690 my $v = $rp->{$_};
2691 if (/^svn:(author|date|log)$/) {
2692 $log_entry{$1} = $v;
2693 } elsif ($_ eq 'svm:headrev') {
2694 $headrev = $v;
2695 } else {
2696 print $un " rev_prop: ", uri_encode($_), ' ',
2697 uri_encode($v), "\n";
2698 }
2699 }
2700 } else {
2701 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2702 }
2703 close $un or croak $!;
2704
2705 $log_entry{date} = parse_svn_date($log_entry{date});
2706 $log_entry{log} .= "\n";
2707 my $author = $log_entry{author} = check_author($log_entry{author});
2708 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2709 : ($author, undef);
2710
2711 my ($commit_name, $commit_email) = ($name, $email);
2712 if ($_use_log_author) {
2713 my $name_field;
2714 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2715 $name_field = $1;
2716 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2717 $name_field = $1;
2718 }
2719 if (!defined $name_field) {
2720 if (!defined $email) {
2721 $email = $name;
2722 }
2723 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2724 ($name, $email) = ($1, $2);
2725 } elsif ($name_field =~ /(.*)@/) {
2726 ($name, $email) = ($1, $name_field);
2727 } else {
2728 ($name, $email) = ($name_field, $name_field);
2729 }
2730 }
2731 if (defined $headrev && $self->use_svm_props) {
2732 if ($self->rewrite_root) {
2733 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2734 "options set!\n";
2735 }
2736 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2737 # we don't want "SVM: initializing mirror for junk" ...
2738 return undef if $r == 0;
2739 my $svm = $self->svm;
2740 if ($uuid ne $svm->{uuid}) {
2741 die "UUID mismatch on SVM path:\n",
2742 "expected: $svm->{uuid}\n",
2743 " got: $uuid\n";
2744 }
2745 my $full_url = $self->full_url;
2746 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2747 die "Failed to replace '$svm->{replace}' with ",
2748 "'$svm->{source}' in $full_url\n";
2749 # throw away username for storing in records
2750 remove_username($full_url);
2751 $log_entry{metadata} = "$full_url\@$r $uuid";
2752 $log_entry{svm_revision} = $r;
2753 $email ||= "$author\@$uuid";
2754 $commit_email ||= "$author\@$uuid";
2755 } elsif ($self->use_svnsync_props) {
2756 my $full_url = $self->svnsync->{url};
2757 $full_url .= "/$self->{path}" if length $self->{path};
2758 remove_username($full_url);
2759 my $uuid = $self->svnsync->{uuid};
2760 $log_entry{metadata} = "$full_url\@$rev $uuid";
2761 $email ||= "$author\@$uuid";
2762 $commit_email ||= "$author\@$uuid";
2763 } else {
2764 my $url = $self->metadata_url;
2765 remove_username($url);
2766 $log_entry{metadata} = "$url\@$rev " .
2767 $self->ra->get_uuid;
2768 $email ||= "$author\@" . $self->ra->get_uuid;
2769 $commit_email ||= "$author\@" . $self->ra->get_uuid;
2770 }
2771 $log_entry{name} = $name;
2772 $log_entry{email} = $email;
2773 $log_entry{commit_name} = $commit_name;
2774 $log_entry{commit_email} = $commit_email;
2775 \%log_entry;
2776 }
2777
2778 sub fetch {
2779 my ($self, $min_rev, $max_rev, @parents) = @_;
2780 my ($last_rev, $last_commit) = $self->last_rev_commit;
2781 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2782 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2783 }
2784
2785 sub set_tree_cb {
2786 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2787 $self->{inject_parents} = { $rev => $tree };
2788 $self->fetch(undef, undef);
2789 }
2790
2791 sub set_tree {
2792 my ($self, $tree) = (shift, shift);
2793 my $log_entry = ::get_commit_entry($tree);
2794 unless ($self->{last_rev}) {
2795 ::fatal("Must have an existing revision to commit");
2796 }
2797 my %ed_opts = ( r => $self->{last_rev},
2798 log => $log_entry->{log},
2799 ra => $self->ra,
2800 tree_a => $self->{last_commit},
2801 tree_b => $tree,
2802 editor_cb => sub {
2803 $self->set_tree_cb($log_entry, $tree, @_) },
2804 svn_path => $self->{path} );
2805 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2806 print "No changes\nr$self->{last_rev} = $tree\n";
2807 }
2808 }
2809
2810 sub rebuild_from_rev_db {
2811 my ($self, $path) = @_;
2812 my $r = -1;
2813 open my $fh, '<', $path or croak "open: $!";
2814 binmode $fh or croak "binmode: $!";
2815 while (<$fh>) {
2816 length($_) == 41 or croak "inconsistent size in ($_) != 41";
2817 chomp($_);
2818 ++$r;
2819 next if $_ eq ('0' x 40);
2820 $self->rev_map_set($r, $_);
2821 print "r$r = $_\n";
2822 }
2823 close $fh or croak "close: $!";
2824 unlink $path or croak "unlink: $!";
2825 }
2826
2827 sub rebuild {
2828 my ($self) = @_;
2829 my $map_path = $self->map_path;
2830 my $partial = (-e $map_path && ! -z $map_path);
2831 return unless ::verify_ref($self->refname.'^0');
2832 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
2833 my $rev_db = $self->rev_db_path;
2834 $self->rebuild_from_rev_db($rev_db);
2835 if ($self->use_svm_props) {
2836 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2837 $self->rebuild_from_rev_db($svm_rev_db);
2838 }
2839 $self->unlink_rev_db_symlink;
2840 return;
2841 }
2842 print "Rebuilding $map_path ...\n" if (!$partial);
2843 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
2844 (undef, undef));
2845 my ($log, $ctx) =
2846 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2847 ($head ? "$head.." : "") . $self->refname,
2848 '--');
2849 my $metadata_url = $self->metadata_url;
2850 remove_username($metadata_url);
2851 my $svn_uuid = $self->ra_uuid;
2852 my $c;
2853 while (<$log>) {
2854 if ( m{^commit ($::sha1)$} ) {
2855 $c = $1;
2856 next;
2857 }
2858 next unless s{^\s*(git-svn-id:)}{$1};
2859 my ($url, $rev, $uuid) = ::extract_metadata($_);
2860 remove_username($url);
2861
2862 # ignore merges (from set-tree)
2863 next if (!defined $rev || !$uuid);
2864
2865 # if we merged or otherwise started elsewhere, this is
2866 # how we break out of it
2867 if (($uuid ne $svn_uuid) ||
2868 ($metadata_url && $url && ($url ne $metadata_url))) {
2869 next;
2870 }
2871 if ($partial && $head) {
2872 print "Partial-rebuilding $map_path ...\n";
2873 print "Currently at $base_rev = $head\n";
2874 $head = undef;
2875 }
2876
2877 $self->rev_map_set($rev, $c);
2878 print "r$rev = $c\n";
2879 }
2880 command_close_pipe($log, $ctx);
2881 print "Done rebuilding $map_path\n" if (!$partial || !$head);
2882 my $rev_db_path = $self->rev_db_path;
2883 if (-f $self->rev_db_path) {
2884 unlink $self->rev_db_path or croak "unlink: $!";
2885 }
2886 $self->unlink_rev_db_symlink;
2887 }
2888
2889 # rev_map:
2890 # Tie::File seems to be prone to offset errors if revisions get sparse,
2891 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
2892 # one of my favorite modules is out :< Next up would be one of the DBM
2893 # modules, but I'm not sure which is most portable...
2894 #
2895 # This is the replacement for the rev_db format, which was too big
2896 # and inefficient for large repositories with a lot of sparse history
2897 # (mainly tags)
2898 #
2899 # The format is this:
2900 # - 24 bytes for every record,
2901 # * 4 bytes for the integer representing an SVN revision number
2902 # * 20 bytes representing the sha1 of a git commit
2903 # - No empty padding records like the old format
2904 # (except the last record, which can be overwritten)
2905 # - new records are written append-only since SVN revision numbers
2906 # increase monotonically
2907 # - lookups on SVN revision number are done via a binary search
2908 # - Piping the file to xxd -c24 is a good way of dumping it for
2909 # viewing or editing (piped back through xxd -r), should the need
2910 # ever arise.
2911 # - The last record can be padding revision with an all-zero sha1
2912 # This is used to optimize fetch performance when using multiple
2913 # "fetch" directives in .git/config
2914 #
2915 # These files are disposable unless noMetadata or useSvmProps is set
2916
2917 sub _rev_map_set {
2918 my ($fh, $rev, $commit) = @_;
2919
2920 binmode $fh or croak "binmode: $!";
2921 my $size = (stat($fh))[7];
2922 ($size % 24) == 0 or croak "inconsistent size: $size";
2923
2924 my $wr_offset = 0;
2925 if ($size > 0) {
2926 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2927 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2928 $read == 24 or croak "read only $read bytes (!= 24)";
2929 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2930 if ($last_commit eq ('0' x40)) {
2931 if ($size >= 48) {
2932 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2933 $read = sysread($fh, $buf, 24) or
2934 croak "read: $!";
2935 $read == 24 or
2936 croak "read only $read bytes (!= 24)";
2937 ($last_rev, $last_commit) =
2938 unpack(rev_map_fmt, $buf);
2939 if ($last_commit eq ('0' x40)) {
2940 croak "inconsistent .rev_map\n";
2941 }
2942 }
2943 if ($last_rev >= $rev) {
2944 croak "last_rev is higher!: $last_rev >= $rev";
2945 }
2946 $wr_offset = -24;
2947 }
2948 }
2949 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2950 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2951 croak "write: $!";
2952 }
2953
2954 sub mkfile {
2955 my ($path) = @_;
2956 unless (-e $path) {
2957 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2958 mkpath([$dir]) unless -d $dir;
2959 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2960 close $fh or die "Couldn't close (create) $path: $!\n";
2961 }
2962 }
2963
2964 sub rev_map_set {
2965 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2966 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2967 my $db = $self->map_path($uuid);
2968 my $db_lock = "$db.lock";
2969 my $sig;
2970 if ($update_ref) {
2971 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2972 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2973 }
2974 mkfile($db);
2975
2976 $LOCKFILES{$db_lock} = 1;
2977 my $sync;
2978 # both of these options make our .rev_db file very, very important
2979 # and we can't afford to lose it because rebuild() won't work
2980 if ($self->use_svm_props || $self->no_metadata) {
2981 $sync = 1;
2982 copy($db, $db_lock) or die "rev_map_set(@_): ",
2983 "Failed to copy: ",
2984 "$db => $db_lock ($!)\n";
2985 } else {
2986 rename $db, $db_lock or die "rev_map_set(@_): ",
2987 "Failed to rename: ",
2988 "$db => $db_lock ($!)\n";
2989 }
2990
2991 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2992 or croak "Couldn't open $db_lock: $!\n";
2993 _rev_map_set($fh, $rev, $commit);
2994 if ($sync) {
2995 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2996 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2997 }
2998 close $fh or croak $!;
2999 if ($update_ref) {
3000 $_head = $self;
3001 command_noisy('update-ref', '-m', "r$rev",
3002 $self->refname, $commit);
3003 }
3004 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3005 "$db_lock => $db ($!)\n";
3006 delete $LOCKFILES{$db_lock};
3007 if ($update_ref) {
3008 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3009 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3010 kill $sig, $$ if defined $sig;
3011 }
3012 }
3013
3014 # If want_commit, this will return an array of (rev, commit) where
3015 # commit _must_ be a valid commit in the archive.
3016 # Otherwise, it'll return the max revision (whether or not the
3017 # commit is valid or just a 0x40 placeholder).
