?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
use Git::SVN{,::RA}->url accessor globally
[alt-git.git] / perl / Git / SVN.pm
blobfc1ac07136af343ba2a373163870b195cea080d4
1 package Git::SVN;
2 use strict;
3 use warnings;
4 use Fcntl qw/:DEFAULT :seek/;
5 use constant rev_map_fmt => 'NH40';
6 use vars qw/$_no_metadata
7 $_repack $_repack_flags $_use_svm_props $_head
8 $_use_svnsync_props $no_reuse_existing
9 $_use_log_author $_add_author_from $_localtime/;
10 use Carp qw/croak/;
11 use File::Path qw/mkpath/;
12 use File::Copy qw/copy/;
13 use IPC::Open3;
14 use Time::Local;
15 use Memoize; # core since 5.8.0, Jul 2002
16 use Memoize::Storable;
17 use POSIX qw(:signal_h);
18
19 use Git qw(
20 command
21 command_oneline
22 command_noisy
23 command_output_pipe
24 command_close_pipe
25 );
26 use Git::SVN::Utils qw(fatal can_compress);
27
28 my $can_use_yaml;
29 BEGIN {
30 $can_use_yaml = eval { require Git::SVN::Memoize::YAML; 1};
31 }
32
33 our $_follow_parent = 1;
34 our $_minimize_url = 'unset';
35 our $default_repo_id = 'svn';
36 our $default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
37
38 my ($_gc_nr, $_gc_period);
39
40 # properties that we do not log:
41 my %SKIP_PROP;
42 BEGIN {
43 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
44 svn:special svn:executable
45 svn:entry:committed-rev
46 svn:entry:last-author
47 svn:entry:uuid
48 svn:entry:committed-date/;
49
50 # some options are read globally, but can be overridden locally
51 # per [svn-remote "..."] section. Command-line options will *NOT*
52 # override options set in an [svn-remote "..."] section
53 no strict 'refs';
54 for my $option (qw/follow_parent no_metadata use_svm_props
55 use_svnsync_props/) {
56 my $key = $option;
57 $key =~ tr/_//d;
58 my $prop = "-$option";
59 *$option = sub {
60 my ($self) = @_;
61 return $self->{$prop} if exists $self->{$prop};
62 my $k = "svn-remote.$self->{repo_id}.$key";
63 eval { command_oneline(qw/config --get/, $k) };
64 if ($@) {
65 $self->{$prop} = ${"Git::SVN::_$option"};
66 } else {
67 my $v = command_oneline(qw/config --bool/,$k);
68 $self->{$prop} = $v eq 'false' ? 0 : 1;
69 }
70 return $self->{$prop};
71 }
72 }
73 }
74
75
76 my (%LOCKFILES, %INDEX_FILES);
77 END {
78 unlink keys %LOCKFILES if %LOCKFILES;
79 unlink keys %INDEX_FILES if %INDEX_FILES;
80 }
81
82 sub resolve_local_globs {
83 my ($url, $fetch, $glob_spec) = @_;
84 return unless defined $glob_spec;
85 my $ref = $glob_spec->{ref};
86 my $path = $glob_spec->{path};
87 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
88 next unless m#^$ref->{regex}$#;
89 my $p = $1;
90 my $pathname = desanitize_refname($path->full_path($p));
91 my $refname = desanitize_refname($ref->full_path($p));
92 if (my $existing = $fetch->{$pathname}) {
93 if ($existing ne $refname) {
94 die "Refspec conflict:\n",
95 "existing: $existing\n",
96 " globbed: $refname\n";
97 }
98 my $u = (::cmt_metadata("$refname"))[0];
99 $u =~ s!^\Q$url\E(/|$)!! or die
100 "$refname: '$url' not found in '$u'\n";
101 if ($pathname ne $u) {
102 warn "W: Refspec glob conflict ",
103 "(ref: $refname):\n",
104 "expected path: $pathname\n",
105 " real path: $u\n",
106 "Continuing ahead with $u\n";
107 next;
108 }
109 } else {
110 $fetch->{$pathname} = $refname;
111 }
112 }
113 }
114
115 sub parse_revision_argument {
116 my ($base, $head) = @_;
117 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
118 return ($base, $head);
119 }
120 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
121 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
122 return ($head, $head) if ($::_revision eq 'HEAD');
123 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
124 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
125 die "revision argument: $::_revision not understood by git-svn\n";
126 }
127
128 sub fetch_all {
129 my ($repo_id, $remotes) = @_;
130 if (ref $repo_id) {
131 my $gs = $repo_id;
132 $repo_id = undef;
133 $repo_id = $gs->{repo_id};
134 }
135 $remotes ||= read_all_remotes();
136 my $remote = $remotes->{$repo_id} or
137 die "[svn-remote \"$repo_id\"] unknown\n";
138 my $fetch = $remote->{fetch};
139 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
140 my (@gs, @globs);
141 my $ra = Git::SVN::Ra->new($url);
142 my $uuid = $ra->get_uuid;
143 my $head = $ra->get_latest_revnum;
144
145 # ignore errors, $head revision may not even exist anymore
146 eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
147 warn "W: $@\n" if $@;
148
149 my $base = defined $fetch ? $head : 0;
150
151 # read the max revs for wildcard expansion (branches/*, tags/*)
152 foreach my $t (qw/branches tags/) {
153 defined $remote->{$t} or next;
154 push @globs, @{$remote->{$t}};
155
156 my $max_rev = eval { tmp_config(qw/--int --get/,
157 "svn-remote.$repo_id.${t}-maxRev") };
158 if (defined $max_rev && ($max_rev < $base)) {
159 $base = $max_rev;
160 } elsif (!defined $max_rev) {
161 $base = 0;
162 }
163 }
164
165 if ($fetch) {
166 foreach my $p (sort keys %$fetch) {
167 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
168 my $lr = $gs->rev_map_max;
169 if (defined $lr) {
170 $base = $lr if ($lr < $base);
171 }
172 push @gs, $gs;
173 }
174 }
175
176 ($base, $head) = parse_revision_argument($base, $head);
177 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
178 }
179
180 sub read_all_remotes {
181 my $r = {};
182 my $use_svm_props = eval { command_oneline(qw/config --bool
183 svn.useSvmProps/) };
184 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
185 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
186 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
187 if (m!^(.+)\.fetch=$svn_refspec$!) {
188 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
189 die("svn-remote.$remote: remote ref '$remote_ref' "
190 . "must start with 'refs/'\n")
191 unless $remote_ref =~ m{^refs/};
192 $local_ref = uri_decode($local_ref);
193 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
194 $r->{$remote}->{svm} = {} if $use_svm_props;
195 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
196 $r->{$1}->{svm} = {};
197 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
198 $r->{$1}->{url} = $2;
199 } elsif (m!^(.+)\.pushurl=\s*(.*)\s*$!) {
200 $r->{$1}->{pushurl} = $2;
201 } elsif (m!^(.+)\.ignore-refs=\s*(.*)\s*$!) {
202 $r->{$1}->{ignore_refs_regex} = $2;
203 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
204 my ($remote, $t, $local_ref, $remote_ref) =
205 ($1, $2, $3, $4);
206 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
207 . "must start with 'refs/'\n")
208 unless $remote_ref =~ m{^refs/};
209 $local_ref = uri_decode($local_ref);
210
211 require Git::SVN::GlobSpec;
212 my $rs = {
213 t => $t,
214 remote => $remote,
215 path => Git::SVN::GlobSpec->new($local_ref, 1),
216 ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
217 if (length($rs->{ref}->{right}) != 0) {
218 die "The '*' glob character must be the last ",
219 "character of '$remote_ref'\n";
220 }
221 push @{ $r->{$remote}->{$t} }, $rs;
222 }
223 }
224
225 map {
226 if (defined $r->{$_}->{svm}) {
227 my $svm;
228 eval {
229 my $section = "svn-remote.$_";
230 $svm = {
231 source => tmp_config('--get',
232 "$section.svm-source"),
233 replace => tmp_config('--get',
234 "$section.svm-replace"),
235 }
236 };
237 $r->{$_}->{svm} = $svm;
238 }
239 } keys %$r;
240
241 foreach my $remote (keys %$r) {
242 foreach ( grep { defined $_ }
243 map { $r->{$remote}->{$_} } qw(branches tags) ) {
244 foreach my $rs ( @$_ ) {
245 $rs->{ignore_refs_regex} =
246 $r->{$remote}->{ignore_refs_regex};
247 }
248 }
249 }
250
251 $r;
252 }
253
254 sub init_vars {
255 $_gc_nr = $_gc_period = 1000;
256 if (defined $_repack || defined $_repack_flags) {
257 warn "Repack options are obsolete; they have no effect.\n";
258 }
259 }
260
261 sub verify_remotes_sanity {
262 return unless -d $ENV{GIT_DIR};
263 my %seen;
264 foreach (command(qw/config -l/)) {
265 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
266 if ($seen{$1}) {
267 die "Remote ref refs/remote/$1 is tracked by",
268 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
269 "Please resolve this ambiguity in ",
270 "your git configuration file before ",
271 "continuing\n";
272 }
273 $seen{$1} = $_;
274 }
275 }
276 }
277
278 sub find_existing_remote {
279 my ($url, $remotes) = @_;
280 return undef if $no_reuse_existing;
281 my $existing;
282 foreach my $repo_id (keys %$remotes) {
283 my $u = $remotes->{$repo_id}->{url} or next;
284 next if $u ne $url;
285 $existing = $repo_id;
286 last;
287 }
288 $existing;
289 }
290
291 sub init_remote_config {
292 my ($self, $url, $no_write) = @_;
293 $url =~ s!/+$!!; # strip trailing slash
294 my $r = read_all_remotes();
295 my $existing = find_existing_remote($url, $r);
296 if ($existing) {
297 unless ($no_write) {
298 print STDERR "Using existing ",
299 "[svn-remote \"$existing\"]\n";
300 }
301 $self->{repo_id} = $existing;
302 } elsif ($_minimize_url) {
303 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
304 $existing = find_existing_remote($min_url, $r);
305 if ($existing) {
306 unless ($no_write) {
307 print STDERR "Using existing ",
308 "[svn-remote \"$existing\"]\n";
309 }
310 $self->{repo_id} = $existing;
311 }
312 if ($min_url ne $url) {
313 unless ($no_write) {
314 print STDERR "Using higher level of URL: ",
315 "$url => $min_url\n";
316 }
317 my $old_path = $self->path;
318 $url =~ s!^\Q$min_url\E(/|$)!!;
319 if (length $old_path) {
320 $url .= "/$old_path";
321 }
322 $self->path($url);
323 $url = $min_url;
324 }
325 }
326 my $orig_url;
327 if (!$existing) {
328 # verify that we aren't overwriting anything:
329 $orig_url = eval {
