?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
builtin-cat-file.c: use parse_options()
[git/gitweb-caching.git] / git-svn.perl
blob33e9266fe689609034bb69da694f877ad6191672
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision $_repository
8 $_q $_authors %users/;
9 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10 $VERSION = '@@GIT_VERSION@@';
11
12 # From which subdir have we been invoked?
13 my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15 } || '';
16
17 my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
18 $ENV{GIT_DIR} ||= '.git';
19 $Git::SVN::default_repo_id = 'svn';
20 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
21 $Git::SVN::Ra::_log_window_size = 100;
22
23 $Git::SVN::Log::TZ = $ENV{TZ};
24 $ENV{TZ} = 'UTC';
25 $| = 1; # unbuffer STDOUT
26
27 sub fatal (@) { print STDERR "@_\n"; exit 1 }
28 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
29 require SVN::Ra;
30 require SVN::Delta;
31 if ($SVN::Core::VERSION lt '1.1.0') {
32 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
33 }
34 push @Git::SVN::Ra::ISA, 'SVN::Ra';
35 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
36 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
37 use Carp qw/croak/;
38 use Digest::MD5;
39 use IO::File qw//;
40 use File::Basename qw/dirname basename/;
41 use File::Path qw/mkpath/;
42 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
43 use IPC::Open3;
44 use Git;
45
46 BEGIN {
47 # import functions from Git into our packages, en masse
48 no strict 'refs';
49 foreach (qw/command command_oneline command_noisy command_output_pipe
50 command_input_pipe command_close_pipe/) {
51 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
52 Git::SVN::Migration Git::SVN::Log Git::SVN),
53 __PACKAGE__) {
54 *{"${package}::$_"} = \&{"Git::$_"};
55 }
56 }
57 }
58
59 my ($SVN);
60
61 $sha1 = qr/[a-f\d]{40}/;
62 $sha1_short = qr/[a-f\d]{4,40}/;
63 my ($_stdin, $_help, $_edit,
64 $_message, $_file,
65 $_template, $_shared,
66 $_version, $_fetch_all, $_no_rebase,
67 $_merge, $_strategy, $_dry_run, $_local,
68 $_prefix, $_no_checkout, $_url, $_verbose);
69 $Git::SVN::_follow_parent = 1;
70 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
71 'config-dir=s' => \$Git::SVN::Ra::config_dir,
72 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
73 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
74 'authors-file|A=s' => \$_authors,
75 'repack:i' => \$Git::SVN::_repack,
76 'noMetadata' => \$Git::SVN::_no_metadata,
77 'useSvmProps' => \$Git::SVN::_use_svm_props,
78 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
79 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
80 'no-checkout' => \$_no_checkout,
81 'quiet|q' => \$_q,
82 'repack-flags|repack-args|repack-opts=s' =>
83 \$Git::SVN::_repack_flags,
84 'use-log-author' => \$Git::SVN::_use_log_author,
85 %remote_opts );
86
87 my ($_trunk, $_tags, $_branches, $_stdlayout);
88 my %icv;
89 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
90 'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
91 'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
92 'stdlayout|s' => \$_stdlayout,
93 'minimize-url|m' => \$Git::SVN::_minimize_url,
94 'no-metadata' => sub { $icv{noMetadata} = 1 },
95 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
96 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
97 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
98 %remote_opts );
99 my %cmt_opts = ( 'edit|e' => \$_edit,
100 'rmdir' => \$SVN::Git::Editor::_rmdir,
101 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
102 'l=i' => \$SVN::Git::Editor::_rename_limit,
103 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
104 );
105
106 my %cmd = (
107 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
108 { 'revision|r=s' => \$_revision,
109 'fetch-all|all' => \$_fetch_all,
110 %fc_opts } ],
111 clone => [ \&cmd_clone, "Initialize and fetch revisions",
112 { 'revision|r=s' => \$_revision,
113 %fc_opts, %init_opts } ],
114 init => [ \&cmd_init, "Initialize a repo for tracking" .
115 " (requires URL argument)",
116 \%init_opts ],
117 'multi-init' => [ \&cmd_multi_init,
118 "Deprecated alias for ".
119 "'$0 init -T<trunk> -b<branches> -t<tags>'",
120 \%init_opts ],
121 dcommit => [ \&cmd_dcommit,
122 'Commit several diffs to merge with upstream',
123 { 'merge|m|M' => \$_merge,
124 'strategy|s=s' => \$_strategy,
125 'verbose|v' => \$_verbose,
126 'dry-run|n' => \$_dry_run,
127 'fetch-all|all' => \$_fetch_all,
128 'no-rebase' => \$_no_rebase,
129 %cmt_opts, %fc_opts } ],
130 'set-tree' => [ \&cmd_set_tree,
131 "Set an SVN repository to a git tree-ish",
132 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
133 'create-ignore' => [ \&cmd_create_ignore,
134 'Create a .gitignore per svn:ignore',
135 { 'revision|r=i' => \$_revision
136 } ],
137 'propget' => [ \&cmd_propget,
138 'Print the value of a property on a file or directory',
139 { 'revision|r=i' => \$_revision } ],
140 'proplist' => [ \&cmd_proplist,
141 'List all properties of a file or directory',
142 { 'revision|r=i' => \$_revision } ],
143 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
144 { 'revision|r=i' => \$_revision
145 } ],
146 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
147 { 'revision|r=i' => \$_revision
148 } ],
149 'multi-fetch' => [ \&cmd_multi_fetch,
150 "Deprecated alias for $0 fetch --all",
151 { 'revision|r=s' => \$_revision, %fc_opts } ],
152 'migrate' => [ sub { },
153 # no-op, we automatically run this anyways,
154 'Migrate configuration/metadata/layout from
155 previous versions of git-svn',
156 { 'minimize' => \$Git::SVN::Migration::_minimize,
157 %remote_opts } ],
158 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
159 { 'limit=i' => \$Git::SVN::Log::limit,
160 'revision|r=s' => \$_revision,
161 'verbose|v' => \$Git::SVN::Log::verbose,
162 'incremental' => \$Git::SVN::Log::incremental,
163 'oneline' => \$Git::SVN::Log::oneline,
164 'show-commit' => \$Git::SVN::Log::show_commit,
165 'non-recursive' => \$Git::SVN::Log::non_recursive,
166 'authors-file|A=s' => \$_authors,
167 'color' => \$Git::SVN::Log::color,
168 'pager=s' => \$Git::SVN::Log::pager
169 } ],
170 'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
171 {} ],
172 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
173 { 'merge|m|M' => \$_merge,
174 'verbose|v' => \$_verbose,
175 'strategy|s=s' => \$_strategy,
176 'local|l' => \$_local,
177 'fetch-all|all' => \$_fetch_all,
178 %fc_opts } ],
179 'commit-diff' => [ \&cmd_commit_diff,
180 'Commit a diff between two trees',
181 { 'message|m=s' => \$_message,
182 'file|F=s' => \$_file,
183 'revision|r=s' => \$_revision,
184 %cmt_opts } ],
185 'info' => [ \&cmd_info,
186 "Show info about the latest SVN revision
187 on the current branch",
188 { 'url' => \$_url, } ],
189 'blame' => [ \&Git::SVN::Log::cmd_blame,
190 "Show what revision and author last modified each line of a file",
191 {} ],
192 );
193
194 my $cmd;
195 for (my $i = 0; $i < @ARGV; $i++) {
196 if (defined $cmd{$ARGV[$i]}) {
197 $cmd = $ARGV[$i];
198 splice @ARGV, $i, 1;
199 last;
200 }
201 };
202
203 # make sure we're always running at the top-level working directory
204 unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
205 unless (-d $ENV{GIT_DIR}) {
206 if ($git_dir_user_set) {
207 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
208 "but it is not a directory\n";
209 }
210 my $git_dir = delete $ENV{GIT_DIR};
211 chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
212 unless (length $cdup) {
213 die "Already at toplevel, but $git_dir ",
214 "not found '$cdup'\n";
215 }
216 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
217 unless (-d $git_dir) {
218 die "$git_dir still not found after going to ",
219 "'$cdup'\n";
220 }
221 $ENV{GIT_DIR} = $git_dir;
222 }
223 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
224 }
225
226 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
227
228 read_repo_config(\%opts);
229 Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
230 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
231 'minimize-connections' => \$Git::SVN::Migration::_minimize,
232 'id|i=s' => \$Git::SVN::default_ref_id,
233 'svn-remote|remote|R=s' => sub {
234 $Git::SVN::no_reuse_existing = 1;
235 $Git::SVN::default_repo_id = $_[1] });
236 exit 1 if (!$rv && $cmd && $cmd ne 'log');
237
238 usage(0) if $_help;
239 version() if $_version;
240 usage(1) unless defined $cmd;
241 load_authors() if $_authors;
242
243 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
244 Git::SVN::Migration::migration_check();
245 }
246 Git::SVN::init_vars();
247 eval {
248 Git::SVN::verify_remotes_sanity();
249 $cmd{$cmd}->[0]->(@ARGV);
250 };
251 fatal $@ if $@;
252 post_fetch_checkout();
253 exit 0;
254
255 ####################### primary functions ######################
256 sub usage {
257 my $exit = shift || 0;
258 my $fd = $exit ? \*STDERR : \*STDOUT;
259 print $fd <<"";
260 git-svn - bidirectional operations between a single Subversion tree and git
261 Usage: $0 <command> [options] [arguments]\n
262
263 print $fd "Available commands:\n" unless $cmd;
264
265 foreach (sort keys %cmd) {
266 next if $cmd && $cmd ne $_;
267 next if /^multi-/; # don't show deprecated commands
268 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
269 foreach (sort keys %{$cmd{$_}->[2]}) {
270 # mixed-case options are for .git/config only
271 next if /[A-Z]/ && /^[a-z]+$/i;
272 # prints out arguments as they should be passed:
273 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
274 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
275 "--$_" : "-$_" }
276 split /\|/,$_)," $x\n";
277 }
278 }
279 print $fd <<"";
280 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
281 arbitrary identifier if you're tracking multiple SVN branches/repositories in
282 one git repository and want to keep them separate. See git-svn(1) for more
283 information.
284
285 exit $exit;
286 }
287
288 sub version {
289 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
290 exit 0;
291 }
292
293 sub do_git_init_db {
294 unless (-d $ENV{GIT_DIR}) {
295 my @init_db = ('init');
296 push @init_db, "--template=$_template" if defined $_template;
297 if (defined $_shared) {
298 if ($_shared =~ /[a-z]/) {
299 push @init_db, "--shared=$_shared";
300 } else {
301 push @init_db, "--shared";
302 }
303 }
304 command_noisy(@init_db);
305 $_repository = Git->repository(Repository => ".git");
306 }
307 my $set;
308 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
309 foreach my $i (keys %icv) {
310 die "'$set' and '$i' cannot both be set\n" if $set;
311 next unless defined $icv{$i};
312 command_noisy('config', "$pfx.$i", $icv{$i});
313 $set = $i;
314 }
315 }
316
317 sub init_subdir {
318 my $repo_path = shift or return;
319 mkpath([$repo_path]) unless -d $repo_path;
320 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
321 $ENV{GIT_DIR} = '.git';
322 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
323 }
324
325 sub cmd_clone {
326 my ($url, $path) = @_;
327 if (!defined $path &&
328 (defined $_trunk || defined $_branches || defined $_tags ||
329 defined $_stdlayout) &&
330 $url !~ m#^[a-z\+]+://#) {
331 $path = $url;
332 }
333 $path = basename($url) if !defined $path || !length $path;
334 cmd_init($url, $path);
335 Git::SVN::fetch_all($Git::SVN::default_repo_id);
336 }
337
338 sub cmd_init {
339 if (defined $_stdlayout) {
340 $_trunk = 'trunk' if (!defined $_trunk);
341 $_tags = 'tags' if (!defined $_tags);
342 $_branches = 'branches' if (!defined $_branches);
343 }
344 if (defined $_trunk || defined $_branches || defined $_tags) {
345 return cmd_multi_init(@_);
346 }
347 my $url = shift or die "SVN repository location required ",
348 "as a command-line argument\n";
349 init_subdir(@_);
350 do_git_init_db();
351
352 Git::SVN->init($url);
353 }
354
355 sub cmd_fetch {
356 if (grep /^\d+=./, @_) {
357 die "'<rev>=<commit>' fetch arguments are ",
358 "no longer supported.\n";
359 }
360 my ($remote) = @_;
361 if (@_ > 1) {
362 die "Usage: $0 fetch [--all] [svn-remote]\n";
363 }
364 $remote ||= $Git::SVN::default_repo_id;
365 if ($_fetch_all) {
366 cmd_multi_fetch();
367 } else {
368 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
369 }
370 }
371
372 sub cmd_set_tree {
373 my (@commits) = @_;
374 if ($_stdin || !@commits) {
375 print "Reading from stdin...\n";
376 @commits = ();
377 while (<STDIN>) {
378 if (/\b($sha1_short)\b/o) {
379 unshift @commits, $1;
380 }
381 }
382 }
383 my @revs;
384 foreach my $c (@commits) {
385 my @tmp = command('rev-parse',$c);
386 if (scalar @tmp == 1) {
387 push @revs, $tmp[0];
388 } elsif (scalar @tmp > 1) {
389 push @revs, reverse(command('rev-list',@tmp));
390 } else {
391 fatal "Failed to rev-parse $c";
392 }
393 }
394 my $gs = Git::SVN->new;
395 my ($r_last, $cmt_last) = $gs->last_rev_commit;
396 $gs->fetch;
397 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
398 fatal "There are new revisions that were fetched ",
399 "and need to be merged (or acknowledged) ",
400 "before committing.\nlast rev: $r_last\n",
401 " current: $gs->{last_rev}";
402 }
403 $gs->set_tree($_) foreach @revs;
404 print "Done committing ",scalar @revs," revisions to SVN\n";
405 unlink $gs->{index};
406 }
407
408 sub cmd_dcommit {
409 my $head = shift;
410 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
411 'Cannot dcommit with a dirty index. Commit your changes first, '
412 . "or stash them with `git stash'.\n";
413 $head ||= 'HEAD';
414 my @refs;
415 my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
416 if ($url) {
417 print "Committing to $url ...\n";
418 }
419 unless ($gs) {
420 die "Unable to determine upstream SVN information from ",
421 "$head history.\nPerhaps the repository is empty.";
422 }
423 my $last_rev;
424 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
425 if ($_no_rebase && scalar(@$linear_refs) > 1) {
426 warn "Attempting to commit more than one change while ",
427 "--no-rebase is enabled.\n",
428 "If these changes depend on each other, re-running ",
429 "without --no-rebase may be required."
430 }
431 while (1) {
432 my $d = shift @$linear_refs or last;
433 unless (defined $last_rev) {
434 (undef, $last_rev, undef) = cmt_metadata("$d~1");
435 unless (defined $last_rev) {
436 fatal "Unable to extract revision information ",
437 "from commit $d~1";
438 }
439 }
440 if ($_dry_run) {
441 print "diff-tree $d~1 $d\n";
442 } else {
443 my $cmt_rev;
444 my %ed_opts = ( r => $last_rev,
445 log => get_commit_entry($d)->{log},
446 ra => Git::SVN::Ra->new($gs->full_url),
447 config => SVN::Core::config_get_config(
448 $Git::SVN::Ra::config_dir
449 ),
450 tree_a => "$d~1",
451 tree_b => $d,
452 editor_cb => sub {
453 print "Committed r$_[0]\n";
454 $cmt_rev = $_[0];
455 },
456 svn_path => '');
457 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
458 print "No changes\n$d~1 == $d\n";
459 } elsif ($parents->{$d} && @{$parents->{$d}}) {
460 $gs->{inject_parents_dcommit}->{$cmt_rev} =
461 $parents->{$d};
462 }
463 $_fetch_all ? $gs->fetch_all : $gs->fetch;
464 $last_rev = $cmt_rev;
465 next if $_no_rebase;
466
467 # we always want to rebase against the current HEAD,
468 # not any head that was passed to us
469 my @diff = command('diff-tree', $d,
470 $gs->refname, '--');
471 my @finish;
472 if (@diff) {
473 @finish = rebase_cmd();
474 print STDERR "W: $d and ", $gs->refname,
475 " differ, using @finish:\n",
476 join("\n", @diff), "\n";
477 } else {
478 print "No changes between current HEAD and ",
479 $gs->refname,
480 "\nResetting to the latest ",
481 $gs->refname, "\n";
482 @finish = qw/reset --mixed/;
483 }
484 command_noisy(@finish, $gs->refname);
485 if (@diff) {
486 @refs = ();
487 my ($url_, $rev_, $uuid_, $gs_) =
488 working_head_info($head, \@refs);
489 my ($linear_refs_, $parents_) =
490 linearize_history($gs_, \@refs);
491 if (scalar(@$linear_refs) !=
492 scalar(@$linear_refs_)) {
493 fatal "# of revisions changed ",
494 "\nbefore:\n",
495 join("\n", @$linear_refs),
496 "\n\nafter:\n",
497 join("\n", @$linear_refs_), "\n",
498 'If you are attempting to commit ',
499 "merges, try running:\n\t",
500 'git rebase --interactive',
501 '--preserve-merges ',
502 $gs->refname,
503 "\nBefore dcommitting";
504 }
505 if ($url_ ne $url) {
506 fatal "URL mismatch after rebase: ",
507 "$url_ != $url";
508 }
509 if ($uuid_ ne $uuid) {
510 fatal "uuid mismatch after rebase: ",
511 "$uuid_ != $uuid";
512 }
513 # remap parents
514 my (%p, @l, $i);
515 for ($i = 0; $i < scalar @$linear_refs; $i++) {
516 my $new = $linear_refs_->[$i] or next;
517 $p{$new} =
518 $parents->{$linear_refs->[$i]};
519 push @l, $new;
520 }
521 $parents = \%p;
522 $linear_refs = \@l;
523 }
524 }
525 }
526 unlink $gs->{index};
527 }
528
529 sub cmd_find_rev {
530 my $revision_or_hash = shift or die "SVN or git revision required ",
531 "as a command-line argument\n";
532 my $result;
533 if ($revision_or_hash =~ /^r\d+$/) {
534 my $head = shift;
535 $head ||= 'HEAD';
536 my @refs;
537 my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
538 unless ($gs) {
539 die "Unable to determine upstream SVN information from ",
540 "$head history\n";
541 }
542 my $desired_revision = substr($revision_or_hash, 1);
543 $result = $gs->rev_map_get($desired_revision);
544 } else {
545 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
546 $result = $rev;
547 }
548 print "$result\n" if $result;
549 }
550
551 sub cmd_rebase {
552 command_noisy(qw/update-index --refresh/);
553 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
554 unless ($gs) {
555 die "Unable to determine upstream SVN information from ",
556 "working tree history\n";
557 }
558 if (command(qw/diff-index HEAD --/)) {
559 print STDERR "Cannot rebase with uncommited changes:\n";
560 command_noisy('status');
561 exit 1;
562 }
563 unless ($_local) {
564 # rebase will checkout for us, so no need to do it explicitly
565 $_no_checkout = 'true';
566 $_fetch_all ? $gs->fetch_all : $gs->fetch;
567 }
568 command_noisy(rebase_cmd(), $gs->refname);
569 }
570
571 sub cmd_show_ignore {
572 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
573 $gs ||= Git::SVN->new;
574 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
575 $gs->prop_walk($gs->{path}, $r, sub {
576 my ($gs, $path, $props) = @_;
577 print STDOUT "\n# $path\n";
578 my $s = $props->{'svn:ignore'} or return;
579 $s =~ s/[\r\n]+/\n/g;
580 chomp $s;
581 $s =~ s#^#$path#gm;
582 print STDOUT "$s\n";
583 });
584 }
585
586 sub cmd_show_externals {
587 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
588 $gs ||= Git::SVN->new;
589 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
590 $gs->prop_walk($gs->{path}, $r, sub {
591 my ($gs, $path, $props) = @_;
592 print STDOUT "\n# $path\n";
593 my $s = $props->{'svn:externals'} or return;
594 $s =~ s/[\r\n]+/\n/g;
595 chomp $s;
596 $s =~ s#^#$path#gm;
597 print STDOUT "$s\n";
598 });
599 }
600
601 sub cmd_create_ignore {
602 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
603 $gs ||= Git::SVN->new;
604 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
605 $gs->prop_walk($gs->{path}, $r, sub {
606 my ($gs, $path, $props) = @_;
607 # $path is of the form /path/to/dir/
608 my $ignore = '.' . $path . '.gitignore';
609 my $s = $props->{'svn:ignore'} or return;
610 open(GITIGNORE, '>', $ignore)
611 or fatal("Failed to open `$ignore' for writing: $!");
612 $s =~ s/[\r\n]+/\n/g;
613 chomp $s;
614 # Prefix all patterns so that the ignore doesn't apply
615 # to sub-directories.
