]>
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 | } | |
523a33ca CC |
100 | my $u = (::cmt_metadata("$refname"))[0] or die |
101 | "$refname: no associated commit metadata\n"; | |
29499c0b MS |
102 | $u =~ s!^\Q$url\E(/|$)!! or die |
103 | "$refname: '$url' not found in '$u'\n"; | |
104 | if ($pathname ne $u) { | |
105 | warn "W: Refspec glob conflict ", | |
106 | "(ref: $refname):\n", | |
107 | "expected path: $pathname\n", | |
108 | " real path: $u\n", | |
109 | "Continuing ahead with $u\n"; | |
110 | next; | |
111 | } | |
112 | } else { | |
113 | $fetch->{$pathname} = $refname; | |
114 | } | |
115 | } | |
116 | } | |
117 | ||
118 | sub parse_revision_argument { | |
119 | my ($base, $head) = @_; | |
120 | if (!defined $::_revision || $::_revision eq 'BASE:HEAD') { | |
121 | return ($base, $head); | |
122 | } | |
123 | return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/); | |
124 | return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/); | |
125 | return ($head, $head) if ($::_revision eq 'HEAD'); | |
126 | return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/); | |
127 | return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/); | |
128 | die "revision argument: $::_revision not understood by git-svn\n"; | |
129 | } | |
130 | ||
131 | sub fetch_all { | |
132 | my ($repo_id, $remotes) = @_; | |
133 | if (ref $repo_id) { | |
134 | my $gs = $repo_id; | |
135 | $repo_id = undef; | |
136 | $repo_id = $gs->{repo_id}; | |
137 | } | |
138 | $remotes ||= read_all_remotes(); | |
139 | my $remote = $remotes->{$repo_id} or | |
140 | die "[svn-remote \"$repo_id\"] unknown\n"; | |
141 | my $fetch = $remote->{fetch}; | |
142 | my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n"; | |
143 | my (@gs, @globs); | |
144 | my $ra = Git::SVN::Ra->new($url); | |
145 | my $uuid = $ra->get_uuid; | |
146 | my $head = $ra->get_latest_revnum; | |
147 | ||
148 | # ignore errors, $head revision may not even exist anymore | |
149 | eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) }; | |
150 | warn "W: $@\n" if $@; | |
151 | ||
152 | my $base = defined $fetch ? $head : 0; | |
153 | ||
154 | # read the max revs for wildcard expansion (branches/*, tags/*) | |
155 | foreach my $t (qw/branches tags/) { | |
156 | defined $remote->{$t} or next; | |
157 | push @globs, @{$remote->{$t}}; | |
158 | ||
159 | my $max_rev = eval { tmp_config(qw/--int --get/, | |
160 | "svn-remote.$repo_id.${t}-maxRev") }; | |
161 | if (defined $max_rev && ($max_rev < $base)) { | |
162 | $base = $max_rev; | |
163 | } elsif (!defined $max_rev) { | |
164 | $base = 0; | |
165 | } | |
166 | } | |
167 | ||
168 | if ($fetch) { | |
169 | foreach my $p (sort keys %$fetch) { | |
170 | my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p); | |
171 | my $lr = $gs->rev_map_max; | |
172 | if (defined $lr) { | |
173 | $base = $lr if ($lr < $base); | |
174 | } | |
175 | push @gs, $gs; | |
176 | } | |
177 | } | |
178 | ||
179 | ($base, $head) = parse_revision_argument($base, $head); | |
180 | $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs); | |
181 | } | |
182 | ||
183 | sub read_all_remotes { | |
184 | my $r = {}; | |
185 | my $use_svm_props = eval { command_oneline(qw/config --bool | |
186 | svn.useSvmProps/) }; | |
187 | $use_svm_props = $use_svm_props eq 'true' if $use_svm_props; | |
188 | my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*}; | |
189 | foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) { | |
190 | if (m!^(.+)\.fetch=$svn_refspec$!) { | |
191 | my ($remote, $local_ref, $remote_ref) = ($1, $2, $3); | |
192 | die("svn-remote.$remote: remote ref '$remote_ref' " | |
193 | . "must start with 'refs/'\n") | |
194 | unless $remote_ref =~ m{^refs/}; | |
195 | $local_ref = uri_decode($local_ref); | |
196 | $r->{$remote}->{fetch}->{$local_ref} = $remote_ref; | |
197 | $r->{$remote}->{svm} = {} if $use_svm_props; | |
198 | } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) { | |
199 | $r->{$1}->{svm} = {}; | |
200 | } elsif (m!^(.+)\.url=\s*(.*)\s*$!) { | |
93c3fcbe | 201 | $r->{$1}->{url} = canonicalize_url($2); |
29499c0b | 202 | } elsif (m!^(.+)\.pushurl=\s*(.*)\s*$!) { |
93c3fcbe | 203 | $r->{$1}->{pushurl} = canonicalize_url($2); |
29499c0b MS |
204 | } elsif (m!^(.+)\.ignore-refs=\s*(.*)\s*$!) { |
205 | $r->{$1}->{ignore_refs_regex} = $2; | |
206 | } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) { | |
207 | my ($remote, $t, $local_ref, $remote_ref) = | |
