?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
gitweb: Use --no-commit-id in git_commit and git_commitdiff
[git/jnareb-git/bp-gitweb.git] / git-svn.perl
blob37ecc517879aa2c18cc909c60d3bf6ceed82fb48
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 $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID
8 $GIT_SVN_INDEX $GIT_SVN
9 $GIT_DIR $GIT_SVN_DIR $REVDB/;
10 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
11 $VERSION = '@@GIT_VERSION@@';
12
13 use Cwd qw/abs_path/;
14 $GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
15 $ENV{GIT_DIR} = $GIT_DIR;
16
17 my $LC_ALL = $ENV{LC_ALL};
18 my $TZ = $ENV{TZ};
19 # make sure the svn binary gives consistent output between locales and TZs:
20 $ENV{TZ} = 'UTC';
21 $ENV{LC_ALL} = 'C';
22 $| = 1; # unbuffer STDOUT
23
24 # If SVN:: library support is added, please make the dependencies
25 # optional and preserve the capability to use the command-line client.
26 # use eval { require SVN::... } to make it lazy load
27 # We don't use any modules not in the standard Perl distribution:
28 use Carp qw/croak/;
29 use IO::File qw//;
30 use File::Basename qw/dirname basename/;
31 use File::Path qw/mkpath/;
32 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
33 use File::Spec qw//;
34 use File::Copy qw/copy/;
35 use POSIX qw/strftime/;
36 use IPC::Open3;
37 use Memoize;
38 memoize('revisions_eq');
39 memoize('cmt_metadata');
40 memoize('get_commit_time');
41
42 my ($SVN_PATH, $SVN, $SVN_LOG, $_use_lib);
43
44 sub nag_lib {
45 print STDERR <<EOF;
46 ! Please consider installing the SVN Perl libraries (version 1.1.0 or
47 ! newer). You will generally get better performance and fewer bugs,
48 ! especially if you:
49 ! 1) have a case-insensitive filesystem
50 ! 2) replace symlinks with files (and vice-versa) in commits
51
52 EOF
53 }
54
55 $_use_lib = 1 unless $ENV{GIT_SVN_NO_LIB};
56 libsvn_load();
57 nag_lib() unless $_use_lib;
58
59 my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
60 my $sha1 = qr/[a-f\d]{40}/;
61 my $sha1_short = qr/[a-f\d]{4,40}/;
62 my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
63 $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
64 $_repack, $_repack_nr, $_repack_flags, $_q,
65 $_message, $_file, $_follow_parent, $_no_metadata,
66 $_template, $_shared, $_no_default_regex, $_no_graft_copy,
67 $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit,
68 $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m,
69 $_merge, $_strategy, $_dry_run, $_ignore_nodate, $_non_recursive);
70 my (@_branch_from, %tree_map, %users, %rusers, %equiv);
71 my ($_svn_co_url_revs, $_svn_pg_peg_revs);
72 my @repo_path_split_cache;
73
74 my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
75 'branch|b=s' => \@_branch_from,
76 'follow-parent|follow' => \$_follow_parent,
77 'branch-all-refs|B' => \$_branch_all_refs,
78 'authors-file|A=s' => \$_authors,
79 'repack:i' => \$_repack,
80 'no-metadata' => \$_no_metadata,
81 'quiet|q' => \$_q,
82 'ignore-nodate' => \$_ignore_nodate,
83 'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
84
85 my ($_trunk, $_tags, $_branches);
86 my %multi_opts = ( 'trunk|T=s' => \$_trunk,
87 'tags|t=s' => \$_tags,
88 'branches|b=s' => \$_branches );
89 my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
90 my %cmt_opts = ( 'edit|e' => \$_edit,
91 'rmdir' => \$_rmdir,
92 'find-copies-harder' => \$_find_copies_harder,
93 'l=i' => \$_l,
94 'copy-similarity|C=i'=> \$_cp_similarity
95 );
96
97 my %cmd = (
98 fetch => [ \&fetch, "Download new revisions from SVN",
99 { 'revision|r=s' => \$_revision, %fc_opts } ],
100 init => [ \&init, "Initialize a repo for tracking" .
101 " (requires URL argument)",
102 \%init_opts ],
103 commit => [ \&commit, "Commit git revisions to SVN",
104 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
105 'show-ignore' => [ \&show_ignore, "Show svn:ignore listings",
106 { 'revision|r=i' => \$_revision } ],
107 rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
108 { 'no-ignore-externals' => \$_no_ignore_ext,
109 'copy-remote|remote=s' => \$_cp_remote,
110 'upgrade' => \$_upgrade } ],
111 'graft-branches' => [ \&graft_branches,
112 'Detect merges/branches from already imported history',
113 { 'merge-rx|m' => \@_opt_m,
114 'branch|b=s' => \@_branch_from,
115 'branch-all-refs|B' => \$_branch_all_refs,
116 'no-default-regex' => \$_no_default_regex,
117 'no-graft-copy' => \$_no_graft_copy } ],
118 'multi-init' => [ \&multi_init,
119 'Initialize multiple trees (like git-svnimport)',
120 { %multi_opts, %fc_opts } ],
121 'multi-fetch' => [ \&multi_fetch,
122 'Fetch multiple trees (like git-svnimport)',
123 \%fc_opts ],
124 'log' => [ \&show_log, 'Show commit logs',
125 { 'limit=i' => \$_limit,
126 'revision|r=s' => \$_revision,
127 'verbose|v' => \$_verbose,
128 'incremental' => \$_incremental,
129 'oneline' => \$_oneline,
130 'show-commit' => \$_show_commit,
131 'non-recursive' => \$_non_recursive,
132 'authors-file|A=s' => \$_authors,
133 } ],
134 'commit-diff' => [ \&commit_diff, 'Commit a diff between two trees',
135 { 'message|m=s' => \$_message,
136 'file|F=s' => \$_file,
137 %cmt_opts } ],
138 dcommit => [ \&dcommit, 'Commit several diffs to merge with upstream',
139 { 'merge|m|M' => \$_merge,
140 'strategy|s=s' => \$_strategy,
141 'dry-run|n' => \$_dry_run,
142 %cmt_opts } ],
143 );
144
145 my $cmd;
146 for (my $i = 0; $i < @ARGV; $i++) {
147 if (defined $cmd{$ARGV[$i]}) {
148 $cmd = $ARGV[$i];
149 splice @ARGV, $i, 1;
150 last;
151 }
152 };
153
154 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
155
156 read_repo_config(\%opts);
157 my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
158 'version|V' => \$_version,
159 'id|i=s' => \$GIT_SVN);
160 exit 1 if (!$rv && $cmd ne 'log');
161
162 set_default_vals();
163 usage(0) if $_help;
164 version() if $_version;
165 usage(1) unless defined $cmd;
166 init_vars();
167 load_authors() if $_authors;
168 load_all_refs() if $_branch_all_refs;
169 svn_compat_check() unless $_use_lib;
170 migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/;
171 $cmd{$cmd}->[0]->(@ARGV);
172 exit 0;
173
174 ####################### primary functions ######################
175 sub usage {
176 my $exit = shift || 0;
177 my $fd = $exit ? \*STDERR : \*STDOUT;
178 print $fd <<"";
179 git-svn - bidirectional operations between a single Subversion tree and git
180 Usage: $0 <command> [options] [arguments]\n
181
182 print $fd "Available commands:\n" unless $cmd;
183
184 foreach (sort keys %cmd) {
185 next if $cmd && $cmd ne $_;
186 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
187 foreach (keys %{$cmd{$_}->[2]}) {
188 # prints out arguments as they should be passed:
189 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
190 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
191 "--$_" : "-$_" }
192 split /\|/,$_)," $x\n";
193 }
194 }
195 print $fd <<"";
196 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
197 arbitrary identifier if you're tracking multiple SVN branches/repositories in
198 one git repository and want to keep them separate. See git-svn(1) for more
199 information.
200
201 exit $exit;
202 }
203
204 sub version {
205 print "git-svn version $VERSION\n";
206 exit 0;
207 }
208
209 sub rebuild {
210 if (quiet_run(qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0")) {
211 copy_remote_ref();
212 }
213 $SVN_URL = shift or undef;
214 my $newest_rev = 0;
215 if ($_upgrade) {
216 sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
217 } else {
218 check_upgrade_needed();
219 }
220
221 my $pid = open(my $rev_list,'-|');
222 defined $pid or croak $!;
223 if ($pid == 0) {
224 exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
225 }
226 my $latest;
227 while (<$rev_list>) {
228 chomp;
229 my $c = $_;
230 croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
231 my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
232 next if (!@commit); # skip merges
233 my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]);
234 if (!$rev || !$uuid) {
235 croak "Unable to extract revision or UUID from ",
236 "$c, $commit[$#commit]\n";
237 }
238
239 # if we merged or otherwise started elsewhere, this is
240 # how we break out of it
241 next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
242 next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
243
244 unless (defined $latest) {
245 if (!$SVN_URL && !$url) {
246 croak "SVN repository location required: $url\n";
247 }
248 $SVN_URL ||= $url;
249 $SVN_UUID ||= $uuid;
250 setup_git_svn();
251 $latest = $rev;
252 }
253 revdb_set($REVDB, $rev, $c);
254 print "r$rev = $c\n";
255 $newest_rev = $rev if ($rev > $newest_rev);
256 }
257 close $rev_list or croak $?;
258
259 goto out if $_use_lib;
260 if (!chdir $SVN_WC) {
261 svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
262 chdir $SVN_WC or croak $!;
263 }
264
265 $pid = fork;
266 defined $pid or croak $!;
267 if ($pid == 0) {
268 my @svn_up = qw(svn up);
269 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
270 sys(@svn_up,"-r$newest_rev");
271 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
272 index_changes();
273 exec('git-write-tree') or croak $!;
274 }
275 waitpid $pid, 0;
276 croak $? if $?;
277 out:
278 if ($_upgrade) {
279 print STDERR <<"";
280 Keeping deprecated refs/head/$GIT_SVN-HEAD for now. Please remove it
281 when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
282
283 }
284 }
285
286 sub init {
287 my $url = shift or die "SVN repository location required " .
288 "as a command-line argument\n";
289 $url =~ s!/+$!!; # strip trailing slash
290
291 if (my $repo_path = shift) {
292 unless (-d $repo_path) {
293 mkpath([$repo_path]);
294 }
295 $GIT_DIR = $ENV{GIT_DIR} = $repo_path . "/.git";
296 init_vars();
297 }
298
299 $SVN_URL = $url;
300 unless (-d $GIT_DIR) {
301 my @init_db = ('git-init-db');
302 push @init_db, "--template=$_template" if defined $_template;
303 push @init_db, "--shared" if defined $_shared;
304 sys(@init_db);
305 }
306 setup_git_svn();
307 }
308
309 sub fetch {
310 check_upgrade_needed();
311 $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
312 my $ret = $_use_lib ? fetch_lib(@_) : fetch_cmd(@_);
313 if ($ret->{commit} && quiet_run(qw(git-rev-parse --verify
314 refs/heads/master^0))) {
315 sys(qw(git-update-ref refs/heads/master),$ret->{commit});
316 }
317 return $ret;
318 }
319
320 sub fetch_cmd {
321 my (@parents) = @_;
322 my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
323 unless ($_revision) {
324 $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
325 }
326 push @log_args, "-r$_revision";
327 push @log_args, '--stop-on-copy' unless $_no_stop_copy;
328
329 my $svn_log = svn_log_raw(@log_args);
330
331 my $base = next_log_entry($svn_log) or croak "No base revision!\n";
332 # don't need last_revision from grab_base_rev() because
333 # user could've specified a different revision to skip (they
334 # didn't want to import certain revisions into git for whatever
335 # reason, so trust $base->{revision} instead.
336 my (undef, $last_commit) = svn_grab_base_rev();
337 unless (-d $SVN_WC) {
338 svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
339 chdir $SVN_WC or croak $!;
340 read_uuid();
341 $last_commit = git_commit($base, @parents);
342 assert_tree($last_commit);
343 } else {
344 chdir $SVN_WC or croak $!;
345 read_uuid();
346 # looks like a user manually cp'd and svn switch'ed
347 unless ($last_commit) {
348 sys(qw/svn revert -R ./);
349 assert_svn_wc_clean($base->{revision});
350 $last_commit = git_commit($base, @parents);
351 assert_tree($last_commit);
352 }
353 }
354 my @svn_up = qw(svn up);
355 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
356 my $last = $base;
357 while (my $log_msg = next_log_entry($svn_log)) {
358 if ($last->{revision} >= $log_msg->{revision}) {
359 croak "Out of order: last >= current: ",
360 "$last->{revision} >= $log_msg->{revision}\n";
361 }
362 # Revert is needed for cases like:
363 # https://svn.musicpd.org/Jamming/trunk (r166:167), but
364 # I can't seem to reproduce something like that on a test...
365 sys(qw/svn revert -R ./);
366 assert_svn_wc_clean($last->{revision});
367 sys(@svn_up,"-r$log_msg->{revision}");
368 $last_commit = git_commit($log_msg, $last_commit, @parents);
369 $last = $log_msg;
370 }
371 close $svn_log->{fh};
372 $last->{commit} = $last_commit;
373 return $last;
374 }
375
376 sub fetch_lib {
377 my (@parents) = @_;
378 $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
379 my $repo;
380 ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
381 $SVN_LOG ||= libsvn_connect($repo);
382 $SVN ||= libsvn_connect($repo);
383 my ($last_rev, $last_commit) = svn_grab_base_rev();
384 my ($base, $head) = libsvn_parse_revision($last_rev);
385 if ($base > $head) {
386 return { revision => $last_rev, commit => $last_commit }
387 }
388 my $index = set_index($GIT_SVN_INDEX);
389
390 # limit ourselves and also fork() since get_log won't release memory
391 # after processing a revision and SVN stuff seems to leak
392 my $inc = 1000;
393 my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
394 read_uuid();
395 if (defined $last_commit) {
396 unless (-e $GIT_SVN_INDEX) {
397 sys(qw/git-read-tree/, $last_commit);
398 }
399 chomp (my $x = `git-write-tree`);
400 my ($y) = (`git-cat-file commit $last_commit`
401 =~ /^tree ($sha1)/m);
402 if ($y ne $x) {
403 unlink $GIT_SVN_INDEX or croak $!;
404 sys(qw/git-read-tree/, $last_commit);
405 }
406 chomp ($x = `git-write-tree`);
407 if ($y ne $x) {
408 print STDERR "trees ($last_commit) $y != $x\n",
409 "Something is seriously wrong...\n";
410 }
411 }
412 while (1) {
413 # fork, because using SVN::Pool with get_log() still doesn't
414 # seem to help enough to keep memory usage down.
415 defined(my $pid = fork) or croak $!;
416 if (!$pid) {
417 $SVN::Error::handler = \&libsvn_skip_unknown_revs;
418
419 # Yes I'm perfectly aware that the fourth argument
420 # below is the limit revisions number. Unfortunately
421 # performance sucks with it enabled, so it's much
422 # faster to fetch revision ranges instead of relying
423 # on the limiter.