616 $s =~ s#^#/#gm;
617 print GITIGNORE "$s\n";
618 close(GITIGNORE)
619 or fatal("Failed to close `$ignore': $!");
620 command_noisy('add', $ignore);
621 });
622 }
623
624 sub canonicalize_path {
625 my ($path) = @_;
626 my $dot_slash_added = 0;
627 if (substr($path, 0, 1) ne "/") {
628 $path = "./" . $path;
629 $dot_slash_added = 1;
630 }
631 # File::Spec->canonpath doesn't collapse x/../y into y (for a
632 # good reason), so let's do this manually.
633 $path =~ s#/+#/#g;
634 $path =~ s#/\.(?:/|$)#/#g;
635 $path =~ s#/[^/]+/\.\.##g;
636 $path =~ s#/$##g;
637 $path =~ s#^\./## if $dot_slash_added;
638 return $path;
639 }
640
641 # get_svnprops(PATH)
642 # ------------------
643 # Helper for cmd_propget and cmd_proplist below.
644 sub get_svnprops {
645 my $path = shift;
646 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
647 $gs ||= Git::SVN->new;
648
649 # prefix THE PATH by the sub-directory from which the user
650 # invoked us.
651 $path = $cmd_dir_prefix . $path;
652 fatal("No such file or directory: $path") unless -e $path;
653 my $is_dir = -d $path ? 1 : 0;
654 $path = $gs->{path} . '/' . $path;
655
656 # canonicalize the path (otherwise libsvn will abort or fail to
657 # find the file)
658 $path = canonicalize_path($path);
659
660 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
661 my $props;
662 if ($is_dir) {
663 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
664 }
665 else {
666 (undef, $props) = $gs->ra->get_file($path, $r, undef);
667 }
668 return $props;
669 }
670
671 # cmd_propget (PROP, PATH)
672 # ------------------------
673 # Print the SVN property PROP for PATH.
674 sub cmd_propget {
675 my ($prop, $path) = @_;
676 $path = '.' if not defined $path;
677 usage(1) if not defined $prop;
678 my $props = get_svnprops($path);
679 if (not defined $props->{$prop}) {
680 fatal("`$path' does not have a `$prop' SVN property.");
681 }
682 print $props->{$prop} . "\n";
683 }
684
685 # cmd_proplist (PATH)
686 # -------------------
687 # Print the list of SVN properties for PATH.
688 sub cmd_proplist {
689 my $path = shift;
690 $path = '.' if not defined $path;
691 my $props = get_svnprops($path);
692 print "Properties on '$path':\n";
693 foreach (sort keys %{$props}) {
694 print " $_\n";
695 }
696 }
697
698 sub cmd_multi_init {
699 my $url = shift;
700 unless (defined $_trunk || defined $_branches || defined $_tags) {
701 usage(1);
702 }
703
704 # there are currently some bugs that prevent multi-init/multi-fetch
705 # setups from working well without this.
706 $Git::SVN::_minimize_url = 1;
707
708 $_prefix = '' unless defined $_prefix;
709 if (defined $url) {
710 $url =~ s#/+$##;
711 init_subdir(@_);
712 }
713 do_git_init_db();
714 if (defined $_trunk) {
715 my $trunk_ref = $_prefix . 'trunk';
716 # try both old-style and new-style lookups:
717 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
718 unless ($gs_trunk) {
719 my ($trunk_url, $trunk_path) =
720 complete_svn_url($url, $_trunk);
721 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
722 undef, $trunk_ref);
723 }
724 }
725 return unless defined $_branches || defined $_tags;
726 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
727 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
728 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
729 }
730
731 sub cmd_multi_fetch {
732 my $remotes = Git::SVN::read_all_remotes();
733 foreach my $repo_id (sort keys %$remotes) {
734 if ($remotes->{$repo_id}->{url}) {
735 Git::SVN::fetch_all($repo_id, $remotes);
736 }
737 }
738 }
739
740 # this command is special because it requires no metadata
741 sub cmd_commit_diff {
742 my ($ta, $tb, $url) = @_;
743 my $usage = "Usage: $0 commit-diff -r<revision> ".
744 "<tree-ish> <tree-ish> [<URL>]";
745 fatal($usage) if (!defined $ta || !defined $tb);
746 my $svn_path;
747 if (!defined $url) {
748 my $gs = eval { Git::SVN->new };
749 if (!$gs) {
750 fatal("Needed URL or usable git-svn --id in ",
751 "the command-line\n", $usage);
752 }
753 $url = $gs->{url};
754 $svn_path = $gs->{path};
755 }
756 unless (defined $_revision) {
757 fatal("-r|--revision is a required argument\n", $usage);
758 }
759 if (defined $_message && defined $_file) {
760 fatal("Both --message/-m and --file/-F specified ",
761 "for the commit message.\n",
762 "I have no idea what you mean");
763 }
764 if (defined $_file) {
765 $_message = file_to_s($_file);
766 } else {
767 $_message ||= get_commit_entry($tb)->{log};
768 }
769 my $ra ||= Git::SVN::Ra->new($url);
770 $svn_path ||= $ra->{svn_path};
771 my $r = $_revision;
772 if ($r eq 'HEAD') {
773 $r = $ra->get_latest_revnum;
774 } elsif ($r !~ /^\d+$/) {
775 die "revision argument: $r not understood by git-svn\n";
776 }
777 my %ed_opts = ( r => $r,
778 log => $_message,
779 ra => $ra,
780 tree_a => $ta,
781 tree_b => $tb,
782 editor_cb => sub { print "Committed r$_[0]\n" },
783 svn_path => $svn_path );
784 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
785 print "No changes\n$ta == $tb\n";
786 }
787 }
788
789 sub cmd_info {
790 my $path = canonicalize_path(shift or ".");
791 unless (scalar(@_) == 0) {
792 die "Too many arguments specified\n";
793 }
794
795 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
796
797 if (!$file_type && !$diff_status) {
798 print STDERR "$path: (Not a versioned resource)\n\n";
799 return;
800 }
801
802 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
803 unless ($gs) {
804 die "Unable to determine upstream SVN information from ",
805 "working tree history\n";
806 }
807 my $full_url = $url . ($path eq "." ? "" : "/$path");
808
809 if ($_url) {
810 print $full_url, "\n";
811 return;
812 }
813
814 my $result = "Path: $path\n";
815 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
816 $result .= "URL: " . $full_url . "\n";
817
818 eval {
819 my $repos_root = $gs->repos_root;
820 Git::SVN::remove_username($repos_root);
821 $result .= "Repository Root: $repos_root\n";
822 };
823 if ($@) {
824 $result .= "Repository Root: (offline)\n";
825 }
826 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
827 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
828
829 $result .= "Node Kind: " .
830 ($file_type eq "dir" ? "directory" : "file") . "\n";
831
832 my $schedule = $diff_status eq "A"
833 ? "add"
834 : ($diff_status eq "D" ? "delete" : "normal");
835 $result .= "Schedule: $schedule\n";
836
837 if ($diff_status eq "A") {
838 print $result, "\n";
839 return;
840 }
841
842 my ($lc_author, $lc_rev, $lc_date_utc);
843 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $path);
844 my $log = command_output_pipe(@args);
845 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
846 while (<$log>) {
847 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
848 $lc_author = $1;
849 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
850 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
851 (undef, $lc_rev, undef) = ::extract_metadata($1);
852 }
853 }
854 close $log;
855
856 Git::SVN::Log::set_local_timezone();
857
858 $result .= "Last Changed Author: $lc_author\n";
859 $result .= "Last Changed Rev: $lc_rev\n";
860 $result .= "Last Changed Date: " .
861 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
862
863 if ($file_type ne "dir") {
864 my $text_last_updated_date =
865 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
866 $result .=
867 "Text Last Updated: " .
868 Git::SVN::Log::format_svn_date($text_last_updated_date) .
869 "\n";
870 my $checksum;
871 if ($diff_status eq "D") {
872 my ($fh, $ctx) =
873 command_output_pipe(qw(cat-file blob), "HEAD:$path");
874 if ($file_type eq "link") {
875 my $file_name = <$fh>;
876 $checksum = md5sum("link $file_name");
877 } else {
878 $checksum = md5sum($fh);
879 }
880 command_close_pipe($fh, $ctx);
881 } elsif ($file_type eq "link") {
882 my $file_name =
883 command(qw(cat-file blob), "HEAD:$path");
884 $checksum =
885 md5sum("link " . $file_name);
886 } else {
887 open FILE, "<", $path or die $!;
888 $checksum = md5sum(\*FILE);
889 close FILE or die $!;
890 }
891 $result .= "Checksum: " . $checksum . "\n";
892 }
893
894 print $result, "\n";
895 }
896
897 ########################### utility functions #########################
898
899 sub rebase_cmd {
900 my @cmd = qw/rebase/;
901 push @cmd, '-v' if $_verbose;
902 push @cmd, qw/--merge/ if $_merge;
903 push @cmd, "--strategy=$_strategy" if $_strategy;
904 @cmd;
905 }
906
907 sub post_fetch_checkout {
908 return if $_no_checkout;
909 my $gs = $Git::SVN::_head or return;
910 return if verify_ref('refs/heads/master^0');
911
912 my $valid_head = verify_ref('HEAD^0');
913 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
914 return if ($valid_head || !verify_ref('HEAD^0'));
915
916 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
917 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
918 return if -f $index;
919
920 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
921 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
922 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
923 print STDERR "Checked out HEAD:\n ",
924 $gs->full_url, " r", $gs->last_rev, "\n";
925 }
926
927 sub complete_svn_url {
928 my ($url, $path) = @_;
929 $path =~ s#/+$##;
930 if ($path !~ m#^[a-z\+]+://#) {
931 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
932 fatal("E: '$path' is not a complete URL ",
933 "and a separate URL is not specified");
934 }
935 return ($url, $path);
936 }
937 return ($path, '');
938 }
939
940 sub complete_url_ls_init {
941 my ($ra, $repo_path, $switch, $pfx) = @_;
942 unless ($repo_path) {
943 print STDERR "W: $switch not specified\n";
944 return;
945 }
946 $repo_path =~ s#/+$##;
947 if ($repo_path =~ m#^[a-z\+]+://#) {
948 $ra = Git::SVN::Ra->new($repo_path);
949 $repo_path = '';
950 } else {
951 $repo_path =~ s#^/+##;
952 unless ($ra) {
953 fatal("E: '$repo_path' is not a complete URL ",
954 "and a separate URL is not specified");
955 }
956 }
957 my $url = $ra->{url};
958 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
959 my $k = "svn-remote.$gs->{repo_id}.url";
960 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
961 if ($orig_url && ($orig_url ne $gs->{url})) {
962 die "$k already set: $orig_url\n",
963 "wanted to set to: $gs->{url}\n";
964 }
965 command_oneline('config', $k, $gs->{url}) unless $orig_url;
966 my $remote_path = "$ra->{svn_path}/$repo_path";
967 $remote_path =~ s#/+#/#g;
968 $remote_path =~ s#^/##g;
969 $remote_path .= "/*" if $remote_path !~ /\*/;
970 my ($n) = ($switch =~ /^--(\w+)/);
971 if (length $pfx && $pfx !~ m#/$#) {
972 die "--prefix='$pfx' must have a trailing slash '/'\n";
973 }
974 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
975 "$remote_path:refs/remotes/$pfx*");
976 }
977
978 sub verify_ref {
979 my ($ref) = @_;
980 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
981 { STDERR => 0 }); };
982 }
983
984 sub get_tree_from_treeish {
985 my ($treeish) = @_;
986 # $treeish can be a symbolic ref, too:
987 my $type = command_oneline(qw/cat-file -t/, $treeish);
988 my $expected;
989 while ($type eq 'tag') {
990 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
991 }
992 if ($type eq 'commit') {
993 $expected = (grep /^tree /, command(qw/cat-file commit/,
994 $treeish))[0];
995 ($expected) = ($expected =~ /^tree ($sha1)$/o);
996 die "Unable to get tree from $treeish\n" unless $expected;
997 } elsif ($type eq 'tree') {
998 $expected = $treeish;
999 } else {
1000 die "$treeish is a $type, expected tree, tag or commit\n";
1001 }
1002 return $expected;
1003 }
1004
1005 sub get_commit_entry {
1006 my ($treeish) = shift;
1007 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1008 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1009 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1010 open my $log_fh, '>', $commit_editmsg or croak $!;
1011
1012 my $type = command_oneline(qw/cat-file -t/, $treeish);
1013 if ($type eq 'commit' || $type eq 'tag') {
1014 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1015 $type, $treeish);
1016 my $in_msg = 0;
1017 while (<$msg_fh>) {
1018 if (!$in_msg) {
1019 $in_msg = 1 if (/^\s*$/);
1020 } elsif (/^git-svn-id: /) {
1021 # skip this for now, we regenerate the
1022 # correct one on re-fetch anyways
1023 # TODO: set *:merge properties or like...
1024 } else {
1025 print $log_fh $_ or croak $!;
1026 }
1027 }
1028 command_close_pipe($msg_fh, $ctx);
1029 }
1030 close $log_fh or croak $!;
1031
1032 if ($_edit || ($type eq 'tree')) {
1033 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1034 # TODO: strip out spaces, comments, like git-commit.sh
1035 system($editor, $commit_editmsg);
1036 }
1037 rename $commit_editmsg, $commit_msg or croak $!;
1038 open $log_fh, '<', $commit_msg or croak $!;
1039 { local $/; chomp($log_entry{log} = <$log_fh>); }
1040 close $log_fh or croak $!;
1041 unlink $commit_msg;
1042 \%log_entry;
1043 }
1044
1045 sub s_to_file {
1046 my ($str, $file, $mode) = @_;
1047 open my $fd,'>',$file or croak $!;
1048 print $fd $str,"\n" or croak $!;
1049 close $fd or croak $!;
1050 chmod ($mode &~ umask, $file) if (defined $mode);
1051 }
1052
1053 sub file_to_s {
1054 my $file = shift;
1055 open my $fd,'<',$file or croak "$!: file: $file\n";
1056 local $/;
1057 my $ret = <$fd>;
1058 close $fd or croak $!;
1059 $ret =~ s/\s*$//s;
1060 return $ret;
1061 }
1062
1063 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1064 sub load_authors {
1065 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1066 my $log = $cmd eq 'log';
1067 while (<$authors>) {
1068 chomp;
1069 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1070 my ($user, $name, $email) = ($1, $2, $3);
1071 if ($log) {
1072 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1073 } else {
1074 $users{$user} = [$name, $email];
1075 }
1076 }
1077 close $authors or croak $!;
1078 }
1079
1080 # convert GetOpt::Long specs for use by git-config
1081 sub read_repo_config {
1082 return unless -d $ENV{GIT_DIR};
1083 my $opts = shift;
1084 my @config_only;
1085 foreach my $o (keys %$opts) {
1086 # if we have mixedCase and a long option-only, then
1087 # it's a config-only variable that we don't need for
1088 # the command-line.