208 | ($1, $2, $3, $4); | |
209 | die("svn-remote.$remote: remote ref '$remote_ref' ($t) " | |
210 | . "must start with 'refs/'\n") | |
211 | unless $remote_ref =~ m{^refs/}; | |
212 | $local_ref = uri_decode($local_ref); | |
3d9be15f MS |
213 | |
214 | require Git::SVN::GlobSpec; | |
29499c0b MS |
215 | my $rs = { |
216 | t => $t, | |
217 | remote => $remote, | |
218 | path => Git::SVN::GlobSpec->new($local_ref, 1), | |
219 | ref => Git::SVN::GlobSpec->new($remote_ref, 0) }; | |
220 | if (length($rs->{ref}->{right}) != 0) { | |
221 | die "The '*' glob character must be the last ", | |
222 | "character of '$remote_ref'\n"; | |
223 | } | |
224 | push @{ $r->{$remote}->{$t} }, $rs; | |
225 | } | |
226 | } | |
227 | ||
228 | map { | |
229 | if (defined $r->{$_}->{svm}) { | |
230 | my $svm; | |
231 | eval { | |
232 | my $section = "svn-remote.$_"; | |
233 | $svm = { | |
234 | source => tmp_config('--get', | |
235 | "$section.svm-source"), | |
236 | replace => tmp_config('--get', | |
237 | "$section.svm-replace"), | |
238 | } | |
239 | }; | |
240 | $r->{$_}->{svm} = $svm; | |
241 | } | |
242 | } keys %$r; | |
243 | ||
244 | foreach my $remote (keys %$r) { | |
245 | foreach ( grep { defined $_ } | |
246 | map { $r->{$remote}->{$_} } qw(branches tags) ) { | |
247 | foreach my $rs ( @$_ ) { | |
248 | $rs->{ignore_refs_regex} = | |
249 | $r->{$remote}->{ignore_refs_regex}; | |
250 | } | |
251 | } | |
252 | } | |
253 | ||
254 | $r; | |
255 | } | |
256 | ||
257 | sub init_vars { | |
258 | $_gc_nr = $_gc_period = 1000; | |
259 | if (defined $_repack || defined $_repack_flags) { | |
260 | warn "Repack options are obsolete; they have no effect.\n"; | |
261 | } | |
262 | } | |
263 | ||
264 | sub verify_remotes_sanity { | |
265 | return unless -d $ENV{GIT_DIR}; | |
266 | my %seen; | |
267 | foreach (command(qw/config -l/)) { | |
268 | if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) { | |
269 | if ($seen{$1}) { | |
270 | die "Remote ref refs/remote/$1 is tracked by", | |
271 | "\n \"$_\"\nand\n \"$seen{$1}\"\n", | |
272 | "Please resolve this ambiguity in ", | |
273 | "your git configuration file before ", | |
274 | "continuing\n"; | |
275 | } | |
276 | $seen{$1} = $_; | |
277 | } | |
278 | } | |
279 | } | |
280 | ||
281 | sub find_existing_remote { | |
282 | my ($url, $remotes) = @_; | |
283 | return undef if $no_reuse_existing; | |
284 | my $existing; | |
285 | foreach my $repo_id (keys %$remotes) { | |
286 | my $u = $remotes->{$repo_id}->{url} or next; | |
287 | next if $u ne $url; | |
288 | $existing = $repo_id; | |
289 | last; | |
290 | } | |
291 | $existing; | |
292 | } | |
293 | ||
294 | sub init_remote_config { | |
295 | my ($self, $url, $no_write) = @_; | |
9c27a57b | 296 | $url = canonicalize_url($url); |
29499c0b MS |
297 | my $r = read_all_remotes(); |
298 | my $existing = find_existing_remote($url, $r); | |
299 | if ($existing) { | |
300 | unless ($no_write) { | |
301 | print STDERR "Using existing ", | |
302 | "[svn-remote \"$existing\"]\n"; | |
303 | } | |
304 | $self->{repo_id} = $existing; | |
305 | } elsif ($_minimize_url) { | |
306 | my $min_url = Git::SVN::Ra->new($url)->minimize_url; | |
307 | $existing = find_existing_remote($min_url, $r); | |
308 | if ($existing) { | |
309 | unless ($no_write) { | |
310 | print STDERR "Using existing ", | |
311 | "[svn-remote \"$existing\"]\n"; | |
312 | } | |
313 | $self->{repo_id} = $existing; | |
314 | } | |
315 | if ($min_url ne $url) { | |
316 | unless ($no_write) { | |
317 | print STDERR "Using higher level of URL: ", | |
318 | "$url => $min_url\n"; | |
319 | } | |
5578ed74 MS |
320 | my $old_path = $self->path; |
321 | $url =~ s!^\Q$min_url\E(/|$)!!; | |
ca475a61 | 322 | $url = join_paths($url, $old_path); |
5578ed74 | 323 | $self->path($url); |
29499c0b MS |
324 | $url = $min_url; |
325 | } | |
326 | } | |
327 | my $orig_url; | |
328 | if (!$existing) { | |
329 | # verify that we aren't overwriting anything: | |
330 | $orig_url = eval { | |
331 | command_oneline('config', '--get', | |
332 | "svn-remote.$self->{repo_id}.url") | |
333 | }; | |
334 | if ($orig_url && ($orig_url ne $url)) { | |
335 | die "svn-remote.$self->{repo_id}.url already set: ", | |
336 | "$orig_url\nwanted to set to: $url\n"; | |
337 | } | |
338 | } | |
339 | my ($xrepo_id, $xpath) = find_ref($self->refname); | |
340 | if (!$no_write && defined $xpath) { | |
341 | die "svn-remote.$xrepo_id.fetch already set to track ", | |