330 command_oneline('config', '--get',
331 "svn-remote.$self->{repo_id}.url")
332 };
333 if ($orig_url && ($orig_url ne $url)) {
334 die "svn-remote.$self->{repo_id}.url already set: ",
335 "$orig_url\nwanted to set to: $url\n";
336 }
337 }
338 my ($xrepo_id, $xpath) = find_ref($self->refname);
339 if (!$no_write && defined $xpath) {
340 die "svn-remote.$xrepo_id.fetch already set to track ",
341 "$xpath:", $self->refname, "\n";
342 }
343 unless ($no_write) {
344 command_noisy('config',
345 "svn-remote.$self->{repo_id}.url", $url);
346 my $path = $self->path;
347 $path =~ s{^/}{};
348 $path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
349 $self->path($path);
350 command_noisy('config', '--add',
351 "svn-remote.$self->{repo_id}.fetch",
352 $self->path.":".$self->refname);
353 }
354 $self->url($url);
355 }
356
357 sub find_by_url { # repos_root and, path are optional
358 my ($class, $full_url, $repos_root, $path) = @_;
359
360 return undef unless defined $full_url;
361 remove_username($full_url);
362 remove_username($repos_root) if defined $repos_root;
363 my $remotes = read_all_remotes();
364 if (defined $full_url && defined $repos_root && !defined $path) {
365 $path = $full_url;
366 $path =~ s#^\Q$repos_root\E(?:/|$)##;
367 }
368 foreach my $repo_id (keys %$remotes) {
369 my $u = $remotes->{$repo_id}->{url} or next;
370 remove_username($u);
371 next if defined $repos_root && $repos_root ne $u;
372
373 my $fetch = $remotes->{$repo_id}->{fetch} || {};
374 foreach my $t (qw/branches tags/) {
375 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
376 resolve_local_globs($u, $fetch, $globspec);
377 }
378 }
379 my $p = $path;
380 my $rwr = rewrite_root({repo_id => $repo_id});
381 my $svm = $remotes->{$repo_id}->{svm}
382 if defined $remotes->{$repo_id}->{svm};
383 unless (defined $p) {
384 $p = $full_url;
385 my $z = $u;
386 my $prefix = '';
387 if ($rwr) {
388 $z = $rwr;
389 remove_username($z);
390 } elsif (defined $svm) {
391 $z = $svm->{source};
392 $prefix = $svm->{replace};
393 $prefix =~ s#^\Q$u\E(?:/|$)##;
394 $prefix =~ s#/$##;
395 }
396 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
397 }
398 foreach my $f (keys %$fetch) {
399 next if $f ne $p;
400 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
401 }
402 }
403 undef;
404 }
405
406 sub init {
407 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
408 my $self = _new($class, $repo_id, $ref_id, $path);
409 if (defined $url) {
410 $self->init_remote_config($url, $no_write);
411 }
412 $self;
413 }
414
415 sub find_ref {
416 my ($ref_id) = @_;
417 foreach (command(qw/config -l/)) {
418 next unless m!^svn-remote\.(.+)\.fetch=
419 \s*(.*?)\s*:\s*(.+?)\s*$!x;
420 my ($repo_id, $path, $ref) = ($1, $2, $3);
421 if ($ref eq $ref_id) {
422 $path = '' if ($path =~ m#^\./?#);
423 return ($repo_id, $path);
424 }
425 }
426 (undef, undef, undef);
427 }
428
429 sub new {
430 my ($class, $ref_id, $repo_id, $path) = @_;
431 if (defined $ref_id && !defined $repo_id && !defined $path) {
432 ($repo_id, $path) = find_ref($ref_id);
433 if (!defined $repo_id) {
434 die "Could not find a \"svn-remote.*.fetch\" key ",
435 "in the repository configuration matching: ",
436 "$ref_id\n";
437 }
438 }
439 my $self = _new($class, $repo_id, $ref_id, $path);
440 if (!defined $self->path || !length $self->path) {
441 my $fetch = command_oneline('config', '--get',
442 "svn-remote.$repo_id.fetch",
443 ":$ref_id\$") or
444 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
445 "\":$ref_id\$\" in config\n";
446 my($path) = split(/\s*:\s*/, $fetch);
447 $self->path($path);
448 }
449 {
450 my $path = $self->path;
451 $path =~ s{/+}{/}g;
452 $path =~ s{\A/}{};
453 $path =~ s{/\z}{};
454 $self->path($path);
455 }
456 my $url = command_oneline('config', '--get',
457 "svn-remote.$repo_id.url") or
458 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
459 $self->url($url);
460 $self->{pushurl} = eval { command_oneline('config', '--get',
461 "svn-remote.$repo_id.pushurl") };
462 $self->rebuild;
463 $self;
464 }
465
466 sub refname {
467 my ($refname) = $_[0]->{ref_id} ;
468
469 # It cannot end with a slash /, we'll throw up on this because
470 # SVN can't have directories with a slash in their name, either:
471 if ($refname =~ m{/$}) {
472 die "ref: '$refname' ends with a trailing slash, this is ",
473 "not permitted by git nor Subversion\n";
474 }
475
476 # It cannot have ASCII control character space, tilde ~, caret ^,
477 # colon :, question-mark ?, asterisk *, space, or open bracket [
478 # anywhere.
479 #
480 # Additionally, % must be escaped because it is used for escaping
481 # and we want our escaped refname to be reversible
482 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
483
484 # no slash-separated component can begin with a dot .
485 # /.* becomes /%2E*
486 $refname =~ s{/\.}{/%2E}g;
487
488 # It cannot have two consecutive dots .. anywhere
489 # .. becomes %2E%2E
490 $refname =~ s{\.\.}{%2E%2E}g;
491
492 # trailing dots and .lock are not allowed
493 # .$ becomes %2E and .lock becomes %2Elock
494 $refname =~ s{\.(?=$|lock$)}{%2E};
495
496 # the sequence @{ is used to access the reflog
497 # @{ becomes %40{
498 $refname =~ s{\@\{}{%40\{}g;
499
500 return $refname;
501 }
502
503 sub desanitize_refname {
504 my ($refname) = @_;
505 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
506 return $refname;
507 }
508
509 sub svm_uuid {
510 my ($self) = @_;
511 return $self->{svm}->{uuid} if $self->svm;
512 $self->ra;
513 unless ($self->{svm}) {
514 die "SVM UUID not cached, and reading remotely failed\n";
515 }
516 $self->{svm}->{uuid};
517 }
518
519 sub svm {
520 my ($self) = @_;
521 return $self->{svm} if $self->{svm};
522 my $svm;
523 # see if we have it in our config, first:
524 eval {
525 my $section = "svn-remote.$self->{repo_id}";
526 $svm = {
527 source => tmp_config('--get', "$section.svm-source"),
528 uuid => tmp_config('--get', "$section.svm-uuid"),
529 replace => tmp_config('--get', "$section.svm-replace"),
530 }
531 };
532 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
533 $self->{svm} = $svm;
534 }
535 $self->{svm};
536 }
537
538 sub _set_svm_vars {
539 my ($self, $ra) = @_;
540 return $ra if $self->svm;
541
542 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
543 "(svm:source, svm:uuid) ",
544 "from the following URLs:\n" );
545 sub read_svm_props {
546 my ($self, $ra, $path, $r) = @_;
547 my $props = ($ra->get_dir($path, $r))[2];
548 my $src = $props->{'svm:source'};
549 my $uuid = $props->{'svm:uuid'};
550 return undef if (!$src || !$uuid);
551
552 chomp($src, $uuid);
553
554 $uuid =~ m{^[0-9a-f\-]{30,}$}i
555 or die "doesn't look right - svm:uuid is '$uuid'\n";
556
557 # the '!' is used to mark the repos_root!/relative/path
558 $src =~ s{/?!/?}{/};
559 $src =~ s{/+$}{}; # no trailing slashes please
560 # username is of no interest
561 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
562
563 my $replace = $ra->url;
564 $replace .= "/$path" if length $path;
565
566 my $section = "svn-remote.$self->{repo_id}";
567 tmp_config("$section.svm-source", $src);
568 tmp_config("$section.svm-replace", $replace);
569 tmp_config("$section.svm-uuid", $uuid);
570 $self->{svm} = {
571 source => $src,
572 uuid => $uuid,
573 replace => $replace
574 };
575 }
576
577 my $r = $ra->get_latest_revnum;
578 my $path = $self->path;
579 my %tried;
580 while (length $path) {
581 my $try = $self->url . "/$path";
582 unless ($tried{$try}) {
583 return $ra if $self->read_svm_props($ra, $path, $r);
584 $tried{$try} = 1;
585 }
586 $path =~ s#/?[^/]+$##;
587 }
588 die "Path: '$path' should be ''\n" if $path ne '';
589 return $ra if $self->read_svm_props($ra, $path, $r);
590 $tried{$self->url."/$path"} = 1;
591
592 if ($ra->{repos_root} eq $self->url) {
593 die @err, (map { " $_\n" } keys %tried), "\n";
594 }
595
596 # nope, make sure we're connected to the repository root:
597 my $ok;
598 my @tried_b;
599 $path = $ra->{svn_path};
600 $ra = Git::SVN::Ra->new($ra->{repos_root});
601 while (length $path) {
602 my $try = $ra->url ."/$path";
603 unless ($tried{$try}) {
604 $ok = $self->read_svm_props($ra, $path, $r);
605 last if $ok;
606 $tried{$try} = 1;
607 }
608 $path =~ s#/?[^/]+$##;
609 }
610 die "Path: '$path' should be ''\n" if $path ne '';
611 $ok ||= $self->read_svm_props($ra, $path, $r);
612 $tried{$ra->url ."/$path"} = 1;
613 if (!$ok) {
614 die @err, (map { " $_\n" } keys %tried), "\n";
615 }
616 Git::SVN::Ra->new($self->url);
617 }
618
619 sub svnsync {
620 my ($self) = @_;
621 return $self->{svnsync} if $self->{svnsync};
622
623 if ($self->no_metadata) {
624 die "Can't have both 'noMetadata' and ",
625 "'useSvnsyncProps' options set!\n";
626 }
627 if ($self->rewrite_root) {
628 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
629 "options set!\n";
630 }
631 if ($self->rewrite_uuid) {
632 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
633 "options set!\n";
634 }
635
636 my $svnsync;
637 # see if we have it in our config, first:
638 eval {
639 my $section = "svn-remote.$self->{repo_id}";
640
641 my $url = tmp_config('--get', "$section.svnsync-url");
642 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
643 die "doesn't look right - svn:sync-from-url is '$url'\n";
644
645 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
646 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
647 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
648
649 $svnsync = { url => $url, uuid => $uuid }
650 };
651 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
652 return $self->{svnsync} = $svnsync;
653 }
654
655 my $err = "useSvnsyncProps set, but failed to read " .
