![[tokkee]](http://tokkee.org/images/avatar.png){kind=avatar}
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision
8 $_q $_authors %users/;
9 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10 $VERSION = '@@GIT_VERSION@@';
12 my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
13 $ENV{GIT_DIR} ||= '.git';
14 $Git::SVN::default_repo_id = 'svn';
15 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
16 $Git::SVN::Ra::_log_window_size = 100;
18 $Git::SVN::Log::TZ = $ENV{TZ};
19 $ENV{TZ} = 'UTC';
20 $| = 1; # unbuffer STDOUT
22 sub fatal (@) { print STDERR @_; exit 1 }
23 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
24 require SVN::Ra;
25 require SVN::Delta;
26 if ($SVN::Core::VERSION lt '1.1.0') {
27 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
28 }
29 push @Git::SVN::Ra::ISA, 'SVN::Ra';
30 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
31 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
32 use Carp qw/croak/;
33 use IO::File qw//;
34 use File::Basename qw/dirname basename/;
35 use File::Path qw/mkpath/;
36 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
37 use IPC::Open3;
38 use Git;
40 BEGIN {
41 # import functions from Git into our packages, en masse
42 no strict 'refs';
43 foreach (qw/command command_oneline command_noisy command_output_pipe
44 command_input_pipe command_close_pipe/) {
45 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
46 Git::SVN::Migration Git::SVN::Log Git::SVN),
47 __PACKAGE__) {
48 *{"${package}::$_"} = \&{"Git::$_"};
49 }
50 }
51 }
53 my ($SVN);
55 $sha1 = qr/[a-f\d]{40}/;
56 $sha1_short = qr/[a-f\d]{4,40}/;
57 my ($_stdin, $_help, $_edit,
58 $_message, $_file,
59 $_template, $_shared,
60 $_version, $_fetch_all, $_no_rebase,
61 $_merge, $_strategy, $_dry_run, $_local,
62 $_prefix, $_no_checkout, $_verbose);
63 $Git::SVN::_follow_parent = 1;
64 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
65 'config-dir=s' => \$Git::SVN::Ra::config_dir,
66 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
67 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
68 'authors-file|A=s' => \$_authors,
69 'repack:i' => \$Git::SVN::_repack,
70 'noMetadata' => \$Git::SVN::_no_metadata,
71 'useSvmProps' => \$Git::SVN::_use_svm_props,
72 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
73 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
74 'no-checkout' => \$_no_checkout,
75 'quiet|q' => \$_q,
76 'repack-flags|repack-args|repack-opts=s' =>
77 \$Git::SVN::_repack_flags,
78 %remote_opts );
80 my ($_trunk, $_tags, $_branches, $_stdlayout);
81 my %icv;
82 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
83 'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
84 'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
85 'stdlayout|s' => \$_stdlayout,
86 'minimize-url|m' => \$Git::SVN::_minimize_url,
87 'no-metadata' => sub { $icv{noMetadata} = 1 },
88 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
89 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
90 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
91 %remote_opts );
92 my %cmt_opts = ( 'edit|e' => \$_edit,
93 'rmdir' => \$SVN::Git::Editor::_rmdir,
94 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
95 'l=i' => \$SVN::Git::Editor::_rename_limit,
96 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
97 );
99 my %cmd = (
100 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
101 { 'revision|r=s' => \$_revision,
102 'fetch-all|all' => \$_fetch_all,
103 %fc_opts } ],
104 clone => [ \&cmd_clone, "Initialize and fetch revisions",
105 { 'revision|r=s' => \$_revision,
106 %fc_opts, %init_opts } ],
107 init => [ \&cmd_init, "Initialize a repo for tracking" .
108 " (requires URL argument)",
109 \%init_opts ],
110 'multi-init' => [ \&cmd_multi_init,
111 "Deprecated alias for ".
112 "'$0 init -T<trunk> -b<branches> -t<tags>'",
113 \%init_opts ],
114 dcommit => [ \&cmd_dcommit,
115 'Commit several diffs to merge with upstream',
116 { 'merge|m|M' => \$_merge,
117 'strategy|s=s' => \$_strategy,
118 'verbose|v' => \$_verbose,
119 'dry-run|n' => \$_dry_run,
120 'fetch-all|all' => \$_fetch_all,
121 'no-rebase' => \$_no_rebase,
122 %cmt_opts, %fc_opts } ],
123 'set-tree' => [ \&cmd_set_tree,
124 "Set an SVN repository to a git tree-ish",
125 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
126 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
127 { 'revision|r=i' => \$_revision
128 } ],
129 'multi-fetch' => [ \&cmd_multi_fetch,
130 "Deprecated alias for $0 fetch --all",
131 { 'revision|r=s' => \$_revision, %fc_opts } ],
132 'migrate' => [ sub { },
133 # no-op, we automatically run this anyways,
134 'Migrate configuration/metadata/layout from
135 previous versions of git-svn',
136 { 'minimize' => \$Git::SVN::Migration::_minimize,
137 %remote_opts } ],
138 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
139 { 'limit=i' => \$Git::SVN::Log::limit,
140 'revision|r=s' => \$_revision,
141 'verbose|v' => \$Git::SVN::Log::verbose,
142 'incremental' => \$Git::SVN::Log::incremental,
143 'oneline' => \$Git::SVN::Log::oneline,
144 'show-commit' => \$Git::SVN::Log::show_commit,
145 'non-recursive' => \$Git::SVN::Log::non_recursive,
146 'authors-file|A=s' => \$_authors,
147 'color' => \$Git::SVN::Log::color,
148 'pager=s' => \$Git::SVN::Log::pager
149 } ],
150 'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
151 {} ],
152 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
153 { 'merge|m|M' => \$_merge,
154 'verbose|v' => \$_verbose,
155 'strategy|s=s' => \$_strategy,
156 'local|l' => \$_local,
157 'fetch-all|all' => \$_fetch_all,
158 %fc_opts } ],
159 'commit-diff' => [ \&cmd_commit_diff,
160 'Commit a diff between two trees',
161 { 'message|m=s' => \$_message,
162 'file|F=s' => \$_file,
163 'revision|r=s' => \$_revision,
164 %cmt_opts } ],
165 );
167 my $cmd;
168 for (my $i = 0; $i < @ARGV; $i++) {
169 if (defined $cmd{$ARGV[$i]}) {
170 $cmd = $ARGV[$i];
171 splice @ARGV, $i, 1;
172 last;
173 }
174 };
176 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
178 read_repo_config(\%opts);
179 Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
180 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
181 'minimize-connections' => \$Git::SVN::Migration::_minimize,
182 'id|i=s' => \$Git::SVN::default_ref_id,
183 'svn-remote|remote|R=s' => sub {
184 $Git::SVN::no_reuse_existing = 1;
185 $Git::SVN::default_repo_id = $_[1] });
186 exit 1 if (!$rv && $cmd && $cmd ne 'log');
188 usage(0) if $_help;
189 version() if $_version;
190 usage(1) unless defined $cmd;
191 load_authors() if $_authors;
193 # make sure we're always running
194 unless ($cmd =~ /(?:clone|init|multi-init)$/) {
195 unless (-d $ENV{GIT_DIR}) {
196 if ($git_dir_user_set) {
197 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
198 "but it is not a directory\n";
199 }
200 my $git_dir = delete $ENV{GIT_DIR};
201 chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
202 unless (length $cdup) {
203 die "Already at toplevel, but $git_dir ",
204 "not found '$cdup'\n";
205 }
206 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
207 unless (-d $git_dir) {
208 die "$git_dir still not found after going to ",
209 "'$cdup'\n";
210 }
211 $ENV{GIT_DIR} = $git_dir;
212 }
213 }
214 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
215 Git::SVN::Migration::migration_check();
216 }
217 Git::SVN::init_vars();
218 eval {
219 Git::SVN::verify_remotes_sanity();
220 $cmd{$cmd}->[0]->(@ARGV);
221 };
222 fatal $@ if $@;
223 post_fetch_checkout();
224 exit 0;
226 ####################### primary functions ######################
227 sub usage {
228 my $exit = shift || 0;
229 my $fd = $exit ? \*STDERR : \*STDOUT;
230 print $fd <<"";
231 git-svn - bidirectional operations between a single Subversion tree and git
232 Usage: $0 <command> [options] [arguments]\n
234 print $fd "Available commands:\n" unless $cmd;
236 foreach (sort keys %cmd) {
237 next if $cmd && $cmd ne $_;
238 next if /^multi-/; # don't show deprecated commands
239 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
240 foreach (keys %{$cmd{$_}->[2]}) {
241 # mixed-case options are for .git/config only
242 next if /[A-Z]/ && /^[a-z]+$/i;
243 # prints out arguments as they should be passed:
244 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
245 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
246 "--$_" : "-$_" }
247 split /\|/,$_)," $x\n";
248 }
249 }
250 print $fd <<"";
251 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
252 arbitrary identifier if you're tracking multiple SVN branches/repositories in
253 one git repository and want to keep them separate. See git-svn(1) for more
254 information.
256 exit $exit;
257 }
259 sub version {
260 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
261 exit 0;
262 }
264 sub do_git_init_db {
265 unless (-d $ENV{GIT_DIR}) {
266 my @init_db = ('init');
267 push @init_db, "--template=$_template" if defined $_template;
268 if (defined $_shared) {
269 if ($_shared =~ /[a-z]/) {
270 push @init_db, "--shared=$_shared";
271 } else {
272 push @init_db, "--shared";
273 }
274 }
275 command_noisy(@init_db);
276 }
277 my $set;
278 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
279 foreach my $i (keys %icv) {
280 die "'$set' and '$i' cannot both be set\n" if $set;
281 next unless defined $icv{$i};
282 command_noisy('config', "$pfx.$i", $icv{$i});
283 $set = $i;
284 }
285 }
287 sub init_subdir {
288 my $repo_path = shift or return;
289 mkpath([$repo_path]) unless -d $repo_path;
290 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
291 $ENV{GIT_DIR} = '.git';
292 }
294 sub cmd_clone {
295 my ($url, $path) = @_;
296 if (!defined $path &&
297 (defined $_trunk || defined $_branches || defined $_tags ||
298 defined $_stdlayout) &&
299 $url !~ m#^[a-z\+]+://#) {
300 $path = $url;
301 }
302 $path = basename($url) if !defined $path || !length $path;
303 cmd_init($url, $path);
304 Git::SVN::fetch_all($Git::SVN::default_repo_id);
305 }
307 sub cmd_init {
308 if (defined $_stdlayout) {
309 $_trunk = 'trunk' if (!defined $_trunk);
310 $_tags = 'tags' if (!defined $_tags);
311 $_branches = 'branches' if (!defined $_branches);
312 }
313 if (defined $_trunk || defined $_branches || defined $_tags) {
314 return cmd_multi_init(@_);
315 }
316 my $url = shift or die "SVN repository location required ",
317 "as a command-line argument\n";
318 init_subdir(@_);
319 do_git_init_db();
321 Git::SVN->init($url);
322 }
324 sub cmd_fetch {
325 if (grep /^\d+=./, @_) {
326 die "'<rev>=<commit>' fetch arguments are ",
327 "no longer supported.\n";
328 }
329 my ($remote) = @_;
330 if (@_ > 1) {
331 die "Usage: $0 fetch [--all] [svn-remote]\n";
332 }
333 $remote ||= $Git::SVN::default_repo_id;
334 if ($_fetch_all) {
335 cmd_multi_fetch();
336 } else {
337 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
338 }
339 }
341 sub cmd_set_tree {
342 my (@commits) = @_;
343 if ($_stdin || !@commits) {
344 print "Reading from stdin...\n";
345 @commits = ();
346 while (<STDIN>) {
347 if (/\b($sha1_short)\b/o) {
348 unshift @commits, $1;
349 }
350 }
351 }
352 my @revs;
353 foreach my $c (@commits) {
354 my @tmp = command('rev-parse',$c);
355 if (scalar @tmp == 1) {
356 push @revs, $tmp[0];
357 } elsif (scalar @tmp > 1) {
358 push @revs, reverse(command('rev-list',@tmp));
359 } else {
360 fatal "Failed to rev-parse $c\n";
361 }
362 }
363 my $gs = Git::SVN->new;
364 my ($r_last, $cmt_last) = $gs->last_rev_commit;
365 $gs->fetch;
366 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
367 fatal "There are new revisions that were fetched ",
368 "and need to be merged (or acknowledged) ",
369 "before committing.\nlast rev: $r_last\n",
370 " current: $gs->{last_rev}\n";
371 }
372 $gs->set_tree($_) foreach @revs;
373 print "Done committing ",scalar @revs," revisions to SVN\n";
374 }
376 sub cmd_dcommit {
377 my $head = shift;
378 $head ||= 'HEAD';
379 my @refs;
380 my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
381 print "Committing to $url ...\n";
382 unless ($gs) {
383 die "Unable to determine upstream SVN information from ",
384 "$head history\n";
385 }
386 my $last_rev;
387 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
388 if ($_no_rebase && scalar(@$linear_refs) > 1) {
389 warn "Attempting to commit more than one change while ",
390 "--no-rebase is enabled.\n",
391 "If these changes depend on each other, re-running ",
392 "without --no-rebase will be required."
393 }
394 foreach my $d (@$linear_refs) {
395 unless (defined $last_rev) {
396 (undef, $last_rev, undef) = cmt_metadata("$d~1");
397 unless (defined $last_rev) {
398 fatal "Unable to extract revision information ",
399 "from commit $d~1\n";
400 }
401 }
402 if ($_dry_run) {
403 print "diff-tree $d~1 $d\n";
404 } else {
405 my $cmt_rev;
406 my %ed_opts = ( r => $last_rev,
407 log => get_commit_entry($d)->{log},
408 ra => Git::SVN::Ra->new($gs->full_url),
409 tree_a => "$d~1",
410 tree_b => $d,
411 editor_cb => sub {
412 print "Committed r$_[0]\n";
413 $cmt_rev = $_[0];
414 },
415 svn_path => '');
416 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
417 print "No changes\n$d~1 == $d\n";
418 } elsif ($parents->{$d} && @{$parents->{$d}}) {
419 $gs->{inject_parents_dcommit}->{$cmt_rev} =
420 $parents->{$d};
421 }
422 $_fetch_all ? $gs->fetch_all : $gs->fetch;
423 next if $_no_rebase;
425 # we always want to rebase against the current HEAD,
426 # not any head that was passed to us
427 my @diff = command('diff-tree', 'HEAD',
428 $gs->refname, '--');
429 my @finish;
430 if (@diff) {
431 @finish = rebase_cmd();
432 print STDERR "W: HEAD and ", $gs->refname,
433 " differ, using @finish:\n",
434 "@diff";
435 } else {
436 print "No changes between current HEAD and ",
437 $gs->refname,
438 "\nResetting to the latest ",
439 $gs->refname, "\n";
440 @finish = qw/reset --mixed/;
441 }
442 command_noisy(@finish, $gs->refname);
443 $last_rev = $cmt_rev;
444 }
445 }
446 }
448 sub cmd_find_rev {
449 my $revision_or_hash = shift;
450 my $result;
451 if ($revision_or_hash =~ /^r\d+$/) {
452 my $head = shift;
453 $head ||= 'HEAD';
454 my @refs;
455 my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
456 unless ($gs) {
457 die "Unable to determine upstream SVN information from ",
458 "$head history\n";
459 }
460 my $desired_revision = substr($revision_or_hash, 1);
461 $result = $gs->rev_db_get($desired_revision);
462 } else {
463 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
464 $result = $rev;
465 }
466 print "$result\n" if $result;
467 }
469 sub cmd_rebase {
470 command_noisy(qw/update-index --refresh/);
471 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
472 unless ($gs) {
473 die "Unable to determine upstream SVN information from ",
474 "working tree history\n";
475 }
476 if (command(qw/diff-index HEAD --/)) {
477 print STDERR "Cannot rebase with uncommited changes:\n";
478 command_noisy('status');
479 exit 1;
480 }
481 unless ($_local) {
482 $_fetch_all ? $gs->fetch_all : $gs->fetch;
483 }
484 command_noisy(rebase_cmd(), $gs->refname);
485 }
487 sub cmd_show_ignore {
488 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
489 $gs ||= Git::SVN->new;
490 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
491 $gs->traverse_ignore(\*STDOUT, $gs->{path}, $r);
492 }
494 sub cmd_multi_init {
495 my $url = shift;
496 unless (defined $_trunk || defined $_branches || defined $_tags) {
497 usage(1);
498 }
500 # there are currently some bugs that prevent multi-init/multi-fetch
501 # setups from working well without this.
502 $Git::SVN::_minimize_url = 1;
504 $_prefix = '' unless defined $_prefix;
505 if (defined $url) {
506 $url =~ s#/+$##;
507 init_subdir(@_);
508 }
509 do_git_init_db();
510 if (defined $_trunk) {
511 my $trunk_ref = $_prefix . 'trunk';
512 # try both old-style and new-style lookups:
513 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
514 unless ($gs_trunk) {
515 my ($trunk_url, $trunk_path) =
516 complete_svn_url($url, $_trunk);
517 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
518 undef, $trunk_ref);
519 }
520 }
521 return unless defined $_branches || defined $_tags;
522 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
523 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
524 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
525 }
527 sub cmd_multi_fetch {
528 my $remotes = Git::SVN::read_all_remotes();
529 foreach my $repo_id (sort keys %$remotes) {
530 if ($remotes->{$repo_id}->{url}) {
531 Git::SVN::fetch_all($repo_id, $remotes);
532 }
533 }
534 }
536 # this command is special because it requires no metadata
537 sub cmd_commit_diff {
538 my ($ta, $tb, $url) = @_;
539 my $usage = "Usage: $0 commit-diff -r<revision> ".
