Merge branch 'ew/rebase'
[git/git.git] / contrib / git-svn / git-svn.perl
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 = '1.1.1-broken';
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 $ENV{LC_ALL} = 'C';
22
23 # If SVN:: library support is added, please make the dependencies
24 # optional and preserve the capability to use the command-line client.
25 # use eval { require SVN::... } to make it lazy load
26 # We don't use any modules not in the standard Perl distribution:
27 use Carp qw/croak/;
28 use IO::File qw//;
29 use File::Basename qw/dirname basename/;
30 use File::Path qw/mkpath/;
31 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
32 use File::Spec qw//;
33 use POSIX qw/strftime/;
34 use IPC::Open3;
35 use Memoize;
36 memoize('revisions_eq');
37
38 my ($SVN_PATH, $SVN, $SVN_LOG, $_use_lib);
39 $_use_lib = 1 unless $ENV{GIT_SVN_NO_LIB};
40 libsvn_load();
41 my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
42 my $sha1 = qr/[a-f\d]{40}/;
43 my $sha1_short = qr/[a-f\d]{4,40}/;
44 my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
45 $_find_copies_harder, $_l, $_cp_similarity, $_cp_remote,
46 $_repack, $_repack_nr, $_repack_flags,
47 $_template, $_shared, $_no_default_regex, $_no_graft_copy,
48 $_limit, $_verbose, $_incremental, $_oneline, $_l_fmt, $_show_commit,
49 $_version, $_upgrade, $_authors, $_branch_all_refs, @_opt_m);
50 my (@_branch_from, %tree_map, %users, %rusers, %equiv);
51 my ($_svn_co_url_revs, $_svn_pg_peg_revs);
52 my @repo_path_split_cache;
53
54 my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
55 'branch|b=s' => \@_branch_from,
56 'branch-all-refs|B' => \$_branch_all_refs,
57 'authors-file|A=s' => \$_authors,
58 'repack:i' => \$_repack,
59 'repack-flags|repack-args|repack-opts=s' => \$_repack_flags);
60
61 my ($_trunk, $_tags, $_branches);
62 my %multi_opts = ( 'trunk|T=s' => \$_trunk,
63 'tags|t=s' => \$_tags,
64 'branches|b=s' => \$_branches );
65 my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
66
67 # yes, 'native' sets "\n". Patches to fix this for non-*nix systems welcome:
68 my %EOL = ( CR => "\015", LF => "\012", CRLF => "\015\012", native => "\012" );
69
70 my %cmd = (
71 fetch => [ \&fetch, "Download new revisions from SVN",
72 { 'revision|r=s' => \$_revision, %fc_opts } ],
73 init => [ \&init, "Initialize a repo for tracking" .
74 " (requires URL argument)",
75 \%init_opts ],
76 commit => [ \&commit, "Commit git revisions to SVN",
77 { 'stdin|' => \$_stdin,
78 'edit|e' => \$_edit,
79 'rmdir' => \$_rmdir,
80 'find-copies-harder' => \$_find_copies_harder,
81 'l=i' => \$_l,
82 'copy-similarity|C=i'=> \$_cp_similarity,
83 %fc_opts,
84 } ],
85 'show-ignore' => [ \&show_ignore, "Show svn:ignore listings",
86 { 'revision|r=i' => \$_revision } ],
87 rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
88 { 'no-ignore-externals' => \$_no_ignore_ext,
89 'copy-remote|remote=s' => \$_cp_remote,
90 'upgrade' => \$_upgrade } ],
91 'graft-branches' => [ \&graft_branches,
92 'Detect merges/branches from already imported history',
93 { 'merge-rx|m' => \@_opt_m,
94 'no-default-regex' => \$_no_default_regex,
95 'no-graft-copy' => \$_no_graft_copy } ],
96 'multi-init' => [ \&multi_init,
97 'Initialize multiple trees (like git-svnimport)',
98 { %multi_opts, %fc_opts } ],
99 'multi-fetch' => [ \&multi_fetch,
100 'Fetch multiple trees (like git-svnimport)',
101 \%fc_opts ],
102 'log' => [ \&show_log, 'Show commit logs',
103 { 'limit=i' => \$_limit,
104 'revision|r=s' => \$_revision,
105 'verbose|v' => \$_verbose,
106 'incremental' => \$_incremental,
107 'oneline' => \$_oneline,
108 'show-commit' => \$_show_commit,
109 'authors-file|A=s' => \$_authors,
110 } ],
111 );
112
113 my $cmd;
114 for (my $i = 0; $i < @ARGV; $i++) {
115 if (defined $cmd{$ARGV[$i]}) {
116 $cmd = $ARGV[$i];
117 splice @ARGV, $i, 1;
118 last;
119 }
120 };
121
122 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
123
124 read_repo_config(\%opts);
125 my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
126 'version|V' => \$_version,
127 'id|i=s' => \$GIT_SVN);
128 exit 1 if (!$rv && $cmd ne 'log');
129
130 set_default_vals();
131 usage(0) if $_help;
132 version() if $_version;
133 usage(1) unless defined $cmd;
134 init_vars();
135 load_authors() if $_authors;
136 load_all_refs() if $_branch_all_refs;
137 svn_compat_check();
138 migration_check() unless $cmd =~ /^(?:init|rebuild|multi-init)$/;
139 $cmd{$cmd}->[0]->(@ARGV);
140 exit 0;
141
142 ####################### primary functions ######################
143 sub usage {
144 my $exit = shift || 0;
145 my $fd = $exit ? \*STDERR : \*STDOUT;
146 print $fd <<"";
147 git-svn - bidirectional operations between a single Subversion tree and git
148 Usage: $0 <command> [options] [arguments]\n
149
150 print $fd "Available commands:\n" unless $cmd;
151
152 foreach (sort keys %cmd) {
153 next if $cmd && $cmd ne $_;
154 print $fd ' ',pack('A13',$_),$cmd{$_}->[1],"\n";
155 foreach (keys %{$cmd{$_}->[2]}) {
156 # prints out arguments as they should be passed:
157 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
158 print $fd ' ' x 17, join(', ', map { length $_ > 1 ?
159 "--$_" : "-$_" }
160 split /\|/,$_)," $x\n";
161 }
162 }
163 print $fd <<"";
164 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
165 arbitrary identifier if you're tracking multiple SVN branches/repositories in
166 one git repository and want to keep them separate. See git-svn(1) for more
167 information.
168
169 exit $exit;
170 }
171
172 sub version {
173 print "git-svn version $VERSION\n";
174 exit 0;
175 }
176
177 sub rebuild {
178 if (quiet_run(qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0")) {
179 copy_remote_ref();
180 }
181 $SVN_URL = shift or undef;
182 my $newest_rev = 0;
183 if ($_upgrade) {
184 sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
185 } else {
186 check_upgrade_needed();
187 }
188
189 my $pid = open(my $rev_list,'-|');
190 defined $pid or croak $!;
191 if ($pid == 0) {
192 exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
193 }
194 my $latest;
195 while (<$rev_list>) {
196 chomp;
197 my $c = $_;
198 croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
199 my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
200 next if (!@commit); # skip merges
201 my ($url, $rev, $uuid) = extract_metadata($commit[$#commit]);
202 if (!$rev || !$uuid) {
203 croak "Unable to extract revision or UUID from ",
204 "$c, $commit[$#commit]\n";
205 }
206
207 # if we merged or otherwise started elsewhere, this is
208 # how we break out of it
209 next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
210 next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
211
212 unless (defined $latest) {
213 if (!$SVN_URL && !$url) {
214 croak "SVN repository location required: $url\n";
215 }
216 $SVN_URL ||= $url;
217 $SVN_UUID ||= $uuid;
218 setup_git_svn();
219 $latest = $rev;
220 }
221 revdb_set($REVDB, $rev, $c);
222 print "r$rev = $c\n";
223 $newest_rev = $rev if ($rev > $newest_rev);
224 }
225 close $rev_list or croak $?;
226
227 goto out if $_use_lib;
228 if (!chdir $SVN_WC) {
229 svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
230 chdir $SVN_WC or croak $!;
231 }
232
233 $pid = fork;
234 defined $pid or croak $!;
235 if ($pid == 0) {
236 my @svn_up = qw(svn up);
237 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
238 sys(@svn_up,"-r$newest_rev");
239 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
240 index_changes();
241 exec('git-write-tree') or croak $!;
242 }
243 waitpid $pid, 0;
244 croak $? if $?;
245 out:
246 if ($_upgrade) {
247 print STDERR <<"";
248 Keeping deprecated refs/head/$GIT_SVN-HEAD for now. Please remove it
249 when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
250
251 }
252 }
253
254 sub init {
255 $SVN_URL = shift or die "SVN repository location required " .
256 "as a command-line argument\n";
257 $SVN_URL =~ s!/+$!!; # strip trailing slash
258 unless (-d $GIT_DIR) {
259 my @init_db = ('git-init-db');
260 push @init_db, "--template=$_template" if defined $_template;
261 push @init_db, "--shared" if defined $_shared;
262 sys(@init_db);
263 }
264 setup_git_svn();
265 }
266
267 sub fetch {
268 check_upgrade_needed();
269 $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
270 my $ret = $_use_lib ? fetch_lib(@_) : fetch_cmd(@_);
271 if ($ret->{commit} && quiet_run(qw(git-rev-parse --verify
272 refs/heads/master^0))) {
273 sys(qw(git-update-ref refs/heads/master),$ret->{commit});
274 }
275 return $ret;
276 }
277
278 sub fetch_cmd {
279 my (@parents) = @_;
280 my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
281 unless ($_revision) {
282 $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
283 }
284 push @log_args, "-r$_revision";
285 push @log_args, '--stop-on-copy' unless $_no_stop_copy;
286
287 my $svn_log = svn_log_raw(@log_args);
288
289 my $base = next_log_entry($svn_log) or croak "No base revision!\n";
290 # don't need last_revision from grab_base_rev() because
291 # user could've specified a different revision to skip (they
292 # didn't want to import certain revisions into git for whatever
293 # reason, so trust $base->{revision} instead.
294 my (undef, $last_commit) = svn_grab_base_rev();
295 unless (-d $SVN_WC) {
296 svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
297 chdir $SVN_WC or croak $!;
298 read_uuid();
299 $last_commit = git_commit($base, @parents);
300 assert_tree($last_commit);
301 } else {
302 chdir $SVN_WC or croak $!;
303 read_uuid();
304 # looks like a user manually cp'd and svn switch'ed
305 unless ($last_commit) {
306 sys(qw/svn revert -R ./);
307 assert_svn_wc_clean($base->{revision});
308 $last_commit = git_commit($base, @parents);
309 assert_tree($last_commit);
310 }
311 }
312 my @svn_up = qw(svn up);
313 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
314 my $last = $base;
315 while (my $log_msg = next_log_entry($svn_log)) {
316 if ($last->{revision} >= $log_msg->{revision}) {
317 croak "Out of order: last >= current: ",
318 "$last->{revision} >= $log_msg->{revision}\n";
319 }
320 # Revert is needed for cases like:
321 # https://svn.musicpd.org/Jamming/trunk (r166:167), but
322 # I can't seem to reproduce something like that on a test...
323 sys(qw/svn revert -R ./);
324 assert_svn_wc_clean($last->{revision});
325 sys(@svn_up,"-r$log_msg->{revision}");
326 $last_commit = git_commit($log_msg, $last_commit, @parents);
327 $last = $log_msg;
328 }
329 close $svn_log->{fh};
330 $last->{commit} = $last_commit;
331 return $last;
332 }
333
334 sub fetch_lib {
335 my (@parents) = @_;
336 $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
337 my $repo;
338 ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
339 $SVN_LOG ||= libsvn_connect($repo);
340 $SVN ||= libsvn_connect($repo);
341 my ($last_rev, $last_commit) = svn_grab_base_rev();
342 my ($base, $head) = libsvn_parse_revision($last_rev);
343 if ($base > $head) {
344 return { revision => $last_rev, commit => $last_commit }
345 }
346 my $index = set_index($GIT_SVN_INDEX);
347
348 # limit ourselves and also fork() since get_log won't release memory
349 # after processing a revision and SVN stuff seems to leak
350 my $inc = 1000;
351 my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
352 read_uuid();
353 if (defined $last_commit) {
354 unless (-e $GIT_SVN_INDEX) {
355 sys(qw/git-read-tree/, $last_commit);
356 }
357 chomp (my $x = `git-write-tree`);
358 my ($y) = (`git-cat-file commit $last_commit`
359 =~ /^tree ($sha1)/m);
360 if ($y ne $x) {
361 unlink $GIT_SVN_INDEX or croak $!;
362 sys(qw/git-read-tree/, $last_commit);
363 }
364 chomp ($x = `git-write-tree`);
365 if ($y ne $x) {
366 print STDERR "trees ($last_commit) $y != $x\n",
367 "Something is seriously wrong...\n";
368 }
369 }
370 while (1) {
371 # fork, because using SVN::Pool with get_log() still doesn't
372 # seem to help enough to keep memory usage down.
373 defined(my $pid = fork) or croak $!;
374 if (!$pid) {
375 $SVN::Error::handler = \&libsvn_skip_unknown_revs;
376
377 # Yes I'm perfectly aware that the fourth argument
378 # below is the limit revisions number. Unfortunately
379 # performance sucks with it enabled, so it's much
380 # faster to fetch revision ranges instead of relying
381 # on the limiter.