342 | "$xpath:", $self->refname, "\n"; | |
343 | } | |
344 | unless ($no_write) { | |
345 | command_noisy('config', | |
346 | "svn-remote.$self->{repo_id}.url", $url); | |
5578ed74 MS |
347 | my $path = $self->path; |
348 | $path =~ s{^/}{}; | |
349 | $path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg; | |
350 | $self->path($path); | |
29499c0b MS |
351 | command_noisy('config', '--add', |
352 | "svn-remote.$self->{repo_id}.fetch", | |
5578ed74 | 353 | $self->path.":".$self->refname); |
29499c0b | 354 | } |
06ee19e8 | 355 | $self->url($url); |
29499c0b MS |
356 | } |
357 | ||
358 | sub find_by_url { # repos_root and, path are optional | |
359 | my ($class, $full_url, $repos_root, $path) = @_; | |
360 | ||
705b49cb MS |
361 | $full_url = canonicalize_url($full_url); |
362 | ||
29499c0b MS |
363 | return undef unless defined $full_url; |
364 | remove_username($full_url); | |
365 | remove_username($repos_root) if defined $repos_root; | |
366 | my $remotes = read_all_remotes(); | |
367 | if (defined $full_url && defined $repos_root && !defined $path) { | |
368 | $path = $full_url; | |
369 | $path =~ s#^\Q$repos_root\E(?:/|$)##; | |
370 | } | |
371 | foreach my $repo_id (keys %$remotes) { | |
372 | my $u = $remotes->{$repo_id}->{url} or next; | |
373 | remove_username($u); | |
374 | next if defined $repos_root && $repos_root ne $u; | |
375 | ||
376 | my $fetch = $remotes->{$repo_id}->{fetch} || {}; | |
377 | foreach my $t (qw/branches tags/) { | |
378 | foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) { | |
379 | resolve_local_globs($u, $fetch, $globspec); | |
380 | } | |
381 | } | |
382 | my $p = $path; | |
383 | my $rwr = rewrite_root({repo_id => $repo_id}); | |
384 | my $svm = $remotes->{$repo_id}->{svm} | |
385 | if defined $remotes->{$repo_id}->{svm}; | |
386 | unless (defined $p) { | |
387 | $p = $full_url; | |
388 | my $z = $u; | |
389 | my $prefix = ''; | |
390 | if ($rwr) { | |
391 | $z = $rwr; | |
392 | remove_username($z); | |
393 | } elsif (defined $svm) { | |
394 | $z = $svm->{source}; | |
395 | $prefix = $svm->{replace}; | |
396 | $prefix =~ s#^\Q$u\E(?:/|$)##; | |
397 | $prefix =~ s#/$##; | |
398 | } | |
399 | $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next; | |
400 | } | |
705b49cb MS |
401 | |
402 | # remote fetch paths are not URI escaped. Decode ours | |
403 | # so they match | |
404 | $p = uri_decode($p); | |
405 | ||
29499c0b MS |
406 | foreach my $f (keys %$fetch) { |
407 | next if $f ne $p; | |
408 | return Git::SVN->new($fetch->{$f}, $repo_id, $f); | |
409 | } | |
410 | } | |
411 | undef; | |
412 | } | |
413 | ||
414 | sub init { | |
415 | my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_; | |
416 | my $self = _new($class, $repo_id, $ref_id, $path); | |
417 | if (defined $url) { | |
418 | $self->init_remote_config($url, $no_write); | |
419 | } | |
420 | $self; | |
421 | } | |
422 | ||
423 | sub find_ref { | |
424 | my ($ref_id) = @_; | |
425 | foreach (command(qw/config -l/)) { | |
426 | next unless m!^svn-remote\.(.+)\.fetch= | |
427 | \s*(.*?)\s*:\s*(.+?)\s*$!x; | |
428 | my ($repo_id, $path, $ref) = ($1, $2, $3); | |
429 | if ($ref eq $ref_id) { | |
430 | $path = '' if ($path =~ m#^\./?#); | |
431 | return ($repo_id, $path); | |
432 | } | |
433 | } | |
434 | (undef, undef, undef); | |
435 | } | |
436 | ||
437 | sub new { | |
438 | my ($class, $ref_id, $repo_id, $path) = @_; | |
439 | if (defined $ref_id && !defined $repo_id && !defined $path) { | |
440 | ($repo_id, $path) = find_ref($ref_id); | |
441 | if (!defined $repo_id) { | |
442 | die "Could not find a \"svn-remote.*.fetch\" key ", | |
443 | "in the repository configuration matching: ", | |
444 | "$ref_id\n"; | |
445 | } | |
446 | } | |
447 | my $self = _new($class, $repo_id, $ref_id, $path); | |
5578ed74 | 448 | if (!defined $self->path || !length $self->path) { |
29499c0b MS |
449 | my $fetch = command_oneline('config', '--get', |
450 | "svn-remote.$repo_id.fetch", | |
451 | ":$ref_id\$") or | |
452 | die "Failed to read \"svn-remote.$repo_id.fetch\" ", | |
453 | "\":$ref_id\$\" in config\n"; | |
5578ed74 MS |
454 | my($path) = split(/\s*:\s*/, $fetch); |
455 | $self->path($path); | |
29499c0b | 456 | } |
5578ed74 MS |
457 | { |
458 | my $path = $self->path; | |
5578ed74 MS |
459 | $path =~ s{\A/}{}; |
460 | $path =~ s{/\z}{}; | |
461 | $self->path($path); | |
29499c0b | 462 | } |
06ee19e8 MS |
463 | my $url = command_oneline('config', '--get', |