540 "<tree-ish> <tree-ish> [<URL>]\n";
541 fatal($usage) if (!defined $ta || !defined $tb);
542 my $svn_path;
543 if (!defined $url) {
544 my $gs = eval { Git::SVN->new };
545 if (!$gs) {
546 fatal("Needed URL or usable git-svn --id in ",
547 "the command-line\n", $usage);
548 }
549 $url = $gs->{url};
550 $svn_path = $gs->{path};
551 }
552 unless (defined $_revision) {
553 fatal("-r|--revision is a required argument\n", $usage);
554 }
555 if (defined $_message && defined $_file) {
556 fatal("Both --message/-m and --file/-F specified ",
557 "for the commit message.\n",
558 "I have no idea what you mean\n");
559 }
560 if (defined $_file) {
561 $_message = file_to_s($_file);
562 } else {
563 $_message ||= get_commit_entry($tb)->{log};
564 }
565 my $ra ||= Git::SVN::Ra->new($url);
566 $svn_path ||= $ra->{svn_path};
567 my $r = $_revision;
568 if ($r eq 'HEAD') {
569 $r = $ra->get_latest_revnum;
570 } elsif ($r !~ /^\d+$/) {
571 die "revision argument: $r not understood by git-svn\n";
572 }
573 my %ed_opts = ( r => $r,
574 log => $_message,
575 ra => $ra,
576 tree_a => $ta,
577 tree_b => $tb,
578 editor_cb => sub { print "Committed r$_[0]\n" },
579 svn_path => $svn_path );
580 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
581 print "No changes\n$ta == $tb\n";
582 }
583 }
585 ########################### utility functions #########################
587 sub rebase_cmd {
588 my @cmd = qw/rebase/;
589 push @cmd, '-v' if $_verbose;
590 push @cmd, qw/--merge/ if $_merge;
591 push @cmd, "--strategy=$_strategy" if $_strategy;
592 @cmd;
593 }
595 sub post_fetch_checkout {
596 return if $_no_checkout;
597 my $gs = $Git::SVN::_head or return;
598 return if verify_ref('refs/heads/master^0');
600 my $valid_head = verify_ref('HEAD^0');
601 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
602 return if ($valid_head || !verify_ref('HEAD^0'));
604 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
605 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
606 return if -f $index;
608 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
609 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
610 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
611 print STDERR "Checked out HEAD:\n ",
612 $gs->full_url, " r", $gs->last_rev, "\n";
613 }
615 sub complete_svn_url {
616 my ($url, $path) = @_;
617 $path =~ s#/+$##;
618 if ($path !~ m#^[a-z\+]+://#) {
619 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
620 fatal("E: '$path' is not a complete URL ",
621 "and a separate URL is not specified\n");
622 }
623 return ($url, $path);
624 }
625 return ($path, '');
626 }
628 sub complete_url_ls_init {
629 my ($ra, $repo_path, $switch, $pfx) = @_;
630 unless ($repo_path) {
631 print STDERR "W: $switch not specified\n";
632 return;
633 }
634 $repo_path =~ s#/+$##;
635 if ($repo_path =~ m#^[a-z\+]+://#) {
636 $ra = Git::SVN::Ra->new($repo_path);
637 $repo_path = '';
638 } else {
639 $repo_path =~ s#^/+##;
640 unless ($ra) {
641 fatal("E: '$repo_path' is not a complete URL ",
642 "and a separate URL is not specified\n");
643 }
644 }
645 my $url = $ra->{url};
646 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
647 my $k = "svn-remote.$gs->{repo_id}.url";
648 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
649 if ($orig_url && ($orig_url ne $gs->{url})) {
650 die "$k already set: $orig_url\n",
651 "wanted to set to: $gs->{url}\n";
652 }
653 command_oneline('config', $k, $gs->{url}) unless $orig_url;
654 my $remote_path = "$ra->{svn_path}/$repo_path/*";
655 $remote_path =~ s#/+#/#g;
656 $remote_path =~ s#^/##g;
657 my ($n) = ($switch =~ /^--(\w+)/);
658 if (length $pfx && $pfx !~ m#/$#) {
659 die "--prefix='$pfx' must have a trailing slash '/'\n";
660 }
661 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
662 "$remote_path:refs/remotes/$pfx*");
663 }
665 sub verify_ref {
666 my ($ref) = @_;
667 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
668 { STDERR => 0 }); };
669 }
671 sub get_tree_from_treeish {
672 my ($treeish) = @_;
673 # $treeish can be a symbolic ref, too:
674 my $type = command_oneline(qw/cat-file -t/, $treeish);
675 my $expected;
676 while ($type eq 'tag') {
677 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
678 }
679 if ($type eq 'commit') {
680 $expected = (grep /^tree /, command(qw/cat-file commit/,
681 $treeish))[0];
682 ($expected) = ($expected =~ /^tree ($sha1)$/o);
683 die "Unable to get tree from $treeish\n" unless $expected;
684 } elsif ($type eq 'tree') {
685 $expected = $treeish;
686 } else {
687 die "$treeish is a $type, expected tree, tag or commit\n";
688 }
689 return $expected;
690 }
692 sub get_commit_entry {
693 my ($treeish) = shift;
694 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
695 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
696 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
697 open my $log_fh, '>', $commit_editmsg or croak $!;
699 my $type = command_oneline(qw/cat-file -t/, $treeish);
700 if ($type eq 'commit' || $type eq 'tag') {
701 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
702 $type, $treeish);
703 my $in_msg = 0;
704 while (<$msg_fh>) {
705 if (!$in_msg) {
706 $in_msg = 1 if (/^\s*$/);
707 } elsif (/^git-svn-id: /) {
708 # skip this for now, we regenerate the
709 # correct one on re-fetch anyways
710 # TODO: set *:merge properties or like...
711 } else {
712 print $log_fh $_ or croak $!;
713 }
714 }
715 command_close_pipe($msg_fh, $ctx);
716 }
717 close $log_fh or croak $!;
719 if ($_edit || ($type eq 'tree')) {
720 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
721 # TODO: strip out spaces, comments, like git-commit.sh
722 system($editor, $commit_editmsg);
723 }
724 rename $commit_editmsg, $commit_msg or croak $!;
725 open $log_fh, '<', $commit_msg or croak $!;
726 { local $/; chomp($log_entry{log} = <$log_fh>); }
727 close $log_fh or croak $!;
728 unlink $commit_msg;
729 \%log_entry;
730 }
732 sub s_to_file {
733 my ($str, $file, $mode) = @_;
734 open my $fd,'>',$file or croak $!;
735 print $fd $str,"\n" or croak $!;
736 close $fd or croak $!;
737 chmod ($mode &~ umask, $file) if (defined $mode);
738 }
740 sub file_to_s {
741 my $file = shift;
742 open my $fd,'<',$file or croak "$!: file: $file\n";
743 local $/;
744 my $ret = <$fd>;
745 close $fd or croak $!;
746 $ret =~ s/\s*$//s;
747 return $ret;
748 }
750 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
751 sub load_authors {
752 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
753 my $log = $cmd eq 'log';
754 while (<$authors>) {
755 chomp;
756 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
757 my ($user, $name, $email) = ($1, $2, $3);
758 if ($log) {
759 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
760 } else {
761 $users{$user} = [$name, $email];
762 }
763 }
764 close $authors or croak $!;
765 }
767 # convert GetOpt::Long specs for use by git-config
768 sub read_repo_config {
769 return unless -d $ENV{GIT_DIR};
770 my $opts = shift;
771 my @config_only;
772 foreach my $o (keys %$opts) {
773 # if we have mixedCase and a long option-only, then
774 # it's a config-only variable that we don't need for
775 # the command-line.
776 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
777 my $v = $opts->{$o};
778 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
779 $key =~ s/-//g;
780 my $arg = 'git-config';
781 $arg .= ' --int' if ($o =~ /[:=]i$/);
782 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
783 if (ref $v eq 'ARRAY') {
784 chomp(my @tmp = `$arg --get-all svn.$key`);
785 @$v = @tmp if @tmp;
786 } else {
787 chomp(my $tmp = `$arg --get svn.$key`);
788 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
789 $$v = $tmp;
790 }
791 }
792 }
793 delete @$opts{@config_only} if @config_only;
794 }
796 sub extract_metadata {
797 my $id = shift or return (undef, undef, undef);
798 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
799 \s([a-f\d\-]+)$/x);
800 if (!defined $rev || !$uuid || !$url) {
801 # some of the original repositories I made had
802 # identifiers like this:
803 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
804 }
805 return ($url, $rev, $uuid);
806 }
808 sub cmt_metadata {
809 return extract_metadata((grep(/^git-svn-id: /,
810 command(qw/cat-file commit/, shift)))[-1]);
811 }
813 sub working_head_info {
814 my ($head, $refs) = @_;
815 my @args = ('log', '--no-color', '--first-parent');
816 my ($fh, $ctx) = command_output_pipe(@args, $head);
817 my $hash;
818 my %max;
819 while (<$fh>) {
820 if ( m{^commit ($::sha1)$} ) {
821 unshift @$refs, $hash if $hash and $refs;
822 $hash = $1;
823 next;
824 }
825 next unless s{^\s*(git-svn-id:)}{$1};
826 my ($url, $rev, $uuid) = extract_metadata($_);
827 if (defined $url && defined $rev) {
828 next if $max{$url} and $max{$url} < $rev;
829 if (my $gs = Git::SVN->find_by_url($url)) {
830 my $c = $gs->rev_db_get($rev);
831 if ($c && $c eq $hash) {
832 close $fh; # break the pipe
833 return ($url, $rev, $uuid, $gs);
834 } else {
835 $max{$url} ||= $gs->rev_db_max;
836 }
837 }
838 }
839 }
840 command_close_pipe($fh, $ctx);
841 (undef, undef, undef, undef);
842 }
844 sub read_commit_parents {
845 my ($parents, $c) = @_;
846 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
847 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
848 @{$parents->{$c}} = split(/ /, $p);
849 }
851 sub linearize_history {
852 my ($gs, $refs) = @_;
853 my %parents;
854 foreach my $c (@$refs) {
855 read_commit_parents(\%parents, $c);
856 }
858 my @linear_refs;
859 my %skip = ();
860 my $last_svn_commit = $gs->last_commit;
861 foreach my $c (reverse @$refs) {
862 next if $c eq $last_svn_commit;
863 last if $skip{$c};
865 unshift @linear_refs, $c;
866 $skip{$c} = 1;
868 # we only want the first parent to diff against for linear
869 # history, we save the rest to inject when we finalize the
870 # svn commit
871 my $fp_a = verify_ref("$c~1");
872 my $fp_b = shift @{$parents{$c}} if $parents{$c};
873 if (!$fp_a || !$fp_b) {
874 die "Commit $c\n",
875 "has no parent commit, and therefore ",
876 "nothing to diff against.\n",
877 "You should be working from a repository ",
878 "originally created by git-svn\n";
879 }
880 if ($fp_a ne $fp_b) {
881 die "$c~1 = $fp_a, however parsing commit $c ",
882 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
883 }
885 foreach my $p (@{$parents{$c}}) {
886 $skip{$p} = 1;
887 }
888 }
889 (\@linear_refs, \%parents);
890 }
892 package Git::SVN;
893 use strict;
894 use warnings;
895 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
896 $_repack $_repack_flags $_use_svm_props $_head
897 $_use_svnsync_props $no_reuse_existing $_minimize_url/;
898 use Carp qw/croak/;
899 use File::Path qw/mkpath/;
900 use File::Copy qw/copy/;
901 use IPC::Open3;
903 my $_repack_nr;
904 # properties that we do not log:
905 my %SKIP_PROP;
906 BEGIN {
907 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
908 svn:special svn:executable
909 svn:entry:committed-rev
910 svn:entry:last-author
911 svn:entry:uuid
912 svn:entry:committed-date/;
914 # some options are read globally, but can be overridden locally
915 # per [svn-remote "..."] section. Command-line options will *NOT*
916 # override options set in an [svn-remote "..."] section
917 no strict 'refs';
918 for my $option (qw/follow_parent no_metadata use_svm_props
919 use_svnsync_props/) {
920 my $key = $option;
921 $key =~ tr/_//d;
922 my $prop = "-$option";
923 *$option = sub {
924 my ($self) = @_;
925 return $self->{$prop} if exists $self->{$prop};
926 my $k = "svn-remote.$self->{repo_id}.$key";
927 eval { command_oneline(qw/config --get/, $k) };
928 if ($@) {
929 $self->{$prop} = ${"Git::SVN::_$option"};
930 } else {
931 my $v = command_oneline(qw/config --bool/,$k);
932 $self->{$prop} = $v eq 'false' ? 0 : 1;
933 }
934 return $self->{$prop};
935 }
936 }
937 }
939 my %LOCKFILES;
940 END { unlink keys %LOCKFILES if %LOCKFILES }
942 sub resolve_local_globs {
943 my ($url, $fetch, $glob_spec) = @_;
944 return unless defined $glob_spec;
945 my $ref = $glob_spec->{ref};
946 my $path = $glob_spec->{path};
947 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
948 next unless m#^refs/remotes/$ref->{regex}$#;
949 my $p = $1;
950 my $pathname = desanitize_refname($path->full_path($p));
951 my $refname = desanitize_refname($ref->full_path($p));
952 if (my $existing = $fetch->{$pathname}) {
953 if ($existing ne $refname) {
954 die "Refspec conflict:\n",
955 "existing: refs/remotes/$existing\n",
956 " globbed: refs/remotes/$refname\n";
957 }
958 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
959 $u =~ s!^\Q$url\E(/|$)!! or die
960 "refs/remotes/$refname: '$url' not found in '$u'\n";
961 if ($pathname ne $u) {
962 warn "W: Refspec glob conflict ",
963 "(ref: refs/remotes/$refname):\n",
964 "expected path: $pathname\n",
965 " real path: $u\n",
966 "Continuing ahead with $u\n";
967 next;
968 }
969 } else {
970 $fetch->{$pathname} = $refname;
971 }
972 }
973 }
975 sub parse_revision_argument {
976 my ($base, $head) = @_;
977 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
978 return ($base, $head);
979 }
980 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
981 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
982 return ($head, $head) if ($::_revision eq 'HEAD');
983 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
984 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
985 die "revision argument: $::_revision not understood by git-svn\n";
986 }
988 sub fetch_all {
989 my ($repo_id, $remotes) = @_;
990 if (ref $repo_id) {
991 my $gs = $repo_id;
992 $repo_id = undef;
993 $repo_id = $gs->{repo_id};
994 }
995 $remotes ||= read_all_remotes();
996 my $remote = $remotes->{$repo_id} or
997 die "[svn-remote \"$repo_id\"] unknown\n";
998 my $fetch = $remote->{fetch};
999 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1000 my (@gs, @globs);
1001 my $ra = Git::SVN::Ra->new($url);
1002 my $uuid = $ra->get_uuid;
1003 my $head = $ra->get_latest_revnum;
1004 my $base = defined $fetch ? $head : 0;
1006 # read the max revs for wildcard expansion (branches/*, tags/*)
1007 foreach my $t (qw/branches tags/) {
1008 defined $remote->{$t} or next;
1009 push @globs, $remote->{$t};
1010 my $max_rev = eval { tmp_config(qw/--int --get/,
1011 "svn-remote.$repo_id.${t}-maxRev") };
1012 if (defined $max_rev && ($max_rev < $base)) {
1013 $base = $max_rev;
1014 } elsif (!defined $max_rev) {
1015 $base = 0;
1016 }
1017 }
1019 if ($fetch) {
1020 foreach my $p (sort keys %$fetch) {
1021 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1022 my $lr = $gs->rev_db_max;
1023 if (defined $lr) {
1024 $base = $lr if ($lr < $base);
1025 }
1026 push @gs, $gs;
1027 }
1028 }
1030 ($base, $head) = parse_revision_argument($base, $head);
1031 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1032 }
1034 sub read_all_remotes {
1035 my $r = {};
1036 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1037 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
1038 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1039 $local_ref =~ s{^/}{};
1040 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1041 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1042 $r->{$1}->{url} = $2;
1043 } elsif (m!^(.+)\.(branches|tags)=
1044 (.*):refs/remotes/(.+)\s*$/!x) {
1045 my ($p, $g) = ($3, $4);
1046 my $rs = $r->{$1}->{$2} = {
1047 t => $2,
1048 remote => $1,
1049 path => Git::SVN::GlobSpec->new($p),
1050 ref => Git::SVN::GlobSpec->new($g) };
1051 if (length($rs->{ref}->{right}) != 0) {
1052 die "The '*' glob character must be the last ",
1053 "character of '$g'\n";
1054 }
1055 }
1056 }
1057 $r;
1058 }
1060 sub init_vars {
1061 if (defined $_repack) {
1062 $_repack = 1000 if ($_repack <= 0);
1063 $_repack_nr = $_repack;
1064 $_repack_flags ||= '-d';
1065 }
1066 }
1068 sub verify_remotes_sanity {
1069 return unless -d $ENV{GIT_DIR};
1070 my %seen;
1071 foreach (command(qw/config -l/)) {
1072 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1073 if ($seen{$1}) {
1074 die "Remote ref refs/remote/$1 is tracked by",
1075 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1076 "Please resolve this ambiguity in ",
1077 "your git configuration file before ",
1078 "continuing\n";
1079 }
1080 $seen{$1} = $_;
1081 }
1082 }
1083 }
1085 # we allow more chars than remotes2config.sh...
