Commit | Line | Data |
---|---|---|
29499c0b MS |
1 | package Git::SVN; |
2 | use strict; | |
3 | use warnings; | |
4 | use Fcntl qw/:DEFAULT :seek/; | |
5 | use constant rev_map_fmt => 'NH40'; | |
5c71028f | 6 | use vars qw/$_no_metadata |
29499c0b | 7 | $_repack $_repack_flags $_use_svm_props $_head |
5c71028f | 8 | $_use_svnsync_props $no_reuse_existing |
29499c0b MS |
9 | $_use_log_author $_add_author_from $_localtime/; |
10 | use Carp qw/croak/; | |
11 | use File::Path qw/mkpath/; | |
29499c0b | 12 | use IPC::Open3; |
29499c0b | 13 | use Memoize; # core since 5.8.0, Jul 2002 |
29499c0b | 14 | use POSIX qw(:signal_h); |
45c956b3 | 15 | use Time::Local; |
29499c0b MS |
16 | |
17 | use Git qw( | |
18 | command | |
19 | command_oneline | |
20 | command_noisy | |
21 | command_output_pipe | |
22 | command_close_pipe | |
68868ff5 | 23 | get_tz_offset |
29499c0b | 24 | ); |
ca475a61 MS |
25 | use Git::SVN::Utils qw( |
26 | fatal | |
27 | can_compress | |
28 | join_paths | |
565e56c2 MS |
29 | canonicalize_path |
30 | canonicalize_url | |
d2fd119c | 31 | add_path_to_url |
ca475a61 | 32 | ); |
29499c0b | 33 | |
47092c10 | 34 | my $memo_backend; |
5c71028f MS |
35 | our $_follow_parent = 1; |
36 | our $_minimize_url = 'unset'; | |
37 | our $default_repo_id = 'svn'; | |
38 | our $default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn'; | |
39 | ||
29499c0b MS |
40 | my ($_gc_nr, $_gc_period); |
41 | ||
42 | # properties that we do not log: | |
43 | my %SKIP_PROP; | |
44 | BEGIN { | |
45 | %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url | |
46 | svn:special svn:executable | |
47 | svn:entry:committed-rev | |
48 | svn:entry:last-author | |
49 | svn:entry:uuid | |
50 | svn:entry:committed-date/; | |
51 | ||
52 | # some options are read globally, but can be overridden locally | |
53 | # per [svn-remote "..."] section. Command-line options will *NOT* | |
54 | # override options set in an [svn-remote "..."] section | |
55 | no strict 'refs'; | |
56 | for my $option (qw/follow_parent no_metadata use_svm_props | |
57 | use_svnsync_props/) { | |
58 | my $key = $option; | |
59 | $key =~ tr/_//d; | |
60 | my $prop = "-$option"; | |
61 | *$option = sub { | |
62 | my ($self) = @_; | |
63 | return $self->{$prop} if exists $self->{$prop}; | |
64 | my $k = "svn-remote.$self->{repo_id}.$key"; | |
65 | eval { command_oneline(qw/config --get/, $k) }; | |
66 | if ($@) { | |
67 | $self->{$prop} = ${"Git::SVN::_$option"}; | |
68 | } else { | |
69 | my $v = command_oneline(qw/config --bool/,$k); | |
70 | $self->{$prop} = $v eq 'false' ? 0 : 1; | |
71 | } | |
72 | return $self->{$prop}; | |
73 | } | |
74 | } | |
75 | } | |
76 | ||
77 | ||
78 | my (%LOCKFILES, %INDEX_FILES); | |
79 | END { | |
80 | unlink keys %LOCKFILES if %LOCKFILES; | |
81 | unlink keys %INDEX_FILES if %INDEX_FILES; | |
82 | } | |
83 | ||
84 | sub resolve_local_globs { | |
85 | my ($url, $fetch, $glob_spec) = @_; | |
86 | return unless defined $glob_spec; | |
87 | my $ref = $glob_spec->{ref}; | |
88 | my $path = $glob_spec->{path}; | |
89 | foreach (command(qw#for-each-ref --format=%(refname) refs/#)) { | |
90 | next unless m#^$ref->{regex}$#; | |
91 | my $p = $1; | |
92 | my $pathname = desanitize_refname($path->full_path($p)); | |
93 | my $refname = desanitize_refname($ref->full_path($p)); | |
94 | if (my $existing = $fetch->{$pathname}) { | |
95 | if ($existing ne $refname) { | |
96 | die "Refspec conflict:\n", | |
97 | "existing: $existing\n", | |
98 | " globbed: $refname\n"; | |
99 | } | |
2af7da9f EW |
100 | my $u = (::cmt_metadata("$refname"))[0]; |
101 | if (!defined($u)) { | |
102 | warn | |
103 | "W: $refname: no associated commit metadata from SVN, skipping\n"; | |
104 | next; | |
105 | } | |
29499c0b MS |
106 | $u =~ s!^\Q$url\E(/|$)!! or die |
107 | "$refname: '$url' not found in '$u'\n"; | |
108 | if ($pathname ne $u) { | |
109 | warn "W: Refspec glob conflict ", | |
110 | "(ref: $refname):\n", | |
111 | "expected path: $pathname\n", | |
112 | " real path: $u\n", | |
113 | "Continuing ahead with $u\n"; | |
114 | next; | |
115 | } | |
116 | } else { | |
117 | $fetch->{$pathname} = $refname; | |
118 | } | |
119 | } | |
120 | } | |
121 | ||
122 | sub parse_revision_argument { | |
123 | my ($base, $head) = @_; | |
124 | if (!defined $::_revision || $::_revision eq 'BASE:HEAD') { | |
125 | return ($base, $head); | |
126 | } | |
127 | return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/); | |
128 | return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/); | |
129 | return ($head, $head) if ($::_revision eq 'HEAD'); | |
130 | return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/); | |
131 | return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/); | |
132 | die "revision argument: $::_revision not understood by git-svn\n"; | |
133 | } | |
134 | ||
135 | sub fetch_all { | |
136 | my ($repo_id, $remotes) = @_; | |
137 | if (ref $repo_id) { | |
138 | my $gs = $repo_id; | |
139 | $repo_id = undef; | |
140 | $repo_id = $gs->{repo_id}; | |
141 | } | |
142 | $remotes ||= read_all_remotes(); | |
143 | my $remote = $remotes->{$repo_id} or | |
144 | die "[svn-remote \"$repo_id\"] unknown\n"; | |
145 | my $fetch = $remote->{fetch}; | |
146 | my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n"; | |
147 | my (@gs, @globs); | |
148 | my $ra = Git::SVN::Ra->new($url); | |
149 | my $uuid = $ra->get_uuid; | |
150 | my $head = $ra->get_latest_revnum; | |
151 | ||
152 | # ignore errors, $head revision may not even exist anymore | |
153 | eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) }; | |
154 | warn "W: $@\n" if $@; | |
155 | ||
156 | my $base = defined $fetch ? $head : 0; | |
157 | ||
158 | # read the max revs for wildcard expansion (branches/*, tags/*) | |
159 | foreach my $t (qw/branches tags/) { | |
160 | defined $remote->{$t} or next; | |
161 | push @globs, @{$remote->{$t}}; | |
162 | ||
163 | my $max_rev = eval { tmp_config(qw/--int --get/, | |
164 | "svn-remote.$repo_id.${t}-maxRev") }; | |
165 | if (defined $max_rev && ($max_rev < $base)) { | |
166 | $base = $max_rev; | |
167 | } elsif (!defined $max_rev) { | |
168 | $base = 0; | |
169 | } | |
170 | } | |
171 | ||
172 | if ($fetch) { | |
173 | foreach my $p (sort keys %$fetch) { | |
174 | my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p); | |
175 | my $lr = $gs->rev_map_max; | |
176 | if (defined $lr) { | |
177 | $base = $lr if ($lr < $base); | |
178 | } | |
179 | push @gs, $gs; | |
180 | } | |
181 | } | |
182 | ||
183 | ($base, $head) = parse_revision_argument($base, $head); | |
184 | $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs); | |
185 | } | |
186 | ||
187 | sub read_all_remotes { | |
188 | my $r = {}; | |
189 | my $use_svm_props = eval { command_oneline(qw/config --bool | |
190 | svn.useSvmProps/) }; | |
191 | $use_svm_props = $use_svm_props eq 'true' if $use_svm_props; | |
192 | my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*}; | |
193 | foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) { | |
194 | if (m!^(.+)\.fetch=$svn_refspec$!) { | |
195 | my ($remote, $local_ref, $remote_ref) = ($1, $2, $3); | |
196 | die("svn-remote.$remote: remote ref '$remote_ref' " | |
197 | . "must start with 'refs/'\n") | |
198 | unless $remote_ref =~ m{^refs/}; | |
199 | $local_ref = uri_decode($local_ref); | |
200 | $r->{$remote}->{fetch}->{$local_ref} = $remote_ref; | |
201 | $r->{$remote}->{svm} = {} if $use_svm_props; | |
202 | } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) { | |
203 | $r->{$1}->{svm} = {}; | |
204 | } elsif (m!^(.+)\.url=\s*(.*)\s*$!) { | |
93c3fcbe | 205 | $r->{$1}->{url} = canonicalize_url($2); |
29499c0b | 206 | } elsif (m!^(.+)\.pushurl=\s*(.*)\s*$!) { |
93c3fcbe | 207 | $r->{$1}->{pushurl} = canonicalize_url($2); |
29499c0b MS |
208 | } elsif (m!^(.+)\.ignore-refs=\s*(.*)\s*$!) { |
209 | $r->{$1}->{ignore_refs_regex} = $2; | |
210 | } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) { | |
211 | my ($remote, $t, $local_ref, $remote_ref) = | |
212 | ($1, $2, $3, $4); | |
213 | die("svn-remote.$remote: remote ref '$remote_ref' ($t) " | |
214 | . "must start with 'refs/'\n") | |
215 | unless $remote_ref =~ m{^refs/}; | |
216 | $local_ref = uri_decode($local_ref); | |
3d9be15f MS |
217 | |
218 | require Git::SVN::GlobSpec; | |
29499c0b MS |
219 | my $rs = { |
220 | t => $t, | |
221 | remote => $remote, | |
222 | path => Git::SVN::GlobSpec->new($local_ref, 1), | |
223 | ref => Git::SVN::GlobSpec->new($remote_ref, 0) }; | |
224 | if (length($rs->{ref}->{right}) != 0) { | |
225 | die "The '*' glob character must be the last ", | |
226 | "character of '$remote_ref'\n"; | |
227 | } | |
228 | push @{ $r->{$remote}->{$t} }, $rs; | |
229 | } | |
230 | } | |
231 | ||
232 | map { | |
233 | if (defined $r->{$_}->{svm}) { | |
234 | my $svm; | |
235 | eval { | |
236 | my $section = "svn-remote.$_"; | |
237 | $svm = { | |
238 | source => tmp_config('--get', | |
239 | "$section.svm-source"), | |
240 | replace => tmp_config('--get', | |
241 | "$section.svm-replace"), | |
242 | } | |
243 | }; | |
244 | $r->{$_}->{svm} = $svm; | |
245 | } | |
246 | } keys %$r; | |
247 | ||
248 | foreach my $remote (keys %$r) { | |
249 | foreach ( grep { defined $_ } | |
250 | map { $r->{$remote}->{$_} } qw(branches tags) ) { | |
251 | foreach my $rs ( @$_ ) { | |
252 | $rs->{ignore_refs_regex} = | |
253 | $r->{$remote}->{ignore_refs_regex}; | |
254 | } | |
255 | } | |
256 | } | |
257 | ||
258 | $r; | |
259 | } | |
260 | ||
261 | sub init_vars { | |
262 | $_gc_nr = $_gc_period = 1000; | |
263 | if (defined $_repack || defined $_repack_flags) { | |
264 | warn "Repack options are obsolete; they have no effect.\n"; | |
265 | } | |
266 | } | |
267 | ||
268 | sub verify_remotes_sanity { | |
269 | return unless -d $ENV{GIT_DIR}; | |
270 | my %seen; | |
271 | foreach (command(qw/config -l/)) { | |
272 | if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) { | |
273 | if ($seen{$1}) { | |
274 | die "Remote ref refs/remote/$1 is tracked by", | |
275 | "\n \"$_\"\nand\n \"$seen{$1}\"\n", | |
276 | "Please resolve this ambiguity in ", | |
277 | "your git configuration file before ", | |
278 | "continuing\n"; | |
279 | } | |
280 | $seen{$1} = $_; | |
281 | } | |
282 | } | |
283 | } | |
284 | ||
285 | sub find_existing_remote { | |
286 | my ($url, $remotes) = @_; | |
287 | return undef if $no_reuse_existing; | |
288 | my $existing; | |
289 | foreach my $repo_id (keys %$remotes) { | |
290 | my $u = $remotes->{$repo_id}->{url} or next; | |
291 | next if $u ne $url; | |
292 | $existing = $repo_id; | |
293 | last; | |
294 | } | |
295 | $existing; | |
296 | } | |
297 | ||
298 | sub init_remote_config { | |
299 | my ($self, $url, $no_write) = @_; | |
9c27a57b | 300 | $url = canonicalize_url($url); |
29499c0b MS |
301 | my $r = read_all_remotes(); |
302 | my $existing = find_existing_remote($url, $r); | |
303 | if ($existing) { | |
304 | unless ($no_write) { | |
305 | print STDERR "Using existing ", | |
306 | "[svn-remote \"$existing\"]\n"; | |
307 | } | |
308 | $self->{repo_id} = $existing; | |
309 | } elsif ($_minimize_url) { | |
310 | my $min_url = Git::SVN::Ra->new($url)->minimize_url; | |
311 | $existing = find_existing_remote($min_url, $r); | |
312 | if ($existing) { | |
313 | unless ($no_write) { | |
314 | print STDERR "Using existing ", | |
315 | "[svn-remote \"$existing\"]\n"; | |
316 | } | |
317 | $self->{repo_id} = $existing; | |
318 | } | |
319 | if ($min_url ne $url) { | |
320 | unless ($no_write) { | |
321 | print STDERR "Using higher level of URL: ", | |
322 | "$url => $min_url\n"; | |
323 | } | |
5578ed74 MS |
324 | my $old_path = $self->path; |
325 | $url =~ s!^\Q$min_url\E(/|$)!!; | |
ca475a61 | 326 | $url = join_paths($url, $old_path); |
5578ed74 | 327 | $self->path($url); |
29499c0b MS |
328 | $url = $min_url; |
329 | } | |
330 | } | |
331 | my $orig_url; | |
332 | if (!$existing) { | |
333 | # verify that we aren't overwriting anything: | |
