?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git svn: Don't create empty directories whose parents were deleted
[git/jrn.git] / git-svn.perl
blobbdd1f966c37cc3454bfcbd0968146dbd73578717
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 $_authors_prog %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 $Git::SVN::_minimize_url = 'unset';
23
24 if (! exists $ENV{SVN_SSH}) {
25 if (exists $ENV{GIT_SSH}) {
26 $ENV{SVN_SSH} = $ENV{GIT_SSH};
27 if ($^O eq 'msys') {
28 $ENV{SVN_SSH} =~ s/\\/\\\\/g;
29 }
30 }
31 }
32
33 $Git::SVN::Log::TZ = $ENV{TZ};
34 $ENV{TZ} = 'UTC';
35 $| = 1; # unbuffer STDOUT
36
37 sub fatal (@) { print STDERR "@_\n"; exit 1 }
38 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
39 require SVN::Ra;
40 require SVN::Delta;
41 if ($SVN::Core::VERSION lt '1.1.0') {
42 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
43 }
44 my $can_compress = eval { require Compress::Zlib; 1};
45 push @Git::SVN::Ra::ISA, 'SVN::Ra';
46 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
47 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
48 use Carp qw/croak/;
49 use Digest::MD5;
50 use IO::File qw//;
51 use File::Basename qw/dirname basename/;
52 use File::Path qw/mkpath/;
53 use File::Spec;
54 use File::Find;
55 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
56 use IPC::Open3;
57 use Git;
58
59 BEGIN {
60 # import functions from Git into our packages, en masse
61 no strict 'refs';
62 foreach (qw/command command_oneline command_noisy command_output_pipe
63 command_input_pipe command_close_pipe
64 command_bidi_pipe command_close_bidi_pipe/) {
65 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
66 Git::SVN::Migration Git::SVN::Log Git::SVN),
67 __PACKAGE__) {
68 *{"${package}::$_"} = \&{"Git::$_"};
69 }
70 }
71 }
72
73 my ($SVN);
74
75 $sha1 = qr/[a-f\d]{40}/;
76 $sha1_short = qr/[a-f\d]{4,40}/;
77 my ($_stdin, $_help, $_edit,
78 $_message, $_file, $_branch_dest,
79 $_template, $_shared,
80 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
81 $_merge, $_strategy, $_dry_run, $_local,
82 $_prefix, $_no_checkout, $_url, $_verbose,
83 $_git_format, $_commit_url, $_tag);
84 $Git::SVN::_follow_parent = 1;
85 $_q ||= 0;
86 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
87 'config-dir=s' => \$Git::SVN::Ra::config_dir,
88 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
89 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
90 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
91 'authors-file|A=s' => \$_authors,
92 'authors-prog=s' => \$_authors_prog,
93 'repack:i' => \$Git::SVN::_repack,
94 'noMetadata' => \$Git::SVN::_no_metadata,
95 'useSvmProps' => \$Git::SVN::_use_svm_props,
96 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
97 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
98 'no-checkout' => \$_no_checkout,
99 'quiet|q+' => \$_q,
100 'repack-flags|repack-args|repack-opts=s' =>
101 \$Git::SVN::_repack_flags,
102 'use-log-author' => \$Git::SVN::_use_log_author,
103 'add-author-from' => \$Git::SVN::_add_author_from,
104 'localtime' => \$Git::SVN::_localtime,
105 %remote_opts );
106
107 my ($_trunk, @_tags, @_branches, $_stdlayout);
108 my %icv;
109 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
110 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
111 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
112 'stdlayout|s' => \$_stdlayout,
113 'minimize-url|m!' => \$Git::SVN::_minimize_url,
114 'no-metadata' => sub { $icv{noMetadata} = 1 },
115 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
116 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
117 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
118 %remote_opts );
119 my %cmt_opts = ( 'edit|e' => \$_edit,
120 'rmdir' => \$SVN::Git::Editor::_rmdir,
121 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
122 'l=i' => \$SVN::Git::Editor::_rename_limit,
123 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
124 );
125
126 my %cmd = (
127 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
128 { 'revision|r=s' => \$_revision,
129 'fetch-all|all' => \$_fetch_all,
130 'parent|p' => \$_fetch_parent,
131 %fc_opts } ],
132 clone => [ \&cmd_clone, "Initialize and fetch revisions",
133 { 'revision|r=s' => \$_revision,
134 %fc_opts, %init_opts } ],
135 init => [ \&cmd_init, "Initialize a repo for tracking" .
136 " (requires URL argument)",
137 \%init_opts ],
138 'multi-init' => [ \&cmd_multi_init,
139 "Deprecated alias for ".
140 "'$0 init -T<trunk> -b<branches> -t<tags>'",
141 \%init_opts ],
142 dcommit => [ \&cmd_dcommit,
143 'Commit several diffs to merge with upstream',
144 { 'merge|m|M' => \$_merge,
145 'strategy|s=s' => \$_strategy,
146 'verbose|v' => \$_verbose,
147 'dry-run|n' => \$_dry_run,
148 'fetch-all|all' => \$_fetch_all,
149 'commit-url=s' => \$_commit_url,
150 'revision|r=i' => \$_revision,
151 'no-rebase' => \$_no_rebase,
152 %cmt_opts, %fc_opts } ],
153 branch => [ \&cmd_branch,
154 'Create a branch in the SVN repository',
155 { 'message|m=s' => \$_message,
156 'destination|d=s' => \$_branch_dest,
157 'dry-run|n' => \$_dry_run,
158 'tag|t' => \$_tag } ],
159 tag => [ sub { $_tag = 1; cmd_branch(@_) },
160 'Create a tag in the SVN repository',
161 { 'message|m=s' => \$_message,
162 'destination|d=s' => \$_branch_dest,
163 'dry-run|n' => \$_dry_run } ],
164 'set-tree' => [ \&cmd_set_tree,
165 "Set an SVN repository to a git tree-ish",
166 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
167 'create-ignore' => [ \&cmd_create_ignore,
168 'Create a .gitignore per svn:ignore',
169 { 'revision|r=i' => \$_revision
170 } ],
171 'mkdirs' => [ \&cmd_mkdirs ,
172 "recreate empty directories after a checkout",
173 { 'revision|r=i' => \$_revision } ],
174 'propget' => [ \&cmd_propget,
175 'Print the value of a property on a file or directory',
176 { 'revision|r=i' => \$_revision } ],
177 'proplist' => [ \&cmd_proplist,
178 'List all properties of a file or directory',
179 { 'revision|r=i' => \$_revision } ],
180 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
181 { 'revision|r=i' => \$_revision
182 } ],
183 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
184 { 'revision|r=i' => \$_revision
185 } ],
186 'multi-fetch' => [ \&cmd_multi_fetch,
187 "Deprecated alias for $0 fetch --all",
188 { 'revision|r=s' => \$_revision, %fc_opts } ],
189 'migrate' => [ sub { },
190 # no-op, we automatically run this anyways,
191 'Migrate configuration/metadata/layout from
192 previous versions of git-svn',
193 { 'minimize' => \$Git::SVN::Migration::_minimize,
194 %remote_opts } ],
195 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
196 { 'limit=i' => \$Git::SVN::Log::limit,
197 'revision|r=s' => \$_revision,
198 'verbose|v' => \$Git::SVN::Log::verbose,
199 'incremental' => \$Git::SVN::Log::incremental,
200 'oneline' => \$Git::SVN::Log::oneline,
201 'show-commit' => \$Git::SVN::Log::show_commit,
202 'non-recursive' => \$Git::SVN::Log::non_recursive,
203 'authors-file|A=s' => \$_authors,
204 'color' => \$Git::SVN::Log::color,
205 'pager=s' => \$Git::SVN::Log::pager
206 } ],
207 'find-rev' => [ \&cmd_find_rev,
208 "Translate between SVN revision numbers and tree-ish",
209 {} ],
210 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
211 { 'merge|m|M' => \$_merge,
212 'verbose|v' => \$_verbose,
213 'strategy|s=s' => \$_strategy,
214 'local|l' => \$_local,
215 'fetch-all|all' => \$_fetch_all,
216 'dry-run|n' => \$_dry_run,
217 %fc_opts } ],
218 'commit-diff' => [ \&cmd_commit_diff,
219 'Commit a diff between two trees',
220 { 'message|m=s' => \$_message,
221 'file|F=s' => \$_file,
222 'revision|r=s' => \$_revision,
223 %cmt_opts } ],
224 'info' => [ \&cmd_info,
225 "Show info about the latest SVN revision
226 on the current branch",
227 { 'url' => \$_url, } ],
228 'blame' => [ \&Git::SVN::Log::cmd_blame,
229 "Show what revision and author last modified each line of a file",
230 { 'git-format' => \$_git_format } ],
231 'reset' => [ \&cmd_reset,
232 "Undo fetches back to the specified SVN revision",
233 { 'revision|r=s' => \$_revision,
234 'parent|p' => \$_fetch_parent } ],
235 'gc' => [ \&cmd_gc,
236 "Compress unhandled.log files in .git/svn and remove " .
237 "index files in .git/svn",
238 {} ],
239 );
240
241 my $cmd;
242 for (my $i = 0; $i < @ARGV; $i++) {
243 if (defined $cmd{$ARGV[$i]}) {
244 $cmd = $ARGV[$i];
245 splice @ARGV, $i, 1;
246 last;
247 } elsif ($ARGV[$i] eq 'help') {
248 $cmd = $ARGV[$i+1];
249 usage(0);
250 }
251 };
252
253 # make sure we're always running at the top-level working directory
254 unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
255 unless (-d $ENV{GIT_DIR}) {
256 if ($git_dir_user_set) {
257 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
258 "but it is not a directory\n";
259 }
260 my $git_dir = delete $ENV{GIT_DIR};
261 my $cdup = undef;
262 git_cmd_try {
263 $cdup = command_oneline(qw/rev-parse --show-cdup/);
264 $git_dir = '.' unless ($cdup);
265 chomp $cdup if ($cdup);
266 $cdup = "." unless ($cdup && length $cdup);
267 } "Already at toplevel, but $git_dir not found\n";
268 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
269 unless (-d $git_dir) {
270 die "$git_dir still not found after going to ",
271 "'$cdup'\n";
272 }
273 $ENV{GIT_DIR} = $git_dir;
274 }
275 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
276 }
277
278 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
279
280 read_git_config(\%opts);
281 if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
282 Getopt::Long::Configure('pass_through');
283 }
284 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
285 'minimize-connections' => \$Git::SVN::Migration::_minimize,
286 'id|i=s' => \$Git::SVN::default_ref_id,
287 'svn-remote|remote|R=s' => sub {
288 $Git::SVN::no_reuse_existing = 1;
289 $Git::SVN::default_repo_id = $_[1] });
290 exit 1 if (!$rv && $cmd && $cmd ne 'log');
291
292 usage(0) if $_help;
293 version() if $_version;
294 usage(1) unless defined $cmd;
295 load_authors() if $_authors;
296 if (defined $_authors_prog) {
297 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
298 }
299
300 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
301 Git::SVN::Migration::migration_check();
302 }
303 Git::SVN::init_vars();
304 eval {
305 Git::SVN::verify_remotes_sanity();
306 $cmd{$cmd}->[0]->(@ARGV);
307 };
308 fatal $@ if $@;
309 post_fetch_checkout();
310 exit 0;
311
312 ####################### primary functions ######################
313 sub usage {
314 my $exit = shift || 0;
315 my $fd = $exit ? \*STDERR : \*STDOUT;
316 print $fd <<"";
317 git-svn - bidirectional operations between a single Subversion tree and git
318 Usage: git svn <command> [options] [arguments]\n
319
320 print $fd "Available commands:\n" unless $cmd;
321
322 foreach (sort keys %cmd) {
323 next if $cmd && $cmd ne $_;
324 next if /^multi-/; # don't show deprecated commands
325 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
326 foreach (sort keys %{$cmd{$_}->[2]}) {
327 # mixed-case options are for .git/config only
328 next if /[A-Z]/ && /^[a-z]+$/i;
329 # prints out arguments as they should be passed:
330 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
331 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
332 "--$_" : "-$_" }
333 split /\|/,$_)," $x\n";
334 }
335 }
336 print $fd <<"";
337 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
338 arbitrary identifier if you're tracking multiple SVN branches/repositories in
339 one git repository and want to keep them separate. See git-svn(1) for more
340 information.
341
342 exit $exit;
343 }
344
345 sub version {
346 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
347 exit 0;
348 }
349
350 sub do_git_init_db {
351 unless (-d $ENV{GIT_DIR}) {
352 my @init_db = ('init');
353 push @init_db, "--template=$_template" if defined $_template;
354 if (defined $_shared) {
355 if ($_shared =~ /[a-z]/) {
356 push @init_db, "--shared=$_shared";
357 } else {
358 push @init_db, "--shared";
359 }
360 }
361 command_noisy(@init_db);
362 $_repository = Git->repository(Repository => ".git");
363 }
364 command_noisy('config', 'core.autocrlf', 'false');
365 my $set;
366 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
367 foreach my $i (keys %icv) {
368 die "'$set' and '$i' cannot both be set\n" if $set;
369 next unless defined $icv{$i};
370 command_noisy('config', "$pfx.$i", $icv{$i});
371 $set = $i;
372 }
373 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
374 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
375 if defined $$ignore_regex;
376 }
377
378 sub init_subdir {
379 my $repo_path = shift or return;
380 mkpath([$repo_path]) unless -d $repo_path;
381 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
382 $ENV{GIT_DIR} = '.git';
383 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
384 }
385
386 sub cmd_clone {
387 my ($url, $path) = @_;
388 if (!defined $path &&
389 (defined $_trunk || @_branches || @_tags ||
390 defined $_stdlayout) &&
391 $url !~ m#^[a-z\+]+://#) {
392 $path = $url;
393 }
394 $path = basename($url) if !defined $path || !length $path;
395 cmd_init($url, $path);
396 Git::SVN::fetch_all($Git::SVN::default_repo_id);
397 command_oneline('config', 'svn.authorsfile', $_authors) if $_authors;
398 }
399
400 sub cmd_init {
401 if (defined $_stdlayout) {
402 $_trunk = 'trunk' if (!defined $_trunk);
403 @_tags = 'tags' if (! @_tags);
404 @_branches = 'branches' if (! @_branches);
405 }
406 if (defined $_trunk || @_branches || @_tags) {
407 return cmd_multi_init(@_);
408 }
409 my $url = shift or die "SVN repository location required ",
410 "as a command-line argument\n";
411 $url = canonicalize_url($url);
412 init_subdir(@_);
413 do_git_init_db();
414
415 if ($Git::SVN::_minimize_url eq 'unset') {
416 $Git::SVN::_minimize_url = 0;
417 }
418
419 Git::SVN->init($url);
420 }
421
422 sub cmd_fetch {
423 if (grep /^\d+=./, @_) {
424 die "'<rev>=<commit>' fetch arguments are ",
425 "no longer supported.\n";
426 }
427 my ($remote) = @_;
428 if (@_ > 1) {
429 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
430 }
431 $Git::SVN::no_reuse_existing = undef;
432 if ($_fetch_parent) {
433 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
434 unless ($gs) {
435 die "Unable to determine upstream SVN information from ",
436 "working tree history\n";
437 }
438 # just fetch, don't checkout.
439 $_no_checkout = 'true';
440 $_fetch_all ? $gs->fetch_all : $gs->fetch;
441 } elsif ($_fetch_all) {
442 cmd_multi_fetch();
443 } else {
444 $remote ||= $Git::SVN::default_repo_id;
445 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
446 }
447 }
448
449 sub cmd_set_tree {
450 my (@commits) = @_;
451 if ($_stdin || !@commits) {
452 print "Reading from stdin...\n";
453 @commits = ();
454 while (<STDIN>) {
455 if (/\b($sha1_short)\b/o) {
456 unshift @commits, $1;
457 }
458 }
459 }
460 my @revs;
461 foreach my $c (@commits) {
462 my @tmp = command('rev-parse',$c);
463 if (scalar @tmp == 1) {
464 push @revs, $tmp[0];
465 } elsif (scalar @tmp > 1) {
466 push @revs, reverse(command('rev-list',@tmp));
467 } else {
468 fatal "Failed to rev-parse $c";
469 }
470 }
471 my $gs = Git::SVN->new;
472 my ($r_last, $cmt_last) = $gs->last_rev_commit;
473 $gs->fetch;
474 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
475 fatal "There are new revisions that were fetched ",
476 "and need to be merged (or acknowledged) ",
477 "before committing.\nlast rev: $r_last\n",
478 " current: $gs->{last_rev}";
479 }
480 $gs->set_tree($_) foreach @revs;
481 print "Done committing ",scalar @revs," revisions to SVN\n";
482 unlink $gs->{index};
483 }
484
485 sub cmd_dcommit {
486 my $head = shift;
487 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
488 'Cannot dcommit with a dirty index. Commit your changes first, '
489 . "or stash them with `git stash'.\n";
490 $head ||= 'HEAD';
491
492 my $old_head;
493 if ($head ne 'HEAD') {
494 $old_head = eval {
495 command_oneline([qw/symbolic-ref -q HEAD/])
496 };
497 if ($old_head) {
498 $old_head =~ s{^refs/heads/}{};
499 } else {
500 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
501 }
502 command(['checkout', $head], STDERR => 0);
503 }
504
505 my @refs;
506 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
507 unless ($gs) {
508 die "Unable to determine upstream SVN information from ",
509 "$head history.\nPerhaps the repository is empty.";
510 }
511
512 if (defined $_commit_url) {
513 $url = $_commit_url;
514 } else {
515 $url = eval { command_oneline('config', '--get',
516 "svn-remote.$gs->{repo_id}.commiturl") };
517 if (!$url) {
518 $url = $gs->full_url
519 }
520 }
521
522 my $last_rev = $_revision if defined $_revision;
523 if ($url) {
524 print "Committing to $url ...\n";
525 }
526 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
527 if ($_no_rebase && scalar(@$linear_refs) > 1) {
528 warn "Attempting to commit more than one change while ",
529 "--no-rebase is enabled.\n",
530 "If these changes depend on each other, re-running ",
531 "without --no-rebase may be required."
532 }
533 my $expect_url = $url;
534 Git::SVN::remove_username($expect_url);
535 while (1) {
536 my $d = shift @$linear_refs or last;
537 unless (defined $last_rev) {
538 (undef, $last_rev, undef) = cmt_metadata("$d~1");
539 unless (defined $last_rev) {
540 fatal "Unable to extract revision information ",
541 "from commit $d~1";
542 }
543 }
544 if ($_dry_run) {
545 print "diff-tree $d~1 $d\n";
546 } else {
547 my $cmt_rev;
548 my %ed_opts = ( r => $last_rev,
549 log => get_commit_entry($d)->{log},
550 ra => Git::SVN::Ra->new($url),
551 config => SVN::Core::config_get_config(
552 $Git::SVN::Ra::config_dir
553 ),
554 tree_a => "$d~1",
555 tree_b => $d,
556 editor_cb => sub {
557 print "Committed r$_[0]\n";
558 $cmt_rev = $_[0];
559 },
560 svn_path => '');
561 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
562 print "No changes\n$d~1 == $d\n";
563 } elsif ($parents->{$d} && @{$parents->{$d}}) {
564 $gs->{inject_parents_dcommit}->{$cmt_rev} =
565 $parents->{$d};
566 }
567 $_fetch_all ? $gs->fetch_all : $gs->fetch;
568 $last_rev = $cmt_rev;
569 next if $_no_rebase;
570
571 # we always want to rebase against the current HEAD,
572 # not any head that was passed to us
573 my @diff = command('diff-tree', $d,
574 $gs->refname, '--');
575 my @finish;
576 if (@diff) {
577 @finish = rebase_cmd();
578 print STDERR "W: $d and ", $gs->refname,
579 " differ, using @finish:\n",
580 join("\n", @diff), "\n";
581 } else {
582 print "No changes between current HEAD and ",
583 $gs->refname,
584 "\nResetting to the latest ",
585 $gs->refname, "\n";
586 @finish = qw/reset --mixed/;
587 }
588 command_noisy(@finish, $gs->refname);
589 if (@diff) {
590 @refs = ();
591 my ($url_, $rev_, $uuid_, $gs_) =
592 working_head_info('HEAD', \@refs);
593 my ($linear_refs_, $parents_) =
594 linearize_history($gs_, \@refs);
595 if (scalar(@$linear_refs) !=
596 scalar(@$linear_refs_)) {
597 fatal "# of revisions changed ",
598 "\nbefore:\n",
599 join("\n", @$linear_refs),
600 "\n\nafter:\n",
601 join("\n", @$linear_refs_), "\n",
602 'If you are attempting to commit ',
603 "merges, try running:\n\t",
604 'git rebase --interactive',
605 '--preserve-merges ',
606 $gs->refname,
607 "\nBefore dcommitting";
608 }
609 if ($url_ ne $expect_url) {
610 if ($url_ eq $gs->metadata_url) {
611 print
612 "Accepting rewritten URL:",
613 " $url_\n";
614 } else {
615 fatal
616 "URL mismatch after rebase:",
617 " $url_ != $expect_url";
618 }
619 }
620 if ($uuid_ ne $uuid) {
621 fatal "uuid mismatch after rebase: ",
622 "$uuid_ != $uuid";
623 }
624 # remap parents
625 my (%p, @l, $i);
626 for ($i = 0; $i < scalar @$linear_refs; $i++) {
627 my $new = $linear_refs_->[$i] or next;
628 $p{$new} =
629 $parents->{$linear_refs->[$i]};
630 push @l, $new;
631 }
632 $parents = \%p;
633 $linear_refs = \@l;
634 }
635 }
636 }
637
638 if ($old_head) {
639 my $new_head = command_oneline(qw/rev-parse HEAD/);
640 my $new_is_symbolic = eval {
641 command_oneline(qw/symbolic-ref -q HEAD/);
642 };
643 if ($new_is_symbolic) {
644 print "dcommitted the branch ", $head, "\n";
645 } else {
646 print "dcommitted on a detached HEAD because you gave ",
647 "a revision argument.\n",
648 "The rewritten commit is: ", $new_head, "\n";
649 }
650 command(['checkout', $old_head], STDERR => 0);
651 }
652
653 unlink $gs->{index};
654 }
655
656 sub cmd_branch {
657 my ($branch_name, $head) = @_;
658
659 unless (defined $branch_name && length $branch_name) {
660 die(($_tag ? "tag" : "branch") . " name required\n");
661 }
662 $head ||= 'HEAD';
663
664 my ($src, $rev, undef, $gs) = working_head_info($head);
665
666 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
667 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
668 my $glob;
669 if ($#{$allglobs} == 0) {
670 $glob = $allglobs->[0];
671 } else {
672 unless(defined $_branch_dest) {
673 die "Multiple ",
674 $_tag ? "tag" : "branch",
675 " paths defined for Subversion repository.\n",
676 "You must specify where you want to create the ",
677 $_tag ? "tag" : "branch",
678 " with the --destination argument.\n";
679 }
680 foreach my $g (@{$allglobs}) {
681 # SVN::Git::Editor could probably be moved to Git.pm..