382 $SVN_LOG->get_log( '/'.$SVN_PATH, $min, $max, 0, 1, 1,
383 sub {
384 my $log_msg;
385 if ($last_commit) {
386 $log_msg = libsvn_fetch(
387 $last_commit, @_);
388 $last_commit = git_commit(
389 $log_msg,
390 $last_commit,
391 @parents);
392 } else {
393 $log_msg = libsvn_new_tree(@_);
394 $last_commit = git_commit(
395 $log_msg, @parents);
396 }
397 });
398 exit 0;
399 }
400 waitpid $pid, 0;
401 croak $? if $?;
402 ($last_rev, $last_commit) = svn_grab_base_rev();
403 last if ($max >= $head);
404 $min = $max + 1;
405 $max += $inc;
406 $max = $head if ($max > $head);
407 }
408 restore_index($index);
409 return { revision => $last_rev, commit => $last_commit };
410 }
411
412 sub commit {
413 my (@commits) = @_;
414 check_upgrade_needed();
415 if ($_stdin || !@commits) {
416 print "Reading from stdin...\n";
417 @commits = ();
418 while (<STDIN>) {
419 if (/\b($sha1_short)\b/o) {
420 unshift @commits, $1;
421 }
422 }
423 }
424 my @revs;
425 foreach my $c (@commits) {
426 chomp(my @tmp = safe_qx('git-rev-parse',$c));
427 if (scalar @tmp == 1) {
428 push @revs, $tmp[0];
429 } elsif (scalar @tmp > 1) {
430 push @revs, reverse (safe_qx('git-rev-list',@tmp));
431 } else {
432 die "Failed to rev-parse $c\n";
433 }
434 }
435 chomp @revs;
436 $_use_lib ? commit_lib(@revs) : commit_cmd(@revs);
437 print "Done committing ",scalar @revs," revisions to SVN\n";
438 }
439
440 sub commit_cmd {
441 my (@revs) = @_;
442
443 chdir $SVN_WC or croak "Unable to chdir $SVN_WC: $!\n";
444 my $info = svn_info('.');
445 my $fetched = fetch();
446 if ($info->{Revision} != $fetched->{revision}) {
447 print STDERR "There are new revisions that were fetched ",
448 "and need to be merged (or acknowledged) ",
449 "before committing.\n";
450 exit 1;
451 }
452 $info = svn_info('.');
453 read_uuid($info);
454 my $last = $fetched;
455 foreach my $c (@revs) {
456 my $mods = svn_checkout_tree($last, $c);
457 if (scalar @$mods == 0) {
458 print "Skipping, no changes detected\n";
459 next;
460 }
461 $last = svn_commit_tree($last, $c);
462 }
463 }
464
465 sub commit_lib {
466 my (@revs) = @_;
467 my ($r_last, $cmt_last) = svn_grab_base_rev();
468 defined $r_last or die "Must have an existing revision to commit\n";
469 my $fetched = fetch();
470 if ($r_last != $fetched->{revision}) {
471 print STDERR "There are new revisions that were fetched ",
472 "and need to be merged (or acknowledged) ",
473 "before committing.\n",
474 "last rev: $r_last\n",
475 " current: $fetched->{revision}\n";
476 exit 1;
477 }
478 read_uuid();
479 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
480 my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
481
482 if (defined $LC_ALL) {
483 $ENV{LC_ALL} = $LC_ALL;
484 } else {
485 delete $ENV{LC_ALL};
486 }
487 foreach my $c (@revs) {
488 my $log_msg = get_commit_message($c, $commit_msg);
489
490 # fork for each commit because there's a memory leak I
491 # can't track down... (it's probably in the SVN code)
492 defined(my $pid = open my $fh, '-|') or croak $!;
493 if (!$pid) {
494 my $ed = SVN::Git::Editor->new(
495 { r => $r_last,
496 ra => $SVN,
497 c => $c,
498 svn_path => $SVN_PATH
499 },
500 $SVN->get_commit_editor(
501 $log_msg->{msg},
502 sub {
503 libsvn_commit_cb(
504 @_, $c,
505 $log_msg->{msg},
506 $r_last,
507 $cmt_last)
508 },
509 @lock)
510 );
511 my $mods = libsvn_checkout_tree($cmt_last, $c, $ed);
512 if (@$mods == 0) {
513 print "No changes\nr$r_last = $cmt_last\n";
514 $ed->abort_edit;
515 } else {
516 $ed->close_edit;
517 }
518 exit 0;
519 }
520 my ($r_new, $cmt_new, $no);
521 while (<$fh>) {
522 print $_;
523 chomp;
524 if (/^r(\d+) = ($sha1)$/o) {
525 ($r_new, $cmt_new) = ($1, $2);
526 } elsif ($_ eq 'No changes') {
527 $no = 1;
528 }
529 }
530 close $fh or croak $?;
531 if (! defined $r_new && ! defined $cmt_new) {
532 unless ($no) {
533 die "Failed to parse revision information\n";
534 }
535 } else {
536 ($r_last, $cmt_last) = ($r_new, $cmt_new);
537 }
538 }
539 $ENV{LC_ALL} = 'C';
540 unlink $commit_msg;
541 }
542
543 sub show_ignore {
544 $SVN_URL ||= file_to_s("$GIT_SVN_DIR/info/url");
545 $_use_lib ? show_ignore_lib() : show_ignore_cmd();
546 }
547
548 sub show_ignore_cmd {
549 require File::Find or die $!;
550 if (defined $_revision) {
551 die "-r/--revision option doesn't work unless the Perl SVN ",
552 "libraries are used\n";
553 }
554 chdir $SVN_WC or croak $!;
555 my %ign;
556 File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
557 s#^\./##;
558 @{$ign{$_}} = svn_propget_base('svn:ignore', $_);
559 }}, no_chdir=>1},'.');
560
561 print "\n# /\n";
562 foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
563 delete $ign{'.'};
564 foreach my $i (sort keys %ign) {
565 print "\n# ",$i,"\n";
566 foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
567 }
568 }
569
570 sub show_ignore_lib {
571 my $repo;
572 ($repo, $SVN_PATH) = repo_path_split($SVN_URL);
573 $SVN ||= libsvn_connect($repo);
574 my $r = defined $_revision ? $_revision : $SVN->get_latest_revnum;
575 libsvn_traverse_ignore(\*STDOUT, $SVN_PATH, $r);
576 }
577
578 sub graft_branches {
579 my $gr_file = "$GIT_DIR/info/grafts";
580 my ($grafts, $comments) = read_grafts($gr_file);
581 my $gr_sha1;
582
583 if (%$grafts) {
584 # temporarily disable our grafts file to make this idempotent
585 chomp($gr_sha1 = safe_qx(qw/git-hash-object -w/,$gr_file));
586 rename $gr_file, "$gr_file~$gr_sha1" or croak $!;
587 }
588
589 my $l_map = read_url_paths();
590 my @re = map { qr/$_/is } @_opt_m if @_opt_m;
591 unless ($_no_default_regex) {
592 push @re, ( qr/\b(?:merge|merging|merged)\s+(\S.+)/is,
593 qr/\b(?:from|of)\s+(\S.+)/is );
594 }
595 foreach my $u (keys %$l_map) {
596 if (@re) {
597 foreach my $p (keys %{$l_map->{$u}}) {
598 graft_merge_msg($grafts,$l_map,$u,$p);
599 }
600 }
601 unless ($_no_graft_copy) {
602 if ($_use_lib) {
603 graft_file_copy_lib($grafts,$l_map,$u);
604 } else {
605 graft_file_copy_cmd($grafts,$l_map,$u);
606 }
607 }
608 }
609
610 write_grafts($grafts, $comments, $gr_file);
611 unlink "$gr_file~$gr_sha1" if $gr_sha1;
612 }
613
614 sub multi_init {
615 my $url = shift;
616 $_trunk ||= 'trunk';
617 $_trunk =~ s#/+$##;
618 $url =~ s#/+$## if $url;
619 if ($_trunk !~ m#^[a-z\+]+://#) {
620 $_trunk = '/' . $_trunk if ($_trunk !~ m#^/#);
621 unless ($url) {
622 print STDERR "E: '$_trunk' is not a complete URL ",
623 "and a separate URL is not specified\n";
624 exit 1;
625 }
626 $_trunk = $url . $_trunk;
627 }
628 if ($GIT_SVN eq 'git-svn') {
629 print "GIT_SVN_ID set to 'trunk' for $_trunk\n";
630 $GIT_SVN = $ENV{GIT_SVN_ID} = 'trunk';
631 }
632 init_vars();
633 init($_trunk);
634 complete_url_ls_init($url, $_branches, '--branches/-b', '');
635 complete_url_ls_init($url, $_tags, '--tags/-t', 'tags/');
636 }
637
638 sub multi_fetch {
639 # try to do trunk first, since branches/tags
640 # may be descended from it.
641 if (-e "$GIT_DIR/svn/trunk/info/url") {
642 fetch_child_id('trunk', @_);
643 }
644 rec_fetch('', "$GIT_DIR/svn", @_);
645 }
646
647 sub show_log {
648 my (@args) = @_;
649 my ($r_min, $r_max);
650 my $r_last = -1; # prevent dupes
651 rload_authors() if $_authors;
652 if (defined $TZ) {
653 $ENV{TZ} = $TZ;
654 } else {
655 delete $ENV{TZ};
656 }
657 if (defined $_revision) {
658 if ($_revision =~ /^(\d+):(\d+)$/) {
659 ($r_min, $r_max) = ($1, $2);
660 } elsif ($_revision =~ /^\d+$/) {
661 $r_min = $r_max = $_revision;
662 } else {
663 print STDERR "-r$_revision is not supported, use ",
664 "standard \'git log\' arguments instead\n";
665 exit 1;
666 }
667 }
668
669 my $pid = open(my $log,'-|');
670 defined $pid or croak $!;
671 if (!$pid) {
672 exec(git_svn_log_cmd($r_min,$r_max), @args) or croak $!;
673 }
674 setup_pager();
675 my (@k, $c, $d);
676
677 while (<$log>) {
678 if (/^commit ($sha1_short)/o) {
679 my $cmt = $1;
680 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
681 $r_last = $c->{r};
682 process_commit($c, $r_min, $r_max, \@k) or
683 goto out;
684 }
685 $d = undef;
686 $c = { c => $cmt };
687 } elsif (/^author (.+) (\d+) ([\-\+]?\d+)$/) {
688 get_author_info($c, $1, $2, $3);
689 } elsif (/^(?:tree|parent|committer) /) {
690 # ignore
691 } elsif (/^:\d{6} \d{6} $sha1_short/o) {
692 push @{$c->{raw}}, $_;
693 } elsif (/^diff /) {
694 $d = 1;
695 push @{$c->{diff}}, $_;
696 } elsif ($d) {
697 push @{$c->{diff}}, $_;
698 } elsif (/^ (git-svn-id:.+)$/) {
699 (undef, $c->{r}, undef) = extract_metadata($1);
700 } elsif (s/^ //) {
701 push @{$c->{l}}, $_;
702 }
703 }
704 if ($c && defined $c->{r} && $c->{r} != $r_last) {
705 $r_last = $c->{r};
706 process_commit($c, $r_min, $r_max, \@k);
707 }
708 if (@k) {
709 my $swap = $r_max;
710 $r_max = $r_min;
711 $r_min = $swap;
712 process_commit($_, $r_min, $r_max) foreach reverse @k;
713 }
714 out:
715 close $log;
716 print '-' x72,"\n" unless $_incremental || $_oneline;
717 }
718
719 ########################### utility functions #########################
720
721 sub cmt_showable {
722 my ($c) = @_;
723 return 1 if defined $c->{r};
724 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
725 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
726 my @msg = safe_qx(qw/git-cat-file commit/, $c->{c});
727 shift @msg while ($msg[0] ne "\n");
728 shift @msg;
729 @{$c->{l}} = grep !/^git-svn-id: /, @msg;
730
731 (undef, $c->{r}, undef) = extract_metadata(
732 (grep(/^git-svn-id: /, @msg))[-1]);
733 }
734 return defined $c->{r};
735 }
736
737 sub git_svn_log_cmd {
738 my ($r_min, $r_max) = @_;
739 my @cmd = (qw/git-log --abbrev-commit --pretty=raw
740 --default/, "refs/remotes/$GIT_SVN");
741 push @cmd, '--summary' if $_verbose;
742 return @cmd unless defined $r_max;
743 if ($r_max == $r_min) {
744 push @cmd, '--max-count=1';
745 if (my $c = revdb_get($REVDB, $r_max)) {
746 push @cmd, $c;
747 }
748 } else {
749 my ($c_min, $c_max);
750 $c_max = revdb_get($REVDB, $r_max);
751 $c_min = revdb_get($REVDB, $r_min);
752 if ($c_min && $c_max) {
753 if ($r_max > $r_max) {
754 push @cmd, "$c_min..$c_max";
755 } else {
756 push @cmd, "$c_max..$c_min";
757 }
758 } elsif ($r_max > $r_min) {
759 push @cmd, $c_max;
760 } else {
761 push @cmd, $c_min;
762 }
763 }
764 return @cmd;
765 }
766
767 sub fetch_child_id {
768 my $id = shift;
769 print "Fetching $id\n";
770 my $ref = "$GIT_DIR/refs/remotes/$id";
771 my $ca = file_to_s($ref) if (-r $ref);
772 defined(my $pid = fork) or croak $!;
773 if (!$pid) {
774 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
775 init_vars();
776 fetch(@_);
777 exit 0;
778 }
779 waitpid $pid, 0;
780 croak $? if $?;
781 return unless $_repack || -r $ref;
782
783 my $cb = file_to_s($ref);
784
785 defined($pid = open my $fh, '-|') or croak $!;
786 my $url = file_to_s("$GIT_DIR/svn/$id/info/url");
787 $url = qr/\Q$url\E/;
788 if (!$pid) {
789 exec qw/git-rev-list --pretty=raw/,
790 $ca ? "$ca..$cb" : $cb or croak $!;
791 }
792 while (<$fh>) {
793 if (/^ git-svn-id: $url\@\d+ [a-f0-9\-]+$/) {
794 check_repack();
795 } elsif (/^ git-svn-id: \S+\@\d+ [a-f0-9\-]+$/) {
796 last;
797 }
798 }
799 close $fh;
800 }
801
802 sub rec_fetch {
803 my ($pfx, $p, @args) = @_;
804 my @dir;
805 foreach (sort <$p/*>) {
806 if (-r "$_/info/url") {
807 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
808 my $id = $pfx . basename $_;
809 next if $id eq 'trunk';
810 fetch_child_id($id, @args);
811 } elsif (-d $_) {
812 push @dir, $_;
813 }
814 }
815 foreach (@dir) {
816 my $x = $_;
817 $x =~ s!^\Q$GIT_DIR\E/svn/!!;
818 rec_fetch($x, $_);
819 }
820 }
821
822 sub complete_url_ls_init {
823 my ($url, $var, $switch, $pfx) = @_;
824 unless ($var) {
825 print STDERR "W: $switch not specified\n";
826 return;
827 }
828 $var =~ s#/+$##;
829 if ($var !~ m#^[a-z\+]+://#) {
830 $var = '/' . $var if ($var !~ m#^/#);
831 unless ($url) {
832 print STDERR "E: '$var' is not a complete URL ",
833 "and a separate URL is not specified\n";
834 exit 1;
835 }
836 $var = $url . $var;
837 }
838 chomp(my @ls = $_use_lib ? libsvn_ls_fullurl($var)
839 : safe_qx(qw/svn ls --non-interactive/, $var));
840 my $old = $GIT_SVN;
841 defined(my $pid = fork) or croak $!;
842 if (!$pid) {
843 foreach my $u (map { "$var/$_" } (grep m!/$!, @ls)) {
844 $u =~ s#/+$##;
845 if ($u !~ m!\Q$var\E/(.+)$!) {
846 print STDERR "W: Unrecognized URL: $u\n";
847 die "This should never happen\n";
848 }
849 my $id = $pfx.$1;
850 print "init $u => $id\n";
851 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
852 init_vars();
853 init($u);
854 }
855 exit 0;
856 }
857 waitpid $pid, 0;
858 croak $? if $?;
859 }
860
861 sub common_prefix {
862 my $paths = shift;
863 my %common;
864 foreach (@$paths) {
865 my @tmp = split m#/#, $_;
866 my $p = '';
867 while (my $x = shift @tmp) {
868 $p .= "/$x";
869 $common{$p} ||= 0;
870 $common{$p}++;
871 }
872 }
873 foreach (sort {length $b <=> length $a} keys %common) {
874 if ($common{$_} == @$paths) {
875 return $_;
876 }
877 }
878 return '';
879 }
880
881 # this isn't funky-filename safe, but good enough for now...