334 | $orig_url = eval { | |
335 | command_oneline('config', '--get', | |
336 | "svn-remote.$self->{repo_id}.url") | |
337 | }; | |
338 | if ($orig_url && ($orig_url ne $url)) { | |
339 | die "svn-remote.$self->{repo_id}.url already set: ", | |
340 | "$orig_url\nwanted to set to: $url\n"; | |
341 | } | |
342 | } | |
343 | my ($xrepo_id, $xpath) = find_ref($self->refname); | |
344 | if (!$no_write && defined $xpath) { | |
345 | die "svn-remote.$xrepo_id.fetch already set to track ", | |
346 | "$xpath:", $self->refname, "\n"; | |
347 | } | |
348 | unless ($no_write) { | |
349 | command_noisy('config', | |
350 | "svn-remote.$self->{repo_id}.url", $url); | |
5578ed74 MS |
351 | my $path = $self->path; |
352 | $path =~ s{^/}{}; | |
353 | $path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg; | |
354 | $self->path($path); | |
29499c0b MS |
355 | command_noisy('config', '--add', |
356 | "svn-remote.$self->{repo_id}.fetch", | |
5578ed74 | 357 | $self->path.":".$self->refname); |
29499c0b | 358 | } |
06ee19e8 | 359 | $self->url($url); |
29499c0b MS |
360 | } |
361 | ||
362 | sub find_by_url { # repos_root and, path are optional | |
363 | my ($class, $full_url, $repos_root, $path) = @_; | |
364 | ||
705b49cb MS |
365 | $full_url = canonicalize_url($full_url); |
366 | ||
29499c0b MS |
367 | return undef unless defined $full_url; |
368 | remove_username($full_url); | |
369 | remove_username($repos_root) if defined $repos_root; | |
370 | my $remotes = read_all_remotes(); | |
371 | if (defined $full_url && defined $repos_root && !defined $path) { | |
372 | $path = $full_url; | |
373 | $path =~ s#^\Q$repos_root\E(?:/|$)##; | |
374 | } | |
375 | foreach my $repo_id (keys %$remotes) { | |
376 | my $u = $remotes->{$repo_id}->{url} or next; | |
377 | remove_username($u); | |
378 | next if defined $repos_root && $repos_root ne $u; | |
379 | ||
380 | my $fetch = $remotes->{$repo_id}->{fetch} || {}; | |
381 | foreach my $t (qw/branches tags/) { | |
382 | foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) { | |
383 | resolve_local_globs($u, $fetch, $globspec); | |
384 | } | |
385 | } | |
386 | my $p = $path; | |
387 | my $rwr = rewrite_root({repo_id => $repo_id}); | |
388 | my $svm = $remotes->{$repo_id}->{svm} | |
389 | if defined $remotes->{$repo_id}->{svm}; | |
390 | unless (defined $p) { | |
391 | $p = $full_url; | |
392 | my $z = $u; | |
393 | my $prefix = ''; | |
394 | if ($rwr) { | |
395 | $z = $rwr; | |
396 | remove_username($z); | |
397 | } elsif (defined $svm) { | |
398 | $z = $svm->{source}; | |
399 | $prefix = $svm->{replace}; | |
400 | $prefix =~ s#^\Q$u\E(?:/|$)##; | |
401 | $prefix =~ s#/$##; | |
402 | } | |
403 | $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next; | |
404 | } | |
705b49cb MS |
405 | |
406 | # remote fetch paths are not URI escaped. Decode ours | |
407 | # so they match | |
408 | $p = uri_decode($p); | |
409 | ||
29499c0b MS |
410 | foreach my $f (keys %$fetch) { |
411 | next if $f ne $p; | |
412 | return Git::SVN->new($fetch->{$f}, $repo_id, $f); | |
413 | } | |
414 | } | |
415 | undef; | |
416 | } | |
417 | ||
418 | sub init { | |
419 | my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_; | |
420 | my $self = _new($class, $repo_id, $ref_id, $path); | |
421 | if (defined $url) { | |
422 | $self->init_remote_config($url, $no_write); | |
423 | } | |
424 | $self; | |
425 | } | |
426 | ||
427 | sub find_ref { | |
428 | my ($ref_id) = @_; | |
429 | foreach (command(qw/config -l/)) { | |
430 | next unless m!^svn-remote\.(.+)\.fetch= | |
431 | \s*(.*?)\s*:\s*(.+?)\s*$!x; | |
432 | my ($repo_id, $path, $ref) = ($1, $2, $3); | |
433 | if ($ref eq $ref_id) { | |
434 | $path = '' if ($path =~ m#^\./?#); | |
435 | return ($repo_id, $path); | |
436 | } | |
437 | } | |
438 | (undef, undef, undef); | |
439 | } | |
440 | ||
441 | sub new { | |
442 | my ($class, $ref_id, $repo_id, $path) = @_; | |
443 | if (defined $ref_id && !defined $repo_id && !defined $path) { | |
444 | ($repo_id, $path) = find_ref($ref_id); | |
445 | if (!defined $repo_id) { | |
446 | die "Could not find a \"svn-remote.*.fetch\" key ", | |
447 | "in the repository configuration matching: ", | |
448 | "$ref_id\n"; | |
449 | } | |
450 | } | |
451 | my $self = _new($class, $repo_id, $ref_id, $path); | |
5578ed74 | 452 | if (!defined $self->path || !length $self->path) { |
29499c0b MS |
453 | my $fetch = command_oneline('config', '--get', |
454 | "svn-remote.$repo_id.fetch", | |
455 | ":$ref_id\$") or | |
456 | die "Failed to read \"svn-remote.$repo_id.fetch\" ", | |
457 | "\":$ref_id\$\" in config\n"; | |
5578ed74 MS |
458 | my($path) = split(/\s*:\s*/, $fetch); |
459 | $self->path($path); | |
29499c0b | 460 | } |
5578ed74 MS |
461 | { |
462 | my $path = $self->path; | |
5578ed74 MS |
463 | $path =~ s{\A/}{}; |
464 | $path =~ s{/\z}{}; | |
465 | $self->path($path); | |
29499c0b | 466 | } |
06ee19e8 MS |
467 | my $url = command_oneline('config', '--get', |
468 | "svn-remote.$repo_id.url") or | |
29499c0b | 469 | die "Failed to read \"svn-remote.$repo_id.url\" in config\n"; |
06ee19e8 | 470 | $self->url($url); |
29499c0b MS |
471 | $self->{pushurl} = eval { command_oneline('config', '--get', |
472 | "svn-remote.$repo_id.pushurl") }; | |
473 | $self->rebuild; | |
474 | $self; | |
475 | } | |
476 | ||
477 | sub refname { | |
478 | my ($refname) = $_[0]->{ref_id} ; | |
479 | ||
480 | # It cannot end with a slash /, we'll throw up on this because | |
481 | # SVN can't have directories with a slash in their name, either: | |
482 | if ($refname =~ m{/$}) { | |
235e8d59 JL |
483 | die "ref: '$refname' ends with a trailing slash; this is ", |
484 | "not permitted by git or Subversion\n"; | |
29499c0b MS |
485 | } |
486 | ||
487 | # It cannot have ASCII control character space, tilde ~, caret ^, | |
488 | # colon :, question-mark ?, asterisk *, space, or open bracket [ | |
489 | # anywhere. | |
490 | # | |
491 | # Additionally, % must be escaped because it is used for escaping | |
492 | # and we want our escaped refname to be reversible | |
22af6fef | 493 | $refname =~ s{([ \%~\^:\?\*\[\t\\])}{sprintf('%%%02X',ord($1))}eg; |
29499c0b MS |
494 | |
495 | # no slash-separated component can begin with a dot . | |
496 | # /.* becomes /%2E* | |
497 | $refname =~ s{/\.}{/%2E}g; | |
498 | ||
499 | # It cannot have two consecutive dots .. anywhere | |
500 | # .. becomes %2E%2E | |
501 | $refname =~ s{\.\.}{%2E%2E}g; | |
502 | ||
503 | # trailing dots and .lock are not allowed | |
504 | # .$ becomes %2E and .lock becomes %2Elock | |
505 | $refname =~ s{\.(?=$|lock$)}{%2E}; | |
506 | ||
507 | # the sequence @{ is used to access the reflog | |
508 | # @{ becomes %40{ | |
509 | $refname =~ s{\@\{}{%40\{}g; | |
510 | ||
511 | return $refname; | |
512 | } | |
513 | ||
514 | sub desanitize_refname { | |
515 | my ($refname) = @_; | |
516 | $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg; | |
517 | return $refname; | |
518 | } | |
519 | ||
520 | sub svm_uuid { | |
521 | my ($self) = @_; | |
522 | return $self->{svm}->{uuid} if $self->svm; | |
523 | $self->ra; | |
524 | unless ($self->{svm}) { | |
525 | die "SVM UUID not cached, and reading remotely failed\n"; | |
526 | } | |
527 | $self->{svm}->{uuid}; | |
528 | } | |
529 | ||
530 | sub svm { | |
531 | my ($self) = @_; | |
532 | return $self->{svm} if $self->{svm}; | |
533 | my $svm; | |
534 | # see if we have it in our config, first: | |
535 | eval { | |
536 | my $section = "svn-remote.$self->{repo_id}"; | |
537 | $svm = { | |
538 | source => tmp_config('--get', "$section.svm-source"), | |
539 | uuid => tmp_config('--get', "$section.svm-uuid"), | |
540 | replace => tmp_config('--get', "$section.svm-replace"), | |
541 | } | |
542 | }; | |
543 | if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) { | |
544 | $self->{svm} = $svm; | |
545 | } | |
546 | $self->{svm}; | |
547 | } | |
548 | ||
549 | sub _set_svm_vars { | |
550 | my ($self, $ra) = @_; | |
551 | return $ra if $self->svm; | |
552 | ||
553 | my @err = ( "useSvmProps set, but failed to read SVM properties\n", | |
554 | "(svm:source, svm:uuid) ", | |
555 | "from the following URLs:\n" ); | |
556 | sub read_svm_props { | |
557 | my ($self, $ra, $path, $r) = @_; | |
558 | my $props = ($ra->get_dir($path, $r))[2]; | |
559 | my $src = $props->{'svm:source'}; | |
560 | my $uuid = $props->{'svm:uuid'}; | |
561 | return undef if (!$src || !$uuid); | |
562 | ||
563 | chomp($src, $uuid); | |
564 | ||
565 | $uuid =~ m{^[0-9a-f\-]{30,}$}i | |
566 | or die "doesn't look right - svm:uuid is '$uuid'\n"; | |
567 | ||
568 | # the '!' is used to mark the repos_root!/relative/path | |
569 | $src =~ s{/?!/?}{/}; | |
570 | $src =~ s{/+$}{}; # no trailing slashes please | |
571 | # username is of no interest | |
572 | $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1}; | |
573 | ||
d2fd119c | 574 | my $replace = add_path_to_url($ra->url, $path); |
29499c0b MS |
575 | |
576 | my $section = "svn-remote.$self->{repo_id}"; | |
577 | tmp_config("$section.svm-source", $src); | |
578 | tmp_config("$section.svm-replace", $replace); | |
579 | tmp_config("$section.svm-uuid", $uuid); | |
580 | $self->{svm} = { | |
581 | source => $src, | |
582 | uuid => $uuid, | |
583 | replace => $replace | |
584 | }; | |
585 | } | |
586 | ||
587 | my $r = $ra->get_latest_revnum; | |
5578ed74 | 588 | my $path = $self->path; |
29499c0b MS |
589 | my %tried; |
590 | while (length $path) { | |
d2fd119c | 591 | my $try = add_path_to_url($self->url, $path); |
06ee19e8 | 592 | unless ($tried{$try}) { |
29499c0b | 593 | return $ra if $self->read_svm_props($ra, $path, $r); |
06ee19e8 | 594 | $tried{$try} = 1; |
29499c0b MS |
595 | } |
596 | $path =~ s#/?[^/]+$##; | |
597 | } | |
598 | die "Path: '$path' should be ''\n" if $path ne ''; | |
599 | return $ra if $self->read_svm_props($ra, $path, $r); | |
d2fd119c | 600 | $tried{ add_path_to_url($self->url, $path) } = 1; |
29499c0b | 601 | |
06ee19e8 | 602 | if ($ra->{repos_root} eq $self->url) { |
29499c0b MS |
603 | die @err, (map { " $_\n" } keys %tried), "\n"; |
604 | } | |
605 | ||
606 | # nope, make sure we're connected to the repository root: | |
607 | my $ok; | |
608 | my @tried_b; | |
609 | $path = $ra->{svn_path}; | |
610 | $ra = Git::SVN::Ra->new($ra->{repos_root}); | |
611 | while (length $path) { | |
d2fd119c | 612 | my $try = add_path_to_url($ra->url, $path); |
b1ea6c38 | 613 | unless ($tried{$try}) { |
29499c0b MS |
614 | $ok = $self->read_svm_props($ra, $path, $r); |
615 | last if $ok; | |
b1ea6c38 | 616 | $tried{$try} = 1; |
29499c0b MS |
617 | } |
618 | $path =~ s#/?[^/]+$##; | |
619 | } | |
620 | die "Path: '$path' should be ''\n" if $path ne ''; | |
621 | $ok ||= $self->read_svm_props($ra, $path, $r); | |
d2fd119c | 622 | $tried{ add_path_to_url($ra->url, $path) } = 1; |
29499c0b MS |
623 | if (!$ok) { |
624 | die @err, (map { " $_\n" } keys %tried), "\n"; | |
625 | } | |
06ee19e8 | 626 | Git::SVN::Ra->new($self->url); |
29499c0b MS |
627 | } |
628 | ||
629 | sub svnsync { | |
630 | my ($self) = @_; | |
631 | return $self->{svnsync} if $self->{svnsync}; | |
632 | ||
633 | if ($self->no_metadata) { | |
634 | die "Can't have both 'noMetadata' and ", | |
635 | "'useSvnsyncProps' options set!\n"; | |
636 | } | |
637 | if ($self->rewrite_root) { | |
638 | die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ", | |
639 | "options set!\n"; | |
640 | } | |
641 | if ($self->rewrite_uuid) { | |
642 | die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ", | |
643 | "options set!\n"; | |
644 | } | |
645 | ||
646 | my $svnsync; | |
647 | # see if we have it in our config, first: | |
648 | eval { | |
649 | my $section = "svn-remote.$self->{repo_id}"; | |
650 | ||
651 | my $url = tmp_config('--get', "$section.svnsync-url"); | |
652 | ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or | |
653 | die "doesn't look right - svn:sync-from-url is '$url'\n"; | |
654 | ||
655 | my $uuid = tmp_config('--get', "$section.svnsync-uuid"); | |