1089 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1090 my $v = $opts->{$o};
1091 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1092 $key =~ s/-//g;
1093 my $arg = 'git-config';
1094 $arg .= ' --int' if ($o =~ /[:=]i$/);
1095 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1096 if (ref $v eq 'ARRAY') {
1097 chomp(my @tmp = `$arg --get-all svn.$key`);
1098 @$v = @tmp if @tmp;
1099 } else {
1100 chomp(my $tmp = `$arg --get svn.$key`);
1101 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1102 $$v = $tmp;
1103 }
1104 }
1105 }
1106 delete @$opts{@config_only} if @config_only;
1107 }
1108
1109 sub extract_metadata {
1110 my $id = shift or return (undef, undef, undef);
1111 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1112 \s([a-f\d\-]+)$/x);
1113 if (!defined $rev || !$uuid || !$url) {
1114 # some of the original repositories I made had
1115 # identifiers like this:
1116 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1117 }
1118 return ($url, $rev, $uuid);
1119 }
1120
1121 sub cmt_metadata {
1122 return extract_metadata((grep(/^git-svn-id: /,
1123 command(qw/cat-file commit/, shift)))[-1]);
1124 }
1125
1126 sub working_head_info {
1127 my ($head, $refs) = @_;
1128 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1129 my ($fh, $ctx) = command_output_pipe(@args, $head);
1130 my $hash;
1131 my %max;
1132 while (<$fh>) {
1133 if ( m{^commit ($::sha1)$} ) {
1134 unshift @$refs, $hash if $hash and $refs;
1135 $hash = $1;
1136 next;
1137 }
1138 next unless s{^\s*(git-svn-id:)}{$1};
1139 my ($url, $rev, $uuid) = extract_metadata($_);
1140 if (defined $url && defined $rev) {
1141 next if $max{$url} and $max{$url} < $rev;
1142 if (my $gs = Git::SVN->find_by_url($url)) {
1143 my $c = $gs->rev_map_get($rev);
1144 if ($c && $c eq $hash) {
1145 close $fh; # break the pipe
1146 return ($url, $rev, $uuid, $gs);
1147 } else {
1148 $max{$url} ||= $gs->rev_map_max;
1149 }
1150 }
1151 }
1152 }
1153 command_close_pipe($fh, $ctx);
1154 (undef, undef, undef, undef);
1155 }
1156
1157 sub read_commit_parents {
1158 my ($parents, $c) = @_;
1159 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1160 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1161 @{$parents->{$c}} = split(/ /, $p);
1162 }
1163
1164 sub linearize_history {
1165 my ($gs, $refs) = @_;
1166 my %parents;
1167 foreach my $c (@$refs) {
1168 read_commit_parents(\%parents, $c);
1169 }
1170
1171 my @linear_refs;
1172 my %skip = ();
1173 my $last_svn_commit = $gs->last_commit;
1174 foreach my $c (reverse @$refs) {
1175 next if $c eq $last_svn_commit;
1176 last if $skip{$c};
1177
1178 unshift @linear_refs, $c;
1179 $skip{$c} = 1;
1180
1181 # we only want the first parent to diff against for linear
1182 # history, we save the rest to inject when we finalize the
1183 # svn commit
1184 my $fp_a = verify_ref("$c~1");
1185 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1186 if (!$fp_a || !$fp_b) {
1187 die "Commit $c\n",
1188 "has no parent commit, and therefore ",
1189 "nothing to diff against.\n",
1190 "You should be working from a repository ",
1191 "originally created by git-svn\n";
1192 }
1193 if ($fp_a ne $fp_b) {
1194 die "$c~1 = $fp_a, however parsing commit $c ",
1195 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1196 }
1197
1198 foreach my $p (@{$parents{$c}}) {
1199 $skip{$p} = 1;
1200 }
1201 }
1202 (\@linear_refs, \%parents);
1203 }
1204
1205 sub find_file_type_and_diff_status {
1206 my ($path) = @_;
1207 return ('dir', '') if $path eq '.';
1208
1209 my $diff_output =
1210 command_oneline(qw(diff --cached --name-status --), $path) || "";
1211 my $diff_status = (split(' ', $diff_output))[0] || "";
1212
1213 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1214
1215 return (undef, undef) if !$diff_status && !$ls_tree;
1216
1217 if ($diff_status eq "A") {
1218 return ("link", $diff_status) if -l $path;
1219 return ("dir", $diff_status) if -d $path;
1220 return ("file", $diff_status);
1221 }
1222
1223 my $mode = (split(' ', $ls_tree))[0] || "";
1224
1225 return ("link", $diff_status) if $mode eq "120000";
1226 return ("dir", $diff_status) if $mode eq "040000";
1227 return ("file", $diff_status);
1228 }
1229
1230 sub md5sum {
1231 my $arg = shift;
1232 my $ref = ref $arg;
1233 my $md5 = Digest::MD5->new();
1234 if ($ref eq 'GLOB' || $ref eq 'IO::File') {
1235 $md5->addfile($arg) or croak $!;
1236 } elsif ($ref eq 'SCALAR') {
1237 $md5->add($$arg) or croak $!;
1238 } elsif (!$ref) {
1239 $md5->add($arg) or croak $!;
1240 } else {
1241 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1242 }
1243 return $md5->hexdigest();
1244 }
1245
1246 package Git::SVN;
1247 use strict;
1248 use warnings;
1249 use Fcntl qw/:DEFAULT :seek/;
1250 use constant rev_map_fmt => 'NH40';
1251 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1252 $_repack $_repack_flags $_use_svm_props $_head
1253 $_use_svnsync_props $no_reuse_existing $_minimize_url
1254 $_use_log_author/;
1255 use Carp qw/croak/;
1256 use File::Path qw/mkpath/;
1257 use File::Copy qw/copy/;
1258 use IPC::Open3;
1259
1260 my ($_gc_nr, $_gc_period);
1261
1262 # properties that we do not log:
1263 my %SKIP_PROP;
1264 BEGIN {
1265 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1266 svn:special svn:executable
1267 svn:entry:committed-rev
1268 svn:entry:last-author
1269 svn:entry:uuid
1270 svn:entry:committed-date/;
1271
1272 # some options are read globally, but can be overridden locally
1273 # per [svn-remote "..."] section. Command-line options will *NOT*
1274 # override options set in an [svn-remote "..."] section
1275 no strict 'refs';
1276 for my $option (qw/follow_parent no_metadata use_svm_props
1277 use_svnsync_props/) {
1278 my $key = $option;
1279 $key =~ tr/_//d;
1280 my $prop = "-$option";
1281 *$option = sub {
1282 my ($self) = @_;
1283 return $self->{$prop} if exists $self->{$prop};
1284 my $k = "svn-remote.$self->{repo_id}.$key";
1285 eval { command_oneline(qw/config --get/, $k) };
1286 if ($@) {
1287 $self->{$prop} = ${"Git::SVN::_$option"};
1288 } else {
1289 my $v = command_oneline(qw/config --bool/,$k);
1290 $self->{$prop} = $v eq 'false' ? 0 : 1;
1291 }
1292 return $self->{$prop};
1293 }
1294 }
1295 }
1296
1297 my (%LOCKFILES, %INDEX_FILES);
1298 END {
1299 unlink keys %LOCKFILES if %LOCKFILES;
1300 unlink keys %INDEX_FILES if %INDEX_FILES;
1301 }
1302
1303 sub resolve_local_globs {
1304 my ($url, $fetch, $glob_spec) = @_;
1305 return unless defined $glob_spec;
1306 my $ref = $glob_spec->{ref};
1307 my $path = $glob_spec->{path};
1308 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1309 next unless m#^refs/remotes/$ref->{regex}$#;
1310 my $p = $1;
1311 my $pathname = desanitize_refname($path->full_path($p));
1312 my $refname = desanitize_refname($ref->full_path($p));
1313 if (my $existing = $fetch->{$pathname}) {
1314 if ($existing ne $refname) {
1315 die "Refspec conflict:\n",
1316 "existing: refs/remotes/$existing\n",
1317 " globbed: refs/remotes/$refname\n";
1318 }
1319 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1320 $u =~ s!^\Q$url\E(/|$)!! or die
1321 "refs/remotes/$refname: '$url' not found in '$u'\n";
1322 if ($pathname ne $u) {
1323 warn "W: Refspec glob conflict ",
1324 "(ref: refs/remotes/$refname):\n",
1325 "expected path: $pathname\n",
1326 " real path: $u\n",
1327 "Continuing ahead with $u\n";
1328 next;
1329 }
1330 } else {
1331 $fetch->{$pathname} = $refname;
1332 }
1333 }
1334 }
1335
1336 sub parse_revision_argument {
1337 my ($base, $head) = @_;
1338 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1339 return ($base, $head);
1340 }
1341 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1342 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1343 return ($head, $head) if ($::_revision eq 'HEAD');
1344 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1345 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1346 die "revision argument: $::_revision not understood by git-svn\n";
1347 }
1348
1349 sub fetch_all {
1350 my ($repo_id, $remotes) = @_;
1351 if (ref $repo_id) {
1352 my $gs = $repo_id;
1353 $repo_id = undef;
1354 $repo_id = $gs->{repo_id};
1355 }
1356 $remotes ||= read_all_remotes();
1357 my $remote = $remotes->{$repo_id} or
1358 die "[svn-remote \"$repo_id\"] unknown\n";
1359 my $fetch = $remote->{fetch};
1360 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1361 my (@gs, @globs);
1362 my $ra = Git::SVN::Ra->new($url);
1363 my $uuid = $ra->get_uuid;
1364 my $head = $ra->get_latest_revnum;
1365 my $base = defined $fetch ? $head : 0;
1366
1367 # read the max revs for wildcard expansion (branches/*, tags/*)
1368 foreach my $t (qw/branches tags/) {
1369 defined $remote->{$t} or next;
1370 push @globs, $remote->{$t};
1371 my $max_rev = eval { tmp_config(qw/--int --get/,
1372 "svn-remote.$repo_id.${t}-maxRev") };
1373 if (defined $max_rev && ($max_rev < $base)) {
1374 $base = $max_rev;
1375 } elsif (!defined $max_rev) {
1376 $base = 0;
1377 }
1378 }
1379
1380 if ($fetch) {
1381 foreach my $p (sort keys %$fetch) {
1382 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1383 my $lr = $gs->rev_map_max;
1384 if (defined $lr) {
1385 $base = $lr if ($lr < $base);
1386 }
1387 push @gs, $gs;
1388 }
1389 }
1390
1391 ($base, $head) = parse_revision_argument($base, $head);
1392 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1393 }
1394
1395 sub read_all_remotes {
1396 my $r = {};
1397 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1398 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
1399 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1400 $local_ref =~ s{^/}{};
1401 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1402 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1403 $r->{$1}->{url} = $2;
1404 } elsif (m!^(.+)\.(branches|tags)=
1405 (.*):refs/remotes/(.+)\s*$/!x) {
1406 my ($p, $g) = ($3, $4);
1407 my $rs = $r->{$1}->{$2} = {
1408 t => $2,
1409 remote => $1,
1410 path => Git::SVN::GlobSpec->new($p),
1411 ref => Git::SVN::GlobSpec->new($g) };
1412 if (length($rs->{ref}->{right}) != 0) {
1413 die "The '*' glob character must be the last ",
1414 "character of '$g'\n";
1415 }
1416 }
1417 }
1418 $r;
1419 }
1420
1421 sub init_vars {
1422 $_gc_nr = $_gc_period = 1000;
1423 if (defined $_repack || defined $_repack_flags) {
1424 warn "Repack options are obsolete; they have no effect.\n";
1425 }
1426 }
1427
1428 sub verify_remotes_sanity {
1429 return unless -d $ENV{GIT_DIR};
1430 my %seen;
1431 foreach (command(qw/config -l/)) {
1432 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1433 if ($seen{$1}) {
1434 die "Remote ref refs/remote/$1 is tracked by",
1435 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1436 "Please resolve this ambiguity in ",
1437 "your git configuration file before ",
1438 "continuing\n";
1439 }
1440 $seen{$1} = $_;
1441 }
1442 }
1443 }
1444
1445 # we allow more chars than remotes2config.sh...
1446 sub sanitize_remote_name {
1447 my ($name) = @_;
1448 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1449 $name;
1450 }
1451
1452 sub find_existing_remote {
1453 my ($url, $remotes) = @_;
1454 return undef if $no_reuse_existing;
1455 my $existing;
1456 foreach my $repo_id (keys %$remotes) {
1457 my $u = $remotes->{$repo_id}->{url} or next;
1458 next if $u ne $url;
1459 $existing = $repo_id;
1460 last;
1461 }
1462 $existing;
1463 }
1464
1465 sub init_remote_config {
1466 my ($self, $url, $no_write) = @_;
1467 $url =~ s!/+$!!; # strip trailing slash
1468 my $r = read_all_remotes();
1469 my $existing = find_existing_remote($url, $r);
1470 if ($existing) {
1471 unless ($no_write) {
1472 print STDERR "Using existing ",
1473 "[svn-remote \"$existing\"]\n";
1474 }
1475 $self->{repo_id} = $existing;
1476 } elsif ($_minimize_url) {
1477 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1478 $existing = find_existing_remote($min_url, $r);
1479 if ($existing) {
1480 unless ($no_write) {
1481 print STDERR "Using existing ",
1482 "[svn-remote \"$existing\"]\n";
1483 }
1484 $self->{repo_id} = $existing;
1485 }
1486 if ($min_url ne $url) {
1487 unless ($no_write) {
1488 print STDERR "Using higher level of URL: ",
1489 "$url => $min_url\n";
1490 }
1491 my $old_path = $self->{path};
1492 $self->{path} = $url;
1493 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1494 if (length $old_path) {
1495 $self->{path} .= "/$old_path";
1496 }
1497 $url = $min_url;
1498 }
1499 }
1500 my $orig_url;
1501 if (!$existing) {
1502 # verify that we aren't overwriting anything:
1503 $orig_url = eval {
1504 command_oneline('config', '--get',
1505 "svn-remote.$self->{repo_id}.url")
1506 };
1507 if ($orig_url && ($orig_url ne $url)) {
1508 die "svn-remote.$self->{repo_id}.url already set: ",
1509 "$orig_url\nwanted to set to: $url\n";
1510 }
1511 }
1512 my ($xrepo_id, $xpath) = find_ref($self->refname);
1513 if (defined $xpath) {
1514 die "svn-remote.$xrepo_id.fetch already set to track ",
1515 "$xpath:refs/remotes/", $self->refname, "\n";
1516 }
1517 unless ($no_write) {
1518 command_noisy('config',
1519 "svn-remote.$self->{repo_id}.url", $url);
1520 $self->{path} =~ s{^/}{};
1521 command_noisy('config', '--add',
1522 "svn-remote.$self->{repo_id}.fetch",
1523 "$self->{path}:".$self->refname);
1524 }
1525 $self->{url} = $url;
1526 }
1527
1528 sub find_by_url { # repos_root and, path are optional
1529 my ($class, $full_url, $repos_root, $path) = @_;
1530
1531 return undef unless defined $full_url;
1532 remove_username($full_url);
1533 remove_username($repos_root) if defined $repos_root;
1534 my $remotes = read_all_remotes();
1535 if (defined $full_url && defined $repos_root && !defined $path) {
1536 $path = $full_url;
1537 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1538 }
1539 foreach my $repo_id (keys %$remotes) {
1540 my $u = $remotes->{$repo_id}->{url} or next;
1541 remove_username($u);
1542 next if defined $repos_root && $repos_root ne $u;
1543
1544 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1545 foreach (qw/branches tags/) {
1546 resolve_local_globs($u, $fetch,
1547 $remotes->{$repo_id}->{$_});
1548 }
1549 my $p = $path;
1550 my $rwr = rewrite_root({repo_id => $repo_id});
1551 unless (defined $p) {
1552 $p = $full_url;
1553 my $z = $u;
1554 if ($rwr) {
1555 $z = $rwr;
1556 }
1557 $p =~ s#^\Q$z\E(?:/|$)## or next;
1558 }
1559 foreach my $f (keys %$fetch) {
1560 next if $f ne $p;
1561 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1562 }
1563 }
1564 undef;
1565 }
1566
1567 sub init {
1568 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1569 my $self = _new($class, $repo_id, $ref_id, $path);
1570 if (defined $url) {
1571 $self->init_remote_config($url, $no_write);
1572 }
1573 $self;
1574 }
1575
1576 sub find_ref {
1577 my ($ref_id) = @_;
1578 foreach (command(qw/config -l/)) {
1579 next unless m!^svn-remote\.(.+)\.fetch=
1580 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1581 my ($repo_id, $path, $ref) = ($1, $2, $3);
1582 if ($ref eq $ref_id) {
1583 $path = '' if ($path =~ m#^\./?#);
1584 return ($repo_id, $path);
1585 }
1586 }
1587 (undef, undef, undef);
1588 }
1589
1590 sub new {
1591 my ($class, $ref_id, $repo_id, $path) = @_;
1592 if (defined $ref_id && !defined $repo_id && !defined $path) {
1593 ($repo_id, $path) = find_ref($ref_id);
1594 if (!defined $repo_id) {
1595 die "Could not find a \"svn-remote.*.fetch\" key ",
1596 "in the repository configuration matching: ",
1597 "refs/remotes/$ref_id\n";
1598 }
1599 }
1600 my $self = _new($class, $repo_id, $ref_id, $path);
1601 if (!defined $self->{path} || !length $self->{path}) {
1602 my $fetch = command_oneline('config', '--get',
1603 "svn-remote.$repo_id.fetch",
1604 ":refs/remotes/$ref_id\$") or
1605 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1606 "\":refs/remotes/$ref_id\$\" in config\n";
1607 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1608 }
1609 $self->{url} = command_oneline('config', '--get',
1610 "svn-remote.$repo_id.url") or
1611 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1612 $self->rebuild;
1613 $self;
1614 }
1615
1616 sub refname {
1617 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1618
1619 # It cannot end with a slash /, we'll throw up on this because
1620 # SVN can't have directories with a slash in their name, either:
1621 if ($refname =~ m{/$}) {
1622 die "ref: '$refname' ends with a trailing slash, this is ",
1623 "not permitted by git nor Subversion\n";
1624 }
1625
1626 # It cannot have ASCII control character space, tilde ~, caret ^,
1627 # colon :, question-mark ?, asterisk *, space, or open bracket [
1628 # anywhere.
1629 #
1630 # Additionally, % must be escaped because it is used for escaping
1631 # and we want our escaped refname to be reversible
1632 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1633
1634 # no slash-separated component can begin with a dot .
1635 # /.* becomes /%2E*
1636 $refname =~ s{/\.}{/%2E}g;
1637
1638 # It cannot have two consecutive dots .. anywhere
1639 # .. becomes %2E%2E
1640 $refname =~ s{\.\.}{%2E%2E}g;
1641
1642 return $refname;
1643 }
1644
1645 sub desanitize_refname {
1646 my ($refname) = @_;
1647 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1648 return $refname;
1649 }
1650
1651 sub svm_uuid {
1652 my ($self) = @_;
1653 return $self->{svm}->{uuid} if $self->svm;
1654 $self->ra;
1655 unless ($self->{svm}) {
1656 die "SVM UUID not cached, and reading remotely failed\n";
1657 }
1658 $self->{svm}->{uuid};
1659 }
1660
1661 sub svm {
1662 my ($self) = @_;
1663 return $self->{svm} if $self->{svm};
1664 my $svm;
1665 # see if we have it in our config, first:
1666 eval {
1667 my $section = "svn-remote.$self->{repo_id}";
1668 $svm = {
1669 source => tmp_config('--get', "$section.svm-source"),
1670 uuid => tmp_config('--get', "$section.svm-uuid"),
1671 replace => tmp_config('--get', "$section.svm-replace"),
1672 }
1673 };
1674 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1675 $self->{svm} = $svm;
1676 }
1677 $self->{svm};
1678 }
1679
1680 sub _set_svm_vars {
1681 my ($self, $ra) = @_;
1682 return $ra if $self->svm;
1683
1684 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1685 "(svm:source, svm:uuid) ",
1686 "from the following URLs:\n" );
1687 sub read_svm_props {
1688 my ($self, $ra, $path, $r) = @_;
1689 my $props = ($ra->get_dir($path, $r))[2];
1690 my $src = $props->{'svm:source'};
1691 my $uuid = $props->{'svm:uuid'};
1692 return undef if (!$src || !$uuid);
1693
1694 chomp($src, $uuid);
1695
1696 $uuid =~ m{^[0-9a-f\-]{30,}$}
1697 or die "doesn't look right - svm:uuid is '$uuid'\n";
1698
1699 # the '!' is used to mark the repos_root!/relative/path
1700 $src =~ s{/?!/?}{/};
1701 $src =~ s{/+$}{}; # no trailing slashes please
1702 # username is of no interest
1703 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1704
1705 my $replace = $ra->{url};
1706 $replace .= "/$path" if length $path;
1707
1708 my $section = "svn-remote.$self->{repo_id}";
1709 tmp_config("$section.svm-source", $src);
1710 tmp_config("$section.svm-replace", $replace);
1711 tmp_config("$section.svm-uuid", $uuid);
1712 $self->{svm} = {
1713 source => $src,
1714 uuid => $uuid,
1715 replace => $replace
1716 };
1717 }
1718
1719 my $r = $ra->get_latest_revnum;
1720 my $path = $self->{path};
1721 my %tried;
1722 while (length $path) {
1723 unless ($tried{"$self->{url}/$path"}) {
1724 return $ra if $self->read_svm_props($ra, $path, $r);
1725 $tried{"$self->{url}/$path"} = 1;
1726 }
1727 $path =~ s#/?[^/]+$##;
1728 }
1729 die "Path: '$path' should be ''\n" if $path ne '';
1730 return $ra if $self->read_svm_props($ra, $path, $r);
1731 $tried{"$self->{url}/$path"} = 1;
1732
1733 if ($ra->{repos_root} eq $self->{url}) {
1734 die @err, (map { " $_\n" } keys %tried), "\n";
1735 }
1736
1737 # nope, make sure we're connected to the repository root:
1738 my $ok;
1739 my @tried_b;
1740 $path = $ra->{svn_path};
1741 $ra = Git::SVN::Ra->new($ra->{repos_root});
1742 while (length $path) {
1743 unless ($tried{"$ra->{url}/$path"}) {
1744 $ok = $self->read_svm_props($ra, $path, $r);
1745 last if $ok;
1746 $tried{"$ra->{url}/$path"} = 1;
1747 }
1748 $path =~ s#/?[^/]+$##;
1749 }
1750 die "Path: '$path' should be ''\n" if $path ne '';
1751 $ok ||= $self->read_svm_props($ra, $path, $r);
1752 $tried{"$ra->{url}/$path"} = 1;
1753 if (!$ok) {
1754 die @err, (map { " $_\n" } keys %tried), "\n";
1755 }
1756 Git::SVN::Ra->new($self->{url});
1757 }
1758
1759 sub svnsync {
1760 my ($self) = @_;
1761 return $self->{svnsync} if $self->{svnsync};
1762
1763 if ($self->no_metadata) {
1764 die "Can't have both 'noMetadata' and ",
1765 "'useSvnsyncProps' options set!\n";
1766 }
1767 if ($self->rewrite_root) {
1768 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1769 "options set!\n";
1770 }
1771
1772 my $svnsync;
1773 # see if we have it in our config, first:
1774 eval {
1775 my $section = "svn-remote.$self->{repo_id}";
1776
1777 my $url = tmp_config('--get', "$section.svnsync-url");
1778 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1779 die "doesn't look right - svn:sync-from-url is '$url'\n";
1780
1781 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
1782 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1783 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1784
1785 $svnsync = { url => $url, uuid => $uuid }
1786 };
1787 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1788 return $self->{svnsync} = $svnsync;
1789 }
1790
1791 my $err = "useSvnsyncProps set, but failed to read " .
1792 "svnsync property: svn:sync-from-";
1793 my $rp = $self->ra->rev_proplist(0);
1794
1795 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1796 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1797 die "doesn't look right - svn:sync-from-url is '$url'\n";
1798
1799 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1800 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1801 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1802
1803 my $section = "svn-remote.$self->{repo_id}";
1804 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1805 tmp_config('--add', "$section.svnsync-url", $url);
1806 return $self->{svnsync} = { url => $url, uuid => $uuid };
1807 }
1808
1809 # this allows us to memoize our SVN::Ra UUID locally and avoid a
1810 # remote lookup (useful for 'git svn log').