682 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
683 if ($_branch_dest =~ /$re/) {
684 $glob = $g;
685 last;
686 }
687 }
688 unless (defined $glob) {
689 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
690 foreach my $g (@{$allglobs}) {
691 $g->{path}->{left} =~ /$dest_re/ or next;
692 if (defined $glob) {
693 die "Ambiguous destination: ",
694 $_branch_dest, "\nmatches both '",
695 $glob->{path}->{left}, "' and '",
696 $g->{path}->{left}, "'\n";
697 }
698 $glob = $g;
699 }
700 unless (defined $glob) {
701 die "Unknown ",
702 $_tag ? "tag" : "branch",
703 " destination $_branch_dest\n";
704 }
705 }
706 }
707 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
708 my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
709
710 my $ctx = SVN::Client->new(
711 auth => Git::SVN::Ra::_auth_providers(),
712 log_msg => sub {
713 ${ $_[0] } = defined $_message
714 ? $_message
715 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
716 . $branch_name;
717 },
718 );
719
720 eval {
721 $ctx->ls($dst, 'HEAD', 0);
722 } and die "branch ${branch_name} already exists\n";
723
724 print "Copying ${src} at r${rev} to ${dst}...\n";
725 $ctx->copy($src, $rev, $dst)
726 unless $_dry_run;
727
728 $gs->fetch_all;
729 }
730
731 sub cmd_find_rev {
732 my $revision_or_hash = shift or die "SVN or git revision required ",
733 "as a command-line argument\n";
734 my $result;
735 if ($revision_or_hash =~ /^r\d+$/) {
736 my $head = shift;
737 $head ||= 'HEAD';
738 my @refs;
739 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
740 unless ($gs) {
741 die "Unable to determine upstream SVN information from ",
742 "$head history\n";
743 }
744 my $desired_revision = substr($revision_or_hash, 1);
745 $result = $gs->rev_map_get($desired_revision, $uuid);
746 } else {
747 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
748 $result = $rev;
749 }
750 print "$result\n" if $result;
751 }
752
753 sub cmd_rebase {
754 command_noisy(qw/update-index --refresh/);
755 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
756 unless ($gs) {
757 die "Unable to determine upstream SVN information from ",
758 "working tree history\n";
759 }
760 if ($_dry_run) {
761 print "Remote Branch: " . $gs->refname . "\n";
762 print "SVN URL: " . $url . "\n";
763 return;
764 }
765 if (command(qw/diff-index HEAD --/)) {
766 print STDERR "Cannot rebase with uncommited changes:\n";
767 command_noisy('status');
768 exit 1;
769 }
770 unless ($_local) {
771 # rebase will checkout for us, so no need to do it explicitly
772 $_no_checkout = 'true';
773 $_fetch_all ? $gs->fetch_all : $gs->fetch;
774 }
775 command_noisy(rebase_cmd(), $gs->refname);
776 $gs->mkemptydirs;
777 }
778
779 sub cmd_show_ignore {
780 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
781 $gs ||= Git::SVN->new;
782 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
783 $gs->prop_walk($gs->{path}, $r, sub {
784 my ($gs, $path, $props) = @_;
785 print STDOUT "\n# $path\n";
786 my $s = $props->{'svn:ignore'} or return;
787 $s =~ s/[\r\n]+/\n/g;
788 $s =~ s/^\n+//;
789 chomp $s;
790 $s =~ s#^#$path#gm;
791 print STDOUT "$s\n";
792 });
793 }
794
795 sub cmd_show_externals {
796 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
797 $gs ||= Git::SVN->new;
798 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
799 $gs->prop_walk($gs->{path}, $r, sub {
800 my ($gs, $path, $props) = @_;
801 print STDOUT "\n# $path\n";
802 my $s = $props->{'svn:externals'} or return;
803 $s =~ s/[\r\n]+/\n/g;
804 chomp $s;
805 $s =~ s#^#$path#gm;
806 print STDOUT "$s\n";
807 });
808 }
809
810 sub cmd_create_ignore {
811 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
812 $gs ||= Git::SVN->new;
813 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
814 $gs->prop_walk($gs->{path}, $r, sub {
815 my ($gs, $path, $props) = @_;
816 # $path is of the form /path/to/dir/
817 $path = '.' . $path;
818 # SVN can have attributes on empty directories,
819 # which git won't track
820 mkpath([$path]) unless -d $path;
821 my $ignore = $path . '.gitignore';
822 my $s = $props->{'svn:ignore'} or return;
823 open(GITIGNORE, '>', $ignore)
824 or fatal("Failed to open `$ignore' for writing: $!");
825 $s =~ s/[\r\n]+/\n/g;
826 $s =~ s/^\n+//;
827 chomp $s;
828 # Prefix all patterns so that the ignore doesn't apply
829 # to sub-directories.
830 $s =~ s#^#/#gm;
831 print GITIGNORE "$s\n";
832 close(GITIGNORE)
833 or fatal("Failed to close `$ignore': $!");
834 command_noisy('add', '-f', $ignore);
835 });
836 }
837
838 sub cmd_mkdirs {
839 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
840 $gs ||= Git::SVN->new;
841 $gs->mkemptydirs($_revision);
842 }
843
844 sub canonicalize_path {
845 my ($path) = @_;
846 my $dot_slash_added = 0;
847 if (substr($path, 0, 1) ne "/") {
848 $path = "./" . $path;
849 $dot_slash_added = 1;
850 }
851 # File::Spec->canonpath doesn't collapse x/../y into y (for a
852 # good reason), so let's do this manually.
853 $path =~ s#/+#/#g;
854 $path =~ s#/\.(?:/|$)#/#g;
855 $path =~ s#/[^/]+/\.\.##g;
856 $path =~ s#/$##g;
857 $path =~ s#^\./## if $dot_slash_added;
858 $path =~ s#^/##;
859 $path =~ s#^\.$##;
860 return $path;
861 }
862
863 sub canonicalize_url {
864 my ($url) = @_;
865 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
866 return $url;
867 }
868
869 # get_svnprops(PATH)
870 # ------------------
871 # Helper for cmd_propget and cmd_proplist below.
872 sub get_svnprops {
873 my $path = shift;
874 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
875 $gs ||= Git::SVN->new;
876
877 # prefix THE PATH by the sub-directory from which the user
878 # invoked us.
879 $path = $cmd_dir_prefix . $path;
880 fatal("No such file or directory: $path") unless -e $path;
881 my $is_dir = -d $path ? 1 : 0;
882 $path = $gs->{path} . '/' . $path;
883
884 # canonicalize the path (otherwise libsvn will abort or fail to
885 # find the file)
886 $path = canonicalize_path($path);
887
888 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
889 my $props;
890 if ($is_dir) {
891 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
892 }
893 else {
894 (undef, $props) = $gs->ra->get_file($path, $r, undef);
895 }
896 return $props;
897 }
898
899 # cmd_propget (PROP, PATH)
900 # ------------------------
901 # Print the SVN property PROP for PATH.
902 sub cmd_propget {
903 my ($prop, $path) = @_;
904 $path = '.' if not defined $path;
905 usage(1) if not defined $prop;
906 my $props = get_svnprops($path);
907 if (not defined $props->{$prop}) {
908 fatal("`$path' does not have a `$prop' SVN property.");
909 }
910 print $props->{$prop} . "\n";
911 }
912
913 # cmd_proplist (PATH)
914 # -------------------
915 # Print the list of SVN properties for PATH.
916 sub cmd_proplist {
917 my $path = shift;
918 $path = '.' if not defined $path;
919 my $props = get_svnprops($path);
920 print "Properties on '$path':\n";
921 foreach (sort keys %{$props}) {
922 print " $_\n";
923 }
924 }
925
926 sub cmd_multi_init {
927 my $url = shift;
928 unless (defined $_trunk || @_branches || @_tags) {
929 usage(1);
930 }
931
932 $_prefix = '' unless defined $_prefix;
933 if (defined $url) {
934 $url = canonicalize_url($url);
935 init_subdir(@_);
936 }
937 do_git_init_db();
938 if (defined $_trunk) {
939 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
940 # try both old-style and new-style lookups:
941 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
942 unless ($gs_trunk) {
943 my ($trunk_url, $trunk_path) =
944 complete_svn_url($url, $_trunk);
945 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
946 undef, $trunk_ref);
947 }
948 }
949 return unless @_branches || @_tags;
950 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
951 foreach my $path (@_branches) {
952 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
953 }
954 foreach my $path (@_tags) {
955 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
956 }
957 }
958
959 sub cmd_multi_fetch {
960 $Git::SVN::no_reuse_existing = undef;
961 my $remotes = Git::SVN::read_all_remotes();
962 foreach my $repo_id (sort keys %$remotes) {
963 if ($remotes->{$repo_id}->{url}) {
964 Git::SVN::fetch_all($repo_id, $remotes);
965 }
966 }
967 }
968
969 # this command is special because it requires no metadata
970 sub cmd_commit_diff {
971 my ($ta, $tb, $url) = @_;
972 my $usage = "Usage: $0 commit-diff -r<revision> ".
973 "<tree-ish> <tree-ish> [<URL>]";
974 fatal($usage) if (!defined $ta || !defined $tb);
975 my $svn_path = '';
976 if (!defined $url) {
977 my $gs = eval { Git::SVN->new };
978 if (!$gs) {
979 fatal("Needed URL or usable git-svn --id in ",
980 "the command-line\n", $usage);
981 }
982 $url = $gs->{url};
983 $svn_path = $gs->{path};
984 }
985 unless (defined $_revision) {
986 fatal("-r|--revision is a required argument\n", $usage);
987 }
988 if (defined $_message && defined $_file) {
989 fatal("Both --message/-m and --file/-F specified ",
990 "for the commit message.\n",
991 "I have no idea what you mean");
992 }
993 if (defined $_file) {
994 $_message = file_to_s($_file);
995 } else {
996 $_message ||= get_commit_entry($tb)->{log};
997 }
998 my $ra ||= Git::SVN::Ra->new($url);
999 my $r = $_revision;
1000 if ($r eq 'HEAD') {
1001 $r = $ra->get_latest_revnum;
1002 } elsif ($r !~ /^\d+$/) {
1003 die "revision argument: $r not understood by git-svn\n";
1004 }
1005 my %ed_opts = ( r => $r,
1006 log => $_message,
1007 ra => $ra,
1008 tree_a => $ta,
1009 tree_b => $tb,
1010 editor_cb => sub { print "Committed r$_[0]\n" },
1011 svn_path => $svn_path );
1012 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1013 print "No changes\n$ta == $tb\n";
1014 }
1015 }
1016
1017 sub escape_uri_only {
1018 my ($uri) = @_;
1019 my @tmp;
1020 foreach (split m{/}, $uri) {
1021 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1022 push @tmp, $_;
1023 }
1024 join('/', @tmp);
1025 }
1026
1027 sub escape_url {
1028 my ($url) = @_;
1029 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1030 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1031 $url = "$scheme://$domain$uri";
1032 }
1033 $url;
1034 }
1035
1036 sub cmd_info {
1037 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1038 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1039 if (exists $_[1]) {
1040 die "Too many arguments specified\n";
1041 }
1042
1043 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1044
1045 if (!$file_type && !$diff_status) {
1046 print STDERR "svn: '$path' is not under version control\n";
1047 exit 1;
1048 }
1049
1050 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1051 unless ($gs) {
1052 die "Unable to determine upstream SVN information from ",
1053 "working tree history\n";
1054 }
1055
1056 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1057 $path = "." if $path eq "";
1058
1059 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1060
1061 if ($_url) {
1062 print escape_url($full_url), "\n";
1063 return;
1064 }
1065
1066 my $result = "Path: $path\n";
1067 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1068 $result .= "URL: " . escape_url($full_url) . "\n";
1069
1070 eval {
1071 my $repos_root = $gs->repos_root;
1072 Git::SVN::remove_username($repos_root);
1073 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1074 };
1075 if ($@) {
1076 $result .= "Repository Root: (offline)\n";
1077 }
1078 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1079 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1080 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1081
1082 $result .= "Node Kind: " .
1083 ($file_type eq "dir" ? "directory" : "file") . "\n";
1084
1085 my $schedule = $diff_status eq "A"
1086 ? "add"
1087 : ($diff_status eq "D" ? "delete" : "normal");
1088 $result .= "Schedule: $schedule\n";
1089
1090 if ($diff_status eq "A") {
1091 print $result, "\n";
1092 return;
1093 }
1094
1095 my ($lc_author, $lc_rev, $lc_date_utc);
1096 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1097 my $log = command_output_pipe(@args);
1098 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1099 while (<$log>) {
1100 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1101 $lc_author = $1;
1102 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1103 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1104 (undef, $lc_rev, undef) = ::extract_metadata($1);
1105 }
1106 }
1107 close $log;
1108
1109 Git::SVN::Log::set_local_timezone();
1110
1111 $result .= "Last Changed Author: $lc_author\n";
1112 $result .= "Last Changed Rev: $lc_rev\n";
1113 $result .= "Last Changed Date: " .
1114 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1115
1116 if ($file_type ne "dir") {
1117 my $text_last_updated_date =
1118 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1119 $result .=
1120 "Text Last Updated: " .
1121 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1122 "\n";
1123 my $checksum;
1124 if ($diff_status eq "D") {
1125 my ($fh, $ctx) =
1126 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1127 if ($file_type eq "link") {
1128 my $file_name = <$fh>;
1129 $checksum = md5sum("link $file_name");
1130 } else {
1131 $checksum = md5sum($fh);
1132 }
1133 command_close_pipe($fh, $ctx);
1134 } elsif ($file_type eq "link") {
1135 my $file_name =
1136 command(qw(cat-file blob), "HEAD:$path");
1137 $checksum =
1138 md5sum("link " . $file_name);
1139 } else {
1140 open FILE, "<", $path or die $!;
1141 $checksum = md5sum(\*FILE);
1142 close FILE or die $!;
1143 }
1144 $result .= "Checksum: " . $checksum . "\n";
1145 }
1146
1147 print $result, "\n";
1148 }
1149
1150 sub cmd_reset {
1151 my $target = shift || $_revision or die "SVN revision required\n";
1152 $target = $1 if $target =~ /^r(\d+)$/;
1153 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1154 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1155 unless ($gs) {
1156 die "Unable to determine upstream SVN information from ".
1157 "history\n";
1158 }
1159 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1160 $gs->rev_map_set($r, $c, 'reset', $uuid);
1161 print "r$r = $c ($gs->{ref_id})\n";
1162 }
1163
1164 sub cmd_gc {
1165 if (!$can_compress) {
1166 warn "Compress::Zlib could not be found; unhandled.log " .
1167 "files will not be compressed.\n";
1168 }
1169 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1170 }
1171
1172 ########################### utility functions #########################
1173
1174 sub rebase_cmd {
1175 my @cmd = qw/rebase/;
1176 push @cmd, '-v' if $_verbose;
1177 push @cmd, qw/--merge/ if $_merge;
1178 push @cmd, "--strategy=$_strategy" if $_strategy;
1179 @cmd;
1180 }
1181
1182 sub post_fetch_checkout {
1183 return if $_no_checkout;
1184 my $gs = $Git::SVN::_head or return;
1185 return if verify_ref('refs/heads/master^0');
1186
1187 # look for "trunk" ref if it exists
1188 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1189 my $fetch = $remote->{fetch};
1190 if ($fetch) {
1191 foreach my $p (keys %$fetch) {
1192 basename($fetch->{$p}) eq 'trunk' or next;
1193 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1194 last;
1195 }
1196 }
1197
1198 my $valid_head = verify_ref('HEAD^0');
1199 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1200 return if ($valid_head || !verify_ref('HEAD^0'));
1201
1202 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1203 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1204 return if -f $index;
1205
1206 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1207 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1208 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1209 print STDERR "Checked out HEAD:\n ",
1210 $gs->full_url, " r", $gs->last_rev, "\n";
1211 $gs->mkemptydirs($gs->last_rev);
1212 }
1213
1214 sub complete_svn_url {
1215 my ($url, $path) = @_;
1216 $path =~ s#/+$##;
1217 if ($path !~ m#^[a-z\+]+://#) {
1218 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1219 fatal("E: '$path' is not a complete URL ",
1220 "and a separate URL is not specified");
1221 }
1222 return ($url, $path);
1223 }
1224 return ($path, '');
1225 }
1226
1227 sub complete_url_ls_init {
1228 my ($ra, $repo_path, $switch, $pfx) = @_;
1229 unless ($repo_path) {
1230 print STDERR "W: $switch not specified\n";
1231 return;
1232 }
1233 $repo_path =~ s#/+$##;
1234 if ($repo_path =~ m#^[a-z\+]+://#) {
1235 $ra = Git::SVN::Ra->new($repo_path);
1236 $repo_path = '';
1237 } else {
1238 $repo_path =~ s#^/+##;
1239 unless ($ra) {
1240 fatal("E: '$repo_path' is not a complete URL ",
1241 "and a separate URL is not specified");
1242 }
1243 }
1244 my $url = $ra->{url};
1245 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1246 my $k = "svn-remote.$gs->{repo_id}.url";
1247 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1248 if ($orig_url && ($orig_url ne $gs->{url})) {
1249 die "$k already set: $orig_url\n",
1250 "wanted to set to: $gs->{url}\n";
1251 }
1252 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1253 my $remote_path = "$gs->{path}/$repo_path";
1254 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1255 $remote_path =~ s#/+#/#g;
1256 $remote_path =~ s#^/##g;
1257 $remote_path .= "/*" if $remote_path !~ /\*/;
1258 my ($n) = ($switch =~ /^--(\w+)/);
1259 if (length $pfx && $pfx !~ m#/$#) {
1260 die "--prefix='$pfx' must have a trailing slash '/'\n";
1261 }
1262 command_noisy('config',
1263 '--add',
1264 "svn-remote.$gs->{repo_id}.$n",
1265 "$remote_path:refs/remotes/$pfx*" .
1266 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1267 }
1268
1269 sub verify_ref {
1270 my ($ref) = @_;
1271 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1272 { STDERR => 0 }); };
1273 }
1274
1275 sub get_tree_from_treeish {
1276 my ($treeish) = @_;
1277 # $treeish can be a symbolic ref, too:
1278 my $type = command_oneline(qw/cat-file -t/, $treeish);
1279 my $expected;
1280 while ($type eq 'tag') {
1281 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1282 }
1283 if ($type eq 'commit') {
1284 $expected = (grep /^tree /, command(qw/cat-file commit/,
1285 $treeish))[0];
1286 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1287 die "Unable to get tree from $treeish\n" unless $expected;
1288 } elsif ($type eq 'tree') {
1289 $expected = $treeish;
1290 } else {
1291 die "$treeish is a $type, expected tree, tag or commit\n";
1292 }
1293 return $expected;
1294 }
1295
1296 sub get_commit_entry {
1297 my ($treeish) = shift;
1298 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1299 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1300 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1301 open my $log_fh, '>', $commit_editmsg or croak $!;
1302
1303 my $type = command_oneline(qw/cat-file -t/, $treeish);
1304 if ($type eq 'commit' || $type eq 'tag') {
1305 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1306 $type, $treeish);
1307 my $in_msg = 0;
1308 my $author;
1309 my $saw_from = 0;
1310 my $msgbuf = "";
1311 while (<$msg_fh>) {
1312 if (!$in_msg) {
1313 $in_msg = 1 if (/^\s*$/);
1314 $author = $1 if (/^author (.*>)/);
1315 } elsif (/^git-svn-id: /) {
1316 # skip this for now, we regenerate the
1317 # correct one on re-fetch anyways
1318 # TODO: set *:merge properties or like...
1319 } else {
1320 if (/^From:/ || /^Signed-off-by:/) {
1321 $saw_from = 1;
1322 }
1323 $msgbuf .= $_;
1324 }
1325 }
1326 $msgbuf =~ s/\s+$//s;
1327 if ($Git::SVN::_add_author_from && defined($author)
1328 && !$saw_from) {
1329 $msgbuf .= "\n\nFrom: $author";
1330 }
1331 print $log_fh $msgbuf or croak $!;
1332 command_close_pipe($msg_fh, $ctx);
1333 }
1334 close $log_fh or croak $!;
1335
1336 if ($_edit || ($type eq 'tree')) {
1337 chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1338 system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1339 }
1340 rename $commit_editmsg, $commit_msg or croak $!;
1341 {
1342 require Encode;
1343 # SVN requires messages to be UTF-8 when entering the repo
1344 local $/;
1345 open $log_fh, '<', $commit_msg or croak $!;
1346 binmode $log_fh;
1347 chomp($log_entry{log} = <$log_fh>);
1348
1349 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1350 my $msg = $log_entry{log};
1351
1352 eval { $msg = Encode::decode($enc, $msg, 1) };
1353 if ($@) {
1354 die "Could not decode as $enc:\n", $msg,
1355 "\nPerhaps you need to set i18n.commitencoding\n";
1356 }
1357
1358 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1359 die "Could not encode as UTF-8:\n$msg\n" if $@;
1360
1361 $log_entry{log} = $msg;
1362
1363 close $log_fh or croak $!;
1364 }
1365 unlink $commit_msg;
1366 \%log_entry;
1367 }
1368
1369 sub s_to_file {
1370 my ($str, $file, $mode) = @_;
1371 open my $fd,'>',$file or croak $!;
1372 print $fd $str,"\n" or croak $!;
1373 close $fd or croak $!;
1374 chmod ($mode &~ umask, $file) if (defined $mode);
1375 }
1376
1377 sub file_to_s {
1378 my $file = shift;
1379 open my $fd,'<',$file or croak "$!: file: $file\n";
1380 local $/;
1381 my $ret = <$fd>;
1382 close $fd or croak $!;
1383 $ret =~ s/\s*$//s;
1384 return $ret;
1385 }
1386
1387 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1388 sub load_authors {
1389 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1390 my $log = $cmd eq 'log';
1391 while (<$authors>) {
1392 chomp;
1393 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1394 my ($user, $name, $email) = ($1, $2, $3);
1395 if ($log) {
1396 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1397 } else {
1398 $users{$user} = [$name, $email];
1399 }
1400 }
1401 close $authors or croak $!;
1402 }
1403
1404 # convert GetOpt::Long specs for use by git-config
1405 sub read_git_config {
1406 my $opts = shift;
1407 my @config_only;
1408 foreach my $o (keys %$opts) {
1409 # if we have mixedCase and a long option-only, then
1410 # it's a config-only variable that we don't need for
1411 # the command-line.