464 | "svn-remote.$repo_id.url") or | |
29499c0b | 465 | die "Failed to read \"svn-remote.$repo_id.url\" in config\n"; |
06ee19e8 | 466 | $self->url($url); |
29499c0b MS |
467 | $self->{pushurl} = eval { command_oneline('config', '--get', |
468 | "svn-remote.$repo_id.pushurl") }; | |
469 | $self->rebuild; | |
470 | $self; | |
471 | } | |
472 | ||
473 | sub refname { | |
474 | my ($refname) = $_[0]->{ref_id} ; | |
475 | ||
476 | # It cannot end with a slash /, we'll throw up on this because | |
477 | # SVN can't have directories with a slash in their name, either: | |
478 | if ($refname =~ m{/$}) { | |
235e8d59 JL |
479 | die "ref: '$refname' ends with a trailing slash; this is ", |
480 | "not permitted by git or Subversion\n"; | |
29499c0b MS |
481 | } |
482 | ||
483 | # It cannot have ASCII control character space, tilde ~, caret ^, | |
484 | # colon :, question-mark ?, asterisk *, space, or open bracket [ | |
485 | # anywhere. | |
486 | # | |
487 | # Additionally, % must be escaped because it is used for escaping | |
488 | # and we want our escaped refname to be reversible | |
1b67bef2 | 489 | $refname =~ s{([ \%~\^:\?\*\[\t])}{sprintf('%%%02X',ord($1))}eg; |
29499c0b MS |
490 | |
491 | # no slash-separated component can begin with a dot . | |
492 | # /.* becomes /%2E* | |
493 | $refname =~ s{/\.}{/%2E}g; | |
494 | ||
495 | # It cannot have two consecutive dots .. anywhere | |
496 | # .. becomes %2E%2E | |
497 | $refname =~ s{\.\.}{%2E%2E}g; | |
498 | ||
499 | # trailing dots and .lock are not allowed | |
500 | # .$ becomes %2E and .lock becomes %2Elock | |
501 | $refname =~ s{\.(?=$|lock$)}{%2E}; | |
502 | ||
503 | # the sequence @{ is used to access the reflog | |
504 | # @{ becomes %40{ | |
505 | $refname =~ s{\@\{}{%40\{}g; | |
506 | ||
507 | return $refname; | |
508 | } | |
509 | ||
510 | sub desanitize_refname { | |
511 | my ($refname) = @_; | |
512 | $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg; | |
513 | return $refname; | |
514 | } | |
515 | ||
516 | sub svm_uuid { | |
517 | my ($self) = @_; | |
518 | return $self->{svm}->{uuid} if $self->svm; | |
519 | $self->ra; | |
520 | unless ($self->{svm}) { | |
521 | die "SVM UUID not cached, and reading remotely failed\n"; | |
522 | } | |
523 | $self->{svm}->{uuid}; | |
524 | } | |
525 | ||
526 | sub svm { | |
527 | my ($self) = @_; | |
528 | return $self->{svm} if $self->{svm}; | |
529 | my $svm; | |
530 | # see if we have it in our config, first: | |
531 | eval { | |
532 | my $section = "svn-remote.$self->{repo_id}"; | |
533 | $svm = { | |
534 | source => tmp_config('--get', "$section.svm-source"), | |
535 | uuid => tmp_config('--get', "$section.svm-uuid"), | |
536 | replace => tmp_config('--get', "$section.svm-replace"), | |
537 | } | |
538 | }; | |
539 | if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) { | |
540 | $self->{svm} = $svm; | |
541 | } | |
542 | $self->{svm}; | |
543 | } | |
544 | ||
545 | sub _set_svm_vars { | |
546 | my ($self, $ra) = @_; | |
547 | return $ra if $self->svm; | |
548 | ||
549 | my @err = ( "useSvmProps set, but failed to read SVM properties\n", | |
550 | "(svm:source, svm:uuid) ", | |
551 | "from the following URLs:\n" ); | |
552 | sub read_svm_props { | |
553 | my ($self, $ra, $path, $r) = @_; | |
554 | my $props = ($ra->get_dir($path, $r))[2]; | |
555 | my $src = $props->{'svm:source'}; | |
556 | my $uuid = $props->{'svm:uuid'}; | |
557 | return undef if (!$src || !$uuid); | |
558 | ||
559 | chomp($src, $uuid); | |
560 | ||
561 | $uuid =~ m{^[0-9a-f\-]{30,}$}i | |
562 | or die "doesn't look right - svm:uuid is '$uuid'\n"; | |
563 | ||
564 | # the '!' is used to mark the repos_root!/relative/path | |
565 | $src =~ s{/?!/?}{/}; | |
566 | $src =~ s{/+$}{}; # no trailing slashes please | |
567 | # username is of no interest | |
568 | $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1}; | |
569 | ||
d2fd119c | 570 | my $replace = add_path_to_url($ra->url, $path); |
29499c0b MS |
571 | |
572 | my $section = "svn-remote.$self->{repo_id}"; | |
573 | tmp_config("$section.svm-source", $src); | |
574 | tmp_config("$section.svm-replace", $replace); | |
575 | tmp_config("$section.svm-uuid", $uuid); | |
576 | $self->{svm} = { | |
577 | source => $src, | |
578 | uuid => $uuid, | |
579 | replace => $replace | |
580 | }; | |
581 | } | |
582 | ||
583 | my $r = $ra->get_latest_revnum; | |
5578ed74 | 584 | my $path = $self->path; |
29499c0b MS |