3018 sub rev_map_max {
3019 my ($self, $want_commit) = @_;
3020 $self->rebuild;
3021 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3022 $want_commit ? ($r, $c) : $r;
3023 }
3024
3025 sub rev_map_max_norebuild {
3026 my ($self, $want_commit) = @_;
3027 my $map_path = $self->map_path;
3028 stat $map_path or return $want_commit ? (0, undef) : 0;
3029 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3030 binmode $fh or croak "binmode: $!";
3031 my $size = (stat($fh))[7];
3032 ($size % 24) == 0 or croak "inconsistent size: $size";
3033
3034 if ($size == 0) {
3035 close $fh or croak "close: $!";
3036 return $want_commit ? (0, undef) : 0;
3037 }
3038
3039 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3040 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3041 my ($r, $c) = unpack(rev_map_fmt, $buf);
3042 if ($want_commit && $c eq ('0' x40)) {
3043 if ($size < 48) {
3044 return $want_commit ? (0, undef) : 0;
3045 }
3046 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3047 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3048 ($r, $c) = unpack(rev_map_fmt, $buf);
3049 if ($c eq ('0'x40)) {
3050 croak "Penultimate record is all-zeroes in $map_path";
3051 }
3052 }
3053 close $fh or croak "close: $!";
3054 $want_commit ? ($r, $c) : $r;
3055 }
3056
3057 sub rev_map_get {
3058 my ($self, $rev, $uuid) = @_;
3059 my $map_path = $self->map_path($uuid);
3060 return undef unless -e $map_path;
3061
3062 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3063 binmode $fh or croak "binmode: $!";
3064 my $size = (stat($fh))[7];
3065 ($size % 24) == 0 or croak "inconsistent size: $size";
3066
3067 if ($size == 0) {
3068 close $fh or croak "close: $fh";
3069 return undef;
3070 }
3071
3072 my ($l, $u) = (0, $size - 24);
3073 my ($r, $c, $buf);
3074
3075 while ($l <= $u) {
3076 my $i = int(($l/24 + $u/24) / 2) * 24;
3077 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3078 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3079 my ($r, $c) = unpack('NH40', $buf);
3080
3081 if ($r < $rev) {
3082 $l = $i + 24;
3083 } elsif ($r > $rev) {
3084 $u = $i - 24;
3085 } else { # $r == $rev
3086 close($fh) or croak "close: $!";
3087 return $c eq ('0' x 40) ? undef : $c;
3088 }
3089 }
3090 close($fh) or croak "close: $!";
3091 undef;
3092 }
3093
3094 # Finds the first svn revision that exists on (if $eq_ok is true) or
3095 # before $rev for the current branch. It will not search any lower
3096 # than $min_rev. Returns the git commit hash and svn revision number
3097 # if found, else (undef, undef).
3098 sub find_rev_before {
3099 my ($self, $rev, $eq_ok, $min_rev) = @_;
3100 --$rev unless $eq_ok;
3101 $min_rev ||= 1;
3102 while ($rev >= $min_rev) {
3103 if (my $c = $self->rev_map_get($rev)) {
3104 return ($rev, $c);
3105 }
3106 --$rev;
3107 }
3108 return (undef, undef);
3109 }
3110
3111 # Finds the first svn revision that exists on (if $eq_ok is true) or
3112 # after $rev for the current branch. It will not search any higher
3113 # than $max_rev. Returns the git commit hash and svn revision number
3114 # if found, else (undef, undef).
3115 sub find_rev_after {
3116 my ($self, $rev, $eq_ok, $max_rev) = @_;
3117 ++$rev unless $eq_ok;
3118 $max_rev ||= $self->rev_map_max;
3119 while ($rev <= $max_rev) {
3120 if (my $c = $self->rev_map_get($rev)) {
3121 return ($rev, $c);
3122 }
3123 ++$rev;
3124 }
3125 return (undef, undef);
3126 }
3127
3128 sub _new {
3129 my ($class, $repo_id, $ref_id, $path) = @_;
3130 unless (defined $repo_id && length $repo_id) {
3131 $repo_id = $Git::SVN::default_repo_id;
3132 }
3133 unless (defined $ref_id && length $ref_id) {
3134 $_[2] = $ref_id = $Git::SVN::default_ref_id;
3135 }
3136 $_[1] = $repo_id;
3137 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3138 $_[3] = $path = '' unless (defined $path);
3139 mkpath(["$ENV{GIT_DIR}/svn"]);
3140 bless {
3141 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3142 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3143 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3144 }
3145
3146 # for read-only access of old .rev_db formats
3147 sub unlink_rev_db_symlink {
3148 my ($self) = @_;
3149 my $link = $self->rev_db_path;
3150 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3151 if (-l $link) {
3152 unlink $link or croak "unlink: $link failed!";
3153 }
3154 }
3155
3156 sub rev_db_path {
3157 my ($self, $uuid) = @_;
3158 my $db_path = $self->map_path($uuid);
3159 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3160 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3161 $db_path;
3162 }
3163
3164 # the new replacement for .rev_db
3165 sub map_path {
3166 my ($self, $uuid) = @_;
3167 $uuid ||= $self->ra_uuid;
3168 "$self->{map_root}.$uuid";
3169 }
3170
3171 sub uri_encode {
3172 my ($f) = @_;
3173 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3174 $f
3175 }
3176
3177 sub remove_username {
3178 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3179 }
3180
3181 package Git::SVN::Prompt;
3182 use strict;
3183 use warnings;
3184 require SVN::Core;
3185 use vars qw/$_no_auth_cache $_username/;
3186
3187 sub simple {
3188 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3189 $may_save = undef if $_no_auth_cache;
3190 $default_username = $_username if defined $_username;
3191 if (defined $default_username && length $default_username) {
3192 if (defined $realm && length $realm) {
3193 print STDERR "Authentication realm: $realm\n";
3194 STDERR->flush;
3195 }
3196 $cred->username($default_username);
3197 } else {
3198 username($cred, $realm, $may_save, $pool);
3199 }
3200 $cred->password(_read_password("Password for '" .
3201 $cred->username . "': ", $realm));
3202 $cred->may_save($may_save);
3203 $SVN::_Core::SVN_NO_ERROR;
3204 }
3205
3206 sub ssl_server_trust {
3207 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3208 $may_save = undef if $_no_auth_cache;
3209 print STDERR "Error validating server certificate for '$realm':\n";
3210 {
3211 no warnings 'once';
3212 # All variables SVN::Auth::SSL::* are used only once,
3213 # so we're shutting up Perl warnings about this.
3214 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3215 print STDERR " - The certificate is not issued ",
3216 "by a trusted authority. Use the\n",
3217 " fingerprint to validate ",
3218 "the certificate manually!\n";
3219 }
3220 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3221 print STDERR " - The certificate hostname ",
3222 "does not match.\n";
3223 }
3224 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3225 print STDERR " - The certificate is not yet valid.\n";
3226 }
3227 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3228 print STDERR " - The certificate has expired.\n";
3229 }
3230 if ($failures & $SVN::Auth::SSL::OTHER) {
3231 print STDERR " - The certificate has ",
3232 "an unknown error.\n";
3233 }
3234 } # no warnings 'once'
3235 printf STDERR
3236 "Certificate information:\n".
3237 " - Hostname: %s\n".
3238 " - Valid: from %s until %s\n".
3239 " - Issuer: %s\n".
3240 " - Fingerprint: %s\n",
3241 map $cert_info->$_, qw(hostname valid_from valid_until
3242 issuer_dname fingerprint);
3243 my $choice;
3244 prompt:
3245 print STDERR $may_save ?
3246 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3247 "(R)eject or accept (t)emporarily? ";
3248 STDERR->flush;
3249 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3250 if ($choice =~ /^t$/i) {
3251 $cred->may_save(undef);
3252 } elsif ($choice =~ /^r$/i) {
3253 return -1;
3254 } elsif ($may_save && $choice =~ /^p$/i) {
3255 $cred->may_save($may_save);
3256 } else {
3257 goto prompt;
3258 }
3259 $cred->accepted_failures($failures);
3260 $SVN::_Core::SVN_NO_ERROR;
3261 }
3262
3263 sub ssl_client_cert {
3264 my ($cred, $realm, $may_save, $pool) = @_;
3265 $may_save = undef if $_no_auth_cache;
3266 print STDERR "Client certificate filename: ";
3267 STDERR->flush;
3268 chomp(my $filename = <STDIN>);
3269 $cred->cert_file($filename);
3270 $cred->may_save($may_save);
3271 $SVN::_Core::SVN_NO_ERROR;
3272 }
3273
3274 sub ssl_client_cert_pw {
3275 my ($cred, $realm, $may_save, $pool) = @_;
3276 $may_save = undef if $_no_auth_cache;
3277 $cred->password(_read_password("Password: ", $realm));
3278 $cred->may_save($may_save);
3279 $SVN::_Core::SVN_NO_ERROR;
3280 }
3281
3282 sub username {
3283 my ($cred, $realm, $may_save, $pool) = @_;
3284 $may_save = undef if $_no_auth_cache;
3285 if (defined $realm && length $realm) {
3286 print STDERR "Authentication realm: $realm\n";
3287 }
3288 my $username;
3289 if (defined $_username) {
3290 $username = $_username;
3291 } else {
3292 print STDERR "Username: ";
3293 STDERR->flush;
3294 chomp($username = <STDIN>);
3295 }
3296 $cred->username($username);
3297 $cred->may_save($may_save);
3298 $SVN::_Core::SVN_NO_ERROR;
3299 }
3300
3301 sub _read_password {
3302 my ($prompt, $realm) = @_;
3303 print STDERR $prompt;
3304 STDERR->flush;
3305 require Term::ReadKey;
3306 Term::ReadKey::ReadMode('noecho');
3307 my $password = '';
3308 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3309 last if $key =~ /[\012\015]/; # \n\r
3310 $password .= $key;
3311 }
3312 Term::ReadKey::ReadMode('restore');
3313 print STDERR "\n";
3314 STDERR->flush;
3315 $password;
3316 }
3317
3318 package SVN::Git::Fetcher;
3319 use vars qw/@ISA/;
3320 use strict;
3321 use warnings;
3322 use Carp qw/croak/;
3323 use File::Temp qw/tempfile/;
3324 use IO::File qw//;
3325 use vars qw/$_ignore_regex/;
3326
3327 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
3328 sub new {
3329 my ($class, $git_svn, $switch_path) = @_;
3330 my $self = SVN::Delta::Editor->new;
3331 bless $self, $class;
3332 if (exists $git_svn->{last_commit}) {
3333 $self->{c} = $git_svn->{last_commit};
3334 $self->{empty_symlinks} =
3335 _mark_empty_symlinks($git_svn, $switch_path);
3336 }
3337 $self->{ignore_regex} = eval { command_oneline('config', '--get',
3338 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3339 $self->{empty} = {};
3340 $self->{dir_prop} = {};
3341 $self->{file_prop} = {};
3342 $self->{absent_dir} = {};
3343 $self->{absent_file} = {};
3344 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3345 $self;
3346 }
3347
3348 # this uses the Ra object, so it must be called before do_{switch,update},
3349 # not inside them (when the Git::SVN::Fetcher object is passed) to
3350 # do_{switch,update}
3351 sub _mark_empty_symlinks {
3352 my ($git_svn, $switch_path) = @_;
3353 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3354 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3355
3356 my %ret;
3357 my ($rev, $cmt) = $git_svn->last_rev_commit;
3358 return {} unless ($rev && $cmt);
3359
3360 # allow the warning to be printed for each revision we fetch to
3361 # ensure the user sees it. The user can also disable the workaround
3362 # on the repository even while git svn is running and the next
3363 # revision fetched will skip this expensive function.