882 sub graft_file_copy_cmd {
883 my ($grafts, $l_map, $u) = @_;
884 my $paths = $l_map->{$u};
885 my $pfx = common_prefix([keys %$paths]);
886 $SVN_URL ||= $u.$pfx;
887 my $pid = open my $fh, '-|';
888 defined $pid or croak $!;
889 unless ($pid) {
890 my @exec = qw/svn log -v/;
891 push @exec, "-r$_revision" if defined $_revision;
892 exec @exec, $u.$pfx or croak $!;
893 }
894 my ($r, $mp) = (undef, undef);
895 while (<$fh>) {
896 chomp;
897 if (/^\-{72}$/) {
898 $mp = $r = undef;
899 } elsif (/^r(\d+) \| /) {
900 $r = $1 unless defined $r;
901 } elsif (/^Changed paths:/) {
902 $mp = 1;
903 } elsif ($mp && m#^ [AR] /(\S.*?) \(from /(\S+?):(\d+)\)$#) {
904 my ($p1, $p0, $r0) = ($1, $2, $3);
905 my $c = find_graft_path_commit($paths, $p1, $r);
906 next unless $c;
907 find_graft_path_parents($grafts, $paths, $c, $p0, $r0);
908 }
909 }
910 }
911
912 sub graft_file_copy_lib {
913 my ($grafts, $l_map, $u) = @_;
914 my $tree_paths = $l_map->{$u};
915 my $pfx = common_prefix([keys %$tree_paths]);
916 my ($repo, $path) = repo_path_split($u.$pfx);
917 $SVN_LOG ||= libsvn_connect($repo);
918 $SVN ||= libsvn_connect($repo);
919
920 my ($base, $head) = libsvn_parse_revision();
921 my $inc = 1000;
922 my ($min, $max) = ($base, $head < $base+$inc ? $head : $base+$inc);
923 my $eh = $SVN::Error::handler;
924 $SVN::Error::handler = \&libsvn_skip_unknown_revs;
925 while (1) {
926 my $pool = SVN::Pool->new;
927 $SVN_LOG->get_log( "/$path", $min, $max, 0, 1, 1,
928 sub {
929 libsvn_graft_file_copies($grafts, $tree_paths,
930 $path, @_);
931 }, $pool);
932 $pool->clear;
933 last if ($max >= $head);
934 $min = $max + 1;
935 $max += $inc;
936 $max = $head if ($max > $head);
937 }
938 $SVN::Error::handler = $eh;
939 }
940
941 sub process_merge_msg_matches {
942 my ($grafts, $l_map, $u, $p, $c, @matches) = @_;
943 my (@strong, @weak);
944 foreach (@matches) {
945 # merging with ourselves is not interesting
946 next if $_ eq $p;
947 if ($l_map->{$u}->{$_}) {
948 push @strong, $_;
949 } else {
950 push @weak, $_;
951 }
952 }
953 foreach my $w (@weak) {
954 last if @strong;
955 # no exact match, use branch name as regexp.
956 my $re = qr/\Q$w\E/i;
957 foreach (keys %{$l_map->{$u}}) {
958 if (/$re/) {
959 push @strong, $_;
960 last;
961 }
962 }
963 last if @strong;
964 $w = basename($w);
965 $re = qr/\Q$w\E/i;
966 foreach (keys %{$l_map->{$u}}) {
967 if (/$re/) {
968 push @strong, $_;
969 last;
970 }
971 }
972 }
973 my ($rev) = ($c->{m} =~ /^git-svn-id:\s(?:\S+?)\@(\d+)
974 \s(?:[a-f\d\-]+)$/xsm);
975 unless (defined $rev) {
976 ($rev) = ($c->{m} =~/^git-svn-id:\s(\d+)
977 \@(?:[a-f\d\-]+)/xsm);
978 return unless defined $rev;
979 }
980 foreach my $m (@strong) {
981 my ($r0, $s0) = find_rev_before($rev, $m);
982 $grafts->{$c->{c}}->{$s0} = 1 if defined $s0;
983 }
984 }
985
986 sub graft_merge_msg {
987 my ($grafts, $l_map, $u, $p, @re) = @_;
988
989 my $x = $l_map->{$u}->{$p};
990 my $rl = rev_list_raw($x);
991 while (my $c = next_rev_list_entry($rl)) {
992 foreach my $re (@re) {
993 my (@br) = ($c->{m} =~ /$re/g);
994 next unless @br;
995 process_merge_msg_matches($grafts,$l_map,$u,$p,$c,@br);
996 }
997 }
998 }
999
1000 sub read_uuid {
1001 return if $SVN_UUID;
1002 if ($_use_lib) {
1003 my $pool = SVN::Pool->new;
1004 $SVN_UUID = $SVN->get_uuid($pool);
1005 $pool->clear;
1006 } else {
1007 my $info = shift || svn_info('.');
1008 $SVN_UUID = $info->{'Repository UUID'} or
1009 croak "Repository UUID unreadable\n";
1010 }
1011 }
1012
1013 sub quiet_run {
1014 my $pid = fork;
1015 defined $pid or croak $!;
1016 if (!$pid) {
1017 open my $null, '>', '/dev/null' or croak $!;
1018 open STDERR, '>&', $null or croak $!;
1019 open STDOUT, '>&', $null or croak $!;
1020 exec @_ or croak $!;
1021 }
1022 waitpid $pid, 0;
1023 return $?;
1024 }
1025
1026 sub repo_path_split {
1027 my $full_url = shift;
1028 $full_url =~ s#/+$##;
1029
1030 foreach (@repo_path_split_cache) {
1031 if ($full_url =~ s#$_##) {
1032 my $u = $1;
1033 $full_url =~ s#^/+##;
1034 return ($u, $full_url);
1035 }
1036 }
1037
1038 my ($url, $path) = ($full_url =~ m!^([a-z\+]+://[^/]*)(.*)$!i);
1039 $path =~ s#^/+##;
1040 my @paths = split(m#/+#, $path);
1041
1042 if ($_use_lib) {
1043 while (1) {
1044 $SVN = libsvn_connect($url);
1045 last if (defined $SVN &&
1046 defined eval { $SVN->get_latest_revnum });
1047 my $n = shift @paths || last;
1048 $url .= "/$n";
1049 }
1050 } else {
1051 while (quiet_run(qw/svn ls --non-interactive/, $url)) {
1052 my $n = shift @paths || last;
1053 $url .= "/$n";
1054 }
1055 }
1056 push @repo_path_split_cache, qr/^(\Q$url\E)/;
1057 $path = join('/',@paths);
1058 return ($url, $path);
1059 }
1060
1061 sub setup_git_svn {
1062 defined $SVN_URL or croak "SVN repository location required\n";
1063 unless (-d $GIT_DIR) {
1064 croak "GIT_DIR=$GIT_DIR does not exist!\n";
1065 }
1066 mkpath([$GIT_SVN_DIR]);
1067 mkpath(["$GIT_SVN_DIR/info"]);
1068 open my $fh, '>>',$REVDB or croak $!;
1069 close $fh;
1070 s_to_file($SVN_URL,"$GIT_SVN_DIR/info/url");
1071
1072 }
1073
1074 sub assert_svn_wc_clean {
1075 return if $_use_lib;
1076 my ($svn_rev) = @_;
1077 croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
1078 my $lcr = svn_info('.')->{'Last Changed Rev'};
1079 if ($svn_rev != $lcr) {
1080 print STDERR "Checking for copy-tree ... ";
1081 my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
1082 "-r$lcr:$svn_rev")));
1083 if (@diff) {
1084 croak "Nope! Expected r$svn_rev, got r$lcr\n";
1085 } else {
1086 print STDERR "OK!\n";
1087 }
1088 }
1089 my @status = grep(!/^Performing status on external/,(`svn status`));
1090 @status = grep(!/^\s*$/,@status);
1091 if (scalar @status) {
1092 print STDERR "Tree ($SVN_WC) is not clean:\n";
1093 print STDERR $_ foreach @status;
1094 croak;
1095 }
1096 }
1097
1098 sub get_tree_from_treeish {
1099 my ($treeish) = @_;
1100 croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
1101 chomp(my $type = `git-cat-file -t $treeish`);
1102 my $expected;
1103 while ($type eq 'tag') {
1104 chomp(($treeish, $type) = `git-cat-file tag $treeish`);
1105 }
1106 if ($type eq 'commit') {
1107 $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
1108 ($expected) = ($expected =~ /^tree ($sha1)$/);
1109 die "Unable to get tree from $treeish\n" unless $expected;
1110 } elsif ($type eq 'tree') {
1111 $expected = $treeish;
1112 } else {
1113 die "$treeish is a $type, expected tree, tag or commit\n";
1114 }
1115 return $expected;
1116 }
1117
1118 sub assert_tree {
1119 return if $_use_lib;
1120 my ($treeish) = @_;
1121 my $expected = get_tree_from_treeish($treeish);
1122
1123 my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
1124 if (-e $tmpindex) {
1125 unlink $tmpindex or croak $!;
1126 }
1127 my $old_index = set_index($tmpindex);
1128 index_changes(1);
1129 chomp(my $tree = `git-write-tree`);
1130 restore_index($old_index);
1131 if ($tree ne $expected) {
1132 croak "Tree mismatch, Got: $tree, Expected: $expected\n";
1133 }
1134 unlink $tmpindex;
1135 }
1136
1137 sub parse_diff_tree {
1138 my $diff_fh = shift;
1139 local $/ = "\0";
1140 my $state = 'meta';
1141 my @mods;
1142 while (<$diff_fh>) {
1143 chomp $_; # this gets rid of the trailing "\0"
1144 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
1145 $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
1146 push @mods, { mode_a => $1, mode_b => $2,
1147 sha1_b => $3, chg => $4 };
1148 if ($4 =~ /^(?:C|R)$/) {
1149 $state = 'file_a';
1150 } else {
1151 $state = 'file_b';
1152 }
1153 } elsif ($state eq 'file_a') {
1154 my $x = $mods[$#mods] or croak "Empty array\n";
1155 if ($x->{chg} !~ /^(?:C|R)$/) {
1156 croak "Error parsing $_, $x->{chg}\n";
1157 }
1158 $x->{file_a} = $_;
1159 $state = 'file_b';
1160 } elsif ($state eq 'file_b') {
1161 my $x = $mods[$#mods] or croak "Empty array\n";
1162 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
1163 croak "Error parsing $_, $x->{chg}\n";
1164 }
1165 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
1166 croak "Error parsing $_, $x->{chg}\n";
1167 }
1168 $x->{file_b} = $_;
1169 $state = 'meta';
1170 } else {
1171 croak "Error parsing $_\n";
1172 }
1173 }
1174 close $diff_fh or croak $?;
1175
1176 return \@mods;
1177 }
1178
1179 sub svn_check_prop_executable {
1180 my $m = shift;
1181 return if -l $m->{file_b};
1182 if ($m->{mode_b} =~ /755$/) {
1183 chmod((0755 &~ umask),$m->{file_b}) or croak $!;
1184 if ($m->{mode_a} !~ /755$/) {
1185 sys(qw(svn propset svn:executable 1), $m->{file_b});
1186 }
1187 -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
1188 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
1189 sys(qw(svn propdel svn:executable), $m->{file_b});
1190 chmod((0644 &~ umask),$m->{file_b}) or croak $!;
1191 -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
1192 }
1193 }
1194
1195 sub svn_ensure_parent_path {
1196 my $dir_b = dirname(shift);
1197 svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
1198 mkpath([$dir_b]) unless (-d $dir_b);
1199 sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
1200 }
1201
1202 sub precommit_check {
1203 my $mods = shift;
1204 my (%rm_file, %rmdir_check, %added_check);
1205
1206 my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
1207 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1208 if ($m->{chg} eq 'R') {
1209 if (-d $m->{file_b}) {
1210 err_dir_to_file("$m->{file_a} => $m->{file_b}");
1211 }
1212 # dir/$file => dir/file/$file
1213 my $dirname = dirname($m->{file_b});
1214 while ($dirname ne File::Spec->curdir) {
1215 if ($dirname ne $m->{file_a}) {