656 | ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or | |
657 | die "doesn't look right - svn:sync-from-uuid is '$uuid'\n"; | |
658 | ||
659 | $svnsync = { url => $url, uuid => $uuid } | |
660 | }; | |
661 | if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) { | |
662 | return $self->{svnsync} = $svnsync; | |
663 | } | |
664 | ||
665 | my $err = "useSvnsyncProps set, but failed to read " . | |
666 | "svnsync property: svn:sync-from-"; | |
667 | my $rp = $self->ra->rev_proplist(0); | |
668 | ||
669 | my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n"; | |
670 | ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or | |
671 | die "doesn't look right - svn:sync-from-url is '$url'\n"; | |
672 | ||
673 | my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n"; | |
674 | ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or | |
675 | die "doesn't look right - svn:sync-from-uuid is '$uuid'\n"; | |
676 | ||
677 | my $section = "svn-remote.$self->{repo_id}"; | |
678 | tmp_config('--add', "$section.svnsync-uuid", $uuid); | |
679 | tmp_config('--add', "$section.svnsync-url", $url); | |
680 | return $self->{svnsync} = { url => $url, uuid => $uuid }; | |
681 | } | |
682 | ||
683 | # this allows us to memoize our SVN::Ra UUID locally and avoid a | |
684 | # remote lookup (useful for 'git svn log'). | |
685 | sub ra_uuid { | |
686 | my ($self) = @_; | |
687 | unless ($self->{ra_uuid}) { | |
688 | my $key = "svn-remote.$self->{repo_id}.uuid"; | |
689 | my $uuid = eval { tmp_config('--get', $key) }; | |
690 | if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) { | |
691 | $self->{ra_uuid} = $uuid; | |
692 | } else { | |
06ee19e8 | 693 | die "ra_uuid called without URL\n" unless $self->url; |
29499c0b MS |
694 | $self->{ra_uuid} = $self->ra->get_uuid; |
695 | tmp_config('--add', $key, $self->{ra_uuid}); | |
696 | } | |
697 | } | |
698 | $self->{ra_uuid}; | |
699 | } | |
700 | ||
701 | sub _set_repos_root { | |
702 | my ($self, $repos_root) = @_; | |
703 | my $k = "svn-remote.$self->{repo_id}.reposRoot"; | |
704 | $repos_root ||= $self->ra->{repos_root}; | |
705 | tmp_config($k, $repos_root); | |
706 | $repos_root; | |
707 | } | |
708 | ||
709 | sub repos_root { | |
710 | my ($self) = @_; | |
711 | my $k = "svn-remote.$self->{repo_id}.reposRoot"; | |
712 | eval { tmp_config('--get', $k) } || $self->_set_repos_root; | |
713 | } | |
714 | ||
715 | sub ra { | |
716 | my ($self) = shift; | |
06ee19e8 | 717 | my $ra = Git::SVN::Ra->new($self->url); |
29499c0b MS |
718 | $self->_set_repos_root($ra->{repos_root}); |
719 | if ($self->use_svm_props && !$self->{svm}) { | |
720 | if ($self->no_metadata) { | |
721 | die "Can't have both 'noMetadata' and ", | |
722 | "'useSvmProps' options set!\n"; | |
723 | } elsif ($self->use_svnsync_props) { | |
724 | die "Can't have both 'useSvnsyncProps' and ", | |
725 | "'useSvmProps' options set!\n"; | |
726 | } | |
727 | $ra = $self->_set_svm_vars($ra); | |
728 | $self->{-want_revprops} = 1; | |
729 | } | |
730 | $ra; | |
731 | } | |
732 | ||
733 | # prop_walk(PATH, REV, SUB) | |
734 | # ------------------------- | |
735 | # Recursively traverse PATH at revision REV and invoke SUB for each | |
736 | # directory that contains a SVN property. SUB will be invoked as | |
737 | # follows: &SUB(gs, path, props); where `gs' is this instance of | |
738 | # Git::SVN, `path' the path to the directory where the properties | |
739 | # `props' were found. The `path' will be relative to point of checkout, | |
740 | # that is, if url://repo/trunk is the current Git branch, and that | |
741 | # directory contains a sub-directory `d', SUB will be invoked with `/d/' | |
742 | # as `path' (note the trailing `/'). | |
743 | sub prop_walk { | |
744 | my ($self, $path, $rev, $sub) = @_; | |
745 | ||
746 | $path =~ s#^/##; | |
747 | my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev); | |
748 | $path =~ s#^/*#/#g; | |
749 | my $p = $path; | |
750 | # Strip the irrelevant part of the path. | |
5578ed74 | 751 | $p =~ s#^/+\Q@{[$self->path]}\E(/|$)#/#; |
29499c0b MS |
752 | # Ensure the path is terminated by a `/'. |
753 | $p =~ s#/*$#/#; | |
754 | ||
755 | # The properties contain all the internal SVN stuff nobody | |
756 | # (usually) cares about. | |
757 | my $interesting_props = 0; | |
758 | foreach (keys %{$props}) { | |
759 | # If it doesn't start with `svn:', it must be a | |
760 | # user-defined property. | |
761 | ++$interesting_props and next if $_ !~ /^svn:/; | |
762 | # FIXME: Fragile, if SVN adds new public properties, | |
763 | # this needs to be updated. | |
764 | ++$interesting_props if /^svn:(?:ignore|keywords|executable | |
765 | |eol-style|mime-type | |
766 | |externals|needs-lock)$/x; | |
767 | } | |
768 | &$sub($self, $p, $props) if $interesting_props; | |
769 | ||
770 | foreach (sort keys %$dirent) { | |
771 | next if $dirent->{$_}->{kind} != $SVN::Node::dir; | |
5578ed74 | 772 | $self->prop_walk($self->path . $p . $_, $rev, $sub); |
29499c0b MS |
773 | } |
774 | } | |
775 | ||
776 | sub last_rev { ($_[0]->last_rev_commit)[0] } | |
777 | sub last_commit { ($_[0]->last_rev_commit)[1] } | |
778 | ||
779 | # returns the newest SVN revision number and newest commit SHA1 | |
780 | sub last_rev_commit { | |
781 | my ($self) = @_; | |
782 | if (defined $self->{last_rev} && defined $self->{last_commit}) { | |
783 | return ($self->{last_rev}, $self->{last_commit}); | |
784 | } | |
785 | my $c = ::verify_ref($self->refname.'^0'); | |
786 | if ($c && !$self->use_svm_props && !$self->no_metadata) { | |
787 | my $rev = (::cmt_metadata($c))[1]; | |
788 | if (defined $rev) { | |
789 | ($self->{last_rev}, $self->{last_commit}) = ($rev, $c); | |
790 | return ($rev, $c); | |
791 | } | |
792 | } | |
793 | my $map_path = $self->map_path; | |
794 | unless (-e $map_path) { | |
795 | ($self->{last_rev}, $self->{last_commit}) = (undef, undef); | |
796 | return (undef, undef); | |
797 | } | |
798 | my ($rev, $commit) = $self->rev_map_max(1); | |
799 | ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit); | |
800 | return ($rev, $commit); | |
801 | } | |
802 | ||
803 | sub get_fetch_range { | |
804 | my ($self, $min, $max) = @_; | |
805 | $max ||= $self->ra->get_latest_revnum; | |
806 | $min ||= $self->rev_map_max; | |
807 | (++$min, $max); | |
808 | } | |
809 | ||
112423eb EW |
810 | sub svn_dir { |
811 | command_oneline(qw(rev-parse --git-path svn)); | |
812 | } | |
813 | ||
29499c0b MS |
814 | sub tmp_config { |
815 | my (@args) = @_; | |
112423eb EW |
816 | my $svn_dir = svn_dir(); |
817 | my $old_def_config = "$svn_dir/config"; | |
818 | my $config = "$svn_dir/.metadata"; | |
29499c0b MS |
819 | if (! -f $config && -f $old_def_config) { |
820 | rename $old_def_config, $config or | |
821 | die "Failed rename $old_def_config => $config: $!\n"; | |
822 | } | |
823 | my $old_config = $ENV{GIT_CONFIG}; | |
824 | $ENV{GIT_CONFIG} = $config; | |
825 | $@ = undef; | |
826 | my @ret = eval { | |
827 | unless (-f $config) { | |
828 | mkfile($config); | |
829 | open my $fh, '>', $config or | |
830 | die "Can't open $config: $!\n"; | |
831 | print $fh "; This file is used internally by ", | |
832 | "git-svn\n" or die | |
833 | "Couldn't write to $config: $!\n"; | |
834 | print $fh "; You should not have to edit it\n" or | |
835 | die "Couldn't write to $config: $!\n"; | |
836 | close $fh or die "Couldn't close $config: $!\n"; | |
837 | } | |
838 | command('config', @args); | |
839 | }; | |
840 | my $err = $@; | |
841 | if (defined $old_config) { | |
842 | $ENV{GIT_CONFIG} = $old_config; | |
843 | } else { | |
844 | delete $ENV{GIT_CONFIG}; | |
845 | } | |
846 | die $err if $err; | |
847 | wantarray ? @ret : $ret[0]; | |
848 | } | |
849 | ||
850 | sub tmp_index_do { | |
851 | my ($self, $sub) = @_; | |
852 | my $old_index = $ENV{GIT_INDEX_FILE}; | |
853 | $ENV{GIT_INDEX_FILE} = $self->{index}; | |
854 | $@ = undef; | |
855 | my @ret = eval { | |
856 | my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#); | |
857 | mkpath([$dir]) unless -d $dir; | |
858 | &$sub; | |
859 | }; | |
860 | my $err = $@; | |
861 | if (defined $old_index) { | |
862 | $ENV{GIT_INDEX_FILE} = $old_index; | |
863 | } else { | |
864 | delete $ENV{GIT_INDEX_FILE}; | |
865 | } | |
866 | die $err if $err; | |
867 | wantarray ? @ret : $ret[0]; | |
868 | } | |
869 | ||
870 | sub assert_index_clean { | |
871 | my ($self, $treeish) = @_; | |
872 | ||
873 | $self->tmp_index_do(sub { | |
874 | command_noisy('read-tree', $treeish) unless -e $self->{index}; | |
875 | my $x = command_oneline('write-tree'); | |
876 | my ($y) = (command(qw/cat-file commit/, $treeish) =~ | |
877 | /^tree ($::sha1)/mo); | |
878 | return if $y eq $x; | |
879 | ||
880 | warn "Index mismatch: $y != $x\nrereading $treeish\n"; | |
881 | unlink $self->{index} or die "unlink $self->{index}: $!\n"; | |
882 | command_noisy('read-tree', $treeish); | |
883 | $x = command_oneline('write-tree'); | |
884 | if ($y ne $x) { | |
885 | fatal "trees ($treeish) $y != $x\n", | |
886 | "Something is seriously wrong..."; | |
887 | } | |
888 | }); | |
889 | } | |
890 | ||
891 | sub get_commit_parents { | |
892 | my ($self, $log_entry) = @_; | |
893 | my (%seen, @ret, @tmp); | |
894 | # legacy support for 'set-tree'; this is only used by set_tree_cb: | |
895 | if (my $ip = $self->{inject_parents}) { | |
896 | if (my $commit = delete $ip->{$log_entry->{revision}}) { | |
897 | push @tmp, $commit; | |
898 | } | |
899 | } | |
900 | if (my $cur = ::verify_ref($self->refname.'^0')) { | |
901 | push @tmp, $cur; | |
902 | } | |
903 | if (my $ipd = $self->{inject_parents_dcommit}) { | |
904 | if (my $commit = delete $ipd->{$log_entry->{revision}}) { | |
905 | push @tmp, @$commit; | |
906 | } | |
907 | } | |
908 | push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp); | |
909 | while (my $p = shift @tmp) { | |
910 | next if $seen{$p}; | |
911 | $seen{$p} = 1; | |
912 | push @ret, $p; | |
913 | } | |
914 | @ret; | |
915 | } | |
916 | ||
917 | sub rewrite_root { | |
918 | my ($self) = @_; | |
919 | return $self->{-rewrite_root} if exists $self->{-rewrite_root}; | |
920 | my $k = "svn-remote.$self->{repo_id}.rewriteRoot"; | |
921 | my $rwr = eval { command_oneline(qw/config --get/, $k) }; | |
922 | if ($rwr) { | |
923 | $rwr =~ s#/+$##; | |
924 | if ($rwr !~ m#^[a-z\+]+://#) { | |
925 | die "$rwr is not a valid URL (key: $k)\n"; | |
926 | } | |
927 | } | |
928 | $self->{-rewrite_root} = $rwr; | |
929 | } | |
930 | ||
931 | sub rewrite_uuid { | |
932 | my ($self) = @_; | |
933 | return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid}; | |
934 | my $k = "svn-remote.$self->{repo_id}.rewriteUUID"; | |
935 | my $rwid = eval { command_oneline(qw/config --get/, $k) }; | |
936 | if ($rwid) { | |
937 | $rwid =~ s#/+$##; | |
938 | if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) { | |
939 | die "$rwid is not a valid UUID (key: $k)\n"; | |
940 | } | |
941 | } | |
942 | $self->{-rewrite_uuid} = $rwid; | |
943 | } | |
944 | ||
945 | sub metadata_url { | |
946 | my ($self) = @_; | |
d2fd119c | 947 | my $url = $self->rewrite_root || $self->url; |
705b49cb | 948 | return canonicalize_url( add_path_to_url( $url, $self->path ) ); |
29499c0b MS |
949 | } |
950 | ||
951 | sub full_url { | |
952 | my ($self) = @_; | |
705b49cb | 953 | return canonicalize_url( add_path_to_url( $self->url, $self->path ) ); |
29499c0b MS |
954 | } |
955 | ||
956 | sub full_pushurl { | |
957 | my ($self) = @_; | |
958 | if ($self->{pushurl}) { | |
705b49cb | 959 | return canonicalize_url( add_path_to_url( $self->{pushurl}, $self->path ) ); |
29499c0b MS |
960 | } else { |
961 | return $self->full_url; | |
962 | } | |
963 | } | |
964 | ||
965 | sub set_commit_header_env { | |
966 | my ($log_entry) = @_; | |
967 | my %env; | |
968 | foreach my $ned (qw/NAME EMAIL DATE/) { | |
969 | foreach my $ac (qw/AUTHOR COMMITTER/) { | |
970 | $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"}; | |
971 | } | |
972 | } | |
973 | ||