3364 my $printed_warning;
3365 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3366 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3367 local $/ = "\0";
3368 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3369 $pfx .= '/' if length($pfx);
3370 while (<$ls>) {
3371 chomp;
3372 s/\A100644 blob $empty_blob\t//o or next;
3373 unless ($printed_warning) {
3374 print STDERR "Scanning for empty symlinks, ",
3375 "this may take a while if you have ",
3376 "many empty files\n",
3377 "You may disable this with `",
3378 "git config svn.brokenSymlinkWorkaround ",
3379 "false'.\n",
3380 "This may be done in a different ",
3381 "terminal without restarting ",
3382 "git svn\n";
3383 $printed_warning = 1;
3384 }
3385 my $path = $_;
3386 my (undef, $props) =
3387 $git_svn->ra->get_file($pfx.$path, $rev, undef);
3388 if ($props->{'svn:special'}) {
3389 $ret{$path} = 1;
3390 }
3391 }
3392 command_close_pipe($ls, $ctx);
3393 \%ret;
3394 }
3395
3396 # returns true if a given path is inside a ".git" directory
3397 sub in_dot_git {
3398 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3399 }
3400
3401 # return value: 0 -- don't ignore, 1 -- ignore
3402 sub is_path_ignored {
3403 my ($self, $path) = @_;
3404 return 1 if in_dot_git($path);
3405 return 1 if defined($self->{ignore_regex}) &&
3406 $path =~ m!$self->{ignore_regex}!;
3407 return 0 unless defined($_ignore_regex);
3408 return 1 if $path =~ m!$_ignore_regex!o;
3409 return 0;
3410 }
3411
3412 sub set_path_strip {
3413 my ($self, $path) = @_;
3414 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3415 }
3416
3417 sub open_root {
3418 { path => '' };
3419 }
3420
3421 sub open_directory {
3422 my ($self, $path, $pb, $rev) = @_;
3423 { path => $path };
3424 }
3425
3426 sub git_path {
3427 my ($self, $path) = @_;
3428 if ($self->{path_strip}) {
3429 $path =~ s!$self->{path_strip}!! or
3430 die "Failed to strip path '$path' ($self->{path_strip})\n";
3431 }
3432 $path;
3433 }
3434
3435 sub delete_entry {
3436 my ($self, $path, $rev, $pb) = @_;
3437 return undef if $self->is_path_ignored($path);
3438
3439 my $gpath = $self->git_path($path);
3440 return undef if ($gpath eq '');
3441
3442 # remove entire directories.
3443 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3444 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3445 if ($tree) {
3446 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3447 -r --name-only -z/,
3448 $tree);
3449 local $/ = "\0";
3450 while (<$ls>) {
3451 chomp;
3452 my $rmpath = "$gpath/$_";
3453 $self->{gii}->remove($rmpath);
3454 print "\tD\t$rmpath\n" unless $::_q;
3455 }
3456 print "\tD\t$gpath/\n" unless $::_q;
3457 command_close_pipe($ls, $ctx);
3458 $self->{empty}->{$path} = 0
3459 } else {
3460 $self->{gii}->remove($gpath);
3461 print "\tD\t$gpath\n" unless $::_q;
3462 }
3463 undef;
3464 }
3465
3466 sub open_file {
3467 my ($self, $path, $pb, $rev) = @_;
3468 my ($mode, $blob);
3469
3470 goto out if $self->is_path_ignored($path);
3471
3472 my $gpath = $self->git_path($path);
3473 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3474 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3475 unless (defined $mode && defined $blob) {
3476 die "$path was not found in commit $self->{c} (r$rev)\n";
3477 }
3478 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3479 $mode = '120000';
3480 }
3481 out:
3482 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3483 pool => SVN::Pool->new, action => 'M' };
3484 }
3485
3486 sub add_file {
3487 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3488 my $mode;
3489
3490 if (!$self->is_path_ignored($path)) {
3491 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3492 delete $self->{empty}->{$dir};
3493 $mode = '100644';
3494 }
3495 { path => $path, mode_a => $mode, mode_b => $mode,
3496 pool => SVN::Pool->new, action => 'A' };
3497 }
3498
3499 sub add_directory {
3500 my ($self, $path, $cp_path, $cp_rev) = @_;
3501 goto out if $self->is_path_ignored($path);
3502 my $gpath = $self->git_path($path);
3503 if ($gpath eq '') {
3504 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3505 -r --name-only -z/,
3506 $self->{c});
3507 local $/ = "\0";
3508 while (<$ls>) {
3509 chomp;
3510 $self->{gii}->remove($_);
3511 print "\tD\t$_\n" unless $::_q;
3512 }
3513 command_close_pipe($ls, $ctx);
3514 $self->{empty}->{$path} = 0;
3515 }
3516 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3517 delete $self->{empty}->{$dir};
3518 $self->{empty}->{$path} = 1;
3519 out:
3520 { path => $path };
3521 }
3522
3523 sub change_dir_prop {
3524 my ($self, $db, $prop, $value) = @_;
3525 return undef if $self->is_path_ignored($db->{path});
3526 $self->{dir_prop}->{$db->{path}} ||= {};
3527 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3528 undef;
3529 }
3530
3531 sub absent_directory {
3532 my ($self, $path, $pb) = @_;
3533 return undef if $self->is_path_ignored($path);
3534 $self->{absent_dir}->{$pb->{path}} ||= [];
3535 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3536 undef;
3537 }
3538
3539 sub absent_file {
3540 my ($self, $path, $pb) = @_;
3541 return undef if $self->is_path_ignored($path);
3542 $self->{absent_file}->{$pb->{path}} ||= [];
3543 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3544 undef;
3545 }
3546
3547 sub change_file_prop {
3548 my ($self, $fb, $prop, $value) = @_;
3549 return undef if $self->is_path_ignored($fb->{path});
3550 if ($prop eq 'svn:executable') {
3551 if ($fb->{mode_b} != 120000) {
3552 $fb->{mode_b} = defined $value ? 100755 : 100644;
3553 }
3554 } elsif ($prop eq 'svn:special') {
3555 $fb->{mode_b} = defined $value ? 120000 : 100644;
3556 } else {
3557 $self->{file_prop}->{$fb->{path}} ||= {};
3558 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3559 }
3560 undef;
3561 }
3562
3563 sub apply_textdelta {
3564 my ($self, $fb, $exp) = @_;
3565 return undef if $self->is_path_ignored($fb->{path});
3566 my $fh = $::_repository->temp_acquire('svn_delta');
3567 # $fh gets auto-closed() by SVN::TxDelta::apply(),
3568 # (but $base does not,) so dup() it for reading in close_file
3569 open my $dup, '<&', $fh or croak $!;
3570 my $base = $::_repository->temp_acquire('git_blob');
3571
3572 if ($fb->{blob}) {
3573 my ($base_is_link, $size);
3574
3575 if ($fb->{mode_a} eq '120000' &&
3576 ! $self->{empty_symlinks}->{$fb->{path}}) {
3577 print $base 'link ' or die "print $!\n";
3578 $base_is_link = 1;
3579 }
3580 retry:
3581 $size = $::_repository->cat_blob($fb->{blob}, $base);
3582 die "Failed to read object $fb->{blob}" if ($size < 0);
3583
3584 if (defined $exp) {
3585 seek $base, 0, 0 or croak $!;
3586 my $got = ::md5sum($base);
3587 if ($got ne $exp) {
3588 my $err = "Checksum mismatch: ".
3589 "$fb->{path} $fb->{blob}\n" .
3590 "expected: $exp\n" .
3591 " got: $got\n";
3592 if ($base_is_link) {
3593 warn $err,
3594 "Retrying... (possibly ",
3595 "a bad symlink from SVN)\n";
3596 $::_repository->temp_reset($base);
3597 $base_is_link = 0;
3598 goto retry;
3599 }
3600 die $err;
3601 }
3602 }
3603 }
3604 seek $base, 0, 0 or croak $!;
3605 $fb->{fh} = $fh;
3606 $fb->{base} = $base;
3607 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3608 }
3609
3610 sub close_file {
3611 my ($self, $fb, $exp) = @_;
3612 return undef if $self->is_path_ignored($fb->{path});
3613
3614 my $hash;
3615 my $path = $self->git_path($fb->{path});
3616 if (my $fh = $fb->{fh}) {
3617 if (defined $exp) {
3618 seek($fh, 0, 0) or croak $!;
3619 my $got = ::md5sum($fh);
3620 if ($got ne $exp) {
3621 die "Checksum mismatch: $path\n",
3622 "expected: $exp\n got: $got\n";
3623 }
3624 }
3625 if ($fb->{mode_b} == 120000) {
3626 sysseek($fh, 0, 0) or croak $!;
3627 my $rd = sysread($fh, my $buf, 5);
3628
3629 if (!defined $rd) {
3630 croak "sysread: $!\n";
3631 } elsif ($rd == 0) {
3632 warn "$path has mode 120000",
3633 " but it points to nothing\n",
3634 "converting to an empty file with mode",
3635 " 100644\n";
3636 $fb->{mode_b} = '100644';
3637 } elsif ($buf ne 'link ') {
3638 warn "$path has mode 120000",
3639 " but is not a link\n";
3640 } else {
3641 my $tmp_fh = $::_repository->temp_acquire(
3642 'svn_hash');
3643 my $res;
3644 while ($res = sysread($fh, my $str, 1024)) {
3645 my $out = syswrite($tmp_fh, $str, $res);
3646 defined($out) && $out == $res
3647 or croak("write ",
3648 Git::temp_path($tmp_fh),
3649 ": $!\n");
3650 }
3651 defined $res or croak $!;
3652
3653 ($fh, $tmp_fh) = ($tmp_fh, $fh);
3654 Git::temp_release($tmp_fh, 1);
3655 }
3656 }
3657
3658 $hash = $::_repository->hash_and_insert_object(
3659 Git::temp_path($fh));
3660 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3661
3662 Git::temp_release($fb->{base}, 1);
3663 Git::temp_release($fh, 1);
3664 } else {
3665 $hash = $fb->{blob} or die "no blob information\n";
3666 }
3667 $fb->{pool}->clear;
3668 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3669 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3670 undef;
3671 }
3672
3673 sub abort_edit {
3674 my $self = shift;
3675 $self->{nr} = $self->{gii}->{nr};
3676 delete $self->{gii};
3677 $self->SUPER::abort_edit(@_);
3678 }
3679
3680 sub close_edit {
3681 my $self = shift;
3682 $self->{git_commit_ok} = 1;
3683 $self->{nr} = $self->{gii}->{nr};
3684 delete $self->{gii};
3685 $self->SUPER::close_edit(@_);
3686 }
3687
3688 package SVN::Git::Editor;
3689 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3690 use strict;
3691 use warnings;
3692 use Carp qw/croak/;
3693 use IO::File;
3694
3695 sub new {
3696 my ($class, $opts) = @_;
3697 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3698 die "$_ required!\n" unless (defined $opts->{$_});
3699 }
3700
3701 my $pool = SVN::Pool->new;
3702 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3703 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3704 $opts->{r}, $mods);
3705
3706 # $opts->{ra} functions should not be used after this:
3707 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3708 $opts->{editor_cb}, $pool);
3709 my $self = SVN::Delta::Editor->new(@ce, $pool);
3710 bless $self, $class;
3711 foreach (qw/svn_path r tree_a tree_b/) {
3712 $self->{$_} = $opts->{$_};
3713 }
3714 $self->{url} = $opts->{ra}->{url};
3715 $self->{mods} = $mods;
3716 $self->{types} = $types;
3717 $self->{pool} = $pool;
3718 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3719 $self->{rm} = { };
3720 $self->{path_prefix} = length $self->{svn_path} ?