1086 sub sanitize_remote_name {
1087 my ($name) = @_;
1088 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1089 $name;
1090 }
1092 sub find_existing_remote {
1093 my ($url, $remotes) = @_;
1094 return undef if $no_reuse_existing;
1095 my $existing;
1096 foreach my $repo_id (keys %$remotes) {
1097 my $u = $remotes->{$repo_id}->{url} or next;
1098 next if $u ne $url;
1099 $existing = $repo_id;
1100 last;
1101 }
1102 $existing;
1103 }
1105 sub init_remote_config {
1106 my ($self, $url, $no_write) = @_;
1107 $url =~ s!/+$!!; # strip trailing slash
1108 my $r = read_all_remotes();
1109 my $existing = find_existing_remote($url, $r);
1110 if ($existing) {
1111 unless ($no_write) {
1112 print STDERR "Using existing ",
1113 "[svn-remote \"$existing\"]\n";
1114 }
1115 $self->{repo_id} = $existing;
1116 } elsif ($_minimize_url) {
1117 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1118 $existing = find_existing_remote($min_url, $r);
1119 if ($existing) {
1120 unless ($no_write) {
1121 print STDERR "Using existing ",
1122 "[svn-remote \"$existing\"]\n";
1123 }
1124 $self->{repo_id} = $existing;
1125 }
1126 if ($min_url ne $url) {
1127 unless ($no_write) {
1128 print STDERR "Using higher level of URL: ",
1129 "$url => $min_url\n";
1130 }
1131 my $old_path = $self->{path};
1132 $self->{path} = $url;
1133 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1134 if (length $old_path) {
1135 $self->{path} .= "/$old_path";
1136 }
1137 $url = $min_url;
1138 }
1139 }
1140 my $orig_url;
1141 if (!$existing) {
1142 # verify that we aren't overwriting anything:
1143 $orig_url = eval {
1144 command_oneline('config', '--get',
1145 "svn-remote.$self->{repo_id}.url")
1146 };
1147 if ($orig_url && ($orig_url ne $url)) {
1148 die "svn-remote.$self->{repo_id}.url already set: ",
1149 "$orig_url\nwanted to set to: $url\n";
1150 }
1151 }
1152 my ($xrepo_id, $xpath) = find_ref($self->refname);
1153 if (defined $xpath) {
1154 die "svn-remote.$xrepo_id.fetch already set to track ",
1155 "$xpath:refs/remotes/", $self->refname, "\n";
1156 }
1157 unless ($no_write) {
1158 command_noisy('config',
1159 "svn-remote.$self->{repo_id}.url", $url);
1160 $self->{path} =~ s{^/}{};
1161 command_noisy('config', '--add',
1162 "svn-remote.$self->{repo_id}.fetch",
1163 "$self->{path}:".$self->refname);
1164 }
1165 $self->{url} = $url;
1166 }
1168 sub find_by_url { # repos_root and, path are optional
1169 my ($class, $full_url, $repos_root, $path) = @_;
1171 return undef unless defined $full_url;
1172 remove_username($full_url);
1173 remove_username($repos_root) if defined $repos_root;
1174 my $remotes = read_all_remotes();
1175 if (defined $full_url && defined $repos_root && !defined $path) {
1176 $path = $full_url;
1177 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1178 }
1179 foreach my $repo_id (keys %$remotes) {
1180 my $u = $remotes->{$repo_id}->{url} or next;
1181 remove_username($u);
1182 next if defined $repos_root && $repos_root ne $u;
1184 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1185 foreach (qw/branches tags/) {
1186 resolve_local_globs($u, $fetch,
1187 $remotes->{$repo_id}->{$_});
1188 }
1189 my $p = $path;
1190 unless (defined $p) {
1191 $p = $full_url;
1192 $p =~ s#^\Q$u\E(?:/|$)## or next;
1193 }
1194 foreach my $f (keys %$fetch) {
1195 next if $f ne $p;
1196 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1197 }
1198 }
1199 undef;
1200 }
1202 sub init {
1203 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1204 my $self = _new($class, $repo_id, $ref_id, $path);
1205 if (defined $url) {
1206 $self->init_remote_config($url, $no_write);
1207 }
1208 $self;
1209 }
1211 sub find_ref {
1212 my ($ref_id) = @_;
1213 foreach (command(qw/config -l/)) {
1214 next unless m!^svn-remote\.(.+)\.fetch=
1215 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1216 my ($repo_id, $path, $ref) = ($1, $2, $3);
1217 if ($ref eq $ref_id) {
1218 $path = '' if ($path =~ m#^\./?#);
1219 return ($repo_id, $path);
1220 }
1221 }
1222 (undef, undef, undef);
1223 }
1225 sub new {
1226 my ($class, $ref_id, $repo_id, $path) = @_;
1227 if (defined $ref_id && !defined $repo_id && !defined $path) {
1228 ($repo_id, $path) = find_ref($ref_id);
1229 if (!defined $repo_id) {
1230 die "Could not find a \"svn-remote.*.fetch\" key ",
1231 "in the repository configuration matching: ",
1232 "refs/remotes/$ref_id\n";
1233 }
1234 }
1235 my $self = _new($class, $repo_id, $ref_id, $path);
1236 if (!defined $self->{path} || !length $self->{path}) {
1237 my $fetch = command_oneline('config', '--get',
1238 "svn-remote.$repo_id.fetch",
1239 ":refs/remotes/$ref_id\$") or
1240 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1241 "\":refs/remotes/$ref_id\$\" in config\n";
1242 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1243 }
1244 $self->{url} = command_oneline('config', '--get',
1245 "svn-remote.$repo_id.url") or
1246 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1247 $self->rebuild;
1248 $self;
1249 }
1251 sub refname {
1252 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1254 # It cannot end with a slash /, we'll throw up on this because
1255 # SVN can't have directories with a slash in their name, either:
1256 if ($refname =~ m{/$}) {
1257 die "ref: '$refname' ends with a trailing slash, this is ",
1258 "not permitted by git nor Subversion\n";
1259 }
1261 # It cannot have ASCII control character space, tilde ~, caret ^,
1262 # colon :, question-mark ?, asterisk *, space, or open bracket [
1263 # anywhere.
1264 #
1265 # Additionally, % must be escaped because it is used for escaping
1266 # and we want our escaped refname to be reversible
1267 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1269 # no slash-separated component can begin with a dot .
1270 # /.* becomes /%2E*
1271 $refname =~ s{/\.}{/%2E}g;
1273 # It cannot have two consecutive dots .. anywhere
1274 # .. becomes %2E%2E
1275 $refname =~ s{\.\.}{%2E%2E}g;
1277 return $refname;
1278 }
1280 sub desanitize_refname {
1281 my ($refname) = @_;
1282 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1283 return $refname;
1284 }
1286 sub svm_uuid {
1287 my ($self) = @_;
1288 return $self->{svm}->{uuid} if $self->svm;
1289 $self->ra;
1290 unless ($self->{svm}) {
1291 die "SVM UUID not cached, and reading remotely failed\n";
1292 }
1293 $self->{svm}->{uuid};
1294 }
1296 sub svm {
1297 my ($self) = @_;
1298 return $self->{svm} if $self->{svm};
1299 my $svm;
1300 # see if we have it in our config, first:
1301 eval {
1302 my $section = "svn-remote.$self->{repo_id}";
1303 $svm = {
1304 source => tmp_config('--get', "$section.svm-source"),
1305 uuid => tmp_config('--get', "$section.svm-uuid"),
1306 replace => tmp_config('--get', "$section.svm-replace"),
1307 }
1308 };
1309 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1310 $self->{svm} = $svm;
1311 }
1312 $self->{svm};
1313 }
1315 sub _set_svm_vars {
1316 my ($self, $ra) = @_;
1317 return $ra if $self->svm;
1319 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1320 "(svm:source, svm:uuid) ",
1321 "from the following URLs:\n" );
1322 sub read_svm_props {
1323 my ($self, $ra, $path, $r) = @_;
1324 my $props = ($ra->get_dir($path, $r))[2];
1325 my $src = $props->{'svm:source'};
1326 my $uuid = $props->{'svm:uuid'};
1327 return undef if (!$src || !$uuid);
1329 chomp($src, $uuid);
1331 $uuid =~ m{^[0-9a-f\-]{30,}$}
1332 or die "doesn't look right - svm:uuid is '$uuid'\n";
1334 # the '!' is used to mark the repos_root!/relative/path
1335 $src =~ s{/?!/?}{/};
1336 $src =~ s{/+$}{}; # no trailing slashes please
1337 # username is of no interest
1338 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1340 my $replace = $ra->{url};
1341 $replace .= "/$path" if length $path;
1343 my $section = "svn-remote.$self->{repo_id}";
1344 tmp_config("$section.svm-source", $src);
1345 tmp_config("$section.svm-replace", $replace);
1346 tmp_config("$section.svm-uuid", $uuid);
1347 $self->{svm} = {
1348 source => $src,
1349 uuid => $uuid,
1350 replace => $replace
1351 };
1352 }
1354 my $r = $ra->get_latest_revnum;
1355 my $path = $self->{path};
1356 my %tried;
1357 while (length $path) {
1358 unless ($tried{"$self->{url}/$path"}) {
1359 return $ra if $self->read_svm_props($ra, $path, $r);
1360 $tried{"$self->{url}/$path"} = 1;
1361 }
1362 $path =~ s#/?[^/]+$##;
1363 }
1364 die "Path: '$path' should be ''\n" if $path ne '';
1365 return $ra if $self->read_svm_props($ra, $path, $r);
1366 $tried{"$self->{url}/$path"} = 1;
1368 if ($ra->{repos_root} eq $self->{url}) {
1369 die @err, (map { " $_\n" } keys %tried), "\n";
1370 }
1372 # nope, make sure we're connected to the repository root:
1373 my $ok;
1374 my @tried_b;
1375 $path = $ra->{svn_path};
1376 $ra = Git::SVN::Ra->new($ra->{repos_root});
1377 while (length $path) {
1378 unless ($tried{"$ra->{url}/$path"}) {
1379 $ok = $self->read_svm_props($ra, $path, $r);
1380 last if $ok;
1381 $tried{"$ra->{url}/$path"} = 1;
1382 }
1383 $path =~ s#/?[^/]+$##;
1384 }
1385 die "Path: '$path' should be ''\n" if $path ne '';
1386 $ok ||= $self->read_svm_props($ra, $path, $r);
1387 $tried{"$ra->{url}/$path"} = 1;
1388 if (!$ok) {
1389 die @err, (map { " $_\n" } keys %tried), "\n";
1390 }
1391 Git::SVN::Ra->new($self->{url});
1392 }
1394 sub svnsync {
1395 my ($self) = @_;
1396 return $self->{svnsync} if $self->{svnsync};
1398 if ($self->no_metadata) {
1399 die "Can't have both 'noMetadata' and ",
1400 "'useSvnsyncProps' options set!\n";
1401 }
1402 if ($self->rewrite_root) {
1403 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1404 "options set!\n";
1405 }
1407 my $svnsync;
1408 # see if we have it in our config, first:
1409 eval {
1410 my $section = "svn-remote.$self->{repo_id}";
1411 $svnsync = {
1412 url => tmp_config('--get', "$section.svnsync-url"),
1413 uuid => tmp_config('--get', "$section.svnsync-uuid"),
1414 }
1415 };
1416 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1417 return $self->{svnsync} = $svnsync;
1418 }
1420 my $err = "useSvnsyncProps set, but failed to read " .
1421 "svnsync property: svn:sync-from-";
1422 my $rp = $self->ra->rev_proplist(0);
1424 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1425 $url =~ m{^[a-z\+]+://} or
1426 die "doesn't look right - svn:sync-from-url is '$url'\n";
1428 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1429 $uuid =~ m{^[0-9a-f\-]{30,}$} or
1430 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1432 my $section = "svn-remote.$self->{repo_id}";
1433 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1434 tmp_config('--add', "$section.svnsync-url", $url);
1435 return $self->{svnsync} = { url => $url, uuid => $uuid };
1436 }
1438 # this allows us to memoize our SVN::Ra UUID locally and avoid a
1439 # remote lookup (useful for 'git svn log').
1440 sub ra_uuid {
1441 my ($self) = @_;
1442 unless ($self->{ra_uuid}) {
1443 my $key = "svn-remote.$self->{repo_id}.uuid";
1444 my $uuid = eval { tmp_config('--get', $key) };
1445 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1446 $self->{ra_uuid} = $uuid;
1447 } else {
1448 die "ra_uuid called without URL\n" unless $self->{url};
1449 $self->{ra_uuid} = $self->ra->get_uuid;
1450 tmp_config('--add', $key, $self->{ra_uuid});
1451 }
1452 }
1453 $self->{ra_uuid};
1454 }
1456 sub ra {
1457 my ($self) = shift;
1458 my $ra = Git::SVN::Ra->new($self->{url});
1459 if ($self->use_svm_props && !$self->{svm}) {
1460 if ($self->no_metadata) {
1461 die "Can't have both 'noMetadata' and ",
1462 "'useSvmProps' options set!\n";
1463 } elsif ($self->use_svnsync_props) {
1464 die "Can't have both 'useSvnsyncProps' and ",
1465 "'useSvmProps' options set!\n";
1466 }
1467 $ra = $self->_set_svm_vars($ra);
1468 $self->{-want_revprops} = 1;
1469 }
1470 $ra;
1471 }
1473 sub rel_path {
1474 my ($self) = @_;
1475 my $repos_root = $self->ra->{repos_root};
1476 return $self->{path} if ($self->{url} eq $repos_root);
1477 my $url = $self->{url} .