1412 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1413 my $v = $opts->{$o};
1414 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1415 $key =~ s/-//g;
1416 my $arg = 'git config';
1417 $arg .= ' --int' if ($o =~ /[:=]i$/);
1418 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1419 if (ref $v eq 'ARRAY') {
1420 chomp(my @tmp = `$arg --get-all svn.$key`);
1421 @$v = @tmp if @tmp;
1422 } else {
1423 chomp(my $tmp = `$arg --get svn.$key`);
1424 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1425 $$v = $tmp;
1426 }
1427 }
1428 }
1429 delete @$opts{@config_only} if @config_only;
1430 }
1431
1432 sub extract_metadata {
1433 my $id = shift or return (undef, undef, undef);
1434 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1435 \s([a-f\d\-]+)$/ix);
1436 if (!defined $rev || !$uuid || !$url) {
1437 # some of the original repositories I made had
1438 # identifiers like this:
1439 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1440 }
1441 return ($url, $rev, $uuid);
1442 }
1443
1444 sub cmt_metadata {
1445 return extract_metadata((grep(/^git-svn-id: /,
1446 command(qw/cat-file commit/, shift)))[-1]);
1447 }
1448
1449 sub cmt_sha2rev_batch {
1450 my %s2r;
1451 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1452 my $list = shift;
1453
1454 foreach my $sha (@{$list}) {
1455 my $first = 1;
1456 my $size = 0;
1457 print $out $sha, "\n";
1458
1459 while (my $line = <$in>) {
1460 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1461 last;
1462 } elsif ($first &&
1463 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1464 $first = 0;
1465 $size = $1;
1466 next;
1467 } elsif ($line =~ /^(git-svn-id: )/) {
1468 my (undef, $rev, undef) =
1469 extract_metadata($line);
1470 $s2r{$sha} = $rev;
1471 }
1472
1473 $size -= length($line);
1474 last if ($size == 0);
1475 }
1476 }
1477
1478 command_close_bidi_pipe($pid, $in, $out, $ctx);
1479
1480 return \%s2r;
1481 }
1482
1483 sub working_head_info {
1484 my ($head, $refs) = @_;
1485 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1486 my ($fh, $ctx) = command_output_pipe(@args, $head);
1487 my $hash;
1488 my %max;
1489 while (<$fh>) {
1490 if ( m{^commit ($::sha1)$} ) {
1491 unshift @$refs, $hash if $hash and $refs;
1492 $hash = $1;
1493 next;
1494 }
1495 next unless s{^\s*(git-svn-id:)}{$1};
1496 my ($url, $rev, $uuid) = extract_metadata($_);
1497 if (defined $url && defined $rev) {
1498 next if $max{$url} and $max{$url} < $rev;
1499 if (my $gs = Git::SVN->find_by_url($url)) {
1500 my $c = $gs->rev_map_get($rev, $uuid);
1501 if ($c && $c eq $hash) {
1502 close $fh; # break the pipe
1503 return ($url, $rev, $uuid, $gs);
1504 } else {
1505 $max{$url} ||= $gs->rev_map_max;
1506 }
1507 }
1508 }
1509 }
1510 command_close_pipe($fh, $ctx);
1511 (undef, undef, undef, undef);
1512 }
1513
1514 sub read_commit_parents {
1515 my ($parents, $c) = @_;
1516 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1517 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1518 @{$parents->{$c}} = split(/ /, $p);
1519 }
1520
1521 sub linearize_history {
1522 my ($gs, $refs) = @_;
1523 my %parents;
1524 foreach my $c (@$refs) {
1525 read_commit_parents(\%parents, $c);
1526 }
1527
1528 my @linear_refs;
1529 my %skip = ();
1530 my $last_svn_commit = $gs->last_commit;
1531 foreach my $c (reverse @$refs) {
1532 next if $c eq $last_svn_commit;
1533 last if $skip{$c};
1534
1535 unshift @linear_refs, $c;
1536 $skip{$c} = 1;
1537
1538 # we only want the first parent to diff against for linear
1539 # history, we save the rest to inject when we finalize the
1540 # svn commit
1541 my $fp_a = verify_ref("$c~1");
1542 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1543 if (!$fp_a || !$fp_b) {
1544 die "Commit $c\n",
1545 "has no parent commit, and therefore ",
1546 "nothing to diff against.\n",
1547 "You should be working from a repository ",
1548 "originally created by git-svn\n";
1549 }
1550 if ($fp_a ne $fp_b) {
1551 die "$c~1 = $fp_a, however parsing commit $c ",
1552 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1553 }
1554
1555 foreach my $p (@{$parents{$c}}) {
1556 $skip{$p} = 1;
1557 }
1558 }
1559 (\@linear_refs, \%parents);
1560 }
1561
1562 sub find_file_type_and_diff_status {
1563 my ($path) = @_;
1564 return ('dir', '') if $path eq '';
1565
1566 my $diff_output =
1567 command_oneline(qw(diff --cached --name-status --), $path) || "";
1568 my $diff_status = (split(' ', $diff_output))[0] || "";
1569
1570 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1571
1572 return (undef, undef) if !$diff_status && !$ls_tree;
1573
1574 if ($diff_status eq "A") {
1575 return ("link", $diff_status) if -l $path;
1576 return ("dir", $diff_status) if -d $path;
1577 return ("file", $diff_status);
1578 }
1579
1580 my $mode = (split(' ', $ls_tree))[0] || "";
1581
1582 return ("link", $diff_status) if $mode eq "120000";
1583 return ("dir", $diff_status) if $mode eq "040000";
1584 return ("file", $diff_status);
1585 }
1586
1587 sub md5sum {
1588 my $arg = shift;
1589 my $ref = ref $arg;
1590 my $md5 = Digest::MD5->new();
1591 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1592 $md5->addfile($arg) or croak $!;
1593 } elsif ($ref eq 'SCALAR') {
1594 $md5->add($$arg) or croak $!;
1595 } elsif (!$ref) {
1596 $md5->add($arg) or croak $!;
1597 } else {
1598 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1599 }
1600 return $md5->hexdigest();
1601 }
1602
1603 sub gc_directory {
1604 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1605 my $out_filename = $_ . ".gz";
1606 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1607 binmode $in_fh;
1608 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1609 die "Unable to open $out_filename: $!\n";
1610
1611 my $res;
1612 while ($res = sysread($in_fh, my $str, 1024)) {
1613 $gz->gzwrite($str) or
1614 die "Unable to write: ".$gz->gzerror()."!\n";
1615 }
1616 unlink $_ or die "unlink $File::Find::name: $!\n";
1617 } elsif (-f $_ && basename($_) eq "index") {
1618 unlink $_ or die "unlink $_: $!\n";
1619 }
1620 }
1621
1622 package Git::SVN;
1623 use strict;
1624 use warnings;
1625 use Fcntl qw/:DEFAULT :seek/;
1626 use constant rev_map_fmt => 'NH40';
1627 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1628 $_repack $_repack_flags $_use_svm_props $_head
1629 $_use_svnsync_props $no_reuse_existing $_minimize_url
1630 $_use_log_author $_add_author_from $_localtime/;
1631 use Carp qw/croak/;
1632 use File::Path qw/mkpath/;
1633 use File::Copy qw/copy/;
1634 use IPC::Open3;
1635
1636 my ($_gc_nr, $_gc_period);
1637
1638 # properties that we do not log:
1639 my %SKIP_PROP;
1640 BEGIN {
1641 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1642 svn:special svn:executable
1643 svn:entry:committed-rev
1644 svn:entry:last-author
1645 svn:entry:uuid
1646 svn:entry:committed-date/;
1647
1648 # some options are read globally, but can be overridden locally
1649 # per [svn-remote "..."] section. Command-line options will *NOT*
1650 # override options set in an [svn-remote "..."] section
1651 no strict 'refs';
1652 for my $option (qw/follow_parent no_metadata use_svm_props
1653 use_svnsync_props/) {
1654 my $key = $option;
1655 $key =~ tr/_//d;
1656 my $prop = "-$option";
1657 *$option = sub {
1658 my ($self) = @_;
1659 return $self->{$prop} if exists $self->{$prop};
1660 my $k = "svn-remote.$self->{repo_id}.$key";
1661 eval { command_oneline(qw/config --get/, $k) };
1662 if ($@) {
1663 $self->{$prop} = ${"Git::SVN::_$option"};
1664 } else {
1665 my $v = command_oneline(qw/config --bool/,$k);
1666 $self->{$prop} = $v eq 'false' ? 0 : 1;
1667 }
1668 return $self->{$prop};
1669 }
1670 }
1671 }
1672
1673
1674 my (%LOCKFILES, %INDEX_FILES);
1675 END {
1676 unlink keys %LOCKFILES if %LOCKFILES;
1677 unlink keys %INDEX_FILES if %INDEX_FILES;
1678 }
1679
1680 sub resolve_local_globs {
1681 my ($url, $fetch, $glob_spec) = @_;
1682 return unless defined $glob_spec;
1683 my $ref = $glob_spec->{ref};
1684 my $path = $glob_spec->{path};
1685 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1686 next unless m#^$ref->{regex}$#;
1687 my $p = $1;
1688 my $pathname = desanitize_refname($path->full_path($p));
1689 my $refname = desanitize_refname($ref->full_path($p));
1690 if (my $existing = $fetch->{$pathname}) {
1691 if ($existing ne $refname) {
1692 die "Refspec conflict:\n",
1693 "existing: $existing\n",
1694 " globbed: $refname\n";
1695 }
1696 my $u = (::cmt_metadata("$refname"))[0];
1697 $u =~ s!^\Q$url\E(/|$)!! or die
1698 "$refname: '$url' not found in '$u'\n";
1699 if ($pathname ne $u) {
1700 warn "W: Refspec glob conflict ",
1701 "(ref: $refname):\n",
1702 "expected path: $pathname\n",
1703 " real path: $u\n",
1704 "Continuing ahead with $u\n";
1705 next;
1706 }
1707 } else {
1708 $fetch->{$pathname} = $refname;
1709 }
1710 }
1711 }
1712
1713 sub parse_revision_argument {
1714 my ($base, $head) = @_;
1715 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1716 return ($base, $head);
1717 }
1718 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1719 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1720 return ($head, $head) if ($::_revision eq 'HEAD');
1721 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1722 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1723 die "revision argument: $::_revision not understood by git-svn\n";
1724 }
1725
1726 sub fetch_all {
1727 my ($repo_id, $remotes) = @_;
1728 if (ref $repo_id) {
1729 my $gs = $repo_id;
1730 $repo_id = undef;
1731 $repo_id = $gs->{repo_id};
1732 }
1733 $remotes ||= read_all_remotes();
1734 my $remote = $remotes->{$repo_id} or
1735 die "[svn-remote \"$repo_id\"] unknown\n";
1736 my $fetch = $remote->{fetch};
1737 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1738 my (@gs, @globs);
1739 my $ra = Git::SVN::Ra->new($url);
1740 my $uuid = $ra->get_uuid;
1741 my $head = $ra->get_latest_revnum;
1742 $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1743 my $base = defined $fetch ? $head : 0;
1744
1745 # read the max revs for wildcard expansion (branches/*, tags/*)
1746 foreach my $t (qw/branches tags/) {
1747 defined $remote->{$t} or next;
1748 push @globs, @{$remote->{$t}};
1749
1750 my $max_rev = eval { tmp_config(qw/--int --get/,
1751 "svn-remote.$repo_id.${t}-maxRev") };
1752 if (defined $max_rev && ($max_rev < $base)) {
1753 $base = $max_rev;
1754 } elsif (!defined $max_rev) {
1755 $base = 0;
1756 }
1757 }
1758
1759 if ($fetch) {
1760 foreach my $p (sort keys %$fetch) {
1761 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1762 my $lr = $gs->rev_map_max;
1763 if (defined $lr) {
1764 $base = $lr if ($lr < $base);
1765 }
1766 push @gs, $gs;
1767 }
1768 }
1769
1770 ($base, $head) = parse_revision_argument($base, $head);
1771 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1772 }
1773
1774 sub read_all_remotes {
1775 my $r = {};
1776 my $use_svm_props = eval { command_oneline(qw/config --bool
1777 svn.useSvmProps/) };
1778 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1779 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1780 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1781 if (m!^(.+)\.fetch=$svn_refspec$!) {
1782 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1783 die("svn-remote.$remote: remote ref '$remote_ref' "
1784 . "must start with 'refs/'\n")
1785 unless $remote_ref =~ m{^refs/};
1786 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1787 $r->{$remote}->{svm} = {} if $use_svm_props;
1788 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1789 $r->{$1}->{svm} = {};
1790 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1791 $r->{$1}->{url} = $2;
1792 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1793 my ($remote, $t, $local_ref, $remote_ref) =
1794 ($1, $2, $3, $4);
1795 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1796 . "must start with 'refs/'\n")
1797 unless $remote_ref =~ m{^refs/};
1798 my $rs = {
1799 t => $t,
1800 remote => $remote,
1801 path => Git::SVN::GlobSpec->new($local_ref),
1802 ref => Git::SVN::GlobSpec->new($remote_ref) };
1803 if (length($rs->{ref}->{right}) != 0) {
1804 die "The '*' glob character must be the last ",
1805 "character of '$remote_ref'\n";
1806 }
1807 push @{ $r->{$remote}->{$t} }, $rs;
1808 }
1809 }
1810
1811 map {
1812 if (defined $r->{$_}->{svm}) {
1813 my $svm;
1814 eval {
1815 my $section = "svn-remote.$_";
1816 $svm = {
1817 source => tmp_config('--get',
1818 "$section.svm-source"),
1819 replace => tmp_config('--get',
1820 "$section.svm-replace"),
1821 }
1822 };
1823 $r->{$_}->{svm} = $svm;
1824 }
1825 } keys %$r;
1826
1827 $r;
1828 }
1829
1830 sub init_vars {
1831 $_gc_nr = $_gc_period = 1000;
1832 if (defined $_repack || defined $_repack_flags) {
1833 warn "Repack options are obsolete; they have no effect.\n";
1834 }
1835 }
1836
1837 sub verify_remotes_sanity {
1838 return unless -d $ENV{GIT_DIR};
1839 my %seen;
1840 foreach (command(qw/config -l/)) {
1841 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1842 if ($seen{$1}) {
1843 die "Remote ref refs/remote/$1 is tracked by",
1844 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1845 "Please resolve this ambiguity in ",
1846 "your git configuration file before ",
1847 "continuing\n";
1848 }
1849 $seen{$1} = $_;
1850 }
1851 }
1852 }
1853
1854 sub find_existing_remote {
1855 my ($url, $remotes) = @_;
1856 return undef if $no_reuse_existing;
1857 my $existing;
1858 foreach my $repo_id (keys %$remotes) {
1859 my $u = $remotes->{$repo_id}->{url} or next;
1860 next if $u ne $url;
1861 $existing = $repo_id;
1862 last;
1863 }
1864 $existing;
1865 }
1866
1867 sub init_remote_config {
1868 my ($self, $url, $no_write) = @_;
1869 $url =~ s!/+$!!; # strip trailing slash
1870 my $r = read_all_remotes();
1871 my $existing = find_existing_remote($url, $r);
1872 if ($existing) {
1873 unless ($no_write) {
1874 print STDERR "Using existing ",
1875 "[svn-remote \"$existing\"]\n";
1876 }
1877 $self->{repo_id} = $existing;
1878 } elsif ($_minimize_url) {
1879 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1880 $existing = find_existing_remote($min_url, $r);
1881 if ($existing) {
1882 unless ($no_write) {
1883 print STDERR "Using existing ",
1884 "[svn-remote \"$existing\"]\n";
1885 }
1886 $self->{repo_id} = $existing;
1887 }
1888 if ($min_url ne $url) {
1889 unless ($no_write) {
1890 print STDERR "Using higher level of URL: ",
1891 "$url => $min_url\n";
1892 }
1893 my $old_path = $self->{path};
1894 $self->{path} = $url;
1895 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1896 if (length $old_path) {
1897 $self->{path} .= "/$old_path";
1898 }
1899 $url = $min_url;
1900 }
1901 }
1902 my $orig_url;
1903 if (!$existing) {
1904 # verify that we aren't overwriting anything:
1905 $orig_url = eval {
1906 command_oneline('config', '--get',
1907 "svn-remote.$self->{repo_id}.url")
1908 };
1909 if ($orig_url && ($orig_url ne $url)) {
1910 die "svn-remote.$self->{repo_id}.url already set: ",
1911 "$orig_url\nwanted to set to: $url\n";
1912 }
1913 }
1914 my ($xrepo_id, $xpath) = find_ref($self->refname);
1915 if (!$no_write && defined $xpath) {
1916 die "svn-remote.$xrepo_id.fetch already set to track ",
1917 "$xpath:", $self->refname, "\n";
1918 }
1919 unless ($no_write) {
1920 command_noisy('config',
1921 "svn-remote.$self->{repo_id}.url", $url);
1922 $self->{path} =~ s{^/}{};
1923 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1924 command_noisy('config', '--add',
1925 "svn-remote.$self->{repo_id}.fetch",
1926 "$self->{path}:".$self->refname);
1927 }
1928 $self->{url} = $url;
1929 }
1930
1931 sub find_by_url { # repos_root and, path are optional
1932 my ($class, $full_url, $repos_root, $path) = @_;
1933
1934 return undef unless defined $full_url;
1935 remove_username($full_url);
1936 remove_username($repos_root) if defined $repos_root;
1937 my $remotes = read_all_remotes();
1938 if (defined $full_url && defined $repos_root && !defined $path) {
1939 $path = $full_url;
1940 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1941 }
1942 foreach my $repo_id (keys %$remotes) {
1943 my $u = $remotes->{$repo_id}->{url} or next;
1944 remove_username($u);
1945 next if defined $repos_root && $repos_root ne $u;
1946
1947 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1948 foreach my $t (qw/branches tags/) {
1949 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1950 resolve_local_globs($u, $fetch, $globspec);
1951 }
1952 }
1953 my $p = $path;
1954 my $rwr = rewrite_root({repo_id => $repo_id});
1955 my $svm = $remotes->{$repo_id}->{svm}
1956 if defined $remotes->{$repo_id}->{svm};
1957 unless (defined $p) {
1958 $p = $full_url;
1959 my $z = $u;
1960 my $prefix = '';
1961 if ($rwr) {
1962 $z = $rwr;
1963 remove_username($z);
1964 } elsif (defined $svm) {
1965 $z = $svm->{source};
1966 $prefix = $svm->{replace};
1967 $prefix =~ s#^\Q$u\E(?:/|$)##;
1968 $prefix =~ s#/$##;
1969 }
1970 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1971 }
1972 foreach my $f (keys %$fetch) {
1973 next if $f ne $p;
1974 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1975 }
1976 }
1977 undef;
1978 }
1979
1980 sub init {
1981 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1982 my $self = _new($class, $repo_id, $ref_id, $path);
1983 if (defined $url) {
1984 $self->init_remote_config($url, $no_write);
1985 }
1986 $self;
1987 }
1988
1989 sub find_ref {
1990 my ($ref_id) = @_;
1991 foreach (command(qw/config -l/)) {
1992 next unless m!^svn-remote\.(.+)\.fetch=
1993 \s*(.*?)\s*:\s*(.+?)\s*$!x;
1994 my ($repo_id, $path, $ref) = ($1, $2, $3);
1995 if ($ref eq $ref_id) {
1996 $path = '' if ($path =~ m#^\./?#);
1997 return ($repo_id, $path);
1998 }
1999 }
2000 (undef, undef, undef);
2001 }
2002
2003 sub new {
2004 my ($class, $ref_id, $repo_id, $path) = @_;
2005 if (defined $ref_id && !defined $repo_id && !defined $path) {
2006 ($repo_id, $path) = find_ref($ref_id);
2007 if (!defined $repo_id) {
2008 die "Could not find a \"svn-remote.*.fetch\" key ",
2009 "in the repository configuration matching: ",
2010 "$ref_id\n";
2011 }
2012 }
2013 my $self = _new($class, $repo_id, $ref_id, $path);
2014 if (!defined $self->{path} || !length $self->{path}) {
2015 my $fetch = command_oneline('config', '--get',
2016 "svn-remote.$repo_id.fetch",
2017 ":$ref_id\$") or
2018 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2019 "\":$ref_id\$\" in config\n";
2020 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2021 }
2022 $self->{url} = command_oneline('config', '--get',
2023 "svn-remote.$repo_id.url") or
2024 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2025 $self->rebuild;
2026 $self;
2027 }
2028
2029 sub refname {
2030 my ($refname) = $_[0]->{ref_id} ;
2031
2032 # It cannot end with a slash /, we'll throw up on this because
2033 # SVN can't have directories with a slash in their name, either:
2034 if ($refname =~ m{/$}) {
2035 die "ref: '$refname' ends with a trailing slash, this is ",
2036 "not permitted by git nor Subversion\n";
2037 }
2038
2039 # It cannot have ASCII control character space, tilde ~, caret ^,
2040 # colon :, question-mark ?, asterisk *, space, or open bracket [
2041 # anywhere.
2042 #
2043 # Additionally, % must be escaped because it is used for escaping
2044 # and we want our escaped refname to be reversible
2045 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2046
2047 # no slash-separated component can begin with a dot .
2048 # /.* becomes /%2E*
2049 $refname =~ s{/\.}{/%2E}g;
2050
2051 # It cannot have two consecutive dots .. anywhere
2052 # .. becomes %2E%2E
2053 $refname =~ s{\.\.}{%2E%2E}g;
2054
2055 return $refname;
2056 }
2057
2058 sub desanitize_refname {
2059 my ($refname) = @_;
2060 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2061 return $refname;
2062 }
2063
2064 sub svm_uuid {
2065 my ($self) = @_;
2066 return $self->{svm}->{uuid} if $self->svm;
2067 $self->ra;
2068 unless ($self->{svm}) {
2069 die "SVM UUID not cached, and reading remotely failed\n";
2070 }
2071 $self->{svm}->{uuid};
2072 }
2073
2074 sub svm {
2075 my ($self) = @_;
2076 return $self->{svm} if $self->{svm};
2077 my $svm;
2078 # see if we have it in our config, first:
2079 eval {
2080 my $section = "svn-remote.$self->{repo_id}";
2081 $svm = {
2082 source => tmp_config('--get', "$section.svm-source"),
2083 uuid => tmp_config('--get', "$section.svm-uuid"),
2084 replace => tmp_config('--get', "$section.svm-replace"),
2085 }
2086 };
2087 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2088 $self->{svm} = $svm;
2089 }
2090 $self->{svm};
2091 }
2092
2093 sub _set_svm_vars {
2094 my ($self, $ra) = @_;
2095 return $ra if $self->svm;
2096
2097 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2098 "(svm:source, svm:uuid) ",
2099 "from the following URLs:\n" );
2100 sub read_svm_props {
2101 my ($self, $ra, $path, $r) = @_;
2102 my $props = ($ra->get_dir($path, $r))[2];
2103 my $src = $props->{'svm:source'};
2104 my $uuid = $props->{'svm:uuid'};
2105 return undef if (!$src || !$uuid);
2106
2107 chomp($src, $uuid);
2108
2109 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2110 or die "doesn't look right - svm:uuid is '$uuid'\n";
2111
2112 # the '!' is used to mark the repos_root!/relative/path
2113 $src =~ s{/?!/?}{/};
2114 $src =~ s{/+$}{}; # no trailing slashes please
2115 # username is of no interest
2116 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2117
2118 my $replace = $ra->{url};
2119 $replace .= "/$path" if length $path;
2120
2121 my $section = "svn-remote.$self->{repo_id}";
2122 tmp_config("$section.svm-source", $src);
2123 tmp_config("$section.svm-replace", $replace);
2124 tmp_config("$section.svm-uuid", $uuid);
2125 $self->{svm} = {
2126 source => $src,
2127 uuid => $uuid,
2128 replace => $replace
2129 };
2130 }
2131
2132 my $r = $ra->get_latest_revnum;
2133 my $path = $self->{path};
2134 my %tried;
2135 while (length $path) {
2136 unless ($tried{"$self->{url}/$path"}) {
2137 return $ra if $self->read_svm_props($ra, $path, $r);
2138 $tried{"$self->{url}/$path"} = 1;
2139 }
2140 $path =~ s#/?[^/]+$##;
2141 }
2142 die "Path: '$path' should be ''\n" if $path ne '';
2143 return $ra if $self->read_svm_props($ra, $path, $r);
2144 $tried{"$self->{url}/$path"} = 1;
2145
2146 if ($ra->{repos_root} eq $self->{url}) {
2147 die @err, (map { " $_\n" } keys %tried), "\n";
2148 }
2149
2150 # nope, make sure we're connected to the repository root:
2151 my $ok;
2152 my @tried_b;
2153 $path = $ra->{svn_path};
2154 $ra = Git::SVN::Ra->new($ra->{repos_root});
2155 while (length $path) {
2156 unless ($tried{"$ra->{url}/$path"}) {
2157 $ok = $self->read_svm_props($ra, $path, $r);
2158 last if $ok;
2159 $tried{"$ra->{url}/$path"} = 1;
2160 }
2161 $path =~ s#/?[^/]+$##;
2162 }
2163 die "Path: '$path' should be ''\n" if $path ne '';
2164 $ok ||= $self->read_svm_props($ra, $path, $r);
2165 $tried{"$ra->{url}/$path"} = 1;
2166 if (!$ok) {
2167 die @err, (map { " $_\n" } keys %tried), "\n";
2168 }
2169 Git::SVN::Ra->new($self->{url});
2170 }
2171
2172 sub svnsync {
2173 my ($self) = @_;
2174 return $self->{svnsync} if $self->{svnsync};
2175
2176 if ($self->no_metadata) {
2177 die "Can't have both 'noMetadata' and ",
2178 "'useSvnsyncProps' options set!\n";
2179 }
2180 if ($self->rewrite_root) {
2181 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2182 "options set!\n";
2183 }
2184
2185 my $svnsync;
2186 # see if we have it in our config, first:
2187 eval {
2188 my $section = "svn-remote.$self->{repo_id}";
2189
2190 my $url = tmp_config('--get', "$section.svnsync-url");
2191 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2192 die "doesn't look right - svn:sync-from-url is '$url'\n";
2193
2194 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2195 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2196 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2197
2198 $svnsync = { url => $url, uuid => $uuid }
2199 };
2200 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2201 return $self->{svnsync} = $svnsync;
2202 }
2203
2204 my $err = "useSvnsyncProps set, but failed to read " .
2205 "svnsync property: svn:sync-from-";
2206 my $rp = $self->ra->rev_proplist(0);
2207
2208 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2209 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2210 die "doesn't look right - svn:sync-from-url is '$url'\n";
2211
2212 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2213 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2214 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2215
2216 my $section = "svn-remote.$self->{repo_id}";
2217 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2218 tmp_config('--add', "$section.svnsync-url", $url);
2219 return $self->{svnsync} = { url => $url, uuid => $uuid };
2220 }
2221
2222 # this allows us to memoize our SVN::Ra UUID locally and avoid a
2223 # remote lookup (useful for 'git svn log').