585 | my %tried; |
586 | while (length $path) { | |
d2fd119c | 587 | my $try = add_path_to_url($self->url, $path); |
06ee19e8 | 588 | unless ($tried{$try}) { |
29499c0b | 589 | return $ra if $self->read_svm_props($ra, $path, $r); |
06ee19e8 | 590 | $tried{$try} = 1; |
29499c0b MS |
591 | } |
592 | $path =~ s#/?[^/]+$##; | |
593 | } | |
594 | die "Path: '$path' should be ''\n" if $path ne ''; | |
595 | return $ra if $self->read_svm_props($ra, $path, $r); | |
d2fd119c | 596 | $tried{ add_path_to_url($self->url, $path) } = 1; |
29499c0b | 597 | |
06ee19e8 | 598 | if ($ra->{repos_root} eq $self->url) { |
29499c0b MS |
599 | die @err, (map { " $_\n" } keys %tried), "\n"; |
600 | } | |
601 | ||
602 | # nope, make sure we're connected to the repository root: | |
603 | my $ok; | |
604 | my @tried_b; | |
605 | $path = $ra->{svn_path}; | |
606 | $ra = Git::SVN::Ra->new($ra->{repos_root}); | |
607 | while (length $path) { | |
d2fd119c | 608 | my $try = add_path_to_url($ra->url, $path); |
b1ea6c38 | 609 | unless ($tried{$try}) { |
29499c0b MS |
610 | $ok = $self->read_svm_props($ra, $path, $r); |
611 | last if $ok; | |
b1ea6c38 | 612 | $tried{$try} = 1; |
29499c0b MS |
613 | } |
614 | $path =~ s#/?[^/]+$##; | |
615 | } | |
616 | die "Path: '$path' should be ''\n" if $path ne ''; | |
617 | $ok ||= $self->read_svm_props($ra, $path, $r); | |
d2fd119c | 618 | $tried{ add_path_to_url($ra->url, $path) } = 1; |
29499c0b MS |
619 | if (!$ok) { |
620 | die @err, (map { " $_\n" } keys %tried), "\n"; | |
621 | } | |
06ee19e8 | 622 | Git::SVN::Ra->new($self->url); |
29499c0b MS |
623 | } |
624 | ||
625 | sub svnsync { | |
626 | my ($self) = @_; | |
627 | return $self->{svnsync} if $self->{svnsync}; | |
628 | ||
629 | if ($self->no_metadata) { | |
630 | die "Can't have both 'noMetadata' and ", | |
631 | "'useSvnsyncProps' options set!\n"; | |
632 | } | |
633 | if ($self->rewrite_root) { | |
634 | die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ", | |
635 | "options set!\n"; | |
636 | } | |
637 | if ($self->rewrite_uuid) { | |
638 | die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ", | |
639 | "options set!\n"; | |
640 | } | |
641 | ||
642 | my $svnsync; | |
643 | # see if we have it in our config, first: | |
644 | eval { | |
645 | my $section = "svn-remote.$self->{repo_id}"; | |
646 | ||
647 | my $url = tmp_config('--get', "$section.svnsync-url"); | |
648 | ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or | |
649 | die "doesn't look right - svn:sync-from-url is '$url'\n"; | |
650 | ||
651 | my $uuid = tmp_config('--get', "$section.svnsync-uuid"); | |
652 | ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or | |
653 | die "doesn't look right - svn:sync-from-uuid is '$uuid'\n"; | |
654 | ||
655 | $svnsync = { url => $url, uuid => $uuid } | |
656 | }; | |
657 | if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) { | |
658 | return $self->{svnsync} = $svnsync; | |
659 | } | |
660 | ||
661 | my $err = "useSvnsyncProps set, but failed to read " . | |
662 | "svnsync property: svn:sync-from-"; | |
663 | my $rp = $self->ra->rev_proplist(0); | |
664 | ||
665 | my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n"; | |
666 | ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or | |
667 | die "doesn't look right - svn:sync-from-url is '$url'\n"; | |
668 | ||
669 | my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n"; | |
670 | ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or | |
671 | die "doesn't look right - svn:sync-from-uuid is '$uuid'\n"; | |
672 | ||
673 | my $section = "svn-remote.$self->{repo_id}"; | |
674 | tmp_config('--add', "$section.svnsync-uuid", $uuid); | |
675 | tmp_config('--add', "$section.svnsync-url", $url); | |
676 | return $self->{svnsync} = { url => $url, uuid => $uuid }; | |
677 | } | |
678 | ||
679 | # this allows us to memoize our SVN::Ra UUID locally and avoid a | |
680 | # remote lookup (useful for 'git svn log'). | |
681 | sub ra_uuid { | |
682 | my ($self) = @_; | |
683 | unless ($self->{ra_uuid}) { | |
684 | my $key = "svn-remote.$self->{repo_id}.uuid"; | |
685 | my $uuid = eval { tmp_config('--get', $key) }; | |
686 | if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) { | |
687 | $self->{ra_uuid} = $uuid; | |
688 | } else { | |
06ee19e8 | 689 | die "ra_uuid called without URL\n" unless $self->url; |
29499c0b MS |
690 | $self->{ra_uuid} = $self->ra->get_uuid; |
691 | tmp_config('--add', $key, $self->{ra_uuid}); | |