1216 $dirname = dirname($dirname);
1217 next;
1218 }
1219 err_file_to_dir("$m->{file_a} => $m->{file_b}");
1220 }
1221 # baz/zzz => baz (baz is a file)
1222 $dirname = dirname($m->{file_a});
1223 while ($dirname ne File::Spec->curdir) {
1224 if ($dirname ne $m->{file_b}) {
1225 $dirname = dirname($dirname);
1226 next;
1227 }
1228 err_dir_to_file("$m->{file_a} => $m->{file_b}");
1229 }
1230 }
1231 if ($m->{chg} =~ /^(D|R)$/) {
1232 my $t = $1 eq 'D' ? 'file_b' : 'file_a';
1233 $rm_file{ $m->{$t} } = 1;
1234 my $dirname = dirname( $m->{$t} );
1235 my $basename = basename( $m->{$t} );
1236 $rmdir_check{$dirname}->{$basename} = 1;
1237 } elsif ($m->{chg} =~ /^(?:A|C)$/) {
1238 if (-d $m->{file_b}) {
1239 err_dir_to_file($m->{file_b});
1240 }
1241 my $dirname = dirname( $m->{file_b} );
1242 my $basename = basename( $m->{file_b} );
1243 $added_check{$dirname}->{$basename} = 1;
1244 while ($dirname ne File::Spec->curdir) {
1245 if ($rm_file{$dirname}) {
1246 err_file_to_dir($m->{file_b});
1247 }
1248 $dirname = dirname $dirname;
1249 }
1250 }
1251 }
1252 return (\%rmdir_check, \%added_check);
1253
1254 sub err_dir_to_file {
1255 my $file = shift;
1256 print STDERR "Node change from directory to file ",
1257 "is not supported by Subversion: ",$file,"\n";
1258 exit 1;
1259 }
1260 sub err_file_to_dir {
1261 my $file = shift;
1262 print STDERR "Node change from file to directory ",
1263 "is not supported by Subversion: ",$file,"\n";
1264 exit 1;
1265 }
1266 }
1267
1268
1269 sub get_diff {
1270 my ($from, $treeish) = @_;
1271 assert_tree($from);
1272 print "diff-tree $from $treeish\n";
1273 my $pid = open my $diff_fh, '-|';
1274 defined $pid or croak $!;
1275 if ($pid == 0) {
1276 my @diff_tree = qw(git-diff-tree -z -r);
1277 if ($_cp_similarity) {
1278 push @diff_tree, "-C$_cp_similarity";
1279 } else {
1280 push @diff_tree, '-C';
1281 }
1282 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
1283 push @diff_tree, "-l$_l" if defined $_l;
1284 exec(@diff_tree, $from, $treeish) or croak $!;
1285 }
1286 return parse_diff_tree($diff_fh);
1287 }
1288
1289 sub svn_checkout_tree {
1290 my ($from, $treeish) = @_;
1291 my $mods = get_diff($from->{commit}, $treeish);
1292 return $mods unless (scalar @$mods);
1293 my ($rm, $add) = precommit_check($mods);
1294
1295 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1296 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1297 if ($m->{chg} eq 'C') {
1298 svn_ensure_parent_path( $m->{file_b} );
1299 sys(qw(svn cp), $m->{file_a}, $m->{file_b});
1300 apply_mod_line_blob($m);
1301 svn_check_prop_executable($m);
1302 } elsif ($m->{chg} eq 'D') {
1303 sys(qw(svn rm --force), $m->{file_b});
1304 } elsif ($m->{chg} eq 'R') {
1305 svn_ensure_parent_path( $m->{file_b} );
1306 sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
1307 apply_mod_line_blob($m);
1308 svn_check_prop_executable($m);
1309 } elsif ($m->{chg} eq 'M') {
1310 apply_mod_line_blob($m);
1311 svn_check_prop_executable($m);
1312 } elsif ($m->{chg} eq 'T') {
1313 sys(qw(svn rm --force),$m->{file_b});
1314 apply_mod_line_blob($m);
1315 sys(qw(svn add), $m->{file_b});
1316 svn_check_prop_executable($m);
1317 } elsif ($m->{chg} eq 'A') {
1318 svn_ensure_parent_path( $m->{file_b} );
1319 apply_mod_line_blob($m);
1320 sys(qw(svn add), $m->{file_b});
1321 svn_check_prop_executable($m);
1322 } else {
1323 croak "Invalid chg: $m->{chg}\n";
1324 }
1325 }
1326
1327 assert_tree($treeish);
1328 if ($_rmdir) { # remove empty directories
1329 handle_rmdir($rm, $add);
1330 }
1331 assert_tree($treeish);
1332 return $mods;
1333 }
1334
1335 sub libsvn_checkout_tree {
1336 my ($from, $treeish, $ed) = @_;
1337 my $mods = get_diff($from, $treeish);
1338 return $mods unless (scalar @$mods);
1339 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
1340 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
1341 my $f = $m->{chg};
1342 if (defined $o{$f}) {
1343 $ed->$f($m);
1344 } else {
1345 croak "Invalid change type: $f\n";
1346 }
1347 }
1348 $ed->rmdirs if $_rmdir;
1349 return $mods;
1350 }
1351
1352 # svn ls doesn't work with respect to the current working tree, but what's
1353 # in the repository. There's not even an option for it... *sigh*
1354 # (added files don't show up and removed files remain in the ls listing)
1355 sub svn_ls_current {
1356 my ($dir, $rm, $add) = @_;
1357 chomp(my @ls = safe_qx('svn','ls',$dir));
1358 my @ret = ();
1359 foreach (@ls) {
1360 s#/$##; # trailing slashes are evil
1361 push @ret, $_ unless $rm->{$dir}->{$_};
1362 }
1363 if (exists $add->{$dir}) {
1364 push @ret, keys %{$add->{$dir}};
1365 }
1366 return \@ret;
1367 }
1368
1369 sub handle_rmdir {
1370 my ($rm, $add) = @_;
1371
1372 foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
1373 my $ls = svn_ls_current($dir, $rm, $add);
1374 next if (scalar @$ls);
1375 sys(qw(svn rm --force),$dir);
1376
1377 my $dn = dirname $dir;
1378 $rm->{ $dn }->{ basename $dir } = 1;
1379 $ls = svn_ls_current($dn, $rm, $add);
1380 while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
1381 sys(qw(svn rm --force),$dn);
1382 $dir = basename $dn;
1383 $dn = dirname $dn;
1384 $rm->{ $dn }->{ $dir } = 1;
1385 $ls = svn_ls_current($dn, $rm, $add);
1386 }
1387 }
1388 }
1389
1390 sub get_commit_message {
1391 my ($commit, $commit_msg) = (@_);
1392 my %log_msg = ( msg => '' );
1393 open my $msg, '>', $commit_msg or croak $!;
1394
1395 print "commit: $commit\n";
1396 chomp(my $type = `git-cat-file -t $commit`);
1397 if ($type eq 'commit') {
1398 my $pid = open my $msg_fh, '-|';
1399 defined $pid or croak $!;
1400
1401 if ($pid == 0) {
1402 exec(qw(git-cat-file commit), $commit) or croak $!;
1403 }
1404 my $in_msg = 0;
1405 while (<$msg_fh>) {
1406 if (!$in_msg) {
1407 $in_msg = 1 if (/^\s*$/);
1408 } elsif (/^git-svn-id: /) {
1409 # skip this, we regenerate the correct one
1410 # on re-fetch anyways
1411 } else {
1412 print $msg $_ or croak $!;
1413 }
1414 }
1415 close $msg_fh or croak $?;
1416 }
1417 close $msg or croak $!;
1418
1419 if ($_edit || ($type eq 'tree')) {
1420 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1421 system($editor, $commit_msg);
1422 }
1423
1424 # file_to_s removes all trailing newlines, so just use chomp() here:
1425 open $msg, '<', $commit_msg or croak $!;
1426 { local $/; chomp($log_msg{msg} = <$msg>); }
1427 close $msg or croak $!;
1428
1429 return \%log_msg;
1430 }
1431
1432 sub svn_commit_tree {
1433 my ($last, $commit) = @_;
1434 my $commit_msg = "$GIT_SVN_DIR/.svn-commit.tmp.$$";
1435 my $log_msg = get_commit_message($commit, $commit_msg);
1436 my ($oneline) = ($log_msg->{msg} =~ /([^\n\r]+)/);
1437 print "Committing $commit: $oneline\n";
1438
1439 if (defined $LC_ALL) {
1440 $ENV{LC_ALL} = $LC_ALL;
1441 } else {
1442 delete $ENV{LC_ALL};
1443 }
1444 my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
1445 $ENV{LC_ALL} = 'C';
1446 unlink $commit_msg;
1447 my ($committed) = ($ci_output[$#ci_output] =~ /(\d+)/);
1448 if (!defined $committed) {
1449 my $out = join("\n",@ci_output);
1450 print STDERR "W: Trouble parsing \`svn commit' output:\n\n",
1451 $out, "\n\nAssuming English locale...";
1452 ($committed) = ($out =~ /^Committed revision \d+\./sm);
1453 defined $committed or die " FAILED!\n",
1454 "Commit output failed to parse committed revision!\n",
1455 print STDERR " OK\n";
1456 }
1457
1458 my @svn_up = qw(svn up);
1459 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
1460 if ($_optimize_commits && ($committed == ($last->{revision} + 1))) {
1461 push @svn_up, "-r$committed";
1462 sys(@svn_up);
1463 my $info = svn_info('.');
1464 my $date = $info->{'Last Changed Date'} or die "Missing date\n";
1465 if ($info->{'Last Changed Rev'} != $committed) {
1466 croak "$info->{'Last Changed Rev'} != $committed\n"
1467 }
1468 my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1469 /(\d{4})\-(\d\d)\-(\d\d)\s
1470 (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1471 or croak "Failed to parse date: $date\n";
1472 $log_msg->{date} = "$tz $Y-$m-$d $H:$M:$S";
1473 $log_msg->{author} = $info->{'Last Changed Author'};
1474 $log_msg->{revision} = $committed;
1475 $log_msg->{msg} .= "\n";
1476 $log_msg->{parents} = [ $last->{commit} ];
1477 $log_msg->{commit} = git_commit($log_msg, $commit);
1478 return $log_msg;
1479 }
1480 # resync immediately
1481 push @svn_up, "-r$last->{revision}";
1482 sys(@svn_up);
1483 return fetch("$committed=$commit");
1484 }
1485
1486 sub rev_list_raw {
1487 my (@args) = @_;
1488 my $pid = open my $fh, '-|';
1489 defined $pid or croak $!;
1490 if (!$pid) {
1491 exec(qw/git-rev-list --pretty=raw/, @args) or croak $!;
1492 }
1493 return { fh => $fh, t => { } };
1494 }
1495
1496 sub next_rev_list_entry {
1497 my $rl = shift;
1498 my $fh = $rl->{fh};
1499 my $x = $rl->{t};
1500 while (<$fh>) {
1501 if (/^commit ($sha1)$/o) {
1502 if ($x->{c}) {
1503 $rl->{t} = { c => $1 };
1504 return $x;
1505 } else {
1506 $x->{c} = $1;
1507 }
1508 } elsif (/^parent ($sha1)$/o) {
1509 $x->{p}->{$1} = 1;
1510 } elsif (s/^ //) {
1511 $x->{m} ||= '';
1512 $x->{m} .= $_;
1513 }
1514 }
1515 return ($x != $rl->{t}) ? $x : undef;
1516 }
1517
1518 # read the entire log into a temporary file (which is removed ASAP)
1519 # and store the file handle + parser state
1520 sub svn_log_raw {
1521 my (@log_args) = @_;
1522 my $log_fh = IO::File->new_tmpfile or croak $!;
1523 my $pid = fork;
1524 defined $pid or croak $!;
1525 if (!$pid) {
1526 open STDOUT, '>&', $log_fh or croak $!;
1527 exec (qw(svn log), @log_args) or croak $!