3721 "$self->{svn_path}/" : '';
3722 $self->{config} = $opts->{config};
3723 return $self;
3724 }
3725
3726 sub generate_diff {
3727 my ($tree_a, $tree_b) = @_;
3728 my @diff_tree = qw(diff-tree -z -r);
3729 if ($_cp_similarity) {
3730 push @diff_tree, "-C$_cp_similarity";
3731 } else {
3732 push @diff_tree, '-C';
3733 }
3734 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3735 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3736 push @diff_tree, $tree_a, $tree_b;
3737 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3738 local $/ = "\0";
3739 my $state = 'meta';
3740 my @mods;
3741 while (<$diff_fh>) {
3742 chomp $_; # this gets rid of the trailing "\0"
3743 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3744 ($::sha1)\s($::sha1)\s
3745 ([MTCRAD])\d*$/xo) {
3746 push @mods, { mode_a => $1, mode_b => $2,
3747 sha1_a => $3, sha1_b => $4,
3748 chg => $5 };
3749 if ($5 =~ /^(?:C|R)$/) {
3750 $state = 'file_a';
3751 } else {
3752 $state = 'file_b';
3753 }
3754 } elsif ($state eq 'file_a') {
3755 my $x = $mods[$#mods] or croak "Empty array\n";
3756 if ($x->{chg} !~ /^(?:C|R)$/) {
3757 croak "Error parsing $_, $x->{chg}\n";
3758 }
3759 $x->{file_a} = $_;
3760 $state = 'file_b';
3761 } elsif ($state eq 'file_b') {
3762 my $x = $mods[$#mods] or croak "Empty array\n";
3763 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3764 croak "Error parsing $_, $x->{chg}\n";
3765 }
3766 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3767 croak "Error parsing $_, $x->{chg}\n";
3768 }
3769 $x->{file_b} = $_;
3770 $state = 'meta';
3771 } else {
3772 croak "Error parsing $_\n";
3773 }
3774 }
3775 command_close_pipe($diff_fh, $ctx);
3776 \@mods;
3777 }
3778
3779 sub check_diff_paths {
3780 my ($ra, $pfx, $rev, $mods) = @_;
3781 my %types;
3782 $pfx .= '/' if length $pfx;
3783
3784 sub type_diff_paths {
3785 my ($ra, $types, $path, $rev) = @_;
3786 my @p = split m#/+#, $path;
3787 my $c = shift @p;
3788 unless (defined $types->{$c}) {
3789 $types->{$c} = $ra->check_path($c, $rev);
3790 }
3791 while (@p) {
3792 $c .= '/' . shift @p;
3793 next if defined $types->{$c};
3794 $types->{$c} = $ra->check_path($c, $rev);
3795 }
3796 }
3797
3798 foreach my $m (@$mods) {
3799 foreach my $f (qw/file_a file_b/) {
3800 next unless defined $m->{$f};
3801 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3802 if (length $pfx.$dir && ! defined $types{$dir}) {
3803 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3804 }
3805 }
3806 }
3807 \%types;
3808 }
3809
3810 sub split_path {
3811 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3812 }
3813
3814 sub repo_path {
3815 my ($self, $path) = @_;
3816 $self->{path_prefix}.(defined $path ? $path : '');
3817 }
3818
3819 sub url_path {
3820 my ($self, $path) = @_;
3821 if ($self->{url} =~ m#^https?://#) {
3822 $path =~ s/([^~a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3823 }
3824 $self->{url} . '/' . $self->repo_path($path);
3825 }
3826
3827 sub rmdirs {
3828 my ($self) = @_;
3829 my $rm = $self->{rm};
3830 delete $rm->{''}; # we never delete the url we're tracking
3831 return unless %$rm;
3832
3833 foreach (keys %$rm) {
3834 my @d = split m#/#, $_;
3835 my $c = shift @d;
3836 $rm->{$c} = 1;
3837 while (@d) {
3838 $c .= '/' . shift @d;
3839 $rm->{$c} = 1;
3840 }
3841 }
3842 delete $rm->{$self->{svn_path}};
3843 delete $rm->{''}; # we never delete the url we're tracking
3844 return unless %$rm;
3845
3846 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3847 $self->{tree_b});
3848 local $/ = "\0";
3849 while (<$fh>) {
3850 chomp;
3851 my @dn = split m#/#, $_;
3852 while (pop @dn) {
3853 delete $rm->{join '/', @dn};
3854 }
3855 unless (%$rm) {
3856 close $fh;
3857 return;
3858 }
3859 }
3860 command_close_pipe($fh, $ctx);
3861
3862 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3863 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3864 $self->close_directory($bat->{$d}, $p);
3865 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3866 print "\tD+\t$d/\n" unless $::_q;
3867 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3868 delete $bat->{$d};
3869 }
3870 }
3871
3872 sub open_or_add_dir {
3873 my ($self, $full_path, $baton) = @_;
3874 my $t = $self->{types}->{$full_path};
3875 if (!defined $t) {
3876 die "$full_path not known in r$self->{r} or we have a bug!\n";
3877 }
3878 {
3879 no warnings 'once';
3880 # SVN::Node::none and SVN::Node::file are used only once,
3881 # so we're shutting up Perl's warnings about them.
3882 if ($t == $SVN::Node::none) {
3883 return $self->add_directory($full_path, $baton,
3884 undef, -1, $self->{pool});
3885 } elsif ($t == $SVN::Node::dir) {
3886 return $self->open_directory($full_path, $baton,
3887 $self->{r}, $self->{pool});
3888 } # no warnings 'once'
3889 print STDERR "$full_path already exists in repository at ",
3890 "r$self->{r} and it is not a directory (",
3891 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3892 } # no warnings 'once'
3893 exit 1;
3894 }
3895
3896 sub ensure_path {
3897 my ($self, $path) = @_;
3898 my $bat = $self->{bat};
3899 my $repo_path = $self->repo_path($path);
3900 return $bat->{''} unless (length $repo_path);
3901 my @p = split m#/+#, $repo_path;
3902 my $c = shift @p;
3903 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3904 while (@p) {
3905 my $c0 = $c;
3906 $c .= '/' . shift @p;
3907 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3908 }
3909 return $bat->{$c};
3910 }
3911
3912 # Subroutine to convert a globbing pattern to a regular expression.
3913 # From perl cookbook.
3914 sub glob2pat {
3915 my $globstr = shift;
3916 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
3917 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
3918 return '^' . $globstr . '$';
3919 }
3920
3921 sub check_autoprop {
3922 my ($self, $pattern, $properties, $file, $fbat) = @_;
3923 # Convert the globbing pattern to a regular expression.
3924 my $regex = glob2pat($pattern);
3925 # Check if the pattern matches the file name.
3926 if($file =~ m/($regex)/) {
3927 # Parse the list of properties to set.
3928 my @props = split(/;/, $properties);
3929 foreach my $prop (@props) {
3930 # Parse 'name=value' syntax and set the property.
3931 if ($prop =~ /([^=]+)=(.*)/) {
3932 my ($n,$v) = ($1,$2);
3933 for ($n, $v) {
3934 s/^\s+//; s/\s+$//;
3935 }
3936 $self->change_file_prop($fbat, $n, $v);
3937 }
3938 }
3939 }
3940 }
3941
3942 sub apply_autoprops {
3943 my ($self, $file, $fbat) = @_;
3944 my $conf_t = ${$self->{config}}{'config'};
3945 no warnings 'once';
3946 # Check [miscellany]/enable-auto-props in svn configuration.
3947 if (SVN::_Core::svn_config_get_bool(
3948 $conf_t,
3949 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
3950 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
3951 0)) {
3952 # Auto-props are enabled. Enumerate them to look for matches.
3953 my $callback = sub {
3954 $self->check_autoprop($_[0], $_[1], $file, $fbat);
3955 };
3956 SVN::_Core::svn_config_enumerate(
3957 $conf_t,
3958 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
3959 $callback);
3960 }
3961 }
3962
3963 sub A {
3964 my ($self, $m) = @_;
3965 my ($dir, $file) = split_path($m->{file_b});
3966 my $pbat = $self->ensure_path($dir);
3967 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3968 undef, -1);
3969 print "\tA\t$m->{file_b}\n" unless $::_q;
3970 $self->apply_autoprops($file, $fbat);
3971 $self->chg_file($fbat, $m);
3972 $self->close_file($fbat,undef,$self->{pool});
3973 }
3974
3975 sub C {
3976 my ($self, $m) = @_;
3977 my ($dir, $file) = split_path($m->{file_b});
3978 my $pbat = $self->ensure_path($dir);
3979 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3980 $self->url_path($m->{file_a}), $self->{r});
3981 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3982 $self->chg_file($fbat, $m);
3983 $self->close_file($fbat,undef,$self->{pool});
3984 }
3985
3986 sub delete_entry {
3987 my ($self, $path, $pbat) = @_;
3988 my $rpath = $self->repo_path($path);
3989 my ($dir, $file) = split_path($rpath);
3990 $self->{rm}->{$dir} = 1;
3991 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3992 }
3993
3994 sub R {
3995 my ($self, $m) = @_;
3996 my ($dir, $file) = split_path($m->{file_b});
3997 my $pbat = $self->ensure_path($dir);
3998 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3999 $self->url_path($m->{file_a}), $self->{r});
4000 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4001 $self->apply_autoprops($file, $fbat);
4002 $self->chg_file($fbat, $m);
4003 $self->close_file($fbat,undef,$self->{pool});
4004
4005 ($dir, $file) = split_path($m->{file_a});
4006 $pbat = $self->ensure_path($dir);
4007 $self->delete_entry($m->{file_a}, $pbat);
4008 }
4009
4010 sub M {
4011 my ($self, $m) = @_;
4012 my ($dir, $file) = split_path($m->{file_b});
4013 my $pbat = $self->ensure_path($dir);
4014 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4015 $pbat,$self->{r},$self->{pool});
4016 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4017 $self->chg_file($fbat, $m);
4018 $self->close_file($fbat,undef,$self->{pool});
4019 }
4020
4021 sub T { shift->M(@_) }
4022
4023 sub change_file_prop {
4024 my ($self, $fbat, $pname, $pval) = @_;
4025 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4026 }
4027
4028 sub _chg_file_get_blob ($$$$) {
4029 my ($self, $fbat, $m, $which) = @_;
4030 my $fh = $::_repository->temp_acquire("git_blob_$which");
4031 if ($m->{"mode_$which"} =~ /^120/) {
4032 print $fh 'link ' or croak $!;
4033 $self->change_file_prop($fbat,'svn:special','*');
4034 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4035 $self->change_file_prop($fbat,'svn:special',undef);