656 "svnsync property: svn:sync-from-";
657 my $rp = $self->ra->rev_proplist(0);
658
659 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
660 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
661 die "doesn't look right - svn:sync-from-url is '$url'\n";
662
663 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
664 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
665 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
666
667 my $section = "svn-remote.$self->{repo_id}";
668 tmp_config('--add', "$section.svnsync-uuid", $uuid);
669 tmp_config('--add', "$section.svnsync-url", $url);
670 return $self->{svnsync} = { url => $url, uuid => $uuid };
671 }
672
673 # this allows us to memoize our SVN::Ra UUID locally and avoid a
674 # remote lookup (useful for 'git svn log').
675 sub ra_uuid {
676 my ($self) = @_;
677 unless ($self->{ra_uuid}) {
678 my $key = "svn-remote.$self->{repo_id}.uuid";
679 my $uuid = eval { tmp_config('--get', $key) };
680 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
681 $self->{ra_uuid} = $uuid;
682 } else {
683 die "ra_uuid called without URL\n" unless $self->url;
684 $self->{ra_uuid} = $self->ra->get_uuid;
685 tmp_config('--add', $key, $self->{ra_uuid});
686 }
687 }
688 $self->{ra_uuid};
689 }
690
691 sub _set_repos_root {
692 my ($self, $repos_root) = @_;
693 my $k = "svn-remote.$self->{repo_id}.reposRoot";
694 $repos_root ||= $self->ra->{repos_root};
695 tmp_config($k, $repos_root);
696 $repos_root;
697 }
698
699 sub repos_root {
700 my ($self) = @_;
701 my $k = "svn-remote.$self->{repo_id}.reposRoot";
702 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
703 }
704
705 sub ra {
706 my ($self) = shift;
707 my $ra = Git::SVN::Ra->new($self->url);
708 $self->_set_repos_root($ra->{repos_root});
709 if ($self->use_svm_props && !$self->{svm}) {
710 if ($self->no_metadata) {
711 die "Can't have both 'noMetadata' and ",
712 "'useSvmProps' options set!\n";
713 } elsif ($self->use_svnsync_props) {
714 die "Can't have both 'useSvnsyncProps' and ",
715 "'useSvmProps' options set!\n";
716 }
717 $ra = $self->_set_svm_vars($ra);
718 $self->{-want_revprops} = 1;
719 }
720 $ra;
721 }
722
723 # prop_walk(PATH, REV, SUB)
724 # -------------------------
725 # Recursively traverse PATH at revision REV and invoke SUB for each
726 # directory that contains a SVN property. SUB will be invoked as
727 # follows: &SUB(gs, path, props); where `gs' is this instance of
728 # Git::SVN, `path' the path to the directory where the properties
729 # `props' were found. The `path' will be relative to point of checkout,
730 # that is, if url://repo/trunk is the current Git branch, and that
731 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
732 # as `path' (note the trailing `/').
733 sub prop_walk {
734 my ($self, $path, $rev, $sub) = @_;
735
736 $path =~ s#^/##;
737 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
738 $path =~ s#^/*#/#g;
739 my $p = $path;
740 # Strip the irrelevant part of the path.
741 $p =~ s#^/+\Q@{[$self->path]}\E(/|$)#/#;
742 # Ensure the path is terminated by a `/'.
743 $p =~ s#/*$#/#;
744
745 # The properties contain all the internal SVN stuff nobody
746 # (usually) cares about.
747 my $interesting_props = 0;
748 foreach (keys %{$props}) {
749 # If it doesn't start with `svn:', it must be a
750 # user-defined property.
751 ++$interesting_props and next if $_ !~ /^svn:/;
752 # FIXME: Fragile, if SVN adds new public properties,
753 # this needs to be updated.
754 ++$interesting_props if /^svn:(?:ignore|keywords|executable
755 |eol-style|mime-type
756 |externals|needs-lock)$/x;
757 }
758 &$sub($self, $p, $props) if $interesting_props;
759
760 foreach (sort keys %$dirent) {
761 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
762 $self->prop_walk($self->path . $p . $_, $rev, $sub);
763 }
764 }
765
766 sub last_rev { ($_[0]->last_rev_commit)[0] }
767 sub last_commit { ($_[0]->last_rev_commit)[1] }
768
769 # returns the newest SVN revision number and newest commit SHA1
770 sub last_rev_commit {
771 my ($self) = @_;
772 if (defined $self->{last_rev} && defined $self->{last_commit}) {
773 return ($self->{last_rev}, $self->{last_commit});
774 }
775 my $c = ::verify_ref($self->refname.'^0');
776 if ($c && !$self->use_svm_props && !$self->no_metadata) {
777 my $rev = (::cmt_metadata($c))[1];
778 if (defined $rev) {
779 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
780 return ($rev, $c);
781 }
782 }
783 my $map_path = $self->map_path;
784 unless (-e $map_path) {
785 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
786 return (undef, undef);
787 }
788 my ($rev, $commit) = $self->rev_map_max(1);
789 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
790 return ($rev, $commit);
791 }
792
793 sub get_fetch_range {
794 my ($self, $min, $max) = @_;
795 $max ||= $self->ra->get_latest_revnum;
796 $min ||= $self->rev_map_max;
797 (++$min, $max);
798 }
799
800 sub tmp_config {
801 my (@args) = @_;
802 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
803 my $config = "$ENV{GIT_DIR}/svn/.metadata";
804 if (! -f $config && -f $old_def_config) {
805 rename $old_def_config, $config or
806 die "Failed rename $old_def_config => $config: $!\n";
807 }
808 my $old_config = $ENV{GIT_CONFIG};
809 $ENV{GIT_CONFIG} = $config;
810 $@ = undef;
811 my @ret = eval {
812 unless (-f $config) {
813 mkfile($config);
814 open my $fh, '>', $config or
815 die "Can't open $config: $!\n";
816 print $fh "; This file is used internally by ",
817 "git-svn\n" or die
818 "Couldn't write to $config: $!\n";
819 print $fh "; You should not have to edit it\n" or
820 die "Couldn't write to $config: $!\n";
821 close $fh or die "Couldn't close $config: $!\n";
822 }
823 command('config', @args);
824 };
825 my $err = $@;
826 if (defined $old_config) {
827 $ENV{GIT_CONFIG} = $old_config;
828 } else {
829 delete $ENV{GIT_CONFIG};
830 }
831 die $err if $err;
832 wantarray ? @ret : $ret[0];
833 }
834
835 sub tmp_index_do {
836 my ($self, $sub) = @_;
837 my $old_index = $ENV{GIT_INDEX_FILE};
838 $ENV{GIT_INDEX_FILE} = $self->{index};
839 $@ = undef;
840 my @ret = eval {
841 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
842 mkpath([$dir]) unless -d $dir;
843 &$sub;
844 };
845 my $err = $@;
846 if (defined $old_index) {
847 $ENV{GIT_INDEX_FILE} = $old_index;
848 } else {
849 delete $ENV{GIT_INDEX_FILE};
850 }
851 die $err if $err;
852 wantarray ? @ret : $ret[0];
853 }
854
855 sub assert_index_clean {
856 my ($self, $treeish) = @_;
857
858 $self->tmp_index_do(sub {
859 command_noisy('read-tree', $treeish) unless -e $self->{index};
860 my $x = command_oneline('write-tree');
861 my ($y) = (command(qw/cat-file commit/, $treeish) =~
862 /^tree ($::sha1)/mo);
863 return if $y eq $x;
864
865 warn "Index mismatch: $y != $x\nrereading $treeish\n";
866 unlink $self->{index} or die "unlink $self->{index}: $!\n";
867 command_noisy('read-tree', $treeish);
868 $x = command_oneline('write-tree');
869 if ($y ne $x) {
870 fatal "trees ($treeish) $y != $x\n",
871 "Something is seriously wrong...";
872 }
873 });
874 }
875
876 sub get_commit_parents {
877 my ($self, $log_entry) = @_;
878 my (%seen, @ret, @tmp);
879 # legacy support for 'set-tree'; this is only used by set_tree_cb:
880 if (my $ip = $self->{inject_parents}) {
881 if (my $commit = delete $ip->{$log_entry->{revision}}) {
882 push @tmp, $commit;
883 }
884 }
885 if (my $cur = ::verify_ref($self->refname.'^0')) {
886 push @tmp, $cur;
887 }
888 if (my $ipd = $self->{inject_parents_dcommit}) {
889 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
890 push @tmp, @$commit;
891 }
892 }
893 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
894 while (my $p = shift @tmp) {
895 next if $seen{$p};
896 $seen{$p} = 1;
897 push @ret, $p;
898 }
899 @ret;
900 }
901
902 sub rewrite_root {
903 my ($self) = @_;
904 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
905 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
906 my $rwr = eval { command_oneline(qw/config --get/, $k) };
907 if ($rwr) {
908 $rwr =~ s#/+$##;
909 if ($rwr !~ m#^[a-z\+]+://#) {
910 die "$rwr is not a valid URL (key: $k)\n";
911 }
912 }
913 $self->{-rewrite_root} = $rwr;
914 }
915
916 sub rewrite_uuid {
917 my ($self) = @_;
918 return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
919 my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
920 my $rwid = eval { command_oneline(qw/config --get/, $k) };
921 if ($rwid) {
922 $rwid =~ s#/+$##;
923 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
924 die "$rwid is not a valid UUID (key: $k)\n";
925 }
926 }
927 $self->{-rewrite_uuid} = $rwid;
928 }
929
930 sub metadata_url {
931 my ($self) = @_;
932 ($self->rewrite_root || $self->url) .