424 libsvn_get_log($SVN_LOG, '/'.$SVN_PATH,
425 $min, $max, 0, 1, 1,
426 sub {
427 my $log_msg;
428 if ($last_commit) {
429 $log_msg = libsvn_fetch(
430 $last_commit, @_);
431 $last_commit = git_commit(
432 $log_msg,
433 $last_commit,
434 @parents);
435 } else {
436 $log_msg = libsvn_new_tree(@_);
437 $last_commit = git_commit(
438 $log_msg, @parents);
439 }
440 });
441 exit 0;
442 }
443 waitpid $pid, 0;
444 croak $? if $?;
445 ($last_rev, $last_commit) = svn_grab_base_rev();
446 last if ($max >= $head);
447 $min = $max + 1;
448 $max += $inc;
449 $max = $head if ($max > $head);
450 }
451 restore_index($index);
452 return { revision => $last_rev, commit => $last_commit };
453 }
454
455 sub commit {
456 my (@commits) = @_;
457 check_upgrade_needed();
458 if ($_stdin || !@commits) {
459 print "Reading from stdin...\n";
460 @commits = ();
461 while (<STDIN>) {
462 if (/\b($sha1_short)\b/o) {
463 unshift @commits, $1;
464 }
465 }
466 }
467 my @revs;
468 foreach my $c (@commits) {
469 chomp(my @tmp = safe_qx('git-rev-parse',$c));
470 if (scalar @tmp == 1) {
471 push @revs, $tmp[0];
472 } elsif (scalar @tmp > 1) {
473 push @revs, reverse (safe_qx('git-rev-list',@tmp));
474 } else {
475 die "Failed to rev-parse $c\n";
476 }
477 }
478 chomp @revs;
479 $_use_lib ? commit_lib(@revs) : commit_cmd(@revs);
480 print "Done committing ",scalar @revs," revisions to SVN\n";
481 }
482
483 sub commit_cmd {
484 my (@revs) = @_;
485
486 chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n";
487 my $info = svn_info('.');
488 my $fetched = fetch();
489 if ($info->{Revision} != $fetched->{revision}) {
490 print STDERR "There are new revisions that were fetched ",
491 "and need to be merged (or acknowledged) ",
492 "before committing.\n";
493 exit 1;
494 }
495 $info = svn_info('.');
496 read_uuid($info);
497 my $last = $fetched;
498 foreach my $c (@revs) {
499 my $mods = svn_checkout_tree($last, $c);
500 if (scalar @$mods == 0) {
501 print "Skipping, no changes detected\n";
502 next;
503 }
504 $last = svn_commit_tree($last, $c);
505 }
506 }
507
508 sub commit_lib {
509 my (@revs) = @_;
510 my ($r_last, $cmt_last) = svn_grab_base_rev();
511 defined $r_last or die "Must have an existing revision to commit\n";
512 my $fetched = fetch();
513 if ($r_last != $fetched->{revision}) {
514 print STDERR "There are new revisions that were fetched ",
515 "and need to be merged (or acknowledged) ",
516 "before committing.\n",
517 "last rev: $r_last\n",
518 " current: $fetched->{revision}\n";
519 exit 1;
520 }
521 read_uuid();
522 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
523 my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
524
525 my $repo;
526 ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
527 set_svn_commit_env();
528 foreach my $c (@revs) {
529 my $log_msg = get_commit_message($c, $commit_msg);
530
531 # fork for each commit because there's a memory leak I
532 # can't track down... (it's probably in the SVN code)
533 defined(my $pid = open my $fh, '-|') or croak $!;
534 if (!$pid) {
535 $SVN_LOG = libsvn_connect($repo);
536 $SVN = libsvn_connect($repo);
537 my $ed = SVN::Git::Editor->new(
538 { r => $r_last,
539 ra => $SVN_LOG,
540 c => $c,
541 svn_path => $SVN_PATH
542 },
543 $SVN->get_commit_editor(
544 $log_msg->{msg},
545 sub {
546 libsvn_commit_cb(
547 @_, $c,
548 $log_msg->{msg},
549 $r_last,
550 $cmt_last)
551 },
552 @lock)
553 );
554 my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
555 if (@$mods == 0) {
556 print "No changes\nr$r_last = $cmt_last\n";
557 $ed->abort_edit;
558 } else {
559 $ed->close_edit;
560 }
561 exit 0;
562 }
563 my ($r_new, $cmt_new, $no);
564 while (<$fh>) {
565 print $_;
566 chomp;
567 if (/^r(\d+) = ($sha1)$/o) {
568 ($r_new, $cmt_new) = ($1, $2);
569 } elsif ($_ eq 'No changes') {
570 $no = 1;
571 }
572 }
573 close $fh or croak $?;
574 if (! defined $r_new && ! defined $cmt_new) {
575 unless ($no) {
576 die "Failed to parse revision information\n";
577 }
578 } else {
579 ($r_last, $cmt_last) = ($r_new, $cmt_new);
580 }
581 }
582 $ENV{LC_ALL} = 'C';
583 unlink $commit_msg;
584 }
585
586 sub dcommit {
587 my $gs = "refs/remotes/$GIT_SVN";
588 chomp(my @refs = safe_qx(qw/git-rev-list --no-merges/, "$gs..HEAD"));
589 foreach my $d (reverse @refs) {
590 if ($_dry_run) {
591 print "diff-tree $d~1 $d\n";
592 } else {
593 commit_diff("$d~1", $d);
594 }
595 }
596 return if $_dry_run;
597 fetch();
598 my @diff = safe_qx(qw/git-diff-tree HEAD/, $gs);
599 my @finish;
600 if (@diff) {
601 @finish = qw/rebase/;
602 push @finish, qw/--merge/ if $_merge;
603 push @finish, "--strategy=$_strategy" if $_strategy;
604 print STDERR "W: HEAD and $gs differ, using @finish:\n", @diff;
605 } else {
606 print "No changes between current HEAD and $gs\n",
607 "Hard resetting to the latest $gs\n";
608 @finish = qw/reset --hard/;
609 }
610 sys('git', @finish, $gs);
611 }
612
613 sub show_ignore {
614 $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
615 $_use_lib ? show_ignore_lib() : show_ignore_cmd();
616 }
617
618 sub show_ignore_cmd {
619 require File::Find or die $!;
620 if (defined $_revision) {
621 die "-r/--revision option doesn't work unless the Perl SVN ",
622 "libraries are used\n";
623 }
624 chdir $SVN_WC or croak $!;
625 my %ign;
626 File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
627 s#^\./##;
628 @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
629 }}, no_chdir=>1},'.');
630
631 print "\n# /\n";
632 foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
633 delete $ign{'.'};
634 foreach my $i (sort keys %ign) {
635 print "\n# ",$i,"\n";
636 foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
637 }
638 }
639
640 sub show_ignore_lib {
641 my $repo;
642 ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
643 $SVN ||= libsvn_connect($repo);
644 my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
645 libsvn_traverse_ignore(\*STDOUT, $SVN_PATH, $r);
646 }
647
648 sub graft_branches {
649 my $gr_file = "$GIT_DIR/info/grafts";
650 my ($grafts, $comments) = read_grafts($gr_file);
651 my $gr_sha1;
652
653 if (%$grafts) {
654 # temporarily disable our grafts file to make this idempotent
655 chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file));
656 rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
657 }
658
659 my $l_map = read_url_paths();
660 my @re = map { qr/$_/is } @_opt_m if @_opt_m;
661 unless ($_no_default_regex) {
662 push @re, (qr/\b(?:merge|merging|merged)\s+with\s+([\w\.\-]+)/i,
663 qr/\b(?:merge|merging|merged)\s+([\w\.\-]+)/i,
664 qr/\b(?:from|of)\s+([\w\.\-]+)/i );
665 }
666 foreach my $u (keys %$l_map) {
667 if (@re) {
668 foreach my $p (keys %{$l_map->{$u}}) {
669 graft_merge_msg($grafts,$l_map,$u,$p,@re);
670 }
671 }
672 unless ($_no_graft_copy) {
673 if ($_use_lib) {
674 graft_file_copy_lib($grafts,$l_map,$u);
675 } else {
676 graft_file_copy_cmd($grafts,$l_map,$u);
677 }
678 }
679 }
680 graft_tree_joins($grafts);
681
682 write_grafts($grafts, $comments, $gr_file);
683 unlink "$gr_file~$gr_sha1" if $gr_sha1;
684 }
685
686 sub multi_init {
687 my $url = shift;
688 $_trunk ||= 'trunk';
689 $_trunk =~ s#/+$##;
690 $url =~ s#/+$## if $url;
691 if ($_trunk !~ m#^[a-z\+]+://#) {
692 $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
693 unless ($url) {
694 print STDERR "E: '$_trunk' is not a complete URL ",
695 "and a separate URL is not specified\n";
696 exit 1;
697 }
698 $_trunk = $url . $_trunk;
699 }
700 my $ch_id;
701 if ($GIT_SVN eq 'git-svn') {
702 $ch_id = 1;
703 $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
704 }
705 init_vars();
706 unless (-d $GIT_SVN_DIR) {
707 print "GIT_SVN_ID set to 'trunk' for $_trunk\n" if $ch_id;
708 init($_trunk);
709 sys('git-repo-config', 'svn.trunk', $_trunk);
710 }
711 complete_url_ls_init($url, $_branches, '--branches/-b', '');
712 complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
713 }
714
715 sub multi_fetch {
716 # try to do trunk first, since branches/tags
717 # may be descended from it.
718 if (-e "$GIT_DIR/svn/trunk/info/url") {
719 fetch_child_id('trunk', @_);
720 }
721 rec_fetch('', "$GIT_DIR/svn", @_);
722 }
723
724 sub show_log {
725 my (@args) = @_;
726 my ($r_min, $r_max);
727 my $r_last = -1; # prevent dupes
728 rload_authors() if $_authors;
729 if (defined $TZ) {
730 $ENV{TZ} = $TZ;
731 } else {
732 delete $ENV{TZ};
733 }
734 if (defined $_revision) {
735 if ($_revision =~ /^(\d+):(\d+)$/) {
736 ($r_min, $r_max) = ($1, $2);
737 } elsif ($_revision =~ /^\d+$/) {
738 $r_min = $r_max = $_revision;
739 } else {
740 print STDERR "-r$_revision is not supported, use ",
741 "standard \'git log\' arguments instead\n";
742 exit 1;
743 }
744 }
745
746 my $pid = open(my $log,'-|');
747 defined $pid or croak $!;
748 if (!$pid) {
749 exec(git_svn_log_cmd($r_min,$r_max), @args) or croak $!;
750 }
751 setup_pager();
752 my (@k, $c, $d);
753
754 while (<$log>) {
755 if (/^commit ($sha1_short)/o) {
756 my $cmt = $1;
757 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
758 $r_last = $c->{r};
759 process_commit($c, $r_min, $r_max, \@k) or
760 goto out;
761 }
762 $d = undef;
763 $c = { c => $cmt };
764 } elsif (/^author (.+) (\d+) ([\-\+]?\d+)$/) {
765 get_author_info($c, $1, $2, $3);
766 } elsif (/^(?:tree|parent|committer) /) {
767 # ignore
768 } elsif (/^:\d{6} \d{6} $sha1_short/o) {
769 push @{$c->{raw}}, $_;
770 } elsif (/^[ACRMDT]\t/) {
771 # we could add $SVN_PATH here, but that requires
772 # remote access at the moment (repo_path_split)...