1811 sub ra_uuid {
1812 my ($self) = @_;
1813 unless ($self->{ra_uuid}) {
1814 my $key = "svn-remote.$self->{repo_id}.uuid";
1815 my $uuid = eval { tmp_config('--get', $key) };
1816 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1817 $self->{ra_uuid} = $uuid;
1818 } else {
1819 die "ra_uuid called without URL\n" unless $self->{url};
1820 $self->{ra_uuid} = $self->ra->get_uuid;
1821 tmp_config('--add', $key, $self->{ra_uuid});
1822 }
1823 }
1824 $self->{ra_uuid};
1825 }
1826
1827 sub _set_repos_root {
1828 my ($self, $repos_root) = @_;
1829 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1830 $repos_root ||= $self->ra->{repos_root};
1831 tmp_config($k, $repos_root);
1832 $repos_root;
1833 }
1834
1835 sub repos_root {
1836 my ($self) = @_;
1837 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1838 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1839 }
1840
1841 sub ra {
1842 my ($self) = shift;
1843 my $ra = Git::SVN::Ra->new($self->{url});
1844 $self->_set_repos_root($ra->{repos_root});
1845 if ($self->use_svm_props && !$self->{svm}) {
1846 if ($self->no_metadata) {
1847 die "Can't have both 'noMetadata' and ",
1848 "'useSvmProps' options set!\n";
1849 } elsif ($self->use_svnsync_props) {
1850 die "Can't have both 'useSvnsyncProps' and ",
1851 "'useSvmProps' options set!\n";
1852 }
1853 $ra = $self->_set_svm_vars($ra);
1854 $self->{-want_revprops} = 1;
1855 }
1856 $ra;
1857 }
1858
1859 sub rel_path {
1860 my ($self) = @_;
1861 my $repos_root = $self->ra->{repos_root};
1862 return $self->{path} if ($self->{url} eq $repos_root);
1863 my $url = $self->{url} .
1864 (length $self->{path} ? "/$self->{path}" : $self->{path});
1865 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1866 $url;
1867 }
1868
1869 # prop_walk(PATH, REV, SUB)
1870 # -------------------------
1871 # Recursively traverse PATH at revision REV and invoke SUB for each
1872 # directory that contains a SVN property. SUB will be invoked as
1873 # follows: &SUB(gs, path, props); where `gs' is this instance of
1874 # Git::SVN, `path' the path to the directory where the properties
1875 # `props' were found. The `path' will be relative to point of checkout,
1876 # that is, if url://repo/trunk is the current Git branch, and that
1877 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
1878 # as `path' (note the trailing `/').
1879 sub prop_walk {
1880 my ($self, $path, $rev, $sub) = @_;
1881
1882 $path =~ s#^/##;
1883 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1884 $path =~ s#^/*#/#g;
1885 my $p = $path;
1886 # Strip the irrelevant part of the path.
1887 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1888 # Ensure the path is terminated by a `/'.
1889 $p =~ s#/*$#/#;
1890
1891 # The properties contain all the internal SVN stuff nobody
1892 # (usually) cares about.
1893 my $interesting_props = 0;
1894 foreach (keys %{$props}) {
1895 # If it doesn't start with `svn:', it must be a
1896 # user-defined property.
1897 ++$interesting_props and next if $_ !~ /^svn:/;
1898 # FIXME: Fragile, if SVN adds new public properties,
1899 # this needs to be updated.
1900 ++$interesting_props if /^svn:(?:ignore|keywords|executable
1901 |eol-style|mime-type
1902 |externals|needs-lock)$/x;
1903 }
1904 &$sub($self, $p, $props) if $interesting_props;
1905
1906 foreach (sort keys %$dirent) {
1907 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1908 $self->prop_walk($p . $_, $rev, $sub);
1909 }
1910 }
1911
1912 sub last_rev { ($_[0]->last_rev_commit)[0] }
1913 sub last_commit { ($_[0]->last_rev_commit)[1] }
1914
1915 # returns the newest SVN revision number and newest commit SHA1
1916 sub last_rev_commit {
1917 my ($self) = @_;
1918 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1919 return ($self->{last_rev}, $self->{last_commit});
1920 }
1921 my $c = ::verify_ref($self->refname.'^0');
1922 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1923 my $rev = (::cmt_metadata($c))[1];
1924 if (defined $rev) {
1925 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1926 return ($rev, $c);
1927 }
1928 }
1929 my $map_path = $self->map_path;
1930 unless (-e $map_path) {
1931 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1932 return (undef, undef);
1933 }
1934 my ($rev, $commit) = $self->rev_map_max(1);
1935 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
1936 return ($rev, $commit);
1937 }
1938
1939 sub get_fetch_range {
1940 my ($self, $min, $max) = @_;
1941 $max ||= $self->ra->get_latest_revnum;
1942 $min ||= $self->rev_map_max;
1943 (++$min, $max);
1944 }
1945
1946 sub tmp_config {
1947 my (@args) = @_;
1948 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1949 my $config = "$ENV{GIT_DIR}/svn/.metadata";
1950 if (! -f $config && -f $old_def_config) {
1951 rename $old_def_config, $config or
1952 die "Failed rename $old_def_config => $config: $!\n";
1953 }
1954 my $old_config = $ENV{GIT_CONFIG};
1955 $ENV{GIT_CONFIG} = $config;
1956 $@ = undef;
1957 my @ret = eval {
1958 unless (-f $config) {
1959 mkfile($config);
1960 open my $fh, '>', $config or
1961 die "Can't open $config: $!\n";
1962 print $fh "; This file is used internally by ",
1963 "git-svn\n" or die
1964 "Couldn't write to $config: $!\n";
1965 print $fh "; You should not have to edit it\n" or
1966 die "Couldn't write to $config: $!\n";
1967 close $fh or die "Couldn't close $config: $!\n";
1968 }
1969 command('config', @args);
1970 };
1971 my $err = $@;
1972 if (defined $old_config) {
1973 $ENV{GIT_CONFIG} = $old_config;
1974 } else {
1975 delete $ENV{GIT_CONFIG};
1976 }
1977 die $err if $err;
1978 wantarray ? @ret : $ret[0];
1979 }
1980
1981 sub tmp_index_do {
1982 my ($self, $sub) = @_;
1983 my $old_index = $ENV{GIT_INDEX_FILE};
1984 $ENV{GIT_INDEX_FILE} = $self->{index};
1985 $@ = undef;
1986 my @ret = eval {
1987 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1988 mkpath([$dir]) unless -d $dir;
1989 &$sub;
1990 };
1991 my $err = $@;
1992 if (defined $old_index) {
1993 $ENV{GIT_INDEX_FILE} = $old_index;
1994 } else {
1995 delete $ENV{GIT_INDEX_FILE};
1996 }
1997 die $err if $err;
1998 wantarray ? @ret : $ret[0];
1999 }
2000
2001 sub assert_index_clean {
2002 my ($self, $treeish) = @_;
2003
2004 $self->tmp_index_do(sub {
2005 command_noisy('read-tree', $treeish) unless -e $self->{index};
2006 my $x = command_oneline('write-tree');
2007 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2008 /^tree ($::sha1)/mo);
2009 return if $y eq $x;
2010
2011 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2012 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2013 command_noisy('read-tree', $treeish);
2014 $x = command_oneline('write-tree');
2015 if ($y ne $x) {
2016 ::fatal "trees ($treeish) $y != $x\n",
2017 "Something is seriously wrong...";
2018 }
2019 });
2020 }
2021
2022 sub get_commit_parents {
2023 my ($self, $log_entry) = @_;
2024 my (%seen, @ret, @tmp);
2025 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2026 if (my $ip = $self->{inject_parents}) {
2027 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2028 push @tmp, $commit;
2029 }
2030 }
2031 if (my $cur = ::verify_ref($self->refname.'^0')) {
2032 push @tmp, $cur;
2033 }
2034 if (my $ipd = $self->{inject_parents_dcommit}) {
2035 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2036 push @tmp, @$commit;
2037 }
2038 }
2039 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2040 while (my $p = shift @tmp) {
2041 next if $seen{$p};
2042 $seen{$p} = 1;
2043 push @ret, $p;
2044 # MAXPARENT is defined to 16 in commit-tree.c:
2045 last if @ret >= 16;
2046 }
2047 if (@tmp) {
2048 die "r$log_entry->{revision}: No room for parents:\n\t",
2049 join("\n\t", @tmp), "\n";
2050 }
2051 @ret;
2052 }
2053
2054 sub rewrite_root {
2055 my ($self) = @_;
2056 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2057 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2058 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2059 if ($rwr) {
2060 $rwr =~ s#/+$##;
2061 if ($rwr !~ m#^[a-z\+]+://#) {
2062 die "$rwr is not a valid URL (key: $k)\n";
2063 }
2064 }
2065 $self->{-rewrite_root} = $rwr;
2066 }
2067
2068 sub metadata_url {
2069 my ($self) = @_;
2070 ($self->rewrite_root || $self->{url}) .
2071 (length $self->{path} ? '/' . $self->{path} : '');
2072 }
2073
2074 sub full_url {
2075 my ($self) = @_;
2076 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2077 }
2078
2079
2080 sub set_commit_header_env {
2081 my ($log_entry) = @_;
2082 my %env;
2083 foreach my $ned (qw/NAME EMAIL DATE/) {
2084 foreach my $ac (qw/AUTHOR COMMITTER/) {
2085 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2086 }
2087 }
2088
2089 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2090 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2091 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2092
2093 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2094 ? $log_entry->{commit_name}
2095 : $log_entry->{name};
2096 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2097 ? $log_entry->{commit_email}
2098 : $log_entry->{email};
2099 \%env;
2100 }
2101
2102 sub restore_commit_header_env {
2103 my ($env) = @_;
2104 foreach my $ned (qw/NAME EMAIL DATE/) {
2105 foreach my $ac (qw/AUTHOR COMMITTER/) {
2106 my $k = "GIT_${ac}_${ned}";
2107 if (defined $env->{$k}) {
2108 $ENV{$k} = $env->{$k};
2109 } else {
2110 delete $ENV{$k};
2111 }
2112 }
2113 }
2114 }
2115
2116 sub gc {
2117 command_noisy('gc', '--auto');
2118 };
2119
2120 sub do_git_commit {
2121 my ($self, $log_entry) = @_;
2122 my $lr = $self->last_rev;
2123 if (defined $lr && $lr >= $log_entry->{revision}) {
2124 die "Last fetched revision of ", $self->refname,
2125 " was r$lr, but we are about to fetch: ",
2126 "r$log_entry->{revision}!\n";
2127 }
2128 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2129 croak "$log_entry->{revision} = $c already exists! ",
2130 "Why are we refetching it?\n";
2131 }
2132 my $old_env = set_commit_header_env($log_entry);
2133 my $tree = $log_entry->{tree};
2134 if (!defined $tree) {
2135 $tree = $self->tmp_index_do(sub {
2136 command_oneline('write-tree') });
2137 }
2138 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2139
2140 my @exec = ('git-commit-tree', $tree);
2141 foreach ($self->get_commit_parents($log_entry)) {
2142 push @exec, '-p', $_;
2143 }
2144 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2145 or croak $!;
2146 print $msg_fh $log_entry->{log} or croak $!;
2147 restore_commit_header_env($old_env);
2148 unless ($self->no_metadata) {
2149 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2150 or croak $!;
2151 }
2152 $msg_fh->flush == 0 or croak $!;
2153 close $msg_fh or croak $!;
2154 chomp(my $commit = do { local $/; <$out_fh> });
2155 close $out_fh or croak $!;
2156 waitpid $pid, 0;
2157 croak $? if $?;
2158 if ($commit !~ /^$::sha1$/o) {
2159 die "Failed to commit, invalid sha1: $commit\n";
2160 }
2161
2162 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2163
2164 $self->{last_rev} = $log_entry->{revision};
2165 $self->{last_commit} = $commit;
2166 print "r$log_entry->{revision}";
2167 if (defined $log_entry->{svm_revision}) {
2168 print " (\@$log_entry->{svm_revision})";
2169 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2170 0, $self->svm_uuid);
2171 }
2172 print " = $commit ($self->{ref_id})\n";
2173 if (--$_gc_nr == 0) {
2174 $_gc_nr = $_gc_period;
2175 gc();
2176 }
2177 return $commit;
2178 }
2179
2180 sub match_paths {
2181 my ($self, $paths, $r) = @_;
2182 return 1 if $self->{path} eq '';
2183 if (my $path = $paths->{"/$self->{path}"}) {
2184 return ($path->{action} eq 'D') ? 0 : 1;
2185 }
2186 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2187 if (grep /$self->{path_regex}/, keys %$paths) {
2188 return 1;
2189 }
2190 my $c = '';
2191 foreach (split m#/#, $self->{path}) {
2192 $c .= "/$_";
2193 next unless ($paths->{$c} &&
2194 ($paths->{$c}->{action} =~ /^[AR]$/));
2195 if ($self->ra->check_path($self->{path}, $r) ==
2196 $SVN::Node::dir) {
2197 return 1;
2198 }
2199 }
2200 return 0;
2201 }
2202
2203 sub find_parent_branch {
2204 my ($self, $paths, $rev) = @_;
2205 return undef unless $self->follow_parent;
2206 unless (defined $paths) {
2207 my $err_handler = $SVN::Error::handler;
2208 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2209 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2210 $paths =
2211 Git::SVN::Ra::dup_changed_paths($_[0]) });
2212 $SVN::Error::handler = $err_handler;
2213 }
2214 return undef unless defined $paths;
2215
2216 # look for a parent from another branch:
2217 my @b_path_components = split m#/#, $self->rel_path;
2218 my @a_path_components;
2219 my $i;
2220 while (@b_path_components) {
2221 $i = $paths->{'/'.join('/', @b_path_components)};
2222 last if $i && defined $i->{copyfrom_path};
2223 unshift(@a_path_components, pop(@b_path_components));
2224 }
2225 return undef unless defined $i && defined $i->{copyfrom_path};
2226 my $branch_from = $i->{copyfrom_path};
2227 if (@a_path_components) {
2228 print STDERR "branch_from: $branch_from => ";
2229 $branch_from .= '/'.join('/', @a_path_components);
2230 print STDERR $branch_from, "\n";
2231 }
2232 my $r = $i->{copyfrom_rev};
2233 my $repos_root = $self->ra->{repos_root};
2234 my $url = $self->ra->{url};
2235 my $new_url = $repos_root . $branch_from;
2236 print STDERR "Found possible branch point: ",
2237 "$new_url => ", $self->full_url, ", $r\n";
2238 $branch_from =~ s#^/##;
2239 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2240 unless ($gs) {
2241 my $ref_id = $self->{ref_id};
2242 $ref_id =~ s/\@\d+$//;
2243 $ref_id .= "\@$r";
2244 # just grow a tail if we're not unique enough :x
2245 $ref_id .= '-' while find_ref($ref_id);
2246 print STDERR "Initializing parent: $ref_id\n";
2247 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2248 if ($u =~ s#^\Q$url\E(/|$)##) {
2249 $p = $u;
2250 $u = $url;
2251 $repo_id = $self->{repo_id};
2252 }
2253 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2254 }
2255 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2256 if (!defined $r0 || !defined $parent) {
2257 my ($base, $head) = parse_revision_argument(0, $r);
2258 if ($base <= $r) {
2259 $gs->fetch($base, $r);
2260 }
2261 ($r0, $parent) = $gs->last_rev_commit;
2262 }
2263 if (defined $r0 && defined $parent) {
2264 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2265 my $ed;
2266 if ($self->ra->can_do_switch) {
2267 $self->assert_index_clean($parent);
2268 print STDERR "Following parent with do_switch\n";
2269 # do_switch works with svn/trunk >= r22312, but that
2270 # is not included with SVN 1.4.3 (the latest version
2271 # at the moment), so we can't rely on it
2272 $self->{last_commit} = $parent;
2273 $ed = SVN::Git::Fetcher->new($self);
2274 $gs->ra->gs_do_switch($r0, $rev, $gs,
2275 $self->full_url, $ed)
2276 or die "SVN connection failed somewhere...\n";
2277 } elsif ($self->ra->trees_match($new_url, $r0,
2278 $self->full_url, $rev)) {
2279 print STDERR "Trees match:\n",
2280 " $new_url\@$r0\n",
2281 " ${\$self->full_url}\@$rev\n",
2282 "Following parent with no changes\n";
2283 $self->tmp_index_do(sub {
2284 command_noisy('read-tree', $parent);
2285 });
2286 $self->{last_commit} = $parent;
2287 } else {
2288 print STDERR "Following parent with do_update\n";
2289 $ed = SVN::Git::Fetcher->new($self);
2290 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2291 or die "SVN connection failed somewhere...\n";
2292 }
2293 print STDERR "Successfully followed parent\n";
2294 return $self->make_log_entry($rev, [$parent], $ed);
2295 }
2296 return undef;
2297 }
2298
2299 sub do_fetch {
2300 my ($self, $paths, $rev) = @_;
2301 my $ed;
2302 my ($last_rev, @parents);
2303 if (my $lc = $self->last_commit) {
2304 # we can have a branch that was deleted, then re-added
2305 # under the same name but copied from another path, in
2306 # which case we'll have multiple parents (we don't
2307 # want to break the original ref, nor lose copypath info):
2308 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2309 push @{$log_entry->{parents}}, $lc;
2310 return $log_entry;
2311 }
2312 $ed = SVN::Git::Fetcher->new($self);
2313 $last_rev = $self->{last_rev};
2314 $ed->{c} = $lc;
2315 @parents = ($lc);
2316 } else {
2317 $last_rev = $rev;
2318 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2319 return $log_entry;
2320 }
2321 $ed = SVN::Git::Fetcher->new($self);
2322 }
2323 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2324 die "SVN connection failed somewhere...\n";
2325 }
2326 $self->make_log_entry($rev, \@parents, $ed);
2327 }
2328
2329 sub get_untracked {
2330 my ($self, $ed) = @_;
2331 my @out;
2332 my $h = $ed->{empty};
2333 foreach (sort keys %$h) {
2334 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2335 push @out, " $act: " . uri_encode($_);
2336 warn "W: $act: $_\n";
2337 }
2338 foreach my $t (qw/dir_prop file_prop/) {
2339 $h = $ed->{$t} or next;
2340 foreach my $path (sort keys %$h) {
2341 my $ppath = $path eq '' ? '.' : $path;
2342 foreach my $prop (sort keys %{$h->{$path}}) {
2343 next if $SKIP_PROP{$prop};
2344 my $v = $h->{$path}->{$prop};
2345 my $t_ppath_prop = "$t: " .
2346 uri_encode($ppath) . ' ' .
2347 uri_encode($prop);
2348 if (defined $v) {
2349 push @out, " +$t_ppath_prop " .
2350 uri_encode($v);
2351 } else {
2352 push @out, " -$t_ppath_prop";
2353 }
2354 }
2355 }
2356 }
2357 foreach my $t (qw/absent_file absent_directory/) {
2358 $h = $ed->{$t} or next;
2359 foreach my $parent (sort keys %$h) {
2360 foreach my $path (sort @{$h->{$parent}}) {
2361 push @out, " $t: " .
2362 uri_encode("$parent/$path");
2363 warn "W: $t: $parent/$path ",
2364 "Insufficient permissions?\n";
2365 }
2366 }
2367 }
2368 \@out;
2369 }
2370
2371 sub parse_svn_date {
2372 my $date = shift || return '+0000 1970-01-01 00:00:00';
2373 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2374 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2375 croak "Unable to parse date: $date\n";
2376 "+0000 $Y-$m-$d $H:$M:$S";
2377 }
2378
2379 sub check_author {
2380 my ($author) = @_;
2381 if (!defined $author || length $author == 0) {
2382 $author = '(no author)';
2383 } elsif (defined $::_authors && ! defined $::users{$author}) {
2384 die "Author: $author not defined in $::_authors file\n";
2385 }
2386 $author;
2387 }
2388
2389 sub make_log_entry {
2390 my ($self, $rev, $parents, $ed) = @_;
2391 my $untracked = $self->get_untracked($ed);
2392
2393 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2394 print $un "r$rev\n" or croak $!;
2395 print $un $_, "\n" foreach @$untracked;
2396 my %log_entry = ( parents => $parents || [], revision => $rev,
2397 log => '');
2398
2399 my $headrev;
2400 my $logged = delete $self->{logged_rev_props};
2401 if (!$logged || $self->{-want_revprops}) {
2402 my $rp = $self->ra->rev_proplist($rev);
2403 foreach (sort keys %$rp) {
2404 my $v = $rp->{$_};
2405 if (/^svn:(author|date|log)$/) {
2406 $log_entry{$1} = $v;
2407 } elsif ($_ eq 'svm:headrev') {
2408 $headrev = $v;
2409 } else {
2410 print $un " rev_prop: ", uri_encode($_), ' ',
2411 uri_encode($v), "\n";
2412 }
2413 }
2414 } else {
2415 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2416 }
2417 close $un or croak $!;
2418
2419 $log_entry{date} = parse_svn_date($log_entry{date});
2420 $log_entry{log} .= "\n";
2421 my $author = $log_entry{author} = check_author($log_entry{author});
2422 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2423 : ($author, undef);
2424
2425 my ($commit_name, $commit_email) = ($name, $email);
2426 if ($_use_log_author) {
2427 my $name_field;
2428 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2429 $name_field = $1;
2430 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2431 $name_field = $1;
2432 }
2433 if (!defined $name_field) {
2434 if (!defined $email) {
2435 $email = $name;
2436 }
2437 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2438 ($name, $email) = ($1, $2);
2439 } elsif ($name_field =~ /(.*)@/) {
2440 ($name, $email) = ($1, $name_field);
2441 } else {
2442 ($name, $email) = ($name_field, $name_field);
2443 }
2444 }
2445 if (defined $headrev && $self->use_svm_props) {
2446 if ($self->rewrite_root) {
2447 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2448 "options set!\n";
2449 }
2450 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2451 # we don't want "SVM: initializing mirror for junk" ...