933 (length $self->path ? '/' . $self->path : '');
934 }
935
936 sub full_url {
937 my ($self) = @_;
938 $self->url . (length $self->path ? '/' . $self->path : '');
939 }
940
941 sub full_pushurl {
942 my ($self) = @_;
943 if ($self->{pushurl}) {
944 return $self->{pushurl} . (length $self->path ? '/' .
945 $self->path : '');
946 } else {
947 return $self->full_url;
948 }
949 }
950
951 sub set_commit_header_env {
952 my ($log_entry) = @_;
953 my %env;
954 foreach my $ned (qw/NAME EMAIL DATE/) {
955 foreach my $ac (qw/AUTHOR COMMITTER/) {
956 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
957 }
958 }
959
960 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
961 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
962 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
963
964 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
965 ? $log_entry->{commit_name}
966 : $log_entry->{name};
967 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
968 ? $log_entry->{commit_email}
969 : $log_entry->{email};
970 \%env;
971 }
972
973 sub restore_commit_header_env {
974 my ($env) = @_;
975 foreach my $ned (qw/NAME EMAIL DATE/) {
976 foreach my $ac (qw/AUTHOR COMMITTER/) {
977 my $k = "GIT_${ac}_${ned}";
978 if (defined $env->{$k}) {
979 $ENV{$k} = $env->{$k};
980 } else {
981 delete $ENV{$k};
982 }
983 }
984 }
985 }
986
987 sub gc {
988 command_noisy('gc', '--auto');
989 };
990
991 sub do_git_commit {
992 my ($self, $log_entry) = @_;
993 my $lr = $self->last_rev;
994 if (defined $lr && $lr >= $log_entry->{revision}) {
995 die "Last fetched revision of ", $self->refname,
996 " was r$lr, but we are about to fetch: ",
997 "r$log_entry->{revision}!\n";
998 }
999 if (my $c = $self->rev_map_get($log_entry->{revision})) {
1000 croak "$log_entry->{revision} = $c already exists! ",
1001 "Why are we refetching it?\n";
1002 }
1003 my $old_env = set_commit_header_env($log_entry);
1004 my $tree = $log_entry->{tree};
1005 if (!defined $tree) {
1006 $tree = $self->tmp_index_do(sub {
1007 command_oneline('write-tree') });
1008 }
1009 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1010
1011 my @exec = ('git', 'commit-tree', $tree);
1012 foreach ($self->get_commit_parents($log_entry)) {
1013 push @exec, '-p', $_;
1014 }
1015 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1016 or croak $!;
1017 binmode $msg_fh;
1018
1019 # we always get UTF-8 from SVN, but we may want our commits in
1020 # a different encoding.
1021 if (my $enc = Git::config('i18n.commitencoding')) {
1022 require Encode;
1023 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
1024 }
1025 print $msg_fh $log_entry->{log} or croak $!;
1026 restore_commit_header_env($old_env);
1027 unless ($self->no_metadata) {
1028 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
1029 or croak $!;
1030 }
1031 $msg_fh->flush == 0 or croak $!;
1032 close $msg_fh or croak $!;
1033 chomp(my $commit = do { local $/; <$out_fh> });
1034 close $out_fh or croak $!;
1035 waitpid $pid, 0;
1036 croak $? if $?;
1037 if ($commit !~ /^$::sha1$/o) {
1038 die "Failed to commit, invalid sha1: $commit\n";
1039 }
1040
1041 $self->rev_map_set($log_entry->{revision}, $commit, 1);
1042
1043 $self->{last_rev} = $log_entry->{revision};
1044 $self->{last_commit} = $commit;
1045 print "r$log_entry->{revision}" unless $::_q > 1;
1046 if (defined $log_entry->{svm_revision}) {
1047 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
1048 $self->rev_map_set($log_entry->{svm_revision}, $commit,
1049 0, $self->svm_uuid);
1050 }
1051 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
1052 if (--$_gc_nr == 0) {
1053 $_gc_nr = $_gc_period;
1054 gc();
1055 }
1056 return $commit;
1057 }
1058
1059 sub match_paths {
1060 my ($self, $paths, $r) = @_;
1061 return 1 if $self->path eq '';
1062 if (my $path = $paths->{"/".$self->path}) {
1063 return ($path->{action} eq 'D') ? 0 : 1;
1064 }
1065 $self->{path_regex} ||= qr{^/\Q@{[$self->path]}\E/};
1066 if (grep /$self->{path_regex}/, keys %$paths) {
1067 return 1;
1068 }
1069 my $c = '';
1070 foreach (split m#/#, $self->path) {
1071 $c .= "/$_";
1072 next unless ($paths->{$c} &&
1073 ($paths->{$c}->{action} =~ /^[AR]$/));
1074 if ($self->ra->check_path($self->path, $r) ==
1075 $SVN::Node::dir) {
1076 return 1;
1077 }
1078 }
1079 return 0;
1080 }
1081
1082 sub find_parent_branch {
1083 my ($self, $paths, $rev) = @_;
1084 return undef unless $self->follow_parent;
1085 unless (defined $paths) {
1086 my $err_handler = $SVN::Error::handler;
1087 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
1088 $self->ra->get_log([$self->path], $rev, $rev, 0, 1, 1,
1089 sub { $paths = $_[0] });
1090 $SVN::Error::handler = $err_handler;
1091 }
1092 return undef unless defined $paths;
1093
1094 # look for a parent from another branch:
1095 my @b_path_components = split m#/#, $self->path;
1096 my @a_path_components;
1097 my $i;
1098 while (@b_path_components) {
1099 $i = $paths->{'/'.join('/', @b_path_components)};
1100 last if $i && defined $i->{copyfrom_path};
1101 unshift(@a_path_components, pop(@b_path_components));
1102 }
1103 return undef unless defined $i && defined $i->{copyfrom_path};
1104 my $branch_from = $i->{copyfrom_path};
1105 if (@a_path_components) {
1106 print STDERR "branch_from: $branch_from => ";
1107 $branch_from .= '/'.join('/', @a_path_components);
1108 print STDERR $branch_from, "\n";
1109 }
1110 my $r = $i->{copyfrom_rev};
1111 my $repos_root = $self->ra->{repos_root};
1112 my $url = $self->ra->url;
1113 my $new_url = $url . $branch_from;
1114 print STDERR "Found possible branch point: ",
1115 "$new_url => ", $self->full_url, ", $r\n"
1116 unless $::_q > 1;
1117 $branch_from =~ s#^/##;
1118 my $gs = $self->other_gs($new_url, $url,
1119 $branch_from, $r, $self->{ref_id});
1120 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1121 {
1122 my ($base, $head);
1123 if (!defined $r0 || !defined $parent) {
1124 ($base, $head) = parse_revision_argument(0, $r);
1125 } else {
1126 if ($r0 < $r) {
1127 $gs->ra->get_log([$gs->path], $r0 + 1, $r, 1,
1128 0, 1, sub { $base = $_[1] - 1 });
1129 }
1130 }
1131 if (defined $base && $base <= $r) {
1132 $gs->fetch($base, $r);
1133 }
1134 ($r0, $parent) = $gs->find_rev_before($r, 1);
1135 }
1136 if (defined $r0 && defined $parent) {
1137 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
1138 unless $::_q > 1;
1139 my $ed;
1140 if ($self->ra->can_do_switch) {
1141 $self->assert_index_clean($parent);
1142 print STDERR "Following parent with do_switch\n"
1143 unless $::_q > 1;
1144 # do_switch works with svn/trunk >= r22312, but that
1145 # is not included with SVN 1.4.3 (the latest version
1146 # at the moment), so we can't rely on it
1147 $self->{last_rev} = $r0;
1148 $self->{last_commit} = $parent;
1149 $ed = Git::SVN::Fetcher->new($self, $gs->path);
1150 $gs->ra->gs_do_switch($r0, $rev, $gs,
1151 $self->full_url, $ed)
1152 or die "SVN connection failed somewhere...\n";
1153 } elsif ($self->ra->trees_match($new_url, $r0,
1154 $self->full_url, $rev)) {
1155 print STDERR "Trees match:\n",
1156 " $new_url\@$r0\n",
1157 " ${\$self->full_url}\@$rev\n",
1158 "Following parent with no changes\n"
1159 unless $::_q > 1;
1160 $self->tmp_index_do(sub {
1161 command_noisy('read-tree', $parent);
1162 });
1163 $self->{last_commit} = $parent;
1164 } else {
1165 print STDERR "Following parent with do_update\n"
1166 unless $::_q > 1;
1167 $ed = Git::SVN::Fetcher->new($self);
1168 $self->ra->gs_do_update($rev, $rev, $self, $ed)
1169 or die "SVN connection failed somewhere...\n";
1170 }
1171 print STDERR "Successfully followed parent\n" unless $::_q > 1;
1172 return $self->make_log_entry($rev, [$parent], $ed);
1173 }
1174 return undef;
1175 }
1176
1177 sub do_fetch {
1178 my ($self, $paths, $rev) = @_;
1179 my $ed;
1180 my ($last_rev, @parents);
1181 if (my $lc = $self->last_commit) {
1182 # we can have a branch that was deleted, then re-added
1183 # under the same name but copied from another path, in
1184 # which case we'll have multiple parents (we don't
1185 # want to break the original ref, nor lose copypath info):
1186 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1187 push @{$log_entry->{parents}}, $lc;
1188 return $log_entry;
1189 }