773 s#^([ACRMDT])\t# $1 #;
774 push @{$c->{changed}}, $_;
775 } elsif (/^diff /) {
776 $d = 1;
777 push @{$c->{diff}}, $_;
778 } elsif ($d) {
779 push @{$c->{diff}}, $_;
780 } elsif (/^ (git-svn-id:.+)$/) {
781 ($c->{url}, $c->{r}, undef) = extract_metadata($1);
782 } elsif (s/^ //) {
783 push @{$c->{l}}, $_;
784 }
785 }
786 if ($c && defined $c->{r} && $c->{r} != $r_last) {
787 $r_last = $c->{r};
788 process_commit($c, $r_min, $r_max, \@k);
789 }
790 if (@k) {
791 my $swap = $r_max;
792 $r_max = $r_min;
793 $r_min = $swap;
794 process_commit($_, $r_min, $r_max) foreach reverse @k;
795 }
796 out:
797 close $log;
798 print '-' x72,"\n" unless $_incremental || $_oneline;
799 }
800
801 sub commit_diff_usage {
802 print STDERR "Usage: $0 commit-diff <tree-ish> <tree-ish> [<URL>]\n";
803 exit 1
804 }
805
806 sub commit_diff {
807 if (!$_use_lib) {
808 print STDERR "commit-diff must be used with SVN libraries\n";
809 exit 1;
810 }
811 my $ta = shift or commit_diff_usage();
812 my $tb = shift or commit_diff_usage();
813 if (!eval { $SVN_URL = shift || file_to_s("$GIT_SVN_DIR/info/url") }) {
814 print STDERR "Needed URL or usable git-svn id command-line\n";
815 commit_diff_usage();
816 }
817 if (defined $_message && defined $_file) {
818 print STDERR "Both --message/-m and --file/-F specified ",
819 "for the commit message.\n",
820 "I have no idea what you mean\n";
821 exit 1;
822 }
823 if (defined $_file) {
824 $_message = file_to_s($_file);
825 } else {
826 $_message ||= get_commit_message($tb,
827 "$GIT_DIR/.svn-commit.tmp.$$")->{msg};
828 }
829 my $repo;
830 ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
831 $SVN_LOG ||= libsvn_connect($repo);
832 $SVN ||= libsvn_connect($repo);
833 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
834 my $ed = SVN::Git::Editor->new({ r => $SVN->get_latest_revnum,
835 ra => $SVN_LOG, c => $tb,
836 svn_path => $SVN_PATH
837 },
838 $SVN->get_commit_editor($_message,
839 sub {print "Committed $_[0]\n"},@lock)
840 );
841 my $mods = libsvn_checkout_tree($ta, $tb, $ed);
842 if (@$mods == 0) {
843 print "No changes\n$ta == $tb\n";
844 $ed->abort_edit;
845 } else {
846 $ed->close_edit;
847 }
848 $_message = $_file = undef;
849 }
850
851 ########################### utility functions #########################
852
853 sub cmt_showable {
854 my ($c) = @_;
855 return 1 if defined $c->{r};
856 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
857 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
858 my @msg = safe_qx(qw/git-cat-file commit/, $c->{c});
859 shift @msg while ($msg[0] ne "\n");
860 shift @msg;
861 @{$c->{l}} = grep !/^git-svn-id: /, @msg;
862
863 (undef, $c->{r}, undef) = extract_metadata(
864 (grep(/^git-svn-id: /, @msg))[-1]);
865 }
866 return defined $c->{r};
867 }
868
869 sub git_svn_log_cmd {
870 my ($r_min, $r_max) = @_;
871 my @cmd = (qw/git-log --abbrev-commit --pretty=raw
872 --default/, "refs/remotes/$GIT_SVN");
873 push @cmd, '-r' unless $_non_recursive;
874 push @cmd, qw/--raw --name-status/ if $_verbose;
875 return @cmd unless defined $r_max;
876 if ($r_max == $r_min) {
877 push @cmd, '--max-count=1';
878 if (my $c = revdb_get($REVDB, $r_max)) {
879 push @cmd, $c;
880 }
881 } else {
882 my ($c_min, $c_max);
883 $c_max = revdb_get($REVDB, $r_max);
884 $c_min = revdb_get($REVDB, $r_min);
885 if (defined $c_min && defined $c_max) {
886 if ($r_max > $r_max) {
887 push @cmd, "$c_min..$c_max";
888 } else {
889 push @cmd, "$c_max..$c_min";
890 }
891 } elsif ($r_max > $r_min) {
892 push @cmd, $c_max;
893 } else {
894 push @cmd, $c_min;
895 }
896 }
897 return @cmd;
898 }
899
900 sub fetch_child_id {
901 my $id = shift;
902 print "Fetching $id\n";
903 my $ref = "$GIT_DIR/refs/remotes/$id";
904 defined(my $pid = open my $fh, '-|') or croak $!;
905 if (!$pid) {
906 $_repack = undef;
907 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
908 init_vars();
909 fetch(@_);
910 exit 0;
911 }
912 while (<$fh>) {
913 print $_;
914 check_repack() if (/^r\d+ = $sha1/);
915 }
916 close $fh or croak $?;
917 }
918
919 sub rec_fetch {
920 my ($pfx, $p, @args) = @_;
921 my @dir;
922 foreach (sort <$p/*>) {
923 if (-r "$_/info/url") {
924 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
925 my $id = $pfx . basename $_;
926 next if $id eq 'trunk';
927 fetch_child_id($id, @args);
928 } elsif (-d $_) {
929 push @dir, $_;
930 }
931 }
932 foreach (@dir) {
933 my $x = $_;
934 $x =~ s!^\Q$GIT_DIR\E/svn/!!;
935 rec_fetch($x, $_);
936 }
937 }
938
939 sub complete_url_ls_init {
940 my ($url, $var, $switch, $pfx) = @_;
941 unless ($var) {
942 print STDERR "W: $switch not specified\n";
943 return;
944 }
945 $var =~ s#/+$##;
946 if ($var !~ m#^[a-z\+]+://#) {
947 $var = '/' . $var if ($var !~ m#^/#);
948 unless ($url) {
949 print STDERR "E: '$var' is not a complete URL ",
950 "and a separate URL is not specified\n";
951 exit 1;
952 }
953 $var = $url . $var;
954 }
955 chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
956 : safe_qx(qw/svn ls --non-interactive/, $var));
957 my $old = $GIT_SVN;
958 defined(my $pid = fork) or croak $!;
959 if (!$pid) {
960 foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
961 $u =~ s#/+$##;
962 if ($u !~ m!\Q$var\E/(.+)$!) {
963 print STDERR "W: Unrecognized URL: $u\n";
964 die "This should never happen\n";
965 }
966 # don't try to init already existing refs
967 my $id = $pfx.$1;
968 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
969 init_vars();
970 unless (-d $GIT_SVN_DIR) {
971 print "init $u => $id\n";
972 init($u);
973 }
974 }
975 exit 0;
976 }
977 waitpid $pid, 0;
978 croak $? if $?;
979 my ($n) = ($switch =~ /^--(\w+)/);
980 sys('git-repo-config', "svn.$n", $var);
981 }
982
983 sub common_prefix {
984 my $paths = shift;
985 my %common;
986 foreach (@$paths) {
987 my @tmp = split m#/#, $_;
988 my $p = '';
989 while (my $x = shift @tmp) {
990 $p .= "/$x";
991 $common{$p} ||= 0;
992 $common{$p}++;
993 }
994 }
995 foreach (sort {length $b <=> length $a} keys %common) {
996 if ($common{$_} == @$paths) {
997 return $_;
998 }
999 }
1000 return '';
1001 }
1002
1003 # grafts set here are 'stronger' in that they're based on actual tree
1004 # matches, and won't be deleted from merge-base checking in write_grafts()
1005 sub graft_tree_joins {
1006 my $grafts = shift;
1007 map_tree_joins() if (@_branch_from && !%tree_map);
1008 return unless %tree_map;
1009
1010 git_svn_each(sub {
1011 my $i = shift;
1012 defined(my $pid = open my $fh, '-|') or croak $!;
1013 if (!$pid) {
1014 exec qw/git-rev-list --pretty=raw/,
1015 "refs/remotes/$i" or croak $!;
1016 }
1017 while (<$fh>) {
1018 next unless /^commit ($sha1)$/o;
1019 my $c = $1;
1020 my ($t) = (<$fh> =~ /^tree ($sha1)$/o);
1021 next unless $tree_map{$t};
1022
1023 my $l;
1024 do {
1025 $l = readline $fh;
1026 } until ($l =~ /^committer (?:.+) (\d+) ([\-\+]?\d+)$/);
1027
1028 my ($s, $tz) = ($1, $2);
1029 if ($tz =~ s/^\+//) {
1030 $s += tz_to_s_offset($tz);
1031 } elsif ($tz =~ s/^\-//) {
1032 $s -= tz_to_s_offset($tz);
1033 }
1034
1035 my ($url_a, $r_a, $uuid_a) = cmt_metadata($c);
1036
1037 foreach my $p (@{$tree_map{$t}}) {
1038 next if $p eq $c;
1039 my $mb = eval {
1040 safe_qx('git-merge-base', $c, $p)
1041 };
1042 next unless ($@ || $?);
1043 if (defined $r_a) {
1044 # see if SVN says it's a relative
1045 my ($url_b, $r_b, $uuid_b) =
1046 cmt_metadata($p);
1047 next if (defined $url_b &&
1048 defined $url_a &&
1049 ($url_a eq $url_b) &&
1050 ($uuid_a eq $uuid_b));
1051 if ($uuid_a eq $uuid_b) {
1052 if ($r_b < $r_a) {
1053 $grafts->{$c}->{$p} = 2;
1054 next;
1055 } elsif ($r_b > $r_a) {
1056 $grafts->{$p}->{$c} = 2;
1057 next;
1058 }
1059 }
1060 }
1061 my $ct = get_commit_time($p);
1062 if ($ct < $s) {
1063 $grafts->{$c}->{$p} = 2;
1064 } elsif ($ct > $s) {
1065 $grafts->{$p}->{$c} = 2;
1066 }
1067 # what should we do when $ct == $s ?
1068 }
1069 }
1070 close $fh or croak $?;
1071 });
1072 }
1073
1074 # this isn't funky-filename safe, but good enough for now...
1075 sub graft_file_copy_cmd {
1076 my ($grafts, $l_map, $u) = @_;
1077 my $paths = $l_map->{$u};
1078 my $pfx = common_prefix([keys %$paths]);
1079 $SVN_URL ||= $u.$pfx;
1080 my $pid = open my $fh, '-|';
1081 defined $pid or croak $!;
1082 unless ($pid) {
1083 my @exec = qw/svn log -v/;
1084 push @exec, "-r$_revision" if defined $_revision;
1085 exec @exec, $u.$pfx or croak $!;
1086 }
1087 my ($r, $mp) = (undef, undef);
1088 while (<$fh>) {
1089 chomp;
1090 if (/^\-{72}$/) {
1091 $mp = $r = undef;
1092 } elsif (/^r(\d+) \| /) {
1093 $r = $1 unless defined $r;
1094 } elsif (/^Changed paths:/) {
1095 $mp = 1;
1096 } elsif ($mp && m#^ [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
1097 my ($p1, $p0, $r0) = ($1, $2, $3);
1098 my $c = find_graft_path_commit($paths, $p1, $r);
1099 next unless $c;
1100 find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
1101 }
1102 }
1103 }
1104
1105 sub graft_file_copy_lib {
1106 my ($grafts, $l_map, $u) = @_;
1107 my $tree_paths = $l_map->{$u};
1108 my $pfx = common_prefix([keys %$tree_paths]);
1109 my ($repo, $path) = repo_path_split($u.$pfx);
1110 $SVN_LOG ||= libsvn_connect($repo);
1111 $SVN ||= libsvn_connect($repo);
1112
1113 my ($base, $head) = libsvn_parse_revision();
1114 my $inc = 1000;
1115 my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
1116 my $eh = $SVN::Error::handler;
1117 $SVN::Error::handler = \&libsvn_skip_unknown_revs;
1118 while (1) {
1119 my $pool = SVN::Pool->new;
1120 libsvn_get_log($SVN_LOG, "/$path", $min, $max, 0, 1, 1,
1121 sub {
1122 libsvn_graft_file_copies($grafts, $tree_paths,
1123 $path, @_);
1124 }, $pool);
1125 $pool->clear;
1126 last if ($max >= $head);
1127 $min = $max + 1;
1128 $max += $inc;
1129 $max = $head if ($max > $head);
1130 }
1131 $SVN::Error::handler = $eh;
1132 }
1133
1134 sub process_merge_msg_matches {
1135 my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
1136 my (@strong, @weak);
1137 foreach (@matches) {
1138 # merging with ourselves is not interesting
1139 next if $_ eq $p;
1140 if ($l_map->{$u}->{$_}) {
1141 push @strong, $_;
1142 } else {
1143 push @weak, $_;
1144 }
1145 }
1146 foreach my $w (@weak) {
1147 last if @strong;
1148 # no exact match, use branch name as regexp.
1149 my $re = qr/\Q$w\E/i;
1150 foreach (keys %{$l_map->{$u}}) {
1151 if (/$re/) {
1152 push @strong, $l_map->{$u}->{$_};
1153 last;
1154 }
1155 }
1156 last if @strong;
1157 $w = basename($w);
1158 $re = qr/\Q$w\E/i;
1159 foreach (keys %{$l_map->{$u}}) {
1160 if (/$re/) {
1161 push @strong, $l_map->{$u}->{$_};
1162 last;
1163 }
1164 }
1165 }
1166 my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
1167 \s(?:[a-f\d\-]+)$/xsm);
1168 unless (defined $rev) {
1169 ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
1170 \@(?:[a-f\d\-]+)/xsm);
1171 return unless defined $rev;
1172 }
1173 foreach my $m (@strong) {
1174 my ($r0, $s0) = find_rev_before($rev, $m, 1);
1175 $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
1176 }
1177 }
1178
1179 sub graft_merge_msg {
1180 my ($grafts, $l_map, $u, $p, @re) = @_;
1181
1182 my $x = $l_map->{$u}->{$p};
1183 my $rl = rev_list_raw($x);
1184 while (my $c = next_rev_list_entry($rl)) {
1185 foreach my $re (@re) {
1186 my (@br) = ($c->{m} =~ /$re/g);
1187 next unless @br;
1188 process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
1189 }
1190 }
1191 }
1192
1193 sub read_uuid {
1194 return if $SVN_UUID;
1195 if ($_use_lib) {
1196 my $pool = SVN::Pool->new;
1197 $SVN_UUID = $SVN->get_uuid($pool);
1198 $pool->clear;
1199 } else {
1200 my $info = shift || svn_info('.');
1201 $SVN_UUID = $info->{'Repository UUID'} or
1202 croak "Repository UUID unreadable\n";
1203 }
1204 }
1205
1206 sub quiet_run {
1207 my $pid = fork;
1208 defined $pid or croak $!;
1209 if (!$pid) {
1210 open my $null, '>', '/dev/null' or croak $!;
1211 open STDERR, '>&', $null or croak $!;
1212 open STDOUT, '>&', $null or croak $!;
1213 exec @_ or croak $!;
1214 }
1215 waitpid $pid, 0;
1216 return $?;
1217 }
1218
1219 sub repo_path_split {
1220 my $full_url = shift;
1221 $full_url =~ s#/+$##;
1222
1223 foreach (@repo_path_split_cache) {
1224 if ($full_url =~ s#$_##) {
1225 my $u = $1;
1226 $full_url =~ s#^/+##;
1227 return ($u, $full_url);
1228 }
1229 }
1230
1231 if ($_use_lib) {
1232 my $tmp = libsvn_connect($full_url);
1233 my $url = $tmp->get_repos_root;
1234 $full_url =~ s#^\Q$url\E/*##;
1235 push @repo_path_split_cache, qr/^(\Q$url\E)/;
1236 return ($url, $full_url);
1237 } else {
1238 my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1239 $path =~ s#^/+##;
1240 my @paths = split(m#/+#, $path);
1241 while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1242 my $n = shift @paths || last;
1243 $url .= "/$n";
1244 }
1245 push @repo_path_split_cache, qr/^(\Q$url\E)/;
1246 $path = join('/',@paths);
1247 return ($url, $path);
1248 }
1249 }
1250
1251 sub setup_git_svn {
1252 defined $SVN_URL or croak "SVN repository location required\n";
1253 unless (-d $GIT_DIR) {
1254 croak "GIT_DIR=$GIT_DIR does not exist!\n";
1255 }
1256 mkpath([$GIT_SVN_DIR]);
1257 mkpath(["$GIT_SVN_DIR/info"]);
1258 open my $fh, '>>',$REVDB or croak $!;
1259 close $fh;
1260 s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1261
1262 }
1263
1264 sub assert_svn_wc_clean {
1265 return if $_use_lib;
1266 my ($svn_rev) = @_;
1267 croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1268 my $lcr = svn_info('.')->{'Last Changed Rev'};
1269 if ($svn_rev != $lcr) {
1270 print STDERR "Checking for copy-tree ... ";
1271 my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1272 "-r$lcr:$svn_rev")));
1273 if (@diff) {
1274 croak "Nope! Expected r$svn_rev, got r$lcr\n";
1275 } else {
1276 print STDERR "OK!\n";
1277 }
1278 }
1279 my @status = grep(!/^Performing status on external/,(`svn status`));