2452 return undef if $r == 0;
2453 my $svm = $self->svm;
2454 if ($uuid ne $svm->{uuid}) {
2455 die "UUID mismatch on SVM path:\n",
2456 "expected: $svm->{uuid}\n",
2457 " got: $uuid\n";
2458 }
2459 my $full_url = $self->full_url;
2460 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2461 die "Failed to replace '$svm->{replace}' with ",
2462 "'$svm->{source}' in $full_url\n";
2463 # throw away username for storing in records
2464 remove_username($full_url);
2465 $log_entry{metadata} = "$full_url\@$r $uuid";
2466 $log_entry{svm_revision} = $r;
2467 $email ||= "$author\@$uuid";
2468 $commit_email ||= "$author\@$uuid";
2469 } elsif ($self->use_svnsync_props) {
2470 my $full_url = $self->svnsync->{url};
2471 $full_url .= "/$self->{path}" if length $self->{path};
2472 remove_username($full_url);
2473 my $uuid = $self->svnsync->{uuid};
2474 $log_entry{metadata} = "$full_url\@$rev $uuid";
2475 $email ||= "$author\@$uuid";
2476 $commit_email ||= "$author\@$uuid";
2477 } else {
2478 my $url = $self->metadata_url;
2479 remove_username($url);
2480 $log_entry{metadata} = "$url\@$rev " .
2481 $self->ra->get_uuid;
2482 $email ||= "$author\@" . $self->ra->get_uuid;
2483 $commit_email ||= "$author\@" . $self->ra->get_uuid;
2484 }
2485 $log_entry{name} = $name;
2486 $log_entry{email} = $email;
2487 $log_entry{commit_name} = $commit_name;
2488 $log_entry{commit_email} = $commit_email;
2489 \%log_entry;
2490 }
2491
2492 sub fetch {
2493 my ($self, $min_rev, $max_rev, @parents) = @_;
2494 my ($last_rev, $last_commit) = $self->last_rev_commit;
2495 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2496 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2497 }
2498
2499 sub set_tree_cb {
2500 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2501 $self->{inject_parents} = { $rev => $tree };
2502 $self->fetch(undef, undef);
2503 }
2504
2505 sub set_tree {
2506 my ($self, $tree) = (shift, shift);
2507 my $log_entry = ::get_commit_entry($tree);
2508 unless ($self->{last_rev}) {
2509 fatal("Must have an existing revision to commit");
2510 }
2511 my %ed_opts = ( r => $self->{last_rev},
2512 log => $log_entry->{log},
2513 ra => $self->ra,
2514 tree_a => $self->{last_commit},
2515 tree_b => $tree,
2516 editor_cb => sub {
2517 $self->set_tree_cb($log_entry, $tree, @_) },
2518 svn_path => $self->{path} );
2519 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2520 print "No changes\nr$self->{last_rev} = $tree\n";
2521 }
2522 }
2523
2524 sub rebuild_from_rev_db {
2525 my ($self, $path) = @_;
2526 my $r = -1;
2527 open my $fh, '<', $path or croak "open: $!";
2528 binmode $fh or croak "binmode: $!";
2529 while (<$fh>) {
2530 length($_) == 41 or croak "inconsistent size in ($_) != 41";
2531 chomp($_);
2532 ++$r;
2533 next if $_ eq ('0' x 40);
2534 $self->rev_map_set($r, $_);
2535 print "r$r = $_\n";
2536 }
2537 close $fh or croak "close: $!";
2538 unlink $path or croak "unlink: $!";
2539 }
2540
2541 sub rebuild {
2542 my ($self) = @_;
2543 my $map_path = $self->map_path;
2544 return if (-e $map_path && ! -z $map_path);
2545 return unless ::verify_ref($self->refname.'^0');
2546 if ($self->use_svm_props || $self->no_metadata) {
2547 my $rev_db = $self->rev_db_path;
2548 $self->rebuild_from_rev_db($rev_db);
2549 if ($self->use_svm_props) {
2550 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2551 $self->rebuild_from_rev_db($svm_rev_db);
2552 }
2553 $self->unlink_rev_db_symlink;
2554 return;
2555 }
2556 print "Rebuilding $map_path ...\n";
2557 my ($log, $ctx) =
2558 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2559 $self->refname, '--');
2560 my $full_url = $self->full_url;
2561 remove_username($full_url);
2562 my $svn_uuid = $self->ra_uuid;
2563 my $c;
2564 while (<$log>) {
2565 if ( m{^commit ($::sha1)$} ) {
2566 $c = $1;
2567 next;
2568 }
2569 next unless s{^\s*(git-svn-id:)}{$1};
2570 my ($url, $rev, $uuid) = ::extract_metadata($_);
2571 remove_username($url);
2572
2573 # ignore merges (from set-tree)
2574 next if (!defined $rev || !$uuid);
2575
2576 # if we merged or otherwise started elsewhere, this is
2577 # how we break out of it
2578 if (($uuid ne $svn_uuid) ||
2579 ($full_url && $url && ($url ne $full_url))) {
2580 next;
2581 }
2582
2583 $self->rev_map_set($rev, $c);
2584 print "r$rev = $c\n";
2585 }
2586 command_close_pipe($log, $ctx);
2587 print "Done rebuilding $map_path\n";
2588 my $rev_db_path = $self->rev_db_path;
2589 if (-f $self->rev_db_path) {
2590 unlink $self->rev_db_path or croak "unlink: $!";
2591 }
2592 $self->unlink_rev_db_symlink;
2593 }
2594
2595 # rev_map:
2596 # Tie::File seems to be prone to offset errors if revisions get sparse,
2597 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
2598 # one of my favorite modules is out :< Next up would be one of the DBM
2599 # modules, but I'm not sure which is most portable...
2600 #
2601 # This is the replacement for the rev_db format, which was too big
2602 # and inefficient for large repositories with a lot of sparse history
2603 # (mainly tags)
2604 #
2605 # The format is this:
2606 # - 24 bytes for every record,
2607 # * 4 bytes for the integer representing an SVN revision number
2608 # * 20 bytes representing the sha1 of a git commit
2609 # - No empty padding records like the old format
2610 # (except the last record, which can be overwritten)
2611 # - new records are written append-only since SVN revision numbers
2612 # increase monotonically
2613 # - lookups on SVN revision number are done via a binary search
2614 # - Piping the file to xxd -c24 is a good way of dumping it for
2615 # viewing or editing (piped back through xxd -r), should the need
2616 # ever arise.
2617 # - The last record can be padding revision with an all-zero sha1
2618 # This is used to optimize fetch performance when using multiple
2619 # "fetch" directives in .git/config
2620 #
2621 # These files are disposable unless noMetadata or useSvmProps is set
2622
2623 sub _rev_map_set {
2624 my ($fh, $rev, $commit) = @_;
2625
2626 binmode $fh or croak "binmode: $!";
2627 my $size = (stat($fh))[7];
2628 ($size % 24) == 0 or croak "inconsistent size: $size";
2629
2630 my $wr_offset = 0;
2631 if ($size > 0) {
2632 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2633 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2634 $read == 24 or croak "read only $read bytes (!= 24)";
2635 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2636 if ($last_commit eq ('0' x40)) {
2637 if ($size >= 48) {
2638 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2639 $read = sysread($fh, $buf, 24) or
2640 croak "read: $!";
2641 $read == 24 or
2642 croak "read only $read bytes (!= 24)";
2643 ($last_rev, $last_commit) =
2644 unpack(rev_map_fmt, $buf);
2645 if ($last_commit eq ('0' x40)) {
2646 croak "inconsistent .rev_map\n";
2647 }
2648 }
2649 if ($last_rev >= $rev) {
2650 croak "last_rev is higher!: $last_rev >= $rev";
2651 }
2652 $wr_offset = -24;
2653 }
2654 }
2655 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2656 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2657 croak "write: $!";
2658 }
2659
2660 sub mkfile {
2661 my ($path) = @_;
2662 unless (-e $path) {
2663 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2664 mkpath([$dir]) unless -d $dir;
2665 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2666 close $fh or die "Couldn't close (create) $path: $!\n";
2667 }
2668 }
2669
2670 sub rev_map_set {
2671 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2672 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2673 my $db = $self->map_path($uuid);
2674 my $db_lock = "$db.lock";
2675 my $sig;
2676 if ($update_ref) {
2677 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2678 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2679 }
2680 mkfile($db);
2681
2682 $LOCKFILES{$db_lock} = 1;
2683 my $sync;
2684 # both of these options make our .rev_db file very, very important
2685 # and we can't afford to lose it because rebuild() won't work
2686 if ($self->use_svm_props || $self->no_metadata) {
2687 $sync = 1;
2688 copy($db, $db_lock) or die "rev_map_set(@_): ",
2689 "Failed to copy: ",
2690 "$db => $db_lock ($!)\n";
2691 } else {
2692 rename $db, $db_lock or die "rev_map_set(@_): ",
2693 "Failed to rename: ",
2694 "$db => $db_lock ($!)\n";
2695 }
2696
2697 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2698 or croak "Couldn't open $db_lock: $!\n";
2699 _rev_map_set($fh, $rev, $commit);
2700 if ($sync) {
2701 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2702 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2703 }
2704 close $fh or croak $!;
2705 if ($update_ref) {
2706 $_head = $self;
2707 command_noisy('update-ref', '-m', "r$rev",
2708 $self->refname, $commit);
2709 }
2710 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2711 "$db_lock => $db ($!)\n";
2712 delete $LOCKFILES{$db_lock};
2713 if ($update_ref) {
2714 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2715 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2716 kill $sig, $$ if defined $sig;
2717 }
2718 }
2719
2720 # If want_commit, this will return an array of (rev, commit) where
2721 # commit _must_ be a valid commit in the archive.
2722 # Otherwise, it'll return the max revision (whether or not the
2723 # commit is valid or just a 0x40 placeholder).
2724 sub rev_map_max {
2725 my ($self, $want_commit) = @_;
2726 $self->rebuild;
2727 my $map_path = $self->map_path;
2728 stat $map_path or return $want_commit ? (0, undef) : 0;
2729 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2730 binmode $fh or croak "binmode: $!";
2731 my $size = (stat($fh))[7];
2732 ($size % 24) == 0 or croak "inconsistent size: $size";
2733
2734 if ($size == 0) {
2735 close $fh or croak "close: $!";
2736 return $want_commit ? (0, undef) : 0;
2737 }
2738
2739 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2740 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2741 my ($r, $c) = unpack(rev_map_fmt, $buf);
2742 if ($want_commit && $c eq ('0' x40)) {
2743 if ($size < 48) {
2744 return $want_commit ? (0, undef) : 0;
2745 }
2746 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2747 sysread($fh, $buf, 24) == 24 or croak "read: $!";
2748 ($r, $c) = unpack(rev_map_fmt, $buf);
2749 if ($c eq ('0'x40)) {
2750 croak "Penultimate record is all-zeroes in $map_path";
2751 }
2752 }
2753 close $fh or croak "close: $!";
2754 $want_commit ? ($r, $c) : $r;
2755 }
2756
2757 sub rev_map_get {
2758 my ($self, $rev, $uuid) = @_;
2759 my $map_path = $self->map_path($uuid);
2760 return undef unless -e $map_path;
2761
2762 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2763 binmode $fh or croak "binmode: $!";
2764 my $size = (stat($fh))[7];
2765 ($size % 24) == 0 or croak "inconsistent size: $size";
2766
2767 if ($size == 0) {
2768 close $fh or croak "close: $fh";
2769 return undef;
2770 }
2771
2772 my ($l, $u) = (0, $size - 24);
2773 my ($r, $c, $buf);
2774
2775 while ($l <= $u) {
2776 my $i = int(($l/24 + $u/24) / 2) * 24;
2777 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2778 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2779 my ($r, $c) = unpack('NH40', $buf);
2780
2781 if ($r < $rev) {
2782 $l = $i + 24;
2783 } elsif ($r > $rev) {
2784 $u = $i - 24;
2785 } else { # $r == $rev
2786 close($fh) or croak "close: $!";
2787 return $c eq ('0' x 40) ? undef : $c;
2788 }
2789 }
2790 close($fh) or croak "close: $!";
2791 undef;
2792 }
2793
2794 # Finds the first svn revision that exists on (if $eq_ok is true) or
2795 # before $rev for the current branch. It will not search any lower
2796 # than $min_rev. Returns the git commit hash and svn revision number
2797 # if found, else (undef, undef).
2798 sub find_rev_before {
2799 my ($self, $rev, $eq_ok, $min_rev) = @_;
2800 --$rev unless $eq_ok;
2801 $min_rev ||= 1;
2802 while ($rev >= $min_rev) {
2803 if (my $c = $self->rev_map_get($rev)) {
2804 return ($rev, $c);
2805 }
2806 --$rev;
2807 }
2808 return (undef, undef);
2809 }
2810
2811 # Finds the first svn revision that exists on (if $eq_ok is true) or
2812 # after $rev for the current branch. It will not search any higher
2813 # than $max_rev. Returns the git commit hash and svn revision number
2814 # if found, else (undef, undef).
2815 sub find_rev_after {
2816 my ($self, $rev, $eq_ok, $max_rev) = @_;
2817 ++$rev unless $eq_ok;
2818 $max_rev ||= $self->rev_map_max;
2819 while ($rev <= $max_rev) {
2820 if (my $c = $self->rev_map_get($rev)) {
2821 return ($rev, $c);
2822 }
2823 ++$rev;
2824 }
2825 return (undef, undef);
2826 }
2827
2828 sub _new {
2829 my ($class, $repo_id, $ref_id, $path) = @_;
2830 unless (defined $repo_id && length $repo_id) {
2831 $repo_id = $Git::SVN::default_repo_id;
2832 }
2833 unless (defined $ref_id && length $ref_id) {
2834 $_[2] = $ref_id = $Git::SVN::default_ref_id;
2835 }
2836 $_[1] = $repo_id = sanitize_remote_name($repo_id);
2837 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2838 $_[3] = $path = '' unless (defined $path);
2839 mkpath(["$ENV{GIT_DIR}/svn"]);
2840 bless {
2841 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2842 path => $path, config => "$ENV{GIT_DIR}/svn/config",
2843 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
2844 }
2845
2846 # for read-only access of old .rev_db formats
2847 sub unlink_rev_db_symlink {
2848 my ($self) = @_;
2849 my $link = $self->rev_db_path;
2850 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2851 if (-l $link) {
2852 unlink $link or croak "unlink: $link failed!";
2853 }
2854 }
2855
2856 sub rev_db_path {
2857 my ($self, $uuid) = @_;
2858 my $db_path = $self->map_path($uuid);
2859 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2860 or croak "map_path: $db_path does not contain '/.rev_map.' !";
2861 $db_path;
2862 }
2863
2864 # the new replacement for .rev_db
2865 sub map_path {
2866 my ($self, $uuid) = @_;
2867 $uuid ||= $self->ra_uuid;
2868 "$self->{map_root}.$uuid";
2869 }
2870
2871 sub uri_encode {
2872 my ($f) = @_;
2873 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2874 $f
2875 }
2876
2877 sub remove_username {
2878 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2879 }
2880
2881 package Git::SVN::Prompt;
2882 use strict;
2883 use warnings;
2884 require SVN::Core;
2885 use vars qw/$_no_auth_cache $_username/;
2886
2887 sub simple {
2888 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2889 $may_save = undef if $_no_auth_cache;
2890 $default_username = $_username if defined $_username;
2891 if (defined $default_username && length $default_username) {
2892 if (defined $realm && length $realm) {
2893 print STDERR "Authentication realm: $realm\n";
2894 STDERR->flush;
2895 }
2896 $cred->username($default_username);
2897 } else {
2898 username($cred, $realm, $may_save, $pool);
2899 }
2900 $cred->password(_read_password("Password for '" .
2901 $cred->username . "': ", $realm));
2902 $cred->may_save($may_save);
2903 $SVN::_Core::SVN_NO_ERROR;
2904 }
2905
2906 sub ssl_server_trust {
2907 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2908 $may_save = undef if $_no_auth_cache;
2909 print STDERR "Error validating server certificate for '$realm':\n";
2910 {
2911 no warnings 'once';
2912 # All variables SVN::Auth::SSL::* are used only once,
2913 # so we're shutting up Perl warnings about this.
2914 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2915 print STDERR " - The certificate is not issued ",
2916 "by a trusted authority. Use the\n",
2917 " fingerprint to validate ",
2918 "the certificate manually!\n";
2919 }
2920 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2921 print STDERR " - The certificate hostname ",
2922 "does not match.\n";
2923 }
2924 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2925 print STDERR " - The certificate is not yet valid.\n";
2926 }
2927 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2928 print STDERR " - The certificate has expired.\n";
2929 }
2930 if ($failures & $SVN::Auth::SSL::OTHER) {
2931 print STDERR " - The certificate has ",
2932 "an unknown error.\n";
2933 }
2934 } # no warnings 'once'
2935 printf STDERR
2936 "Certificate information:\n".
2937 " - Hostname: %s\n".
2938 " - Valid: from %s until %s\n".
2939 " - Issuer: %s\n".
2940 " - Fingerprint: %s\n",
2941 map $cert_info->$_, qw(hostname valid_from valid_until
2942 issuer_dname fingerprint);
2943 my $choice;
2944 prompt:
2945 print STDERR $may_save ?
2946 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2947 "(R)eject or accept (t)emporarily? ";
2948 STDERR->flush;
2949 $choice = lc(substr(<STDIN> || 'R', 0, 1));
2950 if ($choice =~ /^t$/i) {
2951 $cred->may_save(undef);
2952 } elsif ($choice =~ /^r$/i) {
2953 return -1;
2954 } elsif ($may_save && $choice =~ /^p$/i) {
2955 $cred->may_save($may_save);
2956 } else {
2957 goto prompt;
2958 }
2959 $cred->accepted_failures($failures);
2960 $SVN::_Core::SVN_NO_ERROR;
2961 }
2962
2963 sub ssl_client_cert {
2964 my ($cred, $realm, $may_save, $pool) = @_;
2965 $may_save = undef if $_no_auth_cache;
2966 print STDERR "Client certificate filename: ";
2967 STDERR->flush;
2968 chomp(my $filename = <STDIN>);
2969 $cred->cert_file($filename);
2970 $cred->may_save($may_save);
2971 $SVN::_Core::SVN_NO_ERROR;
2972 }
2973
2974 sub ssl_client_cert_pw {
2975 my ($cred, $realm, $may_save, $pool) = @_;
2976 $may_save = undef if $_no_auth_cache;
2977 $cred->password(_read_password("Password: ", $realm));
2978 $cred->may_save($may_save);
2979 $SVN::_Core::SVN_NO_ERROR;
2980 }
2981
2982 sub username {
2983 my ($cred, $realm, $may_save, $pool) = @_;
2984 $may_save = undef if $_no_auth_cache;
2985 if (defined $realm && length $realm) {
2986 print STDERR "Authentication realm: $realm\n";
2987 }
2988 my $username;
2989 if (defined $_username) {
2990 $username = $_username;
2991 } else {
2992 print STDERR "Username: ";
2993 STDERR->flush;
2994 chomp($username = <STDIN>);
2995 }
2996 $cred->username($username);
2997 $cred->may_save($may_save);
2998 $SVN::_Core::SVN_NO_ERROR;
2999 }
3000
3001 sub _read_password {
3002 my ($prompt, $realm) = @_;
3003 print STDERR $prompt;
3004 STDERR->flush;
3005 require Term::ReadKey;
3006 Term::ReadKey::ReadMode('noecho');
3007 my $password = '';
3008 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3009 last if $key =~ /[\012\015]/; # \n\r
3010 $password .= $key;
3011 }
3012 Term::ReadKey::ReadMode('restore');
3013 print STDERR "\n";
3014 STDERR->flush;
3015 $password;
3016 }
3017
3018 package SVN::Git::Fetcher;
3019 use vars qw/@ISA/;
3020 use strict;
3021 use warnings;
3022 use Carp qw/croak/;
3023 use File::Temp qw/tempfile/;
3024 use IO::File qw//;
3025
3026 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
3027 sub new {
3028 my ($class, $git_svn) = @_;
3029 my $self = SVN::Delta::Editor->new;
3030 bless $self, $class;
3031 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
3032 $self->{empty} = {};
3033 $self->{dir_prop} = {};
3034 $self->{file_prop} = {};
3035 $self->{absent_dir} = {};
3036 $self->{absent_file} = {};
3037 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3038 $self;
3039 }
3040
3041 sub set_path_strip {
3042 my ($self, $path) = @_;
3043 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3044 }
3045
3046 sub open_root {
3047 { path => '' };
3048 }
3049
3050 sub open_directory {
3051 my ($self, $path, $pb, $rev) = @_;
3052 { path => $path };
3053 }
3054
3055 sub git_path {
3056 my ($self, $path) = @_;
3057 if ($self->{path_strip}) {
3058 $path =~ s!$self->{path_strip}!! or
3059 die "Failed to strip path '$path' ($self->{path_strip})\n";
3060 }
3061 $path;
3062 }
3063
3064 sub delete_entry {
3065 my ($self, $path, $rev, $pb) = @_;
3066
3067 my $gpath = $self->git_path($path);
3068 return undef if ($gpath eq '');
3069
3070 # remove entire directories.