4036 }
4037 my $blob = $m->{"sha1_$which"};
4038 return ($fh,) if ($blob =~ /^0{40}$/);
4039 my $size = $::_repository->cat_blob($blob, $fh);
4040 croak "Failed to read object $blob" if ($size < 0);
4041 $fh->flush == 0 or croak $!;
4042 seek $fh, 0, 0 or croak $!;
4043
4044 my $exp = ::md5sum($fh);
4045 seek $fh, 0, 0 or croak $!;
4046 return ($fh, $exp);
4047 }
4048
4049 sub chg_file {
4050 my ($self, $fbat, $m) = @_;
4051 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4052 $self->change_file_prop($fbat,'svn:executable','*');
4053 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4054 $self->change_file_prop($fbat,'svn:executable',undef);
4055 }
4056 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4057 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4058 my $pool = SVN::Pool->new;
4059 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4060 if (-s $fh_a) {
4061 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4062 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4063 if (defined $res) {
4064 die "Unexpected result from send_txstream: $res\n",
4065 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4066 }
4067 } else {
4068 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4069 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4070 if ($got ne $exp_b);
4071 }
4072 Git::temp_release($fh_b, 1);
4073 Git::temp_release($fh_a, 1);
4074 $pool->clear;
4075 }
4076
4077 sub D {
4078 my ($self, $m) = @_;
4079 my ($dir, $file) = split_path($m->{file_b});
4080 my $pbat = $self->ensure_path($dir);
4081 print "\tD\t$m->{file_b}\n" unless $::_q;
4082 $self->delete_entry($m->{file_b}, $pbat);
4083 }
4084
4085 sub close_edit {
4086 my ($self) = @_;
4087 my ($p,$bat) = ($self->{pool}, $self->{bat});
4088 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4089 next if $_ eq '';
4090 $self->close_directory($bat->{$_}, $p);
4091 }
4092 $self->close_directory($bat->{''}, $p);
4093 $self->SUPER::close_edit($p);
4094 $p->clear;
4095 }
4096
4097 sub abort_edit {
4098 my ($self) = @_;
4099 $self->SUPER::abort_edit($self->{pool});
4100 }
4101
4102 sub DESTROY {
4103 my $self = shift;
4104 $self->SUPER::DESTROY(@_);
4105 $self->{pool}->clear;
4106 }
4107
4108 # this drives the editor
4109 sub apply_diff {
4110 my ($self) = @_;
4111 my $mods = $self->{mods};
4112 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4113 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4114 my $f = $m->{chg};
4115 if (defined $o{$f}) {
4116 $self->$f($m);
4117 } else {
4118 fatal("Invalid change type: $f");
4119 }
4120 }
4121 $self->rmdirs if $_rmdir;
4122 if (@$mods == 0) {
4123 $self->abort_edit;
4124 } else {
4125 $self->close_edit;
4126 }
4127 return scalar @$mods;
4128 }
4129
4130 package Git::SVN::Ra;
4131 use vars qw/@ISA $config_dir $_log_window_size/;
4132 use strict;
4133 use warnings;
4134 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4135
4136 BEGIN {
4137 # enforce temporary pool usage for some simple functions
4138 no strict 'refs';
4139 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4140 get_file/) {
4141 my $SUPER = "SUPER::$f";
4142 *$f = sub {
4143 my $self = shift;
4144 my $pool = SVN::Pool->new;
4145 my @ret = $self->$SUPER(@_,$pool);
4146 $pool->clear;
4147 wantarray ? @ret : $ret[0];
4148 };
4149 }
4150 }
4151
4152 sub _auth_providers () {
4153 [
4154 SVN::Client::get_simple_provider(),
4155 SVN::Client::get_ssl_server_trust_file_provider(),
4156 SVN::Client::get_simple_prompt_provider(
4157 \&Git::SVN::Prompt::simple, 2),
4158 SVN::Client::get_ssl_client_cert_file_provider(),
4159 SVN::Client::get_ssl_client_cert_prompt_provider(
4160 \&Git::SVN::Prompt::ssl_client_cert, 2),
4161 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4162 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4163 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4164 SVN::Client::get_username_provider(),
4165 SVN::Client::get_ssl_server_trust_prompt_provider(
4166 \&Git::SVN::Prompt::ssl_server_trust),
4167 SVN::Client::get_username_prompt_provider(
4168 \&Git::SVN::Prompt::username, 2)
4169 ]
4170 }
4171
4172 sub escape_uri_only {
4173 my ($uri) = @_;
4174 my @tmp;
4175 foreach (split m{/}, $uri) {
4176 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4177 push @tmp, $_;
4178 }
4179 join('/', @tmp);
4180 }
4181
4182 sub escape_url {
4183 my ($url) = @_;
4184 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4185 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4186 $url = "$scheme://$domain$uri";
4187 }
4188 $url;
4189 }
4190
4191 sub new {
4192 my ($class, $url) = @_;
4193 $url =~ s!/+$!!;
4194 return $RA if ($RA && $RA->{url} eq $url);
4195
4196 SVN::_Core::svn_config_ensure($config_dir, undef);
4197 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4198 my $config = SVN::Core::config_get_config($config_dir);
4199 $RA = undef;
4200 my $dont_store_passwords = 1;
4201 my $conf_t = ${$config}{'config'};
4202 {
4203 no warnings 'once';
4204 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4205 # produces warnings that variables are used only once.
4206 # I had not found the better way to shut them up, so
4207 # the warnings of type 'once' are disabled in this block.
4208 if (SVN::_Core::svn_config_get_bool($conf_t,
4209 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4210 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4211 1) == 0) {
4212 SVN::_Core::svn_auth_set_parameter($baton,
4213 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4214 bless (\$dont_store_passwords, "_p_void"));
4215 }
4216 if (SVN::_Core::svn_config_get_bool($conf_t,
4217 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4218 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4219 1) == 0) {
4220 $Git::SVN::Prompt::_no_auth_cache = 1;
4221 }
4222 } # no warnings 'once'
4223 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4224 config => $config,
4225 pool => SVN::Pool->new,
4226 auth_provider_callbacks => $callbacks);
4227 $self->{url} = $url;
4228 $self->{svn_path} = $url;
4229 $self->{repos_root} = $self->get_repos_root;
4230 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4231 $self->{cache} = { check_path => { r => 0, data => {} },
4232 get_dir => { r => 0, data => {} } };
4233 $RA = bless $self, $class;
4234 }
4235
4236 sub check_path {
4237 my ($self, $path, $r) = @_;
4238 my $cache = $self->{cache}->{check_path};
4239 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4240 return $cache->{data}->{$path};
4241 }
4242 my $pool = SVN::Pool->new;
4243 my $t = $self->SUPER::check_path($path, $r, $pool);
4244 $pool->clear;
4245 if ($r != $cache->{r}) {
4246 %{$cache->{data}} = ();
4247 $cache->{r} = $r;
4248 }
4249 $cache->{data}->{$path} = $t;
4250 }
4251
4252 sub get_dir {
4253 my ($self, $dir, $r) = @_;
4254 my $cache = $self->{cache}->{get_dir};
4255 if ($r == $cache->{r}) {
4256 if (my $x = $cache->{data}->{$dir}) {
4257 return wantarray ? @$x : $x->[0];
4258 }
4259 }
4260 my $pool = SVN::Pool->new;
4261 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4262 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4263 $pool->clear;
4264 if ($r != $cache->{r}) {
4265 %{$cache->{data}} = ();
4266 $cache->{r} = $r;
4267 }
4268 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4269 wantarray ? (\%dirents, $r, $props) : \%dirents;
4270 }
4271
4272 sub DESTROY {
4273 # do not call the real DESTROY since we store ourselves in $RA
4274 }
4275
4276 # get_log(paths, start, end, limit,
4277 # discover_changed_paths, strict_node_history, receiver)
4278 sub get_log {
4279 my ($self, @args) = @_;
4280 my $pool = SVN::Pool->new;
4281
4282 # the limit parameter was not supported in SVN 1.1.x, so we
4283 # drop it. Therefore, the receiver callback passed to it
4284 # is made aware of this limitation by being wrapped if
4285 # the limit passed to is being wrapped.
4286 if ($SVN::Core::VERSION le '1.2.0') {
4287 my $limit = splice(@args, 3, 1);
4288 if ($limit > 0) {
4289 my $receiver = pop @args;
4290 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4291 }
4292 }
4293 my $ret = $self->SUPER::get_log(@args, $pool);
4294 $pool->clear;
4295 $ret;
4296 }
4297
4298 sub trees_match {
4299 my ($self, $url1, $rev1, $url2, $rev2) = @_;
4300 my $ctx = SVN::Client->new(auth => _auth_providers);
4301 my $out = IO::File->new_tmpfile;
4302
4303 # older SVN (1.1.x) doesn't take $pool as the last parameter for
4304 # $ctx->diff(), so we'll create a default one
4305 my $pool = SVN::Pool->new_default_sub;
4306
4307 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4308 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4309 $out->flush;
4310 my $ret = (($out->stat)[7] == 0);
4311 close $out or croak $!;
4312
4313 $ret;
4314 }
4315
4316 sub get_commit_editor {
4317 my ($self, $log, $cb, $pool) = @_;
4318 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4319 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4320 }
4321
4322 sub gs_do_update {
4323 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4324 my $new = ($rev_a == $rev_b);
4325 my $path = $gs->{path};
4326
4327 if ($new && -e $gs->{index}) {
4328 unlink $gs->{index} or die
4329 "Couldn't unlink index: $gs->{index}: $!\n";
4330 }
4331 my $pool = SVN::Pool->new;
4332 $editor->set_path_strip($path);
4333 my (@pc) = split m#/#, $path;
4334 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4335 1, $editor, $pool);
4336 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4337
4338 # Since we can't rely on svn_ra_reparent being available, we'll
4339 # just have to do some magic with set_path to make it so
4340 # we only want a partial path.