974 | $ENV{GIT_AUTHOR_NAME} = $log_entry->{name}; | |
975 | $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email}; | |
976 | $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date}; | |
977 | ||
978 | $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name}) | |
979 | ? $log_entry->{commit_name} | |
980 | : $log_entry->{name}; | |
981 | $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email}) | |
982 | ? $log_entry->{commit_email} | |
983 | : $log_entry->{email}; | |
984 | \%env; | |
985 | } | |
986 | ||
987 | sub restore_commit_header_env { | |
988 | my ($env) = @_; | |
989 | foreach my $ned (qw/NAME EMAIL DATE/) { | |
990 | foreach my $ac (qw/AUTHOR COMMITTER/) { | |
991 | my $k = "GIT_${ac}_${ned}"; | |
992 | if (defined $env->{$k}) { | |
993 | $ENV{$k} = $env->{$k}; | |
994 | } else { | |
995 | delete $ENV{$k}; | |
996 | } | |
997 | } | |
998 | } | |
999 | } | |
1000 | ||
1001 | sub gc { | |
1002 | command_noisy('gc', '--auto'); | |
1003 | }; | |
1004 | ||
1005 | sub do_git_commit { | |
1006 | my ($self, $log_entry) = @_; | |
1007 | my $lr = $self->last_rev; | |
1008 | if (defined $lr && $lr >= $log_entry->{revision}) { | |
1009 | die "Last fetched revision of ", $self->refname, | |
1010 | " was r$lr, but we are about to fetch: ", | |
1011 | "r$log_entry->{revision}!\n"; | |
1012 | } | |
1013 | if (my $c = $self->rev_map_get($log_entry->{revision})) { | |
1014 | croak "$log_entry->{revision} = $c already exists! ", | |
1015 | "Why are we refetching it?\n"; | |
1016 | } | |
1017 | my $old_env = set_commit_header_env($log_entry); | |
1018 | my $tree = $log_entry->{tree}; | |
1019 | if (!defined $tree) { | |
1020 | $tree = $self->tmp_index_do(sub { | |
1021 | command_oneline('write-tree') }); | |
1022 | } | |
1023 | die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o; | |
1024 | ||
1025 | my @exec = ('git', 'commit-tree', $tree); | |
1026 | foreach ($self->get_commit_parents($log_entry)) { | |
1027 | push @exec, '-p', $_; | |
1028 | } | |
1029 | defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec)) | |
1030 | or croak $!; | |
1031 | binmode $msg_fh; | |
1032 | ||
1033 | # we always get UTF-8 from SVN, but we may want our commits in | |
1034 | # a different encoding. | |
1035 | if (my $enc = Git::config('i18n.commitencoding')) { | |
1036 | require Encode; | |
1037 | Encode::from_to($log_entry->{log}, 'UTF-8', $enc); | |
1038 | } | |
1039 | print $msg_fh $log_entry->{log} or croak $!; | |
1040 | restore_commit_header_env($old_env); | |
1041 | unless ($self->no_metadata) { | |
1042 | print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n" | |
1043 | or croak $!; | |
1044 | } | |
1045 | $msg_fh->flush == 0 or croak $!; | |
1046 | close $msg_fh or croak $!; | |
1047 | chomp(my $commit = do { local $/; <$out_fh> }); | |
1048 | close $out_fh or croak $!; | |
1049 | waitpid $pid, 0; | |
1050 | croak $? if $?; | |
1051 | if ($commit !~ /^$::sha1$/o) { | |
1052 | die "Failed to commit, invalid sha1: $commit\n"; | |
1053 | } | |
1054 | ||
1055 | $self->rev_map_set($log_entry->{revision}, $commit, 1); | |
1056 | ||
1057 | $self->{last_rev} = $log_entry->{revision}; | |
1058 | $self->{last_commit} = $commit; | |
1059 | print "r$log_entry->{revision}" unless $::_q > 1; | |
1060 | if (defined $log_entry->{svm_revision}) { | |
1061 | print " (\@$log_entry->{svm_revision})" unless $::_q > 1; | |
1062 | $self->rev_map_set($log_entry->{svm_revision}, $commit, | |
1063 | 0, $self->svm_uuid); | |
1064 | } | |
1065 | print " = $commit ($self->{ref_id})\n" unless $::_q > 1; | |
1066 | if (--$_gc_nr == 0) { | |
1067 | $_gc_nr = $_gc_period; | |
1068 | gc(); | |
1069 | } | |
1070 | return $commit; | |
1071 | } | |
1072 | ||
1073 | sub match_paths { | |
1074 | my ($self, $paths, $r) = @_; | |
5578ed74 MS |
1075 | return 1 if $self->path eq ''; |
1076 | if (my $path = $paths->{"/".$self->path}) { | |
29499c0b MS |
1077 | return ($path->{action} eq 'D') ? 0 : 1; |
1078 | } | |
5578ed74 | 1079 | $self->{path_regex} ||= qr{^/\Q@{[$self->path]}\E/}; |
29499c0b MS |
1080 | if (grep /$self->{path_regex}/, keys %$paths) { |
1081 | return 1; | |
1082 | } | |
1083 | my $c = ''; | |
5578ed74 | 1084 | foreach (split m#/#, $self->path) { |
29499c0b MS |
1085 | $c .= "/$_"; |
1086 | next unless ($paths->{$c} && | |
1087 | ($paths->{$c}->{action} =~ /^[AR]$/)); | |
5578ed74 | 1088 | if ($self->ra->check_path($self->path, $r) == |
29499c0b MS |
1089 | $SVN::Node::dir) { |
1090 | return 1; | |
1091 | } | |
1092 | } | |
1093 | return 0; | |
1094 | } | |
1095 | ||
1096 | sub find_parent_branch { | |
1097 | my ($self, $paths, $rev) = @_; | |
1098 | return undef unless $self->follow_parent; | |
1099 | unless (defined $paths) { | |
1100 | my $err_handler = $SVN::Error::handler; | |
1101 | $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs; | |
5578ed74 | 1102 | $self->ra->get_log([$self->path], $rev, $rev, 0, 1, 1, |
29499c0b MS |
1103 | sub { $paths = $_[0] }); |
1104 | $SVN::Error::handler = $err_handler; | |
1105 | } | |
1106 | return undef unless defined $paths; | |
1107 | ||
1108 | # look for a parent from another branch: | |
5578ed74 | 1109 | my @b_path_components = split m#/#, $self->path; |
29499c0b MS |
1110 | my @a_path_components; |
1111 | my $i; | |
1112 | while (@b_path_components) { | |
1113 | $i = $paths->{'/'.join('/', @b_path_components)}; | |
1114 | last if $i && defined $i->{copyfrom_path}; | |
1115 | unshift(@a_path_components, pop(@b_path_components)); | |
1116 | } | |
1117 | return undef unless defined $i && defined $i->{copyfrom_path}; | |
1118 | my $branch_from = $i->{copyfrom_path}; | |
1119 | if (@a_path_components) { | |
1120 | print STDERR "branch_from: $branch_from => "; | |
1121 | $branch_from .= '/'.join('/', @a_path_components); | |
1122 | print STDERR $branch_from, "\n"; | |
1123 | } | |
1124 | my $r = $i->{copyfrom_rev}; | |
1125 | my $repos_root = $self->ra->{repos_root}; | |
b1ea6c38 | 1126 | my $url = $self->ra->url; |
705b49cb | 1127 | my $new_url = canonicalize_url( add_path_to_url( $url, $branch_from ) ); |
29499c0b MS |
1128 | print STDERR "Found possible branch point: ", |
1129 | "$new_url => ", $self->full_url, ", $r\n" | |
1130 | unless $::_q > 1; | |
1131 | $branch_from =~ s#^/##; | |
1132 | my $gs = $self->other_gs($new_url, $url, | |
1133 | $branch_from, $r, $self->{ref_id}); | |
1134 | my ($r0, $parent) = $gs->find_rev_before($r, 1); | |
1135 | { | |
1136 | my ($base, $head); | |
1137 | if (!defined $r0 || !defined $parent) { | |
1138 | ($base, $head) = parse_revision_argument(0, $r); | |
1139 | } else { | |
1140 | if ($r0 < $r) { | |
6a8d999e | 1141 | $gs->ra->get_log([$gs->path], $r0 + 1, $r, 1, |
29499c0b MS |
1142 | 0, 1, sub { $base = $_[1] - 1 }); |
1143 | } | |
1144 | } | |
1145 | if (defined $base && $base <= $r) { | |
1146 | $gs->fetch($base, $r); | |
1147 | } | |
1148 | ($r0, $parent) = $gs->find_rev_before($r, 1); | |
1149 | } | |
1150 | if (defined $r0 && defined $parent) { | |
1151 | print STDERR "Found branch parent: ($self->{ref_id}) $parent\n" | |
1152 | unless $::_q > 1; | |
1153 | my $ed; | |
1154 | if ($self->ra->can_do_switch) { | |
1155 | $self->assert_index_clean($parent); | |
1156 | print STDERR "Following parent with do_switch\n" | |
1157 | unless $::_q > 1; | |
1158 | # do_switch works with svn/trunk >= r22312, but that | |
1159 | # is not included with SVN 1.4.3 (the latest version | |
1160 | # at the moment), so we can't rely on it | |
1161 | $self->{last_rev} = $r0; | |
1162 | $self->{last_commit} = $parent; | |
6a8d999e | 1163 | $ed = Git::SVN::Fetcher->new($self, $gs->path); |
29499c0b MS |
1164 | $gs->ra->gs_do_switch($r0, $rev, $gs, |
1165 | $self->full_url, $ed) | |
1166 | or die "SVN connection failed somewhere...\n"; | |
1167 | } elsif ($self->ra->trees_match($new_url, $r0, | |
1168 | $self->full_url, $rev)) { | |
1169 | print STDERR "Trees match:\n", | |
1170 | " $new_url\@$r0\n", | |
1171 | " ${\$self->full_url}\@$rev\n", | |
1172 | "Following parent with no changes\n" | |
1173 | unless $::_q > 1; | |
1174 | $self->tmp_index_do(sub { | |
1175 | command_noisy('read-tree', $parent); | |
1176 | }); | |
1177 | $self->{last_commit} = $parent; | |
1178 | } else { | |
1179 | print STDERR "Following parent with do_update\n" | |
1180 | unless $::_q > 1; | |
1181 | $ed = Git::SVN::Fetcher->new($self); | |
1182 | $self->ra->gs_do_update($rev, $rev, $self, $ed) | |
1183 | or die "SVN connection failed somewhere...\n"; | |
1184 | } | |
1185 | print STDERR "Successfully followed parent\n" unless $::_q > 1; | |
abfef3bb | 1186 | return $self->make_log_entry($rev, [$parent], $ed, $r0, $branch_from); |
29499c0b MS |
1187 | } |
1188 | return undef; | |
1189 | } | |
1190 | ||
1191 | sub do_fetch { | |
1192 | my ($self, $paths, $rev) = @_; | |
1193 | my $ed; | |
1194 | my ($last_rev, @parents); | |
1195 | if (my $lc = $self->last_commit) { | |
1196 | # we can have a branch that was deleted, then re-added | |
1197 | # under the same name but copied from another path, in | |
1198 | # which case we'll have multiple parents (we don't | |
01689909 | 1199 | # want to break the original ref or lose copypath info): |
29499c0b MS |
1200 | if (my $log_entry = $self->find_parent_branch($paths, $rev)) { |
1201 | push @{$log_entry->{parents}}, $lc; | |
1202 | return $log_entry; | |
1203 | } | |
1204 | $ed = Git::SVN::Fetcher->new($self); | |
1205 | $last_rev = $self->{last_rev}; | |
1206 | $ed->{c} = $lc; | |
1207 | @parents = ($lc); | |
1208 | } else { | |
1209 | $last_rev = $rev; | |
1210 | if (my $log_entry = $self->find_parent_branch($paths, $rev)) { | |
1211 | return $log_entry; | |
1212 | } | |
1213 | $ed = Git::SVN::Fetcher->new($self); | |
1214 | } | |
1215 | unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) { | |
1216 | die "SVN connection failed somewhere...\n"; | |
1217 | } | |
9ee13a93 | 1218 | $self->make_log_entry($rev, \@parents, $ed, $last_rev, $self->path); |
29499c0b MS |
1219 | } |
1220 | ||
1221 | sub mkemptydirs { | |
1222 | my ($self, $r) = @_; | |
1223 | ||
82625748 DG |
1224 | # add/remove/collect a paths table |
1225 | # | |
1226 | # Paths are split into a tree of nodes, stored as a hash of hashes. | |
1227 | # | |
1228 | # Each node contains a 'path' entry for the path (if any) associated | |
1229 | # with that node and a 'children' entry for any nodes under that | |
1230 | # location. | |
1231 | # | |
1232 | # Removing a path requires a hash lookup for each component then | |
1233 | # dropping that node (and anything under it), which is substantially | |
1234 | # faster than a grep slice into a single hash of paths for large | |
1235 | # numbers of paths. | |
1236 | # | |
1237 | # For a large (200K) number of empty_dir directives this reduces | |
1238 | # scanning time to 3 seconds vs 10 minutes for grep+delete on a single | |
1239 | # hash of paths. | |
1240 | sub add_path { | |
1241 | my ($paths_table, $path) = @_; | |
1242 | my $node_ref; | |
1243 | ||
1244 | foreach my $x (split('/', $path)) { | |
1245 | if (!exists($paths_table->{$x})) { | |
1246 | $paths_table->{$x} = { children => {} }; | |
1247 | } | |
1248 | ||
1249 | $node_ref = $paths_table->{$x}; | |
1250 | $paths_table = $paths_table->{$x}->{children}; | |
1251 | } | |
1252 | ||
1253 | $node_ref->{path} = $path; | |
1254 | } | |
1255 | ||
1256 | sub remove_path { | |
1257 | my ($paths_table, $path) = @_; | |
1258 | my $nodes_ref; | |
1259 | my $node_name; | |
1260 | ||
1261 | foreach my $x (split('/', $path)) { | |
1262 | if (!exists($paths_table->{$x})) { | |
1263 | return; | |
1264 | } | |
1265 | ||
1266 | $nodes_ref = $paths_table; | |
1267 | $node_name = $x; | |
1268 | ||
1269 | $paths_table = $paths_table->{$x}->{children}; | |
1270 | } | |
1271 | ||
1272 | delete($nodes_ref->{$node_name}); | |
1273 | } | |
1274 | ||
1275 | sub collect_paths { | |
1276 | my ($paths_table, $paths_ref) = @_; | |
1277 | ||
1278 | foreach my $v (values %$paths_table) { | |
1279 | my $p = $v->{path}; | |