2224 sub ra_uuid {
2225 my ($self) = @_;
2226 unless ($self->{ra_uuid}) {
2227 my $key = "svn-remote.$self->{repo_id}.uuid";
2228 my $uuid = eval { tmp_config('--get', $key) };
2229 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2230 $self->{ra_uuid} = $uuid;
2231 } else {
2232 die "ra_uuid called without URL\n" unless $self->{url};
2233 $self->{ra_uuid} = $self->ra->get_uuid;
2234 tmp_config('--add', $key, $self->{ra_uuid});
2235 }
2236 }
2237 $self->{ra_uuid};
2238 }
2239
2240 sub _set_repos_root {
2241 my ($self, $repos_root) = @_;
2242 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2243 $repos_root ||= $self->ra->{repos_root};
2244 tmp_config($k, $repos_root);
2245 $repos_root;
2246 }
2247
2248 sub repos_root {
2249 my ($self) = @_;
2250 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2251 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2252 }
2253
2254 sub ra {
2255 my ($self) = shift;
2256 my $ra = Git::SVN::Ra->new($self->{url});
2257 $self->_set_repos_root($ra->{repos_root});
2258 if ($self->use_svm_props && !$self->{svm}) {
2259 if ($self->no_metadata) {
2260 die "Can't have both 'noMetadata' and ",
2261 "'useSvmProps' options set!\n";
2262 } elsif ($self->use_svnsync_props) {
2263 die "Can't have both 'useSvnsyncProps' and ",
2264 "'useSvmProps' options set!\n";
2265 }
2266 $ra = $self->_set_svm_vars($ra);
2267 $self->{-want_revprops} = 1;
2268 }
2269 $ra;
2270 }
2271
2272 # prop_walk(PATH, REV, SUB)
2273 # -------------------------
2274 # Recursively traverse PATH at revision REV and invoke SUB for each
2275 # directory that contains a SVN property. SUB will be invoked as
2276 # follows: &SUB(gs, path, props); where `gs' is this instance of
2277 # Git::SVN, `path' the path to the directory where the properties
2278 # `props' were found. The `path' will be relative to point of checkout,
2279 # that is, if url://repo/trunk is the current Git branch, and that
2280 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
2281 # as `path' (note the trailing `/').
2282 sub prop_walk {
2283 my ($self, $path, $rev, $sub) = @_;
2284
2285 $path =~ s#^/##;
2286 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2287 $path =~ s#^/*#/#g;
2288 my $p = $path;
2289 # Strip the irrelevant part of the path.
2290 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2291 # Ensure the path is terminated by a `/'.
2292 $p =~ s#/*$#/#;
2293
2294 # The properties contain all the internal SVN stuff nobody
2295 # (usually) cares about.
2296 my $interesting_props = 0;
2297 foreach (keys %{$props}) {
2298 # If it doesn't start with `svn:', it must be a
2299 # user-defined property.
2300 ++$interesting_props and next if $_ !~ /^svn:/;
2301 # FIXME: Fragile, if SVN adds new public properties,
2302 # this needs to be updated.
2303 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2304 |eol-style|mime-type
2305 |externals|needs-lock)$/x;
2306 }
2307 &$sub($self, $p, $props) if $interesting_props;
2308
2309 foreach (sort keys %$dirent) {
2310 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2311 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2312 }
2313 }
2314
2315 sub last_rev { ($_[0]->last_rev_commit)[0] }
2316 sub last_commit { ($_[0]->last_rev_commit)[1] }
2317
2318 # returns the newest SVN revision number and newest commit SHA1
2319 sub last_rev_commit {
2320 my ($self) = @_;
2321 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2322 return ($self->{last_rev}, $self->{last_commit});
2323 }
2324 my $c = ::verify_ref($self->refname.'^0');
2325 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2326 my $rev = (::cmt_metadata($c))[1];
2327 if (defined $rev) {
2328 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2329 return ($rev, $c);
2330 }
2331 }
2332 my $map_path = $self->map_path;
2333 unless (-e $map_path) {
2334 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2335 return (undef, undef);
2336 }
2337 my ($rev, $commit) = $self->rev_map_max(1);
2338 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2339 return ($rev, $commit);
2340 }
2341
2342 sub get_fetch_range {
2343 my ($self, $min, $max) = @_;
2344 $max ||= $self->ra->get_latest_revnum;
2345 $min ||= $self->rev_map_max;
2346 (++$min, $max);
2347 }
2348
2349 sub tmp_config {
2350 my (@args) = @_;
2351 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2352 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2353 if (! -f $config && -f $old_def_config) {
2354 rename $old_def_config, $config or
2355 die "Failed rename $old_def_config => $config: $!\n";
2356 }
2357 my $old_config = $ENV{GIT_CONFIG};
2358 $ENV{GIT_CONFIG} = $config;
2359 $@ = undef;
2360 my @ret = eval {
2361 unless (-f $config) {
2362 mkfile($config);
2363 open my $fh, '>', $config or
2364 die "Can't open $config: $!\n";
2365 print $fh "; This file is used internally by ",
2366 "git-svn\n" or die
2367 "Couldn't write to $config: $!\n";
2368 print $fh "; You should not have to edit it\n" or
2369 die "Couldn't write to $config: $!\n";
2370 close $fh or die "Couldn't close $config: $!\n";
2371 }
2372 command('config', @args);
2373 };
2374 my $err = $@;
2375 if (defined $old_config) {
2376 $ENV{GIT_CONFIG} = $old_config;
2377 } else {
2378 delete $ENV{GIT_CONFIG};
2379 }
2380 die $err if $err;
2381 wantarray ? @ret : $ret[0];
2382 }
2383
2384 sub tmp_index_do {
2385 my ($self, $sub) = @_;
2386 my $old_index = $ENV{GIT_INDEX_FILE};
2387 $ENV{GIT_INDEX_FILE} = $self->{index};
2388 $@ = undef;
2389 my @ret = eval {
2390 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2391 mkpath([$dir]) unless -d $dir;
2392 &$sub;
2393 };
2394 my $err = $@;
2395 if (defined $old_index) {
2396 $ENV{GIT_INDEX_FILE} = $old_index;
2397 } else {
2398 delete $ENV{GIT_INDEX_FILE};
2399 }
2400 die $err if $err;
2401 wantarray ? @ret : $ret[0];
2402 }
2403
2404 sub assert_index_clean {
2405 my ($self, $treeish) = @_;
2406
2407 $self->tmp_index_do(sub {
2408 command_noisy('read-tree', $treeish) unless -e $self->{index};
2409 my $x = command_oneline('write-tree');
2410 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2411 /^tree ($::sha1)/mo);
2412 return if $y eq $x;
2413
2414 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2415 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2416 command_noisy('read-tree', $treeish);
2417 $x = command_oneline('write-tree');
2418 if ($y ne $x) {
2419 ::fatal "trees ($treeish) $y != $x\n",
2420 "Something is seriously wrong...";
2421 }
2422 });
2423 }
2424
2425 sub get_commit_parents {
2426 my ($self, $log_entry) = @_;
2427 my (%seen, @ret, @tmp);
2428 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2429 if (my $ip = $self->{inject_parents}) {
2430 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2431 push @tmp, $commit;
2432 }
2433 }
2434 if (my $cur = ::verify_ref($self->refname.'^0')) {
2435 push @tmp, $cur;
2436 }
2437 if (my $ipd = $self->{inject_parents_dcommit}) {
2438 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2439 push @tmp, @$commit;
2440 }
2441 }
2442 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2443 while (my $p = shift @tmp) {
2444 next if $seen{$p};
2445 $seen{$p} = 1;
2446 push @ret, $p;
2447 # MAXPARENT is defined to 16 in commit-tree.c:
2448 last if @ret >= 16;
2449 }
2450 if (@tmp) {
2451 die "r$log_entry->{revision}: No room for parents:\n\t",
2452 join("\n\t", @tmp), "\n";
2453 }
2454 @ret;
2455 }
2456
2457 sub rewrite_root {
2458 my ($self) = @_;
2459 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2460 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2461 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2462 if ($rwr) {
2463 $rwr =~ s#/+$##;
2464 if ($rwr !~ m#^[a-z\+]+://#) {
2465 die "$rwr is not a valid URL (key: $k)\n";
2466 }
2467 }
2468 $self->{-rewrite_root} = $rwr;
2469 }
2470
2471 sub metadata_url {
2472 my ($self) = @_;
2473 ($self->rewrite_root || $self->{url}) .
2474 (length $self->{path} ? '/' . $self->{path} : '');
2475 }
2476
2477 sub full_url {
2478 my ($self) = @_;
2479 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2480 }
2481
2482
2483 sub set_commit_header_env {
2484 my ($log_entry) = @_;
2485 my %env;
2486 foreach my $ned (qw/NAME EMAIL DATE/) {
2487 foreach my $ac (qw/AUTHOR COMMITTER/) {
2488 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2489 }
2490 }
2491
2492 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2493 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2494 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2495
2496 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2497 ? $log_entry->{commit_name}
2498 : $log_entry->{name};
2499 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2500 ? $log_entry->{commit_email}
2501 : $log_entry->{email};
2502 \%env;
2503 }
2504
2505 sub restore_commit_header_env {
2506 my ($env) = @_;
2507 foreach my $ned (qw/NAME EMAIL DATE/) {
2508 foreach my $ac (qw/AUTHOR COMMITTER/) {
2509 my $k = "GIT_${ac}_${ned}";
2510 if (defined $env->{$k}) {
2511 $ENV{$k} = $env->{$k};
2512 } else {
2513 delete $ENV{$k};
2514 }
2515 }
2516 }
2517 }
2518
2519 sub gc {
2520 command_noisy('gc', '--auto');
2521 };
2522
2523 sub do_git_commit {
2524 my ($self, $log_entry) = @_;
2525 my $lr = $self->last_rev;
2526 if (defined $lr && $lr >= $log_entry->{revision}) {
2527 die "Last fetched revision of ", $self->refname,
2528 " was r$lr, but we are about to fetch: ",
2529 "r$log_entry->{revision}!\n";
2530 }
2531 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2532 croak "$log_entry->{revision} = $c already exists! ",
2533 "Why are we refetching it?\n";
2534 }
2535 my $old_env = set_commit_header_env($log_entry);
2536 my $tree = $log_entry->{tree};
2537 if (!defined $tree) {
2538 $tree = $self->tmp_index_do(sub {
2539 command_oneline('write-tree') });
2540 }
2541 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2542
2543 my @exec = ('git', 'commit-tree', $tree);
2544 foreach ($self->get_commit_parents($log_entry)) {
2545 push @exec, '-p', $_;
2546 }
2547 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2548 or croak $!;
2549 binmode $msg_fh;
2550
2551 # we always get UTF-8 from SVN, but we may want our commits in
2552 # a different encoding.
2553 if (my $enc = Git::config('i18n.commitencoding')) {
2554 require Encode;
2555 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2556 }
2557 print $msg_fh $log_entry->{log} or croak $!;
2558 restore_commit_header_env($old_env);
2559 unless ($self->no_metadata) {
2560 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2561 or croak $!;
2562 }
2563 $msg_fh->flush == 0 or croak $!;
2564 close $msg_fh or croak $!;
2565 chomp(my $commit = do { local $/; <$out_fh> });
2566 close $out_fh or croak $!;
2567 waitpid $pid, 0;
2568 croak $? if $?;
2569 if ($commit !~ /^$::sha1$/o) {
2570 die "Failed to commit, invalid sha1: $commit\n";
2571 }
2572
2573 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2574
2575 $self->{last_rev} = $log_entry->{revision};
2576 $self->{last_commit} = $commit;
2577 print "r$log_entry->{revision}" unless $::_q > 1;
2578 if (defined $log_entry->{svm_revision}) {
2579 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2580 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2581 0, $self->svm_uuid);
2582 }
2583 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2584 if (--$_gc_nr == 0) {
2585 $_gc_nr = $_gc_period;
2586 gc();
2587 }
2588 return $commit;
2589 }
2590
2591 sub match_paths {
2592 my ($self, $paths, $r) = @_;
2593 return 1 if $self->{path} eq '';
2594 if (my $path = $paths->{"/$self->{path}"}) {
2595 return ($path->{action} eq 'D') ? 0 : 1;
2596 }
2597 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2598 if (grep /$self->{path_regex}/, keys %$paths) {
2599 return 1;
2600 }
2601 my $c = '';
2602 foreach (split m#/#, $self->{path}) {
2603 $c .= "/$_";
2604 next unless ($paths->{$c} &&
2605 ($paths->{$c}->{action} =~ /^[AR]$/));
2606 if ($self->ra->check_path($self->{path}, $r) ==
2607 $SVN::Node::dir) {
2608 return 1;
2609 }
2610 }
2611 return 0;
2612 }
2613
2614 sub find_parent_branch {
2615 my ($self, $paths, $rev) = @_;
2616 return undef unless $self->follow_parent;
2617 unless (defined $paths) {
2618 my $err_handler = $SVN::Error::handler;
2619 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2620 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2621 sub { $paths = $_[0] });
2622 $SVN::Error::handler = $err_handler;
2623 }
2624 return undef unless defined $paths;
2625
2626 # look for a parent from another branch:
2627 my @b_path_components = split m#/#, $self->{path};
2628 my @a_path_components;
2629 my $i;
2630 while (@b_path_components) {
2631 $i = $paths->{'/'.join('/', @b_path_components)};
2632 last if $i && defined $i->{copyfrom_path};
2633 unshift(@a_path_components, pop(@b_path_components));
2634 }
2635 return undef unless defined $i && defined $i->{copyfrom_path};
2636 my $branch_from = $i->{copyfrom_path};
2637 if (@a_path_components) {
2638 print STDERR "branch_from: $branch_from => ";
2639 $branch_from .= '/'.join('/', @a_path_components);
2640 print STDERR $branch_from, "\n";
2641 }
2642 my $r = $i->{copyfrom_rev};
2643 my $repos_root = $self->ra->{repos_root};
2644 my $url = $self->ra->{url};
2645 my $new_url = $url . $branch_from;
2646 print STDERR "Found possible branch point: ",
2647 "$new_url => ", $self->full_url, ", $r\n"
2648 unless $::_q > 1;
2649 $branch_from =~ s#^/##;
2650 my $gs = $self->other_gs($new_url, $url,
2651 $branch_from, $r, $self->{ref_id});
2652 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2653 {
2654 my ($base, $head);
2655 if (!defined $r0 || !defined $parent) {
2656 ($base, $head) = parse_revision_argument(0, $r);
2657 } else {
2658 if ($r0 < $r) {
2659 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2660 0, 1, sub { $base = $_[1] - 1 });
2661 }
2662 }
2663 if (defined $base && $base <= $r) {
2664 $gs->fetch($base, $r);
2665 }
2666 ($r0, $parent) = $gs->find_rev_before($r, 1);
2667 }
2668 if (defined $r0 && defined $parent) {
2669 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2670 unless $::_q > 1;
2671 my $ed;
2672 if ($self->ra->can_do_switch) {
2673 $self->assert_index_clean($parent);
2674 print STDERR "Following parent with do_switch\n"
2675 unless $::_q > 1;
2676 # do_switch works with svn/trunk >= r22312, but that
2677 # is not included with SVN 1.4.3 (the latest version
2678 # at the moment), so we can't rely on it
2679 $self->{last_rev} = $r0;
2680 $self->{last_commit} = $parent;
2681 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2682 $gs->ra->gs_do_switch($r0, $rev, $gs,
2683 $self->full_url, $ed)
2684 or die "SVN connection failed somewhere...\n";
2685 } elsif ($self->ra->trees_match($new_url, $r0,
2686 $self->full_url, $rev)) {
2687 print STDERR "Trees match:\n",
2688 " $new_url\@$r0\n",
2689 " ${\$self->full_url}\@$rev\n",
2690 "Following parent with no changes\n"
2691 unless $::_q > 1;
2692 $self->tmp_index_do(sub {
2693 command_noisy('read-tree', $parent);
2694 });
2695 $self->{last_commit} = $parent;
2696 } else {
2697 print STDERR "Following parent with do_update\n"
2698 unless $::_q > 1;
2699 $ed = SVN::Git::Fetcher->new($self);
2700 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2701 or die "SVN connection failed somewhere...\n";
2702 }
2703 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2704 return $self->make_log_entry($rev, [$parent], $ed);
2705 }
2706 return undef;
2707 }
2708
2709 sub do_fetch {
2710 my ($self, $paths, $rev) = @_;
2711 my $ed;
2712 my ($last_rev, @parents);
2713 if (my $lc = $self->last_commit) {
2714 # we can have a branch that was deleted, then re-added
2715 # under the same name but copied from another path, in
2716 # which case we'll have multiple parents (we don't
2717 # want to break the original ref, nor lose copypath info):
2718 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2719 push @{$log_entry->{parents}}, $lc;
2720 return $log_entry;
2721 }
2722 $ed = SVN::Git::Fetcher->new($self);
2723 $last_rev = $self->{last_rev};
2724 $ed->{c} = $lc;
2725 @parents = ($lc);
2726 } else {
2727 $last_rev = $rev;
2728 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2729 return $log_entry;
2730 }
2731 $ed = SVN::Git::Fetcher->new($self);
2732 }
2733 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2734 die "SVN connection failed somewhere...\n";
2735 }
2736 $self->make_log_entry($rev, \@parents, $ed);
2737 }
2738
2739 sub mkemptydirs {
2740 my ($self, $r) = @_;
2741 my %empty_dirs = ();
2742
2743 open my $fh, '<', "$self->{dir}/unhandled.log" or return;
2744 binmode $fh or croak "binmode: $!";
2745 while (<$fh>) {
2746 if (defined $r && /^r(\d+)$/) {
2747 last if $1 > $r;
2748 } elsif (/^ \+empty_dir: (.+)$/) {
2749 $empty_dirs{$1} = 1;
2750 } elsif (/^ \-empty_dir: (.+)$/) {
2751 my @d = grep {m[^\Q$1\E(/|$)]} (keys %empty_dirs);
2752 delete @empty_dirs{@d};
2753 }
2754 }
2755 close $fh;
2756
2757 my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2758 foreach my $d (sort keys %empty_dirs) {
2759 $d = uri_decode($d);
2760 $d =~ s/$strip//;
2761 next if -d $d;
2762 if (-e _) {
2763 warn "$d exists but is not a directory\n";
2764 } else {
2765 print "creating empty directory: $d\n";
2766 mkpath([$d]);
2767 }
2768 }
2769 }
2770
2771 sub get_untracked {
2772 my ($self, $ed) = @_;
2773 my @out;
2774 my $h = $ed->{empty};
2775 foreach (sort keys %$h) {
2776 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2777 push @out, " $act: " . uri_encode($_);
2778 warn "W: $act: $_\n";
2779 }
2780 foreach my $t (qw/dir_prop file_prop/) {
2781 $h = $ed->{$t} or next;
2782 foreach my $path (sort keys %$h) {
2783 my $ppath = $path eq '' ? '.' : $path;
2784 foreach my $prop (sort keys %{$h->{$path}}) {
2785 next if $SKIP_PROP{$prop};
2786 my $v = $h->{$path}->{$prop};
2787 my $t_ppath_prop = "$t: " .
2788 uri_encode($ppath) . ' ' .
2789 uri_encode($prop);
2790 if (defined $v) {
2791 push @out, " +$t_ppath_prop " .
2792 uri_encode($v);
2793 } else {
2794 push @out, " -$t_ppath_prop";
2795 }
2796 }
2797 }
2798 }
2799 foreach my $t (qw/absent_file absent_directory/) {
2800 $h = $ed->{$t} or next;
2801 foreach my $parent (sort keys %$h) {
2802 foreach my $path (sort @{$h->{$parent}}) {
2803 push @out, " $t: " .
2804 uri_encode("$parent/$path");
2805 warn "W: $t: $parent/$path ",
2806 "Insufficient permissions?\n";
2807 }
2808 }
2809 }
2810 \@out;
2811 }
2812
2813 # parse_svn_date(DATE)
2814 # --------------------
2815 # Given a date (in UTC) from Subversion, return a string in the format
2816 # "<TZ Offset> <local date/time>" that Git will use.
2817 #
2818 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
2819 # is true we'll convert it to the local timezone instead.
2820 sub parse_svn_date {
2821 my $date = shift || return '+0000 1970-01-01 00:00:00';
2822 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2823 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2824 croak "Unable to parse date: $date\n";
2825 my $parsed_date; # Set next.
2826
2827 if ($Git::SVN::_localtime) {
2828 # Translate the Subversion datetime to an epoch time.
2829 # Begin by switching ourselves to $date's timezone, UTC.
2830 my $old_env_TZ = $ENV{TZ};
2831 $ENV{TZ} = 'UTC';
2832
2833 my $epoch_in_UTC =
2834 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2835
2836 # Determine our local timezone (including DST) at the
2837 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2838 # value of TZ, if any, at the time we were run.
2839 if (defined $Git::SVN::Log::TZ) {
2840 $ENV{TZ} = $Git::SVN::Log::TZ;
2841 } else {
2842 delete $ENV{TZ};
2843 }
2844
2845 my $our_TZ =
2846 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2847
2848 # This converts $epoch_in_UTC into our local timezone.
2849 my ($sec, $min, $hour, $mday, $mon, $year,
2850 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2851
2852 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2853 $our_TZ, $year + 1900, $mon + 1,
2854 $mday, $hour, $min, $sec);
2855
2856 # Reset us to the timezone in effect when we entered
2857 # this routine.
2858 if (defined $old_env_TZ) {
2859 $ENV{TZ} = $old_env_TZ;
2860 } else {
2861 delete $ENV{TZ};
2862 }
2863 } else {
2864 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2865 }
2866
2867 return $parsed_date;
2868 }
2869
2870 sub other_gs {
2871 my ($self, $new_url, $url,
2872 $branch_from, $r, $old_ref_id) = @_;
2873 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2874 unless ($gs) {
2875 my $ref_id = $old_ref_id;
2876 $ref_id =~ s/\@\d+$//;
2877 $ref_id .= "\@$r";
2878 # just grow a tail if we're not unique enough :x
2879 $ref_id .= '-' while find_ref($ref_id);
2880 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2881 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2882 if ($u =~ s#^\Q$url\E(/|$)##) {
2883 $p = $u;
2884 $u = $url;
2885 $repo_id = $self->{repo_id};
2886 }
2887 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2888 }
2889 $gs
2890 }
2891
2892 sub call_authors_prog {
2893 my ($orig_author) = @_;
2894 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2895 my $author = `$::_authors_prog $orig_author`;
2896 if ($? != 0) {
2897 die "$::_authors_prog failed with exit code $?\n"
2898 }
2899 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2900 my ($name, $email) = ($1, $2);
2901 $email = undef if length $2 == 0;
2902 return [$name, $email];
2903 } else {
2904 die "Author: $orig_author: $::_authors_prog returned "
2905 . "invalid author format: $author\n";
2906 }
2907 }
2908
2909 sub check_author {
2910 my ($author) = @_;
2911 if (!defined $author || length $author == 0) {
2912 $author = '(no author)';
2913 }
2914 if (!defined $::users{$author}) {
2915 if (defined $::_authors_prog) {
2916 $::users{$author} = call_authors_prog($author);
2917 } elsif (defined $::_authors) {
2918 die "Author: $author not defined in $::_authors file\n";
2919 }
2920 }
2921 $author;
2922 }
2923
2924 sub find_extra_svk_parents {
2925 my ($self, $ed, $tickets, $parents) = @_;
2926 # aha! svk:merge property changed...
2927 my @tickets = split "\n", $tickets;
2928 my @known_parents;
2929 for my $ticket ( @tickets ) {
2930 my ($uuid, $path, $rev) = split /:/, $ticket;
2931 if ( $uuid eq $self->ra_uuid ) {
2932 my $url = $self->rewrite_root || $self->{url};
2933 my $repos_root = $url;
2934 my $branch_from = $path;
2935 $branch_from =~ s{^/}{};
2936 my $gs = $self->other_gs($repos_root."/".$branch_from,
2937 $url,
2938 $branch_from,
2939 $rev,
2940 $self->{ref_id});
2941 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
2942 # wahey! we found it, but it might be
2943 # an old one (!)
2944 push @known_parents, [ $rev, $commit ];
2945 }
2946 }
2947 }
2948 # Ordering matters; highest-numbered commit merge tickets
2949 # first, as they may account for later merge ticket additions
2950 # or changes.