4341 my $sp = '';
4342 my $final = join('/', @pc);
4343 while (@pc) {
4344 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4345 $sp .= '/' if length $sp;
4346 $sp .= shift @pc;
4347 }
4348 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4349
4350 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4351
4352 $reporter->finish_report($pool);
4353 $pool->clear;
4354 $editor->{git_commit_ok};
4355 }
4356
4357 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4358 # svn_ra_reparent didn't work before 1.4)
4359 sub gs_do_switch {
4360 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4361 my $path = $gs->{path};
4362 my $pool = SVN::Pool->new;
4363
4364 my $full_url = $self->{url};
4365 my $old_url = $full_url;
4366 $full_url .= '/' . escape_uri_only($path) if length $path;
4367 my ($ra, $reparented);
4368
4369 if ($old_url =~ m#^svn(\+ssh)?://#) {
4370 $_[0] = undef;
4371 $self = undef;
4372 $RA = undef;
4373 $ra = Git::SVN::Ra->new($full_url);
4374 $ra_invalid = 1;
4375 } elsif ($old_url ne $full_url) {
4376 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4377 $self->{url} = $full_url;
4378 $reparented = 1;
4379 }
4380
4381 $ra ||= $self;
4382 $url_b = escape_url($url_b);
4383 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4384 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4385 $reporter->set_path('', $rev_a, 0, @lock, $pool);
4386 $reporter->finish_report($pool);
4387
4388 if ($reparented) {
4389 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4390 $self->{url} = $old_url;
4391 }
4392
4393 $pool->clear;
4394 $editor->{git_commit_ok};
4395 }
4396
4397 sub longest_common_path {
4398 my ($gsv, $globs) = @_;
4399 my %common;
4400 my $common_max = scalar @$gsv;
4401
4402 foreach my $gs (@$gsv) {
4403 my @tmp = split m#/#, $gs->{path};
4404 my $p = '';
4405 foreach (@tmp) {
4406 $p .= length($p) ? "/$_" : $_;
4407 $common{$p} ||= 0;
4408 $common{$p}++;
4409 }
4410 }
4411 $globs ||= [];
4412 $common_max += scalar @$globs;
4413 foreach my $glob (@$globs) {
4414 my @tmp = split m#/#, $glob->{path}->{left};
4415 my $p = '';
4416 foreach (@tmp) {
4417 $p .= length($p) ? "/$_" : $_;
4418 $common{$p} ||= 0;
4419 $common{$p}++;
4420 }
4421 }
4422
4423 my $longest_path = '';
4424 foreach (sort {length $b <=> length $a} keys %common) {
4425 if ($common{$_} == $common_max) {
4426 $longest_path = $_;
4427 last;
4428 }
4429 }
4430 $longest_path;
4431 }
4432
4433 sub gs_fetch_loop_common {
4434 my ($self, $base, $head, $gsv, $globs) = @_;
4435 return if ($base > $head);
4436 my $inc = $_log_window_size;
4437 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4438 my $longest_path = longest_common_path($gsv, $globs);
4439 my $ra_url = $self->{url};
4440 while (1) {
4441 my %revs;
4442 my $err;
4443 my $err_handler = $SVN::Error::handler;
4444 $SVN::Error::handler = sub {
4445 ($err) = @_;
4446 skip_unknown_revs($err);
4447 };
4448 sub _cb {
4449 my ($paths, $r, $author, $date, $log) = @_;
4450 [ dup_changed_paths($paths),
4451 { author => $author, date => $date, log => $log } ];
4452 }
4453 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4454 sub { $revs{$_[1]} = _cb(@_) });
4455 if ($err) {
4456 print "Checked through r$max\r";
4457 }
4458 if ($err && $max >= $head) {
4459 print STDERR "Path '$longest_path' ",
4460 "was probably deleted:\n",
4461 $err->expanded_message,
4462 "\nWill attempt to follow ",
4463 "revisions r$min .. r$max ",
4464 "committed before the deletion\n";
4465 my $hi = $max;
4466 while (--$hi >= $min) {
4467 my $ok;
4468 $self->get_log([$longest_path], $min, $hi,
4469 0, 1, 1, sub {
4470 $ok ||= $_[1];
4471 $revs{$_[1]} = _cb(@_) });
4472 if ($ok) {
4473 print STDERR "r$min .. r$ok OK\n";
4474 last;
4475 }
4476 }
4477 }
4478 $SVN::Error::handler = $err_handler;
4479
4480 my %exists = map { $_->{path} => $_ } @$gsv;
4481 foreach my $r (sort {$a <=> $b} keys %revs) {
4482 my ($paths, $logged) = @{$revs{$r}};
4483
4484 foreach my $gs ($self->match_globs(\%exists, $paths,
4485 $globs, $r)) {
4486 if ($gs->rev_map_max >= $r) {
4487 next;
4488 }
4489 next unless $gs->match_paths($paths, $r);
4490 $gs->{logged_rev_props} = $logged;
4491 if (my $last_commit = $gs->last_commit) {
4492 $gs->assert_index_clean($last_commit);
4493 }
4494 my $log_entry = $gs->do_fetch($paths, $r);
4495 if ($log_entry) {
4496 $gs->do_git_commit($log_entry);
4497 }
4498 $INDEX_FILES{$gs->{index}} = 1;
4499 }
4500 foreach my $g (@$globs) {
4501 my $k = "svn-remote.$g->{remote}." .
4502 "$g->{t}-maxRev";
4503 Git::SVN::tmp_config($k, $r);
4504 }
4505 if ($ra_invalid) {
4506 $_[0] = undef;
4507 $self = undef;
4508 $RA = undef;
4509 $self = Git::SVN::Ra->new($ra_url);
4510 $ra_invalid = undef;
4511 }
4512 }
4513 # pre-fill the .rev_db since it'll eventually get filled in
4514 # with '0' x40 if something new gets committed
4515 foreach my $gs (@$gsv) {
4516 next if $gs->rev_map_max >= $max;
4517 next if defined $gs->rev_map_get($max);
4518 $gs->rev_map_set($max, 0 x40);
4519 }
4520 foreach my $g (@$globs) {
4521 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4522 Git::SVN::tmp_config($k, $max);
4523 }
4524 last if $max >= $head;
4525 $min = $max + 1;
4526 $max += $inc;
4527 $max = $head if ($max > $head);
4528 }
4529 Git::SVN::gc();
4530 }
4531
4532 sub get_dir_globbed {
4533 my ($self, $left, $depth, $r) = @_;
4534
4535 my @x = eval { $self->get_dir($left, $r) };
4536 return unless scalar @x == 3;
4537 my $dirents = $x[0];
4538 my @finalents;
4539 foreach my $de (keys %$dirents) {
4540 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4541 if ($depth > 1) {
4542 my @args = ("$left/$de", $depth - 1, $r);
4543 foreach my $dir ($self->get_dir_globbed(@args)) {
4544 push @finalents, "$de/$dir";
4545 }
4546 } else {
4547 push @finalents, $de;
4548 }
4549 }
4550 @finalents;
4551 }
4552
4553 sub match_globs {
4554 my ($self, $exists, $paths, $globs, $r) = @_;
4555
4556 sub get_dir_check {
4557 my ($self, $exists, $g, $r) = @_;
4558
4559 my @dirs = $self->get_dir_globbed($g->{path}->{left},
4560 $g->{path}->{depth},
4561 $r);
4562
4563 foreach my $de (@dirs) {
4564 my $p = $g->{path}->full_path($de);
4565 next if $exists->{$p};
4566 next if (length $g->{path}->{right} &&
4567 ($self->check_path($p, $r) !=
4568 $SVN::Node::dir));
4569 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4570 $g->{ref}->full_path($de), 1);
4571 }
4572 }
4573 foreach my $g (@$globs) {
4574 if (my $path = $paths->{"/$g->{path}->{left}"}) {
4575 if ($path->{action} =~ /^[AR]$/) {
4576 get_dir_check($self, $exists, $g, $r);
4577 }
4578 }
4579 foreach (keys %$paths) {
4580 if (/$g->{path}->{left_regex}/ &&
4581 !/$g->{path}->{regex}/) {
4582 next if $paths->{$_}->{action} !~ /^[AR]$/;
4583 get_dir_check($self, $exists, $g, $r);
4584 }
4585 next unless /$g->{path}->{regex}/;
4586 my $p = $1;
4587 my $pathname = $g->{path}->full_path($p);
4588 next if $exists->{$pathname};
4589 next if ($self->check_path($pathname, $r) !=
4590 $SVN::Node::dir);
4591 $exists->{$pathname} = Git::SVN->init(
4592 $self->{url}, $pathname, undef,
4593 $g->{ref}->full_path($p), 1);
4594 }
4595 my $c = '';
4596 foreach (split m#/#, $g->{path}->{left}) {
4597 $c .= "/$_";
4598 next unless ($paths->{$c} &&
4599 ($paths->{$c}->{action} =~ /^[AR]$/));
4600 get_dir_check($self, $exists, $g, $r);
4601 }
4602 }
4603 values %$exists;
4604 }
4605
4606 sub minimize_url {
4607 my ($self) = @_;
4608 return $self->{url} if ($self->{url} eq $self->{repos_root});
4609 my $url = $self->{repos_root};
4610 my @components = split(m!/!, $self->{svn_path});
4611 my $c = '';
4612 do {
4613 $url .= "/$c" if length $c;
4614 eval { (ref $self)->new($url)->get_latest_revnum };
4615 } while ($@ && ($c = shift @components));
4616 $url;
4617 }
4618
4619 sub can_do_switch {
4620 my $self = shift;
4621 unless (defined $can_do_switch) {
4622 my $pool = SVN::Pool->new;
4623 my $rep = eval {
4624 $self->do_switch(1, '', 0, $self->{url},
4625 SVN::Delta::Editor->new, $pool);
4626 };
4627 if ($@) {
4628 $can_do_switch = 0;
4629 } else {
4630 $rep->abort_report($pool);
4631 $can_do_switch = 1;
4632 }
4633 $pool->clear;
4634 }
4635 $can_do_switch;
4636 }
4637
4638 sub skip_unknown_revs {
4639 my ($err) = @_;
4640 my $errno = $err->apr_err();
4641 # Maybe the branch we're tracking didn't
4642 # exist when the repo started, so it's
4643 # not an error if it doesn't, just continue
4644 #
4645 # Wonderfully consistent library, eh?
4646 # 160013 - svn:// and file://
4647 # 175002 - http(s)://
4648 # 175007 - http(s):// (this repo required authorization, too...)
4649 # More codes may be discovered later...
4650 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4651 my $err_key = $err->expanded_message;
4652 # revision numbers change every time, filter them out
4653 $err_key =~ s/\d+/\0/g;
4654 $err_key = "$errno\0$err_key";
4655 unless ($ignored_err{$err_key}) {
4656 warn "W: Ignoring error from SVN, path probably ",
4657 "does not exist: ($errno): ",
4658 $err->expanded_message,"\n";
4659 warn "W: Do not be alarmed at the above message ",
4660 "git-svn is just searching aggressively for ",
4661 "old history.\n",
4662 "This may take a while on large repositories\n";
4663 $ignored_err{$err_key} = 1;
4664 }
4665 return;
4666 }
4667 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4668 }
4669
4670 # svn_log_changed_path_t objects passed to get_log are likely to be
4671 # overwritten even if only the refs are copied to an external variable,
4672 # so we should dup the structures in their entirety. Using an externally
4673 # passed pool (instead of our temporary and quickly cleared pool in
4674 # Git::SVN::Ra) does not help matters at all...
4675 sub dup_changed_paths {
4676 my ($paths) = @_;
4677 return undef unless $paths;
4678 my %ret;
4679 foreach my $p (keys %$paths) {
4680 my $i = $paths->{$p};
4681 my %s = map { $_ => $i->$_ }
4682 qw/copyfrom_path copyfrom_rev action/;
4683 $ret{$p} = \%s;
4684 }
4685 \%ret;
4686 }
4687
4688 package Git::SVN::Log;
4689 use strict;
4690 use warnings;
4691 use POSIX qw/strftime/;
4692 use Time::Local;
4693 use constant commit_log_separator => ('-' x 72) . "\n";
4694 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4695 %rusers $show_commit $incremental/;
4696 my $l_fmt;
4697
4698 sub cmt_showable {
4699 my ($c) = @_;
4700 return 1 if defined $c->{r};
4701
4702 # big commit message got truncated by the 16k pretty buffer in rev-list
4703 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4704 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4705 @{$c->{l}} = ();
4706 my @log = command(qw/cat-file commit/, $c->{c});
4707
4708 # shift off the headers
4709 shift @log while ($log[0] ne '');
4710 shift @log;
4711
4712 # TODO: make $c->{l} not have a trailing newline in the future
4713 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4714
4715 (undef, $c->{r}, undef) = ::extract_metadata(
4716 (grep(/^git-svn-id: /, @log))[-1]);
4717 }
4718 return defined $c->{r};
4719 }
4720
4721 sub log_use_color {
4722 return $color || Git->repository->get_colorbool('color.diff');
4723 }
4724
4725 sub git_svn_log_cmd {
4726 my ($r_min, $r_max, @args) = @_;
4727 my $head = 'HEAD';
4728 my (@files, @log_opts);
4729 foreach my $x (@args) {
4730 if ($x eq '--' || @files) {
4731 push @files, $x;
4732 } else {
4733 if (::verify_ref("$x^0")) {
4734 $head = $x;
4735 } else {
4736 push @log_opts, $x;
4737 }
4738 }
4739 }
4740
4741 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4742 $gs ||= Git::SVN->_new;
4743 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4744 $gs->refname);
4745 push @cmd, '-r' unless $non_recursive;
4746 push @cmd, qw/--raw --name-status/ if $verbose;
4747 push @cmd, '--color' if log_use_color();
4748 push @cmd, @log_opts;
4749 if (defined $r_max && $r_max == $r_min) {
4750 push @cmd, '--max-count=1';
4751 if (my $c = $gs->rev_map_get($r_max)) {
4752 push @cmd, $c;
4753 }
4754 } elsif (defined $r_max) {
4755 if ($r_max < $r_min) {
4756 ($r_min, $r_max) = ($r_max, $r_min);
4757 }
4758 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4759 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4760 # If there are no commits in the range, both $c_max and $c_min
4761 # will be undefined. If there is at least 1 commit in the
4762 # range, both will be defined.