1280 | my $c = $v->{children}; | |
1281 | ||
1282 | collect_paths($c, $paths_ref); | |
1283 | ||
1284 | if (defined($p)) { | |
1285 | push(@$paths_ref, $p); | |
1286 | } | |
1287 | } | |
1288 | } | |
1289 | ||
29499c0b | 1290 | sub scan { |
82625748 | 1291 | my ($r, $paths_table, $line) = @_; |
29499c0b MS |
1292 | if (defined $r && $line =~ /^r(\d+)$/) { |
1293 | return 0 if $1 > $r; | |
1294 | } elsif ($line =~ /^ \+empty_dir: (.+)$/) { | |
82625748 | 1295 | add_path($paths_table, $1); |
29499c0b | 1296 | } elsif ($line =~ /^ \-empty_dir: (.+)$/) { |
82625748 | 1297 | remove_path($paths_table, $1); |
29499c0b MS |
1298 | } |
1299 | 1; # continue | |
1300 | }; | |
1301 | ||
82625748 DG |
1302 | my @empty_dirs; |
1303 | my %paths_table; | |
1304 | ||
29499c0b MS |
1305 | my $gz_file = "$self->{dir}/unhandled.log.gz"; |
1306 | if (-f $gz_file) { | |
1307 | if (!can_compress()) { | |
1308 | warn "Compress::Zlib could not be found; ", | |
1309 | "empty directories in $gz_file will not be read\n"; | |
1310 | } else { | |
1311 | my $gz = Compress::Zlib::gzopen($gz_file, "rb") or | |
1312 | die "Unable to open $gz_file: $!\n"; | |
1313 | my $line; | |
1314 | while ($gz->gzreadline($line) > 0) { | |
82625748 | 1315 | scan($r, \%paths_table, $line) or last; |
29499c0b MS |
1316 | } |
1317 | $gz->gzclose; | |
1318 | } | |
1319 | } | |
1320 | ||
1321 | if (open my $fh, '<', "$self->{dir}/unhandled.log") { | |
1322 | binmode $fh or croak "binmode: $!"; | |
1323 | while (<$fh>) { | |
82625748 | 1324 | scan($r, \%paths_table, $_) or last; |
29499c0b MS |
1325 | } |
1326 | close $fh; | |
1327 | } | |
1328 | ||
82625748 | 1329 | collect_paths(\%paths_table, \@empty_dirs); |
5578ed74 | 1330 | my $strip = qr/\A\Q@{[$self->path]}\E(?:\/|$)/; |
82625748 | 1331 | foreach my $d (sort @empty_dirs) { |
29499c0b MS |
1332 | $d = uri_decode($d); |
1333 | $d =~ s/$strip//; | |
1334 | next unless length($d); | |
1335 | next if -d $d; | |
1336 | if (-e $d) { | |
1337 | warn "$d exists but is not a directory\n"; | |
1338 | } else { | |
1339 | print "creating empty directory: $d\n"; | |
1340 | mkpath([$d]); | |
1341 | } | |
1342 | } | |
1343 | } | |
1344 | ||
1345 | sub get_untracked { | |
1346 | my ($self, $ed) = @_; | |
1347 | my @out; | |
1348 | my $h = $ed->{empty}; | |
1349 | foreach (sort keys %$h) { | |
1350 | my $act = $h->{$_} ? '+empty_dir' : '-empty_dir'; | |
1351 | push @out, " $act: " . uri_encode($_); | |
1352 | warn "W: $act: $_\n"; | |
1353 | } | |
1354 | foreach my $t (qw/dir_prop file_prop/) { | |
1355 | $h = $ed->{$t} or next; | |
1356 | foreach my $path (sort keys %$h) { | |
1357 | my $ppath = $path eq '' ? '.' : $path; | |
1358 | foreach my $prop (sort keys %{$h->{$path}}) { | |
1359 | next if $SKIP_PROP{$prop}; | |
1360 | my $v = $h->{$path}->{$prop}; | |
1361 | my $t_ppath_prop = "$t: " . | |
1362 | uri_encode($ppath) . ' ' . | |
1363 | uri_encode($prop); | |
1364 | if (defined $v) { | |
1365 | push @out, " +$t_ppath_prop " . | |
1366 | uri_encode($v); | |
1367 | } else { | |
1368 | push @out, " -$t_ppath_prop"; | |
1369 | } | |
1370 | } | |
1371 | } | |
1372 | } | |
1373 | foreach my $t (qw/absent_file absent_directory/) { | |
1374 | $h = $ed->{$t} or next; | |
1375 | foreach my $parent (sort keys %$h) { | |
1376 | foreach my $path (sort @{$h->{$parent}}) { | |
1377 | push @out, " $t: " . | |
1378 | uri_encode("$parent/$path"); | |
1379 | warn "W: $t: $parent/$path ", | |
1380 | "Insufficient permissions?\n"; | |
1381 | } | |
1382 | } | |
1383 | } | |
1384 | \@out; | |
1385 | } | |
1386 | ||
29499c0b MS |
1387 | # parse_svn_date(DATE) |
1388 | # -------------------- | |
1389 | # Given a date (in UTC) from Subversion, return a string in the format | |
1390 | # "<TZ Offset> <local date/time>" that Git will use. | |
1391 | # | |
1392 | # By default the parsed date will be in UTC; if $Git::SVN::_localtime | |
1393 | # is true we'll convert it to the local timezone instead. | |
1394 | sub parse_svn_date { | |
1395 | my $date = shift || return '+0000 1970-01-01 00:00:00'; | |
1396 | my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T | |
784f4b6f | 1397 | (\d\d?)\:(\d\d)\:(\d\d)\.\d*Z$/x) or |
29499c0b MS |
1398 | croak "Unable to parse date: $date\n"; |
1399 | my $parsed_date; # Set next. | |
1400 | ||
1401 | if ($Git::SVN::_localtime) { | |
1402 | # Translate the Subversion datetime to an epoch time. | |
1403 | # Begin by switching ourselves to $date's timezone, UTC. | |
1404 | my $old_env_TZ = $ENV{TZ}; | |
1405 | $ENV{TZ} = 'UTC'; | |
1406 | ||
1407 | my $epoch_in_UTC = | |
45c956b3 | 1408 | Time::Local::timelocal($S, $M, $H, $d, $m - 1, $Y - 1900); |
29499c0b MS |
1409 | |
1410 | # Determine our local timezone (including DST) at the | |
1411 | # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the | |
1412 | # value of TZ, if any, at the time we were run. | |
1413 | if (defined $Git::SVN::Log::TZ) { | |
1414 | $ENV{TZ} = $Git::SVN::Log::TZ; | |
1415 | } else { | |
1416 | delete $ENV{TZ}; | |
1417 | } | |
1418 | ||
68868ff5 | 1419 | my $our_TZ = get_tz_offset(); |
29499c0b MS |
1420 | |
1421 | # This converts $epoch_in_UTC into our local timezone. | |
1422 | my ($sec, $min, $hour, $mday, $mon, $year, | |
1423 | $wday, $yday, $isdst) = localtime($epoch_in_UTC); | |
1424 | ||
1425 | $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d', | |
1426 | $our_TZ, $year + 1900, $mon + 1, | |
1427 | $mday, $hour, $min, $sec); | |
1428 | ||
1429 | # Reset us to the timezone in effect when we entered | |
1430 | # this routine. | |
1431 | if (defined $old_env_TZ) { | |
1432 | $ENV{TZ} = $old_env_TZ; | |
1433 | } else { | |
1434 | delete $ENV{TZ}; | |
1435 | } | |
1436 | } else { | |
1437 | $parsed_date = "+0000 $Y-$m-$d $H:$M:$S"; | |
1438 | } | |
1439 | ||
1440 | return $parsed_date; | |
1441 | } | |
1442 | ||
1443 | sub other_gs { | |
1444 | my ($self, $new_url, $url, | |
1445 | $branch_from, $r, $old_ref_id) = @_; | |
1446 | my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from); | |
1447 | unless ($gs) { | |
1448 | my $ref_id = $old_ref_id; | |
1449 | $ref_id =~ s/\@\d+-*$//; | |
1450 | $ref_id .= "\@$r"; | |
1451 | # just grow a tail if we're not unique enough :x | |
1452 | $ref_id .= '-' while find_ref($ref_id); | |
1453 | my ($u, $p, $repo_id) = ($new_url, '', $ref_id); | |
1454 | if ($u =~ s#^\Q$url\E(/|$)##) { | |
1455 | $p = $u; | |
1456 | $u = $url; | |
1457 | $repo_id = $self->{repo_id}; | |
1458 | } | |
1459 | while (1) { | |
1460 | # It is possible to tag two different subdirectories at | |
1461 | # the same revision. If the url for an existing ref | |
1462 | # does not match, we must either find a ref with a | |
1463 | # matching url or create a new ref by growing a tail. | |
1464 | $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1); | |
1465 | my (undef, $max_commit) = $gs->rev_map_max(1); | |
1466 | last if (!$max_commit); | |
1467 | my ($url) = ::cmt_metadata($max_commit); | |
1468 | last if ($url eq $gs->metadata_url); | |
1469 | $ref_id .= '-'; | |
1470 | } | |
1471 | print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1; | |
1472 | } | |
1473 | $gs | |
1474 | } | |
1475 | ||
1476 | sub call_authors_prog { | |
1477 | my ($orig_author) = @_; | |
1478 | $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author); | |
1479 | my $author = `$::_authors_prog $orig_author`; | |
1480 | if ($? != 0) { | |
1481 | die "$::_authors_prog failed with exit code $?\n" | |
1482 | } | |
1483 | if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) { | |
1484 | my ($name, $email) = ($1, $2); | |
1485 | $email = undef if length $2 == 0; | |
1486 | return [$name, $email]; | |
1487 | } else { | |
1488 | die "Author: $orig_author: $::_authors_prog returned " | |
1489 | . "invalid author format: $author\n"; | |
1490 | } | |
1491 | } | |
1492 | ||
1493 | sub check_author { | |
1494 | my ($author) = @_; | |
1495 | if (!defined $author || length $author == 0) { | |
1496 | $author = '(no author)'; | |
1497 | } | |
1498 | if (!defined $::users{$author}) { | |
1499 | if (defined $::_authors_prog) { | |
1500 | $::users{$author} = call_authors_prog($author); | |
1501 | } elsif (defined $::_authors) { | |
1502 | die "Author: $author not defined in $::_authors file\n"; | |
1503 | } | |
1504 | } | |
1505 | $author; | |
1506 | } | |
1507 | ||
1508 | sub find_extra_svk_parents { | |
4ae9a7b9 | 1509 | my ($self, $tickets, $parents) = @_; |
29499c0b MS |
1510 | # aha! svk:merge property changed... |
1511 | my @tickets = split "\n", $tickets; | |
1512 | my @known_parents; | |
1513 | for my $ticket ( @tickets ) { | |
1514 | my ($uuid, $path, $rev) = split /:/, $ticket; | |
1515 | if ( $uuid eq $self->ra_uuid ) { | |
d2fd119c | 1516 | my $repos_root = $self->url; |
29499c0b MS |
1517 | my $branch_from = $path; |
1518 | $branch_from =~ s{^/}{}; | |
d2fd119c MS |
1519 | my $gs = $self->other_gs(add_path_to_url( $repos_root, $branch_from ), |
1520 | $repos_root, | |
29499c0b MS |
1521 | $branch_from, |
1522 | $rev, | |
1523 | $self->{ref_id}); | |
1524 | if ( my $commit = $gs->rev_map_get($rev, $uuid) ) { | |
1525 | # wahey! we found it, but it might be | |
1526 | # an old one (!) | |
1527 | push @known_parents, [ $rev, $commit ]; | |
1528 | } | |
1529 | } | |
1530 | } | |
1531 | # Ordering matters; highest-numbered commit merge tickets | |
1532 | # first, as they may account for later merge ticket additions | |
1533 | # or changes. | |
1534 | @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents; | |
1535 | for my $parent ( @known_parents ) { | |
1536 | my @cmd = ('rev-list', $parent, map { "^$_" } @$parents ); | |
1537 | my ($msg_fh, $ctx) = command_output_pipe(@cmd); | |
1538 | my $new; | |
1539 | while ( <$msg_fh> ) { | |
1540 | $new=1;last; | |
1541 | } | |
1542 | command_close_pipe($msg_fh, $ctx); | |
1543 | if ( $new ) { | |
1544 | print STDERR | |
1545 | "Found merge parent (svk:merge ticket): $parent\n"; | |
1546 | push @$parents, $parent; | |
1547 | } | |
1548 | } | |
1549 | } | |
1550 | ||
1551 | sub lookup_svn_merge { | |
1552 | my $uuid = shift; | |
1553 | my $url = shift; | |
abfef3bb JSO |
1554 | my $source = shift; |
1555 | my $revs = shift; | |
29499c0b | 1556 | |
29499c0b MS |
1557 | my $path = $source; |
1558 | $path =~ s{^/}{}; | |
1559 | my $gs = Git::SVN->find_by_url($url.$source, $url, $path); | |
1560 | if ( !$gs ) { | |
1561 | warn "Couldn't find revmap for $url$source\n"; | |
1562 | return; | |
1563 | } | |
1564 | my @ranges = split ",", $revs; | |
1565 | my ($tip, $tip_commit); | |
1566 | my @merged_commit_ranges; | |
1567 | # find the tip | |
1568 | for my $range ( @ranges ) { | |
47543d16 JP |
1569 | if ($range =~ /[*]$/) { |
1570 | warn "W: Ignoring partial merge in svn:mergeinfo " | |
1571 | ."dirprop: $source:$range\n"; | |
1572 | next; | |
1573 | } | |
29499c0b MS |
1574 | my ($bottom, $top) = split "-", $range; |
1575 | $top ||= $bottom; | |
1576 | my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top ); | |
1577 | my $top_commit = $gs->find_rev_before( $top, 1, $bottom ); | |
1578 | ||
1579 | unless ($top_commit and $bottom_commit) { | |
eae6cf5a | 1580 | warn "W: unknown path/rev in svn:mergeinfo " |
29499c0b MS |
1581 | ."dirprop: $source:$range\n"; |
1582 | next; | |
1583 | } | |
1584 | ||
1585 | if (scalar(command('rev-parse', "$bottom_commit^@"))) { | |
1586 | push @merged_commit_ranges, | |
1587 | "$bottom_commit^..$top_commit"; | |
1588 | } else { | |
1589 | push @merged_commit_ranges, "$top_commit"; | |
1590 | } | |
1591 | ||
1592 | if ( !defined $tip or $top > $tip ) { | |
1593 | $tip = $top; | |
1594 | $tip_commit = $top_commit; | |
1595 | } | |
1596 | } | |
1597 | return ($tip_commit, @merged_commit_ranges); | |
1598 | } | |
1599 | ||
1600 | sub _rev_list { | |
1601 | my ($msg_fh, $ctx) = command_output_pipe( | |
1602 | "rev-list", @_, | |
1603 | ); | |
1604 | my @rv; | |
1605 | while ( <$msg_fh> ) { | |
1606 | chomp; | |
1607 | push @rv, $_; | |
1608 | } | |