1280 @status = grep(!/^\s*$/,@status);
1281 @status = grep(!/^X/,@status) if $_no_ignore_ext;
1282 if (scalar @status) {
1283 print STDERR "Tree ($SVN_WC) is not clean:\n";
1284 print STDERR $_ foreach @status;
1285 croak;
1286 }
1287 }
1288
1289 sub get_tree_from_treeish {
1290 my ($treeish) = @_;
1291 croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1292 chomp(my $type = `git-cat-file -t $treeish`);
1293 my $expected;
1294 while ($type eq 'tag') {
1295 chomp(($treeish, $type) = `git-cat-file tag $treeish`);
1296 }
1297 if ($type eq 'commit') {
1298 $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
1299 ($expected) = ($expected =~ /^tree ($sha1)$/);
1300 die "Unable to get tree from $treeish\n" unless $expected;
1301 } elsif ($type eq 'tree') {
1302 $expected = $treeish;
1303 } else {
1304 die "$treeish is a $type, expected tree, tag or commit\n";
1305 }
1306 return $expected;
1307 }
1308
1309 sub assert_tree {
1310 return if $_use_lib;
1311 my ($treeish) = @_;
1312 my $expected = get_tree_from_treeish($treeish);
1313
1314 my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1315 if (-e $tmpindex) {
1316 unlink $tmpindex or croak $!;
1317 }
1318 my $old_index = set_index($tmpindex);
1319 index_changes(1);
1320 chomp(my $tree = `git-write-tree`);
1321 restore_index($old_index);
1322 if ($tree ne $expected) {
1323 croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1324 }
1325 unlink $tmpindex;
1326 }
1327
1328 sub parse_diff_tree {
1329 my $diff_fh = shift;
1330 local $/ = "\0";
1331 my $state = 'meta';
1332 my @mods;
1333 while (<$diff_fh>) {
1334 chomp $_; # this gets rid of the trailing "\0"
1335 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1336 $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1337 push @mods, { mode_a => $1, mode_b => $2,
1338 sha1_b => $3, chg => $4 };
1339 if ($4 =~ /^(?:C|R)$/) {
1340 $state = 'file_a';
1341 } else {
1342 $state = 'file_b';
1343 }
1344 } elsif ($state eq 'file_a') {
1345 my $x = $mods[$#mods] or croak "Empty array\n";
1346 if ($x->{chg} !~ /^(?:C|R)$/) {
1347 croak "Error parsing $_, $x->{chg}\n";
1348 }
1349 $x->{file_a} = $_;
1350 $state = 'file_b';
1351 } elsif ($state eq 'file_b') {
1352 my $x = $mods[$#mods] or croak "Empty array\n";
1353 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1354 croak "Error parsing $_, $x->{chg}\n";
1355 }
1356 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1357 croak "Error parsing $_, $x->{chg}\n";
1358 }
1359 $x->{file_b} = $_;
1360 $state = 'meta';
1361 } else {
1362 croak "Error parsing $_\n";
1363 }
1364 }
1365 close $diff_fh or croak $?;
1366
1367 return \@mods;
1368 }
1369
1370 sub svn_check_prop_executable {
1371 my $m = shift;
1372 return if -l $m->{file_b};
1373 if ($m->{mode_b} =~ /755$/) {
1374 chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1375 if ($m->{mode_a} !~ /755$/) {
1376 sys(qw(svn propset svn:executable 1), $m->{file_b});
1377 }
1378 -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1379 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1380 sys(qw(svn propdel svn:executable), $m->{file_b});
1381 chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1382 -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1383 }
1384 }
1385
1386 sub svn_ensure_parent_path {
1387 my $dir_b = dirname(shift);
1388 svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1389 mkpath([$dir_b]) unless (-d $dir_b);
1390 sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1391 }
1392
1393 sub precommit_check {
1394 my $mods = shift;
1395 my (%rm_file, %rmdir_check, %added_check);
1396
1397 my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1398 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1399 if ($m->{chg} eq 'R') {
1400 if (-d $m->{file_b}) {
1401 err_dir_to_file("$m->{file_a} => $m->{file_b}");
1402 }
1403 # dir/$file => dir/file/$file
1404 my $dirname = dirname($m->{file_b});
1405 while ($dirname ne File::Spec->curdir) {
1406 if ($dirname ne $m->{file_a}) {
1407 $dirname = dirname($dirname);
1408 next;
1409 }
1410 err_file_to_dir("$m->{file_a} => $m->{file_b}");
1411 }
1412 # baz/zzz => baz (baz is a file)
1413 $dirname = dirname($m->{file_a});
1414 while ($dirname ne File::Spec->curdir) {
1415 if ($dirname ne $m->{file_b}) {
1416 $dirname = dirname($dirname);
1417 next;
1418 }
1419 err_dir_to_file("$m->{file_a} => $m->{file_b}");
1420 }
1421 }
1422 if ($m->{chg} =~ /^(D|R)$/) {
1423 my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1424 $rm_file{ $m->{$t} } = 1;
1425 my $dirname = dirname( $m->{$t} );
1426 my $basename = basename( $m->{$t} );
1427 $rmdir_check{$dirname}->{$basename} = 1;
1428 } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1429 if (-d $m->{file_b}) {
1430 err_dir_to_file($m->{file_b});
1431 }
1432 my $dirname = dirname( $m->{file_b} );
1433 my $basename = basename( $m->{file_b} );
1434 $added_check{$dirname}->{$basename} = 1;
1435 while ($dirname ne File::Spec->curdir) {
1436 if ($rm_file{$dirname}) {
1437 err_file_to_dir($m->{file_b});
1438 }
1439 $dirname = dirname $dirname;
1440 }
1441 }
1442 }
1443 return (\%rmdir_check, \%added_check);
1444
1445 sub err_dir_to_file {
1446 my $file = shift;
1447 print STDERR "Node change from directory to file ",
1448 "is not supported by Subversion: ",$file,"\n";
1449 exit 1;
1450 }
1451 sub err_file_to_dir {
1452 my $file = shift;
1453 print STDERR "Node change from file to directory ",
1454 "is not supported by Subversion: ",$file,"\n";
1455 exit 1;
1456 }
1457 }
1458
1459
1460 sub get_diff {
1461 my ($from, $treeish) = @_;
1462 assert_tree($from);
1463 print "diff-tree $from $treeish\n";
1464 my $pid = open my $diff_fh, '-|';
1465 defined $pid or croak $!;
1466 if ($pid == 0) {
1467 my @diff_tree = qw(git-diff-tree -z -r);
1468 if ($_cp_similarity) {
1469 push @diff_tree, "-C$_cp_similarity";
1470 } else {
1471 push @diff_tree, '-C';
1472 }
1473 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1474 push @diff_tree, "-l$_l" if defined $_l;
1475 exec(@diff_tree, $from, $treeish) or croak $!;
1476 }
1477 return parse_diff_tree($diff_fh);
1478 }
1479
1480 sub svn_checkout_tree {
1481 my ($from, $treeish) = @_;
1482 my $mods = get_diff($from->{commit}, $treeish);
1483 return $mods unless (scalar @$mods);
1484 my ($rm, $add) = precommit_check($mods);
1485
1486 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1487 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1488 if ($m->{chg} eq 'C') {
1489 svn_ensure_parent_path( $m->{file_b} );
1490 sys(qw(svn cp), $m->{file_a}, $m->{file_b});
1491 apply_mod_line_blob($m);
1492 svn_check_prop_executable($m);
1493 } elsif ($m->{chg} eq 'D') {
1494 sys(qw(svn rm --force), $m->{file_b});
1495 } elsif ($m->{chg} eq 'R') {
1496 svn_ensure_parent_path( $m->{file_b} );
1497 sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1498 apply_mod_line_blob($m);
1499 svn_check_prop_executable($m);
1500 } elsif ($m->{chg} eq 'M') {
1501 apply_mod_line_blob($m);
1502 svn_check_prop_executable($m);
1503 } elsif ($m->{chg} eq 'T') {
1504 svn_check_prop_executable($m);
1505 apply_mod_line_blob($m);
1506 if ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
1507 sys(qw(svn propdel svn:special), $m->{file_b});
1508 } else {
1509 sys(qw(svn propset svn:special *),$m->{file_b});
1510 }
1511 } elsif ($m->{chg} eq 'A') {
1512 svn_ensure_parent_path( $m->{file_b} );
1513 apply_mod_line_blob($m);
1514 sys(qw(svn add), $m->{file_b});
1515 svn_check_prop_executable($m);
1516 } else {
1517 croak "Invalid chg: $m->{chg}\n";
1518 }
1519 }
1520
1521 assert_tree($treeish);
1522 if ($_rmdir) { # remove empty directories
1523 handle_rmdir($rm, $add);
1524 }
1525 assert_tree($treeish);
1526 return $mods;
1527 }
1528
1529 sub libsvn_checkout_tree {
1530 my ($from, $treeish, $ed) = @_;
1531 my $mods = get_diff($from, $treeish);
1532 return $mods unless (scalar @$mods);
1533 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1534 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1535 my $f = $m->{chg};
1536 if (defined $o{$f}) {
1537 $ed->$f($m, $_q);
1538 } else {
1539 croak "Invalid change type: $f\n";
1540 }
1541 }
1542 $ed->rmdirs($_q) if $_rmdir;
1543 return $mods;
1544 }
1545
1546 # svn ls doesn't work with respect to the current working tree, but what's
1547 # in the repository. There's not even an option for it... *sigh*
1548 # (added files don't show up and removed files remain in the ls listing)
1549 sub svn_ls_current {
1550 my ($dir, $rm, $add) = @_;
1551 chomp(my @ls = safe_qx('svn','ls',$dir));
1552 my @ret = ();
1553 foreach (@ls) {
1554 s#/$##; # trailing slashes are evil
1555 push @ret, $_ unless $rm->{$dir}->{$_};
1556 }
1557 if (exists $add->{$dir}) {
1558 push @ret, keys %{$add->{$dir}};
1559 }
1560 return \@ret;
1561 }
1562
1563 sub handle_rmdir {
1564 my ($rm, $add) = @_;
1565
1566 foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1567 my $ls = svn_ls_current($dir, $rm, $add);
1568 next if (scalar @$ls);
1569 sys(qw(svn rm --force),$dir);
1570
1571 my $dn = dirname $dir;
1572 $rm->{ $dn }->{ basename $dir } = 1;
1573 $ls = svn_ls_current($dn, $rm, $add);
1574 while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1575 sys(qw(svn rm --force),$dn);
1576 $dir = basename $dn;
1577 $dn = dirname $dn;
1578 $rm->{ $dn }->{ $dir } = 1;
1579 $ls = svn_ls_current($dn, $rm, $add);
1580 }
1581 }
1582 }
1583
1584 sub get_commit_message {
1585 my ($commit, $commit_msg) = (@_);
1586 my %log_msg = ( msg => '' );
1587 open my $msg, '>', $commit_msg or croak $!;
1588
1589 chomp(my $type = `git-cat-file -t $commit`);
1590 if ($type eq 'commit' || $type eq 'tag') {
1591 my $pid = open my $msg_fh, '-|';
1592 defined $pid or croak $!;
1593
1594 if ($pid == 0) {
1595 exec('git-cat-file', $type, $commit) or croak $!;
1596 }
1597 my $in_msg = 0;
1598 while (<$msg_fh>) {
1599 if (!$in_msg) {
1600 $in_msg = 1 if (/^\s*$/);
1601 } elsif (/^git-svn-id: /) {
1602 # skip this, we regenerate the correct one
1603 # on re-fetch anyways
1604 } else {
1605 print $msg $_ or croak $!;
1606 }
1607 }
1608 close $msg_fh or croak $?;
1609 }
1610 close $msg or croak $!;
1611
1612 if ($_edit || ($type eq 'tree')) {
1613 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1614 system($editor, $commit_msg);
1615 }
1616
1617 # file_to_s removes all trailing newlines, so just use chomp() here:
1618 open $msg, '<', $commit_msg or croak $!;
1619 { local $/; chomp($log_msg{msg} = <$msg>); }
1620 close $msg or croak $!;
1621
1622 return \%log_msg;
1623 }
1624
1625 sub set_svn_commit_env {
1626 if (defined $LC_ALL) {
1627 $ENV{LC_ALL} = $LC_ALL;
1628 } else {
1629 delete $ENV{LC_ALL};
1630 }
1631 }
1632
1633 sub svn_commit_tree {
1634 my ($last, $commit) = @_;
1635 my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1636 my $log_msg = get_commit_message($commit, $commit_msg);
1637 my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1638 print "Committing $commit: $oneline\n";
1639
1640 set_svn_commit_env();
1641 my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1642 $ENV{LC_ALL} = 'C';
1643 unlink $commit_msg;
1644 my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1645 if (!defined $committed) {
1646 my $out = join("\n",@ci_output);
1647 print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1648 $out, "\n\nAssuming English locale...";
1649 ($committed) = ($out =~ /^Committed revision \d+\./sm);
1650 defined $committed or die " FAILED!\n",
1651 "Commit output failed to parse committed revision!\n",
1652 print STDERR " OK\n";
1653 }
1654
1655 my @svn_up = qw(svn up);
1656 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1657 if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1658 push @svn_up, "-r$committed";
1659 sys(@svn_up);
1660 my $info = svn_info('.');
1661 my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1662 if ($info->{'Last Changed Rev'} != $committed) {
1663 croak "$info->{'Last Changed Rev'} != $committed\n"
1664 }
1665 my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1666 /(\d{4})\-(\d\d)\-(\d\d)\s
1667 (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1668 or croak "Failed to parse date: $date\n";
1669 $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1670 $log_msg->{author} = $info->{'Last Changed Author'};
1671 $log_msg->{revision} = $committed;
1672 $log_msg->{msg} .= "\n";
1673 $log_msg->{parents} = [ $last->{commit} ];
1674 $log_msg->{commit} = git_commit($log_msg, $commit);
1675 return $log_msg;
1676 }
1677 # resync immediately
1678 push @svn_up, "-r$last->{revision}";
1679 sys(@svn_up);
1680 return fetch("$committed=$commit");
1681 }
1682
1683 sub rev_list_raw {
1684 my (@args) = @_;
1685 my $pid = open my $fh, '-|';
1686 defined $pid or croak $!;
1687 if (!$pid) {
1688 exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1689 }
1690 return { fh => $fh, t => { } };
1691 }
1692
1693 sub next_rev_list_entry {
1694 my $rl = shift;
1695 my $fh = $rl->{fh};
1696 my $x = $rl->{t};
1697 while (<$fh>) {
1698 if (/^commit ($sha1)$/o) {
1699 if ($x->{c}) {
1700 $rl->{t} = { c => $1 };
1701 return $x;
1702 } else {
1703 $x->{c} = $1;
1704 }
1705 } elsif (/^parent ($sha1)$/o) {
1706 $x->{p}->{$1} = 1;
1707 } elsif (s/^ //) {
1708 $x->{m} ||= '';
1709 $x->{m} .= $_;
1710 }
1711 }
1712 return ($x != $rl->{t}) ? $x : undef;
1713 }
1714
1715 # read the entire log into a temporary file (which is removed ASAP)
1716 # and store the file handle + parser state
1717 sub svn_log_raw {
1718 my (@log_args) = @_;
1719 my $log_fh = IO::File->new_tmpfile or croak $!;
1720 my $pid = fork;
1721 defined $pid or croak $!;
1722 if (!$pid) {
1723 open STDOUT, '>&', $log_fh or croak $!;
1724 exec (qw(svn log), @log_args) or croak $!