1190 $ed = Git::SVN::Fetcher->new($self);
1191 $last_rev = $self->{last_rev};
1192 $ed->{c} = $lc;
1193 @parents = ($lc);
1194 } else {
1195 $last_rev = $rev;
1196 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1197 return $log_entry;
1198 }
1199 $ed = Git::SVN::Fetcher->new($self);
1200 }
1201 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
1202 die "SVN connection failed somewhere...\n";
1203 }
1204 $self->make_log_entry($rev, \@parents, $ed);
1205 }
1206
1207 sub mkemptydirs {
1208 my ($self, $r) = @_;
1209
1210 sub scan {
1211 my ($r, $empty_dirs, $line) = @_;
1212 if (defined $r && $line =~ /^r(\d+)$/) {
1213 return 0 if $1 > $r;
1214 } elsif ($line =~ /^ \+empty_dir: (.+)$/) {
1215 $empty_dirs->{$1} = 1;
1216 } elsif ($line =~ /^ \-empty_dir: (.+)$/) {
1217 my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
1218 delete @$empty_dirs{@d};
1219 }
1220 1; # continue
1221 };
1222
1223 my %empty_dirs = ();
1224 my $gz_file = "$self->{dir}/unhandled.log.gz";
1225 if (-f $gz_file) {
1226 if (!can_compress()) {
1227 warn "Compress::Zlib could not be found; ",
1228 "empty directories in $gz_file will not be read\n";
1229 } else {
1230 my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
1231 die "Unable to open $gz_file: $!\n";
1232 my $line;
1233 while ($gz->gzreadline($line) > 0) {
1234 scan($r, \%empty_dirs, $line) or last;
1235 }
1236 $gz->gzclose;
1237 }
1238 }
1239
1240 if (open my $fh, '<', "$self->{dir}/unhandled.log") {
1241 binmode $fh or croak "binmode: $!";
1242 while (<$fh>) {
1243 scan($r, \%empty_dirs, $_) or last;
1244 }
1245 close $fh;
1246 }
1247
1248 my $strip = qr/\A\Q@{[$self->path]}\E(?:\/|$)/;
1249 foreach my $d (sort keys %empty_dirs) {
1250 $d = uri_decode($d);
1251 $d =~ s/$strip//;
1252 next unless length($d);
1253 next if -d $d;
1254 if (-e $d) {
1255 warn "$d exists but is not a directory\n";
1256 } else {
1257 print "creating empty directory: $d\n";
1258 mkpath([$d]);
1259 }
1260 }
1261 }
1262
1263 sub get_untracked {
1264 my ($self, $ed) = @_;
1265 my @out;
1266 my $h = $ed->{empty};
1267 foreach (sort keys %$h) {
1268 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1269 push @out, " $act: " . uri_encode($_);
1270 warn "W: $act: $_\n";
1271 }
1272 foreach my $t (qw/dir_prop file_prop/) {
1273 $h = $ed->{$t} or next;
1274 foreach my $path (sort keys %$h) {
1275 my $ppath = $path eq '' ? '.' : $path;
1276 foreach my $prop (sort keys %{$h->{$path}}) {
1277 next if $SKIP_PROP{$prop};
1278 my $v = $h->{$path}->{$prop};
1279 my $t_ppath_prop = "$t: " .
1280 uri_encode($ppath) . ' ' .
1281 uri_encode($prop);
1282 if (defined $v) {
1283 push @out, " +$t_ppath_prop " .
1284 uri_encode($v);
1285 } else {
1286 push @out, " -$t_ppath_prop";
1287 }
1288 }
1289 }
1290 }
1291 foreach my $t (qw/absent_file absent_directory/) {
1292 $h = $ed->{$t} or next;
1293 foreach my $parent (sort keys %$h) {
1294 foreach my $path (sort @{$h->{$parent}}) {
1295 push @out, " $t: " .
1296 uri_encode("$parent/$path");
1297 warn "W: $t: $parent/$path ",
1298 "Insufficient permissions?\n";
1299 }
1300 }
1301 }
1302 \@out;
1303 }
1304
1305 sub get_tz {
1306 # some systmes don't handle or mishandle %z, so be creative.
1307 my $t = shift || time;
1308 my $gm = timelocal(gmtime($t));
1309 my $sign = qw( + + - )[ $t <=> $gm ];
1310 return sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
1311 }
1312
1313 # parse_svn_date(DATE)
1314 # --------------------
1315 # Given a date (in UTC) from Subversion, return a string in the format
1316 # "<TZ Offset> <local date/time>" that Git will use.
1317 #
1318 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
1319 # is true we'll convert it to the local timezone instead.
1320 sub parse_svn_date {
1321 my $date = shift || return '+0000 1970-01-01 00:00:00';
1322 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1323 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
1324 croak "Unable to parse date: $date\n";
1325 my $parsed_date; # Set next.
1326
1327 if ($Git::SVN::_localtime) {
1328 # Translate the Subversion datetime to an epoch time.
1329 # Begin by switching ourselves to $date's timezone, UTC.
1330 my $old_env_TZ = $ENV{TZ};
1331 $ENV{TZ} = 'UTC';
1332
1333 my $epoch_in_UTC =
1334 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
1335
1336 # Determine our local timezone (including DST) at the
1337 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
1338 # value of TZ, if any, at the time we were run.
1339 if (defined $Git::SVN::Log::TZ) {
1340 $ENV{TZ} = $Git::SVN::Log::TZ;
1341 } else {
1342 delete $ENV{TZ};
1343 }
1344
1345 my $our_TZ = get_tz();
1346
1347 # This converts $epoch_in_UTC into our local timezone.
1348 my ($sec, $min, $hour, $mday, $mon, $year,
1349 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
1350
1351 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
1352 $our_TZ, $year + 1900, $mon + 1,
1353 $mday, $hour, $min, $sec);
1354
1355 # Reset us to the timezone in effect when we entered
1356 # this routine.
1357 if (defined $old_env_TZ) {
1358 $ENV{TZ} = $old_env_TZ;
1359 } else {
1360 delete $ENV{TZ};
1361 }
1362 } else {
1363 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
1364 }
1365
1366 return $parsed_date;
1367 }
1368
1369 sub other_gs {
1370 my ($self, $new_url, $url,
1371 $branch_from, $r, $old_ref_id) = @_;
1372 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
1373 unless ($gs) {
1374 my $ref_id = $old_ref_id;
1375 $ref_id =~ s/\@\d+-*$//;
1376 $ref_id .= "\@$r";
1377 # just grow a tail if we're not unique enough :x
1378 $ref_id .= '-' while find_ref($ref_id);
1379 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
1380 if ($u =~ s#^\Q$url\E(/|$)##) {
1381 $p = $u;
1382 $u = $url;
1383 $repo_id = $self->{repo_id};
1384 }
1385 while (1) {
1386 # It is possible to tag two different subdirectories at
1387 # the same revision. If the url for an existing ref
1388 # does not match, we must either find a ref with a
1389 # matching url or create a new ref by growing a tail.
1390 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
1391 my (undef, $max_commit) = $gs->rev_map_max(1);
1392 last if (!$max_commit);
1393 my ($url) = ::cmt_metadata($max_commit);
1394 last if ($url eq $gs->metadata_url);
1395 $ref_id .= '-';
1396 }
1397 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
1398 }
1399 $gs
1400 }
1401
1402 sub call_authors_prog {
1403 my ($orig_author) = @_;
1404 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
1405 my $author = `$::_authors_prog $orig_author`;
1406 if ($? != 0) {
1407 die "$::_authors_prog failed with exit code $?\n"
1408 }
1409 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
1410 my ($name, $email) = ($1, $2);
1411 $email = undef if length $2 == 0;
1412 return [$name, $email];
1413 } else {
1414 die "Author: $orig_author: $::_authors_prog returned "
1415 . "invalid author format: $author\n";
1416 }
1417 }
1418
1419 sub check_author {
1420 my ($author) = @_;
1421 if (!defined $author || length $author == 0) {
1422 $author = '(no author)';
1423 }
1424 if (!defined $::users{$author}) {
1425 if (defined $::_authors_prog) {
1426 $::users{$author} = call_authors_prog($author);
1427 } elsif (defined $::_authors) {
1428 die "Author: $author not defined in $::_authors file\n";
1429 }
1430 }
1431 $author;
1432 }
1433
1434 sub find_extra_svk_parents {
1435 my ($self, $ed, $tickets, $parents) = @_;
1436 # aha! svk:merge property changed...
1437 my @tickets = split "\n", $tickets;
1438 my @known_parents;
1439 for my $ticket ( @tickets ) {
1440 my ($uuid, $path, $rev) = split /:/, $ticket;
1441 if ( $uuid eq $self->ra_uuid ) {
1442 my $url = $self->url;
1443 my $repos_root = $url;
1444 my $branch_from = $path;
1445 $branch_from =~ s{^/}{};
1446 my $gs = $self->other_gs($repos_root."/".$branch_from,
1447 $url,
1448 $branch_from,
1449 $rev,
1450 $self->{ref_id});
1451 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
1452 # wahey! we found it, but it might be
1453 # an old one (!)
1454 push @known_parents, [ $rev, $commit ];
1455 }
1456 }
1457 }
1458 # Ordering matters; highest-numbered commit merge tickets
1459 # first, as they may account for later merge ticket additions
1460 # or changes.