692 | } | |
693 | } | |
694 | $self->{ra_uuid}; | |
695 | } | |
696 | ||
697 | sub _set_repos_root { | |
698 | my ($self, $repos_root) = @_; | |
699 | my $k = "svn-remote.$self->{repo_id}.reposRoot"; | |
700 | $repos_root ||= $self->ra->{repos_root}; | |
701 | tmp_config($k, $repos_root); | |
702 | $repos_root; | |
703 | } | |
704 | ||
705 | sub repos_root { | |
706 | my ($self) = @_; | |
707 | my $k = "svn-remote.$self->{repo_id}.reposRoot"; | |
708 | eval { tmp_config('--get', $k) } || $self->_set_repos_root; | |
709 | } | |
710 | ||
711 | sub ra { | |
712 | my ($self) = shift; | |
06ee19e8 | 713 | my $ra = Git::SVN::Ra->new($self->url); |
29499c0b MS |
714 | $self->_set_repos_root($ra->{repos_root}); |
715 | if ($self->use_svm_props && !$self->{svm}) { | |
716 | if ($self->no_metadata) { | |
717 | die "Can't have both 'noMetadata' and ", | |
718 | "'useSvmProps' options set!\n"; | |
719 | } elsif ($self->use_svnsync_props) { | |
720 | die "Can't have both 'useSvnsyncProps' and ", | |
721 | "'useSvmProps' options set!\n"; | |
722 | } | |
723 | $ra = $self->_set_svm_vars($ra); | |
724 | $self->{-want_revprops} = 1; | |
725 | } | |
726 | $ra; | |
727 | } | |
728 | ||
729 | # prop_walk(PATH, REV, SUB) | |
730 | # ------------------------- | |
731 | # Recursively traverse PATH at revision REV and invoke SUB for each | |
732 | # directory that contains a SVN property. SUB will be invoked as | |
733 | # follows: &SUB(gs, path, props); where `gs' is this instance of | |
734 | # Git::SVN, `path' the path to the directory where the properties | |
735 | # `props' were found. The `path' will be relative to point of checkout, | |
736 | # that is, if url://repo/trunk is the current Git branch, and that | |
737 | # directory contains a sub-directory `d', SUB will be invoked with `/d/' | |
738 | # as `path' (note the trailing `/'). | |
739 | sub prop_walk { | |
740 | my ($self, $path, $rev, $sub) = @_; | |
741 | ||
742 | $path =~ s#^/##; | |
743 | my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev); | |
744 | $path =~ s#^/*#/#g; | |
745 | my $p = $path; | |
746 | # Strip the irrelevant part of the path. | |
5578ed74 | 747 | $p =~ s#^/+\Q@{[$self->path]}\E(/|$)#/#; |
29499c0b MS |
748 | # Ensure the path is terminated by a `/'. |
749 | $p =~ s#/*$#/#; | |
750 | ||
751 | # The properties contain all the internal SVN stuff nobody | |
752 | # (usually) cares about. | |
753 | my $interesting_props = 0; | |
754 | foreach (keys %{$props}) { | |
755 | # If it doesn't start with `svn:', it must be a | |
756 | # user-defined property. | |
757 | ++$interesting_props and next if $_ !~ /^svn:/; | |
758 | # FIXME: Fragile, if SVN adds new public properties, | |
759 | # this needs to be updated. | |
760 | ++$interesting_props if /^svn:(?:ignore|keywords|executable | |
761 | |eol-style|mime-type | |
762 | |externals|needs-lock)$/x; | |
763 | } | |
764 | &$sub($self, $p, $props) if $interesting_props; | |
765 | ||
766 | foreach (sort keys %$dirent) { | |
767 | next if $dirent->{$_}->{kind} != $SVN::Node::dir; | |
5578ed74 | 768 | $self->prop_walk($self->path . $p . $_, $rev, $sub); |
29499c0b MS |
769 | } |
770 | } | |
771 | ||
772 | sub last_rev { ($_[0]->last_rev_commit)[0] } | |
773 | sub last_commit { ($_[0]->last_rev_commit)[1] } | |
774 | ||
775 | # returns the newest SVN revision number and newest commit SHA1 | |
776 | sub last_rev_commit { | |
777 | my ($self) = @_; | |
778 | if (defined $self->{last_rev} && defined $self->{last_commit}) { | |
779 | return ($self->{last_rev}, $self->{last_commit}); | |
780 | } | |
781 | my $c = ::verify_ref($self->refname.'^0'); | |
782 | if ($c && !$self->use_svm_props && !$self->no_metadata) { | |
783 | my $rev = (::cmt_metadata($c))[1]; | |
784 | if (defined $rev) { | |
785 | ($self->{last_rev}, $self->{last_commit}) = ($rev, $c); | |
786 | return ($rev, $c); | |
787 | } | |
788 | } | |
789 | my $map_path = $self->map_path; | |
790 | unless (-e $map_path) { | |
791 | ($self->{last_rev}, $self->{last_commit}) = (undef, undef); | |
792 | return (undef, undef); | |
793 | } | |
794 | my ($rev, $commit) = $self->rev_map_max(1); | |
795 | ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit); | |
796 | return ($rev, $commit); | |
797 | } | |
798 | ||
799 | sub get_fetch_range { | |
800 | my ($self, $min, $max) = @_; | |
801 | $max ||= $self->ra->get_latest_revnum; | |
802 | $min ||= $self->rev_map_max; | |
803 | (++$min, $max); | |