1528 }
1529 waitpid $pid, 0;
1530 croak $? if $?;
1531 seek $log_fh, 0, 0 or croak $!;
1532 return { state => 'sep', fh => $log_fh };
1533 }
1534
1535 sub next_log_entry {
1536 my $log = shift; # retval of svn_log_raw()
1537 my $ret = undef;
1538 my $fh = $log->{fh};
1539
1540 while (<$fh>) {
1541 chomp;
1542 if (/^\-{72}$/) {
1543 if ($log->{state} eq 'msg') {
1544 if ($ret->{lines}) {
1545 $ret->{msg} .= $_."\n";
1546 unless(--$ret->{lines}) {
1547 $log->{state} = 'sep';
1548 }
1549 } else {
1550 croak "Log parse error at: $_\n",
1551 $ret->{revision},
1552 "\n";
1553 }
1554 next;
1555 }
1556 if ($log->{state} ne 'sep') {
1557 croak "Log parse error at: $_\n",
1558 "state: $log->{state}\n",
1559 $ret->{revision},
1560 "\n";
1561 }
1562 $log->{state} = 'rev';
1563
1564 # if we have an empty log message, put something there:
1565 if ($ret) {
1566 $ret->{msg} ||= "\n";
1567 delete $ret->{lines};
1568 return $ret;
1569 }
1570 next;
1571 }
1572 if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
1573 my $rev = $1;
1574 my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
1575 ($lines) = ($lines =~ /(\d+)/);
1576 my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
1577 /(\d{4})\-(\d\d)\-(\d\d)\s
1578 (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
1579 or croak "Failed to parse date: $date\n";
1580 $ret = { revision => $rev,
1581 date => "$tz $Y-$m-$d $H:$M:$S",
1582 author => $author,
1583 lines => $lines,
1584 msg => '' };
1585 if (defined $_authors && ! defined $users{$author}) {
1586 die "Author: $author not defined in ",
1587 "$_authors file\n";
1588 }
1589 $log->{state} = 'msg_start';
1590 next;
1591 }
1592 # skip the first blank line of the message:
1593 if ($log->{state} eq 'msg_start' && /^$/) {
1594 $log->{state} = 'msg';
1595 } elsif ($log->{state} eq 'msg') {
1596 if ($ret->{lines}) {
1597 $ret->{msg} .= $_."\n";
1598 unless (--$ret->{lines}) {
1599 $log->{state} = 'sep';
1600 }
1601 } else {
1602 croak "Log parse error at: $_\n",
1603 $ret->{revision},"\n";
1604 }
1605 }
1606 }
1607 return $ret;
1608 }
1609
1610 sub svn_info {
1611 my $url = shift || $SVN_URL;
1612
1613 my $pid = open my $info_fh, '-|';
1614 defined $pid or croak $!;
1615
1616 if ($pid == 0) {
1617 exec(qw(svn info),$url) or croak $!;
1618 }
1619
1620 my $ret = {};
1621 # only single-lines seem to exist in svn info output
1622 while (<$info_fh>) {
1623 chomp $_;
1624 if (m#^([^:]+)\s*:\s*(\S.*)$#) {
1625 $ret->{$1} = $2;
1626 push @{$ret->{-order}}, $1;
1627 }
1628 }
1629 close $info_fh or croak $?;
1630 return $ret;
1631 }
1632
1633 sub sys { system(@_) == 0 or croak $? }
1634
1635 sub eol_cp {
1636 my ($from, $to) = @_;
1637 my $es = svn_propget_base('svn:eol-style', $to);
1638 open my $rfd, '<', $from or croak $!;
1639 binmode $rfd or croak $!;
1640 open my $wfd, '>', $to or croak $!;
1641 binmode $wfd or croak $!;
1642 eol_cp_fd($rfd, $wfd, $es);
1643 close $rfd or croak $!;
1644 close $wfd or croak $!;
1645 }
1646
1647 sub eol_cp_fd {
1648 my ($rfd, $wfd, $es) = @_;
1649 my $eol = defined $es ? $EOL{$es} : undef;
1650 my $buf;
1651 use bytes;
1652 while (1) {
1653 my ($r, $w, $t);
1654 defined($r = sysread($rfd, $buf, 4096)) or croak $!;
1655 return unless $r;
1656 if ($eol) {
1657 if ($buf =~ /\015$/) {
1658 my $c;
1659 defined($r = sysread($rfd,$c,1)) or croak $!;
1660 $buf .= $c if $r > 0;
1661 }
1662 $buf =~ s/(?:\015\012|\015|\012)/$eol/gs;
1663 $r = length($buf);
1664 }
1665 for ($w = 0; $w < $r; $w += $t) {
1666 $t = syswrite($wfd, $buf, $r - $w, $w) or croak $!;
1667 }
1668 }
1669 no bytes;
1670 }
1671
1672 sub do_update_index {
1673 my ($z_cmd, $cmd, $no_text_base) = @_;
1674
1675 my $z = open my $p, '-|';
1676 defined $z or croak $!;
1677 unless ($z) { exec @$z_cmd or croak $! }
1678
1679 my $pid = open my $ui, '|-';
1680 defined $pid or croak $!;
1681 unless ($pid) {
1682 exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
1683 }
1684 local $/ = "\0";
1685 while (my $x = <$p>) {
1686 chomp $x;
1687 if (!$no_text_base && lstat $x && ! -l _ &&
1688 svn_propget_base('svn:keywords', $x)) {
1689 my $mode = -x _ ? 0755 : 0644;
1690 my ($v,$d,$f) = File::Spec->splitpath($x);
1691 my $tb = File::Spec->catfile($d, '.svn', 'tmp',
1692 'text-base',"$f.svn-base");
1693 $tb =~ s#^/##;
1694 unless (-f $tb) {
1695 $tb = File::Spec->catfile($d, '.svn',
1696 'text-base',"$f.svn-base");
1697 $tb =~ s#^/##;
1698 }
1699 unlink $x or croak $!;
1700 eol_cp($tb, $x);
1701 chmod(($mode &~ umask), $x) or croak $!;
1702 }
1703 print $ui $x,"\0";
1704 }
1705 close $ui or croak $?;
1706 }
1707
1708 sub index_changes {
1709 return if $_use_lib;
1710
1711 if (!-f "$GIT_SVN_DIR/info/exclude") {
1712 open my $fd, '>>', "$GIT_SVN_DIR/info/exclude" or croak $!;
1713 print $fd '.svn',"\n";
1714 close $fd or croak $!;
1715 }
1716 my $no_text_base = shift;
1717 do_update_index([qw/git-diff-files --name-only -z/],
1718 'remove',
1719 $no_text_base);
1720 do_update_index([qw/git-ls-files -z --others/,
1721 "--exclude-from=$GIT_SVN_DIR/info/exclude"],
1722 'add',
1723 $no_text_base);
1724 }
1725
1726 sub s_to_file {
1727 my ($str, $file, $mode) = @_;
1728 open my $fd,'>',$file or croak $!;
1729 print $fd $str,"\n" or croak $!;
1730 close $fd or croak $!;
1731 chmod ($mode &~ umask, $file) if (defined $mode);
1732 }
1733
1734 sub file_to_s {
1735 my $file = shift;
1736 open my $fd,'<',$file or croak "$!: file: $file\n";
1737 local $/;
1738 my $ret = <$fd>;
1739 close $fd or croak $!;
1740 $ret =~ s/\s*$//s;
1741 return $ret;
1742 }
1743
1744 sub assert_revision_unknown {
1745 my $r = shift;
1746 if (my $c = revdb_get($REVDB, $r)) {
1747 croak "$r = $c already exists! Why are we refetching it?";
1748 }
1749 }
1750
1751 sub trees_eq {
1752 my ($x, $y) = @_;
1753 my @x = safe_qx('git-cat-file','commit',$x);
1754 my @y = safe_qx('git-cat-file','commit',$y);
1755 if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
1756 || $y[0] !~ /^tree $sha1\n$/) {
1757 print STDERR "Trees not equal: $y[0] != $x[0]\n";
1758 return 0
1759 }
1760 return 1;
1761 }
1762
1763 sub git_commit {
1764 my ($log_msg, @parents) = @_;
1765 assert_revision_unknown($log_msg->{revision});
1766 map_tree_joins() if (@_branch_from && !%tree_map);
1767
1768 my (@tmp_parents, @exec_parents, %seen_parent);
1769 if (my $lparents = $log_msg->{parents}) {
1770 @tmp_parents = @$lparents
1771 }
1772 # commit parents can be conditionally bound to a particular
1773 # svn revision via: "svn_revno=commit_sha1", filter them out here:
1774 foreach my $p (@parents) {
1775 next unless defined $p;
1776 if ($p =~ /^(\d+)=($sha1_short)$/o) {
1777 if ($1 == $log_msg->{revision}) {
1778 push @tmp_parents, $2;
1779 }
1780 } else {
1781 push @tmp_parents, $p if $p =~ /$sha1_short/o;
1782 }
1783 }
1784 my $tree = $log_msg->{tree};
1785 if (!defined $tree) {
1786 my $index = set_index($GIT_SVN_INDEX);
1787 index_changes();
1788 chomp($tree = `git-write-tree`);
1789 croak $? if $?;
1790 restore_index($index);
1791 }
1792 if (exists $tree_map{$tree}) {
1793 push @tmp_parents, @{$tree_map{$tree}};
1794 }
1795 foreach (@tmp_parents) {
1796 next if $seen_parent{$_};
1797 $seen_parent{$_} = 1;
1798 push @exec_parents, $_;
1799 # MAXPARENT is defined to 16 in commit-tree.c:
1800 last if @exec_parents > 16;
1801 }
1802
1803 defined(my $pid = open my $out_fh, '-|') or croak $!;
1804 if ($pid == 0) {
1805 my $msg_fh = IO::File->new_tmpfile or croak $!;
1806 print $msg_fh $log_msg->{msg}, "\ngit-svn-id: ",
1807 "$SVN_URL\@$log_msg->{revision}",
1808 " $SVN_UUID\n" or croak $!;
1809 $msg_fh->flush == 0 or croak $!;
1810 seek $msg_fh, 0, 0 or croak $!;
1811 set_commit_env($log_msg);
1812 my @exec = ('git-commit-tree',$tree);
1813 push @exec, '-p', $_ foreach @exec_parents;
1814 open STDIN, '<&', $msg_fh or croak $!;
1815 exec @exec or croak $!;
1816 }
1817 chomp(my $commit = do { local $/; <$out_fh> });
1818 close $out_fh or croak $?;
1819 if ($commit !~ /^$sha1$/o) {
1820 croak "Failed to commit, invalid sha1: $commit\n";
1821 }
1822 my @update_ref = ('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
1823 if (my $primary_parent = shift @exec_parents) {
1824 quiet_run(qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0");
1825 push @update_ref, $primary_parent unless $?;
1826 }
1827 sys(@update_ref);
1828 revdb_set($REVDB, $log_msg->{revision}, $commit);
1829
1830 # this output is read via pipe, do not change:
1831 print "r$log_msg->{revision} = $commit\n";
1832 check_repack();
1833 return $commit;
1834 }
1835
1836 sub check_repack {
1837 if ($_repack && (--$_repack_nr == 0)) {
1838 $_repack_nr = $_repack;
1839 sys("git repack $_repack_flags");
1840 }
1841 }
1842
1843 sub set_commit_env {
1844 my ($log_msg) = @_;
1845 my $author = $log_msg->{author};
1846 if (!defined $author || length $author == 0) {
1847 $author = '(no author)';
1848 }
1849 my ($name,$email) = defined $users{$author} ? @{$users{$author}}
1850 : ($author,"$author\@$SVN_UUID");
1851 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1852 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1853 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
1854 }
1855
1856 sub apply_mod_line_blob {
1857 my $m = shift;
1858 if ($m->{mode_b} =~ /^120/) {
1859 blob_to_symlink($m->{sha1_b}, $m->{file_b});
1860 } else {
1861 blob_to_file($m->{sha1_b}, $m->{file_b});
1862 }
1863 }
1864
1865 sub blob_to_symlink {
1866 my ($blob, $link) = @_;
1867 defined $link or croak "\$link not defined!\n";
1868 croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
1869 if (-l $link || -f _) {
1870 unlink $link or croak $!;
1871 }
1872
1873 my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
1874 symlink $dest, $link or croak $!;
1875 }
1876
1877 sub blob_to_file {
1878 my ($blob, $file) = @_;
1879 defined $file or croak "\$file not defined!\n";
1880 croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
1881 if (-l $file || -f _) {
1882 unlink $file or croak $!;
1883 }
1884
1885 open my $blob_fh, '>', $file or croak "$!: $file\n";
1886 my $pid = fork;
1887 defined $pid or croak $!;
1888
1889 if ($pid == 0) {
1890 open STDOUT, '>&', $blob_fh or croak $!;
1891 exec('git-cat-file','blob',$blob) or croak $!;
1892 }
1893 waitpid $pid, 0;
1894 croak $? if $?;
1895
1896 close $blob_fh or croak $!;
1897 }
1898
1899 sub safe_qx {
1900 my $pid = open my $child, '-|';
1901 defined $pid or croak $!;
1902 if ($pid == 0) {
1903 exec(@_) or croak $!;
1904 }
1905 my @ret = (<$child>);
1906 close $child or croak $?;
1907 die $? if $?; # just in case close didn't error out
1908 return wantarray ? @ret : join('',@ret);
1909 }
1910
1911 sub svn_compat_check {
1912 my @co_help = safe_qx(qw(svn co -h));
1913 unless (grep /ignore-externals/,@co_help) {
1914 print STDERR "W: Installed svn version does not support ",
1915 "--ignore-externals\n";
1916 $_no_ignore_ext = 1;
1917 }
1918 if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
1919 $_svn_co_url_revs = 1;
1920 }
1921 if (grep /\[TARGET\[\@REV\]\.\.\.\]/, `svn propget -h`) {
1922 $_svn_pg_peg_revs = 1;
1923 }
1924
1925 # I really, really hope nobody hits this...
1926 unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
1927 print STDERR <<'';
1928 W: The installed svn version does not support the --stop-on-copy flag in
1929 the log command.
1930 Lets hope the directory you're tracking is not a branch or tag
1931 and was never moved within the repository...
1932
1933 $_no_stop_copy = 1;
1934 }
1935 }
1936
1937 # *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
1938 # (and they won't honor URL@<rev> without -r<rev>, too!)