1461 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
1462 for my $parent ( @known_parents ) {
1463 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
1464 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
1465 my $new;
1466 while ( <$msg_fh> ) {
1467 $new=1;last;
1468 }
1469 command_close_pipe($msg_fh, $ctx);
1470 if ( $new ) {
1471 print STDERR
1472 "Found merge parent (svk:merge ticket): $parent\n";
1473 push @$parents, $parent;
1474 }
1475 }
1476 }
1477
1478 sub lookup_svn_merge {
1479 my $uuid = shift;
1480 my $url = shift;
1481 my $merge = shift;
1482
1483 my ($source, $revs) = split ":", $merge;
1484 my $path = $source;
1485 $path =~ s{^/}{};
1486 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
1487 if ( !$gs ) {
1488 warn "Couldn't find revmap for $url$source\n";
1489 return;
1490 }
1491 my @ranges = split ",", $revs;
1492 my ($tip, $tip_commit);
1493 my @merged_commit_ranges;
1494 # find the tip
1495 for my $range ( @ranges ) {
1496 my ($bottom, $top) = split "-", $range;
1497 $top ||= $bottom;
1498 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
1499 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
1500
1501 unless ($top_commit and $bottom_commit) {
1502 warn "W:unknown path/rev in svn:mergeinfo "
1503 ."dirprop: $source:$range\n";
1504 next;
1505 }
1506
1507 if (scalar(command('rev-parse', "$bottom_commit^@"))) {
1508 push @merged_commit_ranges,
1509 "$bottom_commit^..$top_commit";
1510 } else {
1511 push @merged_commit_ranges, "$top_commit";
1512 }
1513
1514 if ( !defined $tip or $top > $tip ) {
1515 $tip = $top;
1516 $tip_commit = $top_commit;
1517 }
1518 }
1519 return ($tip_commit, @merged_commit_ranges);
1520 }
1521
1522 sub _rev_list {
1523 my ($msg_fh, $ctx) = command_output_pipe(
1524 "rev-list", @_,
1525 );
1526 my @rv;
1527 while ( <$msg_fh> ) {
1528 chomp;
1529 push @rv, $_;
1530 }
1531 command_close_pipe($msg_fh, $ctx);
1532 @rv;
1533 }
1534
1535 sub check_cherry_pick {
1536 my $base = shift;
1537 my $tip = shift;
1538 my $parents = shift;
1539 my @ranges = @_;
1540 my %commits = map { $_ => 1 }
1541 _rev_list("--no-merges", $tip, "--not", $base, @$parents, "--");
1542 for my $range ( @ranges ) {
1543 delete @commits{_rev_list($range, "--")};
1544 }
1545 for my $commit (keys %commits) {
1546 if (has_no_changes($commit)) {
1547 delete $commits{$commit};
1548 }
1549 }
1550 return (keys %commits);
1551 }
1552
1553 sub has_no_changes {
1554 my $commit = shift;
1555
1556 my @revs = split / /, command_oneline(
1557 qw(rev-list --parents -1 -m), $commit);
1558
1559 # Commits with no parents, e.g. the start of a partial branch,
1560 # have changes by definition.
1561 return 1 if (@revs < 2);
1562
1563 # Commits with multiple parents, e.g a merge, have no changes
1564 # by definition.
1565 return 0 if (@revs > 2);
1566
1567 return (command_oneline("rev-parse", "$commit^{tree}") eq
1568 command_oneline("rev-parse", "$commit~1^{tree}"));
1569 }
1570
1571 sub tie_for_persistent_memoization {
1572 my $hash = shift;
1573 my $path = shift;
1574
1575 if ($can_use_yaml) {
1576 tie %$hash => 'Git::SVN::Memoize::YAML', "$path.yaml";
1577 } else {
1578 tie %$hash => 'Memoize::Storable', "$path.db", 'nstore';
1579 }
1580 }
1581
1582 # The GIT_DIR environment variable is not always set until after the command
1583 # line arguments are processed, so we can't memoize in a BEGIN block.
1584 {
1585 my $memoized = 0;
1586
1587 sub memoize_svn_mergeinfo_functions {
1588 return if $memoized;
1589 $memoized = 1;
1590
1591 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
1592 mkpath([$cache_path]) unless -d $cache_path;
1593
1594 my %lookup_svn_merge_cache;
1595 my %check_cherry_pick_cache;
1596 my %has_no_changes_cache;
1597
1598 tie_for_persistent_memoization(\%lookup_svn_merge_cache,
1599 "$cache_path/lookup_svn_merge");
1600 memoize 'lookup_svn_merge',
1601 SCALAR_CACHE => 'FAULT',
1602 LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
1603 ;
1604
1605 tie_for_persistent_memoization(\%check_cherry_pick_cache,
1606 "$cache_path/check_cherry_pick");
1607 memoize 'check_cherry_pick',
1608 SCALAR_CACHE => 'FAULT',
1609 LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
1610 ;
1611
1612 tie_for_persistent_memoization(\%has_no_changes_cache,
1613 "$cache_path/has_no_changes");
1614 memoize 'has_no_changes',
1615 SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
1616 LIST_CACHE => 'FAULT',
1617 ;
1618 }
1619
1620 sub unmemoize_svn_mergeinfo_functions {
1621 return if not $memoized;
1622 $memoized = 0;
1623
1624 Memoize::unmemoize 'lookup_svn_merge';
1625 Memoize::unmemoize 'check_cherry_pick';
1626 Memoize::unmemoize 'has_no_changes';
1627 }
1628
1629 Memoize::memoize 'Git::SVN::repos_root';
1630 }
1631
1632 END {
1633 # Force cache writeout explicitly instead of waiting for
1634 # global destruction to avoid segfault in Storable:
1635 # http://rt.cpan.org/Public/Bug/Display.html?id=36087
1636 unmemoize_svn_mergeinfo_functions();
1637 }
1638
1639 sub parents_exclude {
1640 my $parents = shift;
1641 my @commits = @_;
1642 return unless @commits;
1643
1644 my @excluded;
1645 my $excluded;
1646 do {
1647 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
1648 $excluded = command_oneline(@cmd);
1649 if ( $excluded ) {
1650 my @new;
1651 my $found;
1652 for my $commit ( @commits ) {
1653 if ( $commit eq $excluded ) {
1654 push @excluded, $commit;
1655 $found++;
1656 last;
1657 }
1658 else {
1659 push @new, $commit;
1660 }
1661 }
1662 die "saw commit '$excluded' in rev-list output, "
1663 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
1664 unless $found;
1665 @commits = @new;
1666 }
1667 }
1668 while ($excluded and @commits);
1669
1670 return @excluded;
1671 }
1672
1673
1674 # note: this function should only be called if the various dirprops
1675 # have actually changed
1676 sub find_extra_svn_parents {
1677 my ($self, $ed, $mergeinfo, $parents) = @_;
1678 # aha! svk:merge property changed...
1679
1680 memoize_svn_mergeinfo_functions();
1681
1682 # We first search for merged tips which are not in our
1683 # history. Then, we figure out which git revisions are in
1684 # that tip, but not this revision. If all of those revisions
1685 # are now marked as merge, we can add the tip as a parent.
1686 my @merges = split "\n", $mergeinfo;
1687 my @merge_tips;
1688 my $url = $self->url;
1689 my $uuid = $self->ra_uuid;
1690 my %ranges;
1691 for my $merge ( @merges ) {
1692 my ($tip_commit, @ranges) =
1693 lookup_svn_merge( $uuid, $url, $merge );
1694 unless (!$tip_commit or
1695 grep { $_ eq $tip_commit } @$parents ) {
1696 push @merge_tips, $tip_commit;
1697 $ranges{$tip_commit} = \@ranges;
1698 } else {
1699 push @merge_tips, undef;
1700 }
1701 }
1702
1703 my %excluded = map { $_ => 1 }
1704 parents_exclude($parents, grep { defined } @merge_tips);
1705
1706 # check merge tips for new parents
1707 my @new_parents;
1708 for my $merge_tip ( @merge_tips ) {
1709 my $spec = shift @merges;
1710 next unless $merge_tip and $excluded{$merge_tip};
1711
1712 my $ranges = $ranges{$merge_tip};
1713
1714 # check out 'new' tips
1715 my $merge_base;
1716 eval {
1717 $merge_base = command_oneline(
1718 "merge-base",
1719 @$parents, $merge_tip,
1720 );
1721 };
1722 if ($@) {
1723 die "An error occurred during merge-base"
1724 unless $@->isa("Git::Error::Command");
1725
1726 warn "W: Cannot find common ancestor between ".