804 | } | |
805 | ||
806 | sub tmp_config { | |
807 | my (@args) = @_; | |
808 | my $old_def_config = "$ENV{GIT_DIR}/svn/config"; | |
809 | my $config = "$ENV{GIT_DIR}/svn/.metadata"; | |
810 | if (! -f $config && -f $old_def_config) { | |
811 | rename $old_def_config, $config or | |
812 | die "Failed rename $old_def_config => $config: $!\n"; | |
813 | } | |
814 | my $old_config = $ENV{GIT_CONFIG}; | |
815 | $ENV{GIT_CONFIG} = $config; | |
816 | $@ = undef; | |
817 | my @ret = eval { | |
818 | unless (-f $config) { | |
819 | mkfile($config); | |
820 | open my $fh, '>', $config or | |
821 | die "Can't open $config: $!\n"; | |
822 | print $fh "; This file is used internally by ", | |
823 | "git-svn\n" or die | |
824 | "Couldn't write to $config: $!\n"; | |
825 | print $fh "; You should not have to edit it\n" or | |
826 | die "Couldn't write to $config: $!\n"; | |
827 | close $fh or die "Couldn't close $config: $!\n"; | |
828 | } | |
829 | command('config', @args); | |
830 | }; | |
831 | my $err = $@; | |
832 | if (defined $old_config) { | |
833 | $ENV{GIT_CONFIG} = $old_config; | |
834 | } else { | |
835 | delete $ENV{GIT_CONFIG}; | |
836 | } | |
837 | die $err if $err; | |
838 | wantarray ? @ret : $ret[0]; | |
839 | } | |
840 | ||
841 | sub tmp_index_do { | |
842 | my ($self, $sub) = @_; | |
843 | my $old_index = $ENV{GIT_INDEX_FILE}; | |
844 | $ENV{GIT_INDEX_FILE} = $self->{index}; | |
845 | $@ = undef; | |
846 | my @ret = eval { | |
847 | my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#); | |
848 | mkpath([$dir]) unless -d $dir; | |
849 | &$sub; | |
850 | }; | |
851 | my $err = $@; | |
852 | if (defined $old_index) { | |
853 | $ENV{GIT_INDEX_FILE} = $old_index; | |
854 | } else { | |
855 | delete $ENV{GIT_INDEX_FILE}; | |
856 | } | |
857 | die $err if $err; | |
858 | wantarray ? @ret : $ret[0]; | |
859 | } | |
860 | ||
861 | sub assert_index_clean { | |
862 | my ($self, $treeish) = @_; | |
863 | ||
864 | $self->tmp_index_do(sub { | |
865 | command_noisy('read-tree', $treeish) unless -e $self->{index}; | |
866 | my $x = command_oneline('write-tree'); | |
867 | my ($y) = (command(qw/cat-file commit/, $treeish) =~ | |
868 | /^tree ($::sha1)/mo); | |
869 | return if $y eq $x; | |
870 | ||
871 | warn "Index mismatch: $y != $x\nrereading $treeish\n"; | |
872 | unlink $self->{index} or die "unlink $self->{index}: $!\n"; | |
873 | command_noisy('read-tree', $treeish); | |
874 | $x = command_oneline('write-tree'); | |
875 | if ($y ne $x) { | |
876 | fatal "trees ($treeish) $y != $x\n", | |
877 | "Something is seriously wrong..."; | |
878 | } | |
879 | }); | |
880 | } | |
881 | ||
882 | sub get_commit_parents { | |
883 | my ($self, $log_entry) = @_; | |
884 | my (%seen, @ret, @tmp); | |
885 | # legacy support for 'set-tree'; this is only used by set_tree_cb: | |
886 | if (my $ip = $self->{inject_parents}) { | |
887 | if (my $commit = delete $ip->{$log_entry->{revision}}) { | |
888 | push @tmp, $commit; | |
889 | } | |
890 | } | |
891 | if (my $cur = ::verify_ref($self->refname.'^0')) { | |
892 | push @tmp, $cur; | |
893 | } | |
894 | if (my $ipd = $self->{inject_parents_dcommit}) { | |
895 | if (my $commit = delete $ipd->{$log_entry->{revision}}) { | |
896 | push @tmp, @$commit; | |
897 | } | |
898 | } | |
899 | push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp); | |
900 | while (my $p = shift @tmp) { | |
901 | next if $seen{$p}; | |
902 | $seen{$p} = 1; | |
903 | push @ret, $p; | |
904 | } | |
905 | @ret; | |
906 | } | |
907 | ||
908 | sub rewrite_root { | |
909 | my ($self) = @_; | |
910 | return $self->{-rewrite_root} if exists $self->{-rewrite_root}; | |
911 | my $k = "svn-remote.$self->{repo_id}.rewriteRoot"; | |
912 | my $rwr = eval { command_oneline(qw/config --get/, $k) }; | |
913 | if ($rwr) { | |
914 | $rwr =~ s#/+$##; | |
915 | if ($rwr !~ m#^[a-z\+]+://#) { | |
916 | die "$rwr is not a valid URL (key: $k)\n"; | |
917 | } | |
918 | } | |
919 | $self->{-rewrite_root} = $rwr; | |
920 | } | |
921 | ||
922 | sub rewrite_uuid { | |
923 | my ($self) = @_; | |
924 | return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid}; | |
925 | my $k = "svn-remote.$self->{repo_id}.rewriteUUID"; | |
926 | my $rwid = eval { command_oneline(qw/config --get/, $k) }; | |
927 | if ($rwid) { | |
928 | $rwid =~ s#/+$##; | |
929 | if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) { | |
930 | die "$rwid is not a valid UUID (key: $k)\n"; | |