1478 (length $self->{path} ? "/$self->{path}" : $self->{path});
1479 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1480 $url;
1481 }
1483 sub traverse_ignore {
1484 my ($self, $fh, $path, $r) = @_;
1485 $path =~ s#^/+##g;
1486 my $ra = $self->ra;
1487 my ($dirent, undef, $props) = $ra->get_dir($path, $r);
1488 my $p = $path;
1489 $p =~ s#^\Q$self->{path}\E(/|$)##;
1490 print $fh length $p ? "\n# $p\n" : "\n# /\n";
1491 if (my $s = $props->{'svn:ignore'}) {
1492 $s =~ s/[\r\n]+/\n/g;
1493 chomp $s;
1494 if (length $p == 0) {
1495 $s =~ s#\n#\n/$p#g;
1496 print $fh "/$s\n";
1497 } else {
1498 $s =~ s#\n#\n/$p/#g;
1499 print $fh "/$p/$s\n";
1500 }
1501 }
1502 foreach (sort keys %$dirent) {
1503 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1504 $self->traverse_ignore($fh, "$path/$_", $r);
1505 }
1506 }
1508 sub last_rev { ($_[0]->last_rev_commit)[0] }
1509 sub last_commit { ($_[0]->last_rev_commit)[1] }
1511 # returns the newest SVN revision number and newest commit SHA1
1512 sub last_rev_commit {
1513 my ($self) = @_;
1514 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1515 return ($self->{last_rev}, $self->{last_commit});
1516 }
1517 my $c = ::verify_ref($self->refname.'^0');
1518 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1519 my $rev = (::cmt_metadata($c))[1];
1520 if (defined $rev) {
1521 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1522 return ($rev, $c);
1523 }
1524 }
1525 my $db_path = $self->db_path;
1526 unless (-e $db_path) {
1527 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1528 return (undef, undef);
1529 }
1530 my $offset = -41; # from tail
1531 my $rl;
1532 open my $fh, '<', $db_path or croak "$db_path not readable: $!\n";
1533 sysseek($fh, $offset, 2); # don't care for errors
1534 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1535 chomp $rl;
1536 while (('0' x40) eq $rl && sysseek($fh, 0, 1) != 0) {
1537 $offset -= 41;
1538 sysseek($fh, $offset, 2); # don't care for errors
1539 sysread($fh, $rl, 41) == 41 or return (undef, undef);
1540 chomp $rl;
1541 }
1542 if ($c && $c ne $rl) {
1543 die "$db_path and ", $self->refname,
1544 " inconsistent!:\n$c != $rl\n";
1545 }
1546 my $rev = sysseek($fh, 0, 1) or croak $!;
1547 $rev = ($rev - 41) / 41;
1548 close $fh or croak $!;
1549 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1550 return ($rev, $c);
1551 }
1553 sub get_fetch_range {
1554 my ($self, $min, $max) = @_;
1555 $max ||= $self->ra->get_latest_revnum;
1556 $min ||= $self->rev_db_max;
1557 (++$min, $max);
1558 }
1560 sub tmp_config {
1561 my (@args) = @_;
1562 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1563 my $config = "$ENV{GIT_DIR}/svn/.metadata";
1564 if (! -f $config && -f $old_def_config) {
1565 rename $old_def_config, $config or
1566 die "Failed rename $old_def_config => $config: $!\n";
1567 }
1568 my $old_config = $ENV{GIT_CONFIG};
1569 $ENV{GIT_CONFIG} = $config;
1570 $@ = undef;
1571 my @ret = eval {
1572 unless (-f $config) {
1573 mkfile($config);
1574 open my $fh, '>', $config or
1575 die "Can't open $config: $!\n";
1576 print $fh "; This file is used internally by ",
1577 "git-svn\n" or die
1578 "Couldn't write to $config: $!\n";
1579 print $fh "; You should not have to edit it\n" or
1580 die "Couldn't write to $config: $!\n";
1581 close $fh or die "Couldn't close $config: $!\n";
1582 }
1583 command('config', @args);
1584 };
1585 my $err = $@;
1586 if (defined $old_config) {
1587 $ENV{GIT_CONFIG} = $old_config;
1588 } else {
1589 delete $ENV{GIT_CONFIG};
1590 }
1591 die $err if $err;
1592 wantarray ? @ret : $ret[0];
1593 }
1595 sub tmp_index_do {
1596 my ($self, $sub) = @_;
1597 my $old_index = $ENV{GIT_INDEX_FILE};
1598 $ENV{GIT_INDEX_FILE} = $self->{index};
1599 $@ = undef;
1600 my @ret = eval {
1601 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1602 mkpath([$dir]) unless -d $dir;
1603 &$sub;
1604 };
1605 my $err = $@;
1606 if (defined $old_index) {
1607 $ENV{GIT_INDEX_FILE} = $old_index;
1608 } else {
1609 delete $ENV{GIT_INDEX_FILE};
1610 }
1611 die $err if $err;
1612 wantarray ? @ret : $ret[0];
1613 }
1615 sub assert_index_clean {
1616 my ($self, $treeish) = @_;
1618 $self->tmp_index_do(sub {
1619 command_noisy('read-tree', $treeish) unless -e $self->{index};
1620 my $x = command_oneline('write-tree');
1621 my ($y) = (command(qw/cat-file commit/, $treeish) =~
1622 /^tree ($::sha1)/mo);
1623 return if $y eq $x;
1625 warn "Index mismatch: $y != $x\nrereading $treeish\n";
1626 unlink $self->{index} or die "unlink $self->{index}: $!\n";
1627 command_noisy('read-tree', $treeish);
1628 $x = command_oneline('write-tree');
1629 if ($y ne $x) {
1630 ::fatal "trees ($treeish) $y != $x\n",
1631 "Something is seriously wrong...\n";
1632 }
1633 });
1634 }
1636 sub get_commit_parents {
1637 my ($self, $log_entry) = @_;
1638 my (%seen, @ret, @tmp);
1639 # legacy support for 'set-tree'; this is only used by set_tree_cb:
1640 if (my $ip = $self->{inject_parents}) {
1641 if (my $commit = delete $ip->{$log_entry->{revision}}) {
1642 push @tmp, $commit;
1643 }
1644 }
1645 if (my $cur = ::verify_ref($self->refname.'^0')) {
1646 push @tmp, $cur;
1647 }
1648 if (my $ipd = $self->{inject_parents_dcommit}) {
1649 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
1650 push @tmp, @$commit;
1651 }
1652 }
1653 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1654 while (my $p = shift @tmp) {
1655 next if $seen{$p};
1656 $seen{$p} = 1;
1657 push @ret, $p;
1658 # MAXPARENT is defined to 16 in commit-tree.c:
1659 last if @ret >= 16;
1660 }
1661 if (@tmp) {
1662 die "r$log_entry->{revision}: No room for parents:\n\t",
1663 join("\n\t", @tmp), "\n";
1664 }
1665 @ret;
1666 }
1668 sub rewrite_root {
1669 my ($self) = @_;
1670 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
1671 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
1672 my $rwr = eval { command_oneline(qw/config --get/, $k) };
1673 if ($rwr) {
1674 $rwr =~ s#/+$##;
1675 if ($rwr !~ m#^[a-z\+]+://#) {
1676 die "$rwr is not a valid URL (key: $k)\n";
1677 }
1678 }
1679 $self->{-rewrite_root} = $rwr;
1680 }
1682 sub metadata_url {
1683 my ($self) = @_;
1684 ($self->rewrite_root || $self->{url}) .
1685 (length $self->{path} ? '/' . $self->{path} : '');
1686 }
1688 sub full_url {
1689 my ($self) = @_;
1690 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1691 }
1693 sub do_git_commit {
1694 my ($self, $log_entry) = @_;
1695 my $lr = $self->last_rev;
1696 if (defined $lr && $lr >= $log_entry->{revision}) {
1697 die "Last fetched revision of ", $self->refname,
1698 " was r$lr, but we are about to fetch: ",
1699 "r$log_entry->{revision}!\n";
1700 }
1701 if (my $c = $self->rev_db_get($log_entry->{revision})) {
1702 croak "$log_entry->{revision} = $c already exists! ",
1703 "Why are we refetching it?\n";
1704 }
1705 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $log_entry->{name};
1706 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} =
1707 $log_entry->{email};
1708 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1710 my $tree = $log_entry->{tree};
1711 if (!defined $tree) {
1712 $tree = $self->tmp_index_do(sub {
1713 command_oneline('write-tree') });
1714 }
1715 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1717 my @exec = ('git-commit-tree', $tree);
1718 foreach ($self->get_commit_parents($log_entry)) {
1719 push @exec, '-p', $_;
1720 }
1721 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1722 or croak $!;
1723 print $msg_fh $log_entry->{log} or croak $!;
1724 unless ($self->no_metadata) {
1725 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1726 or croak $!;
1727 }
1728 $msg_fh->flush == 0 or croak $!;
1729 close $msg_fh or croak $!;
1730 chomp(my $commit = do { local $/; <$out_fh> });
1731 close $out_fh or croak $!;
1732 waitpid $pid, 0;
1733 croak $? if $?;
1734 if ($commit !~ /^$::sha1$/o) {
1735 die "Failed to commit, invalid sha1: $commit\n";
1736 }
1738 $self->rev_db_set($log_entry->{revision}, $commit, 1);
1740 $self->{last_rev} = $log_entry->{revision};
1741 $self->{last_commit} = $commit;
1742 print "r$log_entry->{revision}";
1743 if (defined $log_entry->{svm_revision}) {
1744 print " (\@$log_entry->{svm_revision})";
1745 $self->rev_db_set($log_entry->{svm_revision}, $commit,
1746 0, $self->svm_uuid);
1747 }
1748 print " = $commit ($self->{ref_id})\n";
1749 if (defined $_repack && (--$_repack_nr == 0)) {
1750 $_repack_nr = $_repack;
1751 # repack doesn't use any arguments with spaces in them, does it?
1752 print "Running git repack $_repack_flags ...\n";
1753 command_noisy('repack', split(/\s+/, $_repack_flags));
1754 print "Done repacking\n";
1755 }
1756 return $commit;
1757 }
1759 sub match_paths {
1760 my ($self, $paths, $r) = @_;
1761 return 1 if $self->{path} eq '';
1762 if (my $path = $paths->{"/$self->{path}"}) {
1763 return ($path->{action} eq 'D') ? 0 : 1;
1764 }
1765 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
1766 if (grep /$self->{path_regex}/, keys %$paths) {
1767 return 1;
1768 }
1769 my $c = '';
1770 foreach (split m#/#, $self->{path}) {
1771 $c .= "/$_";
1772 next unless ($paths->{$c} &&
1773 ($paths->{$c}->{action} =~ /^[AR]$/));
1774 if ($self->ra->check_path($self->{path}, $r) ==
1775 $SVN::Node::dir) {
1776 return 1;
1777 }
1778 }
1779 return 0;
1780 }
1782 sub find_parent_branch {
1783 my ($self, $paths, $rev) = @_;
1784 return undef unless $self->follow_parent;
1785 unless (defined $paths) {
1786 my $err_handler = $SVN::Error::handler;
1787 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1788 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
1789 $paths =
1790 Git::SVN::Ra::dup_changed_paths($_[0]) });
1791 $SVN::Error::handler = $err_handler;
1792 }
1793 return undef unless defined $paths;
1795 # look for a parent from another branch:
1796 my @b_path_components = split m#/#, $self->rel_path;
1797 my @a_path_components;
1798 my $i;
1799 while (@b_path_components) {
1800 $i = $paths->{'/'.join('/', @b_path_components)};
1801 last if $i && defined $i->{copyfrom_path};
1802 unshift(@a_path_components, pop(@b_path_components));
1803 }
1804 return undef unless defined $i && defined $i->{copyfrom_path};
1805 my $branch_from = $i->{copyfrom_path};
1806 if (@a_path_components) {
1807 print STDERR "branch_from: $branch_from => ";
1808 $branch_from .= '/'.join('/', @a_path_components);
1809 print STDERR $branch_from, "\n";
1810 }
1811 my $r = $i->{copyfrom_rev};
1812 my $repos_root = $self->ra->{repos_root};
1813 my $url = $self->ra->{url};
1814 my $new_url = $repos_root . $branch_from;
1815 print STDERR "Found possible branch point: ",
1816 "$new_url => ", $self->full_url, ", $r\n";
1817 $branch_from =~ s#^/##;
1818 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
1819 unless ($gs) {
1820 my $ref_id = $self->{ref_id};
1821 $ref_id =~ s/\@\d+$//;
1822 $ref_id .= "\@$r";
1823 # just grow a tail if we're not unique enough :x
1824 $ref_id .= '-' while find_ref($ref_id);
1825 print STDERR "Initializing parent: $ref_id\n";
1826 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id, 1);
1827 }
1828 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1829 if (!defined $r0 || !defined $parent) {
1830 my ($base, $head) = parse_revision_argument(0, $r);
1831 if ($base <= $r) {
1832 $gs->fetch($base, $r);
1833 }
1834 ($r0, $parent) = $gs->last_rev_commit;
1835 }
1836 if (defined $r0 && defined $parent) {
1837 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1838 my $ed;
1839 if ($self->ra->can_do_switch) {
1840 $self->assert_index_clean($parent);
1841 print STDERR "Following parent with do_switch\n";
1842 # do_switch works with svn/trunk >= r22312, but that
1843 # is not included with SVN 1.4.3 (the latest version
1844 # at the moment), so we can't rely on it
1845 $self->{last_commit} = $parent;
1846 $ed = SVN::Git::Fetcher->new($self);
1847 $gs->ra->gs_do_switch($r0, $rev, $gs,
1848 $self->full_url, $ed)
1849 or die "SVN connection failed somewhere...\n";
1850 } else {
1851 print STDERR "Following parent with do_update\n";
1852 $ed = SVN::Git::Fetcher->new($self);
1853 $self->ra->gs_do_update($rev, $rev, $self, $ed)
1854 or die "SVN connection failed somewhere...\n";
1855 }
1856 print STDERR "Successfully followed parent\n";
1857 return $self->make_log_entry($rev, [$parent], $ed);
1858 }
1859 return undef;
1860 }
1862 sub do_fetch {
1863 my ($self, $paths, $rev) = @_;
1864 my $ed;
1865 my ($last_rev, @parents);
1866 if (my $lc = $self->last_commit) {
1867 # we can have a branch that was deleted, then re-added
1868 # under the same name but copied from another path, in
1869 # which case we'll have multiple parents (we don't
1870 # want to break the original ref, nor lose copypath info):
1871 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1872 push @{$log_entry->{parents}}, $lc;
1873 return $log_entry;
1874 }
1875 $ed = SVN::Git::Fetcher->new($self);
1876 $last_rev = $self->{last_rev};
1877 $ed->{c} = $lc;
1878 @parents = ($lc);
1879 } else {
1880 $last_rev = $rev;
1881 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1882 return $log_entry;
1883 }
1884 $ed = SVN::Git::Fetcher->new($self);
1885 }
1886 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1887 die "SVN connection failed somewhere...\n";
1888 }
1889 $self->make_log_entry($rev, \@parents, $ed);
1890 }
1892 sub get_untracked {
1893 my ($self, $ed) = @_;
1894 my @out;
1895 my $h = $ed->{empty};
1896 foreach (sort keys %$h) {
1897 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1898 push @out, " $act: " . uri_encode($_);
1899 warn "W: $act: $_\n";
1900 }
1901 foreach my $t (qw/dir_prop file_prop/) {
1902 $h = $ed->{$t} or next;
1903 foreach my $path (sort keys %$h) {
1904 my $ppath = $path eq '' ? '.' : $path;
1905 foreach my $prop (sort keys %{$h->{$path}}) {
1906 next if $SKIP_PROP{$prop};
1907 my $v = $h->{$path}->{$prop};
1908 my $t_ppath_prop = "$t: " .
1909 uri_encode($ppath) . ' ' .
1910 uri_encode($prop);
1911 if (defined $v) {
1912 push @out, " +$t_ppath_prop " .
1913 uri_encode($v);
1914 } else {
1915 push @out, " -$t_ppath_prop";
1916 }
1917 }
1918 }
1919 }
1920 foreach my $t (qw/absent_file absent_directory/) {
1921 $h = $ed->{$t} or next;
1922 foreach my $parent (sort keys %$h) {
1923 foreach my $path (sort @{$h->{$parent}}) {
1924 push @out, " $t: " .
1925 uri_encode("$parent/$path");
1926 warn "W: $t: $parent/$path ",
1927 "Insufficient permissions?\n";
1928 }
1929 }
1930 }
1931 \@out;
1932 }
1934 sub parse_svn_date {
1935 my $date = shift || return '+0000 1970-01-01 00:00:00';
1936 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1937 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1938 croak "Unable to parse date: $date\n";
1939 "+0000 $Y-$m-$d $H:$M:$S";
1940 }
1942 sub check_author {
1943 my ($author) = @_;
1944 if (!defined $author || length $author == 0) {
1945 $author = '(no author)';
1946 }
1947 if (defined $::_authors && ! defined $::users{$author}) {
1948 die "Author: $author not defined in $::_authors file\n";
1949 }
1950 $author;
1951 }
1953 sub make_log_entry {
1954 my ($self, $rev, $parents, $ed) = @_;
1955 my $untracked = $self->get_untracked($ed);
1957 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1958 print $un "r$rev\n" or croak $!;
1959 print $un $_, "\n" foreach @$untracked;
1960 my %log_entry = ( parents => $parents || [], revision => $rev,
1961 log => '');
1963 my $headrev;
1964 my $logged = delete $self->{logged_rev_props};
1965 if (!$logged || $self->{-want_revprops}) {
1966 my $rp = $self->ra->rev_proplist($rev);
1967 foreach (sort keys %$rp) {
1968 my $v = $rp->{$_};
1969 if (/^svn:(author|date|log)$/) {
1970 $log_entry{$1} = $v;
1971 } elsif ($_ eq 'svm:headrev') {
1972 $headrev = $v;
1973 } else {
1974 print $un " rev_prop: ", uri_encode($_), ' ',
1975 uri_encode($v), "\n";
1976 }
1977 }
1978 } else {
1979 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1980 }
1981 close $un or croak $!;
1983 $log_entry{date} = parse_svn_date($log_entry{date});
1984 $log_entry{log} .= "\n";
1985 my $author = $log_entry{author} = check_author($log_entry{author});
1986 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1987 : ($author, undef);
1988 if (defined $headrev && $self->use_svm_props) {
1989 if ($self->rewrite_root) {
1990 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
1991 "options set!\n";
1992 }
1993 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
1994 # we don't want "SVM: initializing mirror for junk" ...
1995 return undef if $r == 0;
1996 my $svm = $self->svm;
1997 if ($uuid ne $svm->{uuid}) {
1998 die "UUID mismatch on SVM path:\n",
1999 "expected: $svm->{uuid}\n",
2000 " got: $uuid\n";
2001 }
2002 my $full_url = $self->full_url;
2003 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2004 die "Failed to replace '$svm->{replace}' with ",
2005 "'$svm->{source}' in $full_url\n";
2006 # throw away username for storing in records
2007 remove_username($full_url);
2008 $log_entry{metadata} = "$full_url\@$r $uuid";
2009 $log_entry{svm_revision} = $r;
2010 $email ||= "$author\@$uuid"
2011 } elsif ($self->use_svnsync_props) {
2012 my $full_url = $self->svnsync->{url};
2013 $full_url .= "/$self->{path}" if length $self->{path};
2014 remove_username($full_url);
2015 my $uuid = $self->svnsync->{uuid};
2016 $log_entry{metadata} = "$full_url\@$rev $uuid";
2017 $email ||= "$author\@$uuid"
2018 } else {
2019 my $url = $self->metadata_url;
2020 remove_username($url);
2021 $log_entry{metadata} = "$url\@$rev " .