1727 "@$parents and $merge_tip. Ignoring merge info.\n";
1728 next;
1729 }
1730
1731 # double check that there are no missing non-merge commits
1732 my (@incomplete) = check_cherry_pick(
1733 $merge_base, $merge_tip,
1734 $parents,
1735 @$ranges,
1736 );
1737
1738 if ( @incomplete ) {
1739 warn "W:svn cherry-pick ignored ($spec) - missing "
1740 .@incomplete." commit(s) (eg $incomplete[0])\n";
1741 } else {
1742 warn
1743 "Found merge parent (svn:mergeinfo prop): ",
1744 $merge_tip, "\n";
1745 push @new_parents, $merge_tip;
1746 }
1747 }
1748
1749 # cater for merges which merge commits from multiple branches
1750 if ( @new_parents > 1 ) {
1751 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
1752 for ( my $j = 0; $j <= $#new_parents; $j++ ) {
1753 next if $i == $j;
1754 next unless $new_parents[$i];
1755 next unless $new_parents[$j];
1756 my $revs = command_oneline(
1757 "rev-list", "-1",
1758 "$new_parents[$i]..$new_parents[$j]",
1759 );
1760 if ( !$revs ) {
1761 undef($new_parents[$j]);
1762 }
1763 }
1764 }
1765 }
1766 push @$parents, grep { defined } @new_parents;
1767 }
1768
1769 sub make_log_entry {
1770 my ($self, $rev, $parents, $ed) = @_;
1771 my $untracked = $self->get_untracked($ed);
1772
1773 my @parents = @$parents;
1774 my $ps = $ed->{path_strip} || "";
1775 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
1776 my $props = $ed->{dir_prop}{$path};
1777 if ( $props->{"svk:merge"} ) {
1778 $self->find_extra_svk_parents
1779 ($ed, $props->{"svk:merge"}, \@parents);
1780 }
1781 if ( $props->{"svn:mergeinfo"} ) {
1782 $self->find_extra_svn_parents
1783 ($ed,
1784 $props->{"svn:mergeinfo"},
1785 \@parents);
1786 }
1787 }
1788
1789 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1790 print $un "r$rev\n" or croak $!;
1791 print $un $_, "\n" foreach @$untracked;
1792 my %log_entry = ( parents => \@parents, revision => $rev,
1793 log => '');
1794
1795 my $headrev;
1796 my $logged = delete $self->{logged_rev_props};
1797 if (!$logged || $self->{-want_revprops}) {
1798 my $rp = $self->ra->rev_proplist($rev);
1799 foreach (sort keys %$rp) {
1800 my $v = $rp->{$_};
1801 if (/^svn:(author|date|log)$/) {
1802 $log_entry{$1} = $v;
1803 } elsif ($_ eq 'svm:headrev') {
1804 $headrev = $v;
1805 } else {
1806 print $un " rev_prop: ", uri_encode($_), ' ',
1807 uri_encode($v), "\n";
1808 }
1809 }
1810 } else {
1811 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
1812 }
1813 close $un or croak $!;
1814
1815 $log_entry{date} = parse_svn_date($log_entry{date});
1816 $log_entry{log} .= "\n";
1817 my $author = $log_entry{author} = check_author($log_entry{author});
1818 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
1819 : ($author, undef);
1820
1821 my ($commit_name, $commit_email) = ($name, $email);
1822 if ($_use_log_author) {
1823 my $name_field;
1824 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
1825 $name_field = $1;
1826 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
1827 $name_field = $1;
1828 }
1829 if (!defined $name_field) {
1830 if (!defined $email) {
1831 $email = $name;
1832 }
1833 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
1834 ($name, $email) = ($1, $2);
1835 } elsif ($name_field =~ /(.*)@/) {
1836 ($name, $email) = ($1, $name_field);
1837 } else {
1838 ($name, $email) = ($name_field, $name_field);
1839 }
1840 }
1841 if (defined $headrev && $self->use_svm_props) {
1842 if ($self->rewrite_root) {
1843 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
1844 "options set!\n";
1845 }
1846 if ($self->rewrite_uuid) {
1847 die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
1848 "options set!\n";
1849 }
1850 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
1851 # we don't want "SVM: initializing mirror for junk" ...
1852 return undef if $r == 0;
1853 my $svm = $self->svm;
1854 if ($uuid ne $svm->{uuid}) {
1855 die "UUID mismatch on SVM path:\n",
1856 "expected: $svm->{uuid}\n",
1857 " got: $uuid\n";
1858 }
1859 my $full_url = $self->full_url;
1860 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
1861 die "Failed to replace '$svm->{replace}' with ",
1862 "'$svm->{source}' in $full_url\n";
1863 # throw away username for storing in records
1864 remove_username($full_url);
1865 $log_entry{metadata} = "$full_url\@$r $uuid";
1866 $log_entry{svm_revision} = $r;
1867 $email ||= "$author\@$uuid";
1868 $commit_email ||= "$author\@$uuid";
1869 } elsif ($self->use_svnsync_props) {
1870 my $full_url = $self->svnsync->{url};
1871 $full_url .= "/".$self->path if length $self->path;
1872 remove_username($full_url);
1873 my $uuid = $self->svnsync->{uuid};
1874 $log_entry{metadata} = "$full_url\@$rev $uuid";
1875 $email ||= "$author\@$uuid";
1876 $commit_email ||= "$author\@$uuid";
1877 } else {
1878 my $url = $self->metadata_url;
1879 remove_username($url);
1880 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
1881 $log_entry{metadata} = "$url\@$rev " . $uuid;
1882 $email ||= "$author\@" . $uuid;
1883 $commit_email ||= "$author\@" . $uuid;
1884 }
1885 $log_entry{name} = $name;
1886 $log_entry{email} = $email;
1887 $log_entry{commit_name} = $commit_name;
1888 $log_entry{commit_email} = $commit_email;
1889 \%log_entry;
1890 }
1891
1892 sub fetch {
1893 my ($self, $min_rev, $max_rev, @parents) = @_;
1894 my ($last_rev, $last_commit) = $self->last_rev_commit;
1895 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
1896 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
1897 }
1898
1899 sub set_tree_cb {
1900 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1901 $self->{inject_parents} = { $rev => $tree };
1902 $self->fetch(undef, undef);
1903 }
1904
1905 sub set_tree {
1906 my ($self, $tree) = (shift, shift);
1907 my $log_entry = ::get_commit_entry($tree);
1908 unless ($self->{last_rev}) {
1909 fatal("Must have an existing revision to commit");
1910 }
1911 my %ed_opts = ( r => $self->{last_rev},
1912 log => $log_entry->{log},
1913 ra => $self->ra,
1914 tree_a => $self->{last_commit},
1915 tree_b => $tree,
1916 editor_cb => sub {
1917 $self->set_tree_cb($log_entry, $tree, @_) },
1918 svn_path => $self->path );
1919 if (!Git::SVN::Editor->new(\%ed_opts)->apply_diff) {
1920 print "No changes\nr$self->{last_rev} = $tree\n";
1921 }
1922 }
1923
1924 sub rebuild_from_rev_db {
1925 my ($self, $path) = @_;
1926 my $r = -1;
1927 open my $fh, '<', $path or croak "open: $!";
1928 binmode $fh or croak "binmode: $!";
1929 while (<$fh>) {
1930 length($_) == 41 or croak "inconsistent size in ($_) != 41";
1931 chomp($_);
1932 ++$r;
1933 next if $_ eq ('0' x 40);
1934 $self->rev_map_set($r, $_);
1935 print "r$r = $_\n";
1936 }
1937 close $fh or croak "close: $!";
1938 unlink $path or croak "unlink: $!";
1939 }
1940
1941 sub rebuild {
1942 my ($self) = @_;
1943 my $map_path = $self->map_path;
1944 my $partial = (-e $map_path && ! -z $map_path);
1945 return unless ::verify_ref($self->refname.'^0');
1946 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
1947 my $rev_db = $self->rev_db_path;
1948 $self->rebuild_from_rev_db($rev_db);
1949 if ($self->use_svm_props) {
1950 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
1951 $self->rebuild_from_rev_db($svm_rev_db);
1952 }
1953 $self->unlink_rev_db_symlink;
1954 return;
1955 }
1956 print "Rebuilding $map_path ...\n" if (!$partial);
1957 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
1958 (undef, undef));
1959 my ($log, $ctx) =
1960 command_output_pipe(qw/rev-list --pretty=raw --reverse/,
1961 ($head ? "$head.." : "") . $self->refname,
1962 '--');
1963 my $metadata_url = $self->metadata_url;
1964 remove_username($metadata_url);
1965 my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
1966 my $c;
1967 while (<$log>) {
1968 if ( m{^commit ($::sha1)$} ) {
1969 $c = $1;
1970 next;
1971 }
1972 next unless s{^\s*(git-svn-id:)}{$1};
1973 my ($url, $rev, $uuid) = ::extract_metadata($_);
1974 remove_username($url);
1975
1976 # ignore merges (from set-tree)
1977 next if (!defined $rev || !$uuid);
1978
1979 # if we merged or otherwise started elsewhere, this is
1980 # how we break out of it
1981 if (($uuid ne $svn_uuid) ||
1982 ($metadata_url && $url && ($url ne $metadata_url))) {
1983 next;
1984 }
1985 if ($partial && $head) {
1986 print "Partial-rebuilding $map_path ...\n";
1987 print "Currently at $base_rev = $head\n";
1988 $head = undef;
1989 }
1990
1991 $self->rev_map_set($rev, $c);
1992 print "r$rev = $c\n";
1993 }
1994 command_close_pipe($log, $ctx);
1995 print "Done rebuilding $map_path\n" if (!$partial || !$head);
1996 my $rev_db_path = $self->rev_db_path;
1997 if (-f $self->rev_db_path) {
1998 unlink $self->rev_db_path or croak "unlink: $!";
1999 }
2000 $self->unlink_rev_db_symlink;
2001 }
2002
2003 # rev_map:
2004 # Tie::File seems to be prone to offset errors if revisions get sparse,
2005 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
2006 # one of my favorite modules is out :< Next up would be one of the DBM
2007 # modules, but I'm not sure which is most portable...
2008 #
2009 # This is the replacement for the rev_db format, which was too big
2010 # and inefficient for large repositories with a lot of sparse history
2011 # (mainly tags)
2012 #
2013 # The format is this:
2014 # - 24 bytes for every record,
2015 # * 4 bytes for the integer representing an SVN revision number
2016 # * 20 bytes representing the sha1 of a git commit
2017 # - No empty padding records like the old format
2018 # (except the last record, which can be overwritten)
2019 # - new records are written append-only since SVN revision numbers
2020 # increase monotonically
2021 # - lookups on SVN revision number are done via a binary search
2022 # - Piping the file to xxd -c24 is a good way of dumping it for
2023 # viewing or editing (piped back through xxd -r), should the need
2024 # ever arise.