1725 }
1726 waitpid $pid, 0;
1727 croak $? if $?;
1728 seek $log_fh, 0, 0 or croak $!;
1729 return { state => 'sep', fh => $log_fh };
1730 }
1731
1732 sub next_log_entry {
1733 my $log = shift; # retval of svn_log_raw()
1734 my $ret = undef;
1735 my $fh = $log->{fh};
1736
1737 while (<$fh>) {
1738 chomp;
1739 if (/^\-{72}$/) {
1740 if ($log->{state} eq 'msg') {
1741 if ($ret->{lines}) {
1742 $ret->{msg} .= $_."\n";
1743 unless(--$ret->{lines}) {
1744 $log->{state} = 'sep';
1745 }
1746 } else {
1747 croak "Log parse error at: $_\n",
1748 $ret->{revision},
1749 "\n";
1750 }
1751 next;
1752 }
1753 if ($log->{state} ne 'sep') {
1754 croak "Log parse error at: $_\n",
1755 "state: $log->{state}\n",
1756 $ret->{revision},
1757 "\n";
1758 }
1759 $log->{state} = 'rev';
1760
1761 # if we have an empty log message, put something there:
1762 if ($ret) {
1763 $ret->{msg} ||= "\n";
1764 delete $ret->{lines};
1765 return $ret;
1766 }
1767 next;
1768 }
1769 if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1770 my $rev = $1;
1771 my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1772 ($lines) = ($lines =~ /(\d+)/);
1773 $date = '1970-01-01 00:00:00 +0000'
1774 if ($_ignore_nodate && $date eq '(no date)');
1775 my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1776 /(\d{4})\-(\d\d)\-(\d\d)\s
1777 (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1778 or croak "Failed to parse date: $date\n";
1779 $ret = { revision => $rev,
1780 date => "$tz $Y-$m-$d $H:$M:$S",
1781 author => $author,
1782 lines => $lines,
1783 msg => '' };
1784 if (defined $_authors && ! defined $users{$author}) {
1785 die "Author: $author not defined in ",
1786 "$_authors file\n";
1787 }
1788 $log->{state} = 'msg_start';
1789 next;
1790 }
1791 # skip the first blank line of the message:
1792 if ($log->{state} eq 'msg_start' && /^$/) {
1793 $log->{state} = 'msg';
1794 } elsif ($log->{state} eq 'msg') {
1795 if ($ret->{lines}) {
1796 $ret->{msg} .= $_."\n";
1797 unless (--$ret->{lines}) {
1798 $log->{state} = 'sep';
1799 }
1800 } else {
1801 croak "Log parse error at: $_\n",
1802 $ret->{revision},"\n";
1803 }
1804 }
1805 }
1806 return $ret;
1807 }
1808
1809 sub svn_info {
1810 my $url = shift || $SVN_URL;
1811
1812 my $pid = open my $info_fh, '-|';
1813 defined $pid or croak $!;
1814
1815 if ($pid == 0) {
1816 exec(qw(svn info),$url) or croak $!;
1817 }
1818
1819 my $ret = {};
1820 # only single-lines seem to exist in svn info output
1821 while (<$info_fh>) {
1822 chomp $_;
1823 if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1824 $ret->{$1} = $2;
1825 push @{$ret->{-order}}, $1;
1826 }
1827 }
1828 close $info_fh or croak $?;
1829 return $ret;
1830 }
1831
1832 sub sys { system(@_) == 0 or croak $? }
1833
1834 sub do_update_index {
1835 my ($z_cmd, $cmd, $no_text_base) = @_;
1836
1837 my $z = open my $p, '-|';
1838 defined $z or croak $!;
1839 unless ($z) { exec @$z_cmd or croak $! }
1840
1841 my $pid = open my $ui, '|-';
1842 defined $pid or croak $!;
1843 unless ($pid) {
1844 exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1845 }
1846 local $/ = "\0";
1847 while (my $x = <$p>) {
1848 chomp $x;
1849 if (!$no_text_base && lstat $x && ! -l _ &&
1850 svn_propget_base('svn:keywords', $x)) {
1851 my $mode = -x _ ? 0755 : 0644;
1852 my ($v,$d,$f) = File::Spec->splitpath($x);
1853 my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1854 'text-base',"$f.svn-base");
1855 $tb =~ s#^/##;
1856 unless (-f $tb) {
1857 $tb = File::Spec->catfile($d, '.svn',
1858 'text-base',"$f.svn-base");
1859 $tb =~ s#^/##;
1860 }
1861 my @s = stat($x);
1862 unlink $x or croak $!;
1863 copy($tb, $x);
1864 chmod(($mode &~ umask), $x) or croak $!;
1865 utime $s[8], $s[9], $x;
1866 }
1867 print $ui $x,"\0";
1868 }
1869 close $ui or croak $?;
1870 }
1871
1872 sub index_changes {
1873 return if $_use_lib;
1874
1875 if (!-f "$GIT_SVN_DIR/info/exclude") {
1876 open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1877 print $fd '.svn',"\n";
1878 close $fd or croak $!;
1879 }
1880 my $no_text_base = shift;
1881 do_update_index([qw/git-diff-files --name-only -z/],
1882 'remove',
1883 $no_text_base);
1884 do_update_index([qw/git-ls-files -z --others/,
1885 "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1886 'add',
1887 $no_text_base);
1888 }
1889
1890 sub s_to_file {
1891 my ($str, $file, $mode) = @_;
1892 open my $fd,'>',$file or croak $!;
1893 print $fd $str,"\n" or croak $!;
1894 close $fd or croak $!;
1895 chmod ($mode &~ umask, $file) if (defined $mode);
1896 }
1897
1898 sub file_to_s {
1899 my $file = shift;
1900 open my $fd,'<',$file or croak "$!: file: $file\n";
1901 local $/;
1902 my $ret = <$fd>;
1903 close $fd or croak $!;
1904 $ret =~ s/\s*$//s;
1905 return $ret;
1906 }
1907
1908 sub assert_revision_unknown {
1909 my $r = shift;
1910 if (my $c = revdb_get($REVDB, $r)) {
1911 croak "$r = $c already exists! Why are we refetching it?";
1912 }
1913 }
1914
1915 sub trees_eq {
1916 my ($x, $y) = @_;
1917 my @x = safe_qx('git-cat-file','commit',$x);
1918 my @y = safe_qx('git-cat-file','commit',$y);
1919 if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
1920 || $y[0] !~ /^tree $sha1\n$/) {
1921 print STDERR "Trees not equal: $y[0] != $x[0]\n";
1922 return 0
1923 }
1924 return 1;
1925 }
1926
1927 sub git_commit {
1928 my ($log_msg, @parents) = @_;
1929 assert_revision_unknown($log_msg->{revision});
1930 map_tree_joins() if (@_branch_from && !%tree_map);
1931
1932 my (@tmp_parents, @exec_parents, %seen_parent);
1933 if (my $lparents = $log_msg->{parents}) {
1934 @tmp_parents = @$lparents
1935 }
1936 # commit parents can be conditionally bound to a particular
1937 # svn revision via: "svn_revno=commit_sha1", filter them out here:
1938 foreach my $p (@parents) {
1939 next unless defined $p;
1940 if ($p =~ /^(\d+)=($sha1_short)$/o) {
1941 if ($1 == $log_msg->{revision}) {
1942 push @tmp_parents, $2;
1943 }
1944 } else {
1945 push @tmp_parents, $p if $p =~ /$sha1_short/o;
1946 }
1947 }
1948 my $tree = $log_msg->{tree};
1949 if (!defined $tree) {
1950 my $index = set_index($GIT_SVN_INDEX);
1951 index_changes();
1952 chomp($tree = `git-write-tree`);
1953 croak $? if $?;
1954 restore_index($index);
1955 }
1956
1957 # just in case we clobber the existing ref, we still want that ref
1958 # as our parent:
1959 if (my $cur = eval { file_to_s("$GIT_DIR/refs/remotes/$GIT_SVN") }) {
1960 push @tmp_parents, $cur;
1961 }
1962
1963 if (exists $tree_map{$tree}) {
1964 foreach my $p (@{$tree_map{$tree}}) {
1965 my $skip;
1966 foreach (@tmp_parents) {
1967 # see if a common parent is found
1968 my $mb = eval {
1969 safe_qx('git-merge-base', $_, $p)
1970 };
1971 next if ($@ || $?);
1972 $skip = 1;
1973 last;
1974 }
1975 next if $skip;
1976 my ($url_p, $r_p, $uuid_p) = cmt_metadata($p);
1977 next if (($SVN_UUID eq $uuid_p) &&
1978 ($log_msg->{revision} > $r_p));
1979 next if (defined $url_p && defined $SVN_URL &&
1980 ($SVN_UUID eq $uuid_p) &&
1981 ($url_p eq $SVN_URL));
1982 push @tmp_parents, $p;
1983 }
1984 }
1985 foreach (@tmp_parents) {
1986 next if $seen_parent{$_};
1987 $seen_parent{$_} = 1;
1988 push @exec_parents, $_;
1989 # MAXPARENT is defined to 16 in commit-tree.c:
1990 last if @exec_parents > 16;
1991 }
1992
1993 set_commit_env($log_msg);
1994 my @exec = ('git-commit-tree', $tree);
1995 push @exec, '-p', $_ foreach @exec_parents;
1996 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1997 or croak $!;
1998 print $msg_fh $log_msg->{msg} or croak $!;
1999 unless ($_no_metadata) {
2000 print $msg_fh "\ngit-svn-id: $SVN_URL\@$log_msg->{revision}",
2001 " $SVN_UUID\n" or croak $!;
2002 }
2003 $msg_fh->flush == 0 or croak $!;
2004 close $msg_fh or croak $!;
2005 chomp(my $commit = do { local $/; <$out_fh> });
2006 close $out_fh or croak $!;
2007 waitpid $pid, 0;
2008 croak $? if $?;
2009 if ($commit !~ /^$sha1$/o) {
2010 die "Failed to commit, invalid sha1: $commit\n";
2011 }
2012 sys('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
2013 revdb_set($REVDB, $log_msg->{revision}, $commit);
2014
2015 # this output is read via pipe, do not change:
2016 print "r$log_msg->{revision} = $commit\n";
2017 check_repack();
2018 return $commit;
2019 }
2020
2021 sub check_repack {
2022 if ($_repack && (--$_repack_nr == 0)) {
2023 $_repack_nr = $_repack;
2024 sys("git repack $_repack_flags");
2025 }
2026 }
2027
2028 sub set_commit_env {
2029 my ($log_msg) = @_;
2030 my $author = $log_msg->{author};
2031 if (!defined $author || length $author == 0) {
2032 $author = '(no author)';
2033 }
2034 my ($name,$email) = defined $users{$author} ? @{$users{$author}}
2035 : ($author,"$author\@$SVN_UUID");
2036 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
2037 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
2038 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
2039 }
2040
2041 sub apply_mod_line_blob {
2042 my $m = shift;
2043 if ($m->{mode_b} =~ /^120/) {
2044 blob_to_symlink($m->{sha1_b}, $m->{file_b});
2045 } else {
2046 blob_to_file($m->{sha1_b}, $m->{file_b});
2047 }
2048 }
2049
2050 sub blob_to_symlink {
2051 my ($blob, $link) = @_;
2052 defined $link or croak "\$link not defined!\n";
2053 croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2054 if (-l $link || -f _) {
2055 unlink $link or croak $!;
2056 }
2057
2058 my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
2059 symlink $dest, $link or croak $!;
2060 }
2061
2062 sub blob_to_file {
2063 my ($blob, $file) = @_;
2064 defined $file or croak "\$file not defined!\n";
2065 croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
2066 if (-l $file || -f _) {
2067 unlink $file or croak $!;
2068 }
2069
2070 open my $blob_fh, '>', $file or croak "$!: $file\n";
2071 my $pid = fork;
2072 defined $pid or croak $!;
2073
2074 if ($pid == 0) {
2075 open STDOUT, '>&', $blob_fh or croak $!;
2076 exec('git-cat-file','blob',$blob) or croak $!;
2077 }
2078 waitpid $pid, 0;
2079 croak $? if $?;
2080
2081 close $blob_fh or croak $!;
2082 }
2083
2084 sub safe_qx {
2085 my $pid = open my $child, '-|';
2086 defined $pid or croak $!;
2087 if ($pid == 0) {
2088 exec(@_) or croak $!;
2089 }
2090 my @ret = (<$child>);
2091 close $child or croak $?;
2092 die $? if $?; # just in case close didn't error out
2093 return wantarray ? @ret : join('',@ret);
2094 }
2095
2096 sub svn_compat_check {
2097 if ($_follow_parent) {
2098 print STDERR 'E: --follow-parent functionality is only ',
2099 "available when SVN libraries are used\n";
2100 exit 1;
2101 }
2102 my @co_help = safe_qx(qw(svn co -h));
2103 unless (grep /ignore-externals/,@co_help) {
2104 print STDERR "W: Installed svn version does not support ",
2105 "--ignore-externals\n";
2106 $_no_ignore_ext = 1;
2107 }
2108 if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
2109 $_svn_co_url_revs = 1;
2110 }
2111 if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
2112 $_svn_pg_peg_revs = 1;
2113 }
2114
2115 # I really, really hope nobody hits this...
2116 unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
2117 print STDERR <<'';
2118 W: The installed svn version does not support the --stop-on-copy flag in
2119 the log command.
2120 Lets hope the directory you're tracking is not a branch or tag
2121 and was never moved within the repository...
2122
2123 $_no_stop_copy = 1;
2124 }
2125 }
2126
2127 # *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
2128 # (and they won't honor URL@<rev> without -r<rev>, too!)
2129 sub svn_cmd_checkout {
2130 my ($url, $rev, $dir) = @_;
2131 my @cmd = ('svn','co', "-r$rev");
2132 push @cmd, '--ignore-externals' unless $_no_ignore_ext;
2133 $url .= "\@$rev" if $_svn_co_url_revs;
2134 sys(@cmd, $url, $dir);
2135 }
2136
2137 sub check_upgrade_needed {
2138 if (!-r $REVDB) {
2139 -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2140 open my $fh, '>>',$REVDB or croak $!;
2141 close $fh;
2142 }
2143 my $old = eval {
2144 my $pid = open my $child, '-|';
2145 defined $pid or croak $!;
2146 if ($pid == 0) {
2147 close STDERR;
2148 exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
2149 }
2150 my @ret = (<$child>);
2151 close $child or croak $?;
2152 die $? if $?; # just in case close didn't error out
2153 return wantarray ? @ret : join('',@ret);
2154 };
2155 return unless $old;
2156 my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
2157 if ($@ || !$head) {
2158 print STDERR "Please run: $0 rebuild --upgrade\n";
2159 exit 1;
2160 }
2161 }
2162
2163 # fills %tree_map with a reverse mapping of trees to commits. Useful
2164 # for finding parents to commit on.