4763 return () if !defined $c_min || !defined $c_max;
4764 if ($c_min eq $c_max) {
4765 push @cmd, '--max-count=1', $c_min;
4766 } else {
4767 push @cmd, '--boundary', "$c_min..$c_max";
4768 }
4769 }
4770 return (@cmd, @files);
4771 }
4772
4773 # adapted from pager.c
4774 sub config_pager {
4775 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4776 if (!defined $pager) {
4777 $pager = 'less';
4778 } elsif (length $pager == 0 || $pager eq 'cat') {
4779 $pager = undef;
4780 }
4781 $ENV{GIT_PAGER_IN_USE} = defined($pager);
4782 }
4783
4784 sub run_pager {
4785 return unless -t *STDOUT && defined $pager;
4786 pipe my ($rfd, $wfd) or return;
4787 defined(my $pid = fork) or ::fatal "Can't fork: $!";
4788 if (!$pid) {
4789 open STDOUT, '>&', $wfd or
4790 ::fatal "Can't redirect to stdout: $!";
4791 return;
4792 }
4793 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4794 $ENV{LESS} ||= 'FRSX';
4795 exec $pager or ::fatal "Can't run pager: $! ($pager)";
4796 }
4797
4798 sub format_svn_date {
4799 # some systmes don't handle or mishandle %z, so be creative.
4800 my $t = shift || time;
4801 my $gm = timelocal(gmtime($t));
4802 my $sign = qw( + + - )[ $t <=> $gm ];
4803 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
4804 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
4805 }
4806
4807 sub parse_git_date {
4808 my ($t, $tz) = @_;
4809 # Date::Parse isn't in the standard Perl distro :(
4810 if ($tz =~ s/^\+//) {
4811 $t += tz_to_s_offset($tz);
4812 } elsif ($tz =~ s/^\-//) {
4813 $t -= tz_to_s_offset($tz);
4814 }
4815 return $t;
4816 }
4817
4818 sub set_local_timezone {
4819 if (defined $TZ) {
4820 $ENV{TZ} = $TZ;
4821 } else {
4822 delete $ENV{TZ};
4823 }
4824 }
4825
4826 sub tz_to_s_offset {
4827 my ($tz) = @_;
4828 $tz =~ s/(\d\d)$//;
4829 return ($1 * 60) + ($tz * 3600);
4830 }
4831
4832 sub get_author_info {
4833 my ($dest, $author, $t, $tz) = @_;
4834 $author =~ s/(?:^\s*|\s*$)//g;
4835 $dest->{a_raw} = $author;
4836 my $au;
4837 if ($::_authors) {
4838 $au = $rusers{$author} || undef;
4839 }
4840 if (!$au) {
4841 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4842 }
4843 $dest->{t} = $t;
4844 $dest->{tz} = $tz;
4845 $dest->{a} = $au;
4846 $dest->{t_utc} = parse_git_date($t, $tz);
4847 }
4848
4849 sub process_commit {
4850 my ($c, $r_min, $r_max, $defer) = @_;
4851 if (defined $r_min && defined $r_max) {
4852 if ($r_min == $c->{r} && $r_min == $r_max) {
4853 show_commit($c);
4854 return 0;
4855 }
4856 return 1 if $r_min == $r_max;
4857 if ($r_min < $r_max) {
4858 # we need to reverse the print order
4859 return 0 if (defined $limit && --$limit < 0);
4860 push @$defer, $c;
4861 return 1;
4862 }
4863 if ($r_min != $r_max) {
4864 return 1 if ($r_min < $c->{r});
4865 return 1 if ($r_max > $c->{r});
4866 }
4867 }
4868 return 0 if (defined $limit && --$limit < 0);
4869 show_commit($c);
4870 return 1;
4871 }
4872
4873 sub show_commit {
4874 my $c = shift;
4875 if ($oneline) {
4876 my $x = "\n";
4877 if (my $l = $c->{l}) {
4878 while ($l->[0] =~ /^\s*$/) { shift @$l }
4879 $x = $l->[0];
4880 }
4881 $l_fmt ||= 'A' . length($c->{r});
4882 print 'r',pack($l_fmt, $c->{r}),' | ';
4883 print "$c->{c} | " if $show_commit;
4884 print $x;
4885 } else {
4886 show_commit_normal($c);
4887 }
4888 }
4889
4890 sub show_commit_changed_paths {
4891 my ($c) = @_;
4892 return unless $c->{changed};
4893 print "Changed paths:\n", @{$c->{changed}};
4894 }
4895
4896 sub show_commit_normal {
4897 my ($c) = @_;
4898 print commit_log_separator, "r$c->{r} | ";
4899 print "$c->{c} | " if $show_commit;
4900 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4901 my $nr_line = 0;
4902
4903 if (my $l = $c->{l}) {
4904 while ($l->[$#$l] eq "\n" && $#$l > 0
4905 && $l->[($#$l - 1)] eq "\n") {
4906 pop @$l;
4907 }
4908 $nr_line = scalar @$l;
4909 if (!$nr_line) {
4910 print "1 line\n\n\n";
4911 } else {
4912 if ($nr_line == 1) {
4913 $nr_line = '1 line';
4914 } else {
4915 $nr_line .= ' lines';
4916 }
4917 print $nr_line, "\n";
4918 show_commit_changed_paths($c);
4919 print "\n";
4920 print $_ foreach @$l;
4921 }
4922 } else {
4923 print "1 line\n";
4924 show_commit_changed_paths($c);
4925 print "\n";
4926
4927 }
4928 foreach my $x (qw/raw stat diff/) {
4929 if ($c->{$x}) {
4930 print "\n";
4931 print $_ foreach @{$c->{$x}}
4932 }
4933 }
4934 }
4935
4936 sub cmd_show_log {
4937 my (@args) = @_;
4938 my ($r_min, $r_max);
4939 my $r_last = -1; # prevent dupes
4940 set_local_timezone();
4941 if (defined $::_revision) {
4942 if ($::_revision =~ /^(\d+):(\d+)$/) {
4943 ($r_min, $r_max) = ($1, $2);
4944 } elsif ($::_revision =~ /^\d+$/) {
4945 $r_min = $r_max = $::_revision;
4946 } else {
4947 ::fatal "-r$::_revision is not supported, use ",
4948 "standard 'git log' arguments instead";
4949 }
4950 }
4951
4952 config_pager();
4953 @args = git_svn_log_cmd($r_min, $r_max, @args);
4954 if (!@args) {
4955 print commit_log_separator unless $incremental || $oneline;
4956 return;
4957 }
4958 my $log = command_output_pipe(@args);
4959 run_pager();
4960 my (@k, $c, $d, $stat);
4961 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4962 while (<$log>) {
4963 if (/^${esc_color}commit -?($::sha1_short)/o) {
4964 my $cmt = $1;
4965 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4966 $r_last = $c->{r};
4967 process_commit($c, $r_min, $r_max, \@k) or
4968 goto out;
4969 }
4970 $d = undef;
4971 $c = { c => $cmt };
4972 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4973 get_author_info($c, $1, $2, $3);
4974 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4975 # ignore
4976 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4977 push @{$c->{raw}}, $_;
4978 } elsif (/^${esc_color}[ACRMDT]\t/) {
4979 # we could add $SVN->{svn_path} here, but that requires
4980 # remote access at the moment (repo_path_split)...
4981 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
4982 push @{$c->{changed}}, $_;
4983 } elsif (/^${esc_color}diff /o) {
4984 $d = 1;
4985 push @{$c->{diff}}, $_;
4986 } elsif ($d) {
4987 push @{$c->{diff}}, $_;
4988 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4989 $esc_color*[\+\-]*$esc_color$/x) {
4990 $stat = 1;
4991 push @{$c->{stat}}, $_;
4992 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4993 push @{$c->{stat}}, $_;
4994 $stat = undef;
4995 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
4996 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4997 } elsif (s/^${esc_color} //o) {
4998 push @{$c->{l}}, $_;
4999 }
5000 }
5001 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5002 $r_last = $c->{r};
5003 process_commit($c, $r_min, $r_max, \@k);
5004 }
5005 if (@k) {
5006 ($r_min, $r_max) = ($r_max, $r_min);
5007 process_commit($_, $r_min, $r_max) foreach reverse @k;
5008 }
5009 out:
5010 close $log;
5011 print commit_log_separator unless $incremental || $oneline;
5012 }
5013
5014 sub cmd_blame {
5015 my $path = pop;
5016
5017 config_pager();
5018 run_pager();
5019
5020 my ($fh, $ctx, $rev);
5021
5022 if ($_git_format) {
5023 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5024 while (my $line = <$fh>) {
5025 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5026 # Uncommitted edits show up as a rev ID of
5027 # all zeros, which we can't look up with
5028 # cmt_metadata
5029 if ($1 !~ /^0+$/) {
5030 (undef, $rev, undef) =
5031 ::cmt_metadata($1);
5032 $rev = '0' if (!$rev);
5033 } else {
5034 $rev = '0';
5035 }
5036 $rev = sprintf('%-10s', $rev);
5037 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5038 }
5039 print $line;
5040 }
5041 } else {
5042 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5043 '--', $path);
5044 my ($sha1);
5045 my %authors;
5046 my @buffer;
5047 my %dsha; #distinct sha keys
5048
5049 while (my $line = <$fh>) {
5050 push @buffer, $line;
5051 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5052 $dsha{$1} = 1;
5053 }
5054 }
5055
5056 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5057
5058 foreach my $line (@buffer) {
5059 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5060 $rev = $s2r->{$1};
5061 $rev = '0' if (!$rev)
5062 }
5063 elsif ($line =~ /^author (.*)/) {
5064 $authors{$rev} = $1;
5065 $authors{$rev} =~ s/\s/_/g;
5066 }
5067 elsif ($line =~ /^\t(.*)$/) {
5068 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5069 }
5070 }
5071 }
5072 command_close_pipe($fh, $ctx);
5073 }
5074
5075 package Git::SVN::Migration;
5076 # these version numbers do NOT correspond to actual version numbers
5077 # of git nor git-svn. They are just relative.
5078 #
5079 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5080 #
5081 # v1 layout: .git/$id/info/url, refs/remotes/$id
5082 #
5083 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5084 #
5085 # v3 layout: .git/svn/$id, refs/remotes/$id
5086 # - info/url may remain for backwards compatibility
5087 # - this is what we migrate up to this layout automatically,
5088 # - this will be used by git svn init on single branches
5089 # v3.1 layout (auto migrated):
5090 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5091 # for backwards compatibility
5092 #
5093 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5094 # - this is only created for newly multi-init-ed
5095 # repositories. Similar in spirit to the
5096 # --use-separate-remotes option in git-clone (now default)
5097 # - we do not automatically migrate to this (following
5098 # the example set by core git)
5099 #
5100 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
5101 # - newer, more-efficient format that uses 24-bytes per record
5102 # with no filler space.
5103 # - use xxd -c24 < .rev_map.$UUID to view and debug
5104 # - This is a one-way migration, repositories updated to the
5105 # new format will not be able to use old git-svn without
5106 # rebuilding the .rev_db. Rebuilding the rev_db is not
5107 # possible if noMetadata or useSvmProps are set; but should
5108 # be no problem for users that use the (sensible) defaults.