2951 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
2952 for my $parent ( @known_parents ) {
2953 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
2954 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
2955 my $new;
2956 while ( <$msg_fh> ) {
2957 $new=1;last;
2958 }
2959 command_close_pipe($msg_fh, $ctx);
2960 if ( $new ) {
2961 print STDERR
2962 "Found merge parent (svk:merge ticket): $parent\n";
2963 push @$parents, $parent;
2964 }
2965 }
2966 }
2967
2968 # note: this function should only be called if the various dirprops
2969 # have actually changed
2970 sub find_extra_svn_parents {
2971 my ($self, $ed, $mergeinfo, $parents) = @_;
2972 # aha! svk:merge property changed...
2973
2974 # We first search for merged tips which are not in our
2975 # history. Then, we figure out which git revisions are in
2976 # that tip, but not this revision. If all of those revisions
2977 # are now marked as merge, we can add the tip as a parent.
2978 my @merges = split "\n", $mergeinfo;
2979 my @merge_tips;
2980 my @merged_commit_ranges;
2981 my $url = $self->rewrite_root || $self->{url};
2982 for my $merge ( @merges ) {
2983 my ($source, $revs) = split ":", $merge;
2984 my $path = $source;
2985 $path =~ s{^/}{};
2986 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
2987 if ( !$gs ) {
2988 warn "Couldn't find revmap for $url$source\n";
2989 next;
2990 }
2991 my @ranges = split ",", $revs;
2992 my ($tip, $tip_commit);
2993 # find the tip
2994 for my $range ( @ranges ) {
2995 my ($bottom, $top) = split "-", $range;
2996 $top ||= $bottom;
2997 my $bottom_commit =
2998 $gs->rev_map_get($bottom, $self->ra_uuid) ||
2999 $gs->rev_map_get($bottom+1, $self->ra_uuid);
3000 my $top_commit;
3001 for (; !$top_commit && $top >= $bottom; --$top) {
3002 $top_commit =
3003 $gs->rev_map_get($top, $self->ra_uuid);
3004 }
3005
3006 unless ($top_commit and $bottom_commit) {
3007 warn "W:unknown path/rev in svn:mergeinfo "
3008 ."dirprop: $source:$range\n";
3009 next;
3010 }
3011
3012 push @merged_commit_ranges,
3013 "$bottom_commit..$top_commit";
3014
3015 if ( !defined $tip or $top > $tip ) {
3016 $tip = $top;
3017 $tip_commit = $top_commit;
3018 }
3019 }
3020 unless (!$tip_commit or
3021 grep { $_ eq $tip_commit } @$parents ) {
3022 push @merge_tips, $tip_commit;
3023 } else {
3024 push @merge_tips, undef;
3025 }
3026 }
3027 for my $merge_tip ( @merge_tips ) {
3028 my $spec = shift @merges;
3029 next unless $merge_tip;
3030 my @cmd = ('rev-list', "-1", $merge_tip,
3031 "--not", @$parents );
3032 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3033 my $new;
3034 while ( <$msg_fh> ) {
3035 $new=1;last;
3036 }
3037 command_close_pipe($msg_fh, $ctx);
3038 if ( $new ) {
3039 push @cmd, @merged_commit_ranges;
3040 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3041 my $unmerged;
3042 while ( <$msg_fh> ) {
3043 $unmerged=1;last;
3044 }
3045 command_close_pipe($msg_fh, $ctx);
3046 if ( $unmerged ) {
3047 warn "W:svn cherry-pick ignored ($spec)\n";
3048 } else {
3049 warn
3050 "Found merge parent (svn:mergeinfo prop): ",
3051 $merge_tip, "\n";
3052 push @$parents, $merge_tip;
3053 }
3054 }
3055 }
3056 }
3057
3058 sub make_log_entry {
3059 my ($self, $rev, $parents, $ed) = @_;
3060 my $untracked = $self->get_untracked($ed);
3061
3062 my @parents = @$parents;
3063 my $ps = $ed->{path_strip} || "";
3064 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3065 my $props = $ed->{dir_prop}{$path};
3066 if ( $props->{"svk:merge"} ) {
3067 $self->find_extra_svk_parents
3068 ($ed, $props->{"svk:merge"}, \@parents);
3069 }
3070 if ( $props->{"svn:mergeinfo"} ) {
3071 $self->find_extra_svn_parents
3072 ($ed,
3073 $props->{"svn:mergeinfo"},
3074 \@parents);
3075 }
3076 }
3077
3078 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3079 print $un "r$rev\n" or croak $!;
3080 print $un $_, "\n" foreach @$untracked;
3081 my %log_entry = ( parents => \@parents, revision => $rev,
3082 log => '');
3083
3084 my $headrev;
3085 my $logged = delete $self->{logged_rev_props};
3086 if (!$logged || $self->{-want_revprops}) {
3087 my $rp = $self->ra->rev_proplist($rev);
3088 foreach (sort keys %$rp) {
3089 my $v = $rp->{$_};
3090 if (/^svn:(author|date|log)$/) {
3091 $log_entry{$1} = $v;
3092 } elsif ($_ eq 'svm:headrev') {
3093 $headrev = $v;
3094 } else {
3095 print $un " rev_prop: ", uri_encode($_), ' ',
3096 uri_encode($v), "\n";
3097 }
3098 }
3099 } else {
3100 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3101 }
3102 close $un or croak $!;
3103
3104 $log_entry{date} = parse_svn_date($log_entry{date});
3105 $log_entry{log} .= "\n";
3106 my $author = $log_entry{author} = check_author($log_entry{author});
3107 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3108 : ($author, undef);
3109
3110 my ($commit_name, $commit_email) = ($name, $email);
3111 if ($_use_log_author) {
3112 my $name_field;
3113 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3114 $name_field = $1;
3115 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3116 $name_field = $1;
3117 }
3118 if (!defined $name_field) {
3119 if (!defined $email) {
3120 $email = $name;
3121 }
3122 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3123 ($name, $email) = ($1, $2);
3124 } elsif ($name_field =~ /(.*)@/) {
3125 ($name, $email) = ($1, $name_field);
3126 } else {
3127 ($name, $email) = ($name_field, $name_field);
3128 }
3129 }
3130 if (defined $headrev && $self->use_svm_props) {
3131 if ($self->rewrite_root) {
3132 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3133 "options set!\n";
3134 }
3135 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3136 # we don't want "SVM: initializing mirror for junk" ...
3137 return undef if $r == 0;
3138 my $svm = $self->svm;
3139 if ($uuid ne $svm->{uuid}) {
3140 die "UUID mismatch on SVM path:\n",
3141 "expected: $svm->{uuid}\n",
3142 " got: $uuid\n";
3143 }
3144 my $full_url = $self->full_url;
3145 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3146 die "Failed to replace '$svm->{replace}' with ",
3147 "'$svm->{source}' in $full_url\n";
3148 # throw away username for storing in records
3149 remove_username($full_url);
3150 $log_entry{metadata} = "$full_url\@$r $uuid";
3151 $log_entry{svm_revision} = $r;
3152 $email ||= "$author\@$uuid";
3153 $commit_email ||= "$author\@$uuid";
3154 } elsif ($self->use_svnsync_props) {
3155 my $full_url = $self->svnsync->{url};
3156 $full_url .= "/$self->{path}" if length $self->{path};
3157 remove_username($full_url);
3158 my $uuid = $self->svnsync->{uuid};
3159 $log_entry{metadata} = "$full_url\@$rev $uuid";
3160 $email ||= "$author\@$uuid";
3161 $commit_email ||= "$author\@$uuid";
3162 } else {
3163 my $url = $self->metadata_url;
3164 remove_username($url);
3165 $log_entry{metadata} = "$url\@$rev " .
3166 $self->ra->get_uuid;
3167 $email ||= "$author\@" . $self->ra->get_uuid;
3168 $commit_email ||= "$author\@" . $self->ra->get_uuid;
3169 }
3170 $log_entry{name} = $name;
3171 $log_entry{email} = $email;
3172 $log_entry{commit_name} = $commit_name;
3173 $log_entry{commit_email} = $commit_email;
3174 \%log_entry;
3175 }
3176
3177 sub fetch {
3178 my ($self, $min_rev, $max_rev, @parents) = @_;
3179 my ($last_rev, $last_commit) = $self->last_rev_commit;
3180 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3181 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3182 }
3183
3184 sub set_tree_cb {
3185 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3186 $self->{inject_parents} = { $rev => $tree };
3187 $self->fetch(undef, undef);
3188 }
3189
3190 sub set_tree {
3191 my ($self, $tree) = (shift, shift);
3192 my $log_entry = ::get_commit_entry($tree);
3193 unless ($self->{last_rev}) {
3194 ::fatal("Must have an existing revision to commit");
3195 }
3196 my %ed_opts = ( r => $self->{last_rev},
3197 log => $log_entry->{log},
3198 ra => $self->ra,
3199 tree_a => $self->{last_commit},
3200 tree_b => $tree,
3201 editor_cb => sub {
3202 $self->set_tree_cb($log_entry, $tree, @_) },
3203 svn_path => $self->{path} );
3204 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3205 print "No changes\nr$self->{last_rev} = $tree\n";
3206 }
3207 }
3208
3209 sub rebuild_from_rev_db {
3210 my ($self, $path) = @_;
3211 my $r = -1;
3212 open my $fh, '<', $path or croak "open: $!";
3213 binmode $fh or croak "binmode: $!";
3214 while (<$fh>) {
3215 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3216 chomp($_);
3217 ++$r;
3218 next if $_ eq ('0' x 40);
3219 $self->rev_map_set($r, $_);
3220 print "r$r = $_\n";
3221 }
3222 close $fh or croak "close: $!";
3223 unlink $path or croak "unlink: $!";
3224 }
3225
3226 sub rebuild {
3227 my ($self) = @_;
3228 my $map_path = $self->map_path;
3229 my $partial = (-e $map_path && ! -z $map_path);
3230 return unless ::verify_ref($self->refname.'^0');
3231 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3232 my $rev_db = $self->rev_db_path;
3233 $self->rebuild_from_rev_db($rev_db);
3234 if ($self->use_svm_props) {
3235 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3236 $self->rebuild_from_rev_db($svm_rev_db);
3237 }
3238 $self->unlink_rev_db_symlink;
3239 return;
3240 }
3241 print "Rebuilding $map_path ...\n" if (!$partial);
3242 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3243 (undef, undef));
3244 my ($log, $ctx) =
3245 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3246 ($head ? "$head.." : "") . $self->refname,
3247 '--');
3248 my $metadata_url = $self->metadata_url;
3249 remove_username($metadata_url);
3250 my $svn_uuid = $self->ra_uuid;
3251 my $c;
3252 while (<$log>) {
3253 if ( m{^commit ($::sha1)$} ) {
3254 $c = $1;
3255 next;
3256 }
3257 next unless s{^\s*(git-svn-id:)}{$1};
3258 my ($url, $rev, $uuid) = ::extract_metadata($_);
3259 remove_username($url);
3260
3261 # ignore merges (from set-tree)
3262 next if (!defined $rev || !$uuid);
3263
3264 # if we merged or otherwise started elsewhere, this is
3265 # how we break out of it
3266 if (($uuid ne $svn_uuid) ||
3267 ($metadata_url && $url && ($url ne $metadata_url))) {
3268 next;
3269 }
3270 if ($partial && $head) {
3271 print "Partial-rebuilding $map_path ...\n";
3272 print "Currently at $base_rev = $head\n";
3273 $head = undef;
3274 }
3275
3276 $self->rev_map_set($rev, $c);
3277 print "r$rev = $c\n";
3278 }
3279 command_close_pipe($log, $ctx);
3280 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3281 my $rev_db_path = $self->rev_db_path;
3282 if (-f $self->rev_db_path) {
3283 unlink $self->rev_db_path or croak "unlink: $!";
3284 }
3285 $self->unlink_rev_db_symlink;
3286 }
3287
3288 # rev_map:
3289 # Tie::File seems to be prone to offset errors if revisions get sparse,
3290 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
3291 # one of my favorite modules is out :< Next up would be one of the DBM
3292 # modules, but I'm not sure which is most portable...
3293 #
3294 # This is the replacement for the rev_db format, which was too big
3295 # and inefficient for large repositories with a lot of sparse history
3296 # (mainly tags)
3297 #
3298 # The format is this:
3299 # - 24 bytes for every record,
3300 # * 4 bytes for the integer representing an SVN revision number
3301 # * 20 bytes representing the sha1 of a git commit
3302 # - No empty padding records like the old format
3303 # (except the last record, which can be overwritten)
3304 # - new records are written append-only since SVN revision numbers
3305 # increase monotonically
3306 # - lookups on SVN revision number are done via a binary search
3307 # - Piping the file to xxd -c24 is a good way of dumping it for
3308 # viewing or editing (piped back through xxd -r), should the need
3309 # ever arise.
3310 # - The last record can be padding revision with an all-zero sha1
3311 # This is used to optimize fetch performance when using multiple
3312 # "fetch" directives in .git/config
3313 #
3314 # These files are disposable unless noMetadata or useSvmProps is set
3315
3316 sub _rev_map_set {
3317 my ($fh, $rev, $commit) = @_;
3318
3319 binmode $fh or croak "binmode: $!";
3320 my $size = (stat($fh))[7];
3321 ($size % 24) == 0 or croak "inconsistent size: $size";
3322
3323 my $wr_offset = 0;
3324 if ($size > 0) {
3325 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3326 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3327 $read == 24 or croak "read only $read bytes (!= 24)";
3328 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3329 if ($last_commit eq ('0' x40)) {
3330 if ($size >= 48) {
3331 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3332 $read = sysread($fh, $buf, 24) or
3333 croak "read: $!";
3334 $read == 24 or
3335 croak "read only $read bytes (!= 24)";
3336 ($last_rev, $last_commit) =
3337 unpack(rev_map_fmt, $buf);
3338 if ($last_commit eq ('0' x40)) {
3339 croak "inconsistent .rev_map\n";
3340 }
3341 }
3342 if ($last_rev >= $rev) {
3343 croak "last_rev is higher!: $last_rev >= $rev";
3344 }
3345 $wr_offset = -24;
3346 }
3347 }
3348 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3349 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3350 croak "write: $!";
3351 }
3352
3353 sub _rev_map_reset {
3354 my ($fh, $rev, $commit) = @_;
3355 my $c = _rev_map_get($fh, $rev);
3356 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3357 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3358 truncate $fh, $offset or croak "truncate: $!";
3359 }
3360
3361 sub mkfile {
3362 my ($path) = @_;
3363 unless (-e $path) {
3364 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3365 mkpath([$dir]) unless -d $dir;
3366 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3367 close $fh or die "Couldn't close (create) $path: $!\n";
3368 }
3369 }
3370
3371 sub rev_map_set {
3372 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3373 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3374 my $db = $self->map_path($uuid);
3375 my $db_lock = "$db.lock";
3376 my $sig;
3377 $update_ref ||= 0;
3378 if ($update_ref) {
3379 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3380 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3381 }
3382 mkfile($db);
3383
3384 $LOCKFILES{$db_lock} = 1;
3385 my $sync;
3386 # both of these options make our .rev_db file very, very important
3387 # and we can't afford to lose it because rebuild() won't work
3388 if ($self->use_svm_props || $self->no_metadata) {
3389 $sync = 1;
3390 copy($db, $db_lock) or die "rev_map_set(@_): ",
3391 "Failed to copy: ",
3392 "$db => $db_lock ($!)\n";
3393 } else {
3394 rename $db, $db_lock or die "rev_map_set(@_): ",
3395 "Failed to rename: ",
3396 "$db => $db_lock ($!)\n";
3397 }
3398
3399 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3400 or croak "Couldn't open $db_lock: $!\n";
3401 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3402 _rev_map_set($fh, $rev, $commit);
3403 if ($sync) {
3404 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3405 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3406 }
3407 close $fh or croak $!;
3408 if ($update_ref) {
3409 $_head = $self;
3410 my $note = "";
3411 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3412 command_noisy('update-ref', '-m', "r$rev$note",
3413 $self->refname, $commit);
3414 }
3415 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3416 "$db_lock => $db ($!)\n";
3417 delete $LOCKFILES{$db_lock};
3418 if ($update_ref) {
3419 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3420 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3421 kill $sig, $$ if defined $sig;
3422 }
3423 }
3424
3425 # If want_commit, this will return an array of (rev, commit) where
3426 # commit _must_ be a valid commit in the archive.
3427 # Otherwise, it'll return the max revision (whether or not the
3428 # commit is valid or just a 0x40 placeholder).
3429 sub rev_map_max {
3430 my ($self, $want_commit) = @_;
3431 $self->rebuild;
3432 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3433 $want_commit ? ($r, $c) : $r;
3434 }
3435
3436 sub rev_map_max_norebuild {
3437 my ($self, $want_commit) = @_;
3438 my $map_path = $self->map_path;
3439 stat $map_path or return $want_commit ? (0, undef) : 0;
3440 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3441 binmode $fh or croak "binmode: $!";
3442 my $size = (stat($fh))[7];
3443 ($size % 24) == 0 or croak "inconsistent size: $size";
3444
3445 if ($size == 0) {
3446 close $fh or croak "close: $!";
3447 return $want_commit ? (0, undef) : 0;
3448 }
3449
3450 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3451 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3452 my ($r, $c) = unpack(rev_map_fmt, $buf);
3453 if ($want_commit && $c eq ('0' x40)) {
3454 if ($size < 48) {
3455 return $want_commit ? (0, undef) : 0;
3456 }
3457 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3458 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3459 ($r, $c) = unpack(rev_map_fmt, $buf);
3460 if ($c eq ('0'x40)) {
3461 croak "Penultimate record is all-zeroes in $map_path";
3462 }
3463 }
3464 close $fh or croak "close: $!";
3465 $want_commit ? ($r, $c) : $r;
3466 }
3467
3468 sub rev_map_get {
3469 my ($self, $rev, $uuid) = @_;
3470 my $map_path = $self->map_path($uuid);
3471 return undef unless -e $map_path;
3472
3473 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3474 my $c = _rev_map_get($fh, $rev);
3475 close($fh) or croak "close: $!";
3476 $c
3477 }
3478
3479 sub _rev_map_get {
3480 my ($fh, $rev) = @_;
3481
3482 binmode $fh or croak "binmode: $!";
3483 my $size = (stat($fh))[7];
3484 ($size % 24) == 0 or croak "inconsistent size: $size";
3485
3486 if ($size == 0) {
3487 return undef;
3488 }
3489
3490 my ($l, $u) = (0, $size - 24);
3491 my ($r, $c, $buf);
3492
3493 while ($l <= $u) {
3494 my $i = int(($l/24 + $u/24) / 2) * 24;
3495 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3496 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3497 my ($r, $c) = unpack(rev_map_fmt, $buf);
3498
3499 if ($r < $rev) {
3500 $l = $i + 24;
3501 } elsif ($r > $rev) {
3502 $u = $i - 24;
3503 } else { # $r == $rev
3504 return $c eq ('0' x 40) ? undef : $c;
3505 }
3506 }
3507 undef;
3508 }
3509
3510 # Finds the first svn revision that exists on (if $eq_ok is true) or
3511 # before $rev for the current branch. It will not search any lower
3512 # than $min_rev. Returns the git commit hash and svn revision number
3513 # if found, else (undef, undef).
3514 sub find_rev_before {
3515 my ($self, $rev, $eq_ok, $min_rev) = @_;
3516 --$rev unless $eq_ok;
3517 $min_rev ||= 1;
3518 my $max_rev = $self->rev_map_max;
3519 $rev = $max_rev if ($rev > $max_rev);
3520 while ($rev >= $min_rev) {
3521 if (my $c = $self->rev_map_get($rev)) {
3522 return ($rev, $c);
3523 }
3524 --$rev;
3525 }
3526 return (undef, undef);
3527 }
3528
3529 # Finds the first svn revision that exists on (if $eq_ok is true) or
3530 # after $rev for the current branch. It will not search any higher
3531 # than $max_rev. Returns the git commit hash and svn revision number
3532 # if found, else (undef, undef).
3533 sub find_rev_after {
3534 my ($self, $rev, $eq_ok, $max_rev) = @_;
3535 ++$rev unless $eq_ok;
3536 $max_rev ||= $self->rev_map_max;
3537 while ($rev <= $max_rev) {
3538 if (my $c = $self->rev_map_get($rev)) {
3539 return ($rev, $c);
3540 }
3541 ++$rev;
3542 }
3543 return (undef, undef);
3544 }
3545
3546 sub _new {
3547 my ($class, $repo_id, $ref_id, $path) = @_;
3548 unless (defined $repo_id && length $repo_id) {
3549 $repo_id = $Git::SVN::default_repo_id;
3550 }
3551 unless (defined $ref_id && length $ref_id) {
3552 $_prefix = '' unless defined($_prefix);
3553 $_[2] = $ref_id =
3554 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3555 }
3556 $_[1] = $repo_id;
3557 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3558
3559 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3560 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3561 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3562 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3563 if (-d $old_dir && ! -d $dir) {
3564 $dir = $old_dir;
3565 }
3566 }
3567
3568 $_[3] = $path = '' unless (defined $path);
3569 mkpath([$dir]);
3570 bless {
3571 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3572 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3573 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3574 }
3575
3576 # for read-only access of old .rev_db formats
3577 sub unlink_rev_db_symlink {
3578 my ($self) = @_;
3579 my $link = $self->rev_db_path;
3580 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3581 if (-l $link) {
3582 unlink $link or croak "unlink: $link failed!";
3583 }
3584 }
3585
3586 sub rev_db_path {
3587 my ($self, $uuid) = @_;
3588 my $db_path = $self->map_path($uuid);
3589 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3590 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3591 $db_path;
3592 }
3593
3594 # the new replacement for .rev_db
3595 sub map_path {
3596 my ($self, $uuid) = @_;
3597 $uuid ||= $self->ra_uuid;
3598 "$self->{map_root}.$uuid";
3599 }
3600
3601 sub uri_encode {
3602 my ($f) = @_;
3603 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3604 $f
3605 }
3606
3607 sub uri_decode {
3608 my ($f) = @_;
3609 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3610 $f
3611 }
3612
3613 sub remove_username {
3614 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3615 }
3616
3617 package Git::SVN::Prompt;
3618 use strict;
3619 use warnings;
3620 require SVN::Core;
3621 use vars qw/$_no_auth_cache $_username/;
3622
3623 sub simple {
3624 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3625 $may_save = undef if $_no_auth_cache;
3626 $default_username = $_username if defined $_username;
3627 if (defined $default_username && length $default_username) {
3628 if (defined $realm && length $realm) {
3629 print STDERR "Authentication realm: $realm\n";
3630 STDERR->flush;
3631 }
3632 $cred->username($default_username);
3633 } else {
3634 username($cred, $realm, $may_save, $pool);
3635 }
3636 $cred->password(_read_password("Password for '" .
3637 $cred->username . "': ", $realm));
3638 $cred->may_save($may_save);
3639 $SVN::_Core::SVN_NO_ERROR;
3640 }
3641
3642 sub ssl_server_trust {
3643 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3644 $may_save = undef if $_no_auth_cache;
3645 print STDERR "Error validating server certificate for '$realm':\n";
3646 {
3647 no warnings 'once';
3648 # All variables SVN::Auth::SSL::* are used only once,
3649 # so we're shutting up Perl warnings about this.
3650 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3651 print STDERR " - The certificate is not issued ",
3652 "by a trusted authority. Use the\n",
3653 " fingerprint to validate ",
3654 "the certificate manually!\n";
3655 }
3656 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3657 print STDERR " - The certificate hostname ",
3658 "does not match.\n";
3659 }
3660 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3661 print STDERR " - The certificate is not yet valid.\n";
3662 }
3663 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3664 print STDERR " - The certificate has expired.\n";
3665 }
3666 if ($failures & $SVN::Auth::SSL::OTHER) {
3667 print STDERR " - The certificate has ",
3668 "an unknown error.\n";
3669 }
3670 } # no warnings 'once'
3671 printf STDERR
3672 "Certificate information:\n".
3673 " - Hostname: %s\n".
3674 " - Valid: from %s until %s\n".
3675 " - Issuer: %s\n".
3676 " - Fingerprint: %s\n",
3677 map $cert_info->$_, qw(hostname valid_from valid_until
3678 issuer_dname fingerprint);
3679 my $choice;
3680 prompt:
3681 print STDERR $may_save ?