1609 | command_close_pipe($msg_fh, $ctx); | |
1610 | @rv; | |
1611 | } | |
1612 | ||
d0b34f24 | 1613 | sub check_cherry_pick2 { |
29499c0b MS |
1614 | my $base = shift; |
1615 | my $tip = shift; | |
1616 | my $parents = shift; | |
1617 | my @ranges = @_; | |
1618 | my %commits = map { $_ => 1 } | |
1619 | _rev_list("--no-merges", $tip, "--not", $base, @$parents, "--"); | |
1620 | for my $range ( @ranges ) { | |
1621 | delete @commits{_rev_list($range, "--")}; | |
1622 | } | |
1623 | for my $commit (keys %commits) { | |
1624 | if (has_no_changes($commit)) { | |
1625 | delete $commits{$commit}; | |
1626 | } | |
1627 | } | |
d0b34f24 EW |
1628 | my @k = (keys %commits); |
1629 | return (scalar @k, $k[0]); | |
29499c0b MS |
1630 | } |
1631 | ||
1632 | sub has_no_changes { | |
1633 | my $commit = shift; | |
1634 | ||
1635 | my @revs = split / /, command_oneline( | |
1636 | qw(rev-list --parents -1 -m), $commit); | |
1637 | ||
1638 | # Commits with no parents, e.g. the start of a partial branch, | |
1639 | # have changes by definition. | |
1640 | return 1 if (@revs < 2); | |
1641 | ||
1642 | # Commits with multiple parents, e.g a merge, have no changes | |
1643 | # by definition. | |
1644 | return 0 if (@revs > 2); | |
1645 | ||
1646 | return (command_oneline("rev-parse", "$commit^{tree}") eq | |
1647 | command_oneline("rev-parse", "$commit~1^{tree}")); | |
1648 | } | |
1649 | ||
1650 | sub tie_for_persistent_memoization { | |
1651 | my $hash = shift; | |
1652 | my $path = shift; | |
1653 | ||
47092c10 EW |
1654 | unless ($memo_backend) { |
1655 | if (eval { require Git::SVN::Memoize::YAML; 1}) { | |
1656 | $memo_backend = 1; | |
1657 | } else { | |
1658 | require Memoize::Storable; | |
1659 | $memo_backend = -1; | |
1660 | } | |
1661 | } | |
1662 | ||
1663 | if ($memo_backend > 0) { | |
29499c0b MS |
1664 | tie %$hash => 'Git::SVN::Memoize::YAML', "$path.yaml"; |
1665 | } else { | |
a2c761ce GL |
1666 | # first verify that any existing file can actually be loaded |
1667 | # (it may have been saved by an incompatible version) | |
1668 | my $db = "$path.db"; | |
1669 | if (-e $db) { | |
1670 | use Storable qw(retrieve); | |
1671 | ||
1672 | if (!eval { retrieve($db); 1 }) { | |
1673 | unlink $db or die "unlink $db failed: $!"; | |
1674 | } | |
1675 | } | |
1676 | tie %$hash => 'Memoize::Storable', $db, 'nstore'; | |
29499c0b MS |
1677 | } |
1678 | } | |
1679 | ||
1680 | # The GIT_DIR environment variable is not always set until after the command | |
1681 | # line arguments are processed, so we can't memoize in a BEGIN block. | |
1682 | { | |
1683 | my $memoized = 0; | |
1684 | ||
1685 | sub memoize_svn_mergeinfo_functions { | |
1686 | return if $memoized; | |
1687 | $memoized = 1; | |
1688 | ||
112423eb | 1689 | my $cache_path = svn_dir() . '/.caches/'; |
29499c0b MS |
1690 | mkpath([$cache_path]) unless -d $cache_path; |
1691 | ||
1692 | my %lookup_svn_merge_cache; | |
d0b34f24 | 1693 | my %check_cherry_pick2_cache; |
29499c0b MS |
1694 | my %has_no_changes_cache; |
1695 | ||
1696 | tie_for_persistent_memoization(\%lookup_svn_merge_cache, | |
1697 | "$cache_path/lookup_svn_merge"); | |
1698 | memoize 'lookup_svn_merge', | |
1699 | SCALAR_CACHE => 'FAULT', | |
1700 | LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache], | |
1701 | ; | |
1702 | ||
d0b34f24 EW |
1703 | tie_for_persistent_memoization(\%check_cherry_pick2_cache, |
1704 | "$cache_path/check_cherry_pick2"); | |
1705 | memoize 'check_cherry_pick2', | |
29499c0b | 1706 | SCALAR_CACHE => 'FAULT', |
d0b34f24 | 1707 | LIST_CACHE => ['HASH' => \%check_cherry_pick2_cache], |
29499c0b MS |
1708 | ; |
1709 | ||
1710 | tie_for_persistent_memoization(\%has_no_changes_cache, | |
1711 | "$cache_path/has_no_changes"); | |
1712 | memoize 'has_no_changes', | |
1713 | SCALAR_CACHE => ['HASH' => \%has_no_changes_cache], | |
1714 | LIST_CACHE => 'FAULT', | |
1715 | ; | |
1716 | } | |
1717 | ||
1718 | sub unmemoize_svn_mergeinfo_functions { | |
1719 | return if not $memoized; | |
1720 | $memoized = 0; | |
1721 | ||
1722 | Memoize::unmemoize 'lookup_svn_merge'; | |
d0b34f24 | 1723 | Memoize::unmemoize 'check_cherry_pick2'; |
29499c0b MS |
1724 | Memoize::unmemoize 'has_no_changes'; |
1725 | } | |
1726 | ||
61b472ed PB |
1727 | sub clear_memoized_mergeinfo_caches { |
1728 | die "Only call this method in non-memoized context" if ($memoized); | |
1729 | ||
112423eb | 1730 | my $cache_path = svn_dir() . '/.caches/'; |
61b472ed PB |
1731 | return unless -d $cache_path; |
1732 | ||
1733 | for my $cache_file (("$cache_path/lookup_svn_merge", | |
d0b34f24 EW |
1734 | "$cache_path/check_cherry_pick", # old |
1735 | "$cache_path/check_cherry_pick2", | |
61b472ed PB |
1736 | "$cache_path/has_no_changes")) { |
1737 | for my $suffix (qw(yaml db)) { | |
1738 | my $file = "$cache_file.$suffix"; | |
1739 | next unless -e $file; | |
1740 | unlink($file) or die "unlink($file) failed: $!\n"; | |
1741 | } | |
1742 | } | |
1743 | } | |
1744 | ||
1745 | ||
29499c0b MS |
1746 | Memoize::memoize 'Git::SVN::repos_root'; |
1747 | } | |
1748 | ||
1749 | END { | |
1750 | # Force cache writeout explicitly instead of waiting for | |
1751 | # global destruction to avoid segfault in Storable: | |
1752 | # http://rt.cpan.org/Public/Bug/Display.html?id=36087 | |
1753 | unmemoize_svn_mergeinfo_functions(); | |
1754 | } | |
1755 | ||
1756 | sub parents_exclude { | |
1757 | my $parents = shift; | |
1758 | my @commits = @_; | |
1759 | return unless @commits; | |
1760 | ||
1761 | my @excluded; | |
1762 | my $excluded; | |
1763 | do { | |
1764 | my @cmd = ('rev-list', "-1", @commits, "--not", @$parents ); | |
1765 | $excluded = command_oneline(@cmd); | |
1766 | if ( $excluded ) { | |
1767 | my @new; | |
1768 | my $found; | |
1769 | for my $commit ( @commits ) { | |
1770 | if ( $commit eq $excluded ) { | |
1771 | push @excluded, $commit; | |
1772 | $found++; | |
29499c0b MS |
1773 | } |
1774 | else { | |
1775 | push @new, $commit; | |
1776 | } | |
1777 | } | |
1778 | die "saw commit '$excluded' in rev-list output, " | |
1779 | ."but we didn't ask for that commit (wanted: @commits --not @$parents)" | |
1780 | unless $found; | |
1781 | @commits = @new; | |
1782 | } | |
1783 | } | |
1784 | while ($excluded and @commits); | |
1785 | ||
1786 | return @excluded; | |
1787 | } | |
1788 | ||
abfef3bb JSO |
1789 | # Compute what's new in svn:mergeinfo. |
1790 | sub mergeinfo_changes { | |
1791 | my ($self, $old_path, $old_rev, $path, $rev, $mergeinfo_prop) = @_; | |
1792 | my %minfo = map {split ":", $_ } split "\n", $mergeinfo_prop; | |
1793 | my $old_minfo = {}; | |
1794 | ||
2b6c613f EW |
1795 | my $ra = $self->ra; |
1796 | # Give up if $old_path isn't in the repo. | |
1797 | # This is probably a merge on a subtree. | |
1798 | if ($ra->check_path($old_path, $old_rev) != $SVN::Node::dir) { | |
1799 | warn "W: ignoring svn:mergeinfo on $old_path, ", | |
1800 | "directory didn't exist in r$old_rev\n"; | |
1801 | return {}; | |
1802 | } | |
1803 | my (undef, undef, $props) = $ra->get_dir($old_path, $old_rev); | |
54b95346 EW |
1804 | if (defined $props->{"svn:mergeinfo"}) { |
1805 | my %omi = map {split ":", $_ } split "\n", | |
1806 | $props->{"svn:mergeinfo"}; | |
1807 | $old_minfo = \%omi; | |
1808 | } | |
abfef3bb JSO |
1809 | |
1810 | my %changes = (); | |
1811 | foreach my $p (keys %minfo) { | |
1812 | my $a = $old_minfo->{$p} || ""; | |
1813 | my $b = $minfo{$p}; | |
1814 | # Omit merged branches whose ranges lists are unchanged. | |
1815 | next if $a eq $b; | |
1816 | # Remove any common range list prefix. | |
1817 | ($a ^ $b) =~ /^[\0]*/; | |
1818 | my $common_prefix = rindex $b, ",", $+[0] - 1; | |
1819 | $changes{$p} = substr $b, $common_prefix + 1; | |
1820 | } | |
1821 | print STDERR "Checking svn:mergeinfo changes since r$old_rev: ", | |
1822 | scalar(keys %minfo), " sources, ", | |
1823 | scalar(keys %changes), " changed\n"; | |
1824 | ||
1825 | return \%changes; | |
1826 | } | |
29499c0b MS |
1827 | |
1828 | # note: this function should only be called if the various dirprops | |
1829 | # have actually changed | |
1830 | sub find_extra_svn_parents { | |
4ae9a7b9 | 1831 | my ($self, $mergeinfo, $parents) = @_; |
29499c0b MS |
1832 | # aha! svk:merge property changed... |
1833 | ||
1834 | memoize_svn_mergeinfo_functions(); | |
1835 | ||
1836 | # We first search for merged tips which are not in our | |
1837 | # history. Then, we figure out which git revisions are in | |
1838 | # that tip, but not this revision. If all of those revisions | |
1839 | # are now marked as merge, we can add the tip as a parent. | |
abfef3bb | 1840 | my @merges = sort keys %$mergeinfo; |
29499c0b | 1841 | my @merge_tips; |
06ee19e8 | 1842 | my $url = $self->url; |
29499c0b | 1843 | my $uuid = $self->ra_uuid; |
f271fad2 | 1844 | my @all_ranges; |
29499c0b MS |
1845 | for my $merge ( @merges ) { |
1846 | my ($tip_commit, @ranges) = | |
abfef3bb JSO |
1847 | lookup_svn_merge( $uuid, $url, |
1848 | $merge, $mergeinfo->{$merge} ); | |
29499c0b MS |
1849 | unless (!$tip_commit or |
1850 | grep { $_ eq $tip_commit } @$parents ) { | |
1851 | push @merge_tips, $tip_commit; | |
f271fad2 | 1852 | push @all_ranges, @ranges; |
29499c0b MS |
1853 | } else { |
1854 | push @merge_tips, undef; | |
1855 | } | |
1856 | } | |
1857 | ||
1858 | my %excluded = map { $_ => 1 } | |
1859 | parents_exclude($parents, grep { defined } @merge_tips); | |
1860 | ||
1861 | # check merge tips for new parents | |
1862 | my @new_parents; | |
1863 | for my $merge_tip ( @merge_tips ) { | |
abfef3bb | 1864 | my $merge = shift @merges; |
29499c0b | 1865 | next unless $merge_tip and $excluded{$merge_tip}; |
abfef3bb | 1866 | my $spec = "$merge:$mergeinfo->{$merge}"; |
29499c0b | 1867 | |
29499c0b MS |
1868 | # check out 'new' tips |
1869 | my $merge_base; | |
1870 | eval { | |
1871 | $merge_base = command_oneline( | |
1872 | "merge-base", | |
1873 | @$parents, $merge_tip, | |
1874 | ); | |
1875 | }; | |
1876 | if ($@) { | |
1877 | die "An error occurred during merge-base" | |
1878 | unless $@->isa("Git::Error::Command"); | |
1879 | ||
1880 | warn "W: Cannot find common ancestor between ". | |
1881 | "@$parents and $merge_tip. Ignoring merge info.\n"; | |
1882 | next; | |
1883 | } | |
1884 | ||
1885 | # double check that there are no missing non-merge commits | |
d0b34f24 | 1886 | my ($ninc, $ifirst) = check_cherry_pick2( |
29499c0b MS |
1887 | $merge_base, $merge_tip, |
1888 | $parents, | |
f271fad2 | 1889 | @all_ranges, |
29499c0b MS |
1890 | ); |
1891 | ||
d0b34f24 | 1892 | if ($ninc) { |
da0bc948 | 1893 | warn "W: svn cherry-pick ignored ($spec) - missing " . |
d0b34f24 | 1894 | "$ninc commit(s) (eg $ifirst)\n"; |
29499c0b | 1895 | } else { |
da0bc948 | 1896 | warn "Found merge parent ($spec): ", $merge_tip, "\n"; |
29499c0b MS |
1897 | push @new_parents, $merge_tip; |
1898 | } | |
1899 | } | |
1900 | ||
1901 | # cater for merges which merge commits from multiple branches | |
1902 | if ( @new_parents > 1 ) { | |
1903 | for ( my $i = 0; $i <= $#new_parents; $i++ ) { | |
1904 | for ( my $j = 0; $j <= $#new_parents; $j++ ) { | |
1905 | next if $i == $j; | |
1906 | next unless $new_parents[$i]; | |
1907 | next unless $new_parents[$j]; | |
1908 | my $revs = command_oneline( | |
1909 | "rev-list", "-1", | |
1910 | "$new_parents[$i]..$new_parents[$j]", | |
1911 | ); | |
1912 | if ( !$revs ) { | |
1913 | undef($new_parents[$j]); | |
1914 | } | |
1915 | } | |
1916 | } | |
1917 | } | |
1918 | push @$parents, grep { defined } @new_parents; | |
1919 | } | |
1920 | ||
1921 | sub make_log_entry { | |
abfef3bb | 1922 | my ($self, $rev, $parents, $ed, $parent_rev, $parent_path) = @_; |
29499c0b MS |
1923 | my $untracked = $self->get_untracked($ed); |
1924 | ||
1925 | my @parents = @$parents; | |
9ee13a93 | 1926 | my $props = $ed->{dir_prop}{$self->path}; |
6d523a3a EW |
1927 | if ($self->follow_parent) { |
1928 | my $tickets = $props->{"svk:merge"}; | |
1929 | if ($tickets) { | |