2022 $self->ra->get_uuid;
2023 $email ||= "$author\@" . $self->ra->get_uuid;
2024 }
2025 $log_entry{name} = $name;
2026 $log_entry{email} = $email;
2027 \%log_entry;
2028 }
2030 sub fetch {
2031 my ($self, $min_rev, $max_rev, @parents) = @_;
2032 my ($last_rev, $last_commit) = $self->last_rev_commit;
2033 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2034 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2035 }
2037 sub set_tree_cb {
2038 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2039 $self->{inject_parents} = { $rev => $tree };
2040 $self->fetch(undef, undef);
2041 }
2043 sub set_tree {
2044 my ($self, $tree) = (shift, shift);
2045 my $log_entry = ::get_commit_entry($tree);
2046 unless ($self->{last_rev}) {
2047 fatal("Must have an existing revision to commit\n");
2048 }
2049 my %ed_opts = ( r => $self->{last_rev},
2050 log => $log_entry->{log},
2051 ra => $self->ra,
2052 tree_a => $self->{last_commit},
2053 tree_b => $tree,
2054 editor_cb => sub {
2055 $self->set_tree_cb($log_entry, $tree, @_) },
2056 svn_path => $self->{path} );
2057 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2058 print "No changes\nr$self->{last_rev} = $tree\n";
2059 }
2060 }
2062 sub rebuild {
2063 my ($self) = @_;
2064 my $db_path = $self->db_path;
2065 return if (-e $db_path && ! -z $db_path);
2066 return unless ::verify_ref($self->refname.'^0');
2067 if (-f $self->{db_root}) {
2068 rename $self->{db_root}, $db_path or die
2069 "rename $self->{db_root} => $db_path failed: $!\n";
2070 my ($dir, $base) = ($db_path =~ m#^(.*?)/?([^/]+)$#);
2071 symlink $base, $self->{db_root} or die
2072 "symlink $base => $self->{db_root} failed: $!\n";
2073 return;
2074 }
2075 print "Rebuilding $db_path ...\n";
2076 my ($log, $ctx) = command_output_pipe("log", '--no-color', $self->refname);
2077 my $latest;
2078 my $full_url = $self->full_url;
2079 remove_username($full_url);
2080 my $svn_uuid;
2081 my $c;
2082 while (<$log>) {
2083 if ( m{^commit ($::sha1)$} ) {
2084 $c = $1;
2085 next;
2086 }
2087 next unless s{^\s*(git-svn-id:)}{$1};
2088 my ($url, $rev, $uuid) = ::extract_metadata($_);
2089 remove_username($url);
2091 # ignore merges (from set-tree)
2092 next if (!defined $rev || !$uuid);
2094 # if we merged or otherwise started elsewhere, this is
2095 # how we break out of it
2096 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
2097 ($full_url && $url && ($url ne $full_url))) {
2098 next;
2099 }
2100 $latest ||= $rev;
2101 $svn_uuid ||= $uuid;
2103 $self->rev_db_set($rev, $c);
2104 print "r$rev = $c\n";
2105 }
2106 command_close_pipe($log, $ctx);
2107 print "Done rebuilding $db_path\n";
2108 }
2110 # rev_db:
2111 # Tie::File seems to be prone to offset errors if revisions get sparse,
2112 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
2113 # one of my favorite modules is out :< Next up would be one of the DBM
2114 # modules, but I'm not sure which is most portable... So I'll just
2115 # go with something that's plain-text, but still capable of
2116 # being randomly accessed. So here's my ultra-simple fixed-width
2117 # database. All records are 40 characters + "\n", so it's easy to seek
2118 # to a revision: (41 * rev) is the byte offset.
2119 # A record of 40 0s denotes an empty revision.
2120 # And yes, it's still pretty fast (faster than Tie::File).
2121 # These files are disposable unless noMetadata or useSvmProps is set
2123 sub _rev_db_set {
2124 my ($fh, $rev, $commit) = @_;
2125 my $offset = $rev * 41;
2126 # assume that append is the common case:
2127 seek $fh, 0, 2 or croak $!;
2128 my $pos = tell $fh;
2129 if ($pos < $offset) {
2130 for (1 .. (($offset - $pos) / 41)) {
2131 print $fh (('0' x 40),"\n") or croak $!;
2132 }
2133 }
2134 seek $fh, $offset, 0 or croak $!;
2135 print $fh $commit,"\n" or croak $!;
2136 }
2138 sub mkfile {
2139 my ($path) = @_;
2140 unless (-e $path) {
2141 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2142 mkpath([$dir]) unless -d $dir;
2143 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2144 close $fh or die "Couldn't close (create) $path: $!\n";
2145 }
2146 }
2148 sub rev_db_set {
2149 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2150 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2151 my $db = $self->db_path($uuid);
2152 my $db_lock = "$db.lock";
2153 my $sig;
2154 if ($update_ref) {
2155 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2156 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2157 }
2158 mkfile($db);
2160 $LOCKFILES{$db_lock} = 1;
2161 my $sync;
2162 # both of these options make our .rev_db file very, very important
2163 # and we can't afford to lose it because rebuild() won't work
2164 if ($self->use_svm_props || $self->no_metadata) {
2165 $sync = 1;
2166 copy($db, $db_lock) or die "rev_db_set(@_): ",
2167 "Failed to copy: ",
2168 "$db => $db_lock ($!)\n";
2169 } else {
2170 rename $db, $db_lock or die "rev_db_set(@_): ",
2171 "Failed to rename: ",
2172 "$db => $db_lock ($!)\n";
2173 }
2174 open my $fh, '+<', $db_lock or die "Couldn't open $db_lock: $!\n";
2175 _rev_db_set($fh, $rev, $commit);
2176 if ($sync) {
2177 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2178 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2179 }
2180 close $fh or croak $!;
2181 if ($update_ref) {
2182 $_head = $self;
2183 command_noisy('update-ref', '-m', "r$rev",
2184 $self->refname, $commit);
2185 }
2186 rename $db_lock, $db or die "rev_db_set(@_): ", "Failed to rename: ",
2187 "$db_lock => $db ($!)\n";
2188 delete $LOCKFILES{$db_lock};
2189 if ($update_ref) {
2190 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2191 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2192 kill $sig, $$ if defined $sig;
2193 }
2194 }
2196 sub rev_db_max {
2197 my ($self) = @_;
2198 $self->rebuild;
2199 my $db_path = $self->db_path;
2200 my @stat = stat $db_path or return 0;
2201 ($stat[7] % 41) == 0 or die "$db_path inconsistent size: $stat[7]\n";
2202 my $max = $stat[7] / 41;
2203 (($max > 0) ? $max - 1 : 0);
2204 }
2206 sub rev_db_get {
2207 my ($self, $rev, $uuid) = @_;
2208 my $ret;
2209 my $offset = $rev * 41;
2210 my $db_path = $self->db_path($uuid);
2211 return undef unless -e $db_path;
2212 open my $fh, '<', $db_path or croak $!;
2213 if (sysseek($fh, $offset, 0) == $offset) {
2214 my $read = sysread($fh, $ret, 40);
2215 $ret = undef if ($read != 40 || $ret eq ('0'x40));
2216 }
2217 close $fh or croak $!;
2218 $ret;
2219 }
2221 sub find_rev_before {
2222 my ($self, $rev, $eq_ok) = @_;
2223 --$rev unless $eq_ok;
2224 while ($rev > 0) {
2225 if (my $c = $self->rev_db_get($rev)) {
2226 return ($rev, $c);
2227 }
2228 --$rev;
2229 }
2230 return (undef, undef);
2231 }
2233 sub _new {
2234 my ($class, $repo_id, $ref_id, $path) = @_;
2235 unless (defined $repo_id && length $repo_id) {
2236 $repo_id = $Git::SVN::default_repo_id;
2237 }
2238 unless (defined $ref_id && length $ref_id) {
2239 $_[2] = $ref_id = $Git::SVN::default_ref_id;
2240 }
2241 $_[1] = $repo_id = sanitize_remote_name($repo_id);
2242 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2243 $_[3] = $path = '' unless (defined $path);
2244 mkpath(["$ENV{GIT_DIR}/svn"]);
2245 bless {
2246 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2247 path => $path, config => "$ENV{GIT_DIR}/svn/config",
2248 db_root => "$dir/.rev_db", repo_id => $repo_id }, $class;
2249 }
2251 sub db_path {
2252 my ($self, $uuid) = @_;
2253 $uuid ||= $self->ra_uuid;
2254 "$self->{db_root}.$uuid";
2255 }
2257 sub uri_encode {
2258 my ($f) = @_;
2259 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2260 $f
2261 }
2263 sub remove_username {
2264 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2265 }
2267 package Git::SVN::Prompt;
2268 use strict;
2269 use warnings;
2270 require SVN::Core;
2271 use vars qw/$_no_auth_cache $_username/;
2273 sub simple {
2274 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2275 $may_save = undef if $_no_auth_cache;
2276 $default_username = $_username if defined $_username;
2277 if (defined $default_username && length $default_username) {
2278 if (defined $realm && length $realm) {
2279 print STDERR "Authentication realm: $realm\n";
2280 STDERR->flush;
2281 }
2282 $cred->username($default_username);
2283 } else {
2284 username($cred, $realm, $may_save, $pool);
2285 }
2286 $cred->password(_read_password("Password for '" .
2287 $cred->username . "': ", $realm));
2288 $cred->may_save($may_save);
2289 $SVN::_Core::SVN_NO_ERROR;
2290 }
2292 sub ssl_server_trust {
2293 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2294 $may_save = undef if $_no_auth_cache;
2295 print STDERR "Error validating server certificate for '$realm':\n";
2296 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2297 print STDERR " - The certificate is not issued by a trusted ",
2298 "authority. Use the\n",
2299 " fingerprint to validate the certificate manually!\n";
2300 }
2301 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2302 print STDERR " - The certificate hostname does not match.\n";
2303 }
2304 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2305 print STDERR " - The certificate is not yet valid.\n";
2306 }
2307 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2308 print STDERR " - The certificate has expired.\n";
2309 }
2310 if ($failures & $SVN::Auth::SSL::OTHER) {
2311 print STDERR " - The certificate has an unknown error.\n";
2312 }
2313 printf STDERR
2314 "Certificate information:\n".
2315 " - Hostname: %s\n".
2316 " - Valid: from %s until %s\n".
2317 " - Issuer: %s\n".
2318 " - Fingerprint: %s\n",
2319 map $cert_info->$_, qw(hostname valid_from valid_until
2320 issuer_dname fingerprint);
2321 my $choice;
2322 prompt:
2323 print STDERR $may_save ?
2324 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2325 "(R)eject or accept (t)emporarily? ";
2326 STDERR->flush;
2327 $choice = lc(substr(<STDIN> || 'R', 0, 1));
2328 if ($choice =~ /^t$/i) {
2329 $cred->may_save(undef);
2330 } elsif ($choice =~ /^r$/i) {
2331 return -1;
2332 } elsif ($may_save && $choice =~ /^p$/i) {
2333 $cred->may_save($may_save);
2334 } else {
2335 goto prompt;
2336 }
2337 $cred->accepted_failures($failures);
2338 $SVN::_Core::SVN_NO_ERROR;
2339 }
2341 sub ssl_client_cert {
2342 my ($cred, $realm, $may_save, $pool) = @_;
2343 $may_save = undef if $_no_auth_cache;
2344 print STDERR "Client certificate filename: ";
2345 STDERR->flush;
2346 chomp(my $filename = <STDIN>);
2347 $cred->cert_file($filename);
2348 $cred->may_save($may_save);
2349 $SVN::_Core::SVN_NO_ERROR;
2350 }
2352 sub ssl_client_cert_pw {
2353 my ($cred, $realm, $may_save, $pool) = @_;
2354 $may_save = undef if $_no_auth_cache;
2355 $cred->password(_read_password("Password: ", $realm));
2356 $cred->may_save($may_save);
2357 $SVN::_Core::SVN_NO_ERROR;
2358 }
2360 sub username {
2361 my ($cred, $realm, $may_save, $pool) = @_;
2362 $may_save = undef if $_no_auth_cache;
2363 if (defined $realm && length $realm) {
2364 print STDERR "Authentication realm: $realm\n";
2365 }
2366 my $username;
2367 if (defined $_username) {
2368 $username = $_username;
2369 } else {
2370 print STDERR "Username: ";
2371 STDERR->flush;
2372 chomp($username = <STDIN>);
2373 }
2374 $cred->username($username);
2375 $cred->may_save($may_save);
2376 $SVN::_Core::SVN_NO_ERROR;
2377 }
2379 sub _read_password {
2380 my ($prompt, $realm) = @_;
2381 print STDERR $prompt;
2382 STDERR->flush;
2383 require Term::ReadKey;
2384 Term::ReadKey::ReadMode('noecho');
2385 my $password = '';
2386 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2387 last if $key =~ /[\012\015]/; # \n\r
2388 $password .= $key;
2389 }
2390 Term::ReadKey::ReadMode('restore');
2391 print STDERR "\n";
2392 STDERR->flush;
2393 $password;
2394 }
2396 package main;
2398 {
2399 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2400 $SVN::Node::dir.$SVN::Node::unknown.
2401 $SVN::Node::none.$SVN::Node::file.
2402 $SVN::Node::dir.$SVN::Node::unknown.
2403 $SVN::Auth::SSL::CNMISMATCH.
2404 $SVN::Auth::SSL::NOTYETVALID.
2405 $SVN::Auth::SSL::EXPIRED.
2406 $SVN::Auth::SSL::UNKNOWNCA.
2407 $SVN::Auth::SSL::OTHER;
2408 }
2410 package SVN::Git::Fetcher;
2411 use vars qw/@ISA/;
2412 use strict;
2413 use warnings;
2414 use Carp qw/croak/;
2415 use IO::File qw//;
2416 use Digest::MD5;
2418 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
2419 sub new {
2420 my ($class, $git_svn) = @_;
2421 my $self = SVN::Delta::Editor->new;
2422 bless $self, $class;
2423 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
2424 $self->{empty} = {};
2425 $self->{dir_prop} = {};
2426 $self->{file_prop} = {};
2427 $self->{absent_dir} = {};
2428 $self->{absent_file} = {};
2429 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
2430 $self;
2431 }
2433 sub set_path_strip {
2434 my ($self, $path) = @_;
2435 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
2436 }
2438 sub open_root {
2439 { path => '' };
2440 }
2442 sub open_directory {
2443 my ($self, $path, $pb, $rev) = @_;
2444 { path => $path };
2445 }
2447 sub git_path {
2448 my ($self, $path) = @_;
2449 if ($self->{path_strip}) {
2450 $path =~ s!$self->{path_strip}!! or
2451 die "Failed to strip path '$path' ($self->{path_strip})\n";
2452 }
2453 $path;
2454 }
2456 sub delete_entry {
2457 my ($self, $path, $rev, $pb) = @_;
2459 my $gpath = $self->git_path($path);
2460 return undef if ($gpath eq '');
2462 # remove entire directories.