3071 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
3072 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3073 -r --name-only -z/,
3074 $self->{c}, '--', $gpath);
3075 local $/ = "\0";
3076 while (<$ls>) {
3077 chomp;
3078 $self->{gii}->remove($_);
3079 print "\tD\t$_\n" unless $::_q;
3080 }
3081 print "\tD\t$gpath/\n" unless $::_q;
3082 command_close_pipe($ls, $ctx);
3083 $self->{empty}->{$path} = 0
3084 } else {
3085 $self->{gii}->remove($gpath);
3086 print "\tD\t$gpath\n" unless $::_q;
3087 }
3088 undef;
3089 }
3090
3091 sub open_file {
3092 my ($self, $path, $pb, $rev) = @_;
3093 my $gpath = $self->git_path($path);
3094 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
3095 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3096 unless (defined $mode && defined $blob) {
3097 die "$path was not found in commit $self->{c} (r$rev)\n";
3098 }
3099 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3100 pool => SVN::Pool->new, action => 'M' };
3101 }
3102
3103 sub add_file {
3104 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3105 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3106 delete $self->{empty}->{$dir};
3107 { path => $path, mode_a => 100644, mode_b => 100644,
3108 pool => SVN::Pool->new, action => 'A' };
3109 }
3110
3111 sub add_directory {
3112 my ($self, $path, $cp_path, $cp_rev) = @_;
3113 my $gpath = $self->git_path($path);
3114 if ($gpath eq '') {
3115 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3116 -r --name-only -z/,
3117 $self->{c});
3118 local $/ = "\0";
3119 while (<$ls>) {
3120 chomp;
3121 $self->{gii}->remove($_);
3122 print "\tD\t$_\n" unless $::_q;
3123 }
3124 command_close_pipe($ls, $ctx);
3125 $self->{empty}->{$path} = 0;
3126 }
3127 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3128 delete $self->{empty}->{$dir};
3129 $self->{empty}->{$path} = 1;
3130 { path => $path };
3131 }
3132
3133 sub change_dir_prop {
3134 my ($self, $db, $prop, $value) = @_;
3135 $self->{dir_prop}->{$db->{path}} ||= {};
3136 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3137 undef;
3138 }
3139
3140 sub absent_directory {
3141 my ($self, $path, $pb) = @_;
3142 $self->{absent_dir}->{$pb->{path}} ||= [];
3143 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3144 undef;
3145 }
3146
3147 sub absent_file {
3148 my ($self, $path, $pb) = @_;
3149 $self->{absent_file}->{$pb->{path}} ||= [];
3150 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3151 undef;
3152 }
3153
3154 sub change_file_prop {
3155 my ($self, $fb, $prop, $value) = @_;
3156 if ($prop eq 'svn:executable') {
3157 if ($fb->{mode_b} != 120000) {
3158 $fb->{mode_b} = defined $value ? 100755 : 100644;
3159 }
3160 } elsif ($prop eq 'svn:special') {
3161 $fb->{mode_b} = defined $value ? 120000 : 100644;
3162 } else {
3163 $self->{file_prop}->{$fb->{path}} ||= {};
3164 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3165 }
3166 undef;
3167 }
3168
3169 sub apply_textdelta {
3170 my ($self, $fb, $exp) = @_;
3171 my $fh = IO::File->new_tmpfile;
3172 $fh->autoflush(1);
3173 # $fh gets auto-closed() by SVN::TxDelta::apply(),
3174 # (but $base does not,) so dup() it for reading in close_file
3175 open my $dup, '<&', $fh or croak $!;
3176 my $base = IO::File->new_tmpfile;
3177 $base->autoflush(1);
3178 if ($fb->{blob}) {
3179 print $base 'link ' if ($fb->{mode_a} == 120000);
3180 my $size = $::_repository->cat_blob($fb->{blob}, $base);
3181 die "Failed to read object $fb->{blob}" unless $size;
3182
3183 if (defined $exp) {
3184 seek $base, 0, 0 or croak $!;
3185 my $got = ::md5sum($base);
3186 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3187 "expected: $exp\n",
3188 " got: $got\n" if ($got ne $exp);
3189 }
3190 }
3191 seek $base, 0, 0 or croak $!;
3192 $fb->{fh} = $dup;
3193 $fb->{base} = $base;
3194 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
3195 }
3196
3197 sub close_file {
3198 my ($self, $fb, $exp) = @_;
3199 my $hash;
3200 my $path = $self->git_path($fb->{path});
3201 if (my $fh = $fb->{fh}) {
3202 if (defined $exp) {
3203 seek($fh, 0, 0) or croak $!;
3204 my $got = ::md5sum($fh);
3205 if ($got ne $exp) {
3206 die "Checksum mismatch: $path\n",
3207 "expected: $exp\n got: $got\n";
3208 }
3209 }
3210 sysseek($fh, 0, 0) or croak $!;
3211 if ($fb->{mode_b} == 120000) {
3212 eval {
3213 sysread($fh, my $buf, 5) == 5 or croak $!;
3214 $buf eq 'link ' or die "$path has mode 120000",
3215 " but is not a link";
3216 };
3217 if ($@) {
3218 warn "$@\n";
3219 sysseek($fh, 0, 0) or croak $!;
3220 }
3221 }
3222
3223 my ($tmp_fh, $tmp_filename) = File::Temp::tempfile(UNLINK => 1);
3224 my $result;
3225 while ($result = sysread($fh, my $string, 1024)) {
3226 syswrite($tmp_fh, $string, $result);
3227 }
3228 defined $result or croak $!;
3229 close $tmp_fh or croak $!;
3230
3231 close $fh or croak $!;
3232
3233 $hash = $::_repository->hash_and_insert_object($tmp_filename);
3234 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3235 close $fb->{base} or croak $!;
3236 } else {
3237 $hash = $fb->{blob} or die "no blob information\n";
3238 }
3239 $fb->{pool}->clear;
3240 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3241 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3242 undef;
3243 }
3244
3245 sub abort_edit {
3246 my $self = shift;
3247 $self->{nr} = $self->{gii}->{nr};
3248 delete $self->{gii};
3249 $self->SUPER::abort_edit(@_);
3250 }
3251
3252 sub close_edit {
3253 my $self = shift;
3254 $self->{git_commit_ok} = 1;
3255 $self->{nr} = $self->{gii}->{nr};
3256 delete $self->{gii};
3257 $self->SUPER::close_edit(@_);
3258 }
3259
3260 package SVN::Git::Editor;
3261 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3262 use strict;
3263 use warnings;
3264 use Carp qw/croak/;
3265 use IO::File;
3266
3267 sub new {
3268 my ($class, $opts) = @_;
3269 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3270 die "$_ required!\n" unless (defined $opts->{$_});
3271 }
3272
3273 my $pool = SVN::Pool->new;
3274 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3275 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3276 $opts->{r}, $mods);
3277
3278 # $opts->{ra} functions should not be used after this:
3279 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3280 $opts->{editor_cb}, $pool);
3281 my $self = SVN::Delta::Editor->new(@ce, $pool);
3282 bless $self, $class;
3283 foreach (qw/svn_path r tree_a tree_b/) {
3284 $self->{$_} = $opts->{$_};
3285 }
3286 $self->{url} = $opts->{ra}->{url};
3287 $self->{mods} = $mods;
3288 $self->{types} = $types;
3289 $self->{pool} = $pool;
3290 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3291 $self->{rm} = { };
3292 $self->{path_prefix} = length $self->{svn_path} ?
3293 "$self->{svn_path}/" : '';
3294 return $self;
3295 }
3296
3297 sub generate_diff {
3298 my ($tree_a, $tree_b) = @_;
3299 my @diff_tree = qw(diff-tree -z -r);
3300 if ($_cp_similarity) {
3301 push @diff_tree, "-C$_cp_similarity";
3302 } else {
3303 push @diff_tree, '-C';
3304 }
3305 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3306 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3307 push @diff_tree, $tree_a, $tree_b;
3308 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3309 local $/ = "\0";
3310 my $state = 'meta';
3311 my @mods;
3312 while (<$diff_fh>) {
3313 chomp $_; # this gets rid of the trailing "\0"
3314 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3315 $::sha1\s($::sha1)\s
3316 ([MTCRAD])\d*$/xo) {
3317 push @mods, { mode_a => $1, mode_b => $2,
3318 sha1_b => $3, chg => $4 };
3319 if ($4 =~ /^(?:C|R)$/) {
3320 $state = 'file_a';
3321 } else {
3322 $state = 'file_b';
3323 }
3324 } elsif ($state eq 'file_a') {
3325 my $x = $mods[$#mods] or croak "Empty array\n";
3326 if ($x->{chg} !~ /^(?:C|R)$/) {
3327 croak "Error parsing $_, $x->{chg}\n";
3328 }
3329 $x->{file_a} = $_;
3330 $state = 'file_b';
3331 } elsif ($state eq 'file_b') {
3332 my $x = $mods[$#mods] or croak "Empty array\n";
3333 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3334 croak "Error parsing $_, $x->{chg}\n";
3335 }
3336 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3337 croak "Error parsing $_, $x->{chg}\n";
3338 }
3339 $x->{file_b} = $_;
3340 $state = 'meta';
3341 } else {
3342 croak "Error parsing $_\n";
3343 }
3344 }
3345 command_close_pipe($diff_fh, $ctx);
3346 \@mods;
3347 }
3348
3349 sub check_diff_paths {
3350 my ($ra, $pfx, $rev, $mods) = @_;
3351 my %types;
3352 $pfx .= '/' if length $pfx;
3353
3354 sub type_diff_paths {
3355 my ($ra, $types, $path, $rev) = @_;
3356 my @p = split m#/+#, $path;
3357 my $c = shift @p;
3358 unless (defined $types->{$c}) {
3359 $types->{$c} = $ra->check_path($c, $rev);
3360 }
3361 while (@p) {
3362 $c .= '/' . shift @p;
3363 next if defined $types->{$c};
3364 $types->{$c} = $ra->check_path($c, $rev);
3365 }
3366 }
3367
3368 foreach my $m (@$mods) {
3369 foreach my $f (qw/file_a file_b/) {
3370 next unless defined $m->{$f};
3371 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3372 if (length $pfx.$dir && ! defined $types{$dir}) {
3373 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3374 }
3375 }
3376 }
3377 \%types;
3378 }
3379
3380 sub split_path {
3381 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3382 }
3383
3384 sub repo_path {
3385 my ($self, $path) = @_;
3386 $self->{path_prefix}.(defined $path ? $path : '');
3387 }
3388
3389 sub url_path {
3390 my ($self, $path) = @_;
3391 if ($self->{url} =~ m#^https?://#) {
3392 $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3393 }
3394 $self->{url} . '/' . $self->repo_path($path);
3395 }
3396
3397 sub rmdirs {
3398 my ($self) = @_;
3399 my $rm = $self->{rm};
3400 delete $rm->{''}; # we never delete the url we're tracking
3401 return unless %$rm;
3402
3403 foreach (keys %$rm) {
3404 my @d = split m#/#, $_;
3405 my $c = shift @d;
3406 $rm->{$c} = 1;
3407 while (@d) {
3408 $c .= '/' . shift @d;
3409 $rm->{$c} = 1;
3410 }
3411 }
3412 delete $rm->{$self->{svn_path}};
3413 delete $rm->{''}; # we never delete the url we're tracking
3414 return unless %$rm;
3415
3416 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3417 $self->{tree_b});
3418 local $/ = "\0";
3419 while (<$fh>) {
3420 chomp;
3421 my @dn = split m#/#, $_;
3422 while (pop @dn) {
3423 delete $rm->{join '/', @dn};
3424 }
3425 unless (%$rm) {
3426 close $fh;
3427 return;
3428 }
3429 }
3430 command_close_pipe($fh, $ctx);
3431
3432 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3433 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3434 $self->close_directory($bat->{$d}, $p);
3435 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3436 print "\tD+\t$d/\n" unless $::_q;
3437 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3438 delete $bat->{$d};
3439 }
3440 }
3441
3442 sub open_or_add_dir {
3443 my ($self, $full_path, $baton) = @_;
3444 my $t = $self->{types}->{$full_path};
3445 if (!defined $t) {
3446 die "$full_path not known in r$self->{r} or we have a bug!\n";
3447 }
3448 {
3449 no warnings 'once';
3450 # SVN::Node::none and SVN::Node::file are used only once,
3451 # so we're shutting up Perl's warnings about them.
3452 if ($t == $SVN::Node::none) {
3453 return $self->add_directory($full_path, $baton,
3454 undef, -1, $self->{pool});
3455 } elsif ($t == $SVN::Node::dir) {
3456 return $self->open_directory($full_path, $baton,
3457 $self->{r}, $self->{pool});
3458 } # no warnings 'once'
3459 print STDERR "$full_path already exists in repository at ",
3460 "r$self->{r} and it is not a directory (",
3461 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3462 } # no warnings 'once'
3463 exit 1;
3464 }
3465
3466 sub ensure_path {
3467 my ($self, $path) = @_;
3468 my $bat = $self->{bat};
3469 my $repo_path = $self->repo_path($path);
3470 return $bat->{''} unless (length $repo_path);
3471 my @p = split m#/+#, $repo_path;
3472 my $c = shift @p;
3473 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3474 while (@p) {
3475 my $c0 = $c;
3476 $c .= '/' . shift @p;
3477 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3478 }
3479 return $bat->{$c};
3480 }
3481
3482 sub A {
3483 my ($self, $m) = @_;
3484 my ($dir, $file) = split_path($m->{file_b});
3485 my $pbat = $self->ensure_path($dir);
3486 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3487 undef, -1);
3488 print "\tA\t$m->{file_b}\n" unless $::_q;
3489 $self->chg_file($fbat, $m);
3490 $self->close_file($fbat,undef,$self->{pool});
3491 }
3492
3493 sub C {
3494 my ($self, $m) = @_;
3495 my ($dir, $file) = split_path($m->{file_b});
3496 my $pbat = $self->ensure_path($dir);
3497 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3498 $self->url_path($m->{file_a}), $self->{r});
3499 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3500 $self->chg_file($fbat, $m);
3501 $self->close_file($fbat,undef,$self->{pool});
3502 }
3503
3504 sub delete_entry {
3505 my ($self, $path, $pbat) = @_;
3506 my $rpath = $self->repo_path($path);
3507 my ($dir, $file) = split_path($rpath);
3508 $self->{rm}->{$dir} = 1;
3509 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3510 }
3511
3512 sub R {
3513 my ($self, $m) = @_;
3514 my ($dir, $file) = split_path($m->{file_b});
3515 my $pbat = $self->ensure_path($dir);
3516 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3517 $self->url_path($m->{file_a}), $self->{r});
3518 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3519 $self->chg_file($fbat, $m);
3520 $self->close_file($fbat,undef,$self->{pool});
3521
3522 ($dir, $file) = split_path($m->{file_a});
3523 $pbat = $self->ensure_path($dir);
3524 $self->delete_entry($m->{file_a}, $pbat);
3525 }
3526
3527 sub M {
3528 my ($self, $m) = @_;
3529 my ($dir, $file) = split_path($m->{file_b});
3530 my $pbat = $self->ensure_path($dir);
3531 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3532 $pbat,$self->{r},$self->{pool});
3533 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3534 $self->chg_file($fbat, $m);
3535 $self->close_file($fbat,undef,$self->{pool});
3536 }
3537
3538 sub T { shift->M(@_) }
3539
3540 sub change_file_prop {
3541 my ($self, $fbat, $pname, $pval) = @_;
3542 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3543 }
3544
3545 sub chg_file {
3546 my ($self, $fbat, $m) = @_;
3547 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3548 $self->change_file_prop($fbat,'svn:executable','*');
3549 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3550 $self->change_file_prop($fbat,'svn:executable',undef);
3551 }
3552 my $fh = IO::File->new_tmpfile or croak $!;
3553 if ($m->{mode_b} =~ /^120/) {
3554 print $fh 'link ' or croak $!;
3555 $self->change_file_prop($fbat,'svn:special','*');
3556 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3557 $self->change_file_prop($fbat,'svn:special',undef);
3558 }
3559 my $size = $::_repository->cat_blob($m->{sha1_b}, $fh);
3560 croak "Failed to read object $m->{sha1_b}" unless $size;
3561 $fh->flush == 0 or croak $!;
3562 seek $fh, 0, 0 or croak $!;
3563
3564 my $exp = ::md5sum($fh);
3565 seek $fh, 0, 0 or croak $!;
3566
3567 my $pool = SVN::Pool->new;
3568 my $atd = $self->apply_textdelta($fbat, undef, $pool);
3569 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3570 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3571 $pool->clear;
3572
3573 close $fh or croak $!;
3574 }
3575
3576 sub D {
3577 my ($self, $m) = @_;
3578 my ($dir, $file) = split_path($m->{file_b});
3579 my $pbat = $self->ensure_path($dir);
3580 print "\tD\t$m->{file_b}\n" unless $::_q;
3581 $self->delete_entry($m->{file_b}, $pbat);
3582 }
3583
3584 sub close_edit {
3585 my ($self) = @_;
3586 my ($p,$bat) = ($self->{pool}, $self->{bat});
3587 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3588 next if $_ eq '';
3589 $self->close_directory($bat->{$_}, $p);
3590 }
3591 $self->close_directory($bat->{''}, $p);
3592 $self->SUPER::close_edit($p);
3593 $p->clear;
3594 }
3595
3596 sub abort_edit {
3597 my ($self) = @_;
3598 $self->SUPER::abort_edit($self->{pool});
3599 }
3600
3601 sub DESTROY {
3602 my $self = shift;
3603 $self->SUPER::DESTROY(@_);
3604 $self->{pool}->clear;
3605 }
3606
3607 # this drives the editor
3608 sub apply_diff {
3609 my ($self) = @_;
3610 my $mods = $self->{mods};
3611 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3612 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3613 my $f = $m->{chg};
3614 if (defined $o{$f}) {
3615 $self->$f($m);
3616 } else {
3617 fatal("Invalid change type: $f");
3618 }
3619 }
3620 $self->rmdirs if $_rmdir;
3621 if (@$mods == 0) {
3622 $self->abort_edit;
3623 } else {
3624 $self->close_edit;
3625 }
3626 return scalar @$mods;
3627 }
3628
3629 package Git::SVN::Ra;
3630 use vars qw/@ISA $config_dir $_log_window_size/;
3631 use strict;
3632 use warnings;
3633 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3634
3635 BEGIN {
3636 # enforce temporary pool usage for some simple functions
3637 no strict 'refs';
3638 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3639 my $SUPER = "SUPER::$f";
3640 *$f = sub {
3641 my $self = shift;
3642 my $pool = SVN::Pool->new;
3643 my @ret = $self->$SUPER(@_,$pool);
3644 $pool->clear;
3645 wantarray ? @ret : $ret[0];
3646 };
3647 }
3648 }
3649
3650 sub _auth_providers () {
3651 [
3652 SVN::Client::get_simple_provider(),
3653 SVN::Client::get_ssl_server_trust_file_provider(),
3654 SVN::Client::get_simple_prompt_provider(
3655 \&Git::SVN::Prompt::simple, 2),
3656 SVN::Client::get_ssl_client_cert_file_provider(),
3657 SVN::Client::get_ssl_client_cert_prompt_provider(
3658 \&Git::SVN::Prompt::ssl_client_cert, 2),
3659 SVN::Client::get_ssl_client_cert_pw_file_provider(),
3660 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3661 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3662 SVN::Client::get_username_provider(),
3663 SVN::Client::get_ssl_server_trust_prompt_provider(
3664 \&Git::SVN::Prompt::ssl_server_trust),
3665 SVN::Client::get_username_prompt_provider(
3666 \&Git::SVN::Prompt::username, 2)
3667 ]
3668 }
3669
3670 sub escape_uri_only {
3671 my ($uri) = @_;
3672 my @tmp;
3673 foreach (split m{/}, $uri) {
3674 s/([^\w.%-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
3675 push @tmp, $_;
3676 }
3677 join('/', @tmp);
3678 }
3679
3680 sub escape_url {
3681 my ($url) = @_;
3682 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3683 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3684 $url = "$scheme://$domain$uri";
3685 }
3686 $url;
3687 }
3688
3689 sub new {
3690 my ($class, $url) = @_;
3691 $url =~ s!/+$!!;
3692 return $RA if ($RA && $RA->{url} eq $url);
3693
3694 SVN::_Core::svn_config_ensure($config_dir, undef);
3695 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3696 my $config = SVN::Core::config_get_config($config_dir);
3697 $RA = undef;
3698 my $dont_store_passwords = 1;
3699 my $conf_t = ${$config}{'config'};
3700 {
3701 no warnings 'once';
3702 # The usage of $SVN::_Core::SVN_CONFIG_* variables
3703 # produces warnings that variables are used only once.