1939 sub svn_cmd_checkout {
1940 my ($url, $rev, $dir) = @_;
1941 my @cmd = ('svn','co', "-r$rev");
1942 push @cmd, '--ignore-externals' unless $_no_ignore_ext;
1943 $url .= "\@$rev" if $_svn_co_url_revs;
1944 sys(@cmd, $url, $dir);
1945 }
1946
1947 sub check_upgrade_needed {
1948 if (!-r $REVDB) {
1949 -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
1950 open my $fh, '>>',$REVDB or croak $!;
1951 close $fh;
1952 }
1953 my $old = eval {
1954 my $pid = open my $child, '-|';
1955 defined $pid or croak $!;
1956 if ($pid == 0) {
1957 close STDERR;
1958 exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $!;
1959 }
1960 my @ret = (<$child>);
1961 close $child or croak $?;
1962 die $? if $?; # just in case close didn't error out
1963 return wantarray ? @ret : join('',@ret);
1964 };
1965 return unless $old;
1966 my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
1967 if ($@ || !$head) {
1968 print STDERR "Please run: $0 rebuild --upgrade\n";
1969 exit 1;
1970 }
1971 }
1972
1973 # fills %tree_map with a reverse mapping of trees to commits. Useful
1974 # for finding parents to commit on.
1975 sub map_tree_joins {
1976 my %seen;
1977 foreach my $br (@_branch_from) {
1978 my $pid = open my $pipe, '-|';
1979 defined $pid or croak $!;
1980 if ($pid == 0) {
1981 exec(qw(git-rev-list --topo-order --pretty=raw), $br)
1982 or croak $!;
1983 }
1984 while (<$pipe>) {
1985 if (/^commit ($sha1)$/o) {
1986 my $commit = $1;
1987
1988 # if we've seen a commit,
1989 # we've seen its parents
1990 last if $seen{$commit};
1991 my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
1992 unless (defined $tree) {
1993 die "Failed to parse commit $commit\n";
1994 }
1995 push @{$tree_map{$tree}}, $commit;
1996 $seen{$commit} = 1;
1997 }
1998 }
1999 close $pipe; # we could be breaking the pipe early
2000 }
2001 }
2002
2003 sub load_all_refs {
2004 if (@_branch_from) {
2005 print STDERR '--branch|-b parameters are ignored when ',
2006 "--branch-all-refs|-B is passed\n";
2007 }
2008
2009 # don't worry about rev-list on non-commit objects/tags,
2010 # it shouldn't blow up if a ref is a blob or tree...
2011 chomp(@_branch_from = `git-rev-parse --symbolic --all`);
2012 }
2013
2014 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
2015 sub load_authors {
2016 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2017 while (<$authors>) {
2018 chomp;
2019 next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2020 my ($user, $name, $email) = ($1, $2, $3);
2021 $users{$user} = [$name, $email];
2022 }
2023 close $authors or croak $!;
2024 }
2025
2026 sub rload_authors {
2027 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
2028 while (<$authors>) {
2029 chomp;
2030 next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
2031 my ($user, $name, $email) = ($1, $2, $3);
2032 $rusers{"$name <$email>"} = $user;
2033 }
2034 close $authors or croak $!;
2035 }
2036
2037 sub svn_propget_base {
2038 my ($p, $f) = @_;
2039 $f .= '@BASE' if $_svn_pg_peg_revs;
2040 return safe_qx(qw/svn propget/, $p, $f);
2041 }
2042
2043 sub git_svn_each {
2044 my $sub = shift;
2045 foreach (`git-rev-parse --symbolic --all`) {
2046 next unless s#^refs/remotes/##;
2047 chomp $_;
2048 next unless -f "$GIT_DIR/svn/$_/info/url";
2049 &$sub($_);
2050 }
2051 }
2052
2053 sub migrate_revdb {
2054 git_svn_each(sub {
2055 my $id = shift;
2056 defined(my $pid = fork) or croak $!;
2057 if (!$pid) {
2058 $GIT_SVN = $ENV{GIT_SVN_ID} = $id;
2059 init_vars();
2060 exit 0 if -r $REVDB;
2061 print "Upgrading svn => git mapping...\n";
2062 -d $GIT_SVN_DIR or mkpath([$GIT_SVN_DIR]);
2063 open my $fh, '>>',$REVDB or croak $!;
2064 close $fh;
2065 rebuild();
2066 print "Done upgrading. You may now delete the ",
2067 "deprecated $GIT_SVN_DIR/revs directory\n";
2068 exit 0;
2069 }
2070 waitpid $pid, 0;
2071 croak $? if $?;
2072 });
2073 }
2074
2075 sub migration_check {
2076 migrate_revdb() unless (-e $REVDB);
2077 return if (-d "$GIT_DIR/svn" || !-d $GIT_DIR);
2078 print "Upgrading repository...\n";
2079 unless (-d "$GIT_DIR/svn") {
2080 mkdir "$GIT_DIR/svn" or croak $!;
2081 }
2082 print "Data from a previous version of git-svn exists, but\n\t",
2083 "$GIT_SVN_DIR\n\t(required for this version ",
2084 "($VERSION) of git-svn) does not.\n";
2085
2086 foreach my $x (`git-rev-parse --symbolic --all`) {
2087 next unless $x =~ s#^refs/remotes/##;
2088 chomp $x;
2089 next unless -f "$GIT_DIR/$x/info/url";
2090 my $u = eval { file_to_s("$GIT_DIR/$x/info/url") };
2091 next unless $u;
2092 my $dn = dirname("$GIT_DIR/svn/$x");
2093 mkpath([$dn]) unless -d $dn;
2094 rename "$GIT_DIR/$x", "$GIT_DIR/svn/$x" or croak "$!: $x";
2095 }
2096 migrate_revdb() if (-d $GIT_SVN_DIR && !-w $REVDB);
2097 print "Done upgrading.\n";
2098 }
2099
2100 sub find_rev_before {
2101 my ($r, $id, $eq_ok) = @_;
2102 my $f = "$GIT_DIR/svn/$id/.rev_db";
2103 return (undef,undef) unless -r $f;
2104 --$r unless $eq_ok;
2105 while ($r > 0) {
2106 if (my $c = revdb_get($f, $r)) {
2107 return ($r, $c);
2108 }
2109 --$r;
2110 }
2111 return (undef, undef);
2112 }
2113
2114 sub init_vars {
2115 $GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
2116 $GIT_SVN_DIR = "$GIT_DIR/svn/$GIT_SVN";
2117 $REVDB = "$GIT_SVN_DIR/.rev_db";
2118 $GIT_SVN_INDEX = "$GIT_SVN_DIR/index";
2119 $SVN_URL = undef;
2120 $SVN_WC = "$GIT_SVN_DIR/tree";
2121 }
2122
2123 # convert GetOpt::Long specs for use by git-repo-config
2124 sub read_repo_config {
2125 return unless -d $GIT_DIR;
2126 my $opts = shift;
2127 foreach my $o (keys %$opts) {
2128 my $v = $opts->{$o};
2129 my ($key) = ($o =~ /^([a-z\-]+)/);
2130 $key =~ s/-//g;
2131 my $arg = 'git-repo-config';
2132 $arg .= ' --int' if ($o =~ /[:=]i$/);
2133 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
2134 if (ref $v eq 'ARRAY') {
2135 chomp(my @tmp = `$arg --get-all svn.$key`);
2136 @$v = @tmp if @tmp;
2137 } else {
2138 chomp(my $tmp = `$arg --get svn.$key`);
2139 if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
2140 $$v = $tmp;
2141 }
2142 }
2143 }
2144 }
2145
2146 sub set_default_vals {
2147 if (defined $_repack) {
2148 $_repack = 1000 if ($_repack <= 0);
2149 $_repack_nr = $_repack;
2150 $_repack_flags ||= '-d';
2151 }
2152 }
2153
2154 sub read_grafts {
2155 my $gr_file = shift;
2156 my ($grafts, $comments) = ({}, {});
2157 if (open my $fh, '<', $gr_file) {
2158 my @tmp;
2159 while (<$fh>) {
2160 if (/^($sha1)\s+/) {
2161 my $c = $1;
2162 if (@tmp) {
2163 @{$comments->{$c}} = @tmp;
2164 @tmp = ();
2165 }
2166 foreach my $p (split /\s+/, $_) {
2167 $grafts->{$c}->{$p} = 1;
2168 }
2169 } else {
2170 push @tmp, $_;
2171 }
2172 }
2173 close $fh or croak $!;
2174 @{$comments->{'END'}} = @tmp if @tmp;
2175 }
2176 return ($grafts, $comments);
2177 }
2178
2179 sub write_grafts {
2180 my ($grafts, $comments, $gr_file) = @_;
2181
2182 open my $fh, '>', $gr_file or croak $!;
2183 foreach my $c (sort keys %$grafts) {
2184 if ($comments->{$c}) {
2185 print $fh $_ foreach @{$comments->{$c}};
2186 }
2187 my $p = $grafts->{$c};
2188 delete $p->{$c}; # commits are not self-reproducing...
2189 my $pid = open my $ch, '-|';
2190 defined $pid or croak $!;
2191 if (!$pid) {
2192 exec(qw/git-cat-file commit/, $c) or croak $!;
2193 }
2194 while (<$ch>) {
2195 if (/^parent ([a-f\d]{40})/) {
2196 $p->{$1} = 1;
2197 } else {
2198 last unless /^\S/i;
2199 }
2200 }
2201 close $ch; # breaking the pipe
2202 print $fh $c, ' ', join(' ', sort keys %$p),"\n";
2203 }
2204 if ($comments->{'END'}) {
2205 print $fh $_ foreach @{$comments->{'END'}};
2206 }
2207 close $fh or croak $!;
2208 }
2209
2210 sub read_url_paths {
2211 my $l_map = {};
2212 git_svn_each(sub { my $x = shift;
2213 my $url = file_to_s("$GIT_DIR/svn/$x/info/url");
2214 my ($u, $p) = repo_path_split($url);
2215 $l_map->{$u}->{$p} = $x;
2216 });
2217 return $l_map;
2218 }
2219
2220 sub extract_metadata {
2221 my $id = shift;
2222 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
2223 \s([a-f\d\-]+)$/x);
2224 if (!$rev || !$uuid || !$url) {
2225 # some of the original repositories I made had
2226 # indentifiers like this:
2227 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
2228 }
2229 return ($url, $rev, $uuid);
2230 }
2231
2232 sub tz_to_s_offset {
2233 my ($tz) = @_;
2234 $tz =~ s/(\d\d)$//;
2235 return ($1 * 60) + ($tz * 3600);
2236 }
2237
2238 sub setup_pager { # translated to Perl from pager.c
2239 return unless (-t *STDOUT);
2240 my $pager = $ENV{PAGER};
2241 if (!defined $pager) {
2242 $pager = 'less';
2243 } elsif (length $pager == 0 || $pager eq 'cat') {
2244 return;
2245 }
2246 pipe my $rfd, my $wfd or return;
2247 defined(my $pid = fork) or croak $!;
2248 if (!$pid) {
2249 open STDOUT, '>&', $wfd or croak $!;
2250 return;
2251 }
2252 open STDIN, '<&', $rfd or croak $!;
2253 $ENV{LESS} ||= '-S';
2254 exec $pager or croak "Can't run pager: $!\n";;
2255 }
2256
2257 sub get_author_info {
2258 my ($dest, $author, $t, $tz) = @_;
2259 $author =~ s/(?:^\s*|\s*$)//g;
2260 $dest->{a_raw} = $author;
2261 my $_a;
2262 if ($_authors) {
2263 $_a = $rusers{$author} || undef;
2264 }
2265 if (!$_a) {
2266 ($_a) = ($author =~ /<([^>]+)\@[^>]+>$/);
2267 }
2268 $dest->{t} = $t;
2269 $dest->{tz} = $tz;
2270 $dest->{a} = $_a;
2271 # Date::Parse isn't in the standard Perl distro :(
2272 if ($tz =~ s/^\+//) {
2273 $t += tz_to_s_offset($tz);
2274 } elsif ($tz =~ s/^\-//) {
2275 $t -= tz_to_s_offset($tz);
2276 }
2277 $dest->{t_utc} = $t;
2278 }
2279
2280 sub process_commit {
2281 my ($c, $r_min, $r_max, $defer) = @_;
2282 if (defined $r_min && defined $r_max) {
2283 if ($r_min == $c->{r} && $r_min == $r_max) {
2284 show_commit($c);
2285 return 0;
2286 }
2287 return 1 if $r_min == $r_max;
2288 if ($r_min < $r_max) {
2289 # we need to reverse the print order
2290 return 0 if (defined $_limit && --$_limit < 0);
2291 push @$defer, $c;
2292 return 1;
2293 }
2294 if ($r_min != $r_max) {
2295 return 1 if ($r_min < $c->{r});
2296 return 1 if ($r_max > $c->{r});
2297 }
2298 }
2299 return 0 if (defined $_limit && --$_limit < 0);
2300 show_commit($c);
2301 return 1;
2302 }
2303
2304 sub show_commit {
2305 my $c = shift;
2306 if ($_oneline) {
2307 my $x = "\n";
2308 if (my $l = $c->{l}) {
2309 while ($l->[0] =~ /^\s*$/) { shift @$l }
2310 $x = $l->[0];
2311 }
2312 $_l_fmt ||= 'A' . length($c->{r});
2313 print 'r',pack($_l_fmt, $c->{r}),' | ';
2314 print "$c->{c} | " if $_show_commit;
2315 print $x;
2316 } else {
2317 show_commit_normal($c);
2318 }
2319 }
2320
2321 sub show_commit_normal {
2322 my ($c) = @_;
2323 print '-' x72, "\nr$c->{r} | ";
2324 print "$c->{c} | " if $_show_commit;
2325 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2326 localtime($c->{t_utc})), ' | ';
2327 my $nr_line = 0;
2328
2329 if (my $l = $c->{l}) {
2330 while ($l->[$#$l] eq "\n" && $l->[($#$l - 1)] eq "\n") {
2331 pop @$l;
2332 }
2333 $nr_line = scalar @$l;
2334 if (!$nr_line) {
2335 print "1 line\n\n\n";
2336 } else {
2337 if ($nr_line == 1) {
2338 $nr_line = '1 line';
2339 } else {
2340 $nr_line .= ' lines';
2341 }
2342 print $nr_line, "\n\n";
2343 print $_ foreach @$l;
2344 }
2345 } else {
2346 print "1 line\n\n";
2347
2348 }
2349 foreach my $x (qw/raw diff/) {
2350 if ($c->{$x}) {
2351 print "\n";
2352 print $_ foreach @{$c->{$x}}
2353 }
2354 }
2355 }
2356
2357 sub libsvn_load {
2358 return unless $_use_lib;
2359 $_use_lib = eval {
2360 require SVN::Core;
2361 if ($SVN::Core::VERSION lt '1.2.1') {
2362 die "Need SVN::Core 1.2.1 or better ",
2363 "(got $SVN::Core::VERSION) ",
2364 "Falling back to command-line svn\n";
2365 }
2366 require SVN::Ra;
2367 require SVN::Delta;
2368 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
2369 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
2370 $SVN::Node::dir.$SVN::Node::unknown.