2025 # - The last record can be padding revision with an all-zero sha1
2026 # This is used to optimize fetch performance when using multiple
2027 # "fetch" directives in .git/config
2028 #
2029 # These files are disposable unless noMetadata or useSvmProps is set
2030
2031 sub _rev_map_set {
2032 my ($fh, $rev, $commit) = @_;
2033
2034 binmode $fh or croak "binmode: $!";
2035 my $size = (stat($fh))[7];
2036 ($size % 24) == 0 or croak "inconsistent size: $size";
2037
2038 my $wr_offset = 0;
2039 if ($size > 0) {
2040 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2041 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2042 $read == 24 or croak "read only $read bytes (!= 24)";
2043 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2044 if ($last_commit eq ('0' x40)) {
2045 if ($size >= 48) {
2046 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2047 $read = sysread($fh, $buf, 24) or
2048 croak "read: $!";
2049 $read == 24 or
2050 croak "read only $read bytes (!= 24)";
2051 ($last_rev, $last_commit) =
2052 unpack(rev_map_fmt, $buf);
2053 if ($last_commit eq ('0' x40)) {
2054 croak "inconsistent .rev_map\n";
2055 }
2056 }
2057 if ($last_rev >= $rev) {
2058 croak "last_rev is higher!: $last_rev >= $rev";
2059 }
2060 $wr_offset = -24;
2061 }
2062 }
2063 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2064 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2065 croak "write: $!";
2066 }
2067
2068 sub _rev_map_reset {
2069 my ($fh, $rev, $commit) = @_;
2070 my $c = _rev_map_get($fh, $rev);
2071 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
2072 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
2073 truncate $fh, $offset or croak "truncate: $!";
2074 }
2075
2076 sub mkfile {
2077 my ($path) = @_;
2078 unless (-e $path) {
2079 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2080 mkpath([$dir]) unless -d $dir;
2081 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2082 close $fh or die "Couldn't close (create) $path: $!\n";
2083 }
2084 }
2085
2086 sub rev_map_set {
2087 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2088 defined $commit or die "missing arg3\n";
2089 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2090 my $db = $self->map_path($uuid);
2091 my $db_lock = "$db.lock";
2092 my $sigmask;
2093 $update_ref ||= 0;
2094 if ($update_ref) {
2095 $sigmask = POSIX::SigSet->new();
2096 my $signew = POSIX::SigSet->new(SIGINT, SIGHUP, SIGTERM,
2097 SIGALRM, SIGUSR1, SIGUSR2);
2098 sigprocmask(SIG_BLOCK, $signew, $sigmask) or
2099 croak "Can't block signals: $!";
2100 }
2101 mkfile($db);
2102
2103 $LOCKFILES{$db_lock} = 1;
2104 my $sync;
2105 # both of these options make our .rev_db file very, very important
2106 # and we can't afford to lose it because rebuild() won't work
2107 if ($self->use_svm_props || $self->no_metadata) {
2108 $sync = 1;
2109 copy($db, $db_lock) or die "rev_map_set(@_): ",
2110 "Failed to copy: ",
2111 "$db => $db_lock ($!)\n";
2112 } else {
2113 rename $db, $db_lock or die "rev_map_set(@_): ",
2114 "Failed to rename: ",
2115 "$db => $db_lock ($!)\n";
2116 }
2117
2118 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2119 or croak "Couldn't open $db_lock: $!\n";
2120 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
2121 _rev_map_set($fh, $rev, $commit);
2122 if ($sync) {
2123 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2124 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2125 }
2126 close $fh or croak $!;
2127 if ($update_ref) {
2128 $_head = $self;
2129 my $note = "";
2130 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
2131 command_noisy('update-ref', '-m', "r$rev$note",
2132 $self->refname, $commit);
2133 }
2134 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2135 "$db_lock => $db ($!)\n";
2136 delete $LOCKFILES{$db_lock};
2137 if ($update_ref) {
2138 sigprocmask(SIG_SETMASK, $sigmask) or
2139 croak "Can't restore signal mask: $!";
2140 }
2141 }
2142
2143 # If want_commit, this will return an array of (rev, commit) where
2144 # commit _must_ be a valid commit in the archive.
2145 # Otherwise, it'll return the max revision (whether or not the
2146 # commit is valid or just a 0x40 placeholder).
2147 sub rev_map_max {
2148 my ($self, $want_commit) = @_;
2149 $self->rebuild;
2150 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
2151 $want_commit ? ($r, $c) : $r;
2152 }
2153
2154 sub rev_map_max_norebuild {
2155 my ($self, $want_commit) = @_;
2156 my $map_path = $self->map_path;
2157 stat $map_path or return $want_commit ? (0, undef) : 0;
2158 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2159 binmode $fh or croak "binmode: $!";
2160 my $size = (stat($fh))[7];
2161 ($size % 24) == 0 or croak "inconsistent size: $size";
2162
2163 if ($size == 0) {
2164 close $fh or croak "close: $!";
2165 return $want_commit ? (0, undef) : 0;
2166 }
2167
2168 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2169 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2170 my ($r, $c) = unpack(rev_map_fmt, $buf);
2171 if ($want_commit && $c eq ('0' x40)) {
2172 if ($size < 48) {
2173 return $want_commit ? (0, undef) : 0;
2174 }
2175 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2176 sysread($fh, $buf, 24) == 24 or croak "read: $!";
2177 ($r, $c) = unpack(rev_map_fmt, $buf);
2178 if ($c eq ('0'x40)) {
2179 croak "Penultimate record is all-zeroes in $map_path";
2180 }
2181 }
2182 close $fh or croak "close: $!";
2183 $want_commit ? ($r, $c) : $r;
2184 }
2185
2186 sub rev_map_get {
2187 my ($self, $rev, $uuid) = @_;
2188 my $map_path = $self->map_path($uuid);
2189 return undef unless -e $map_path;
2190
2191 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2192 my $c = _rev_map_get($fh, $rev);
2193 close($fh) or croak "close: $!";
2194 $c
2195 }
2196
2197 sub _rev_map_get {
2198 my ($fh, $rev) = @_;
2199
2200 binmode $fh or croak "binmode: $!";
2201 my $size = (stat($fh))[7];
2202 ($size % 24) == 0 or croak "inconsistent size: $size";
2203
2204 if ($size == 0) {
2205 return undef;
2206 }
2207
2208 my ($l, $u) = (0, $size - 24);
2209 my ($r, $c, $buf);
2210
2211 while ($l <= $u) {
2212 my $i = int(($l/24 + $u/24) / 2) * 24;
2213 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2214 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2215 my ($r, $c) = unpack(rev_map_fmt, $buf);
2216
2217 if ($r < $rev) {
2218 $l = $i + 24;
2219 } elsif ($r > $rev) {
2220 $u = $i - 24;
2221 } else { # $r == $rev
2222 return $c eq ('0' x 40) ? undef : $c;
2223 }
2224 }
2225 undef;
2226 }
2227
2228 # Finds the first svn revision that exists on (if $eq_ok is true) or
2229 # before $rev for the current branch. It will not search any lower
2230 # than $min_rev. Returns the git commit hash and svn revision number
2231 # if found, else (undef, undef).
2232 sub find_rev_before {
2233 my ($self, $rev, $eq_ok, $min_rev) = @_;
2234 --$rev unless $eq_ok;
2235 $min_rev ||= 1;
2236 my $max_rev = $self->rev_map_max;
2237 $rev = $max_rev if ($rev > $max_rev);
2238 while ($rev >= $min_rev) {
2239 if (my $c = $self->rev_map_get($rev)) {
2240 return ($rev, $c);
2241 }
2242 --$rev;
2243 }
2244 return (undef, undef);
2245 }
2246
2247 # Finds the first svn revision that exists on (if $eq_ok is true) or
2248 # after $rev for the current branch. It will not search any higher
2249 # than $max_rev. Returns the git commit hash and svn revision number
2250 # if found, else (undef, undef).
2251 sub find_rev_after {
2252 my ($self, $rev, $eq_ok, $max_rev) = @_;
2253 ++$rev unless $eq_ok;
2254 $max_rev ||= $self->rev_map_max;
2255 while ($rev <= $max_rev) {
2256 if (my $c = $self->rev_map_get($rev)) {
2257 return ($rev, $c);
2258 }
2259 ++$rev;
2260 }
2261 return (undef, undef);
2262 }
2263
2264 sub _new {
2265 my ($class, $repo_id, $ref_id, $path) = @_;
2266 unless (defined $repo_id && length $repo_id) {
2267 $repo_id = $default_repo_id;
2268 }
2269 unless (defined $ref_id && length $ref_id) {
2270 # Access the prefix option from the git-svn main program if it's loaded.
2271 my $prefix = defined &::opt_prefix ? ::opt_prefix() : "";
2272 $_[2] = $ref_id =
2273 "refs/remotes/$prefix$default_ref_id";
2274 }
2275 $_[1] = $repo_id;
2276 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2277
2278 # Older repos imported by us used $GIT_DIR/svn/foo instead of
2279 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
2280 if ($ref_id =~ m{^refs/remotes/(.*)}) {
2281 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
2282 if (-d $old_dir && ! -d $dir) {
2283 $dir = $old_dir;
2284 }
2285 }
2286
2287 $_[3] = $path = '' unless (defined $path);
2288 mkpath([$dir]);
2289 my $obj = bless {
2290 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2291 config => "$ENV{GIT_DIR}/svn/config",
2292 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
2293
2294 # Ensure it gets canonicalized
2295 $obj->path($path);
2296
2297 return $obj;
2298 }
2299
2300 sub path {
2301 my $self = shift;
2302
2303 if (@_) {
2304 my $path = shift;
2305 $self->{path} = $path;
2306 return;
2307 }
2308
2309 return $self->{path};
2310 }
2311
2312 sub url {
2313 my $self = shift;
2314
2315 if (@_) {
2316 my $url = shift;
2317 $self->{url} = $url;
2318 return;
2319 }
2320
2321 return $self->{url};
2322 }
2323
2324 # for read-only access of old .rev_db formats
2325 sub unlink_rev_db_symlink {
2326 my ($self) = @_;
2327 my $link = $self->rev_db_path;
2328 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2329 if (-l $link) {
2330 unlink $link or croak "unlink: $link failed!";
2331 }
2332 }
2333
2334 sub rev_db_path {
2335 my ($self, $uuid) = @_;
2336 my $db_path = $self->map_path($uuid);
2337 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2338 or croak "map_path: $db_path does not contain '/.rev_map.' !";
2339 $db_path;
2340 }
2341
2342 # the new replacement for .rev_db
2343 sub map_path {
2344 my ($self, $uuid) = @_;
2345 $uuid ||= $self->ra_uuid;
2346 "$self->{map_root}.$uuid";
2347 }
2348
2349 sub uri_encode {
2350 my ($f) = @_;
2351 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2352 $f
2353 }
2354
2355 sub uri_decode {
2356 my ($f) = @_;
2357 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
2358 $f
2359 }
2360
2361 sub remove_username {
2362 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2363 }
2364
2365 1;
Alternate source for your git.git needs