2463 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
2464 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
2465 -r --name-only -z/,
2466 $self->{c}, '--', $gpath);
2467 local $/ = "\0";
2468 while (<$ls>) {
2469 chomp;
2470 $self->{gii}->remove($_);
2471 print "\tD\t$_\n" unless $::_q;
2472 }
2473 print "\tD\t$gpath/\n" unless $::_q;
2474 command_close_pipe($ls, $ctx);
2475 $self->{empty}->{$path} = 0
2476 } else {
2477 $self->{gii}->remove($gpath);
2478 print "\tD\t$gpath\n" unless $::_q;
2479 }
2480 undef;
2481 }
2483 sub open_file {
2484 my ($self, $path, $pb, $rev) = @_;
2485 my $gpath = $self->git_path($path);
2486 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
2487 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
2488 unless (defined $mode && defined $blob) {
2489 die "$path was not found in commit $self->{c} (r$rev)\n";
2490 }
2491 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
2492 pool => SVN::Pool->new, action => 'M' };
2493 }
2495 sub add_file {
2496 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
2497 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2498 delete $self->{empty}->{$dir};
2499 { path => $path, mode_a => 100644, mode_b => 100644,
2500 pool => SVN::Pool->new, action => 'A' };
2501 }
2503 sub add_directory {
2504 my ($self, $path, $cp_path, $cp_rev) = @_;
2505 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
2506 delete $self->{empty}->{$dir};
2507 $self->{empty}->{$path} = 1;
2508 { path => $path };
2509 }
2511 sub change_dir_prop {
2512 my ($self, $db, $prop, $value) = @_;
2513 $self->{dir_prop}->{$db->{path}} ||= {};
2514 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
2515 undef;
2516 }
2518 sub absent_directory {
2519 my ($self, $path, $pb) = @_;
2520 $self->{absent_dir}->{$pb->{path}} ||= [];
2521 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
2522 undef;
2523 }
2525 sub absent_file {
2526 my ($self, $path, $pb) = @_;
2527 $self->{absent_file}->{$pb->{path}} ||= [];
2528 push @{$self->{absent_file}->{$pb->{path}}}, $path;
2529 undef;
2530 }
2532 sub change_file_prop {
2533 my ($self, $fb, $prop, $value) = @_;
2534 if ($prop eq 'svn:executable') {
2535 if ($fb->{mode_b} != 120000) {
2536 $fb->{mode_b} = defined $value ? 100755 : 100644;
2537 }
2538 } elsif ($prop eq 'svn:special') {
2539 $fb->{mode_b} = defined $value ? 120000 : 100644;
2540 } else {
2541 $self->{file_prop}->{$fb->{path}} ||= {};
2542 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
2543 }
2544 undef;
2545 }
2547 sub apply_textdelta {
2548 my ($self, $fb, $exp) = @_;
2549 my $fh = IO::File->new_tmpfile;
2550 $fh->autoflush(1);
2551 # $fh gets auto-closed() by SVN::TxDelta::apply(),
2552 # (but $base does not,) so dup() it for reading in close_file
2553 open my $dup, '<&', $fh or croak $!;
2554 my $base = IO::File->new_tmpfile;
2555 $base->autoflush(1);
2556 if ($fb->{blob}) {
2557 defined (my $pid = fork) or croak $!;
2558 if (!$pid) {
2559 open STDOUT, '>&', $base or croak $!;
2560 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
2561 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
2562 }
2563 waitpid $pid, 0;
2564 croak $? if $?;
2566 if (defined $exp) {
2567 seek $base, 0, 0 or croak $!;
2568 my $md5 = Digest::MD5->new;
2569 $md5->addfile($base);
2570 my $got = $md5->hexdigest;
2571 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
2572 "expected: $exp\n",
2573 " got: $got\n" if ($got ne $exp);
2574 }
2575 }
2576 seek $base, 0, 0 or croak $!;
2577 $fb->{fh} = $dup;
2578 $fb->{base} = $base;
2579 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
2580 }
2582 sub close_file {
2583 my ($self, $fb, $exp) = @_;
2584 my $hash;
2585 my $path = $self->git_path($fb->{path});
2586 if (my $fh = $fb->{fh}) {
2587 if (defined $exp) {
2588 seek($fh, 0, 0) or croak $!;
2589 my $md5 = Digest::MD5->new;
2590 $md5->addfile($fh);
2591 my $got = $md5->hexdigest;
2592 if ($got ne $exp) {
2593 die "Checksum mismatch: $path\n",
2594 "expected: $exp\n got: $got\n";
2595 }
2596 }
2597 sysseek($fh, 0, 0) or croak $!;
2598 if ($fb->{mode_b} == 120000) {
2599 sysread($fh, my $buf, 5) == 5 or croak $!;
2600 $buf eq 'link ' or die "$path has mode 120000",
2601 "but is not a link\n";
2602 }
2603 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
2604 if (!$pid) {
2605 open STDIN, '<&', $fh or croak $!;
2606 exec qw/git-hash-object -w --stdin/ or croak $!;
2607 }
2608 chomp($hash = do { local $/; <$out> });
2609 close $out or croak $!;
2610 close $fh or croak $!;
2611 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
2612 close $fb->{base} or croak $!;
2613 } else {
2614 $hash = $fb->{blob} or die "no blob information\n";
2615 }
2616 $fb->{pool}->clear;
2617 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
2618 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
2619 undef;
2620 }
2622 sub abort_edit {
2623 my $self = shift;
2624 $self->{nr} = $self->{gii}->{nr};
2625 delete $self->{gii};
2626 $self->SUPER::abort_edit(@_);
2627 }
2629 sub close_edit {
2630 my $self = shift;
2631 $self->{git_commit_ok} = 1;
2632 $self->{nr} = $self->{gii}->{nr};
2633 delete $self->{gii};
2634 $self->SUPER::close_edit(@_);
2635 }
2637 package SVN::Git::Editor;
2638 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
2639 use strict;
2640 use warnings;
2641 use Carp qw/croak/;
2642 use IO::File;
2643 use Digest::MD5;
2645 sub new {
2646 my ($class, $opts) = @_;
2647 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
2648 die "$_ required!\n" unless (defined $opts->{$_});
2649 }
2651 my $pool = SVN::Pool->new;
2652 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
2653 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
2654 $opts->{r}, $mods);
2656 # $opts->{ra} functions should not be used after this:
2657 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
2658 $opts->{editor_cb}, $pool);
2659 my $self = SVN::Delta::Editor->new(@ce, $pool);
2660 bless $self, $class;
2661 foreach (qw/svn_path r tree_a tree_b/) {
2662 $self->{$_} = $opts->{$_};
2663 }
2664 $self->{url} = $opts->{ra}->{url};
2665 $self->{mods} = $mods;
2666 $self->{types} = $types;
2667 $self->{pool} = $pool;
2668 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2669 $self->{rm} = { };
2670 $self->{path_prefix} = length $self->{svn_path} ?
2671 "$self->{svn_path}/" : '';
2672 return $self;
2673 }
2675 sub generate_diff {
2676 my ($tree_a, $tree_b) = @_;
2677 my @diff_tree = qw(diff-tree -z -r);
2678 if ($_cp_similarity) {
2679 push @diff_tree, "-C$_cp_similarity";
2680 } else {
2681 push @diff_tree, '-C';
2682 }
2683 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
2684 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
2685 push @diff_tree, $tree_a, $tree_b;
2686 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2687 local $/ = "\0";
2688 my $state = 'meta';
2689 my @mods;
2690 while (<$diff_fh>) {
2691 chomp $_; # this gets rid of the trailing "\0"
2692 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2693 $::sha1\s($::sha1)\s
2694 ([MTCRAD])\d*$/xo) {
2695 push @mods, { mode_a => $1, mode_b => $2,
2696 sha1_b => $3, chg => $4 };
2697 if ($4 =~ /^(?:C|R)$/) {
2698 $state = 'file_a';
2699 } else {
2700 $state = 'file_b';
2701 }
2702 } elsif ($state eq 'file_a') {
2703 my $x = $mods[$#mods] or croak "Empty array\n";
2704 if ($x->{chg} !~ /^(?:C|R)$/) {
2705 croak "Error parsing $_, $x->{chg}\n";
2706 }
2707 $x->{file_a} = $_;
2708 $state = 'file_b';
2709 } elsif ($state eq 'file_b') {
2710 my $x = $mods[$#mods] or croak "Empty array\n";
2711 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2712 croak "Error parsing $_, $x->{chg}\n";
2713 }
2714 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2715 croak "Error parsing $_, $x->{chg}\n";
2716 }
2717 $x->{file_b} = $_;
2718 $state = 'meta';
2719 } else {
2720 croak "Error parsing $_\n";
2721 }
2722 }
2723 command_close_pipe($diff_fh, $ctx);
2724 \@mods;
2725 }
2727 sub check_diff_paths {
2728 my ($ra, $pfx, $rev, $mods) = @_;
2729 my %types;
2730 $pfx .= '/' if length $pfx;
2732 sub type_diff_paths {
2733 my ($ra, $types, $path, $rev) = @_;
2734 my @p = split m#/+#, $path;
2735 my $c = shift @p;
2736 unless (defined $types->{$c}) {
2737 $types->{$c} = $ra->check_path($c, $rev);
2738 }
2739 while (@p) {
2740 $c .= '/' . shift @p;
2741 next if defined $types->{$c};
2742 $types->{$c} = $ra->check_path($c, $rev);
2743 }
2744 }
2746 foreach my $m (@$mods) {
2747 foreach my $f (qw/file_a file_b/) {
2748 next unless defined $m->{$f};
2749 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
2750 if (length $pfx.$dir && ! defined $types{$dir}) {
2751 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
2752 }
2753 }
2754 }
2755 \%types;
2756 }
2758 sub split_path {
2759 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2760 }
2762 sub repo_path {
2763 my ($self, $path) = @_;
2764 $self->{path_prefix}.(defined $path ? $path : '');
2765 }
2767 sub url_path {
2768 my ($self, $path) = @_;
2769 if ($self->{url} =~ m#^https?://#) {
2770 $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
2771 }
2772 $self->{url} . '/' . $self->repo_path($path);
2773 }
2775 sub rmdirs {
2776 my ($self) = @_;
2777 my $rm = $self->{rm};
2778 delete $rm->{''}; # we never delete the url we're tracking
2779 return unless %$rm;
2781 foreach (keys %$rm) {
2782 my @d = split m#/#, $_;
2783 my $c = shift @d;
2784 $rm->{$c} = 1;
2785 while (@d) {
2786 $c .= '/' . shift @d;
2787 $rm->{$c} = 1;
2788 }
2789 }
2790 delete $rm->{$self->{svn_path}};
2791 delete $rm->{''}; # we never delete the url we're tracking
2792 return unless %$rm;
2794 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
2795 $self->{tree_b});
2796 local $/ = "\0";
2797 while (<$fh>) {
2798 chomp;
2799 my @dn = split m#/#, $_;
2800 while (pop @dn) {
2801 delete $rm->{join '/', @dn};
2802 }
2803 unless (%$rm) {
2804 close $fh;
2805 return;
2806 }
2807 }
2808 command_close_pipe($fh, $ctx);
2810 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2811 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2812 $self->close_directory($bat->{$d}, $p);
2813 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2814 print "\tD+\t$d/\n" unless $::_q;
2815 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2816 delete $bat->{$d};
2817 }
2818 }
2820 sub open_or_add_dir {
2821 my ($self, $full_path, $baton) = @_;
2822 my $t = $self->{types}->{$full_path};
2823 if (!defined $t) {
2824 die "$full_path not known in r$self->{r} or we have a bug!\n";
2825 }
2826 if ($t == $SVN::Node::none) {
2827 return $self->add_directory($full_path, $baton,
2828 undef, -1, $self->{pool});
2829 } elsif ($t == $SVN::Node::dir) {
2830 return $self->open_directory($full_path, $baton,
2831 $self->{r}, $self->{pool});
2832 }
2833 print STDERR "$full_path already exists in repository at ",
2834 "r$self->{r} and it is not a directory (",
2835 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2836 exit 1;
2837 }
2839 sub ensure_path {
2840 my ($self, $path) = @_;
2841 my $bat = $self->{bat};
2842 my $repo_path = $self->repo_path($path);
2843 return $bat->{''} unless (length $repo_path);
2844 my @p = split m#/+#, $repo_path;
2845 my $c = shift @p;
2846 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2847 while (@p) {
2848 my $c0 = $c;
2849 $c .= '/' . shift @p;
2850 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2851 }
2852 return $bat->{$c};
2853 }
2855 sub A {
2856 my ($self, $m) = @_;
2857 my ($dir, $file) = split_path($m->{file_b});
2858 my $pbat = $self->ensure_path($dir);
2859 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2860 undef, -1);
2861 print "\tA\t$m->{file_b}\n" unless $::_q;
2862 $self->chg_file($fbat, $m);
2863 $self->close_file($fbat,undef,$self->{pool});
2864 }
2866 sub C {
2867 my ($self, $m) = @_;
2868 my ($dir, $file) = split_path($m->{file_b});
2869 my $pbat = $self->ensure_path($dir);
2870 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2871 $self->url_path($m->{file_a}), $self->{r});
2872 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2873 $self->chg_file($fbat, $m);
2874 $self->close_file($fbat,undef,$self->{pool});
2875 }
2877 sub delete_entry {
2878 my ($self, $path, $pbat) = @_;
2879 my $rpath = $self->repo_path($path);
2880 my ($dir, $file) = split_path($rpath);
2881 $self->{rm}->{$dir} = 1;
2882 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2883 }
2885 sub R {
2886 my ($self, $m) = @_;
2887 my ($dir, $file) = split_path($m->{file_b});
2888 my $pbat = $self->ensure_path($dir);
2889 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2890 $self->url_path($m->{file_a}), $self->{r});
2891 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2892 $self->chg_file($fbat, $m);
2893 $self->close_file($fbat,undef,$self->{pool});
2895 ($dir, $file) = split_path($m->{file_a});
2896 $pbat = $self->ensure_path($dir);
2897 $self->delete_entry($m->{file_a}, $pbat);
2898 }
2900 sub M {
2901 my ($self, $m) = @_;
2902 my ($dir, $file) = split_path($m->{file_b});
2903 my $pbat = $self->ensure_path($dir);
2904 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2905 $pbat,$self->{r},$self->{pool});
2906 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2907 $self->chg_file($fbat, $m);
2908 $self->close_file($fbat,undef,$self->{pool});
2909 }
2911 sub T { shift->M(@_) }
2913 sub change_file_prop {
2914 my ($self, $fbat, $pname, $pval) = @_;
2915 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2916 }
2918 sub chg_file {
2919 my ($self, $fbat, $m) = @_;
2920 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2921 $self->change_file_prop($fbat,'svn:executable','*');
2922 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2923 $self->change_file_prop($fbat,'svn:executable',undef);
2924 }
2925 my $fh = IO::File->new_tmpfile or croak $!;
2926 if ($m->{mode_b} =~ /^120/) {
2927 print $fh 'link ' or croak $!;
2928 $self->change_file_prop($fbat,'svn:special','*');
2929 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2930 $self->change_file_prop($fbat,'svn:special',undef);
2931 }
2932 defined(my $pid = fork) or croak $!;
2933 if (!$pid) {
2934 open STDOUT, '>&', $fh or croak $!;
2935 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2936 }
2937 waitpid $pid, 0;
2938 croak $? if $?;
2939 $fh->flush == 0 or croak $!;
2940 seek $fh, 0, 0 or croak $!;
2942 my $md5 = Digest::MD5->new;
2943 $md5->addfile($fh) or croak $!;
2944 seek $fh, 0, 0 or croak $!;
2946 my $exp = $md5->hexdigest;
2947 my $pool = SVN::Pool->new;
2948 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2949 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2950 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2951 $pool->clear;
2953 close $fh or croak $!;
2954 }
2956 sub D {
2957 my ($self, $m) = @_;
2958 my ($dir, $file) = split_path($m->{file_b});
2959 my $pbat = $self->ensure_path($dir);
2960 print "\tD\t$m->{file_b}\n" unless $::_q;
2961 $self->delete_entry($m->{file_b}, $pbat);
2962 }
2964 sub close_edit {
2965 my ($self) = @_;
2966 my ($p,$bat) = ($self->{pool}, $self->{bat});
2967 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2968 next if $_ eq '';
2969 $self->close_directory($bat->{$_}, $p);
2970 }
2971 $self->close_directory($bat->{''}, $p);
2972 $self->SUPER::close_edit($p);
2973 $p->clear;
2974 }
2976 sub abort_edit {
2977 my ($self) = @_;
2978 $self->SUPER::abort_edit($self->{pool});
2979 }
2981 sub DESTROY {
2982 my $self = shift;
2983 $self->SUPER::DESTROY(@_);
2984 $self->{pool}->clear;
2985 }
2987 # this drives the editor
2988 sub apply_diff {
2989 my ($self) = @_;
2990 my $mods = $self->{mods};
2991 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2992 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
2993 my $f = $m->{chg};
2994 if (defined $o{$f}) {
2995 $self->$f($m);
2996 } else {
2997 fatal("Invalid change type: $f\n");
2998 }
2999 }
3000 $self->rmdirs if $_rmdir;
3001 if (@$mods == 0) {
3002 $self->abort_edit;
3003 } else {
3004 $self->close_edit;
3005 }
3006 return scalar @$mods;
3007 }
3009 package Git::SVN::Ra;
3010 use vars qw/@ISA $config_dir $_log_window_size/;
3011 use strict;
3012 use warnings;
3013 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3015 BEGIN {
3016 # enforce temporary pool usage for some simple functions
3017 no strict 'refs';
3018 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3019 my $SUPER = "SUPER::$f";
3020 *$f = sub {
3021 my $self = shift;
3022 my $pool = SVN::Pool->new;
3023 my @ret = $self->$SUPER(@_,$pool);
3024 $pool->clear;
3025 wantarray ? @ret : $ret[0];
3026 };
3027 }
3028 }
3030 sub new {
3031 my ($class, $url) = @_;
3032 $url =~ s!/+$!!;
3033 return $RA if ($RA && $RA->{url} eq $url);
3035 SVN::_Core::svn_config_ensure($config_dir, undef);
3036 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
3037 SVN::Client::get_simple_provider(),
3038 SVN::Client::get_ssl_server_trust_file_provider(),
3039 SVN::Client::get_simple_prompt_provider(
3040 \&Git::SVN::Prompt::simple, 2),
3041 SVN::Client::get_ssl_client_cert_file_provider(),
3042 SVN::Client::get_ssl_client_cert_prompt_provider(
3043 \&Git::SVN::Prompt::ssl_client_cert, 2),
3044 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3045 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3046 SVN::Client::get_username_provider(),
3047 SVN::Client::get_ssl_server_trust_prompt_provider(
3048 \&Git::SVN::Prompt::ssl_server_trust),
3049 SVN::Client::get_username_prompt_provider(
3050 \&Git::SVN::Prompt::username, 2),
3051 ]);
3052 my $config = SVN::Core::config_get_config($config_dir);
3053 $RA = undef;
3054 my $self = SVN::Ra->new(url => $url, auth => $baton,
3055 config => $config,
3056 pool => SVN::Pool->new,
3057 auth_provider_callbacks => $callbacks);
3058 $self->{svn_path} = $url;
3059 $self->{repos_root} = $self->get_repos_root;
3060 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3061 $self->{cache} = { check_path => { r => 0, data => {} },
3062 get_dir => { r => 0, data => {} } };
3063 $RA = bless $self, $class;
3064 }
3066 sub check_path {
3067 my ($self, $path, $r) = @_;
3068 my $cache = $self->{cache}->{check_path};
3069 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3070 return $cache->{data}->{$path};
3071 }
3072 my $pool = SVN::Pool->new;
3073 my $t = $self->SUPER::check_path($path, $r, $pool);
3074 $pool->clear;
3075 if ($r != $cache->{r}) {
3076 %{$cache->{data}} = ();
3077 $cache->{r} = $r;
3078 }
3079 $cache->{data}->{$path} = $t;
3080 }
3082 sub get_dir {
3083 my ($self, $dir, $r) = @_;
3084 my $cache = $self->{cache}->{get_dir};
3085 if ($r == $cache->{r}) {
3086 if (my $x = $cache->{data}->{$dir}) {
3087 return wantarray ? @$x : $x->[0];
3088 }
3089 }
3090 my $pool = SVN::Pool->new;
3091 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3092 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3093 $pool->clear;
3094 if ($r != $cache->{r}) {
3095 %{$cache->{data}} = ();
3096 $cache->{r} = $r;
3097 }
3098 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3099 wantarray ? (\%dirents, $r, $props) : \%dirents;
3100 }
3102 sub DESTROY {
3103 # do not call the real DESTROY since we store ourselves in $RA
3104 }
3106 sub get_log {
3107 my ($self, @args) = @_;
3108 my $pool = SVN::Pool->new;
3109 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3110 my $ret = $self->SUPER::get_log(@args, $pool);
3111 $pool->clear;
3112 $ret;
3113 }
3115 sub get_commit_editor {
3116 my ($self, $log, $cb, $pool) = @_;
3117 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3118 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3119 }
3121 sub gs_do_update {
3122 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3123 my $new = ($rev_a == $rev_b);
3124 my $path = $gs->{path};
3126 if ($new && -e $gs->{index}) {
3127 unlink $gs->{index} or die
3128 "Couldn't unlink index: $gs->{index}: $!\n";
3129 }
3130 my $pool = SVN::Pool->new;
3131 $editor->set_path_strip($path);
3132 my (@pc) = split m#/#, $path;
3133 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3134 1, $editor, $pool);
3135 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3137 # Since we can't rely on svn_ra_reparent being available, we'll
3138 # just have to do some magic with set_path to make it so
3139 # we only want a partial path.