3704 # I had not found the better way to shut them up, so
3705 # the warnings of type 'once' are disabled in this block.
3706 if (SVN::_Core::svn_config_get_bool($conf_t,
3707 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3708 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3709 1) == 0) {
3710 SVN::_Core::svn_auth_set_parameter($baton,
3711 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3712 bless (\$dont_store_passwords, "_p_void"));
3713 }
3714 if (SVN::_Core::svn_config_get_bool($conf_t,
3715 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3716 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3717 1) == 0) {
3718 $Git::SVN::Prompt::_no_auth_cache = 1;
3719 }
3720 } # no warnings 'once'
3721 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3722 config => $config,
3723 pool => SVN::Pool->new,
3724 auth_provider_callbacks => $callbacks);
3725 $self->{url} = $url;
3726 $self->{svn_path} = $url;
3727 $self->{repos_root} = $self->get_repos_root;
3728 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3729 $self->{cache} = { check_path => { r => 0, data => {} },
3730 get_dir => { r => 0, data => {} } };
3731 $RA = bless $self, $class;
3732 }
3733
3734 sub check_path {
3735 my ($self, $path, $r) = @_;
3736 my $cache = $self->{cache}->{check_path};
3737 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3738 return $cache->{data}->{$path};
3739 }
3740 my $pool = SVN::Pool->new;
3741 my $t = $self->SUPER::check_path($path, $r, $pool);
3742 $pool->clear;
3743 if ($r != $cache->{r}) {
3744 %{$cache->{data}} = ();
3745 $cache->{r} = $r;
3746 }
3747 $cache->{data}->{$path} = $t;
3748 }
3749
3750 sub get_dir {
3751 my ($self, $dir, $r) = @_;
3752 my $cache = $self->{cache}->{get_dir};
3753 if ($r == $cache->{r}) {
3754 if (my $x = $cache->{data}->{$dir}) {
3755 return wantarray ? @$x : $x->[0];
3756 }
3757 }
3758 my $pool = SVN::Pool->new;
3759 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3760 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3761 $pool->clear;
3762 if ($r != $cache->{r}) {
3763 %{$cache->{data}} = ();
3764 $cache->{r} = $r;
3765 }
3766 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3767 wantarray ? (\%dirents, $r, $props) : \%dirents;
3768 }
3769
3770 sub DESTROY {
3771 # do not call the real DESTROY since we store ourselves in $RA
3772 }
3773
3774 sub get_log {
3775 my ($self, @args) = @_;
3776 my $pool = SVN::Pool->new;
3777 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3778 my $ret = $self->SUPER::get_log(@args, $pool);
3779 $pool->clear;
3780 $ret;
3781 }
3782
3783 sub trees_match {
3784 my ($self, $url1, $rev1, $url2, $rev2) = @_;
3785 my $ctx = SVN::Client->new(auth => _auth_providers);
3786 my $out = IO::File->new_tmpfile;
3787
3788 # older SVN (1.1.x) doesn't take $pool as the last parameter for
3789 # $ctx->diff(), so we'll create a default one
3790 my $pool = SVN::Pool->new_default_sub;
3791
3792 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3793 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3794 $out->flush;
3795 my $ret = (($out->stat)[7] == 0);
3796 close $out or croak $!;
3797
3798 $ret;
3799 }
3800
3801 sub get_commit_editor {
3802 my ($self, $log, $cb, $pool) = @_;
3803 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3804 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3805 }
3806
3807 sub gs_do_update {
3808 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3809 my $new = ($rev_a == $rev_b);
3810 my $path = $gs->{path};
3811
3812 if ($new && -e $gs->{index}) {
3813 unlink $gs->{index} or die
3814 "Couldn't unlink index: $gs->{index}: $!\n";
3815 }
3816 my $pool = SVN::Pool->new;
3817 $editor->set_path_strip($path);
3818 my (@pc) = split m#/#, $path;
3819 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3820 1, $editor, $pool);
3821 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3822
3823 # Since we can't rely on svn_ra_reparent being available, we'll
3824 # just have to do some magic with set_path to make it so
3825 # we only want a partial path.
3826 my $sp = '';
3827 my $final = join('/', @pc);
3828 while (@pc) {
3829 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3830 $sp .= '/' if length $sp;
3831 $sp .= shift @pc;
3832 }
3833 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3834
3835 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3836
3837 $reporter->finish_report($pool);
3838 $pool->clear;
3839 $editor->{git_commit_ok};
3840 }
3841
3842 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3843 # svn_ra_reparent didn't work before 1.4)
3844 sub gs_do_switch {
3845 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3846 my $path = $gs->{path};
3847 my $pool = SVN::Pool->new;
3848
3849 my $full_url = $self->{url};
3850 my $old_url = $full_url;
3851 $full_url .= '/' . escape_uri_only($path) if length $path;
3852 my ($ra, $reparented);
3853 if ($old_url ne $full_url) {
3854 if ($old_url !~ m#^svn(\+ssh)?://#) {
3855 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3856 $pool);
3857 $self->{url} = $full_url;
3858 $reparented = 1;
3859 } else {
3860 $_[0] = undef;
3861 $self = undef;
3862 $RA = undef;
3863 $ra = Git::SVN::Ra->new($full_url);
3864 $ra_invalid = 1;
3865 }
3866 }
3867 $ra ||= $self;
3868 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3869 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3870 $reporter->set_path('', $rev_a, 0, @lock, $pool);
3871 $reporter->finish_report($pool);
3872
3873 if ($reparented) {
3874 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3875 $self->{url} = $old_url;
3876 }
3877
3878 $pool->clear;
3879 $editor->{git_commit_ok};
3880 }
3881
3882 sub longest_common_path {
3883 my ($gsv, $globs) = @_;
3884 my %common;
3885 my $common_max = scalar @$gsv;
3886
3887 foreach my $gs (@$gsv) {
3888 my @tmp = split m#/#, $gs->{path};
3889 my $p = '';
3890 foreach (@tmp) {
3891 $p .= length($p) ? "/$_" : $_;
3892 $common{$p} ||= 0;
3893 $common{$p}++;
3894 }
3895 }
3896 $globs ||= [];
3897 $common_max += scalar @$globs;
3898 foreach my $glob (@$globs) {
3899 my @tmp = split m#/#, $glob->{path}->{left};
3900 my $p = '';
3901 foreach (@tmp) {
3902 $p .= length($p) ? "/$_" : $_;
3903 $common{$p} ||= 0;
3904 $common{$p}++;
3905 }
3906 }
3907
3908 my $longest_path = '';
3909 foreach (sort {length $b <=> length $a} keys %common) {
3910 if ($common{$_} == $common_max) {
3911 $longest_path = $_;
3912 last;
3913 }
3914 }
3915 $longest_path;
3916 }
3917
3918 sub gs_fetch_loop_common {
3919 my ($self, $base, $head, $gsv, $globs) = @_;
3920 return if ($base > $head);
3921 my $inc = $_log_window_size;
3922 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3923 my $longest_path = longest_common_path($gsv, $globs);
3924 my $ra_url = $self->{url};
3925 while (1) {
3926 my %revs;
3927 my $err;
3928 my $err_handler = $SVN::Error::handler;
3929 $SVN::Error::handler = sub {
3930 ($err) = @_;
3931 skip_unknown_revs($err);
3932 };
3933 sub _cb {
3934 my ($paths, $r, $author, $date, $log) = @_;
3935 [ dup_changed_paths($paths),
3936 { author => $author, date => $date, log => $log } ];
3937 }
3938 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3939 sub { $revs{$_[1]} = _cb(@_) });
3940 if ($err && $max >= $head) {
3941 print STDERR "Path '$longest_path' ",
3942 "was probably deleted:\n",
3943 $err->expanded_message,
3944 "\nWill attempt to follow ",
3945 "revisions r$min .. r$max ",
3946 "committed before the deletion\n";
3947 my $hi = $max;
3948 while (--$hi >= $min) {
3949 my $ok;
3950 $self->get_log([$longest_path], $min, $hi,
3951 0, 1, 1, sub {
3952 $ok ||= $_[1];
3953 $revs{$_[1]} = _cb(@_) });
3954 if ($ok) {
3955 print STDERR "r$min .. r$ok OK\n";
3956 last;
3957 }
3958 }
3959 }
3960 $SVN::Error::handler = $err_handler;
3961
3962 my %exists = map { $_->{path} => $_ } @$gsv;
3963 foreach my $r (sort {$a <=> $b} keys %revs) {
3964 my ($paths, $logged) = @{$revs{$r}};
3965
3966 foreach my $gs ($self->match_globs(\%exists, $paths,
3967 $globs, $r)) {
3968 if ($gs->rev_map_max >= $r) {
3969 next;
3970 }
3971 next unless $gs->match_paths($paths, $r);
3972 $gs->{logged_rev_props} = $logged;
3973 if (my $last_commit = $gs->last_commit) {
3974 $gs->assert_index_clean($last_commit);
3975 }
3976 my $log_entry = $gs->do_fetch($paths, $r);
3977 if ($log_entry) {
3978 $gs->do_git_commit($log_entry);
3979 }
3980 $INDEX_FILES{$gs->{index}} = 1;
3981 }
3982 foreach my $g (@$globs) {
3983 my $k = "svn-remote.$g->{remote}." .
3984 "$g->{t}-maxRev";
3985 Git::SVN::tmp_config($k, $r);
3986 }
3987 if ($ra_invalid) {
3988 $_[0] = undef;
3989 $self = undef;
3990 $RA = undef;
3991 $self = Git::SVN::Ra->new($ra_url);
3992 $ra_invalid = undef;
3993 }
3994 }
3995 # pre-fill the .rev_db since it'll eventually get filled in
3996 # with '0' x40 if something new gets committed
3997 foreach my $gs (@$gsv) {
3998 next if $gs->rev_map_max >= $max;
3999 next if defined $gs->rev_map_get($max);
4000 $gs->rev_map_set($max, 0 x40);
4001 }
4002 foreach my $g (@$globs) {
4003 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4004 Git::SVN::tmp_config($k, $max);
4005 }
4006 last if $max >= $head;
4007 $min = $max + 1;
4008 $max += $inc;
4009 $max = $head if ($max > $head);
4010 }
4011 Git::SVN::gc();
4012 }
4013
4014 sub match_globs {
4015 my ($self, $exists, $paths, $globs, $r) = @_;
4016
4017 sub get_dir_check {
4018 my ($self, $exists, $g, $r) = @_;
4019 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
4020 return unless scalar @x == 3;
4021 my $dirents = $x[0];
4022 foreach my $de (keys %$dirents) {
4023 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4024 my $p = $g->{path}->full_path($de);
4025 next if $exists->{$p};
4026 next if (length $g->{path}->{right} &&
4027 ($self->check_path($p, $r) !=
4028 $SVN::Node::dir));
4029 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4030 $g->{ref}->full_path($de), 1);
4031 }
4032 }
4033 foreach my $g (@$globs) {
4034 if (my $path = $paths->{"/$g->{path}->{left}"}) {
4035 if ($path->{action} =~ /^[AR]$/) {
4036 get_dir_check($self, $exists, $g, $r);
4037 }
4038 }
4039 foreach (keys %$paths) {
4040 if (/$g->{path}->{left_regex}/ &&
4041 !/$g->{path}->{regex}/) {
4042 next if $paths->{$_}->{action} !~ /^[AR]$/;
4043 get_dir_check($self, $exists, $g, $r);
4044 }
4045 next unless /$g->{path}->{regex}/;
4046 my $p = $1;
4047 my $pathname = $g->{path}->full_path($p);
4048 next if $exists->{$pathname};
4049 next if ($self->check_path($pathname, $r) !=
4050 $SVN::Node::dir);
4051 $exists->{$pathname} = Git::SVN->init(
4052 $self->{url}, $pathname, undef,
4053 $g->{ref}->full_path($p), 1);
4054 }
4055 my $c = '';
4056 foreach (split m#/#, $g->{path}->{left}) {
4057 $c .= "/$_";
4058 next unless ($paths->{$c} &&
4059 ($paths->{$c}->{action} =~ /^[AR]$/));
4060 get_dir_check($self, $exists, $g, $r);
4061 }
4062 }
4063 values %$exists;
4064 }
4065
4066 sub minimize_url {
4067 my ($self) = @_;
4068 return $self->{url} if ($self->{url} eq $self->{repos_root});
4069 my $url = $self->{repos_root};
4070 my @components = split(m!/!, $self->{svn_path});
4071 my $c = '';
4072 do {
4073 $url .= "/$c" if length $c;
4074 eval { (ref $self)->new($url)->get_latest_revnum };
4075 } while ($@ && ($c = shift @components));
4076 $url;
4077 }
4078
4079 sub can_do_switch {
4080 my $self = shift;
4081 unless (defined $can_do_switch) {
4082 my $pool = SVN::Pool->new;
4083 my $rep = eval {
4084 $self->do_switch(1, '', 0, $self->{url},
4085 SVN::Delta::Editor->new, $pool);
4086 };
4087 if ($@) {
4088 $can_do_switch = 0;
4089 } else {
4090 $rep->abort_report($pool);
4091 $can_do_switch = 1;
4092 }
4093 $pool->clear;
4094 }
4095 $can_do_switch;
4096 }
4097
4098 sub skip_unknown_revs {
4099 my ($err) = @_;
4100 my $errno = $err->apr_err();
4101 # Maybe the branch we're tracking didn't
4102 # exist when the repo started, so it's
4103 # not an error if it doesn't, just continue
4104 #
4105 # Wonderfully consistent library, eh?
4106 # 160013 - svn:// and file://
4107 # 175002 - http(s)://
4108 # 175007 - http(s):// (this repo required authorization, too...)
4109 # More codes may be discovered later...
4110 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4111 my $err_key = $err->expanded_message;
4112 # revision numbers change every time, filter them out
4113 $err_key =~ s/\d+/\0/g;
4114 $err_key = "$errno\0$err_key";
4115 unless ($ignored_err{$err_key}) {
4116 warn "W: Ignoring error from SVN, path probably ",
4117 "does not exist: ($errno): ",
4118 $err->expanded_message,"\n";
4119 warn "W: Do not be alarmed at the above message ",
4120 "git-svn is just searching aggressively for ",
4121 "old history.\n",
4122 "This may take a while on large repositories\n";
4123 $ignored_err{$err_key} = 1;
4124 }
4125 return;
4126 }
4127 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4128 }
4129
4130 # svn_log_changed_path_t objects passed to get_log are likely to be
4131 # overwritten even if only the refs are copied to an external variable,
4132 # so we should dup the structures in their entirety. Using an externally
4133 # passed pool (instead of our temporary and quickly cleared pool in
4134 # Git::SVN::Ra) does not help matters at all...
4135 sub dup_changed_paths {
4136 my ($paths) = @_;
4137 return undef unless $paths;
4138 my %ret;
4139 foreach my $p (keys %$paths) {
4140 my $i = $paths->{$p};
4141 my %s = map { $_ => $i->$_ }
4142 qw/copyfrom_path copyfrom_rev action/;
4143 $ret{$p} = \%s;
4144 }
4145 \%ret;
4146 }
4147
4148 package Git::SVN::Log;
4149 use strict;
4150 use warnings;
4151 use POSIX qw/strftime/;
4152 use constant commit_log_separator => ('-' x 72) . "\n";
4153 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4154 %rusers $show_commit $incremental/;
4155 my $l_fmt;
4156
4157 sub cmt_showable {
4158 my ($c) = @_;
4159 return 1 if defined $c->{r};
4160
4161 # big commit message got truncated by the 16k pretty buffer in rev-list
4162 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4163 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4164 @{$c->{l}} = ();
4165 my @log = command(qw/cat-file commit/, $c->{c});
4166
4167 # shift off the headers
4168 shift @log while ($log[0] ne '');
4169 shift @log;
4170
4171 # TODO: make $c->{l} not have a trailing newline in the future
4172 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4173
4174 (undef, $c->{r}, undef) = ::extract_metadata(
4175 (grep(/^git-svn-id: /, @log))[-1]);
4176 }
4177 return defined $c->{r};
4178 }
4179
4180 sub log_use_color {
4181 return $color || Git->repository->get_colorbool('color.diff');
4182 }
4183
4184 sub git_svn_log_cmd {
4185 my ($r_min, $r_max, @args) = @_;
4186 my $head = 'HEAD';
4187 my (@files, @log_opts);
4188 foreach my $x (@args) {
4189 if ($x eq '--' || @files) {
4190 push @files, $x;
4191 } else {
4192 if (::verify_ref("$x^0")) {
4193 $head = $x;
4194 } else {
4195 push @log_opts, $x;
4196 }
4197 }
4198 }
4199
4200 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4201 $gs ||= Git::SVN->_new;
4202 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4203 $gs->refname);
4204 push @cmd, '-r' unless $non_recursive;
4205 push @cmd, qw/--raw --name-status/ if $verbose;
4206 push @cmd, '--color' if log_use_color();
4207 push @cmd, @log_opts;
4208 if (defined $r_max && $r_max == $r_min) {
4209 push @cmd, '--max-count=1';
4210 if (my $c = $gs->rev_map_get($r_max)) {
4211 push @cmd, $c;
4212 }
4213 } elsif (defined $r_max) {
4214 if ($r_max < $r_min) {
4215 ($r_min, $r_max) = ($r_max, $r_min);
4216 }
4217 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4218 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4219 # If there are no commits in the range, both $c_max and $c_min
4220 # will be undefined. If there is at least 1 commit in the
4221 # range, both will be defined.