2371 $SVN::Node::none.$SVN::Node::file.
2372 $SVN::Node::dir.$SVN::Node::unknown;
2373 1;
2374 };
2375 }
2376
2377 sub libsvn_connect {
2378 my ($url) = @_;
2379 my $auth = SVN::Core::auth_open([SVN::Client::get_simple_provider(),
2380 SVN::Client::get_ssl_server_trust_file_provider(),
2381 SVN::Client::get_username_provider()]);
2382 my $s = eval { SVN::Ra->new(url => $url, auth => $auth) };
2383 return $s;
2384 }
2385
2386 sub libsvn_get_file {
2387 my ($gui, $f, $rev) = @_;
2388 my $p = $f;
2389 return unless ($p =~ s#^\Q$SVN_PATH\E/?##);
2390
2391 my ($hash, $pid, $in, $out);
2392 my $pool = SVN::Pool->new;
2393 defined($pid = open3($in, $out, '>&STDERR',
2394 qw/git-hash-object -w --stdin/)) or croak $!;
2395 my ($r, $props) = $SVN->get_file($f, $rev, $in, $pool);
2396 $in->flush == 0 or croak $!;
2397 close $in or croak $!;
2398 $pool->clear;
2399 chomp($hash = do { local $/; <$out> });
2400 close $out or croak $!;
2401 waitpid $pid, 0;
2402 $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2403
2404 my $mode = exists $props->{'svn:executable'} ? '100755' : '100644';
2405 if (exists $props->{'svn:special'}) {
2406 $mode = '120000';
2407 my $link = `git-cat-file blob $hash`;
2408 $link =~ s/^link // or die "svn:special file with contents: <",
2409 $link, "> is not understood\n";
2410 defined($pid = open3($in, $out, '>&STDERR',
2411 qw/git-hash-object -w --stdin/)) or croak $!;
2412 print $in $link;
2413 $in->flush == 0 or croak $!;
2414 close $in or croak $!;
2415 chomp($hash = do { local $/; <$out> });
2416 close $out or croak $!;
2417 waitpid $pid, 0;
2418 $hash =~ /^$sha1$/o or die "not a sha1: $hash\n";
2419 }
2420 print $gui $mode,' ',$hash,"\t",$p,"\0" or croak $!;
2421 }
2422
2423 sub libsvn_log_entry {
2424 my ($rev, $author, $date, $msg, $parents) = @_;
2425 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2426 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x)
2427 or die "Unable to parse date: $date\n";
2428 if (defined $_authors && ! defined $users{$author}) {
2429 die "Author: $author not defined in $_authors file\n";
2430 }
2431 return { revision => $rev, date => "+0000 $Y-$m-$d $H:$M:$S",
2432 author => $author, msg => $msg."\n", parents => $parents || [] }
2433 }
2434
2435 sub process_rm {
2436 my ($gui, $last_commit, $f) = @_;
2437 $f =~ s#^\Q$SVN_PATH\E/?## or return;
2438 # remove entire directories.
2439 if (safe_qx('git-ls-tree',$last_commit,'--',$f) =~ /^040000 tree/) {
2440 defined(my $pid = open my $ls, '-|') or croak $!;
2441 if (!$pid) {
2442 exec(qw/git-ls-tree -r --name-only -z/,
2443 $last_commit,'--',$f) or croak $!;
2444 }
2445 local $/ = "\0";
2446 while (<$ls>) {
2447 print $gui '0 ',0 x 40,"\t",$_ or croak $!;
2448 }
2449 close $ls or croak $?;
2450 } else {
2451 print $gui '0 ',0 x 40,"\t",$f,"\0" or croak $!;
2452 }
2453 }
2454
2455 sub libsvn_fetch {
2456 my ($last_commit, $paths, $rev, $author, $date, $msg) = @_;
2457 open my $gui, '| git-update-index -z --index-info' or croak $!;
2458 my @amr;
2459 foreach my $f (keys %$paths) {
2460 my $m = $paths->{$f}->action();
2461 $f =~ s#^/+##;
2462 if ($m =~ /^[DR]$/) {
2463 process_rm($gui, $last_commit, $f);
2464 next if $m eq 'D';
2465 # 'R' can be file replacements, too, right?
2466 }
2467 my $pool = SVN::Pool->new;
2468 my $t = $SVN->check_path($f, $rev, $pool);
2469 if ($t == $SVN::Node::file) {
2470 if ($m =~ /^[AMR]$/) {
2471 push @amr, $f;
2472 } else {
2473 die "Unrecognized action: $m, ($f r$rev)\n";
2474 }
2475 }
2476 $pool->clear;
2477 }
2478 libsvn_get_file($gui, $_, $rev) foreach (@amr);
2479 close $gui or croak $?;
2480 return libsvn_log_entry($rev, $author, $date, $msg, [$last_commit]);
2481 }
2482
2483 sub svn_grab_base_rev {
2484 defined(my $pid = open my $fh, '-|') or croak $!;
2485 if (!$pid) {
2486 open my $null, '>', '/dev/null' or croak $!;
2487 open STDERR, '>&', $null or croak $!;
2488 exec qw/git-rev-parse --verify/,"refs/remotes/$GIT_SVN^0"
2489 or croak $!;
2490 }
2491 chomp(my $c = do { local $/; <$fh> });
2492 close $fh;
2493 if (defined $c && length $c) {
2494 my ($url, $rev, $uuid) = extract_metadata((grep(/^git-svn-id: /,
2495 safe_qx(qw/git-cat-file commit/, $c)))[-1]);
2496 return ($rev, $c);
2497 }
2498 return (undef, undef);
2499 }
2500
2501 sub libsvn_parse_revision {
2502 my $base = shift;
2503 my $head = $SVN->get_latest_revnum();
2504 if (!defined $_revision || $_revision eq 'BASE:HEAD') {
2505 return ($base + 1, $head) if (defined $base);
2506 return (0, $head);
2507 }
2508 return ($1, $2) if ($_revision =~ /^(\d+):(\d+)$/);
2509 return ($_revision, $_revision) if ($_revision =~ /^\d+$/);
2510 if ($_revision =~ /^BASE:(\d+)$/) {
2511 return ($base + 1, $1) if (defined $base);
2512 return (0, $head);
2513 }
2514 return ($1, $head) if ($_revision =~ /^(\d+):HEAD$/);
2515 die "revision argument: $_revision not understood by git-svn\n",
2516 "Try using the command-line svn client instead\n";
2517 }
2518
2519 sub libsvn_traverse {
2520 my ($gui, $pfx, $path, $rev) = @_;
2521 my $cwd = "$pfx/$path";
2522 my $pool = SVN::Pool->new;
2523 $cwd =~ s#^/+##g;
2524 my ($dirent, $r, $props) = $SVN->get_dir($cwd, $rev, $pool);
2525 foreach my $d (keys %$dirent) {
2526 my $t = $dirent->{$d}->kind;
2527 if ($t == $SVN::Node::dir) {
2528 libsvn_traverse($gui, $cwd, $d, $rev);
2529 } elsif ($t == $SVN::Node::file) {
2530 libsvn_get_file($gui, "$cwd/$d", $rev);
2531 }
2532 }
2533 $pool->clear;
2534 }
2535
2536 sub libsvn_traverse_ignore {
2537 my ($fh, $path, $r) = @_;
2538 $path =~ s#^/+##g;
2539 my $pool = SVN::Pool->new;
2540 my ($dirent, undef, $props) = $SVN->get_dir($path, $r, $pool);
2541 my $p = $path;
2542 $p =~ s#^\Q$SVN_PATH\E/?##;
2543 print $fh length $p ? "\n# $p\n" : "\n# /\n";
2544 if (my $s = $props->{'svn:ignore'}) {
2545 $s =~ s/[\r\n]+/\n/g;
2546 chomp $s;
2547 if (length $p == 0) {
2548 $s =~ s#\n#\n/$p#g;
2549 print $fh "/$s\n";
2550 } else {
2551 $s =~ s#\n#\n/$p/#g;
2552 print $fh "/$p/$s\n";
2553 }
2554 }
2555 foreach (sort keys %$dirent) {
2556 next if $dirent->{$_}->kind != $SVN::Node::dir;
2557 libsvn_traverse_ignore($fh, "$path/$_", $r);
2558 }
2559 $pool->clear;
2560 }
2561
2562 sub revisions_eq {
2563 my ($path, $r0, $r1) = @_;
2564 return 1 if $r0 == $r1;
2565 my $nr = 0;
2566 if ($_use_lib) {
2567 # should be OK to use Pool here (r1 - r0) should be small
2568 my $pool = SVN::Pool->new;
2569 $SVN->get_log("/$path", $r0, $r1, 0, 1, 1, sub {$nr++},$pool);
2570 $pool->clear;
2571 } else {
2572 my ($url, undef) = repo_path_split($SVN_URL);
2573 my $svn_log = svn_log_raw("$url/$path","-r$r0:$r1");
2574 while (next_log_entry($svn_log)) { $nr++ }
2575 close $svn_log->{fh};
2576 }
2577 return 0 if ($nr > 1);
2578 return 1;
2579 }
2580
2581 sub libsvn_find_parent_branch {
2582 my ($paths, $rev, $author, $date, $msg) = @_;
2583 my $svn_path = '/'.$SVN_PATH;
2584
2585 # look for a parent from another branch:
2586 my $i = $paths->{$svn_path} or return;
2587 my $branch_from = $i->copyfrom_path or return;
2588 my $r = $i->copyfrom_rev;
2589 print STDERR "Found possible branch point: ",
2590 "$branch_from => $svn_path, $r\n";
2591 $branch_from =~ s#^/##;
2592 my $l_map = read_url_paths();
2593 my $url = $SVN->{url};
2594 defined $l_map->{$url} or return;
2595 my $id = $l_map->{$url}->{$branch_from} or return;
2596 my ($r0, $parent) = find_rev_before($r,$id,1);
2597 return unless (defined $r0 && defined $parent);
2598 if (revisions_eq($branch_from, $r0, $r)) {
2599 unlink $GIT_SVN_INDEX;
2600 print STDERR "Found branch parent: $parent\n";
2601 sys(qw/git-read-tree/, $parent);
2602 return libsvn_fetch($parent, $paths, $rev,
2603 $author, $date, $msg);
2604 }
2605 print STDERR "Nope, branch point not imported or unknown\n";
2606 return undef;
2607 }
2608
2609 sub libsvn_new_tree {
2610 if (my $log_entry = libsvn_find_parent_branch(@_)) {
2611 return $log_entry;
2612 }
2613 my ($paths, $rev, $author, $date, $msg) = @_;
2614 open my $gui, '| git-update-index -z --index-info' or croak $!;
2615 my $pool = SVN::Pool->new;
2616 libsvn_traverse($gui, '', $SVN_PATH, $rev, $pool);
2617 $pool->clear;
2618 close $gui or croak $?;
2619 return libsvn_log_entry($rev, $author, $date, $msg);
2620 }
2621
2622 sub find_graft_path_commit {
2623 my ($tree_paths, $p1, $r1) = @_;
2624 foreach my $x (keys %$tree_paths) {
2625 next unless ($p1 =~ /^\Q$x\E/);
2626 my $i = $tree_paths->{$x};
2627 my ($r0, $parent) = find_rev_before($r1,$i,1);
2628 return $parent if (defined $r0 && $r0 == $r1);
2629 print STDERR "r$r1 of $i not imported\n";
2630 next;
2631 }
2632 return undef;
2633 }
2634
2635 sub find_graft_path_parents {
2636 my ($grafts, $tree_paths, $c, $p0, $r0) = @_;
2637 foreach my $x (keys %$tree_paths) {
2638 next unless ($p0 =~ /^\Q$x\E/);
2639 my $i = $tree_paths->{$x};
2640 my ($r, $parent) = find_rev_before($r0, $i, 1);
2641 if (defined $r && defined $parent && revisions_eq($x,$r,$r0)) {
2642 $grafts->{$c}->{$parent} = 1;
2643 }
2644 }
2645 }
2646
2647 sub libsvn_graft_file_copies {
2648 my ($grafts, $tree_paths, $path, $paths, $rev) = @_;
2649 foreach (keys %$paths) {
2650 my $i = $paths->{$_};
2651 my ($m, $p0, $r0) = ($i->action, $i->copyfrom_path,
2652 $i->copyfrom_rev);
2653 next unless (defined $p0 && defined $r0);
2654
2655 my $p1 = $_;
2656 $p1 =~ s#^/##;
2657 $p0 =~ s#^/##;
2658 my $c = find_graft_path_commit($tree_paths, $p1, $rev);
2659 next unless $c;
2660 find_graft_path_parents($grafts, $tree_paths, $c, $p0, $r0);
2661 }
2662 }
2663
2664 sub set_index {
2665 my $old = $ENV{GIT_INDEX_FILE};
2666 $ENV{GIT_INDEX_FILE} = shift;
2667 return $old;
2668 }
2669
2670 sub restore_index {
2671 my ($old) = @_;
2672 if (defined $old) {
2673 $ENV{GIT_INDEX_FILE} = $old;
2674 } else {
2675 delete $ENV{GIT_INDEX_FILE};
2676 }
2677 }
2678
2679 sub libsvn_commit_cb {
2680 my ($rev, $date, $committer, $c, $msg, $r_last, $cmt_last) = @_;
2681 if ($_optimize_commits && $rev == ($r_last + 1)) {
2682 my $log = libsvn_log_entry($rev,$committer,$date,$msg);
2683 $log->{tree} = get_tree_from_treeish($c);
2684 my $cmt = git_commit($log, $cmt_last, $c);
2685 my @diff = safe_qx('git-diff-tree', $cmt, $c);
2686 if (@diff) {
2687 print STDERR "Trees differ: $cmt $c\n",
2688 join('',@diff),"\n";
2689 exit 1;
2690 }
2691 } else {
2692 fetch("$rev=$c");
2693 }
2694 }
2695
2696 sub libsvn_ls_fullurl {
2697 my $fullurl = shift;
2698 my ($repo, $path) = repo_path_split($fullurl);
2699 $SVN ||= libsvn_connect($repo);
2700 my @ret;
2701 my $pool = SVN::Pool->new;
2702 my ($dirent, undef, undef) = $SVN->get_dir($path,
2703 $SVN->get_latest_revnum, $pool);
2704 foreach my $d (keys %$dirent) {
2705 if ($dirent->{$d}->kind == $SVN::Node::dir) {
2706 push @ret, "$d/"; # add '/' for compat with cli svn
2707 }
2708 }
2709 $pool->clear;
2710 return @ret;
2711 }
2712
2713
2714 sub libsvn_skip_unknown_revs {
2715 my $err = shift;
2716 my $errno = $err->apr_err();
2717 # Maybe the branch we're tracking didn't
2718 # exist when the repo started, so it's
2719 # not an error if it doesn't, just continue
2720 #
2721 # Wonderfully consistent library, eh?