5109 use strict;
5110 use warnings;
5111 use Carp qw/croak/;
5112 use File::Path qw/mkpath/;
5113 use File::Basename qw/dirname basename/;
5114 use vars qw/$_minimize/;
5115
5116 sub migrate_from_v0 {
5117 my $git_dir = $ENV{GIT_DIR};
5118 return undef unless -d $git_dir;
5119 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5120 my $migrated = 0;
5121 while (<$fh>) {
5122 chomp;
5123 my ($id, $orig_ref) = ($_, $_);
5124 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5125 next unless -f "$git_dir/$id/info/url";
5126 my $new_ref = "refs/remotes/$id";
5127 if (::verify_ref("$new_ref^0")) {
5128 print STDERR "W: $orig_ref is probably an old ",
5129 "branch used by an ancient version of ",
5130 "git-svn.\n",
5131 "However, $new_ref also exists.\n",
5132 "We will not be able ",
5133 "to use this branch until this ",
5134 "ambiguity is resolved.\n";
5135 next;
5136 }
5137 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5138 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5139 command_noisy('update-ref', $new_ref, $orig_ref);
5140 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5141 $migrated++;
5142 }
5143 command_close_pipe($fh, $ctx);
5144 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5145 $migrated;
5146 }
5147
5148 sub migrate_from_v1 {
5149 my $git_dir = $ENV{GIT_DIR};
5150 my $migrated = 0;
5151 return $migrated unless -d $git_dir;
5152 my $svn_dir = "$git_dir/svn";
5153
5154 # just in case somebody used 'svn' as their $id at some point...
5155 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5156
5157 print STDERR "Migrating from a git-svn v1 layout...\n";
5158 mkpath([$svn_dir]);
5159 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5160 "$svn_dir\n\t(required for this version ",
5161 "($::VERSION) of git-svn) does not exist.\n";
5162 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5163 while (<$fh>) {
5164 my $x = $_;
5165 next unless $x =~ s#^refs/remotes/##;
5166 chomp $x;
5167 next unless -f "$git_dir/$x/info/url";
5168 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5169 next unless $u;
5170 my $dn = dirname("$git_dir/svn/$x");
5171 mkpath([$dn]) unless -d $dn;
5172 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5173 mkpath(["$git_dir/svn/svn"]);
5174 print STDERR " - $git_dir/$x/info => ",
5175 "$git_dir/svn/$x/info\n";
5176 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5177 croak "$!: $x";
5178 # don't worry too much about these, they probably
5179 # don't exist with repos this old (save for index,
5180 # and we can easily regenerate that)
5181 foreach my $f (qw/unhandled.log index .rev_db/) {
5182 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5183 }
5184 } else {
5185 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5186 rename "$git_dir/$x", "$git_dir/svn/$x" or
5187 croak "$!: $x";
5188 }
5189 $migrated++;
5190 }
5191 command_close_pipe($fh, $ctx);
5192 print STDERR "Done migrating from a git-svn v1 layout\n";
5193 $migrated;
5194 }
5195
5196 sub read_old_urls {
5197 my ($l_map, $pfx, $path) = @_;
5198 my @dir;
5199 foreach (<$path/*>) {
5200 if (-r "$_/info/url") {
5201 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5202 my $ref_id = $pfx . basename $_;
5203 my $url = ::file_to_s("$_/info/url");
5204 $l_map->{$ref_id} = $url;
5205 } elsif (-d $_) {
5206 push @dir, $_;
5207 }
5208 }
5209 foreach (@dir) {
5210 my $x = $_;
5211 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5212 read_old_urls($l_map, $x, $_);
5213 }
5214 }
5215
5216 sub migrate_from_v2 {
5217 my @cfg = command(qw/config -l/);
5218 return if grep /^svn-remote\..+\.url=/, @cfg;
5219 my %l_map;
5220 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5221 my $migrated = 0;
5222
5223 foreach my $ref_id (sort keys %l_map) {
5224 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5225 if ($@) {
5226 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5227 }
5228 $migrated++;
5229 }
5230 $migrated;
5231 }
5232
5233 sub minimize_connections {
5234 my $r = Git::SVN::read_all_remotes();
5235 my $new_urls = {};
5236 my $root_repos = {};
5237 foreach my $repo_id (keys %$r) {
5238 my $url = $r->{$repo_id}->{url} or next;
5239 my $fetch = $r->{$repo_id}->{fetch} or next;
5240 my $ra = Git::SVN::Ra->new($url);
5241
5242 # skip existing cases where we already connect to the root
5243 if (($ra->{url} eq $ra->{repos_root}) ||
5244 ($ra->{repos_root} eq $repo_id)) {
5245 $root_repos->{$ra->{url}} = $repo_id;
5246 next;
5247 }
5248
5249 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5250 my $root_path = $ra->{url};
5251 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5252 foreach my $path (keys %$fetch) {
5253 my $ref_id = $fetch->{$path};
5254 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5255
5256 # make sure we can read when connecting to
5257 # a higher level of a repository
5258 my ($last_rev, undef) = $gs->last_rev_commit;
5259 if (!defined $last_rev) {
5260 $last_rev = eval {
5261 $root_ra->get_latest_revnum;
5262 };
5263 next if $@;
5264 }
5265 my $new = $root_path;
5266 $new .= length $path ? "/$path" : '';
5267 eval {
5268 $root_ra->get_log([$new], $last_rev, $last_rev,
5269 0, 0, 1, sub { });
5270 };
5271 next if $@;
5272 $new_urls->{$ra->{repos_root}}->{$new} =
5273 { ref_id => $ref_id,
5274 old_repo_id => $repo_id,
5275 old_path => $path };
5276 }
5277 }
5278
5279 my @emptied;
5280 foreach my $url (keys %$new_urls) {
5281 # see if we can re-use an existing [svn-remote "repo_id"]
5282 # instead of creating a(n ugly) new section:
5283 my $repo_id = $root_repos->{$url} || $url;
5284
5285 my $fetch = $new_urls->{$url};
5286 foreach my $path (keys %$fetch) {
5287 my $x = $fetch->{$path};
5288 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5289 my $pfx = "svn-remote.$x->{old_repo_id}";
5290
5291 my $old_fetch = quotemeta("$x->{old_path}:".
5292 "refs/remotes/$x->{ref_id}");
5293 command_noisy(qw/config --unset/,
5294 "$pfx.fetch", '^'. $old_fetch . '$');
5295 delete $r->{$x->{old_repo_id}}->
5296 {fetch}->{$x->{old_path}};
5297 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5298 command_noisy(qw/config --unset/,
5299 "$pfx.url");
5300 push @emptied, $x->{old_repo_id}
5301 }
5302 }
5303 }
5304 if (@emptied) {
5305 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5306 print STDERR <<EOF;
5307 The following [svn-remote] sections in your config file ($file) are empty
5308 and can be safely removed:
5309 EOF
5310 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5311 }
5312 }
5313
5314 sub migration_check {
5315 migrate_from_v0();
5316 migrate_from_v1();
5317 migrate_from_v2();
5318 minimize_connections() if $_minimize;
5319 }
5320
5321 package Git::IndexInfo;
5322 use strict;
5323 use warnings;
5324 use Git qw/command_input_pipe command_close_pipe/;
5325
5326 sub new {
5327 my ($class) = @_;
5328 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5329 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5330 }
5331
5332 sub remove {
5333 my ($self, $path) = @_;
5334 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5335 return ++$self->{nr};
5336 }
5337 undef;
5338 }
5339
5340 sub update {
5341 my ($self, $mode, $hash, $path) = @_;
5342 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5343 return ++$self->{nr};
5344 }
5345 undef;
5346 }
5347
5348 sub DESTROY {
5349 my ($self) = @_;
5350 command_close_pipe($self->{gui}, $self->{ctx});
5351 }
5352
5353 package Git::SVN::GlobSpec;
5354 use strict;
5355 use warnings;
5356
5357 sub new {
5358 my ($class, $glob) = @_;
5359 my $re = $glob;
5360 $re =~ s!/+$!!g; # no need for trailing slashes
5361 $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
5362 my $temp = $re;
5363 my ($left, $right) = ($1, $3);
5364 $re = $2;
5365 my $depth = $re =~ tr/*/*/;
5366 if ($depth != $temp =~ tr/*/*/) {
5367 die "Only one set of wildcard directories " .
5368 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5369 }
5370 if ($depth == 0) {
5371 die "One '*' is needed for glob: '$glob'\n";
5372 }
5373 $re =~ s!\*!\[^/\]*!g;
5374 $re = quotemeta($left) . "($re)" . quotemeta($right);
5375 if (length $left && !($left =~ s!/+$!!g)) {
5376 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5377 }
5378 if (length $right && !($right =~ s!^/+!!g)) {
5379 die "Missing leading '/' on right side of: '$glob' ($right)\n";
5380 }
5381 my $left_re = qr/^\/\Q$left\E(\/|$)/;
5382 bless { left => $left, right => $right, left_regex => $left_re,
5383 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5384 }
5385
5386 sub full_path {
5387 my ($self, $path) = @_;
5388 return (length $self->{left} ? "$self->{left}/" : '') .
5389 $path . (length $self->{right} ? "/$self->{right}" : '');
5390 }
5391
5392 __END__
5393
5394 Data structures:
5395
5396
5397 $remotes = { # returned by read_all_remotes()
5398 'svn' => {
5399 # svn-remote.svn.url=https://svn.musicpd.org
5400 url => 'https://svn.musicpd.org',
5401 # svn-remote.svn.fetch=mpd/trunk:trunk
5402 fetch => {
5403 'mpd/trunk' => 'trunk',
5404 },
5405 # svn-remote.svn.tags=mpd/tags/*:tags/*
5406 tags => {
5407 path => {
5408 left => 'mpd/tags',
5409 right => '',
5410 regex => qr!mpd/tags/([^/]+)$!,
5411 glob => 'tags/*',
5412 },
5413 ref => {
5414 left => 'tags',
5415 right => '',
5416 regex => qr!tags/([^/]+)$!,
5417 glob => 'tags/*',
5418 },
5419 }
5420 }
5421 };
5422
5423 $log_entry hashref as returned by libsvn_log_entry()
5424 {
5425 log => 'whitespace-formatted log entry
5426 ', # trailing newline is preserved
5427 revision => '8', # integer
5428 date => '2004-02-24T17:01:44.108345Z', # commit date
5429 author => 'committer name'
5430 };
5431
5432
5433 # this is generated by generate_diff();
5434 @mods = array of diff-index line hashes, each element represents one line
5435 of diff-index output
5436
5437 diff-index line ($m hash)
5438 {
5439 mode_a => first column of diff-index output, no leading ':',
5440 mode_b => second column of diff-index output,
5441 sha1_b => sha1sum of the final blob,
5442 chg => change type [MCRADT],
5443 file_a => original file name of a file (iff chg is 'C' or 'R')
5444 file_b => new/current file name of a file (any chg)
5445 }
5446 ;
5447
5448 # retval of read_url_paths{,_all}();
5449 $l_map = {
5450 # repository root url
5451 'https://svn.musicpd.org' => {
5452 # repository path # GIT_SVN_ID
5453 'mpd/trunk' => 'trunk',
5454 'mpd/tags/0.11.5' => 'tags/0.11.5',
5455 },
5456 }
5457
5458 Notes:
5459 I don't trust the each() function on unless I created %hash myself
5460 because the internal iterator may not have started at base.
Jonathan Nieder’s miscellaneous git branches