3682 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3683 "(R)eject or accept (t)emporarily? ";
3684 STDERR->flush;
3685 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3686 if ($choice =~ /^t$/i) {
3687 $cred->may_save(undef);
3688 } elsif ($choice =~ /^r$/i) {
3689 return -1;
3690 } elsif ($may_save && $choice =~ /^p$/i) {
3691 $cred->may_save($may_save);
3692 } else {
3693 goto prompt;
3694 }
3695 $cred->accepted_failures($failures);
3696 $SVN::_Core::SVN_NO_ERROR;
3697 }
3698
3699 sub ssl_client_cert {
3700 my ($cred, $realm, $may_save, $pool) = @_;
3701 $may_save = undef if $_no_auth_cache;
3702 print STDERR "Client certificate filename: ";
3703 STDERR->flush;
3704 chomp(my $filename = <STDIN>);
3705 $cred->cert_file($filename);
3706 $cred->may_save($may_save);
3707 $SVN::_Core::SVN_NO_ERROR;
3708 }
3709
3710 sub ssl_client_cert_pw {
3711 my ($cred, $realm, $may_save, $pool) = @_;
3712 $may_save = undef if $_no_auth_cache;
3713 $cred->password(_read_password("Password: ", $realm));
3714 $cred->may_save($may_save);
3715 $SVN::_Core::SVN_NO_ERROR;
3716 }
3717
3718 sub username {
3719 my ($cred, $realm, $may_save, $pool) = @_;
3720 $may_save = undef if $_no_auth_cache;
3721 if (defined $realm && length $realm) {
3722 print STDERR "Authentication realm: $realm\n";
3723 }
3724 my $username;
3725 if (defined $_username) {
3726 $username = $_username;
3727 } else {
3728 print STDERR "Username: ";
3729 STDERR->flush;
3730 chomp($username = <STDIN>);
3731 }
3732 $cred->username($username);
3733 $cred->may_save($may_save);
3734 $SVN::_Core::SVN_NO_ERROR;
3735 }
3736
3737 sub _read_password {
3738 my ($prompt, $realm) = @_;
3739 print STDERR $prompt;
3740 STDERR->flush;
3741 require Term::ReadKey;
3742 Term::ReadKey::ReadMode('noecho');
3743 my $password = '';
3744 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3745 last if $key =~ /[\012\015]/; # \n\r
3746 $password .= $key;
3747 }
3748 Term::ReadKey::ReadMode('restore');
3749 print STDERR "\n";
3750 STDERR->flush;
3751 $password;
3752 }
3753
3754 package SVN::Git::Fetcher;
3755 use vars qw/@ISA/;
3756 use strict;
3757 use warnings;
3758 use Carp qw/croak/;
3759 use File::Temp qw/tempfile/;
3760 use IO::File qw//;
3761 use vars qw/$_ignore_regex/;
3762
3763 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
3764 sub new {
3765 my ($class, $git_svn, $switch_path) = @_;
3766 my $self = SVN::Delta::Editor->new;
3767 bless $self, $class;
3768 if (exists $git_svn->{last_commit}) {
3769 $self->{c} = $git_svn->{last_commit};
3770 $self->{empty_symlinks} =
3771 _mark_empty_symlinks($git_svn, $switch_path);
3772 }
3773 $self->{ignore_regex} = eval { command_oneline('config', '--get',
3774 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3775 $self->{empty} = {};
3776 $self->{dir_prop} = {};
3777 $self->{file_prop} = {};
3778 $self->{absent_dir} = {};
3779 $self->{absent_file} = {};
3780 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3781 $self;
3782 }
3783
3784 # this uses the Ra object, so it must be called before do_{switch,update},
3785 # not inside them (when the Git::SVN::Fetcher object is passed) to
3786 # do_{switch,update}
3787 sub _mark_empty_symlinks {
3788 my ($git_svn, $switch_path) = @_;
3789 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3790 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3791
3792 my %ret;
3793 my ($rev, $cmt) = $git_svn->last_rev_commit;
3794 return {} unless ($rev && $cmt);
3795
3796 # allow the warning to be printed for each revision we fetch to
3797 # ensure the user sees it. The user can also disable the workaround
3798 # on the repository even while git svn is running and the next
3799 # revision fetched will skip this expensive function.
3800 my $printed_warning;
3801 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3802 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3803 local $/ = "\0";
3804 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3805 $pfx .= '/' if length($pfx);
3806 while (<$ls>) {
3807 chomp;
3808 s/\A100644 blob $empty_blob\t//o or next;
3809 unless ($printed_warning) {
3810 print STDERR "Scanning for empty symlinks, ",
3811 "this may take a while if you have ",
3812 "many empty files\n",
3813 "You may disable this with `",
3814 "git config svn.brokenSymlinkWorkaround ",
3815 "false'.\n",
3816 "This may be done in a different ",
3817 "terminal without restarting ",
3818 "git svn\n";
3819 $printed_warning = 1;
3820 }
3821 my $path = $_;
3822 my (undef, $props) =
3823 $git_svn->ra->get_file($pfx.$path, $rev, undef);
3824 if ($props->{'svn:special'}) {
3825 $ret{$path} = 1;
3826 }
3827 }
3828 command_close_pipe($ls, $ctx);
3829 \%ret;
3830 }
3831
3832 # returns true if a given path is inside a ".git" directory
3833 sub in_dot_git {
3834 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3835 }
3836
3837 # return value: 0 -- don't ignore, 1 -- ignore
3838 sub is_path_ignored {
3839 my ($self, $path) = @_;
3840 return 1 if in_dot_git($path);
3841 return 1 if defined($self->{ignore_regex}) &&
3842 $path =~ m!$self->{ignore_regex}!;
3843 return 0 unless defined($_ignore_regex);
3844 return 1 if $path =~ m!$_ignore_regex!o;
3845 return 0;
3846 }
3847
3848 sub set_path_strip {
3849 my ($self, $path) = @_;
3850 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3851 }
3852
3853 sub open_root {
3854 { path => '' };
3855 }
3856
3857 sub open_directory {
3858 my ($self, $path, $pb, $rev) = @_;
3859 { path => $path };
3860 }
3861
3862 sub git_path {
3863 my ($self, $path) = @_;
3864 if ($self->{path_strip}) {
3865 $path =~ s!$self->{path_strip}!! or
3866 die "Failed to strip path '$path' ($self->{path_strip})\n";
3867 }
3868 $path;
3869 }
3870
3871 sub delete_entry {
3872 my ($self, $path, $rev, $pb) = @_;
3873 return undef if $self->is_path_ignored($path);
3874
3875 my $gpath = $self->git_path($path);
3876 return undef if ($gpath eq '');
3877
3878 # remove entire directories.
3879 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3880 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3881 if ($tree) {
3882 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3883 -r --name-only -z/,
3884 $tree);
3885 local $/ = "\0";
3886 while (<$ls>) {
3887 chomp;
3888 my $rmpath = "$gpath/$_";
3889 $self->{gii}->remove($rmpath);
3890 print "\tD\t$rmpath\n" unless $::_q;
3891 }
3892 print "\tD\t$gpath/\n" unless $::_q;
3893 command_close_pipe($ls, $ctx);
3894 $self->{empty}->{$path} = 0
3895 } else {
3896 $self->{gii}->remove($gpath);
3897 print "\tD\t$gpath\n" unless $::_q;
3898 }
3899 undef;
3900 }
3901
3902 sub open_file {
3903 my ($self, $path, $pb, $rev) = @_;
3904 my ($mode, $blob);
3905
3906 goto out if $self->is_path_ignored($path);
3907
3908 my $gpath = $self->git_path($path);
3909 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3910 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3911 unless (defined $mode && defined $blob) {
3912 die "$path was not found in commit $self->{c} (r$rev)\n";
3913 }
3914 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3915 $mode = '120000';
3916 }
3917 out:
3918 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3919 pool => SVN::Pool->new, action => 'M' };
3920 }
3921
3922 sub add_file {
3923 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3924 my $mode;
3925
3926 if (!$self->is_path_ignored($path)) {
3927 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3928 delete $self->{empty}->{$dir};
3929 $mode = '100644';
3930 }
3931 { path => $path, mode_a => $mode, mode_b => $mode,
3932 pool => SVN::Pool->new, action => 'A' };
3933 }
3934
3935 sub add_directory {
3936 my ($self, $path, $cp_path, $cp_rev) = @_;
3937 goto out if $self->is_path_ignored($path);
3938 my $gpath = $self->git_path($path);
3939 if ($gpath eq '') {
3940 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3941 -r --name-only -z/,
3942 $self->{c});
3943 local $/ = "\0";
3944 while (<$ls>) {
3945 chomp;
3946 $self->{gii}->remove($_);
3947 print "\tD\t$_\n" unless $::_q;
3948 }
3949 command_close_pipe($ls, $ctx);
3950 $self->{empty}->{$path} = 0;
3951 }
3952 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3953 delete $self->{empty}->{$dir};
3954 $self->{empty}->{$path} = 1;
3955 out:
3956 { path => $path };
3957 }
3958
3959 sub change_dir_prop {
3960 my ($self, $db, $prop, $value) = @_;
3961 return undef if $self->is_path_ignored($db->{path});
3962 $self->{dir_prop}->{$db->{path}} ||= {};
3963 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3964 undef;
3965 }
3966
3967 sub absent_directory {
3968 my ($self, $path, $pb) = @_;
3969 return undef if $self->is_path_ignored($path);
3970 $self->{absent_dir}->{$pb->{path}} ||= [];
3971 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3972 undef;
3973 }
3974
3975 sub absent_file {
3976 my ($self, $path, $pb) = @_;
3977 return undef if $self->is_path_ignored($path);
3978 $self->{absent_file}->{$pb->{path}} ||= [];
3979 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3980 undef;
3981 }
3982
3983 sub change_file_prop {
3984 my ($self, $fb, $prop, $value) = @_;
3985 return undef if $self->is_path_ignored($fb->{path});
3986 if ($prop eq 'svn:executable') {
3987 if ($fb->{mode_b} != 120000) {
3988 $fb->{mode_b} = defined $value ? 100755 : 100644;
3989 }
3990 } elsif ($prop eq 'svn:special') {
3991 $fb->{mode_b} = defined $value ? 120000 : 100644;
3992 } else {
3993 $self->{file_prop}->{$fb->{path}} ||= {};
3994 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3995 }
3996 undef;
3997 }
3998
3999 sub apply_textdelta {
4000 my ($self, $fb, $exp) = @_;
4001 return undef if $self->is_path_ignored($fb->{path});
4002 my $fh = $::_repository->temp_acquire('svn_delta');
4003 # $fh gets auto-closed() by SVN::TxDelta::apply(),
4004 # (but $base does not,) so dup() it for reading in close_file
4005 open my $dup, '<&', $fh or croak $!;
4006 my $base = $::_repository->temp_acquire('git_blob');
4007
4008 if ($fb->{blob}) {
4009 my ($base_is_link, $size);
4010
4011 if ($fb->{mode_a} eq '120000' &&
4012 ! $self->{empty_symlinks}->{$fb->{path}}) {
4013 print $base 'link ' or die "print $!\n";
4014 $base_is_link = 1;
4015 }
4016 retry:
4017 $size = $::_repository->cat_blob($fb->{blob}, $base);
4018 die "Failed to read object $fb->{blob}" if ($size < 0);
4019
4020 if (defined $exp) {
4021 seek $base, 0, 0 or croak $!;
4022 my $got = ::md5sum($base);
4023 if ($got ne $exp) {
4024 my $err = "Checksum mismatch: ".
4025 "$fb->{path} $fb->{blob}\n" .
4026 "expected: $exp\n" .
4027 " got: $got\n";
4028 if ($base_is_link) {
4029 warn $err,
4030 "Retrying... (possibly ",
4031 "a bad symlink from SVN)\n";
4032 $::_repository->temp_reset($base);
4033 $base_is_link = 0;
4034 goto retry;
4035 }
4036 die $err;
4037 }
4038 }
4039 }
4040 seek $base, 0, 0 or croak $!;
4041 $fb->{fh} = $fh;
4042 $fb->{base} = $base;
4043 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4044 }
4045
4046 sub close_file {
4047 my ($self, $fb, $exp) = @_;
4048 return undef if $self->is_path_ignored($fb->{path});
4049
4050 my $hash;
4051 my $path = $self->git_path($fb->{path});
4052 if (my $fh = $fb->{fh}) {
4053 if (defined $exp) {
4054 seek($fh, 0, 0) or croak $!;
4055 my $got = ::md5sum($fh);
4056 if ($got ne $exp) {
4057 die "Checksum mismatch: $path\n",
4058 "expected: $exp\n got: $got\n";
4059 }
4060 }
4061 if ($fb->{mode_b} == 120000) {
4062 sysseek($fh, 0, 0) or croak $!;
4063 my $rd = sysread($fh, my $buf, 5);
4064
4065 if (!defined $rd) {
4066 croak "sysread: $!\n";
4067 } elsif ($rd == 0) {
4068 warn "$path has mode 120000",
4069 " but it points to nothing\n",
4070 "converting to an empty file with mode",
4071 " 100644\n";
4072 $fb->{mode_b} = '100644';
4073 } elsif ($buf ne 'link ') {
4074 warn "$path has mode 120000",
4075 " but is not a link\n";
4076 } else {
4077 my $tmp_fh = $::_repository->temp_acquire(
4078 'svn_hash');
4079 my $res;
4080 while ($res = sysread($fh, my $str, 1024)) {
4081 my $out = syswrite($tmp_fh, $str, $res);
4082 defined($out) && $out == $res
4083 or croak("write ",
4084 Git::temp_path($tmp_fh),
4085 ": $!\n");
4086 }
4087 defined $res or croak $!;
4088
4089 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4090 Git::temp_release($tmp_fh, 1);
4091 }
4092 }
4093
4094 $hash = $::_repository->hash_and_insert_object(
4095 Git::temp_path($fh));
4096 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4097
4098 Git::temp_release($fb->{base}, 1);
4099 Git::temp_release($fh, 1);
4100 } else {
4101 $hash = $fb->{blob} or die "no blob information\n";
4102 }
4103 $fb->{pool}->clear;
4104 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4105 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4106 undef;
4107 }
4108
4109 sub abort_edit {
4110 my $self = shift;
4111 $self->{nr} = $self->{gii}->{nr};
4112 delete $self->{gii};
4113 $self->SUPER::abort_edit(@_);
4114 }
4115
4116 sub close_edit {
4117 my $self = shift;
4118 $self->{git_commit_ok} = 1;
4119 $self->{nr} = $self->{gii}->{nr};
4120 delete $self->{gii};
4121 $self->SUPER::close_edit(@_);
4122 }
4123
4124 package SVN::Git::Editor;
4125 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4126 use strict;
4127 use warnings;
4128 use Carp qw/croak/;
4129 use IO::File;
4130
4131 sub new {
4132 my ($class, $opts) = @_;
4133 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4134 die "$_ required!\n" unless (defined $opts->{$_});
4135 }
4136
4137 my $pool = SVN::Pool->new;
4138 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4139 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4140 $opts->{r}, $mods);
4141
4142 # $opts->{ra} functions should not be used after this:
4143 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4144 $opts->{editor_cb}, $pool);
4145 my $self = SVN::Delta::Editor->new(@ce, $pool);
4146 bless $self, $class;
4147 foreach (qw/svn_path r tree_a tree_b/) {
4148 $self->{$_} = $opts->{$_};
4149 }
4150 $self->{url} = $opts->{ra}->{url};
4151 $self->{mods} = $mods;
4152 $self->{types} = $types;
4153 $self->{pool} = $pool;
4154 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4155 $self->{rm} = { };
4156 $self->{path_prefix} = length $self->{svn_path} ?
4157 "$self->{svn_path}/" : '';
4158 $self->{config} = $opts->{config};
4159 return $self;
4160 }
4161
4162 sub generate_diff {
4163 my ($tree_a, $tree_b) = @_;
4164 my @diff_tree = qw(diff-tree -z -r);
4165 if ($_cp_similarity) {
4166 push @diff_tree, "-C$_cp_similarity";
4167 } else {
4168 push @diff_tree, '-C';
4169 }
4170 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4171 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4172 push @diff_tree, $tree_a, $tree_b;
4173 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4174 local $/ = "\0";
4175 my $state = 'meta';
4176 my @mods;
4177 while (<$diff_fh>) {
4178 chomp $_; # this gets rid of the trailing "\0"
4179 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4180 ($::sha1)\s($::sha1)\s
4181 ([MTCRAD])\d*$/xo) {
4182 push @mods, { mode_a => $1, mode_b => $2,
4183 sha1_a => $3, sha1_b => $4,
4184 chg => $5 };
4185 if ($5 =~ /^(?:C|R)$/) {
4186 $state = 'file_a';
4187 } else {
4188 $state = 'file_b';
4189 }
4190 } elsif ($state eq 'file_a') {
4191 my $x = $mods[$#mods] or croak "Empty array\n";
4192 if ($x->{chg} !~ /^(?:C|R)$/) {
4193 croak "Error parsing $_, $x->{chg}\n";
4194 }
4195 $x->{file_a} = $_;
4196 $state = 'file_b';
4197 } elsif ($state eq 'file_b') {
4198 my $x = $mods[$#mods] or croak "Empty array\n";
4199 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4200 croak "Error parsing $_, $x->{chg}\n";
4201 }
4202 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4203 croak "Error parsing $_, $x->{chg}\n";
4204 }
4205 $x->{file_b} = $_;
4206 $state = 'meta';
4207 } else {
4208 croak "Error parsing $_\n";
4209 }
4210 }
4211 command_close_pipe($diff_fh, $ctx);
4212 \@mods;
4213 }
4214
4215 sub check_diff_paths {
4216 my ($ra, $pfx, $rev, $mods) = @_;
4217 my %types;
4218 $pfx .= '/' if length $pfx;
4219
4220 sub type_diff_paths {
4221 my ($ra, $types, $path, $rev) = @_;
4222 my @p = split m#/+#, $path;
4223 my $c = shift @p;
4224 unless (defined $types->{$c}) {
4225 $types->{$c} = $ra->check_path($c, $rev);
4226 }
4227 while (@p) {
4228 $c .= '/' . shift @p;
4229 next if defined $types->{$c};
4230 $types->{$c} = $ra->check_path($c, $rev);
4231 }
4232 }
4233
4234 foreach my $m (@$mods) {
4235 foreach my $f (qw/file_a file_b/) {
4236 next unless defined $m->{$f};
4237 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4238 if (length $pfx.$dir && ! defined $types{$dir}) {
4239 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4240 }
4241 }
4242 }
4243 \%types;
4244 }
4245
4246 sub split_path {
4247 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4248 }
4249
4250 sub repo_path {
4251 my ($self, $path) = @_;
4252 $self->{path_prefix}.(defined $path ? $path : '');
4253 }
4254
4255 sub url_path {
4256 my ($self, $path) = @_;
4257 if ($self->{url} =~ m#^https?://#) {
4258 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4259 }
4260 $self->{url} . '/' . $self->repo_path($path);
4261 }
4262
4263 sub rmdirs {
4264 my ($self) = @_;
4265 my $rm = $self->{rm};
4266 delete $rm->{''}; # we never delete the url we're tracking
4267 return unless %$rm;
4268
4269 foreach (keys %$rm) {
4270 my @d = split m#/#, $_;
4271 my $c = shift @d;
4272 $rm->{$c} = 1;
4273 while (@d) {
4274 $c .= '/' . shift @d;
4275 $rm->{$c} = 1;
4276 }
4277 }
4278 delete $rm->{$self->{svn_path}};
4279 delete $rm->{''}; # we never delete the url we're tracking
4280 return unless %$rm;
4281
4282 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4283 $self->{tree_b});
4284 local $/ = "\0";
4285 while (<$fh>) {
4286 chomp;
4287 my @dn = split m#/#, $_;
4288 while (pop @dn) {
4289 delete $rm->{join '/', @dn};
4290 }
4291 unless (%$rm) {
4292 close $fh;
4293 return;
4294 }
4295 }
4296 command_close_pipe($fh, $ctx);
4297
4298 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4299 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4300 $self->close_directory($bat->{$d}, $p);
4301 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4302 print "\tD+\t$d/\n" unless $::_q;
4303 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4304 delete $bat->{$d};
4305 }
4306 }
4307
4308 sub open_or_add_dir {
4309 my ($self, $full_path, $baton) = @_;
4310 my $t = $self->{types}->{$full_path};
4311 if (!defined $t) {
4312 die "$full_path not known in r$self->{r} or we have a bug!\n";
4313 }
4314 {
4315 no warnings 'once';
4316 # SVN::Node::none and SVN::Node::file are used only once,
4317 # so we're shutting up Perl's warnings about them.
4318 if ($t == $SVN::Node::none) {
4319 return $self->add_directory($full_path, $baton,
4320 undef, -1, $self->{pool});
4321 } elsif ($t == $SVN::Node::dir) {
4322 return $self->open_directory($full_path, $baton,
4323 $self->{r}, $self->{pool});
4324 } # no warnings 'once'
4325 print STDERR "$full_path already exists in repository at ",
4326 "r$self->{r} and it is not a directory (",
4327 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4328 } # no warnings 'once'
4329 exit 1;
4330 }
4331
4332 sub ensure_path {
4333 my ($self, $path) = @_;
4334 my $bat = $self->{bat};
4335 my $repo_path = $self->repo_path($path);
4336 return $bat->{''} unless (length $repo_path);
4337 my @p = split m#/+#, $repo_path;
4338 my $c = shift @p;
4339 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4340 while (@p) {
4341 my $c0 = $c;
4342 $c .= '/' . shift @p;
4343 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4344 }
4345 return $bat->{$c};
4346 }
4347
4348 # Subroutine to convert a globbing pattern to a regular expression.
4349 # From perl cookbook.
4350 sub glob2pat {
4351 my $globstr = shift;
4352 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4353 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4354 return '^' . $globstr . '$';
4355 }
4356
4357 sub check_autoprop {
4358 my ($self, $pattern, $properties, $file, $fbat) = @_;
4359 # Convert the globbing pattern to a regular expression.
4360 my $regex = glob2pat($pattern);
4361 # Check if the pattern matches the file name.
4362 if($file =~ m/($regex)/) {
4363 # Parse the list of properties to set.
4364 my @props = split(/;/, $properties);
4365 foreach my $prop (@props) {
4366 # Parse 'name=value' syntax and set the property.
4367 if ($prop =~ /([^=]+)=(.*)/) {
4368 my ($n,$v) = ($1,$2);
4369 for ($n, $v) {
4370 s/^\s+//; s/\s+$//;
4371 }
4372 $self->change_file_prop($fbat, $n, $v);
4373 }
4374 }
4375 }
4376 }
4377
4378 sub apply_autoprops {
4379 my ($self, $file, $fbat) = @_;
4380 my $conf_t = ${$self->{config}}{'config'};
4381 no warnings 'once';
4382 # Check [miscellany]/enable-auto-props in svn configuration.
4383 if (SVN::_Core::svn_config_get_bool(
4384 $conf_t,
4385 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4386 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4387 0)) {
4388 # Auto-props are enabled. Enumerate them to look for matches.