2165 sub map_tree_joins {
2166 my %seen;
2167 foreach my $br (@_branch_from) {
2168 my $pid = open my $pipe, '-|';
2169 defined $pid or croak $!;
2170 if ($pid == 0) {
2171 exec(qw(git-rev-list --topo-order --pretty=raw), $br)
2172 or croak $!;
2173 }
2174 while (<$pipe>) {
2175 if (/^commit ($sha1)$/o) {
2176 my $commit = $1;
2177
2178 # if we've seen a commit,
2179 # we've seen its parents
2180 last if $seen{$commit};
2181 my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
2182 unless (defined $tree) {
2183 die "Failed to parse commit $commit\n";
2184 }
2185 push @{$tree_map{$tree}}, $commit;
2186 $seen{$commit} = 1;
2187 }
2188 }
2189 close $pipe; # we could be breaking the pipe early
2190 }
2191 }
2192
2193 sub load_all_refs {
2194 if (@_branch_from) {
2195 print STDERR '--branch|-b parameters are ignored when ',
2196 "--branch-all-refs|-B is passed\n";
2197 }
2198
2199 # don't worry about rev-list on non-commit objects/tags,
2200 # it shouldn't blow up if a ref is a blob or tree...
2201 chomp(@_branch_from = `git-rev-parse --symbolic --all`);
2202 }
2203
2204 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
2205 sub load_authors {
2206 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2207 while (<$authors>) {
2208 chomp;
2209 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
2210 my ($user, $name, $email) = ($1, $2, $3);
2211 $users{$user} = [$name, $email];
2212 }
2213 close $authors or croak $!;
2214 }
2215
2216 sub rload_authors {
2217 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2218 while (<$authors>) {
2219 chomp;
2220 next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2221 my ($user, $name, $email) = ($1, $2, $3);
2222 $rusers{"$name <$email>"} = $user;
2223 }
2224 close $authors or croak $!;
2225 }
2226
2227 sub svn_propget_base {
2228 my ($p, $f) = @_;
2229 $f .= '@BASE' if $_svn_pg_peg_revs;
2230 return safe_qx(qw/svn propget/, $p, $f);
2231 }
2232
2233 sub git_svn_each {
2234 my $sub = shift;
2235 foreach (`git-rev-parse --symbolic --all`) {
2236 next unless s#^refs/remotes/##;
2237 chomp $_;
2238 next unless -f "$GIT_DIR/svn/$_/info/url";
2239 &$sub($_);
2240 }
2241 }
2242
2243 sub migrate_revdb {
2244 git_svn_each(sub {
2245 my $id = shift;
2246 defined(my $pid = fork) or croak $!;
2247 if (!$pid) {
2248 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2249 init_vars();
2250 exit 0 if -r $REVDB;
2251 print "Upgrading svn => git mapping...\n";
2252 -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2253 open my $fh, '>>',$REVDB or croak $!;
2254 close $fh;
2255 rebuild();
2256 print "Done upgrading. You may now delete the ",
2257 "deprecated $GIT_SVN_DIR/revs directory\n";
2258 exit 0;
2259 }
2260 waitpid $pid, 0;
2261 croak $? if $?;
2262 });
2263 }
2264
2265 sub migration_check {
2266 migrate_revdb() unless (-e $REVDB);
2267 return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2268 print "Upgrading repository...\n";
2269 unless (-d "$GIT_DIR/svn") {
2270 mkdir "$GIT_DIR/svn" or croak $!;
2271 }
2272 print "Data from a previous version of git-svn exists, but\n\t",
2273 "$GIT_SVN_DIR\n\t(required for this version ",
2274 "($VERSION) of git-svn) does not.\n";
2275
2276 foreach my $x (`git-rev-parse --symbolic --all`) {
2277 next unless $x =~ s#^refs/remotes/##;
2278 chomp $x;
2279 next unless -f "$GIT_DIR/$x/info/url";
2280 my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2281 next unless $u;
2282 my $dn = dirname("$GIT_DIR/svn/$x");
2283 mkpath([$dn]) unless -d $dn;
2284 rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2285 }
2286 migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2287 print "Done upgrading.\n";
2288 }
2289
2290 sub find_rev_before {
2291 my ($r, $id, $eq_ok) = @_;
2292 my $f = "$GIT_DIR/svn/$id/.rev_db";
2293 return (undef,undef) unless -r $f;
2294 --$r unless $eq_ok;
2295 while ($r > 0) {
2296 if (my $c = revdb_get($f, $r)) {
2297 return ($r, $c);
2298 }
2299 --$r;
2300 }
2301 return (undef, undef);
2302 }
2303
2304 sub init_vars {
2305 $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2306 $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2307 $REVDB = "$GIT_SVN_DIR/.rev_db";
2308 $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2309 $SVN_URL = undef;
2310 $SVN_WC = "$GIT_SVN_DIR/tree";
2311 %tree_map = ();
2312 }
2313
2314 # convert GetOpt::Long specs for use by git-repo-config
2315 sub read_repo_config {
2316 return unless -d $GIT_DIR;
2317 my $opts = shift;
2318 foreach my $o (keys %$opts) {
2319 my $v = $opts->{$o};
2320 my ($key) = ($o =~ /^([a-z\-]+)/);
2321 $key =~ s/-//g;
2322 my $arg = 'git-repo-config';
2323 $arg .= ' --int' if ($o =~ /[:=]i$/);
2324 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2325 if (ref $v eq 'ARRAY') {
2326 chomp(my @tmp = `$arg --get-all svn.$key`);
2327 @$v = @tmp if @tmp;
2328 } else {
2329 chomp(my $tmp = `$arg --get svn.$key`);
2330 if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2331 $$v = $tmp;
2332 }
2333 }
2334 }
2335 }
2336
2337 sub set_default_vals {
2338 if (defined $_repack) {
2339 $_repack = 1000 if ($_repack <= 0);
2340 $_repack_nr = $_repack;
2341 $_repack_flags ||= '-d';
2342 }
2343 }
2344
2345 sub read_grafts {
2346 my $gr_file = shift;
2347 my ($grafts, $comments) = ({}, {});
2348 if (open my $fh, '<', $gr_file) {
2349 my @tmp;
2350 while (<$fh>) {
2351 if (/^($sha1)\s+/) {
2352 my $c = $1;
2353 if (@tmp) {
2354 @{$comments->{$c}} = @tmp;
2355 @tmp = ();
2356 }
2357 foreach my $p (split /\s+/, $_) {
2358 $grafts->{$c}->{$p} = 1;
2359 }
2360 } else {
2361 push @tmp, $_;
2362 }
2363 }
2364 close $fh or croak $!;
2365 @{$comments->{'END'}} = @tmp if @tmp;
2366 }
2367 return ($grafts, $comments);
2368 }
2369
2370 sub write_grafts {
2371 my ($grafts, $comments, $gr_file) = @_;
2372
2373 open my $fh, '>', $gr_file or croak $!;
2374 foreach my $c (sort keys %$grafts) {
2375 if ($comments->{$c}) {
2376 print $fh $_ foreach @{$comments->{$c}};
2377 }
2378 my $p = $grafts->{$c};
2379 my %x; # real parents
2380 delete $p->{$c}; # commits are not self-reproducing...
2381 my $pid = open my $ch, '-|';
2382 defined $pid or croak $!;
2383 if (!$pid) {
2384 exec(qw/git-cat-file commit/, $c) or croak $!;
2385 }
2386 while (<$ch>) {
2387 if (/^parent ($sha1)/) {
2388 $x{$1} = $p->{$1} = 1;
2389 } else {
2390 last unless /^\S/;
2391 }
2392 }
2393 close $ch; # breaking the pipe
2394
2395 # if real parents are the only ones in the grafts, drop it
2396 next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2397
2398 my (@ip, @jp, $mb);
2399 my %del = %x;
2400 @ip = @jp = keys %$p;
2401 foreach my $i (@ip) {
2402 next if $del{$i} || $p->{$i} == 2;
2403 foreach my $j (@jp) {
2404 next if $i eq $j || $del{$j} || $p->{$j} == 2;
2405 $mb = eval { safe_qx('git-merge-base',$i,$j) };
2406 next unless $mb;
2407 chomp $mb;
2408 next if $x{$mb};
2409 if ($mb eq $j) {
2410 delete $p->{$i};
2411 $del{$i} = 1;
2412 } elsif ($mb eq $i) {
2413 delete $p->{$j};
2414 $del{$j} = 1;
2415 }
2416 }
2417 }
2418
2419 # if real parents are the only ones in the grafts, drop it
2420 next if join(' ',sort keys %$p) eq join(' ',sort keys %x);
2421
2422 print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2423 }
2424 if ($comments->{'END'}) {
2425 print $fh $_ foreach @{$comments->{'END'}};
2426 }
2427 close $fh or croak $!;
2428 }
2429
2430 sub read_url_paths_all {
2431 my ($l_map, $pfx, $p) = @_;
2432 my @dir;
2433 foreach (<$p/*>) {
2434 if (-r "$_/info/url") {
2435 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2436 my $id = $pfx . basename $_;
2437 my $url = file_to_s("$_/info/url");
2438 my ($u, $p) = repo_path_split($url);
2439 $l_map->{$u}->{$p} = $id;
2440 } elsif (-d $_) {
2441 push @dir, $_;
2442 }
2443 }
2444 foreach (@dir) {
2445 my $x = $_;
2446 $x =~ s!^\Q$GIT_DIR\E/svn/!!o;
2447 read_url_paths_all($l_map, $x, $_);
2448 }
2449 }
2450
2451 # this one only gets ids that have been imported, not new ones
2452 sub read_url_paths {
2453 my $l_map = {};
2454 git_svn_each(sub { my $x = shift;
2455 my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2456 my ($u, $p) = repo_path_split($url);
2457 $l_map->{$u}->{$p} = $x;
2458 });
2459 return $l_map;
2460 }
2461
2462 sub extract_metadata {
2463 my $id = shift or return (undef, undef, undef);
2464 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2465 \s([a-f\d\-]+)$/x);
2466 if (!$rev || !$uuid || !$url) {
2467 # some of the original repositories I made had
2468 # identifiers like this:
2469 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2470 }
2471 return ($url, $rev, $uuid);
2472 }
2473
2474 sub cmt_metadata {
2475 return extract_metadata((grep(/^git-svn-id: /,
2476 safe_qx(qw/git-cat-file commit/, shift)))[-1]);
2477 }
2478
2479 sub get_commit_time {
2480 my $cmt = shift;
2481 defined(my $pid = open my $fh, '-|') or croak $!;
2482 if (!$pid) {
2483 exec qw/git-rev-list --pretty=raw -n1/, $cmt or croak $!;
2484 }
2485 while (<$fh>) {
2486 /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
2487 my ($s, $tz) = ($1, $2);
2488 if ($tz =~ s/^\+//) {
2489 $s += tz_to_s_offset($tz);
2490 } elsif ($tz =~ s/^\-//) {
2491 $s -= tz_to_s_offset($tz);
2492 }
2493 close $fh;
2494 return $s;
2495 }
2496 die "Can't get commit time for commit: $cmt\n";
2497 }
2498
2499 sub tz_to_s_offset {
2500 my ($tz) = @_;
2501 $tz =~ s/(\d\d)$//;
2502 return ($1 * 60) + ($tz * 3600);
2503 }
2504
2505 sub setup_pager { # translated to Perl from pager.c
2506 return unless (-t *STDOUT);
2507 my $pager = $ENV{PAGER};
2508 if (!defined $pager) {
2509 $pager = 'less';
2510 } elsif (length $pager == 0 || $pager eq 'cat') {
2511 return;
2512 }
2513 pipe my $rfd, my $wfd or return;
2514 defined(my $pid = fork) or croak $!;
2515 if (!$pid) {
2516 open STDOUT, '>&', $wfd or croak $!;
2517 return;
2518 }
2519 open STDIN, '<&', $rfd or croak $!;
2520 $ENV{LESS} ||= '-S';
2521 exec $pager or croak "Can't run pager: $!\n";;
2522 }
2523
2524 sub get_author_info {
2525 my ($dest, $author, $t, $tz) = @_;
2526 $author =~ s/(?:^\s*|\s*$)//g;
2527 $dest->{a_raw} = $author;
2528 my $_a;
2529 if ($_authors) {
2530 $_a = $rusers{$author} || undef;
2531 }
2532 if (!$_a) {
2533 ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2534 }
2535 $dest->{t} = $t;
2536 $dest->{tz} = $tz;
2537 $dest->{a} = $_a;
2538 # Date::Parse isn't in the standard Perl distro :(
2539 if ($tz =~ s/^\+//) {
2540 $t += tz_to_s_offset($tz);
2541 } elsif ($tz =~ s/^\-//) {
2542 $t -= tz_to_s_offset($tz);
2543 }
2544 $dest->{t_utc} = $t;
2545 }
2546
2547 sub process_commit {
2548 my ($c, $r_min, $r_max, $defer) = @_;
2549 if (defined $r_min && defined $r_max) {
2550 if ($r_min == $c->{r} && $r_min == $r_max) {
2551 show_commit($c);
2552 return 0;
2553 }
2554 return 1 if $r_min == $r_max;
2555 if ($r_min < $r_max) {
2556 # we need to reverse the print order
2557 return 0 if (defined $_limit && --$_limit < 0);
2558 push @$defer, $c;
2559 return 1;
2560 }
2561 if ($r_min != $r_max) {
2562 return 1 if ($r_min < $c->{r});
2563 return 1 if ($r_max > $c->{r});
2564 }
2565 }
2566 return 0 if (defined $_limit && --$_limit < 0);
2567 show_commit($c);
2568 return 1;
2569 }
2570
2571 sub show_commit {
2572 my $c = shift;
2573 if ($_oneline) {
2574 my $x = "\n";
2575 if (my $l = $c->{l}) {
2576 while ($l->[0] =~ /^\s*$/) { shift @$l }
2577 $x = $l->[0];
2578 }
2579 $_l_fmt ||= 'A' . length($c->{r});
2580 print 'r',pack($_l_fmt, $c->{r}),' | ';
2581 print "$c->{c} | " if $_show_commit;
2582 print $x;
2583 } else {
2584 show_commit_normal($c);
2585 }
2586 }
2587
2588 sub show_commit_changed_paths {
2589 my ($c) = @_;
2590 return unless $c->{changed};
2591 print "Changed paths:\n", @{$c->{changed}};
2592 }
2593
2594 sub show_commit_normal {
2595 my ($c) = @_;
2596 print '-' x72, "\nr$c->{r} | ";
2597 print "$c->{c} | " if $_show_commit;
2598 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2599 localtime($c->{t_utc})), ' | ';
2600 my $nr_line = 0;
2601
2602 if (my $l = $c->{l}) {
2603 while ($l->[$#$l] eq "\n" && $#$l > 0
2604 && $l->[($#$l - 1)] eq "\n") {
2605 pop @$l;
2606 }
2607 $nr_line = scalar @$l;
2608 if (!$nr_line) {
2609 print "1 line\n\n\n";
2610 } else {
2611 if ($nr_line == 1) {
2612 $nr_line = '1 line';
2613 } else {
2614 $nr_line .= ' lines';
2615 }
2616 print $nr_line, "\n";
2617 show_commit_changed_paths($c);
2618 print "\n";
2619 print $_ foreach @$l;
2620 }
2621 } else {
2622 print "1 line\n";
2623 show_commit_changed_paths($c);
2624 print "\n";
2625
2626 }
2627 foreach my $x (qw/raw diff/) {
2628 if ($c->{$x}) {
2629 print "\n";
2630 print $_ foreach @{$c->{$x}}
2631 }
2632 }
2633 }
2634
2635 sub libsvn_load {
2636 return unless $_use_lib;
2637 $_use_lib = eval {
2638 require SVN::Core;
2639 if ($SVN::Core::VERSION lt '1.1.0') {
2640 die "Need SVN::Core 1.1.0 or better ",
2641 "(got $SVN::Core::VERSION) ",
2642 "Falling back to command-line svn\n";
2643 }
2644 require SVN::Ra;
2645 require SVN::Delta;
2646 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2647 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2648 $SVN::Node::dir.$SVN::Node::unknown.