3140 my $sp = '';
3141 my $final = join('/', @pc);
3142 while (@pc) {
3143 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3144 $sp .= '/' if length $sp;
3145 $sp .= shift @pc;
3146 }
3147 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3149 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3151 $reporter->finish_report($pool);
3152 $pool->clear;
3153 $editor->{git_commit_ok};
3154 }
3156 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3157 # svn_ra_reparent didn't work before 1.4)
3158 sub gs_do_switch {
3159 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3160 my $path = $gs->{path};
3161 my $pool = SVN::Pool->new;
3163 my $full_url = $self->{url};
3164 my $old_url = $full_url;
3165 $full_url .= "/$path" if length $path;
3166 my ($ra, $reparented);
3167 if ($old_url ne $full_url) {
3168 if ($old_url !~ m#^svn(\+ssh)?://#) {
3169 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3170 $pool);
3171 $self->{url} = $full_url;
3172 $reparented = 1;
3173 } else {
3174 $_[0] = undef;
3175 $self = undef;
3176 $RA = undef;
3177 $ra = Git::SVN::Ra->new($full_url);
3178 $ra_invalid = 1;
3179 }
3180 }
3181 $ra ||= $self;
3182 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3183 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3184 $reporter->set_path('', $rev_a, 0, @lock, $pool);
3185 $reporter->finish_report($pool);
3187 if ($reparented) {
3188 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3189 $self->{url} = $old_url;
3190 }
3192 $pool->clear;
3193 $editor->{git_commit_ok};
3194 }
3196 sub longest_common_path {
3197 my ($gsv, $globs) = @_;
3198 my %common;
3199 my $common_max = scalar @$gsv;
3201 foreach my $gs (@$gsv) {
3202 my @tmp = split m#/#, $gs->{path};
3203 my $p = '';
3204 foreach (@tmp) {
3205 $p .= length($p) ? "/$_" : $_;
3206 $common{$p} ||= 0;
3207 $common{$p}++;
3208 }
3209 }
3210 $globs ||= [];
3211 $common_max += scalar @$globs;
3212 foreach my $glob (@$globs) {
3213 my @tmp = split m#/#, $glob->{path}->{left};
3214 my $p = '';
3215 foreach (@tmp) {
3216 $p .= length($p) ? "/$_" : $_;
3217 $common{$p} ||= 0;
3218 $common{$p}++;
3219 }
3220 }
3222 my $longest_path = '';
3223 foreach (sort {length $b <=> length $a} keys %common) {
3224 if ($common{$_} == $common_max) {
3225 $longest_path = $_;
3226 last;
3227 }
3228 }
3229 $longest_path;
3230 }
3232 sub gs_fetch_loop_common {
3233 my ($self, $base, $head, $gsv, $globs) = @_;
3234 return if ($base > $head);
3235 my $inc = $_log_window_size;
3236 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3237 my $longest_path = longest_common_path($gsv, $globs);
3238 my $ra_url = $self->{url};
3239 while (1) {
3240 my %revs;
3241 my $err;
3242 my $err_handler = $SVN::Error::handler;
3243 $SVN::Error::handler = sub {
3244 ($err) = @_;
3245 skip_unknown_revs($err);
3246 };
3247 sub _cb {
3248 my ($paths, $r, $author, $date, $log) = @_;
3249 [ dup_changed_paths($paths),
3250 { author => $author, date => $date, log => $log } ];
3251 }
3252 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3253 sub { $revs{$_[1]} = _cb(@_) });
3254 if ($err && $max >= $head) {
3255 print STDERR "Path '$longest_path' ",
3256 "was probably deleted:\n",
3257 $err->expanded_message,
3258 "\nWill attempt to follow ",
3259 "revisions r$min .. r$max ",
3260 "committed before the deletion\n";
3261 my $hi = $max;
3262 while (--$hi >= $min) {
3263 my $ok;
3264 $self->get_log([$longest_path], $min, $hi,
3265 0, 1, 1, sub {
3266 $ok ||= $_[1];
3267 $revs{$_[1]} = _cb(@_) });
3268 if ($ok) {
3269 print STDERR "r$min .. r$ok OK\n";
3270 last;
3271 }
3272 }
3273 }
3274 $SVN::Error::handler = $err_handler;
3276 my %exists = map { $_->{path} => $_ } @$gsv;
3277 foreach my $r (sort {$a <=> $b} keys %revs) {
3278 my ($paths, $logged) = @{$revs{$r}};
3280 foreach my $gs ($self->match_globs(\%exists, $paths,
3281 $globs, $r)) {
3282 if ($gs->rev_db_max >= $r) {
3283 next;
3284 }
3285 next unless $gs->match_paths($paths, $r);
3286 $gs->{logged_rev_props} = $logged;
3287 if (my $last_commit = $gs->last_commit) {
3288 $gs->assert_index_clean($last_commit);
3289 }
3290 my $log_entry = $gs->do_fetch($paths, $r);
3291 if ($log_entry) {
3292 $gs->do_git_commit($log_entry);
3293 }
3294 }
3295 foreach my $g (@$globs) {
3296 my $k = "svn-remote.$g->{remote}." .
3297 "$g->{t}-maxRev";
3298 Git::SVN::tmp_config($k, $r);
3299 }
3300 if ($ra_invalid) {
3301 $_[0] = undef;
3302 $self = undef;
3303 $RA = undef;
3304 $self = Git::SVN::Ra->new($ra_url);
3305 $ra_invalid = undef;
3306 }
3307 }
3308 # pre-fill the .rev_db since it'll eventually get filled in
3309 # with '0' x40 if something new gets committed
3310 foreach my $gs (@$gsv) {
3311 next if defined $gs->rev_db_get($max);
3312 $gs->rev_db_set($max, 0 x40);
3313 }
3314 foreach my $g (@$globs) {
3315 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3316 Git::SVN::tmp_config($k, $max);
3317 }
3318 last if $max >= $head;
3319 $min = $max + 1;
3320 $max += $inc;
3321 $max = $head if ($max > $head);
3322 }
3323 }
3325 sub match_globs {
3326 my ($self, $exists, $paths, $globs, $r) = @_;
3328 sub get_dir_check {
3329 my ($self, $exists, $g, $r) = @_;
3330 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
3331 return unless scalar @x == 3;
3332 my $dirents = $x[0];
3333 foreach my $de (keys %$dirents) {
3334 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
3335 my $p = $g->{path}->full_path($de);
3336 next if $exists->{$p};
3337 next if (length $g->{path}->{right} &&
3338 ($self->check_path($p, $r) !=
3339 $SVN::Node::dir));
3340 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
3341 $g->{ref}->full_path($de), 1);
3342 }
3343 }
3344 foreach my $g (@$globs) {
3345 if (my $path = $paths->{"/$g->{path}->{left}"}) {
3346 if ($path->{action} =~ /^[AR]$/) {
3347 get_dir_check($self, $exists, $g, $r);
3348 }
3349 }
3350 foreach (keys %$paths) {
3351 if (/$g->{path}->{left_regex}/ &&
3352 !/$g->{path}->{regex}/) {
3353 next if $paths->{$_}->{action} !~ /^[AR]$/;
3354 get_dir_check($self, $exists, $g, $r);
3355 }
3356 next unless /$g->{path}->{regex}/;
3357 my $p = $1;
3358 my $pathname = $g->{path}->full_path($p);
3359 next if $exists->{$pathname};
3360 next if ($self->check_path($pathname, $r) !=
3361 $SVN::Node::dir);
3362 $exists->{$pathname} = Git::SVN->init(
3363 $self->{url}, $pathname, undef,
3364 $g->{ref}->full_path($p), 1);
3365 }
3366 my $c = '';
3367 foreach (split m#/#, $g->{path}->{left}) {
3368 $c .= "/$_";
3369 next unless ($paths->{$c} &&
3370 ($paths->{$c}->{action} =~ /^[AR]$/));
3371 get_dir_check($self, $exists, $g, $r);
3372 }
3373 }
3374 values %$exists;
3375 }
3377 sub minimize_url {
3378 my ($self) = @_;
3379 return $self->{url} if ($self->{url} eq $self->{repos_root});
3380 my $url = $self->{repos_root};
3381 my @components = split(m!/!, $self->{svn_path});
3382 my $c = '';
3383 do {
3384 $url .= "/$c" if length $c;
3385 eval { (ref $self)->new($url)->get_latest_revnum };
3386 } while ($@ && ($c = shift @components));
3387 $url;
3388 }
3390 sub can_do_switch {
3391 my $self = shift;
3392 unless (defined $can_do_switch) {
3393 my $pool = SVN::Pool->new;
3394 my $rep = eval {
3395 $self->do_switch(1, '', 0, $self->{url},
3396 SVN::Delta::Editor->new, $pool);
3397 };
3398 if ($@) {
3399 $can_do_switch = 0;
3400 } else {
3401 $rep->abort_report($pool);
3402 $can_do_switch = 1;
3403 }
3404 $pool->clear;
3405 }
3406 $can_do_switch;
3407 }
3409 sub skip_unknown_revs {
3410 my ($err) = @_;
3411 my $errno = $err->apr_err();
3412 # Maybe the branch we're tracking didn't
3413 # exist when the repo started, so it's
3414 # not an error if it doesn't, just continue
3415 #
3416 # Wonderfully consistent library, eh?
3417 # 160013 - svn:// and file://
3418 # 175002 - http(s)://
3419 # 175007 - http(s):// (this repo required authorization, too...)
3420 # More codes may be discovered later...
3421 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
3422 my $err_key = $err->expanded_message;
3423 # revision numbers change every time, filter them out
3424 $err_key =~ s/\d+/\0/g;
3425 $err_key = "$errno\0$err_key";
3426 unless ($ignored_err{$err_key}) {
3427 warn "W: Ignoring error from SVN, path probably ",
3428 "does not exist: ($errno): ",
3429 $err->expanded_message,"\n";
3430 $ignored_err{$err_key} = 1;
3431 }
3432 return;
3433 }
3434 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3435 }
3437 # svn_log_changed_path_t objects passed to get_log are likely to be
3438 # overwritten even if only the refs are copied to an external variable,
3439 # so we should dup the structures in their entirety. Using an externally
3440 # passed pool (instead of our temporary and quickly cleared pool in
3441 # Git::SVN::Ra) does not help matters at all...
3442 sub dup_changed_paths {
3443 my ($paths) = @_;
3444 return undef unless $paths;
3445 my %ret;
3446 foreach my $p (keys %$paths) {
3447 my $i = $paths->{$p};
3448 my %s = map { $_ => $i->$_ }
3449 qw/copyfrom_path copyfrom_rev action/;
3450 $ret{$p} = \%s;
3451 }
3452 \%ret;
3453 }
3455 package Git::SVN::Log;
3456 use strict;
3457 use warnings;
3458 use POSIX qw/strftime/;
3459 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
3460 %rusers $show_commit $incremental/;
3461 my $l_fmt;
3463 sub cmt_showable {
3464 my ($c) = @_;
3465 return 1 if defined $c->{r};
3467 # big commit message got truncated by the 16k pretty buffer in rev-list
3468 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
3469 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
3470 @{$c->{l}} = ();
3471 my @log = command(qw/cat-file commit/, $c->{c});
3473 # shift off the headers
3474 shift @log while ($log[0] ne '');
3475 shift @log;
3477 # TODO: make $c->{l} not have a trailing newline in the future
3478 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
3480 (undef, $c->{r}, undef) = ::extract_metadata(
3481 (grep(/^git-svn-id: /, @log))[-1]);
3482 }
3483 return defined $c->{r};
3484 }
3486 sub log_use_color {
3487 return 1 if $color;
3488 my ($dc, $dcvar);
3489 $dcvar = 'color.diff';
3490 $dc = `git-config --get $dcvar`;
3491 if ($dc eq '') {
3492 # nothing at all; fallback to "diff.color"
3493 $dcvar = 'diff.color';
3494 $dc = `git-config --get $dcvar`;
3495 }
3496 chomp($dc);
3497 if ($dc eq 'auto') {
3498 my $pc;
3499 $pc = `git-config --get color.pager`;
3500 if ($pc eq '') {
3501 # does not have it -- fallback to pager.color
3502 $pc = `git-config --bool --get pager.color`;
3503 }
3504 else {
3505 $pc = `git-config --bool --get color.pager`;
3506 if ($?) {
3507 $pc = 'false';
3508 }
3509 }
3510 chomp($pc);
3511 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
3512 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
3513 }
3514 return 0;
3515 }
3516 return 0 if $dc eq 'never';
3517 return 1 if $dc eq 'always';
3518 chomp($dc = `git-config --bool --get $dcvar`);
3519 return ($dc eq 'true');
3520 }
3522 sub git_svn_log_cmd {
3523 my ($r_min, $r_max, @args) = @_;
3524 my $head = 'HEAD';
3525 my (@files, @log_opts);
3526 foreach my $x (@args) {
3527 if ($x eq '--' || @files) {
3528 push @files, $x;
3529 } else {
3530 if (::verify_ref("$x^0")) {
3531 $head = $x;
3532 } else {
3533 push @log_opts, $x;
3534 }
3535 }
3536 }
3538 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
3539 $gs ||= Git::SVN->_new;
3540 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
3541 $gs->refname);
3542 push @cmd, '-r' unless $non_recursive;
3543 push @cmd, qw/--raw --name-status/ if $verbose;
3544 push @cmd, '--color' if log_use_color();
3545 push @cmd, @log_opts;
3546 if (defined $r_max && $r_max == $r_min) {
3547 push @cmd, '--max-count=1';
3548 if (my $c = $gs->rev_db_get($r_max)) {
3549 push @cmd, $c;
3550 }
3551 } elsif (defined $r_max) {
3552 my ($c_min, $c_max);
3553 $c_max = $gs->rev_db_get($r_max);
3554 $c_min = $gs->rev_db_get($r_min);
3555 if (defined $c_min && defined $c_max) {
3556 if ($r_max > $r_max) {
3557 push @cmd, "$c_min..$c_max";
3558 } else {
3559 push @cmd, "$c_max..$c_min";
3560 }
3561 } elsif ($r_max > $r_min) {
3562 push @cmd, $c_max;
3563 } else {
3564 push @cmd, $c_min;
3565 }
3566 }
3567 return (@cmd, @files);
3568 }
3570 # adapted from pager.c
3571 sub config_pager {
3572 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
3573 if (!defined $pager) {
3574 $pager = 'less';
3575 } elsif (length $pager == 0 || $pager eq 'cat') {
3576 $pager = undef;
3577 }
3578 }
3580 sub run_pager {
3581 return unless -t *STDOUT && defined $pager;
3582 pipe my $rfd, my $wfd or return;
3583 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
3584 if (!$pid) {
3585 open STDOUT, '>&', $wfd or
3586 ::fatal "Can't redirect to stdout: $!\n";
3587 return;
3588 }
3589 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
3590 $ENV{LESS} ||= 'FRSX';
3591 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
3592 }
3594 sub tz_to_s_offset {
3595 my ($tz) = @_;
3596 $tz =~ s/(\d\d)$//;
3597 return ($1 * 60) + ($tz * 3600);
3598 }
3600 sub get_author_info {
3601 my ($dest, $author, $t, $tz) = @_;
3602 $author =~ s/(?:^\s*|\s*$)//g;
3603 $dest->{a_raw} = $author;
3604 my $au;
3605 if ($::_authors) {
3606 $au = $rusers{$author} || undef;
3607 }
3608 if (!$au) {
3609 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
3610 }
3611 $dest->{t} = $t;
3612 $dest->{tz} = $tz;
3613 $dest->{a} = $au;
3614 # Date::Parse isn't in the standard Perl distro :(
3615 if ($tz =~ s/^\+//) {
3616 $t += tz_to_s_offset($tz);
3617 } elsif ($tz =~ s/^\-//) {
3618 $t -= tz_to_s_offset($tz);
3619 }
3620 $dest->{t_utc} = $t;
3621 }
3623 sub process_commit {
3624 my ($c, $r_min, $r_max, $defer) = @_;
3625 if (defined $r_min && defined $r_max) {
3626 if ($r_min == $c->{r} && $r_min == $r_max) {
3627 show_commit($c);
3628 return 0;
3629 }
3630 return 1 if $r_min == $r_max;
3631 if ($r_min < $r_max) {
3632 # we need to reverse the print order
3633 return 0 if (defined $limit && --$limit < 0);
3634 push @$defer, $c;
3635 return 1;
3636 }
3637 if ($r_min != $r_max) {
3638 return 1 if ($r_min < $c->{r});
3639 return 1 if ($r_max > $c->{r});
3640 }
3641 }
3642 return 0 if (defined $limit && --$limit < 0);
3643 show_commit($c);
3644 return 1;
3645 }
3647 sub show_commit {
3648 my $c = shift;
3649 if ($oneline) {
3650 my $x = "\n";
3651 if (my $l = $c->{l}) {
3652 while ($l->[0] =~ /^\s*$/) { shift @$l }
3653 $x = $l->[0];
3654 }
3655 $l_fmt ||= 'A' . length($c->{r});
3656 print 'r',pack($l_fmt, $c->{r}),' | ';
3657 print "$c->{c} | " if $show_commit;
3658 print $x;
3659 } else {
3660 show_commit_normal($c);
3661 }
3662 }
3664 sub show_commit_changed_paths {
3665 my ($c) = @_;
3666 return unless $c->{changed};
3667 print "Changed paths:\n", @{$c->{changed}};
3668 }
3670 sub show_commit_normal {
3671 my ($c) = @_;
3672 print '-' x72, "\nr$c->{r} | ";
3673 print "$c->{c} | " if $show_commit;
3674 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
3675 localtime($c->{t_utc})), ' | ';
3676 my $nr_line = 0;
3678 if (my $l = $c->{l}) {
3679 while ($l->[$#$l] eq "\n" && $#$l > 0
3680 && $l->[($#$l - 1)] eq "\n") {
3681 pop @$l;
3682 }
3683 $nr_line = scalar @$l;
3684 if (!$nr_line) {
3685 print "1 line\n\n\n";
3686 } else {
3687 if ($nr_line == 1) {
3688 $nr_line = '1 line';
3689 } else {
3690 $nr_line .= ' lines';
3691 }
3692 print $nr_line, "\n";
3693 show_commit_changed_paths($c);
3694 print "\n";
3695 print $_ foreach @$l;
3696 }
3697 } else {
3698 print "1 line\n";
3699 show_commit_changed_paths($c);
3700 print "\n";
3702 }
3703 foreach my $x (qw/raw stat diff/) {
3704 if ($c->{$x}) {
3705 print "\n";
3706 print $_ foreach @{$c->{$x}}
3707 }
3708 }
3709 }
3711 sub cmd_show_log {
3712 my (@args) = @_;
3713 my ($r_min, $r_max);
3714 my $r_last = -1; # prevent dupes
3715 if (defined $TZ) {
3716 $ENV{TZ} = $TZ;
3717 } else {
3718 delete $ENV{TZ};
3719 }
3720 if (defined $::_revision) {
3721 if ($::_revision =~ /^(\d+):(\d+)$/) {
3722 ($r_min, $r_max) = ($1, $2);
3723 } elsif ($::_revision =~ /^\d+$/) {
3724 $r_min = $r_max = $::_revision;
3725 } else {
3726 ::fatal "-r$::_revision is not supported, use ",
3727 "standard \'git log\' arguments instead\n";
3728 }
3729 }
3731 config_pager();
3732 @args = git_svn_log_cmd($r_min, $r_max, @args);
3733 my $log = command_output_pipe(@args);
3734 run_pager();
3735 my (@k, $c, $d, $stat);
3736 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
3737 while (<$log>) {
3738 if (/^${esc_color}commit ($::sha1_short)/o) {
3739 my $cmt = $1;
3740 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
3741 $r_last = $c->{r};
3742 process_commit($c, $r_min, $r_max, \@k) or
3743 goto out;
3744 }
3745 $d = undef;
3746 $c = { c => $cmt };
3747 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
3748 get_author_info($c, $1, $2, $3);
3749 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
3750 # ignore
3751 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
3752 push @{$c->{raw}}, $_;
3753 } elsif (/^${esc_color}[ACRMDT]\t/) {
3754 # we could add $SVN->{svn_path} here, but that requires
3755 # remote access at the moment (repo_path_split)...