2722 # 160013 - svn:// and file://
2723 # 175002 - http(s)://
2724 # More codes may be discovered later...
2725 if ($errno == 175002 || $errno == 160013) {
2726 return;
2727 }
2728 croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
2729 };
2730
2731 # Tie::File seems to be prone to offset errors if revisions get sparse,
2732 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
2733 # one of my favorite modules is out :< Next up would be one of the DBM
2734 # modules, but I'm not sure which is most portable... So I'll just
2735 # go with something that's plain-text, but still capable of
2736 # being randomly accessed. So here's my ultra-simple fixed-width
2737 # database. All records are 40 characters + "\n", so it's easy to seek
2738 # to a revision: (41 * rev) is the byte offset.
2739 # A record of 40 0s denotes an empty revision.
2740 # And yes, it's still pretty fast (faster than Tie::File).
2741 sub revdb_set {
2742 my ($file, $rev, $commit) = @_;
2743 length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
2744 open my $fh, '+<', $file or croak $!;
2745 my $offset = $rev * 41;
2746 # assume that append is the common case:
2747 seek $fh, 0, 2 or croak $!;
2748 my $pos = tell $fh;
2749 if ($pos < $offset) {
2750 print $fh (('0' x 40),"\n") x (($offset - $pos) / 41);
2751 }
2752 seek $fh, $offset, 0 or croak $!;
2753 print $fh $commit,"\n";
2754 close $fh or croak $!;
2755 }
2756
2757 sub revdb_get {
2758 my ($file, $rev) = @_;
2759 my $ret;
2760 my $offset = $rev * 41;
2761 open my $fh, '<', $file or croak $!;
2762 seek $fh, $offset, 0;
2763 if (tell $fh == $offset) {
2764 $ret = readline $fh;
2765 if (defined $ret) {
2766 chomp $ret;
2767 $ret = undef if ($ret =~ /^0{40}$/);
2768 }
2769 }
2770 close $fh or croak $!;
2771 return $ret;
2772 }
2773
2774 sub copy_remote_ref {
2775 my $origin = $_cp_remote ? $_cp_remote : 'origin';
2776 my $ref = "refs/remotes/$GIT_SVN";
2777 if (safe_qx('git-ls-remote', $origin, $ref)) {
2778 sys(qw/git fetch/, $origin, "$ref:$ref");
2779 } else {
2780 die "Unable to find remote reference: ",
2781 "refs/remotes/$GIT_SVN on $origin\n";
2782 }
2783 }
2784
2785 package SVN::Git::Editor;
2786 use vars qw/@ISA/;
2787 use strict;
2788 use warnings;
2789 use Carp qw/croak/;
2790 use IO::File;
2791
2792 sub new {
2793 my $class = shift;
2794 my $git_svn = shift;
2795 my $self = SVN::Delta::Editor->new(@_);
2796 bless $self, $class;
2797 foreach (qw/svn_path c r ra /) {
2798 die "$_ required!\n" unless (defined $git_svn->{$_});
2799 $self->{$_} = $git_svn->{$_};
2800 }
2801 $self->{pool} = SVN::Pool->new;
2802 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
2803 $self->{rm} = { };
2804 require Digest::MD5;
2805 return $self;
2806 }
2807
2808 sub split_path {
2809 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
2810 }
2811
2812 sub repo_path {
2813 (defined $_[1] && length $_[1]) ? "$_[0]->{svn_path}/$_[1]"
2814 : $_[0]->{svn_path}
2815 }
2816
2817 sub url_path {
2818 my ($self, $path) = @_;
2819 $self->{ra}->{url} . '/' . $self->repo_path($path);
2820 }
2821
2822 sub rmdirs {
2823 my ($self) = @_;
2824 my $rm = $self->{rm};
2825 delete $rm->{''}; # we never delete the url we're tracking
2826 return unless %$rm;
2827
2828 foreach (keys %$rm) {
2829 my @d = split m#/#, $_;
2830 my $c = shift @d;
2831 $rm->{$c} = 1;
2832 while (@d) {
2833 $c .= '/' . shift @d;
2834 $rm->{$c} = 1;
2835 }
2836 }
2837 delete $rm->{$self->{svn_path}};
2838 delete $rm->{''}; # we never delete the url we're tracking
2839 return unless %$rm;
2840
2841 defined(my $pid = open my $fh,'-|') or croak $!;
2842 if (!$pid) {
2843 exec qw/git-ls-tree --name-only -r -z/, $self->{c} or croak $!;
2844 }
2845 local $/ = "\0";
2846 my @svn_path = split m#/#, $self->{svn_path};
2847 while (<$fh>) {
2848 chomp;
2849 my @dn = (@svn_path, (split m#/#, $_));
2850 while (pop @dn) {
2851 delete $rm->{join '/', @dn};
2852 }
2853 unless (%$rm) {
2854 close $fh;
2855 return;
2856 }
2857 }
2858 close $fh;
2859
2860 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
2861 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
2862 $self->close_directory($bat->{$d}, $p);
2863 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
2864 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
2865 delete $bat->{$d};
2866 }
2867 }
2868
2869 sub open_or_add_dir {
2870 my ($self, $full_path, $baton) = @_;
2871 my $p = SVN::Pool->new;
2872 my $t = $self->{ra}->check_path($full_path, $self->{r}, $p);
2873 $p->clear;
2874 if ($t == $SVN::Node::none) {
2875 return $self->add_directory($full_path, $baton,
2876 undef, -1, $self->{pool});
2877 } elsif ($t == $SVN::Node::dir) {
2878 return $self->open_directory($full_path, $baton,
2879 $self->{r}, $self->{pool});
2880 }
2881 print STDERR "$full_path already exists in repository at ",
2882 "r$self->{r} and it is not a directory (",
2883 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
2884 exit 1;
2885 }
2886
2887 sub ensure_path {
2888 my ($self, $path) = @_;
2889 my $bat = $self->{bat};
2890 $path = $self->repo_path($path);
2891 return $bat->{''} unless (length $path);
2892 my @p = split m#/+#, $path;
2893 my $c = shift @p;
2894 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
2895 while (@p) {
2896 my $c0 = $c;
2897 $c .= '/' . shift @p;
2898 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
2899 }
2900 return $bat->{$c};
2901 }
2902
2903 sub A {
2904 my ($self, $m) = @_;
2905 my ($dir, $file) = split_path($m->{file_b});
2906 my $pbat = $self->ensure_path($dir);
2907 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2908 undef, -1);
2909 $self->chg_file($fbat, $m);
2910 $self->close_file($fbat,undef,$self->{pool});
2911 }
2912
2913 sub C {
2914 my ($self, $m) = @_;
2915 my ($dir, $file) = split_path($m->{file_b});
2916 my $pbat = $self->ensure_path($dir);
2917 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2918 $self->url_path($m->{file_a}), $self->{r});
2919 $self->chg_file($fbat, $m);
2920 $self->close_file($fbat,undef,$self->{pool});
2921 }
2922
2923 sub delete_entry {
2924 my ($self, $path, $pbat) = @_;
2925 my $rpath = $self->repo_path($path);
2926 my ($dir, $file) = split_path($rpath);
2927 $self->{rm}->{$dir} = 1;
2928 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
2929 }
2930
2931 sub R {
2932 my ($self, $m) = @_;
2933 my ($dir, $file) = split_path($m->{file_b});
2934 my $pbat = $self->ensure_path($dir);
2935 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
2936 $self->url_path($m->{file_a}), $self->{r});
2937 $self->chg_file($fbat, $m);
2938 $self->close_file($fbat,undef,$self->{pool});
2939
2940 ($dir, $file) = split_path($m->{file_a});
2941 $pbat = $self->ensure_path($dir);
2942 $self->delete_entry($m->{file_a}, $pbat);
2943 }
2944
2945 sub M {
2946 my ($self, $m) = @_;
2947 my ($dir, $file) = split_path($m->{file_b});
2948 my $pbat = $self->ensure_path($dir);
2949 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2950 $pbat,$self->{r},$self->{pool});
2951 $self->chg_file($fbat, $m);
2952 $self->close_file($fbat,undef,$self->{pool});
2953 }
2954
2955 sub T { shift->M(@_) }
2956
2957 sub change_file_prop {
2958 my ($self, $fbat, $pname, $pval) = @_;
2959 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2960 }
2961
2962 sub chg_file {
2963 my ($self, $fbat, $m) = @_;
2964 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2965 $self->change_file_prop($fbat,'svn:executable','*');
2966 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2967 $self->change_file_prop($fbat,'svn:executable',undef);
2968 }
2969 my $fh = IO::File->new_tmpfile or croak $!;
2970 if ($m->{mode_b} =~ /^120/) {
2971 print $fh 'link ' or croak $!;
2972 $self->change_file_prop($fbat,'svn:special','*');
2973 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2974 $self->change_file_prop($fbat,'svn:special',undef);
2975 }
2976 defined(my $pid = fork) or croak $!;
2977 if (!$pid) {
2978 open STDOUT, '>&', $fh or croak $!;
2979 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2980 }
2981 waitpid $pid, 0;
2982 croak $? if $?;
2983 $fh->flush == 0 or croak $!;
2984 seek $fh, 0, 0 or croak $!;
2985
2986 my $md5 = Digest::MD5->new;
2987 $md5->addfile($fh) or croak $!;
2988 seek $fh, 0, 0 or croak $!;
2989
2990 my $exp = $md5->hexdigest;
2991 my $atd = $self->apply_textdelta($fbat, undef, $self->{pool});
2992 my $got = SVN::TxDelta::send_stream($fh, @$atd, $self->{pool});
2993 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2994
2995 close $fh or croak $!;
2996 }
2997
2998 sub D {
2999 my ($self, $m) = @_;
3000 my ($dir, $file) = split_path($m->{file_b});
3001 my $pbat = $self->ensure_path($dir);
3002 $self->delete_entry($m->{file_b}, $pbat);
3003 }
3004
3005 sub close_edit {
3006 my ($self) = @_;
3007 my ($p,$bat) = ($self->{pool}, $self->{bat});
3008 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3009 $self->close_directory($bat->{$_}, $p);
3010 }
3011 $self->SUPER::close_edit($p);
3012 $p->clear;
3013 }
3014
3015 sub abort_edit {
3016 my ($self) = @_;
3017 $self->SUPER::abort_edit($self->{pool});
3018 $self->{pool}->clear;
3019 }
3020
3021 __END__
3022
3023 Data structures:
3024
3025 $svn_log hashref (as returned by svn_log_raw)
3026 {
3027 fh => file handle of the log file,
3028 state => state of the log file parser (sep/msg/rev/msg_start...)
3029 }
3030
3031 $log_msg hashref as returned by next_log_entry($svn_log)
3032 {
3033 msg => 'whitespace-formatted log entry
3034 ', # trailing newline is preserved
3035 revision => '8', # integer
3036 date => '2004-02-24T17:01:44.108345Z', # commit date
3037 author => 'committer name'
3038 };
3039
3040
3041 @mods = array of diff-index line hashes, each element represents one line
3042 of diff-index output
3043
3044 diff-index line ($m hash)
3045 {
3046 mode_a => first column of diff-index output, no leading ':',
3047 mode_b => second column of diff-index output,
3048 sha1_b => sha1sum of the final blob,
3049 chg => change type [MCRADT],
3050 file_a => original file name of a file (iff chg is 'C' or 'R')
3051 file_b => new/current file name of a file (any chg)
3052 }
3053 ;
3054
3055 Notes:
3056 I don't trust the each() function on unless I created %hash myself
3057 because the internal iterator may not have started at base.