2649 $SVN::Node::none.$SVN::Node::file.
2650 $SVN::Node::dir.$SVN::Node::unknown;
2651 1;
2652 };
2653 }
2654
2655 sub libsvn_connect {
2656 my ($url) = @_;
2657 my $auth = SVN::Core::auth_open([SVN::Client::get_simple_provider(),
2658 SVN::Client::get_ssl_server_trust_file_provider(),
2659 SVN::Client::get_username_provider()]);
2660 my $s = eval { SVN::Ra->new(url => $url, auth => $auth) };
2661 return $s;
2662 }
2663
2664 sub libsvn_get_file {
2665 my ($gui, $f, $rev) = @_;
2666 my $p = $f;
2667 if (length $SVN_PATH > 0) {
2668 return unless ($p =~ s#^\Q$SVN_PATH\E/##);
2669 }
2670
2671 my ($hash, $pid, $in, $out);
2672 my $pool = SVN::Pool->new;
2673 defined($pid = open3($in, $out, '>&STDERR',
2674 qw/git-hash-object -w --stdin/)) or croak $!;
2675 # redirect STDOUT for SVN 1.1.x compatibility
2676 open my $stdout, '>&', \*STDOUT or croak $!;
2677 open STDOUT, '>&', $in or croak $!;
2678 my ($r, $props) = $SVN->get_file($f, $rev, \*STDOUT, $pool);
2679 $in->flush == 0 or croak $!;
2680 open STDOUT, '>&', $stdout or croak $!;
2681 close $in or croak $!;
2682 close $stdout or croak $!;
2683 $pool->clear;
2684 chomp($hash = do { local $/; <$out> });
2685 close $out or croak $!;
2686 waitpid $pid, 0;
2687 $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2688
2689 my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2690 if (exists $props->{'svn:special'}) {
2691 $mode = '120000';
2692 my $link = `git-cat-file blob $hash`;
2693 $link =~ s/^link // or die "svn:special file with contents: <",
2694 $link, "> is not understood\n";
2695 defined($pid = open3($in, $out, '>&STDERR',
2696 qw/git-hash-object -w --stdin/)) or croak $!;
2697 print $in $link;
2698 $in->flush == 0 or croak $!;
2699 close $in or croak $!;
2700 chomp($hash = do { local $/; <$out> });
2701 close $out or croak $!;
2702 waitpid $pid, 0;
2703 $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2704 }
2705 print $gui $mode,' ',$hash,"\t",$p,"\0" or croak $!;
2706 }
2707
2708 sub libsvn_log_entry {
2709 my ($rev, $author, $date, $msg, $parents) = @_;
2710 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2711 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2712 or die "Unable to parse date: $date\n";
2713 if (defined $_authors && ! defined $users{$author}) {
2714 die "Author: $author not defined in $_authors file\n";
2715 }
2716 $msg = '' if ($rev == 0 && !defined $msg);
2717 return { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
2718 author => $author, msg => $msg."\n", parents => $parents || [] }
2719 }
2720
2721 sub process_rm {
2722 my ($gui, $last_commit, $f) = @_;
2723 $f =~ s#^\Q$SVN_PATH\E/?## or return;
2724 # remove entire directories.
2725 if (safe_qx('git-ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
2726 defined(my $pid = open my $ls, '-|') or croak $!;
2727 if (!$pid) {
2728 exec(qw/git-ls-tree -r --name-only -z/,
2729 $last_commit,'--',$f) or croak $!;
2730 }
2731 local $/ = "\0";
2732 while (<$ls>) {
2733 print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2734 }
2735 close $ls or croak $?;
2736 } else {
2737 print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
2738 }
2739 }
2740
2741 sub libsvn_fetch {
2742 my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
2743 open my $gui, '| git-update-index -z --index-info' or croak $!;
2744 my @amr;
2745 foreach my $f (keys %$paths) {
2746 my $m = $paths->{$f}->action();
2747 $f =~ s#^/+##;
2748 if ($m =~ /^[DR]$/) {
2749 print "\t$m\t$f\n" unless $_q;
2750 process_rm($gui, $last_commit, $f);
2751 next if $m eq 'D';
2752 # 'R' can be file replacements, too, right?
2753 }
2754 my $pool = SVN::Pool->new;
2755 my $t = $SVN->check_path($f, $rev, $pool);
2756 if ($t == $SVN::Node::file) {
2757 if ($m =~ /^[AMR]$/) {
2758 push @amr, [ $m, $f ];
2759 } else {
2760 die "Unrecognized action: $m, ($f r$rev)\n";
2761 }
2762 } elsif ($t == $SVN::Node::dir && $m =~ /^[AR]$/) {
2763 my @traversed = ();
2764 libsvn_traverse($gui, '', $f, $rev, \@traversed);
2765 foreach (@traversed) {
2766 push @amr, [ $m, $_ ]
2767 }
2768 }
2769 $pool->clear;
2770 }
2771 foreach (@amr) {
2772 print "\t$_->[0]\t$_->[1]\n" unless $_q;
2773 libsvn_get_file($gui, $_->[1], $rev)
2774 }
2775 close $gui or croak $?;
2776 return libsvn_log_entry($rev, $author, $date, $msg, [$last_commit]);
2777 }
2778
2779 sub svn_grab_base_rev {
2780 defined(my $pid = open my $fh, '-|') or croak $!;
2781 if (!$pid) {
2782 open my $null, '>', '/dev/null' or croak $!;
2783 open STDERR, '>&', $null or croak $!;
2784 exec qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0"
2785 or croak $!;
2786 }
2787 chomp(my $c = do { local $/; <$fh> });
2788 close $fh;
2789 if (defined $c && length $c) {
2790 my ($url, $rev, $uuid) = cmt_metadata($c);
2791 return ($rev, $c) if defined $rev;
2792 }
2793 if ($_no_metadata) {
2794 my $offset = -41; # from tail
2795 my $rl;
2796 open my $fh, '<', $REVDB or
2797 die "--no-metadata specified and $REVDB not readable\n";
2798 seek $fh, $offset, 2;
2799 $rl = readline $fh;
2800 defined $rl or return (undef, undef);
2801 chomp $rl;
2802 while ($c ne $rl && tell $fh != 0) {
2803 $offset -= 41;
2804 seek $fh, $offset, 2;
2805 $rl = readline $fh;
2806 defined $rl or return (undef, undef);
2807 chomp $rl;
2808 }
2809 my $rev = tell $fh;
2810 croak $! if ($rev < -1);
2811 $rev = ($rev - 41) / 41;
2812 close $fh or croak $!;
2813 return ($rev, $c);
2814 }
2815 return (undef, undef);
2816 }
2817
2818 sub libsvn_parse_revision {
2819 my $base = shift;
2820 my $head = $SVN->get_latest_revnum();
2821 if (!defined $_revision || $_revision eq 'BASE:HEAD') {
2822 return ($base + 1, $head) if (defined $base);
2823 return (0, $head);
2824 }
2825 return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
2826 return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
2827 if ($_revision =~ /^BASE:(\d+)$/) {
2828 return ($base + 1, $1) if (defined $base);
2829 return (0, $head);
2830 }
2831 return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
2832 die "revision argument: $_revision not understood by git-svn\n",
2833 "Try using the command-line svn client instead\n";
2834 }
2835
2836 sub libsvn_traverse {
2837 my ($gui, $pfx, $path, $rev, $files) = @_;
2838 my $cwd = "$pfx/$path";
2839 my $pool = SVN::Pool->new;
2840 $cwd =~ s#^/+##g;
2841 my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
2842 foreach my $d (keys %$dirent) {
2843 my $t = $dirent->{$d}->kind;
2844 if ($t == $SVN::Node::dir) {
2845 libsvn_traverse($gui, $cwd, $d, $rev, $files);
2846 } elsif ($t == $SVN::Node::file) {
2847 my $file = "$cwd/$d";
2848 if (defined $files) {
2849 push @$files, $file;
2850 } else {
2851 print "\tA\t$file\n" unless $_q;
2852 libsvn_get_file($gui, $file, $rev);
2853 }
2854 }
2855 }
2856 $pool->clear;
2857 }
2858
2859 sub libsvn_traverse_ignore {
2860 my ($fh, $path, $r) = @_;
2861 $path =~ s#^/+##g;
2862 my $pool = SVN::Pool->new;
2863 my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
2864 my $p = $path;
2865 $p =~ s#^\Q$SVN_PATH\E/?##;
2866 print $fh length $p ? "\n# $p\n" : "\n# /\n";
2867 if (my $s = $props->{'svn:ignore'}) {
2868 $s =~ s/[\r\n]+/\n/g;
2869 chomp $s;
2870 if (length $p == 0) {
2871 $s =~ s#\n#\n/$p#g;
2872 print $fh "/$s\n";
2873 } else {
2874 $s =~ s#\n#\n/$p/#g;
2875 print $fh "/$p/$s\n";
2876 }
2877 }
2878 foreach (sort keys %$dirent) {
2879 next if $dirent->{$_}->kind != $SVN::Node::dir;
2880 libsvn_traverse_ignore($fh, "$path/$_", $r);
2881 }
2882 $pool->clear;
2883 }
2884
2885 sub revisions_eq {
2886 my ($path, $r0, $r1) = @_;
2887 return 1 if $r0 == $r1;
2888 my $nr = 0;
2889 if ($_use_lib) {
2890 # should be OK to use Pool here (r1 - r0) should be small
2891 my $pool = SVN::Pool->new;
2892 libsvn_get_log($SVN, "/$path", $r0, $r1,
2893 0, 1, 1, sub {$nr++}, $pool);
2894 $pool->clear;
2895 } else {
2896 my ($url, undef) = repo_path_split($SVN_URL);
2897 my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
2898 while (next_log_entry($svn_log)) { $nr++ }
2899 close $svn_log->{fh};
2900 }
2901 return 0 if ($nr > 1);
2902 return 1;
2903 }
2904
2905 sub libsvn_find_parent_branch {
2906 my ($paths, $rev, $author, $date, $msg) = @_;
2907 my $svn_path = '/'.$SVN_PATH;
2908
2909 # look for a parent from another branch:
2910 my $i = $paths->{$svn_path} or return;
2911 my $branch_from = $i->copyfrom_path or return;
2912 my $r = $i->copyfrom_rev;
2913 print STDERR "Found possible branch point: ",
2914 "$branch_from => $svn_path, $r\n";
2915 $branch_from =~ s#^/##;
2916 my $l_map = {};
2917 read_url_paths_all($l_map, '', "$GIT_DIR/svn");
2918 my $url = $SVN->{url};
2919 defined $l_map->{$url} or return;
2920 my $id = $l_map->{$url}->{$branch_from};
2921 if (!defined $id && $_follow_parent) {
2922 print STDERR "Following parent: $branch_from\@$r\n";
2923 # auto create a new branch and follow it
2924 $id = basename($branch_from);
2925 $id .= '@'.$r if -r "$GIT_DIR/svn/$id";
2926 while (-r "$GIT_DIR/svn/$id") {
2927 # just grow a tail if we're not unique enough :x
2928 $id .= '-';
2929 }
2930 }
2931 return unless defined $id;
2932
2933 my ($r0, $parent) = find_rev_before($r,$id,1);
2934 if ($_follow_parent && (!defined $r0 || !defined $parent)) {
2935 defined(my $pid = fork) or croak $!;
2936 if (!$pid) {
2937 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2938 init_vars();
2939 $SVN_URL = "$url/$branch_from";
2940 $SVN_LOG = $SVN = undef;
2941 setup_git_svn();
2942 # we can't assume SVN_URL exists at r+1:
2943 $_revision = "0:$r";
2944 fetch_lib();
2945 exit 0;
2946 }
2947 waitpid $pid, 0;
2948 croak $? if $?;
2949 ($r0, $parent) = find_rev_before($r,$id,1);
2950 }
2951 return unless (defined $r0 && defined $parent);
2952 if (revisions_eq($branch_from, $r0, $r)) {
2953 unlink $GIT_SVN_INDEX;
2954 print STDERR "Found branch parent: ($GIT_SVN) $parent\n";
2955 sys(qw/git-read-tree/, $parent);
2956 return libsvn_fetch($parent, $paths, $rev,
2957 $author, $date, $msg);
2958 }
2959 print STDERR "Nope, branch point not imported or unknown\n";
2960 return undef;
2961 }
2962
2963 sub libsvn_get_log {
2964 my ($ra, @args) = @_;
2965 if ($SVN::Core::VERSION le '1.2.0') {
2966 splice(@args, 3, 1);
2967 }
2968 $ra->get_log(@args);
2969 }
2970
2971 sub libsvn_new_tree {
2972 if (my $log_entry = libsvn_find_parent_branch(@_)) {
2973 return $log_entry;
2974 }
2975 my ($paths, $rev, $author, $date, $msg) = @_;
2976 open my $gui, '| git-update-index -z --index-info' or croak $!;
2977 libsvn_traverse($gui, '', $SVN_PATH, $rev);
2978 close $gui or croak $?;
2979 return libsvn_log_entry($rev, $author, $date, $msg);
2980 }
2981
2982 sub find_graft_path_commit {
2983 my ($tree_paths, $p1, $r1) = @_;
2984 foreach my $x (keys %$tree_paths) {
2985 next unless ($p1 =~ /^\Q$x\E/);
2986 my $i = $tree_paths->{$x};
2987 my ($r0, $parent) = find_rev_before($r1,$i,1);
2988 return $parent if (defined $r0 && $r0 == $r1);
2989 print STDERR "r$r1 of $i not imported\n";
2990 next;
2991 }
2992 return undef;
2993 }
2994
2995 sub find_graft_path_parents {
2996 my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
2997 foreach my $x (keys %$tree_paths) {
2998 next unless ($p0 =~ /^\Q$x\E/);
2999 my $i = $tree_paths->{$x};
3000 my ($r, $parent) = find_rev_before($r0, $i, 1);
3001 if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
3002 my ($url_b, undef, $uuid_b) = cmt_metadata($c);
3003 my ($url_a, undef, $uuid_a) = cmt_metadata($parent);
3004 next if ($url_a && $url_b && $url_a eq $url_b &&
3005 $uuid_b eq $uuid_a);
3006 $grafts->{$c}->{$parent} = 1;
3007 }
3008 }
3009 }
3010
3011 sub libsvn_graft_file_copies {
3012 my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
3013 foreach (keys %$paths) {
3014 my $i = $paths->{$_};
3015 my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
3016 $i->copyfrom_rev);
3017 next unless (defined $p0 && defined $r0);
3018
3019 my $p1 = $_;
3020 $p1 =~ s#^/##;
3021 $p0 =~ s#^/##;
3022 my $c = find_graft_path_commit($tree_paths, $p1, $rev);
3023 next unless $c;
3024 find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
3025 }
3026 }
3027
3028 sub set_index {
3029 my $old = $ENV{GIT_INDEX_FILE};
3030 $ENV{GIT_INDEX_FILE} = shift;
3031 return $old;
3032 }
3033
3034 sub restore_index {
3035 my ($old) = @_;
3036 if (defined $old) {
3037 $ENV{GIT_INDEX_FILE} = $old;
3038 } else {
3039 delete $ENV{GIT_INDEX_FILE};
3040 }
3041 }
3042
3043 sub libsvn_commit_cb {
3044 my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
3045 if ($_optimize_commits && $rev == ($r_last + 1)) {
3046 my $log = libsvn_log_entry($rev,$committer,$date,$msg);
3047 $log->{tree} = get_tree_from_treeish($c);
3048 my $cmt = git_commit($log, $cmt_last, $c);
3049 my @diff = safe_qx('git-diff-tree', $cmt, $c);
3050 if (@diff) {
3051 print STDERR "Trees differ: $cmt $c\n",
3052 join('',@diff),"\n";
3053 exit 1;
3054 }
3055 } else {
3056 fetch("$rev=$c");
3057 }
3058 }
3059
3060 sub libsvn_ls_fullurl {
3061 my $fullurl = shift;
3062 my ($repo, $path) = repo_path_split($fullurl);
3063 $SVN ||= libsvn_connect($repo);
3064 my @ret;
3065 my $pool = SVN::Pool->new;
3066 my ($dirent, undef, undef) = $SVN->get_dir($path,
3067 $SVN->get_latest_revnum, $pool);
3068 foreach my $d (keys %$dirent) {
3069 if ($dirent->{$d}->kind == $SVN::Node::dir) {
3070 push @ret, "$d/"; # add '/' for compat with cli svn
3071 }
3072 }
3073 $pool->clear;
3074 return @ret;
3075 }
3076
3077
3078 sub libsvn_skip_unknown_revs {
3079 my $err = shift;
3080 my $errno = $err->apr_err();
3081 # Maybe the branch we're tracking didn't
3082 # exist when the repo started, so it's
3083 # not an error if it doesn't, just continue
3084 #
3085 # Wonderfully consistent library, eh?