3756 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
3757 push @{$c->{changed}}, $_;
3758 } elsif (/^${esc_color}diff /o) {
3759 $d = 1;
3760 push @{$c->{diff}}, $_;
3761 } elsif ($d) {
3762 push @{$c->{diff}}, $_;
3763 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
3764 $esc_color*[\+\-]*$esc_color$/x) {
3765 $stat = 1;
3766 push @{$c->{stat}}, $_;
3767 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
3768 push @{$c->{stat}}, $_;
3769 $stat = undef;
3770 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
3771 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
3772 } elsif (s/^${esc_color} //o) {
3773 push @{$c->{l}}, $_;
3774 }
3775 }
3776 if ($c && defined $c->{r} && $c->{r} != $r_last) {
3777 $r_last = $c->{r};
3778 process_commit($c, $r_min, $r_max, \@k);
3779 }
3780 if (@k) {
3781 my $swap = $r_max;
3782 $r_max = $r_min;
3783 $r_min = $swap;
3784 process_commit($_, $r_min, $r_max) foreach reverse @k;
3785 }
3786 out:
3787 close $log;
3788 print '-' x72,"\n" unless $incremental || $oneline;
3789 }
3791 package Git::SVN::Migration;
3792 # these version numbers do NOT correspond to actual version numbers
3793 # of git nor git-svn. They are just relative.
3794 #
3795 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
3796 #
3797 # v1 layout: .git/$id/info/url, refs/remotes/$id
3798 #
3799 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
3800 #
3801 # v3 layout: .git/svn/$id, refs/remotes/$id
3802 # - info/url may remain for backwards compatibility
3803 # - this is what we migrate up to this layout automatically,
3804 # - this will be used by git svn init on single branches
3805 # v3.1 layout (auto migrated):
3806 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
3807 # for backwards compatibility
3808 #
3809 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
3810 # - this is only created for newly multi-init-ed
3811 # repositories. Similar in spirit to the
3812 # --use-separate-remotes option in git-clone (now default)
3813 # - we do not automatically migrate to this (following
3814 # the example set by core git)
3815 use strict;
3816 use warnings;
3817 use Carp qw/croak/;
3818 use File::Path qw/mkpath/;
3819 use File::Basename qw/dirname basename/;
3820 use vars qw/$_minimize/;
3822 sub migrate_from_v0 {
3823 my $git_dir = $ENV{GIT_DIR};
3824 return undef unless -d $git_dir;
3825 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3826 my $migrated = 0;
3827 while (<$fh>) {
3828 chomp;
3829 my ($id, $orig_ref) = ($_, $_);
3830 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
3831 next unless -f "$git_dir/$id/info/url";
3832 my $new_ref = "refs/remotes/$id";
3833 if (::verify_ref("$new_ref^0")) {
3834 print STDERR "W: $orig_ref is probably an old ",
3835 "branch used by an ancient version of ",
3836 "git-svn.\n",
3837 "However, $new_ref also exists.\n",
3838 "We will not be able ",
3839 "to use this branch until this ",
3840 "ambiguity is resolved.\n";
3841 next;
3842 }
3843 print STDERR "Migrating from v0 layout...\n" if !$migrated;
3844 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
3845 command_noisy('update-ref', $new_ref, $orig_ref);
3846 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
3847 $migrated++;
3848 }
3849 command_close_pipe($fh, $ctx);
3850 print STDERR "Done migrating from v0 layout...\n" if $migrated;
3851 $migrated;
3852 }
3854 sub migrate_from_v1 {
3855 my $git_dir = $ENV{GIT_DIR};
3856 my $migrated = 0;
3857 return $migrated unless -d $git_dir;
3858 my $svn_dir = "$git_dir/svn";
3860 # just in case somebody used 'svn' as their $id at some point...
3861 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
3863 print STDERR "Migrating from a git-svn v1 layout...\n";
3864 mkpath([$svn_dir]);
3865 print STDERR "Data from a previous version of git-svn exists, but\n\t",
3866 "$svn_dir\n\t(required for this version ",
3867 "($::VERSION) of git-svn) does not. exist\n";
3868 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
3869 while (<$fh>) {
3870 my $x = $_;
3871 next unless $x =~ s#^refs/remotes/##;
3872 chomp $x;
3873 next unless -f "$git_dir/$x/info/url";
3874 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
3875 next unless $u;
3876 my $dn = dirname("$git_dir/svn/$x");
3877 mkpath([$dn]) unless -d $dn;
3878 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
3879 mkpath(["$git_dir/svn/svn"]);
3880 print STDERR " - $git_dir/$x/info => ",
3881 "$git_dir/svn/$x/info\n";
3882 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
3883 croak "$!: $x";
3884 # don't worry too much about these, they probably
3885 # don't exist with repos this old (save for index,
3886 # and we can easily regenerate that)
3887 foreach my $f (qw/unhandled.log index .rev_db/) {
3888 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
3889 }
3890 } else {
3891 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
3892 rename "$git_dir/$x", "$git_dir/svn/$x" or
3893 croak "$!: $x";
3894 }
3895 $migrated++;
3896 }
3897 command_close_pipe($fh, $ctx);
3898 print STDERR "Done migrating from a git-svn v1 layout\n";
3899 $migrated;
3900 }
3902 sub read_old_urls {
3903 my ($l_map, $pfx, $path) = @_;
3904 my @dir;
3905 foreach (<$path/*>) {
3906 if (-r "$_/info/url") {
3907 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
3908 my $ref_id = $pfx . basename $_;
3909 my $url = ::file_to_s("$_/info/url");
3910 $l_map->{$ref_id} = $url;
3911 } elsif (-d $_) {
3912 push @dir, $_;
3913 }
3914 }
3915 foreach (@dir) {
3916 my $x = $_;
3917 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
3918 read_old_urls($l_map, $x, $_);
3919 }
3920 }
3922 sub migrate_from_v2 {
3923 my @cfg = command(qw/config -l/);
3924 return if grep /^svn-remote\..+\.url=/, @cfg;
3925 my %l_map;
3926 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
3927 my $migrated = 0;
3929 foreach my $ref_id (sort keys %l_map) {
3930 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
3931 if ($@) {
3932 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
3933 }
3934 $migrated++;
3935 }
3936 $migrated;
3937 }
3939 sub minimize_connections {
3940 my $r = Git::SVN::read_all_remotes();
3941 my $new_urls = {};
3942 my $root_repos = {};
3943 foreach my $repo_id (keys %$r) {
3944 my $url = $r->{$repo_id}->{url} or next;
3945 my $fetch = $r->{$repo_id}->{fetch} or next;
3946 my $ra = Git::SVN::Ra->new($url);
3948 # skip existing cases where we already connect to the root
3949 if (($ra->{url} eq $ra->{repos_root}) ||
3950 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
3951 $repo_id)) {
3952 $root_repos->{$ra->{url}} = $repo_id;
3953 next;
3954 }
3956 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
3957 my $root_path = $ra->{url};
3958 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
3959 foreach my $path (keys %$fetch) {
3960 my $ref_id = $fetch->{$path};
3961 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
3963 # make sure we can read when connecting to
3964 # a higher level of a repository
3965 my ($last_rev, undef) = $gs->last_rev_commit;
3966 if (!defined $last_rev) {
3967 $last_rev = eval {
3968 $root_ra->get_latest_revnum;
3969 };
3970 next if $@;
3971 }
3972 my $new = $root_path;
3973 $new .= length $path ? "/$path" : '';
3974 eval {
3975 $root_ra->get_log([$new], $last_rev, $last_rev,
3976 0, 0, 1, sub { });
3977 };
3978 next if $@;
3979 $new_urls->{$ra->{repos_root}}->{$new} =
3980 { ref_id => $ref_id,
3981 old_repo_id => $repo_id,
3982 old_path => $path };
3983 }
3984 }
3986 my @emptied;
3987 foreach my $url (keys %$new_urls) {
3988 # see if we can re-use an existing [svn-remote "repo_id"]
3989 # instead of creating a(n ugly) new section:
3990 my $repo_id = $root_repos->{$url} ||
3991 Git::SVN::sanitize_remote_name($url);
3993 my $fetch = $new_urls->{$url};
3994 foreach my $path (keys %$fetch) {
3995 my $x = $fetch->{$path};
3996 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
3997 my $pfx = "svn-remote.$x->{old_repo_id}";
3999 my $old_fetch = quotemeta("$x->{old_path}:".
4000 "refs/remotes/$x->{ref_id}");
4001 command_noisy(qw/config --unset/,
4002 "$pfx.fetch", '^'. $old_fetch . '$');
4003 delete $r->{$x->{old_repo_id}}->
4004 {fetch}->{$x->{old_path}};
4005 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4006 command_noisy(qw/config --unset/,
4007 "$pfx.url");
4008 push @emptied, $x->{old_repo_id}
4009 }
4010 }
4011 }
4012 if (@emptied) {
4013 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4014 "$ENV{GIT_DIR}/config";
4015 print STDERR <<EOF;
4016 The following [svn-remote] sections in your config file ($file) are empty
4017 and can be safely removed:
4018 EOF
4019 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4020 }
4021 }
4023 sub migration_check {
4024 migrate_from_v0();
4025 migrate_from_v1();
4026 migrate_from_v2();
4027 minimize_connections() if $_minimize;
4028 }
4030 package Git::IndexInfo;
4031 use strict;
4032 use warnings;
4033 use Git qw/command_input_pipe command_close_pipe/;
4035 sub new {
4036 my ($class) = @_;
4037 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4038 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4039 }
4041 sub remove {
4042 my ($self, $path) = @_;
4043 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4044 return ++$self->{nr};
4045 }
4046 undef;
4047 }
4049 sub update {
4050 my ($self, $mode, $hash, $path) = @_;
4051 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4052 return ++$self->{nr};
4053 }
4054 undef;
4055 }
4057 sub DESTROY {
4058 my ($self) = @_;
4059 command_close_pipe($self->{gui}, $self->{ctx});
4060 }
4062 package Git::SVN::GlobSpec;
4063 use strict;
4064 use warnings;
4066 sub new {
4067 my ($class, $glob) = @_;
4068 my $re = $glob;
4069 $re =~ s!/+$!!g; # no need for trailing slashes
4070 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
4071 my ($left, $right) = ($1, $2);
4072 if ($nr > 1) {
4073 die "Only one '*' wildcard expansion ",
4074 "is supported (got $nr): '$glob'\n";
4075 } elsif ($nr == 0) {
4076 die "One '*' is needed for glob: '$glob'\n";
4077 }
4078 $re = quotemeta($left) . $re . quotemeta($right);
4079 if (length $left && !($left =~ s!/+$!!g)) {
4080 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4081 }
4082 if (length $right && !($right =~ s!^/+!!g)) {
4083 die "Missing leading '/' on right side of: '$glob' ($right)\n";
4084 }
4085 my $left_re = qr/^\/\Q$left\E(\/|$)/;
4086 bless { left => $left, right => $right, left_regex => $left_re,
4087 regex => qr/$re/, glob => $glob }, $class;
4088 }
4090 sub full_path {
4091 my ($self, $path) = @_;
4092 return (length $self->{left} ? "$self->{left}/" : '') .
4093 $path . (length $self->{right} ? "/$self->{right}" : '');
4094 }
4096 __END__
4098 Data structures:
4101 $remotes = { # returned by read_all_remotes()
4102 'svn' => {
4103 # svn-remote.svn.url=https://svn.musicpd.org
4104 url => 'https://svn.musicpd.org',
4105 # svn-remote.svn.fetch=mpd/trunk:trunk
4106 fetch => {
4107 'mpd/trunk' => 'trunk',
4108 },
4109 # svn-remote.svn.tags=mpd/tags/*:tags/*
4110 tags => {
4111 path => {
4112 left => 'mpd/tags',
4113 right => '',
4114 regex => qr!mpd/tags/([^/]+)$!,
4115 glob => 'tags/*',
4116 },
4117 ref => {
4118 left => 'tags',
4119 right => '',
4120 regex => qr!tags/([^/]+)$!,
4121 glob => 'tags/*',
4122 },
4123 }
4124 }
4125 };
4127 $log_entry hashref as returned by libsvn_log_entry()
4128 {
4129 log => 'whitespace-formatted log entry
4130 ', # trailing newline is preserved
4131 revision => '8', # integer
4132 date => '2004-02-24T17:01:44.108345Z', # commit date
4133 author => 'committer name'
4134 };
4137 # this is generated by generate_diff();
4138 @mods = array of diff-index line hashes, each element represents one line
4139 of diff-index output
4141 diff-index line ($m hash)
4142 {
4143 mode_a => first column of diff-index output, no leading ':',
4144 mode_b => second column of diff-index output,
4145 sha1_b => sha1sum of the final blob,
4146 chg => change type [MCRADT],
4147 file_a => original file name of a file (iff chg is 'C' or 'R')
4148 file_b => new/current file name of a file (any chg)
4149 }
4150 ;
4152 # retval of read_url_paths{,_all}();
4153 $l_map = {
4154 # repository root url
4155 'https://svn.musicpd.org' => {
4156 # repository path # GIT_SVN_ID
4157 'mpd/trunk' => 'trunk',
4158 'mpd/tags/0.11.5' => 'tags/0.11.5',
4159 },
4160 }
4162 Notes:
4163 I don't trust the each() function on unless I created %hash myself
4164 because the internal iterator may not have started at base.