4389 my $callback = sub {
4390 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4391 };
4392 SVN::_Core::svn_config_enumerate(
4393 $conf_t,
4394 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4395 $callback);
4396 }
4397 }
4398
4399 sub A {
4400 my ($self, $m) = @_;
4401 my ($dir, $file) = split_path($m->{file_b});
4402 my $pbat = $self->ensure_path($dir);
4403 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4404 undef, -1);
4405 print "\tA\t$m->{file_b}\n" unless $::_q;
4406 $self->apply_autoprops($file, $fbat);
4407 $self->chg_file($fbat, $m);
4408 $self->close_file($fbat,undef,$self->{pool});
4409 }
4410
4411 sub C {
4412 my ($self, $m) = @_;
4413 my ($dir, $file) = split_path($m->{file_b});
4414 my $pbat = $self->ensure_path($dir);
4415 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4416 $self->url_path($m->{file_a}), $self->{r});
4417 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4418 $self->chg_file($fbat, $m);
4419 $self->close_file($fbat,undef,$self->{pool});
4420 }
4421
4422 sub delete_entry {
4423 my ($self, $path, $pbat) = @_;
4424 my $rpath = $self->repo_path($path);
4425 my ($dir, $file) = split_path($rpath);
4426 $self->{rm}->{$dir} = 1;
4427 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4428 }
4429
4430 sub R {
4431 my ($self, $m) = @_;
4432 my ($dir, $file) = split_path($m->{file_b});
4433 my $pbat = $self->ensure_path($dir);
4434 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4435 $self->url_path($m->{file_a}), $self->{r});
4436 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4437 $self->apply_autoprops($file, $fbat);
4438 $self->chg_file($fbat, $m);
4439 $self->close_file($fbat,undef,$self->{pool});
4440
4441 ($dir, $file) = split_path($m->{file_a});
4442 $pbat = $self->ensure_path($dir);
4443 $self->delete_entry($m->{file_a}, $pbat);
4444 }
4445
4446 sub M {
4447 my ($self, $m) = @_;
4448 my ($dir, $file) = split_path($m->{file_b});
4449 my $pbat = $self->ensure_path($dir);
4450 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4451 $pbat,$self->{r},$self->{pool});
4452 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4453 $self->chg_file($fbat, $m);
4454 $self->close_file($fbat,undef,$self->{pool});
4455 }
4456
4457 sub T { shift->M(@_) }
4458
4459 sub change_file_prop {
4460 my ($self, $fbat, $pname, $pval) = @_;
4461 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4462 }
4463
4464 sub _chg_file_get_blob ($$$$) {
4465 my ($self, $fbat, $m, $which) = @_;
4466 my $fh = $::_repository->temp_acquire("git_blob_$which");
4467 if ($m->{"mode_$which"} =~ /^120/) {
4468 print $fh 'link ' or croak $!;
4469 $self->change_file_prop($fbat,'svn:special','*');
4470 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4471 $self->change_file_prop($fbat,'svn:special',undef);
4472 }
4473 my $blob = $m->{"sha1_$which"};
4474 return ($fh,) if ($blob =~ /^0{40}$/);
4475 my $size = $::_repository->cat_blob($blob, $fh);
4476 croak "Failed to read object $blob" if ($size < 0);
4477 $fh->flush == 0 or croak $!;
4478 seek $fh, 0, 0 or croak $!;
4479
4480 my $exp = ::md5sum($fh);
4481 seek $fh, 0, 0 or croak $!;
4482 return ($fh, $exp);
4483 }
4484
4485 sub chg_file {
4486 my ($self, $fbat, $m) = @_;
4487 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4488 $self->change_file_prop($fbat,'svn:executable','*');
4489 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4490 $self->change_file_prop($fbat,'svn:executable',undef);
4491 }
4492 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4493 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4494 my $pool = SVN::Pool->new;
4495 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4496 if (-s $fh_a) {
4497 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4498 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4499 if (defined $res) {
4500 die "Unexpected result from send_txstream: $res\n",
4501 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4502 }
4503 } else {
4504 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4505 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4506 if ($got ne $exp_b);
4507 }
4508 Git::temp_release($fh_b, 1);
4509 Git::temp_release($fh_a, 1);
4510 $pool->clear;
4511 }
4512
4513 sub D {
4514 my ($self, $m) = @_;
4515 my ($dir, $file) = split_path($m->{file_b});
4516 my $pbat = $self->ensure_path($dir);
4517 print "\tD\t$m->{file_b}\n" unless $::_q;
4518 $self->delete_entry($m->{file_b}, $pbat);
4519 }
4520
4521 sub close_edit {
4522 my ($self) = @_;
4523 my ($p,$bat) = ($self->{pool}, $self->{bat});
4524 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4525 next if $_ eq '';
4526 $self->close_directory($bat->{$_}, $p);
4527 }
4528 $self->close_directory($bat->{''}, $p);
4529 $self->SUPER::close_edit($p);
4530 $p->clear;
4531 }
4532
4533 sub abort_edit {
4534 my ($self) = @_;
4535 $self->SUPER::abort_edit($self->{pool});
4536 }
4537
4538 sub DESTROY {
4539 my $self = shift;
4540 $self->SUPER::DESTROY(@_);
4541 $self->{pool}->clear;
4542 }
4543
4544 # this drives the editor
4545 sub apply_diff {
4546 my ($self) = @_;
4547 my $mods = $self->{mods};
4548 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4549 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4550 my $f = $m->{chg};
4551 if (defined $o{$f}) {
4552 $self->$f($m);
4553 } else {
4554 fatal("Invalid change type: $f");
4555 }
4556 }
4557 $self->rmdirs if $_rmdir;
4558 if (@$mods == 0) {
4559 $self->abort_edit;
4560 } else {
4561 $self->close_edit;
4562 }
4563 return scalar @$mods;
4564 }
4565
4566 package Git::SVN::Ra;
4567 use vars qw/@ISA $config_dir $_log_window_size/;
4568 use strict;
4569 use warnings;
4570 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4571
4572 BEGIN {
4573 # enforce temporary pool usage for some simple functions
4574 no strict 'refs';
4575 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4576 get_file/) {
4577 my $SUPER = "SUPER::$f";
4578 *$f = sub {
4579 my $self = shift;
4580 my $pool = SVN::Pool->new;
4581 my @ret = $self->$SUPER(@_,$pool);
4582 $pool->clear;
4583 wantarray ? @ret : $ret[0];
4584 };
4585 }
4586 }
4587
4588 sub _auth_providers () {
4589 [
4590 SVN::Client::get_simple_provider(),
4591 SVN::Client::get_ssl_server_trust_file_provider(),
4592 SVN::Client::get_simple_prompt_provider(
4593 \&Git::SVN::Prompt::simple, 2),
4594 SVN::Client::get_ssl_client_cert_file_provider(),
4595 SVN::Client::get_ssl_client_cert_prompt_provider(
4596 \&Git::SVN::Prompt::ssl_client_cert, 2),
4597 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4598 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4599 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4600 SVN::Client::get_username_provider(),
4601 SVN::Client::get_ssl_server_trust_prompt_provider(
4602 \&Git::SVN::Prompt::ssl_server_trust),
4603 SVN::Client::get_username_prompt_provider(
4604 \&Git::SVN::Prompt::username, 2)
4605 ]
4606 }
4607
4608 sub escape_uri_only {
4609 my ($uri) = @_;
4610 my @tmp;
4611 foreach (split m{/}, $uri) {
4612 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4613 push @tmp, $_;
4614 }
4615 join('/', @tmp);
4616 }
4617
4618 sub escape_url {
4619 my ($url) = @_;
4620 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4621 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4622 $url = "$scheme://$domain$uri";
4623 }
4624 $url;
4625 }
4626
4627 sub new {
4628 my ($class, $url) = @_;
4629 $url =~ s!/+$!!;
4630 return $RA if ($RA && $RA->{url} eq $url);
4631
4632 SVN::_Core::svn_config_ensure($config_dir, undef);
4633 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4634 my $config = SVN::Core::config_get_config($config_dir);
4635 $RA = undef;
4636 my $dont_store_passwords = 1;
4637 my $conf_t = ${$config}{'config'};
4638 {
4639 no warnings 'once';
4640 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4641 # produces warnings that variables are used only once.
4642 # I had not found the better way to shut them up, so
4643 # the warnings of type 'once' are disabled in this block.
4644 if (SVN::_Core::svn_config_get_bool($conf_t,
4645 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4646 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4647 1) == 0) {
4648 SVN::_Core::svn_auth_set_parameter($baton,
4649 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4650 bless (\$dont_store_passwords, "_p_void"));
4651 }
4652 if (SVN::_Core::svn_config_get_bool($conf_t,
4653 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4654 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4655 1) == 0) {
4656 $Git::SVN::Prompt::_no_auth_cache = 1;
4657 }
4658 } # no warnings 'once'
4659 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4660 config => $config,
4661 pool => SVN::Pool->new,
4662 auth_provider_callbacks => $callbacks);
4663 $self->{url} = $url;
4664 $self->{svn_path} = $url;
4665 $self->{repos_root} = $self->get_repos_root;
4666 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4667 $self->{cache} = { check_path => { r => 0, data => {} },
4668 get_dir => { r => 0, data => {} } };
4669 $RA = bless $self, $class;
4670 }
4671
4672 sub check_path {
4673 my ($self, $path, $r) = @_;
4674 my $cache = $self->{cache}->{check_path};
4675 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4676 return $cache->{data}->{$path};
4677 }
4678 my $pool = SVN::Pool->new;
4679 my $t = $self->SUPER::check_path($path, $r, $pool);
4680 $pool->clear;
4681 if ($r != $cache->{r}) {
4682 %{$cache->{data}} = ();
4683 $cache->{r} = $r;
4684 }
4685 $cache->{data}->{$path} = $t;
4686 }
4687
4688 sub get_dir {
4689 my ($self, $dir, $r) = @_;
4690 my $cache = $self->{cache}->{get_dir};
4691 if ($r == $cache->{r}) {
4692 if (my $x = $cache->{data}->{$dir}) {
4693 return wantarray ? @$x : $x->[0];
4694 }
4695 }
4696 my $pool = SVN::Pool->new;
4697 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4698 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4699 $pool->clear;
4700 if ($r != $cache->{r}) {
4701 %{$cache->{data}} = ();
4702 $cache->{r} = $r;
4703 }
4704 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4705 wantarray ? (\%dirents, $r, $props) : \%dirents;
4706 }
4707
4708 sub DESTROY {
4709 # do not call the real DESTROY since we store ourselves in $RA
4710 }
4711
4712 # get_log(paths, start, end, limit,
4713 # discover_changed_paths, strict_node_history, receiver)
4714 sub get_log {
4715 my ($self, @args) = @_;
4716 my $pool = SVN::Pool->new;
4717
4718 # svn_log_changed_path_t objects passed to get_log are likely to be
4719 # overwritten even if only the refs are copied to an external variable,
4720 # so we should dup the structures in their entirety. Using an
4721 # externally passed pool (instead of our temporary and quickly cleared
4722 # pool in Git::SVN::Ra) does not help matters at all...
4723 my $receiver = pop @args;
4724 my $prefix = "/".$self->{svn_path};
4725 $prefix =~ s#/+($)##;
4726 my $prefix_regex = qr#^\Q$prefix\E#;
4727 push(@args, sub {
4728 my ($paths) = $_[0];
4729 return &$receiver(@_) unless $paths;
4730 $_[0] = ();
4731 foreach my $p (keys %$paths) {
4732 my $i = $paths->{$p};
4733 # Make path relative to our url, not repos_root
4734 $p =~ s/$prefix_regex//;
4735 my %s = map { $_ => $i->$_; }
4736 qw/copyfrom_path copyfrom_rev action/;
4737 if ($s{'copyfrom_path'}) {
4738 $s{'copyfrom_path'} =~ s/$prefix_regex//;
4739 }
4740 $_[0]{$p} = \%s;
4741 }
4742 &$receiver(@_);
4743 });
4744
4745
4746 # the limit parameter was not supported in SVN 1.1.x, so we
4747 # drop it. Therefore, the receiver callback passed to it
4748 # is made aware of this limitation by being wrapped if
4749 # the limit passed to is being wrapped.
4750 if ($SVN::Core::VERSION le '1.2.0') {
4751 my $limit = splice(@args, 3, 1);
4752 if ($limit > 0) {
4753 my $receiver = pop @args;
4754 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4755 }
4756 }
4757 my $ret = $self->SUPER::get_log(@args, $pool);
4758 $pool->clear;
4759 $ret;
4760 }
4761
4762 sub trees_match {
4763 my ($self, $url1, $rev1, $url2, $rev2) = @_;
4764 my $ctx = SVN::Client->new(auth => _auth_providers);
4765 my $out = IO::File->new_tmpfile;
4766
4767 # older SVN (1.1.x) doesn't take $pool as the last parameter for
4768 # $ctx->diff(), so we'll create a default one
4769 my $pool = SVN::Pool->new_default_sub;
4770
4771 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4772 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4773 $out->flush;
4774 my $ret = (($out->stat)[7] == 0);
4775 close $out or croak $!;
4776
4777 $ret;
4778 }
4779
4780 sub get_commit_editor {
4781 my ($self, $log, $cb, $pool) = @_;
4782 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4783 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4784 }
4785
4786 sub gs_do_update {
4787 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4788 my $new = ($rev_a == $rev_b);
4789 my $path = $gs->{path};
4790
4791 if ($new && -e $gs->{index}) {
4792 unlink $gs->{index} or die
4793 "Couldn't unlink index: $gs->{index}: $!\n";
4794 }
4795 my $pool = SVN::Pool->new;
4796 $editor->set_path_strip($path);
4797 my (@pc) = split m#/#, $path;
4798 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4799 1, $editor, $pool);
4800 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4801
4802 # Since we can't rely on svn_ra_reparent being available, we'll
4803 # just have to do some magic with set_path to make it so
4804 # we only want a partial path.
4805 my $sp = '';
4806 my $final = join('/', @pc);
4807 while (@pc) {
4808 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4809 $sp .= '/' if length $sp;
4810 $sp .= shift @pc;
4811 }
4812 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4813
4814 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4815
4816 $reporter->finish_report($pool);
4817 $pool->clear;
4818 $editor->{git_commit_ok};
4819 }
4820
4821 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4822 # svn_ra_reparent didn't work before 1.4)
4823 sub gs_do_switch {
4824 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4825 my $path = $gs->{path};
4826 my $pool = SVN::Pool->new;
4827
4828 my $full_url = $self->{url};
4829 my $old_url = $full_url;
4830 $full_url .= '/' . $path if length $path;
4831 my ($ra, $reparented);
4832
4833 if ($old_url =~ m#^svn(\+ssh)?://# ||
4834 ($full_url =~ m#^https?://# &&
4835 escape_url($full_url) ne $full_url)) {
4836 $_[0] = undef;
4837 $self = undef;
4838 $RA = undef;
4839 $ra = Git::SVN::Ra->new($full_url);
4840 $ra_invalid = 1;
4841 } elsif ($old_url ne $full_url) {
4842 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4843 $self->{url} = $full_url;
4844 $reparented = 1;
4845 }
4846
4847 $ra ||= $self;
4848 $url_b = escape_url($url_b);
4849 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4850 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4851 $reporter->set_path('', $rev_a, 0, @lock, $pool);
4852 $reporter->finish_report($pool);
4853
4854 if ($reparented) {
4855 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4856 $self->{url} = $old_url;
4857 }
4858
4859 $pool->clear;
4860 $editor->{git_commit_ok};
4861 }
4862
4863 sub longest_common_path {
4864 my ($gsv, $globs) = @_;
4865 my %common;
4866 my $common_max = scalar @$gsv;
4867
4868 foreach my $gs (@$gsv) {
4869 my @tmp = split m#/#, $gs->{path};
4870 my $p = '';
4871 foreach (@tmp) {
4872 $p .= length($p) ? "/$_" : $_;
4873 $common{$p} ||= 0;
4874 $common{$p}++;
4875 }
4876 }
4877 $globs ||= [];
4878 $common_max += scalar @$globs;
4879 foreach my $glob (@$globs) {
4880 my @tmp = split m#/#, $glob->{path}->{left};
4881 my $p = '';
4882 foreach (@tmp) {
4883 $p .= length($p) ? "/$_" : $_;
4884 $common{$p} ||= 0;
4885 $common{$p}++;
4886 }
4887 }
4888
4889 my $longest_path = '';
4890 foreach (sort {length $b <=> length $a} keys %common) {
4891 if ($common{$_} == $common_max) {
4892 $longest_path = $_;
4893 last;
4894 }
4895 }
4896 $longest_path;
4897 }
4898
4899 sub gs_fetch_loop_common {
4900 my ($self, $base, $head, $gsv, $globs) = @_;
4901 return if ($base > $head);
4902 my $inc = $_log_window_size;
4903 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4904 my $longest_path = longest_common_path($gsv, $globs);
4905 my $ra_url = $self->{url};
4906 my $find_trailing_edge;
4907 while (1) {
4908 my %revs;
4909 my $err;
4910 my $err_handler = $SVN::Error::handler;
4911 $SVN::Error::handler = sub {
4912 ($err) = @_;
4913 skip_unknown_revs($err);
4914 };
4915 sub _cb {
4916 my ($paths, $r, $author, $date, $log) = @_;
4917 [ $paths,
4918 { author => $author, date => $date, log => $log } ];
4919 }
4920 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4921 sub { $revs{$_[1]} = _cb(@_) });
4922 if ($err) {
4923 print "Checked through r$max\r";
4924 } else {
4925 $find_trailing_edge = 1;
4926 }
4927 if ($err and $find_trailing_edge) {
4928 print STDERR "Path '$longest_path' ",
4929 "was probably deleted:\n",
4930 $err->expanded_message,
4931 "\nWill attempt to follow ",
4932 "revisions r$min .. r$max ",
4933 "committed before the deletion\n";
4934 my $hi = $max;
4935 while (--$hi >= $min) {
4936 my $ok;
4937 $self->get_log([$longest_path], $min, $hi,
4938 0, 1, 1, sub {
4939 $ok = $_[1];
4940 $revs{$_[1]} = _cb(@_) });
4941 if ($ok) {
4942 print STDERR "r$min .. r$ok OK\n";
4943 last;
4944 }
4945 }
4946 $find_trailing_edge = 0;
4947 }
4948 $SVN::Error::handler = $err_handler;
4949
4950 my %exists = map { $_->{path} => $_ } @$gsv;
4951 foreach my $r (sort {$a <=> $b} keys %revs) {
4952 my ($paths, $logged) = @{$revs{$r}};
4953
4954 foreach my $gs ($self->match_globs(\%exists, $paths,
4955 $globs, $r)) {
4956 if ($gs->rev_map_max >= $r) {
4957 next;
4958 }
4959 next unless $gs->match_paths($paths, $r);
4960 $gs->{logged_rev_props} = $logged;
4961 if (my $last_commit = $gs->last_commit) {
4962 $gs->assert_index_clean($last_commit);
4963 }
4964 my $log_entry = $gs->do_fetch($paths, $r);
4965 if ($log_entry) {
4966 $gs->do_git_commit($log_entry);
4967 }
4968 $INDEX_FILES{$gs->{index}} = 1;
4969 }
4970 foreach my $g (@$globs) {
4971 my $k = "svn-remote.$g->{remote}." .
4972 "$g->{t}-maxRev";
4973 Git::SVN::tmp_config($k, $r);
4974 }
4975 if ($ra_invalid) {
4976 $_[0] = undef;
4977 $self = undef;
4978 $RA = undef;
4979 $self = Git::SVN::Ra->new($ra_url);
4980 $ra_invalid = undef;
4981 }
4982 }
4983 # pre-fill the .rev_db since it'll eventually get filled in
4984 # with '0' x40 if something new gets committed
4985 foreach my $gs (@$gsv) {
4986 next if $gs->rev_map_max >= $max;
4987 next if defined $gs->rev_map_get($max);
4988 $gs->rev_map_set($max, 0 x40);
4989 }
4990 foreach my $g (@$globs) {
4991 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4992 Git::SVN::tmp_config($k, $max);
4993 }
4994 last if $max >= $head;
4995 $min = $max + 1;
4996 $max += $inc;
4997 $max = $head if ($max > $head);
4998 }
4999 Git::SVN::gc();
5000 }
5001
5002 sub get_dir_globbed {
5003 my ($self, $left, $depth, $r) = @_;
5004
5005 my @x = eval { $self->get_dir($left, $r) };
5006 return unless scalar @x == 3;
5007 my $dirents = $x[0];
5008 my @finalents;
5009 foreach my $de (keys %$dirents) {
5010 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5011 if ($depth > 1) {
5012 my @args = ("$left/$de", $depth - 1, $r);
5013 foreach my $dir ($self->get_dir_globbed(@args)) {
5014 push @finalents, "$de/$dir";
5015 }
5016 } else {
5017 push @finalents, $de;
5018 }
5019 }
5020 @finalents;
5021 }
5022
5023 sub match_globs {
5024 my ($self, $exists, $paths, $globs, $r) = @_;
5025
5026 sub get_dir_check {
5027 my ($self, $exists, $g, $r) = @_;
5028
5029 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5030 $g->{path}->{depth},
5031 $r);
5032
5033 foreach my $de (@dirs) {
5034 my $p = $g->{path}->full_path($de);
5035 next if $exists->{$p};
5036 next if (length $g->{path}->{right} &&
5037 ($self->check_path($p, $r) !=
5038 $SVN::Node::dir));
5039 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5040 $g->{ref}->full_path($de), 1);
5041 }
5042 }
5043 foreach my $g (@$globs) {
5044 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5045 if ($path->{action} =~ /^[AR]$/) {
5046 get_dir_check($self, $exists, $g, $r);
5047 }
5048 }
5049 foreach (keys %$paths) {
5050 if (/$g->{path}->{left_regex}/ &&
5051 !/$g->{path}->{regex}/) {
5052 next if $paths->{$_}->{action} !~ /^[AR]$/;
5053 get_dir_check($self, $exists, $g, $r);
5054 }
5055 next unless /$g->{path}->{regex}/;
5056 my $p = $1;
5057 my $pathname = $g->{path}->full_path($p);
5058 next if $exists->{$pathname};
5059 next if ($self->check_path($pathname, $r) !=
5060 $SVN::Node::dir);
5061 $exists->{$pathname} = Git::SVN->init(
5062 $self->{url}, $pathname, undef,
5063 $g->{ref}->full_path($p), 1);
5064 }
5065 my $c = '';
5066 foreach (split m#/#, $g->{path}->{left}) {
5067 $c .= "/$_";
5068 next unless ($paths->{$c} &&
5069 ($paths->{$c}->{action} =~ /^[AR]$/));
5070 get_dir_check($self, $exists, $g, $r);
5071 }
5072 }
5073 values %$exists;
5074 }
5075
5076 sub minimize_url {
5077 my ($self) = @_;
5078 return $self->{url} if ($self->{url} eq $self->{repos_root});
5079 my $url = $self->{repos_root};
5080 my @components = split(m!/!, $self->{svn_path});
5081 my $c = '';
5082 do {
5083 $url .= "/$c" if length $c;
5084 eval {
5085 my $ra = (ref $self)->new($url);
5086 my $latest = $ra->get_latest_revnum;
5087 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5088 };
5089 } while ($@ && ($c = shift @components));
5090 $url;
5091 }
5092
5093 sub can_do_switch {
5094 my $self = shift;
5095 unless (defined $can_do_switch) {
5096 my $pool = SVN::Pool->new;
5097 my $rep = eval {
5098 $self->do_switch(1, '', 0, $self->{url},
5099 SVN::Delta::Editor->new, $pool);
5100 };
5101 if ($@) {
5102 $can_do_switch = 0;
5103 } else {
5104 $rep->abort_report($pool);
5105 $can_do_switch = 1;
5106 }
5107 $pool->clear;
5108 }
5109 $can_do_switch;
5110 }
5111
5112 sub skip_unknown_revs {
5113 my ($err) = @_;
5114 my $errno = $err->apr_err();
5115 # Maybe the branch we're tracking didn't
5116 # exist when the repo started, so it's
5117 # not an error if it doesn't, just continue
5118 #
5119 # Wonderfully consistent library, eh?
5120 # 160013 - svn:// and file://
5121 # 175002 - http(s)://
5122 # 175007 - http(s):// (this repo required authorization, too...)
5123 # More codes may be discovered later...