1930 | $self->find_extra_svk_parents($tickets, \@parents); | |
1931 | } | |
1932 | ||
1933 | my $mergeinfo_prop = $props->{"svn:mergeinfo"}; | |
1934 | if ($mergeinfo_prop) { | |
1935 | my $mi_changes = $self->mergeinfo_changes( | |
1936 | $parent_path, | |
1937 | $parent_rev, | |
1938 | $self->path, | |
1939 | $rev, | |
1940 | $mergeinfo_prop); | |
1941 | $self->find_extra_svn_parents($mi_changes, \@parents); | |
1942 | } | |
29499c0b MS |
1943 | } |
1944 | ||
1945 | open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!; | |
1946 | print $un "r$rev\n" or croak $!; | |
1947 | print $un $_, "\n" foreach @$untracked; | |
1948 | my %log_entry = ( parents => \@parents, revision => $rev, | |
1949 | log => ''); | |
1950 | ||
1951 | my $headrev; | |
1952 | my $logged = delete $self->{logged_rev_props}; | |
1953 | if (!$logged || $self->{-want_revprops}) { | |
1954 | my $rp = $self->ra->rev_proplist($rev); | |
1955 | foreach (sort keys %$rp) { | |
1956 | my $v = $rp->{$_}; | |
1957 | if (/^svn:(author|date|log)$/) { | |
1958 | $log_entry{$1} = $v; | |
1959 | } elsif ($_ eq 'svm:headrev') { | |
1960 | $headrev = $v; | |
1961 | } else { | |
1962 | print $un " rev_prop: ", uri_encode($_), ' ', | |
1963 | uri_encode($v), "\n"; | |
1964 | } | |
1965 | } | |
1966 | } else { | |
1967 | map { $log_entry{$_} = $logged->{$_} } keys %$logged; | |
1968 | } | |
1969 | close $un or croak $!; | |
1970 | ||
1971 | $log_entry{date} = parse_svn_date($log_entry{date}); | |
1972 | $log_entry{log} .= "\n"; | |
1973 | my $author = $log_entry{author} = check_author($log_entry{author}); | |
1974 | my ($name, $email) = defined $::users{$author} ? @{$::users{$author}} | |
1975 | : ($author, undef); | |
1976 | ||
1977 | my ($commit_name, $commit_email) = ($name, $email); | |
1978 | if ($_use_log_author) { | |
1979 | my $name_field; | |
1980 | if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) { | |
1981 | $name_field = $1; | |
1982 | } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) { | |
1983 | $name_field = $1; | |
1984 | } | |
1985 | if (!defined $name_field) { | |
1986 | if (!defined $email) { | |
1987 | $email = $name; | |
1988 | } | |
1989 | } elsif ($name_field =~ /(.*?)\s+<(.*)>/) { | |
1990 | ($name, $email) = ($1, $2); | |
1991 | } elsif ($name_field =~ /(.*)@/) { | |
1992 | ($name, $email) = ($1, $name_field); | |
1993 | } else { | |
1994 | ($name, $email) = ($name_field, $name_field); | |
1995 | } | |
1996 | } | |
1997 | if (defined $headrev && $self->use_svm_props) { | |
1998 | if ($self->rewrite_root) { | |
1999 | die "Can't have both 'useSvmProps' and 'rewriteRoot' ", | |
2000 | "options set!\n"; | |
2001 | } | |
2002 | if ($self->rewrite_uuid) { | |
2003 | die "Can't have both 'useSvmProps' and 'rewriteUUID' ", | |
2004 | "options set!\n"; | |
2005 | } | |
2006 | my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i; | |
2007 | # we don't want "SVM: initializing mirror for junk" ... | |
2008 | return undef if $r == 0; | |
2009 | my $svm = $self->svm; | |
2010 | if ($uuid ne $svm->{uuid}) { | |
2011 | die "UUID mismatch on SVM path:\n", | |
2012 | "expected: $svm->{uuid}\n", | |
2013 | " got: $uuid\n"; | |
2014 | } | |
2015 | my $full_url = $self->full_url; | |
2016 | $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or | |
2017 | die "Failed to replace '$svm->{replace}' with ", | |
2018 | "'$svm->{source}' in $full_url\n"; | |
2019 | # throw away username for storing in records | |
2020 | remove_username($full_url); | |
2021 | $log_entry{metadata} = "$full_url\@$r $uuid"; | |
2022 | $log_entry{svm_revision} = $r; | |
2023 | $email ||= "$author\@$uuid"; | |
2024 | $commit_email ||= "$author\@$uuid"; | |
2025 | } elsif ($self->use_svnsync_props) { | |
705b49cb MS |
2026 | my $full_url = canonicalize_url( |
2027 | add_path_to_url( $self->svnsync->{url}, $self->path ) | |
2028 | ); | |
29499c0b MS |
2029 | remove_username($full_url); |
2030 | my $uuid = $self->svnsync->{uuid}; | |
2031 | $log_entry{metadata} = "$full_url\@$rev $uuid"; | |
2032 | $email ||= "$author\@$uuid"; | |
2033 | $commit_email ||= "$author\@$uuid"; | |
2034 | } else { | |
2035 | my $url = $self->metadata_url; | |
2036 | remove_username($url); | |
2037 | my $uuid = $self->rewrite_uuid || $self->ra->get_uuid; | |
2038 | $log_entry{metadata} = "$url\@$rev " . $uuid; | |
2039 | $email ||= "$author\@" . $uuid; | |
2040 | $commit_email ||= "$author\@" . $uuid; | |
2041 | } | |
2042 | $log_entry{name} = $name; | |
2043 | $log_entry{email} = $email; | |
2044 | $log_entry{commit_name} = $commit_name; | |
2045 | $log_entry{commit_email} = $commit_email; | |
2046 | \%log_entry; | |
2047 | } | |
2048 | ||
2049 | sub fetch { | |
2050 | my ($self, $min_rev, $max_rev, @parents) = @_; | |
2051 | my ($last_rev, $last_commit) = $self->last_rev_commit; | |
2052 | my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev); | |
2053 | $self->ra->gs_fetch_loop_common($base, $head, [$self]); | |
2054 | } | |
2055 | ||
2056 | sub set_tree_cb { | |
2057 | my ($self, $log_entry, $tree, $rev, $date, $author) = @_; | |
2058 | $self->{inject_parents} = { $rev => $tree }; | |
2059 | $self->fetch(undef, undef); | |
2060 | } | |
2061 | ||
2062 | sub set_tree { | |
2063 | my ($self, $tree) = (shift, shift); | |
2064 | my $log_entry = ::get_commit_entry($tree); | |
2065 | unless ($self->{last_rev}) { | |
2066 | fatal("Must have an existing revision to commit"); | |
2067 | } | |
2068 | my %ed_opts = ( r => $self->{last_rev}, | |
2069 | log => $log_entry->{log}, | |
2070 | ra => $self->ra, | |
2071 | tree_a => $self->{last_commit}, | |
2072 | tree_b => $tree, | |
2073 | editor_cb => sub { | |
2074 | $self->set_tree_cb($log_entry, $tree, @_) }, | |
5578ed74 | 2075 | svn_path => $self->path ); |
29499c0b MS |
2076 | if (!Git::SVN::Editor->new(\%ed_opts)->apply_diff) { |
2077 | print "No changes\nr$self->{last_rev} = $tree\n"; | |
2078 | } | |
2079 | } | |
2080 | ||
2081 | sub rebuild_from_rev_db { | |
2082 | my ($self, $path) = @_; | |
2083 | my $r = -1; | |
2084 | open my $fh, '<', $path or croak "open: $!"; | |
2085 | binmode $fh or croak "binmode: $!"; | |
2086 | while (<$fh>) { | |
2087 | length($_) == 41 or croak "inconsistent size in ($_) != 41"; | |
2088 | chomp($_); | |
2089 | ++$r; | |
2090 | next if $_ eq ('0' x 40); | |
2091 | $self->rev_map_set($r, $_); | |
2092 | print "r$r = $_\n"; | |
2093 | } | |
2094 | close $fh or croak "close: $!"; | |
2095 | unlink $path or croak "unlink: $!"; | |
2096 | } | |
2097 | ||
ab0bcec9 | 2098 | #define a global associate map to record rebuild status |
2099 | my %rebuild_status; | |
2100 | #define a global associate map to record rebuild verify status | |
2101 | my %rebuild_verify_status; | |
2102 | ||
29499c0b MS |
2103 | sub rebuild { |
2104 | my ($self) = @_; | |
2105 | my $map_path = $self->map_path; | |
2106 | my $partial = (-e $map_path && ! -z $map_path); | |
ab0bcec9 | 2107 | my $verify_key = $self->refname.'^0'; |
2108 | if (!$rebuild_verify_status{$verify_key}) { | |
2109 | my $verify_result = ::verify_ref($verify_key); | |
2110 | if ($verify_result) { | |
2111 | $rebuild_verify_status{$verify_key} = 1; | |
2112 | } | |
2113 | } | |
2114 | if (!$rebuild_verify_status{$verify_key}) { | |
2115 | return; | |
2116 | } | |
29499c0b MS |
2117 | if (!$partial && ($self->use_svm_props || $self->no_metadata)) { |
2118 | my $rev_db = $self->rev_db_path; | |
2119 | $self->rebuild_from_rev_db($rev_db); | |
2120 | if ($self->use_svm_props) { | |
2121 | my $svm_rev_db = $self->rev_db_path($self->svm_uuid); | |
2122 | $self->rebuild_from_rev_db($svm_rev_db); | |
2123 | } | |
2124 | $self->unlink_rev_db_symlink; | |
2125 | return; | |
2126 | } | |
2127 | print "Rebuilding $map_path ...\n" if (!$partial); | |
2128 | my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) : | |
2129 | (undef, undef)); | |
ab0bcec9 | 2130 | my $key_value = ($head ? "$head.." : "") . $self->refname; |
2131 | if (exists $rebuild_status{$key_value}) { | |
2132 | print "Done rebuilding $map_path\n" if (!$partial || !$head); | |
2133 | my $rev_db_path = $self->rev_db_path; | |
2134 | if (-f $self->rev_db_path) { | |
2135 | unlink $self->rev_db_path or croak "unlink: $!"; | |
2136 | } | |
2137 | $self->unlink_rev_db_symlink; | |
2138 | return; | |
2139 | } | |
29499c0b | 2140 | my ($log, $ctx) = |
ab0bcec9 | 2141 | command_output_pipe(qw/rev-list --pretty=raw --reverse/, |
2142 | $key_value, | |
29499c0b | 2143 | '--'); |
ab0bcec9 | 2144 | $rebuild_status{$key_value} = 1; |
29499c0b MS |
2145 | my $metadata_url = $self->metadata_url; |
2146 | remove_username($metadata_url); | |
2147 | my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid; | |
2148 | my $c; | |
2149 | while (<$log>) { | |
2150 | if ( m{^commit ($::sha1)$} ) { | |
2151 | $c = $1; | |
2152 | next; | |
2153 | } | |
2154 | next unless s{^\s*(git-svn-id:)}{$1}; | |
2155 | my ($url, $rev, $uuid) = ::extract_metadata($_); | |
2156 | remove_username($url); | |
2157 | ||
2158 | # ignore merges (from set-tree) | |
2159 | next if (!defined $rev || !$uuid); | |
2160 | ||
2161 | # if we merged or otherwise started elsewhere, this is | |
2162 | # how we break out of it | |
2163 | if (($uuid ne $svn_uuid) || | |
2164 | ($metadata_url && $url && ($url ne $metadata_url))) { | |
2165 | next; | |
2166 | } | |
2167 | if ($partial && $head) { | |
2168 | print "Partial-rebuilding $map_path ...\n"; | |
2169 | print "Currently at $base_rev = $head\n"; | |
2170 | $head = undef; | |
2171 | } | |
2172 | ||
2173 | $self->rev_map_set($rev, $c); | |
2174 | print "r$rev = $c\n"; | |
2175 | } | |
2176 | command_close_pipe($log, $ctx); | |
2177 | print "Done rebuilding $map_path\n" if (!$partial || !$head); | |
2178 | my $rev_db_path = $self->rev_db_path; | |
2179 | if (-f $self->rev_db_path) { | |
2180 | unlink $self->rev_db_path or croak "unlink: $!"; | |
2181 | } | |
2182 | $self->unlink_rev_db_symlink; | |
2183 | } | |
2184 | ||
2185 | # rev_map: | |
2186 | # Tie::File seems to be prone to offset errors if revisions get sparse, | |
2187 | # it's not that fast, either. Tie::File is also not in Perl 5.6. So | |
2188 | # one of my favorite modules is out :< Next up would be one of the DBM | |
2189 | # modules, but I'm not sure which is most portable... | |
2190 | # | |
2191 | # This is the replacement for the rev_db format, which was too big | |
2192 | # and inefficient for large repositories with a lot of sparse history | |
2193 | # (mainly tags) | |
2194 | # | |
2195 | # The format is this: | |
2196 | # - 24 bytes for every record, | |
2197 | # * 4 bytes for the integer representing an SVN revision number | |
2198 | # * 20 bytes representing the sha1 of a git commit | |
2199 | # - No empty padding records like the old format | |
2200 | # (except the last record, which can be overwritten) | |
2201 | # - new records are written append-only since SVN revision numbers | |
2202 | # increase monotonically | |
2203 | # - lookups on SVN revision number are done via a binary search | |
2204 | # - Piping the file to xxd -c24 is a good way of dumping it for | |
2205 | # viewing or editing (piped back through xxd -r), should the need | |
2206 | # ever arise. | |
2207 | # - The last record can be padding revision with an all-zero sha1 | |
2208 | # This is used to optimize fetch performance when using multiple | |
2209 | # "fetch" directives in .git/config | |
2210 | # | |
2211 | # These files are disposable unless noMetadata or useSvmProps is set | |
2212 | ||
2213 | sub _rev_map_set { | |
2214 | my ($fh, $rev, $commit) = @_; | |
2215 | ||
2216 | binmode $fh or croak "binmode: $!"; | |
2217 | my $size = (stat($fh))[7]; | |
2218 | ($size % 24) == 0 or croak "inconsistent size: $size"; | |
2219 | ||
2220 | my $wr_offset = 0; | |
2221 | if ($size > 0) { | |
2222 | sysseek($fh, -24, SEEK_END) or croak "seek: $!"; | |
2223 | my $read = sysread($fh, my $buf, 24) or croak "read: $!"; | |
2224 | $read == 24 or croak "read only $read bytes (!= 24)"; | |