4222 return () if !defined $c_min || !defined $c_max;
4223 if ($c_min eq $c_max) {
4224 push @cmd, '--max-count=1', $c_min;
4225 } else {
4226 push @cmd, '--boundary', "$c_min..$c_max";
4227 }
4228 }
4229 return (@cmd, @files);
4230 }
4231
4232 # adapted from pager.c
4233 sub config_pager {
4234 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4235 if (!defined $pager) {
4236 $pager = 'less';
4237 } elsif (length $pager == 0 || $pager eq 'cat') {
4238 $pager = undef;
4239 }
4240 $ENV{GIT_PAGER_IN_USE} = defined($pager);
4241 }
4242
4243 sub run_pager {
4244 return unless -t *STDOUT && defined $pager;
4245 pipe my $rfd, my $wfd or return;
4246 defined(my $pid = fork) or ::fatal "Can't fork: $!";
4247 if (!$pid) {
4248 open STDOUT, '>&', $wfd or
4249 ::fatal "Can't redirect to stdout: $!";
4250 return;
4251 }
4252 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4253 $ENV{LESS} ||= 'FRSX';
4254 exec $pager or ::fatal "Can't run pager: $! ($pager)";
4255 }
4256
4257 sub format_svn_date {
4258 return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4259 }
4260
4261 sub parse_git_date {
4262 my ($t, $tz) = @_;
4263 # Date::Parse isn't in the standard Perl distro :(
4264 if ($tz =~ s/^\+//) {
4265 $t += tz_to_s_offset($tz);
4266 } elsif ($tz =~ s/^\-//) {
4267 $t -= tz_to_s_offset($tz);
4268 }
4269 return $t;
4270 }
4271
4272 sub set_local_timezone {
4273 if (defined $TZ) {
4274 $ENV{TZ} = $TZ;
4275 } else {
4276 delete $ENV{TZ};
4277 }
4278 }
4279
4280 sub tz_to_s_offset {
4281 my ($tz) = @_;
4282 $tz =~ s/(\d\d)$//;
4283 return ($1 * 60) + ($tz * 3600);
4284 }
4285
4286 sub get_author_info {
4287 my ($dest, $author, $t, $tz) = @_;
4288 $author =~ s/(?:^\s*|\s*$)//g;
4289 $dest->{a_raw} = $author;
4290 my $au;
4291 if ($::_authors) {
4292 $au = $rusers{$author} || undef;
4293 }
4294 if (!$au) {
4295 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4296 }
4297 $dest->{t} = $t;
4298 $dest->{tz} = $tz;
4299 $dest->{a} = $au;
4300 $dest->{t_utc} = parse_git_date($t, $tz);
4301 }
4302
4303 sub process_commit {
4304 my ($c, $r_min, $r_max, $defer) = @_;
4305 if (defined $r_min && defined $r_max) {
4306 if ($r_min == $c->{r} && $r_min == $r_max) {
4307 show_commit($c);
4308 return 0;
4309 }
4310 return 1 if $r_min == $r_max;
4311 if ($r_min < $r_max) {
4312 # we need to reverse the print order
4313 return 0 if (defined $limit && --$limit < 0);
4314 push @$defer, $c;
4315 return 1;
4316 }
4317 if ($r_min != $r_max) {
4318 return 1 if ($r_min < $c->{r});
4319 return 1 if ($r_max > $c->{r});
4320 }
4321 }
4322 return 0 if (defined $limit && --$limit < 0);
4323 show_commit($c);
4324 return 1;
4325 }
4326
4327 sub show_commit {
4328 my $c = shift;
4329 if ($oneline) {
4330 my $x = "\n";
4331 if (my $l = $c->{l}) {
4332 while ($l->[0] =~ /^\s*$/) { shift @$l }
4333 $x = $l->[0];
4334 }
4335 $l_fmt ||= 'A' . length($c->{r});
4336 print 'r',pack($l_fmt, $c->{r}),' | ';
4337 print "$c->{c} | " if $show_commit;
4338 print $x;
4339 } else {
4340 show_commit_normal($c);
4341 }
4342 }
4343
4344 sub show_commit_changed_paths {
4345 my ($c) = @_;
4346 return unless $c->{changed};
4347 print "Changed paths:\n", @{$c->{changed}};
4348 }
4349
4350 sub show_commit_normal {
4351 my ($c) = @_;
4352 print commit_log_separator, "r$c->{r} | ";
4353 print "$c->{c} | " if $show_commit;
4354 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4355 my $nr_line = 0;
4356
4357 if (my $l = $c->{l}) {
4358 while ($l->[$#$l] eq "\n" && $#$l > 0
4359 && $l->[($#$l - 1)] eq "\n") {
4360 pop @$l;
4361 }
4362 $nr_line = scalar @$l;
4363 if (!$nr_line) {
4364 print "1 line\n\n\n";
4365 } else {
4366 if ($nr_line == 1) {
4367 $nr_line = '1 line';
4368 } else {
4369 $nr_line .= ' lines';
4370 }
4371 print $nr_line, "\n";
4372 show_commit_changed_paths($c);
4373 print "\n";
4374 print $_ foreach @$l;
4375 }
4376 } else {
4377 print "1 line\n";
4378 show_commit_changed_paths($c);
4379 print "\n";
4380
4381 }
4382 foreach my $x (qw/raw stat diff/) {
4383 if ($c->{$x}) {
4384 print "\n";
4385 print $_ foreach @{$c->{$x}}
4386 }
4387 }
4388 }
4389
4390 sub cmd_show_log {
4391 my (@args) = @_;
4392 my ($r_min, $r_max);
4393 my $r_last = -1; # prevent dupes
4394 set_local_timezone();
4395 if (defined $::_revision) {
4396 if ($::_revision =~ /^(\d+):(\d+)$/) {
4397 ($r_min, $r_max) = ($1, $2);
4398 } elsif ($::_revision =~ /^\d+$/) {
4399 $r_min = $r_max = $::_revision;
4400 } else {
4401 ::fatal "-r$::_revision is not supported, use ",
4402 "standard 'git log' arguments instead";
4403 }
4404 }
4405
4406 config_pager();
4407 @args = git_svn_log_cmd($r_min, $r_max, @args);
4408 if (!@args) {
4409 print commit_log_separator unless $incremental || $oneline;
4410 return;
4411 }
4412 my $log = command_output_pipe(@args);
4413 run_pager();
4414 my (@k, $c, $d, $stat);
4415 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4416 while (<$log>) {
4417 if (/^${esc_color}commit -?($::sha1_short)/o) {
4418 my $cmt = $1;
4419 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4420 $r_last = $c->{r};
4421 process_commit($c, $r_min, $r_max, \@k) or
4422 goto out;
4423 }
4424 $d = undef;
4425 $c = { c => $cmt };
4426 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4427 get_author_info($c, $1, $2, $3);
4428 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4429 # ignore
4430 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4431 push @{$c->{raw}}, $_;
4432 } elsif (/^${esc_color}[ACRMDT]\t/) {
4433 # we could add $SVN->{svn_path} here, but that requires
4434 # remote access at the moment (repo_path_split)...
4435 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
4436 push @{$c->{changed}}, $_;
4437 } elsif (/^${esc_color}diff /o) {
4438 $d = 1;
4439 push @{$c->{diff}}, $_;
4440 } elsif ($d) {
4441 push @{$c->{diff}}, $_;
4442 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4443 $esc_color*[\+\-]*$esc_color$/x) {
4444 $stat = 1;
4445 push @{$c->{stat}}, $_;
4446 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4447 push @{$c->{stat}}, $_;
4448 $stat = undef;
4449 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
4450 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4451 } elsif (s/^${esc_color} //o) {
4452 push @{$c->{l}}, $_;
4453 }
4454 }
4455 if ($c && defined $c->{r} && $c->{r} != $r_last) {
4456 $r_last = $c->{r};
4457 process_commit($c, $r_min, $r_max, \@k);
4458 }
4459 if (@k) {
4460 ($r_min, $r_max) = ($r_max, $r_min);
4461 process_commit($_, $r_min, $r_max) foreach reverse @k;
4462 }
4463 out:
4464 close $log;
4465 print commit_log_separator unless $incremental || $oneline;
4466 }
4467
4468 sub cmd_blame {
4469 my $path = shift;
4470
4471 config_pager();
4472 run_pager();
4473
4474 my ($fh, $ctx) = command_output_pipe('blame', @_, $path);
4475 while (my $line = <$fh>) {
4476 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
4477 my (undef, $rev, undef) = ::cmt_metadata($1);
4478 $rev = sprintf('%-10s', $rev);
4479 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
4480 }
4481 print $line;
4482 }
4483 command_close_pipe($fh, $ctx);
4484 }
4485
4486 package Git::SVN::Migration;
4487 # these version numbers do NOT correspond to actual version numbers
4488 # of git nor git-svn. They are just relative.
4489 #
4490 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4491 #
4492 # v1 layout: .git/$id/info/url, refs/remotes/$id
4493 #
4494 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4495 #
4496 # v3 layout: .git/svn/$id, refs/remotes/$id
4497 # - info/url may remain for backwards compatibility
4498 # - this is what we migrate up to this layout automatically,
4499 # - this will be used by git svn init on single branches
4500 # v3.1 layout (auto migrated):
4501 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4502 # for backwards compatibility
4503 #
4504 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4505 # - this is only created for newly multi-init-ed
4506 # repositories. Similar in spirit to the
4507 # --use-separate-remotes option in git-clone (now default)
4508 # - we do not automatically migrate to this (following
4509 # the example set by core git)
4510 #
4511 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
4512 # - newer, more-efficient format that uses 24-bytes per record
4513 # with no filler space.
4514 # - use xxd -c24 < .rev_map.$UUID to view and debug
4515 # - This is a one-way migration, repositories updated to the
4516 # new format will not be able to use old git-svn without
4517 # rebuilding the .rev_db. Rebuilding the rev_db is not
4518 # possible if noMetadata or useSvmProps are set; but should
4519 # be no problem for users that use the (sensible) defaults.
4520 use strict;
4521 use warnings;
4522 use Carp qw/croak/;
4523 use File::Path qw/mkpath/;
4524 use File::Basename qw/dirname basename/;
4525 use vars qw/$_minimize/;
4526
4527 sub migrate_from_v0 {
4528 my $git_dir = $ENV{GIT_DIR};
4529 return undef unless -d $git_dir;
4530 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4531 my $migrated = 0;
4532 while (<$fh>) {
4533 chomp;
4534 my ($id, $orig_ref) = ($_, $_);
4535 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4536 next unless -f "$git_dir/$id/info/url";
4537 my $new_ref = "refs/remotes/$id";
4538 if (::verify_ref("$new_ref^0")) {
4539 print STDERR "W: $orig_ref is probably an old ",
4540 "branch used by an ancient version of ",
4541 "git-svn.\n",
4542 "However, $new_ref also exists.\n",
4543 "We will not be able ",
4544 "to use this branch until this ",
4545 "ambiguity is resolved.\n";
4546 next;
4547 }
4548 print STDERR "Migrating from v0 layout...\n" if !$migrated;
4549 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4550 command_noisy('update-ref', $new_ref, $orig_ref);
4551 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4552 $migrated++;
4553 }
4554 command_close_pipe($fh, $ctx);
4555 print STDERR "Done migrating from v0 layout...\n" if $migrated;
4556 $migrated;
4557 }
4558
4559 sub migrate_from_v1 {
4560 my $git_dir = $ENV{GIT_DIR};
4561 my $migrated = 0;
4562 return $migrated unless -d $git_dir;
4563 my $svn_dir = "$git_dir/svn";
4564
4565 # just in case somebody used 'svn' as their $id at some point...
4566 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4567
4568 print STDERR "Migrating from a git-svn v1 layout...\n";
4569 mkpath([$svn_dir]);
4570 print STDERR "Data from a previous version of git-svn exists, but\n\t",
4571 "$svn_dir\n\t(required for this version ",
4572 "($::VERSION) of git-svn) does not. exist\n";
4573 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4574 while (<$fh>) {
4575 my $x = $_;
4576 next unless $x =~ s#^refs/remotes/##;
4577 chomp $x;
4578 next unless -f "$git_dir/$x/info/url";
4579 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4580 next unless $u;
4581 my $dn = dirname("$git_dir/svn/$x");
4582 mkpath([$dn]) unless -d $dn;
4583 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4584 mkpath(["$git_dir/svn/svn"]);
4585 print STDERR " - $git_dir/$x/info => ",
4586 "$git_dir/svn/$x/info\n";
4587 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4588 croak "$!: $x";
4589 # don't worry too much about these, they probably
4590 # don't exist with repos this old (save for index,
4591 # and we can easily regenerate that)
4592 foreach my $f (qw/unhandled.log index .rev_db/) {
4593 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4594 }
4595 } else {
4596 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4597 rename "$git_dir/$x", "$git_dir/svn/$x" or
4598 croak "$!: $x";
4599 }
4600 $migrated++;
4601 }
4602 command_close_pipe($fh, $ctx);
4603 print STDERR "Done migrating from a git-svn v1 layout\n";
4604 $migrated;
4605 }
4606
4607 sub read_old_urls {
4608 my ($l_map, $pfx, $path) = @_;
4609 my @dir;
4610 foreach (<$path/*>) {
4611 if (-r "$_/info/url") {
4612 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4613 my $ref_id = $pfx . basename $_;
4614 my $url = ::file_to_s("$_/info/url");
4615 $l_map->{$ref_id} = $url;
4616 } elsif (-d $_) {
4617 push @dir, $_;
4618 }
4619 }
4620 foreach (@dir) {
4621 my $x = $_;
4622 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4623 read_old_urls($l_map, $x, $_);
4624 }
4625 }
4626
4627 sub migrate_from_v2 {
4628 my @cfg = command(qw/config -l/);
4629 return if grep /^svn-remote\..+\.url=/, @cfg;
4630 my %l_map;
4631 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4632 my $migrated = 0;
4633
4634 foreach my $ref_id (sort keys %l_map) {
4635 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4636 if ($@) {
4637 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4638 }
4639 $migrated++;
4640 }
4641 $migrated;
4642 }
4643
4644 sub minimize_connections {
4645 my $r = Git::SVN::read_all_remotes();
4646 my $new_urls = {};
4647 my $root_repos = {};
4648 foreach my $repo_id (keys %$r) {
4649 my $url = $r->{$repo_id}->{url} or next;
4650 my $fetch = $r->{$repo_id}->{fetch} or next;
4651 my $ra = Git::SVN::Ra->new($url);
4652
4653 # skip existing cases where we already connect to the root
4654 if (($ra->{url} eq $ra->{repos_root}) ||
4655 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
4656 $repo_id)) {
4657 $root_repos->{$ra->{url}} = $repo_id;
4658 next;
4659 }
4660
4661 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4662 my $root_path = $ra->{url};
4663 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4664 foreach my $path (keys %$fetch) {
4665 my $ref_id = $fetch->{$path};
4666 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4667
4668 # make sure we can read when connecting to
4669 # a higher level of a repository
4670 my ($last_rev, undef) = $gs->last_rev_commit;
4671 if (!defined $last_rev) {
4672 $last_rev = eval {
4673 $root_ra->get_latest_revnum;
4674 };
4675 next if $@;
4676 }
4677 my $new = $root_path;
4678 $new .= length $path ? "/$path" : '';
4679 eval {
4680 $root_ra->get_log([$new], $last_rev, $last_rev,
4681 0, 0, 1, sub { });
4682 };
4683 next if $@;
4684 $new_urls->{$ra->{repos_root}}->{$new} =
4685 { ref_id => $ref_id,
4686 old_repo_id => $repo_id,
4687 old_path => $path };
4688 }
4689 }
4690
4691 my @emptied;
4692 foreach my $url (keys %$new_urls) {
4693 # see if we can re-use an existing [svn-remote "repo_id"]
4694 # instead of creating a(n ugly) new section:
4695 my $repo_id = $root_repos->{$url} ||
4696 Git::SVN::sanitize_remote_name($url);
4697
4698 my $fetch = $new_urls->{$url};
4699 foreach my $path (keys %$fetch) {
4700 my $x = $fetch->{$path};
4701 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4702 my $pfx = "svn-remote.$x->{old_repo_id}";
4703
4704 my $old_fetch = quotemeta("$x->{old_path}:".
4705 "refs/remotes/$x->{ref_id}");
4706 command_noisy(qw/config --unset/,
4707 "$pfx.fetch", '^'. $old_fetch . '$');
4708 delete $r->{$x->{old_repo_id}}->
4709 {fetch}->{$x->{old_path}};
4710 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4711 command_noisy(qw/config --unset/,
4712 "$pfx.url");
4713 push @emptied, $x->{old_repo_id}
4714 }
4715 }
4716 }
4717 if (@emptied) {
4718 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4719 "$ENV{GIT_DIR}/config";
4720 print STDERR <<EOF;
4721 The following [svn-remote] sections in your config file ($file) are empty
4722 and can be safely removed:
4723 EOF
4724 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4725 }
4726 }
4727
4728 sub migration_check {
4729 migrate_from_v0();
4730 migrate_from_v1();
4731 migrate_from_v2();
4732 minimize_connections() if $_minimize;
4733 }
4734
4735 package Git::IndexInfo;
4736 use strict;
4737 use warnings;
4738 use Git qw/command_input_pipe command_close_pipe/;
4739
4740 sub new {
4741 my ($class) = @_;
4742 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4743 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4744 }
4745
4746 sub remove {
4747 my ($self, $path) = @_;
4748 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4749 return ++$self->{nr};
4750 }
4751 undef;
4752 }
4753
4754 sub update {
4755 my ($self, $mode, $hash, $path) = @_;
4756 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4757 return ++$self->{nr};
4758 }
4759 undef;
4760 }
4761
4762 sub DESTROY {
4763 my ($self) = @_;
4764 command_close_pipe($self->{gui}, $self->{ctx});
4765 }
4766
4767 package Git::SVN::GlobSpec;
4768 use strict;
4769 use warnings;
4770
4771 sub new {
4772 my ($class, $glob) = @_;
4773 my $re = $glob;
4774 $re =~ s!/+$!!g; # no need for trailing slashes
4775 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
4776 my ($left, $right) = ($1, $2);
4777 if ($nr > 1) {
4778 die "Only one '*' wildcard expansion ",
4779 "is supported (got $nr): '$glob'\n";
4780 } elsif ($nr == 0) {
4781 die "One '*' is needed for glob: '$glob'\n";
4782 }
4783 $re = quotemeta($left) . $re . quotemeta($right);
4784 if (length $left && !($left =~ s!/+$!!g)) {
4785 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4786 }
4787 if (length $right && !($right =~ s!^/+!!g)) {
4788 die "Missing leading '/' on right side of: '$glob' ($right)\n";
4789 }
4790 my $left_re = qr/^\/\Q$left\E(\/|$)/;
4791 bless { left => $left, right => $right, left_regex => $left_re,
4792 regex => qr/$re/, glob => $glob }, $class;
4793 }
4794
4795 sub full_path {
4796 my ($self, $path) = @_;
4797 return (length $self->{left} ? "$self->{left}/" : '') .
4798 $path . (length $self->{right} ? "/$self->{right}" : '');
4799 }
4800
4801 __END__
4802
4803 Data structures:
4804
4805
4806 $remotes = { # returned by read_all_remotes()
4807 'svn' => {
4808 # svn-remote.svn.url=https://svn.musicpd.org
4809 url => 'https://svn.musicpd.org',
4810 # svn-remote.svn.fetch=mpd/trunk:trunk
4811 fetch => {
4812 'mpd/trunk' => 'trunk',
4813 },
4814 # svn-remote.svn.tags=mpd/tags/*:tags/*
4815 tags => {
4816 path => {
4817 left => 'mpd/tags',
4818 right => '',
4819 regex => qr!mpd/tags/([^/]+)$!,
4820 glob => 'tags/*',
4821 },
4822 ref => {
4823 left => 'tags',
4824 right => '',
4825 regex => qr!tags/([^/]+)$!,
4826 glob => 'tags/*',
4827 },
4828 }
4829 }
4830 };
4831
4832 $log_entry hashref as returned by libsvn_log_entry()
4833 {
4834 log => 'whitespace-formatted log entry
4835 ', # trailing newline is preserved
4836 revision => '8', # integer
4837 date => '2004-02-24T17:01:44.108345Z', # commit date
4838 author => 'committer name'
4839 };
4840
4841
4842 # this is generated by generate_diff();
4843 @mods = array of diff-index line hashes, each element represents one line
4844 of diff-index output
4845
4846 diff-index line ($m hash)
4847 {
4848 mode_a => first column of diff-index output, no leading ':',
4849 mode_b => second column of diff-index output,
4850 sha1_b => sha1sum of the final blob,
4851 chg => change type [MCRADT],
4852 file_a => original file name of a file (iff chg is 'C' or 'R')
4853 file_b => new/current file name of a file (any chg)
4854 }
4855 ;
4856
4857 # retval of read_url_paths{,_all}();
4858 $l_map = {
4859 # repository root url
4860 'https://svn.musicpd.org' => {
4861 # repository path # GIT_SVN_ID
4862 'mpd/trunk' => 'trunk',
4863 'mpd/tags/0.11.5' => 'tags/0.11.5',
4864 },
4865 }
4866
4867 Notes:
4868 I don't trust the each() function on unless I created %hash myself
4869 because the internal iterator may not have started at base.
Gitweb caching