5124 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5125 my $err_key = $err->expanded_message;
5126 # revision numbers change every time, filter them out
5127 $err_key =~ s/\d+/\0/g;
5128 $err_key = "$errno\0$err_key";
5129 unless ($ignored_err{$err_key}) {
5130 warn "W: Ignoring error from SVN, path probably ",
5131 "does not exist: ($errno): ",
5132 $err->expanded_message,"\n";
5133 warn "W: Do not be alarmed at the above message ",
5134 "git-svn is just searching aggressively for ",
5135 "old history.\n",
5136 "This may take a while on large repositories\n";
5137 $ignored_err{$err_key} = 1;
5138 }
5139 return;
5140 }
5141 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5142 }
5143
5144 package Git::SVN::Log;
5145 use strict;
5146 use warnings;
5147 use POSIX qw/strftime/;
5148 use Time::Local;
5149 use constant commit_log_separator => ('-' x 72) . "\n";
5150 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5151 %rusers $show_commit $incremental/;
5152 my $l_fmt;
5153
5154 sub cmt_showable {
5155 my ($c) = @_;
5156 return 1 if defined $c->{r};
5157
5158 # big commit message got truncated by the 16k pretty buffer in rev-list
5159 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5160 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5161 @{$c->{l}} = ();
5162 my @log = command(qw/cat-file commit/, $c->{c});
5163
5164 # shift off the headers
5165 shift @log while ($log[0] ne '');
5166 shift @log;
5167
5168 # TODO: make $c->{l} not have a trailing newline in the future
5169 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5170
5171 (undef, $c->{r}, undef) = ::extract_metadata(
5172 (grep(/^git-svn-id: /, @log))[-1]);
5173 }
5174 return defined $c->{r};
5175 }
5176
5177 sub log_use_color {
5178 return $color || Git->repository->get_colorbool('color.diff');
5179 }
5180
5181 sub git_svn_log_cmd {
5182 my ($r_min, $r_max, @args) = @_;
5183 my $head = 'HEAD';
5184 my (@files, @log_opts);
5185 foreach my $x (@args) {
5186 if ($x eq '--' || @files) {
5187 push @files, $x;
5188 } else {
5189 if (::verify_ref("$x^0")) {
5190 $head = $x;
5191 } else {
5192 push @log_opts, $x;
5193 }
5194 }
5195 }
5196
5197 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5198 $gs ||= Git::SVN->_new;
5199 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5200 $gs->refname);
5201 push @cmd, '-r' unless $non_recursive;
5202 push @cmd, qw/--raw --name-status/ if $verbose;
5203 push @cmd, '--color' if log_use_color();
5204 push @cmd, @log_opts;
5205 if (defined $r_max && $r_max == $r_min) {
5206 push @cmd, '--max-count=1';
5207 if (my $c = $gs->rev_map_get($r_max)) {
5208 push @cmd, $c;
5209 }
5210 } elsif (defined $r_max) {
5211 if ($r_max < $r_min) {
5212 ($r_min, $r_max) = ($r_max, $r_min);
5213 }
5214 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5215 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5216 # If there are no commits in the range, both $c_max and $c_min
5217 # will be undefined. If there is at least 1 commit in the
5218 # range, both will be defined.
5219 return () if !defined $c_min || !defined $c_max;
5220 if ($c_min eq $c_max) {
5221 push @cmd, '--max-count=1', $c_min;
5222 } else {
5223 push @cmd, '--boundary', "$c_min..$c_max";
5224 }
5225 }
5226 return (@cmd, @files);
5227 }
5228
5229 # adapted from pager.c
5230 sub config_pager {
5231 chomp(my $pager = command_oneline(qw(var GIT_PAGER)));
5232 if ($pager eq 'cat') {
5233 $pager = undef;
5234 }
5235 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5236 }
5237
5238 sub run_pager {
5239 return unless -t *STDOUT && defined $pager;
5240 pipe my ($rfd, $wfd) or return;
5241 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5242 if (!$pid) {
5243 open STDOUT, '>&', $wfd or
5244 ::fatal "Can't redirect to stdout: $!";
5245 return;
5246 }
5247 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5248 $ENV{LESS} ||= 'FRSX';
5249 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5250 }
5251
5252 sub format_svn_date {
5253 # some systmes don't handle or mishandle %z, so be creative.
5254 my $t = shift || time;
5255 my $gm = timelocal(gmtime($t));
5256 my $sign = qw( + + - )[ $t <=> $gm ];
5257 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5258 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5259 }
5260
5261 sub parse_git_date {
5262 my ($t, $tz) = @_;
5263 # Date::Parse isn't in the standard Perl distro :(
5264 if ($tz =~ s/^\+//) {
5265 $t += tz_to_s_offset($tz);
5266 } elsif ($tz =~ s/^\-//) {
5267 $t -= tz_to_s_offset($tz);
5268 }
5269 return $t;
5270 }
5271
5272 sub set_local_timezone {
5273 if (defined $TZ) {
5274 $ENV{TZ} = $TZ;
5275 } else {
5276 delete $ENV{TZ};
5277 }
5278 }
5279
5280 sub tz_to_s_offset {
5281 my ($tz) = @_;
5282 $tz =~ s/(\d\d)$//;
5283 return ($1 * 60) + ($tz * 3600);
5284 }
5285
5286 sub get_author_info {
5287 my ($dest, $author, $t, $tz) = @_;
5288 $author =~ s/(?:^\s*|\s*$)//g;
5289 $dest->{a_raw} = $author;
5290 my $au;
5291 if ($::_authors) {
5292 $au = $rusers{$author} || undef;
5293 }
5294 if (!$au) {
5295 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5296 }
5297 $dest->{t} = $t;
5298 $dest->{tz} = $tz;
5299 $dest->{a} = $au;
5300 $dest->{t_utc} = parse_git_date($t, $tz);
5301 }
5302
5303 sub process_commit {
5304 my ($c, $r_min, $r_max, $defer) = @_;
5305 if (defined $r_min && defined $r_max) {
5306 if ($r_min == $c->{r} && $r_min == $r_max) {
5307 show_commit($c);
5308 return 0;
5309 }
5310 return 1 if $r_min == $r_max;
5311 if ($r_min < $r_max) {
5312 # we need to reverse the print order
5313 return 0 if (defined $limit && --$limit < 0);
5314 push @$defer, $c;
5315 return 1;
5316 }
5317 if ($r_min != $r_max) {
5318 return 1 if ($r_min < $c->{r});
5319 return 1 if ($r_max > $c->{r});
5320 }
5321 }
5322 return 0 if (defined $limit && --$limit < 0);
5323 show_commit($c);
5324 return 1;
5325 }
5326
5327 sub show_commit {
5328 my $c = shift;
5329 if ($oneline) {
5330 my $x = "\n";
5331 if (my $l = $c->{l}) {
5332 while ($l->[0] =~ /^\s*$/) { shift @$l }
5333 $x = $l->[0];
5334 }
5335 $l_fmt ||= 'A' . length($c->{r});
5336 print 'r',pack($l_fmt, $c->{r}),' | ';
5337 print "$c->{c} | " if $show_commit;
5338 print $x;
5339 } else {
5340 show_commit_normal($c);
5341 }
5342 }
5343
5344 sub show_commit_changed_paths {
5345 my ($c) = @_;
5346 return unless $c->{changed};
5347 print "Changed paths:\n", @{$c->{changed}};
5348 }
5349
5350 sub show_commit_normal {
5351 my ($c) = @_;
5352 print commit_log_separator, "r$c->{r} | ";
5353 print "$c->{c} | " if $show_commit;
5354 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5355 my $nr_line = 0;
5356
5357 if (my $l = $c->{l}) {
5358 while ($l->[$#$l] eq "\n" && $#$l > 0
5359 && $l->[($#$l - 1)] eq "\n") {
5360 pop @$l;
5361 }
5362 $nr_line = scalar @$l;
5363 if (!$nr_line) {
5364 print "1 line\n\n\n";
5365 } else {
5366 if ($nr_line == 1) {
5367 $nr_line = '1 line';
5368 } else {
5369 $nr_line .= ' lines';
5370 }
5371 print $nr_line, "\n";
5372 show_commit_changed_paths($c);
5373 print "\n";
5374 print $_ foreach @$l;
5375 }
5376 } else {
5377 print "1 line\n";
5378 show_commit_changed_paths($c);
5379 print "\n";
5380
5381 }
5382 foreach my $x (qw/raw stat diff/) {
5383 if ($c->{$x}) {
5384 print "\n";
5385 print $_ foreach @{$c->{$x}}
5386 }
5387 }
5388 }
5389
5390 sub cmd_show_log {
5391 my (@args) = @_;
5392 my ($r_min, $r_max);
5393 my $r_last = -1; # prevent dupes
5394 set_local_timezone();
5395 if (defined $::_revision) {
5396 if ($::_revision =~ /^(\d+):(\d+)$/) {
5397 ($r_min, $r_max) = ($1, $2);
5398 } elsif ($::_revision =~ /^\d+$/) {
5399 $r_min = $r_max = $::_revision;
5400 } else {
5401 ::fatal "-r$::_revision is not supported, use ",
5402 "standard 'git log' arguments instead";
5403 }
5404 }
5405
5406 config_pager();
5407 @args = git_svn_log_cmd($r_min, $r_max, @args);
5408 if (!@args) {
5409 print commit_log_separator unless $incremental || $oneline;
5410 return;
5411 }
5412 my $log = command_output_pipe(@args);
5413 run_pager();
5414 my (@k, $c, $d, $stat);
5415 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5416 while (<$log>) {
5417 if (/^${esc_color}commit -?($::sha1_short)/o) {
5418 my $cmt = $1;
5419 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5420 $r_last = $c->{r};
5421 process_commit($c, $r_min, $r_max, \@k) or
5422 goto out;
5423 }
5424 $d = undef;
5425 $c = { c => $cmt };
5426 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5427 get_author_info($c, $1, $2, $3);
5428 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5429 # ignore
5430 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5431 push @{$c->{raw}}, $_;
5432 } elsif (/^${esc_color}[ACRMDT]\t/) {
5433 # we could add $SVN->{svn_path} here, but that requires
5434 # remote access at the moment (repo_path_split)...
5435 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5436 push @{$c->{changed}}, $_;
5437 } elsif (/^${esc_color}diff /o) {
5438 $d = 1;
5439 push @{$c->{diff}}, $_;
5440 } elsif ($d) {
5441 push @{$c->{diff}}, $_;
5442 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5443 $esc_color*[\+\-]*$esc_color$/x) {
5444 $stat = 1;
5445 push @{$c->{stat}}, $_;
5446 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5447 push @{$c->{stat}}, $_;
5448 $stat = undef;
5449 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5450 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5451 } elsif (s/^${esc_color} //o) {
5452 push @{$c->{l}}, $_;
5453 }
5454 }
5455 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5456 $r_last = $c->{r};
5457 process_commit($c, $r_min, $r_max, \@k);
5458 }
5459 if (@k) {
5460 ($r_min, $r_max) = ($r_max, $r_min);
5461 process_commit($_, $r_min, $r_max) foreach reverse @k;
5462 }
5463 out:
5464 close $log;
5465 print commit_log_separator unless $incremental || $oneline;
5466 }
5467
5468 sub cmd_blame {
5469 my $path = pop;
5470
5471 config_pager();
5472 run_pager();
5473
5474 my ($fh, $ctx, $rev);
5475
5476 if ($_git_format) {
5477 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5478 while (my $line = <$fh>) {
5479 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5480 # Uncommitted edits show up as a rev ID of
5481 # all zeros, which we can't look up with
5482 # cmt_metadata
5483 if ($1 !~ /^0+$/) {
5484 (undef, $rev, undef) =
5485 ::cmt_metadata($1);
5486 $rev = '0' if (!$rev);
5487 } else {
5488 $rev = '0';
5489 }
5490 $rev = sprintf('%-10s', $rev);
5491 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5492 }
5493 print $line;
5494 }
5495 } else {
5496 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5497 '--', $path);
5498 my ($sha1);
5499 my %authors;
5500 my @buffer;
5501 my %dsha; #distinct sha keys
5502
5503 while (my $line = <$fh>) {
5504 push @buffer, $line;
5505 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5506 $dsha{$1} = 1;
5507 }
5508 }
5509
5510 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5511
5512 foreach my $line (@buffer) {
5513 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5514 $rev = $s2r->{$1};
5515 $rev = '0' if (!$rev)
5516 }
5517 elsif ($line =~ /^author (.*)/) {
5518 $authors{$rev} = $1;
5519 $authors{$rev} =~ s/\s/_/g;
5520 }
5521 elsif ($line =~ /^\t(.*)$/) {
5522 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5523 }
5524 }
5525 }
5526 command_close_pipe($fh, $ctx);
5527 }
5528
5529 package Git::SVN::Migration;
5530 # these version numbers do NOT correspond to actual version numbers
5531 # of git nor git-svn. They are just relative.
5532 #
5533 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5534 #
5535 # v1 layout: .git/$id/info/url, refs/remotes/$id
5536 #
5537 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5538 #
5539 # v3 layout: .git/svn/$id, refs/remotes/$id
5540 # - info/url may remain for backwards compatibility
5541 # - this is what we migrate up to this layout automatically,
5542 # - this will be used by git svn init on single branches
5543 # v3.1 layout (auto migrated):
5544 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5545 # for backwards compatibility
5546 #
5547 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5548 # - this is only created for newly multi-init-ed
5549 # repositories. Similar in spirit to the
5550 # --use-separate-remotes option in git-clone (now default)
5551 # - we do not automatically migrate to this (following
5552 # the example set by core git)
5553 #
5554 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
5555 # - newer, more-efficient format that uses 24-bytes per record
5556 # with no filler space.
5557 # - use xxd -c24 < .rev_map.$UUID to view and debug
5558 # - This is a one-way migration, repositories updated to the
5559 # new format will not be able to use old git-svn without
5560 # rebuilding the .rev_db. Rebuilding the rev_db is not
5561 # possible if noMetadata or useSvmProps are set; but should
5562 # be no problem for users that use the (sensible) defaults.
5563 use strict;
5564 use warnings;
5565 use Carp qw/croak/;
5566 use File::Path qw/mkpath/;
5567 use File::Basename qw/dirname basename/;
5568 use vars qw/$_minimize/;
5569
5570 sub migrate_from_v0 {
5571 my $git_dir = $ENV{GIT_DIR};
5572 return undef unless -d $git_dir;
5573 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5574 my $migrated = 0;
5575 while (<$fh>) {
5576 chomp;
5577 my ($id, $orig_ref) = ($_, $_);
5578 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5579 next unless -f "$git_dir/$id/info/url";
5580 my $new_ref = "refs/remotes/$id";
5581 if (::verify_ref("$new_ref^0")) {
5582 print STDERR "W: $orig_ref is probably an old ",
5583 "branch used by an ancient version of ",
5584 "git-svn.\n",
5585 "However, $new_ref also exists.\n",
5586 "We will not be able ",
5587 "to use this branch until this ",
5588 "ambiguity is resolved.\n";
5589 next;
5590 }
5591 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5592 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5593 command_noisy('update-ref', $new_ref, $orig_ref);
5594 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5595 $migrated++;
5596 }
5597 command_close_pipe($fh, $ctx);
5598 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5599 $migrated;
5600 }
5601
5602 sub migrate_from_v1 {
5603 my $git_dir = $ENV{GIT_DIR};
5604 my $migrated = 0;
5605 return $migrated unless -d $git_dir;
5606 my $svn_dir = "$git_dir/svn";
5607
5608 # just in case somebody used 'svn' as their $id at some point...
5609 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5610
5611 print STDERR "Migrating from a git-svn v1 layout...\n";
5612 mkpath([$svn_dir]);
5613 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5614 "$svn_dir\n\t(required for this version ",
5615 "($::VERSION) of git-svn) does not exist.\n";
5616 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5617 while (<$fh>) {
5618 my $x = $_;
5619 next unless $x =~ s#^refs/remotes/##;
5620 chomp $x;
5621 next unless -f "$git_dir/$x/info/url";
5622 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5623 next unless $u;
5624 my $dn = dirname("$git_dir/svn/$x");
5625 mkpath([$dn]) unless -d $dn;
5626 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5627 mkpath(["$git_dir/svn/svn"]);
5628 print STDERR " - $git_dir/$x/info => ",
5629 "$git_dir/svn/$x/info\n";
5630 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5631 croak "$!: $x";
5632 # don't worry too much about these, they probably
5633 # don't exist with repos this old (save for index,
5634 # and we can easily regenerate that)
5635 foreach my $f (qw/unhandled.log index .rev_db/) {
5636 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5637 }
5638 } else {
5639 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5640 rename "$git_dir/$x", "$git_dir/svn/$x" or
5641 croak "$!: $x";
5642 }
5643 $migrated++;
5644 }
5645 command_close_pipe($fh, $ctx);
5646 print STDERR "Done migrating from a git-svn v1 layout\n";
5647 $migrated;
5648 }
5649
5650 sub read_old_urls {
5651 my ($l_map, $pfx, $path) = @_;
5652 my @dir;
5653 foreach (<$path/*>) {
5654 if (-r "$_/info/url") {
5655 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5656 my $ref_id = $pfx . basename $_;
5657 my $url = ::file_to_s("$_/info/url");
5658 $l_map->{$ref_id} = $url;
5659 } elsif (-d $_) {
5660 push @dir, $_;
5661 }
5662 }
5663 foreach (@dir) {
5664 my $x = $_;
5665 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5666 read_old_urls($l_map, $x, $_);
5667 }
5668 }
5669
5670 sub migrate_from_v2 {
5671 my @cfg = command(qw/config -l/);
5672 return if grep /^svn-remote\..+\.url=/, @cfg;
5673 my %l_map;
5674 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5675 my $migrated = 0;
5676
5677 foreach my $ref_id (sort keys %l_map) {
5678 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5679 if ($@) {
5680 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5681 }
5682 $migrated++;
5683 }
5684 $migrated;
5685 }
5686
5687 sub minimize_connections {
5688 my $r = Git::SVN::read_all_remotes();
5689 my $new_urls = {};
5690 my $root_repos = {};
5691 foreach my $repo_id (keys %$r) {
5692 my $url = $r->{$repo_id}->{url} or next;
5693 my $fetch = $r->{$repo_id}->{fetch} or next;
5694 my $ra = Git::SVN::Ra->new($url);
5695
5696 # skip existing cases where we already connect to the root
5697 if (($ra->{url} eq $ra->{repos_root}) ||
5698 ($ra->{repos_root} eq $repo_id)) {
5699 $root_repos->{$ra->{url}} = $repo_id;
5700 next;
5701 }
5702
5703 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5704 my $root_path = $ra->{url};
5705 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5706 foreach my $path (keys %$fetch) {
5707 my $ref_id = $fetch->{$path};
5708 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5709
5710 # make sure we can read when connecting to
5711 # a higher level of a repository
5712 my ($last_rev, undef) = $gs->last_rev_commit;
5713 if (!defined $last_rev) {
5714 $last_rev = eval {
5715 $root_ra->get_latest_revnum;
5716 };
5717 next if $@;
5718 }
5719 my $new = $root_path;
5720 $new .= length $path ? "/$path" : '';
5721 eval {
5722 $root_ra->get_log([$new], $last_rev, $last_rev,
5723 0, 0, 1, sub { });
5724 };
5725 next if $@;
5726 $new_urls->{$ra->{repos_root}}->{$new} =
5727 { ref_id => $ref_id,
5728 old_repo_id => $repo_id,
5729 old_path => $path };
5730 }
5731 }
5732
5733 my @emptied;
5734 foreach my $url (keys %$new_urls) {
5735 # see if we can re-use an existing [svn-remote "repo_id"]
5736 # instead of creating a(n ugly) new section:
5737 my $repo_id = $root_repos->{$url} || $url;
5738
5739 my $fetch = $new_urls->{$url};
5740 foreach my $path (keys %$fetch) {
5741 my $x = $fetch->{$path};
5742 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5743 my $pfx = "svn-remote.$x->{old_repo_id}";
5744
5745 my $old_fetch = quotemeta("$x->{old_path}:".
5746 "$x->{ref_id}");
5747 command_noisy(qw/config --unset/,
5748 "$pfx.fetch", '^'. $old_fetch . '$');
5749 delete $r->{$x->{old_repo_id}}->
5750 {fetch}->{$x->{old_path}};
5751 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5752 command_noisy(qw/config --unset/,
5753 "$pfx.url");
5754 push @emptied, $x->{old_repo_id}
5755 }
5756 }
5757 }
5758 if (@emptied) {
5759 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5760 print STDERR <<EOF;
5761 The following [svn-remote] sections in your config file ($file) are empty
5762 and can be safely removed:
5763 EOF
5764 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5765 }
5766 }
5767
5768 sub migration_check {
5769 migrate_from_v0();
5770 migrate_from_v1();
5771 migrate_from_v2();
5772 minimize_connections() if $_minimize;
5773 }
5774
5775 package Git::IndexInfo;
5776 use strict;
5777 use warnings;
5778 use Git qw/command_input_pipe command_close_pipe/;
5779
5780 sub new {
5781 my ($class) = @_;
5782 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5783 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5784 }
5785
5786 sub remove {
5787 my ($self, $path) = @_;
5788 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5789 return ++$self->{nr};
5790 }
5791 undef;
5792 }
5793
5794 sub update {
5795 my ($self, $mode, $hash, $path) = @_;
5796 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5797 return ++$self->{nr};
5798 }
5799 undef;
5800 }
5801
5802 sub DESTROY {
5803 my ($self) = @_;
5804 command_close_pipe($self->{gui}, $self->{ctx});
5805 }
5806
5807 package Git::SVN::GlobSpec;
5808 use strict;
5809 use warnings;
5810
5811 sub new {
5812 my ($class, $glob) = @_;
5813 my $re = $glob;
5814 $re =~ s!/+$!!g; # no need for trailing slashes
5815 $re =~ m!^([^*]*)(\*(?:/\*)*)(.*)$!;
5816 my $temp = $re;
5817 my ($left, $right) = ($1, $3);
5818 $re = $2;
5819 my $depth = $re =~ tr/*/*/;
5820 if ($depth != $temp =~ tr/*/*/) {
5821 die "Only one set of wildcard directories " .
5822 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5823 }
5824 if ($depth == 0) {
5825 die "One '*' is needed for glob: '$glob'\n";
5826 }
5827 $re =~ s!\*!\[^/\]*!g;
5828 $re = quotemeta($left) . "($re)" . quotemeta($right);
5829 if (length $left && !($left =~ s!/+$!!g)) {
5830 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5831 }
5832 if (length $right && !($right =~ s!^/+!!g)) {
5833 die "Missing leading '/' on right side of: '$glob' ($right)\n";
5834 }
5835 my $left_re = qr/^\/\Q$left\E(\/|$)/;
5836 bless { left => $left, right => $right, left_regex => $left_re,
5837 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5838 }
5839
5840 sub full_path {
5841 my ($self, $path) = @_;
5842 return (length $self->{left} ? "$self->{left}/" : '') .
5843 $path . (length $self->{right} ? "/$self->{right}" : '');
5844 }
5845
5846 __END__
5847
5848 Data structures:
5849
5850
5851 $remotes = { # returned by read_all_remotes()
5852 'svn' => {
5853 # svn-remote.svn.url=https://svn.musicpd.org
5854 url => 'https://svn.musicpd.org',
5855 # svn-remote.svn.fetch=mpd/trunk:trunk
5856 fetch => {
5857 'mpd/trunk' => 'trunk',
5858 },
5859 # svn-remote.svn.tags=mpd/tags/*:tags/*
5860 tags => {
5861 path => {
5862 left => 'mpd/tags',
5863 right => '',
5864 regex => qr!mpd/tags/([^/]+)$!,
5865 glob => 'tags/*',
5866 },
5867 ref => {
5868 left => 'tags',
5869 right => '',
5870 regex => qr!tags/([^/]+)$!,
5871 glob => 'tags/*',
5872 },
5873 }
5874 }
5875 };
5876
5877 $log_entry hashref as returned by libsvn_log_entry()
5878 {
5879 log => 'whitespace-formatted log entry
5880 ', # trailing newline is preserved
5881 revision => '8', # integer
5882 date => '2004-02-24T17:01:44.108345Z', # commit date
5883 author => 'committer name'
5884 };
5885
5886
5887 # this is generated by generate_diff();
5888 @mods = array of diff-index line hashes, each element represents one line
5889 of diff-index output
5890
5891 diff-index line ($m hash)
5892 {
5893 mode_a => first column of diff-index output, no leading ':',
5894 mode_b => second column of diff-index output,
5895 sha1_b => sha1sum of the final blob,
5896 chg => change type [MCRADT],
5897 file_a => original file name of a file (iff chg is 'C' or 'R')
5898 file_b => new/current file name of a file (any chg)
5899 }
5900 ;
5901
5902 # retval of read_url_paths{,_all}();
5903 $l_map = {
5904 # repository root url
5905 'https://svn.musicpd.org' => {
5906 # repository path # GIT_SVN_ID
5907 'mpd/trunk' => 'trunk',
5908 'mpd/tags/0.11.5' => 'tags/0.11.5',
5909 },
5910 }
5911
5912 Notes:
5913 I don't trust the each() function on unless I created %hash myself
5914 because the internal iterator may not have started at base.
Jonathan Nieder’s miscellaneous git branches