3086 # 160013 - svn:// and file://
3087 # 175002 - http(s)://
3088 # More codes may be discovered later...
3089 if ($errno == 175002 || $errno == 160013) {
3090 return;
3091 }
3092 croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
3093 };
3094
3095 # Tie::File seems to be prone to offset errors if revisions get sparse,
3096 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
3097 # one of my favorite modules is out :< Next up would be one of the DBM
3098 # modules, but I'm not sure which is most portable... So I'll just
3099 # go with something that's plain-text, but still capable of
3100 # being randomly accessed. So here's my ultra-simple fixed-width
3101 # database. All records are 40 characters + "\n", so it's easy to seek
3102 # to a revision: (41 * rev) is the byte offset.
3103 # A record of 40 0s denotes an empty revision.
3104 # And yes, it's still pretty fast (faster than Tie::File).
3105 sub revdb_set {
3106 my ($file, $rev, $commit) = @_;
3107 length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
3108 open my $fh, '+<', $file or croak $!;
3109 my $offset = $rev * 41;
3110 # assume that append is the common case:
3111 seek $fh, 0, 2 or croak $!;
3112 my $pos = tell $fh;
3113 if ($pos < $offset) {
3114 print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
3115 }
3116 seek $fh, $offset, 0 or croak $!;
3117 print $fh $commit,"\n";
3118 close $fh or croak $!;
3119 }
3120
3121 sub revdb_get {
3122 my ($file, $rev) = @_;
3123 my $ret;
3124 my $offset = $rev * 41;
3125 open my $fh, '<', $file or croak $!;
3126 seek $fh, $offset, 0;
3127 if (tell $fh == $offset) {
3128 $ret = readline $fh;
3129 if (defined $ret) {
3130 chomp $ret;
3131 $ret = undef if ($ret =~ /^0{40}$/);
3132 }
3133 }
3134 close $fh or croak $!;
3135 return $ret;
3136 }
3137
3138 sub copy_remote_ref {
3139 my $origin = $_cp_remote ? $_cp_remote : 'origin';
3140 my $ref = "refs/remotes/$GIT_SVN";
3141 if (safe_qx('git-ls-remote', $origin, $ref)) {
3142 sys(qw/git fetch/, $origin, "$ref:$ref");
3143 } else {
3144 die "Unable to find remote reference: ",
3145 "refs/remotes/$GIT_SVN on $origin\n";
3146 }
3147 }
3148
3149 package SVN::Git::Editor;
3150 use vars qw/@ISA/;
3151 use strict;
3152 use warnings;
3153 use Carp qw/croak/;
3154 use IO::File;
3155
3156 sub new {
3157 my $class = shift;
3158 my $git_svn = shift;
3159 my $self = SVN::Delta::Editor->new(@_);
3160 bless $self, $class;
3161 foreach (qw/svn_path c r ra /) {
3162 die "$_ required!\n" unless (defined $git_svn->{$_});
3163 $self->{$_} = $git_svn->{$_};
3164 }
3165 $self->{pool} = SVN::Pool->new;
3166 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3167 $self->{rm} = { };
3168 require Digest::MD5;
3169 return $self;
3170 }
3171
3172 sub split_path {
3173 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3174 }
3175
3176 sub repo_path {
3177 (defined $_[1] && length $_[1]) ? "$_[0]->{svn_path}/$_[1]"
3178 : $_[0]->{svn_path}
3179 }
3180
3181 sub url_path {
3182 my ($self, $path) = @_;
3183 $self->{ra}->{url} . '/' . $self->repo_path($path);
3184 }
3185
3186 sub rmdirs {
3187 my ($self, $q) = @_;
3188 my $rm = $self->{rm};
3189 delete $rm->{''}; # we never delete the url we're tracking
3190 return unless %$rm;
3191
3192 foreach (keys %$rm) {
3193 my @d = split m#/#, $_;
3194 my $c = shift @d;
3195 $rm->{$c} = 1;
3196 while (@d) {
3197 $c .= '/' . shift @d;
3198 $rm->{$c} = 1;
3199 }
3200 }
3201 delete $rm->{$self->{svn_path}};
3202 delete $rm->{''}; # we never delete the url we're tracking
3203 return unless %$rm;
3204
3205 defined(my $pid = open my $fh,'-|') or croak $!;
3206 if (!$pid) {
3207 exec qw/git-ls-tree --name-only -r -z/, $self->{c} or croak $!;
3208 }
3209 local $/ = "\0";
3210 my @svn_path = split m#/#, $self->{svn_path};
3211 while (<$fh>) {
3212 chomp;
3213 my @dn = (@svn_path, (split m#/#, $_));
3214 while (pop @dn) {
3215 delete $rm->{join '/', @dn};
3216 }
3217 unless (%$rm) {
3218 close $fh;
3219 return;
3220 }
3221 }
3222 close $fh;
3223
3224 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3225 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3226 $self->close_directory($bat->{$d}, $p);
3227 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3228 print "\tD+\t/$d/\n" unless $q;
3229 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3230 delete $bat->{$d};
3231 }
3232 }
3233
3234 sub open_or_add_dir {
3235 my ($self, $full_path, $baton) = @_;
3236 my $p = SVN::Pool->new;
3237 my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
3238 $p->clear;
3239 if ($t == $SVN::Node::none) {
3240 return $self->add_directory($full_path, $baton,
3241 undef, -1, $self->{pool});
3242 } elsif ($t == $SVN::Node::dir) {
3243 return $self->open_directory($full_path, $baton,
3244 $self->{r}, $self->{pool});
3245 }
3246 print STDERR "$full_path already exists in repository at ",
3247 "r$self->{r} and it is not a directory (",
3248 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3249 exit 1;
3250 }
3251
3252 sub ensure_path {
3253 my ($self, $path) = @_;
3254 my $bat = $self->{bat};
3255 $path = $self->repo_path($path);
3256 return $bat->{''} unless (length $path);
3257 my @p = split m#/+#, $path;
3258 my $c = shift @p;
3259 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3260 while (@p) {
3261 my $c0 = $c;
3262 $c .= '/' . shift @p;
3263 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3264 }
3265 return $bat->{$c};
3266 }
3267
3268 sub A {
3269 my ($self, $m, $q) = @_;
3270 my ($dir, $file) = split_path($m->{file_b});
3271 my $pbat = $self->ensure_path($dir);
3272 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3273 undef, -1);
3274 print "\tA\t$m->{file_b}\n" unless $q;
3275 $self->chg_file($fbat, $m);
3276 $self->close_file($fbat,undef,$self->{pool});
3277 }
3278
3279 sub C {
3280 my ($self, $m, $q) = @_;
3281 my ($dir, $file) = split_path($m->{file_b});
3282 my $pbat = $self->ensure_path($dir);
3283 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3284 $self->url_path($m->{file_a}), $self->{r});
3285 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $q;
3286 $self->chg_file($fbat, $m);
3287 $self->close_file($fbat,undef,$self->{pool});
3288 }
3289
3290 sub delete_entry {
3291 my ($self, $path, $pbat) = @_;
3292 my $rpath = $self->repo_path($path);
3293 my ($dir, $file) = split_path($rpath);
3294 $self->{rm}->{$dir} = 1;
3295 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3296 }
3297
3298 sub R {
3299 my ($self, $m, $q) = @_;
3300 my ($dir, $file) = split_path($m->{file_b});
3301 my $pbat = $self->ensure_path($dir);
3302 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3303 $self->url_path($m->{file_a}), $self->{r});
3304 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $q;
3305 $self->chg_file($fbat, $m);
3306 $self->close_file($fbat,undef,$self->{pool});
3307
3308 ($dir, $file) = split_path($m->{file_a});
3309 $pbat = $self->ensure_path($dir);
3310 $self->delete_entry($m->{file_a}, $pbat);
3311 }
3312
3313 sub M {
3314 my ($self, $m, $q) = @_;
3315 my ($dir, $file) = split_path($m->{file_b});
3316 my $pbat = $self->ensure_path($dir);
3317 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3318 $pbat,$self->{r},$self->{pool});
3319 print "\t$m->{chg}\t$m->{file_b}\n" unless $q;
3320 $self->chg_file($fbat, $m);
3321 $self->close_file($fbat,undef,$self->{pool});
3322 }
3323
3324 sub T { shift->M(@_) }
3325
3326 sub change_file_prop {
3327 my ($self, $fbat, $pname, $pval) = @_;
3328 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3329 }
3330
3331 sub chg_file {
3332 my ($self, $fbat, $m) = @_;
3333 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3334 $self->change_file_prop($fbat,'svn:executable','*');
3335 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3336 $self->change_file_prop($fbat,'svn:executable',undef);
3337 }
3338 my $fh = IO::File->new_tmpfile or croak $!;
3339 if ($m->{mode_b} =~ /^120/) {
3340 print $fh 'link ' or croak $!;
3341 $self->change_file_prop($fbat,'svn:special','*');
3342 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3343 $self->change_file_prop($fbat,'svn:special',undef);
3344 }
3345 defined(my $pid = fork) or croak $!;
3346 if (!$pid) {
3347 open STDOUT, '>&', $fh or croak $!;
3348 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3349 }
3350 waitpid $pid, 0;
3351 croak $? if $?;
3352 $fh->flush == 0 or croak $!;
3353 seek $fh, 0, 0 or croak $!;
3354
3355 my $md5 = Digest::MD5->new;
3356 $md5->addfile($fh) or croak $!;
3357 seek $fh, 0, 0 or croak $!;
3358
3359 my $exp = $md5->hexdigest;
3360 my $pool = SVN::Pool->new;
3361 my $atd = $self->apply_textdelta($fbat, undef, $pool);
3362 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3363 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3364 $pool->clear;
3365
3366 close $fh or croak $!;
3367 }
3368
3369 sub D {
3370 my ($self, $m, $q) = @_;
3371 my ($dir, $file) = split_path($m->{file_b});
3372 my $pbat = $self->ensure_path($dir);
3373 print "\tD\t$m->{file_b}\n" unless $q;
3374 $self->delete_entry($m->{file_b}, $pbat);
3375 }
3376
3377 sub close_edit {
3378 my ($self) = @_;
3379 my ($p,$bat) = ($self->{pool}, $self->{bat});
3380 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3381 $self->close_directory($bat->{$_}, $p);
3382 }
3383 $self->SUPER::close_edit($p);
3384 $p->clear;
3385 }
3386
3387 sub abort_edit {
3388 my ($self) = @_;
3389 $self->SUPER::abort_edit($self->{pool});
3390 $self->{pool}->clear;
3391 }
3392
3393 __END__
3394
3395 Data structures:
3396
3397 $svn_log hashref (as returned by svn_log_raw)
3398 {
3399 fh => file handle of the log file,
3400 state => state of the log file parser (sep/msg/rev/msg_start...)
3401 }
3402
3403 $log_msg hashref as returned by next_log_entry($svn_log)
3404 {
3405 msg => 'whitespace-formatted log entry
3406 ', # trailing newline is preserved
3407 revision => '8', # integer
3408 date => '2004-02-24T17:01:44.108345Z', # commit date
3409 author => 'committer name'
3410 };
3411
3412
3413 @mods = array of diff-index line hashes, each element represents one line
3414 of diff-index output
3415
3416 diff-index line ($m hash)
3417 {
3418 mode_a => first column of diff-index output, no leading ':',
3419 mode_b => second column of diff-index output,
3420 sha1_b => sha1sum of the final blob,
3421 chg => change type [MCRADT],
3422 file_a => original file name of a file (iff chg is 'C' or 'R')
3423 file_b => new/current file name of a file (any chg)
3424 }
3425 ;
3426
3427 # retval of read_url_paths{,_all}();
3428 $l_map = {
3429 # repository root url
3430 'https://svn.musicpd.org' => {
3431 # repository path # GIT_SVN_ID
3432 'mpd/trunk' => 'trunk',
3433 'mpd/tags/0.11.5' => 'tags/0.11.5',
3434 },
3435 }
3436
3437 Notes:
3438 I don't trust the each() function on unless I created %hash myself
3439 because the internal iterator may not have started at base.