2225 | my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf); | |
2226 | if ($last_commit eq ('0' x40)) { | |
2227 | if ($size >= 48) { | |
2228 | sysseek($fh, -48, SEEK_END) or croak "seek: $!"; | |
2229 | $read = sysread($fh, $buf, 24) or | |
2230 | croak "read: $!"; | |
2231 | $read == 24 or | |
2232 | croak "read only $read bytes (!= 24)"; | |
2233 | ($last_rev, $last_commit) = | |
2234 | unpack(rev_map_fmt, $buf); | |
2235 | if ($last_commit eq ('0' x40)) { | |
2236 | croak "inconsistent .rev_map\n"; | |
2237 | } | |
2238 | } | |
2239 | if ($last_rev >= $rev) { | |
2240 | croak "last_rev is higher!: $last_rev >= $rev"; | |
2241 | } | |
2242 | $wr_offset = -24; | |
2243 | } | |
2244 | } | |
2245 | sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!"; | |
2246 | syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or | |
2247 | croak "write: $!"; | |
2248 | } | |
2249 | ||
2250 | sub _rev_map_reset { | |
2251 | my ($fh, $rev, $commit) = @_; | |
2252 | my $c = _rev_map_get($fh, $rev); | |
2253 | $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n"; | |
2254 | my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!"; | |
2255 | truncate $fh, $offset or croak "truncate: $!"; | |
2256 | } | |
2257 | ||
2258 | sub mkfile { | |
2259 | my ($path) = @_; | |
2260 | unless (-e $path) { | |
2261 | my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#); | |
2262 | mkpath([$dir]) unless -d $dir; | |
2263 | open my $fh, '>>', $path or die "Couldn't create $path: $!\n"; | |
2264 | close $fh or die "Couldn't close (create) $path: $!\n"; | |
2265 | } | |
2266 | } | |
2267 | ||
2268 | sub rev_map_set { | |
2269 | my ($self, $rev, $commit, $update_ref, $uuid) = @_; | |
2270 | defined $commit or die "missing arg3\n"; | |
2271 | length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n"; | |
2272 | my $db = $self->map_path($uuid); | |
2273 | my $db_lock = "$db.lock"; | |
2274 | my $sigmask; | |
2275 | $update_ref ||= 0; | |
2276 | if ($update_ref) { | |
2277 | $sigmask = POSIX::SigSet->new(); | |
2278 | my $signew = POSIX::SigSet->new(SIGINT, SIGHUP, SIGTERM, | |
2279 | SIGALRM, SIGUSR1, SIGUSR2); | |
2280 | sigprocmask(SIG_BLOCK, $signew, $sigmask) or | |
2281 | croak "Can't block signals: $!"; | |
2282 | } | |
2283 | mkfile($db); | |
2284 | ||
2285 | $LOCKFILES{$db_lock} = 1; | |
2286 | my $sync; | |
2287 | # both of these options make our .rev_db file very, very important | |
2288 | # and we can't afford to lose it because rebuild() won't work | |
2289 | if ($self->use_svm_props || $self->no_metadata) { | |
47092c10 | 2290 | require File::Copy; |
29499c0b | 2291 | $sync = 1; |
47092c10 | 2292 | File::Copy::copy($db, $db_lock) or die "rev_map_set(@_): ", |
29499c0b MS |
2293 | "Failed to copy: ", |
2294 | "$db => $db_lock ($!)\n"; | |
2295 | } else { | |
2296 | rename $db, $db_lock or die "rev_map_set(@_): ", | |
2297 | "Failed to rename: ", | |
2298 | "$db => $db_lock ($!)\n"; | |
2299 | } | |
2300 | ||
2301 | sysopen(my $fh, $db_lock, O_RDWR | O_CREAT) | |
2302 | or croak "Couldn't open $db_lock: $!\n"; | |
61b472ed PB |
2303 | if ($update_ref eq 'reset') { |
2304 | clear_memoized_mergeinfo_caches(); | |
2305 | _rev_map_reset($fh, $rev, $commit); | |
2306 | } else { | |
2307 | _rev_map_set($fh, $rev, $commit); | |
2308 | } | |
2309 | ||
29499c0b MS |
2310 | if ($sync) { |
2311 | $fh->flush or die "Couldn't flush $db_lock: $!\n"; | |
2312 | $fh->sync or die "Couldn't sync $db_lock: $!\n"; | |
2313 | } | |
2314 | close $fh or croak $!; | |
2315 | if ($update_ref) { | |
2316 | $_head = $self; | |
2317 | my $note = ""; | |
2318 | $note = " ($update_ref)" if ($update_ref !~ /^\d*$/); | |
2319 | command_noisy('update-ref', '-m', "r$rev$note", | |
2320 | $self->refname, $commit); | |
2321 | } | |
2322 | rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ", | |
2323 | "$db_lock => $db ($!)\n"; | |
2324 | delete $LOCKFILES{$db_lock}; | |
2325 | if ($update_ref) { | |
2326 | sigprocmask(SIG_SETMASK, $sigmask) or | |
2327 | croak "Can't restore signal mask: $!"; | |
2328 | } | |
2329 | } | |
2330 | ||
2331 | # If want_commit, this will return an array of (rev, commit) where | |
2332 | # commit _must_ be a valid commit in the archive. | |
2333 | # Otherwise, it'll return the max revision (whether or not the | |
2334 | # commit is valid or just a 0x40 placeholder). | |
2335 | sub rev_map_max { | |
2336 | my ($self, $want_commit) = @_; | |
2337 | $self->rebuild; | |
2338 | my ($r, $c) = $self->rev_map_max_norebuild($want_commit); | |
2339 | $want_commit ? ($r, $c) : $r; | |
2340 | } | |
2341 | ||
2342 | sub rev_map_max_norebuild { | |
2343 | my ($self, $want_commit) = @_; | |
2344 | my $map_path = $self->map_path; | |
2345 | stat $map_path or return $want_commit ? (0, undef) : 0; | |
2346 | sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!"; | |
2347 | binmode $fh or croak "binmode: $!"; | |
2348 | my $size = (stat($fh))[7]; | |
2349 | ($size % 24) == 0 or croak "inconsistent size: $size"; | |
2350 | ||
2351 | if ($size == 0) { | |
2352 | close $fh or croak "close: $!"; | |
2353 | return $want_commit ? (0, undef) : 0; | |
2354 | } | |
2355 | ||
2356 | sysseek($fh, -24, SEEK_END) or croak "seek: $!"; | |
2357 | sysread($fh, my $buf, 24) == 24 or croak "read: $!"; | |
2358 | my ($r, $c) = unpack(rev_map_fmt, $buf); | |
2359 | if ($want_commit && $c eq ('0' x40)) { | |
2360 | if ($size < 48) { | |
2361 | return $want_commit ? (0, undef) : 0; | |
2362 | } | |
2363 | sysseek($fh, -48, SEEK_END) or croak "seek: $!"; | |
2364 | sysread($fh, $buf, 24) == 24 or croak "read: $!"; | |
2365 | ($r, $c) = unpack(rev_map_fmt, $buf); | |
2366 | if ($c eq ('0'x40)) { | |
2367 | croak "Penultimate record is all-zeroes in $map_path"; | |
2368 | } | |
2369 | } | |
2370 | close $fh or croak "close: $!"; | |
2371 | $want_commit ? ($r, $c) : $r; | |
2372 | } | |
2373 | ||
2374 | sub rev_map_get { | |
2375 | my ($self, $rev, $uuid) = @_; | |
2376 | my $map_path = $self->map_path($uuid); | |
2377 | return undef unless -e $map_path; | |
2378 | ||
2379 | sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!"; | |
2380 | my $c = _rev_map_get($fh, $rev); | |
2381 | close($fh) or croak "close: $!"; | |
2382 | $c | |
2383 | } | |
2384 | ||
2385 | sub _rev_map_get { | |
2386 | my ($fh, $rev) = @_; | |
2387 | ||
2388 | binmode $fh or croak "binmode: $!"; | |
2389 | my $size = (stat($fh))[7]; | |
2390 | ($size % 24) == 0 or croak "inconsistent size: $size"; | |
2391 | ||
2392 | if ($size == 0) { | |
2393 | return undef; | |
2394 | } | |
2395 | ||
2396 | my ($l, $u) = (0, $size - 24); | |
2397 | my ($r, $c, $buf); | |
2398 | ||
2399 | while ($l <= $u) { | |
2400 | my $i = int(($l/24 + $u/24) / 2) * 24; | |
2401 | sysseek($fh, $i, SEEK_SET) or croak "seek: $!"; | |
2402 | sysread($fh, my $buf, 24) == 24 or croak "read: $!"; | |
2403 | my ($r, $c) = unpack(rev_map_fmt, $buf); | |
2404 | ||
2405 | if ($r < $rev) { | |
2406 | $l = $i + 24; | |
2407 | } elsif ($r > $rev) { | |
2408 | $u = $i - 24; | |
2409 | } else { # $r == $rev | |
2410 | return $c eq ('0' x 40) ? undef : $c; | |
2411 | } | |
2412 | } | |
2413 | undef; | |
2414 | } | |
2415 | ||
2416 | # Finds the first svn revision that exists on (if $eq_ok is true) or | |
2417 | # before $rev for the current branch. It will not search any lower | |
2418 | # than $min_rev. Returns the git commit hash and svn revision number | |
2419 | # if found, else (undef, undef). | |
2420 | sub find_rev_before { | |
2421 | my ($self, $rev, $eq_ok, $min_rev) = @_; | |
2422 | --$rev unless $eq_ok; | |
2423 | $min_rev ||= 1; | |
2424 | my $max_rev = $self->rev_map_max; | |
2425 | $rev = $max_rev if ($rev > $max_rev); | |
2426 | while ($rev >= $min_rev) { | |
2427 | if (my $c = $self->rev_map_get($rev)) { | |
2428 | return ($rev, $c); | |
2429 | } | |
2430 | --$rev; | |
2431 | } | |
2432 | return (undef, undef); | |
2433 | } | |
2434 | ||
2435 | # Finds the first svn revision that exists on (if $eq_ok is true) or | |
2436 | # after $rev for the current branch. It will not search any higher | |
2437 | # than $max_rev. Returns the git commit hash and svn revision number | |
2438 | # if found, else (undef, undef). | |
2439 | sub find_rev_after { | |
2440 | my ($self, $rev, $eq_ok, $max_rev) = @_; | |
2441 | ++$rev unless $eq_ok; | |
2442 | $max_rev ||= $self->rev_map_max; | |
2443 | while ($rev <= $max_rev) { | |
2444 | if (my $c = $self->rev_map_get($rev)) { | |
2445 | return ($rev, $c); | |
2446 | } | |
2447 | ++$rev; | |
2448 | } | |
2449 | return (undef, undef); | |
2450 | } | |
2451 | ||
2452 | sub _new { | |
2453 | my ($class, $repo_id, $ref_id, $path) = @_; | |
2454 | unless (defined $repo_id && length $repo_id) { | |
2455 | $repo_id = $default_repo_id; | |
2456 | } | |
2457 | unless (defined $ref_id && length $ref_id) { | |
2458 | # Access the prefix option from the git-svn main program if it's loaded. | |
2459 | my $prefix = defined &::opt_prefix ? ::opt_prefix() : ""; | |
2460 | $_[2] = $ref_id = | |
2461 | "refs/remotes/$prefix$default_ref_id"; | |
2462 | } | |
2463 | $_[1] = $repo_id; | |
112423eb EW |
2464 | my $svn_dir = svn_dir(); |
2465 | my $dir = "$svn_dir/$ref_id"; | |
29499c0b | 2466 | |
112423eb EW |
2467 | # Older repos imported by us used $svn_dir/foo instead of |
2468 | # $svn_dir/refs/remotes/foo when tracking refs/remotes/foo | |
9a2bb059 | 2469 | if ($ref_id =~ m{^refs/remotes/(.+)}) { |
112423eb | 2470 | my $old_dir = "$svn_dir/$1"; |
29499c0b MS |
2471 | if (-d $old_dir && ! -d $dir) { |
2472 | $dir = $old_dir; | |
2473 | } | |
2474 | } | |
2475 | ||
2476 | $_[3] = $path = '' unless (defined $path); | |
2477 | mkpath([$dir]); | |
5578ed74 | 2478 | my $obj = bless { |
29499c0b | 2479 | ref_id => $ref_id, dir => $dir, index => "$dir/index", |
112423eb | 2480 | config => "$svn_dir/config", |
29499c0b | 2481 | map_root => "$dir/.rev_map", repo_id => $repo_id }, $class; |
5578ed74 MS |
2482 | |
2483 | # Ensure it gets canonicalized | |
2484 | $obj->path($path); | |
2485 | ||
2486 | return $obj; | |
2487 | } | |
2488 | ||
2489 | sub path { | |
2490 | my $self = shift; | |
2491 | ||
2492 | if (@_) { | |
2493 | my $path = shift; | |
52de6fa2 | 2494 | $self->{_path} = canonicalize_path($path); |
5578ed74 MS |
2495 | return; |
2496 | } | |
2497 | ||
52de6fa2 | 2498 | return $self->{_path}; |
29499c0b MS |
2499 | } |
2500 | ||
06ee19e8 MS |
2501 | sub url { |
2502 | my $self = shift; | |
2503 | ||
2504 | if (@_) { | |
2505 | my $url = shift; | |
565e56c2 | 2506 | $self->{url} = canonicalize_url($url); |
06ee19e8 MS |
2507 | return; |
2508 | } | |
2509 | ||
2510 | return $self->{url}; | |
29499c0b MS |
2511 | } |
2512 | ||
2513 | # for read-only access of old .rev_db formats | |
2514 | sub unlink_rev_db_symlink { | |
2515 | my ($self) = @_; | |
2516 | my $link = $self->rev_db_path; | |
2517 | $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link"; | |
2518 | if (-l $link) { | |
2519 | unlink $link or croak "unlink: $link failed!"; | |
2520 | } | |
2521 | } | |
2522 | ||
2523 | sub rev_db_path { | |
2524 | my ($self, $uuid) = @_; | |
2525 | my $db_path = $self->map_path($uuid); | |
2526 | $db_path =~ s{/\.rev_map\.}{/\.rev_db\.} | |
2527 | or croak "map_path: $db_path does not contain '/.rev_map.' !"; | |
2528 | $db_path; | |
2529 | } | |
2530 | ||
2531 | # the new replacement for .rev_db | |
2532 | sub map_path { | |
2533 | my ($self, $uuid) = @_; | |
2534 | $uuid ||= $self->ra_uuid; | |
2535 | "$self->{map_root}.$uuid"; | |
2536 | } | |
2537 | ||
2538 | sub uri_encode { | |
2539 | my ($f) = @_; | |
1b67bef2 | 2540 | $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#sprintf("%%%02X",ord($1))#eg; |
29499c0b MS |
2541 | $f |
2542 | } | |
2543 | ||
2544 | sub uri_decode { | |
2545 | my ($f) = @_; | |
2546 | $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg; | |
2547 | $f | |
2548 | } | |
2549 | ||
2550 | sub remove_username { | |
2551 | $_[0] =~ s{^([^:]*://)[^@]+@}{$1}; | |
2552 | } | |
2553 | ||
2554 | 1; |