931 | } | |
932 | } | |
933 | $self->{-rewrite_uuid} = $rwid; | |
934 | } | |
935 | ||
936 | sub metadata_url { | |
937 | my ($self) = @_; | |
d2fd119c | 938 | my $url = $self->rewrite_root || $self->url; |
705b49cb | 939 | return canonicalize_url( add_path_to_url( $url, $self->path ) ); |
29499c0b MS |
940 | } |
941 | ||
942 | sub full_url { | |
943 | my ($self) = @_; | |
705b49cb | 944 | return canonicalize_url( add_path_to_url( $self->url, $self->path ) ); |
29499c0b MS |
945 | } |
946 | ||
947 | sub full_pushurl { | |
948 | my ($self) = @_; | |
949 | if ($self->{pushurl}) { | |
705b49cb | 950 | return canonicalize_url( add_path_to_url( $self->{pushurl}, $self->path ) ); |
29499c0b MS |
951 | } else { |
952 | return $self->full_url; | |
953 | } | |
954 | } | |
955 | ||
956 | sub set_commit_header_env { | |
957 | my ($log_entry) = @_; | |
958 | my %env; | |
959 | foreach my $ned (qw/NAME EMAIL DATE/) { | |
960 | foreach my $ac (qw/AUTHOR COMMITTER/) { | |
961 | $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"}; | |
962 | } | |
963 | } | |
964 | ||
965 | $ENV{GIT_AUTHOR_NAME} = $log_entry->{name}; | |
966 | $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email}; | |
967 | $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date}; | |
968 | ||
969 | $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name}) | |
970 | ? $log_entry->{commit_name} | |
971 | : $log_entry->{name}; | |
972 | $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email}) | |
973 | ? $log_entry->{commit_email} | |
974 | : $log_entry->{email}; | |
975 | \%env; | |
976 | } | |
977 | ||
978 | sub restore_commit_header_env { | |
979 | my ($env) = @_; | |
980 | foreach my $ned (qw/NAME EMAIL DATE/) { | |
981 | foreach my $ac (qw/AUTHOR COMMITTER/) { | |
982 | my $k = "GIT_${ac}_${ned}"; | |
983 | if (defined $env->{$k}) { | |
984 | $ENV{$k} = $env->{$k}; | |
985 | } else { | |
986 | delete $ENV{$k}; | |
987 | } | |
988 | } | |
989 | } | |
990 | } | |
991 | ||
992 | sub gc { | |
993 | command_noisy('gc', '--auto'); | |
994 | }; | |
995 | ||
996 | sub do_git_commit { | |
997 | my ($self, $log_entry) = @_; | |
998 | my $lr = $self->last_rev; | |
999 | if (defined $lr && $lr >= $log_entry->{revision}) { | |
1000 | die "Last fetched revision of ", $self->refname, | |
1001 | " was r$lr, but we are about to fetch: ", | |
1002 | "r$log_entry->{revision}!\n"; | |
1003 | } | |
1004 | if (my $c = $self->rev_map_get($log_entry->{revision})) { | |
1005 | croak "$log_entry->{revision} = $c already exists! ", | |
1006 | "Why are we refetching it?\n"; | |
1007 | } | |
1008 | my $old_env = set_commit_header_env($log_entry); | |
1009 | my $tree = $log_entry->{tree}; | |
1010 | if (!defined $tree) { | |
1011 | $tree = $self->tmp_index_do(sub { | |
1012 | command_oneline('write-tree') }); | |
1013 | } | |
1014 | die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o; | |
1015 | ||
1016 | my @exec = ('git', 'commit-tree', $tree); | |
1017 | foreach ($self->get_commit_parents($log_entry)) { | |
1018 | push @exec, '-p', $_; | |
1019 | } | |
1020 | defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec)) | |
1021 | or croak $!; | |
1022 | binmode $msg_fh; | |
1023 | ||
1024 | # we always get UTF-8 from SVN, but we may want our commits in | |
1025 | # a different encoding. | |
1026 | if (my $enc = Git::config('i18n.commitencoding')) { | |
1027 | require Encode; | |
1028 | Encode::from_to($log_entry->{log}, 'UTF-8', $enc); | |
1029 | } | |
1030 | print $msg_fh $log_entry->{log} or croak $!; | |
1031 | restore_commit_header_env($old_env); | |
1032 | unless ($self->no_metadata) { | |
1033 | print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n" | |
1034 | or croak $!; | |
1035 | } | |
1036 | $msg_fh->flush == 0 or croak $!; | |
1037 | close $msg_fh or croak $!; | |
1038 | chomp(my $commit = do { local $/; <$out_fh> }); | |
1039 | close $out_fh or croak $!; | |
1040 | waitpid $pid, 0; | |
1041 | croak $? if $?; | |
1042 | if ($commit !~ /^$::sha1$/o) { | |
1043 | die "Failed to commit, invalid sha1: $commit\n"; | |
1044 | } | |
1045 | ||
1046 | $self->rev_map_set($log_entry->{revision}, $commit, 1); | |
1047 | ||
1048 | $self->{last_rev} = $log_entry->{revision}; | |
1049 | $self->{last_commit} = $commit; | |
1050 | print "r$log_entry->{revision}" unless $::_q > 1; | |
1051 | if (defined $log_entry->{svm_revision}) { | |
1052 | print " (\@$log_entry->{svm_revision})" unless $::_q > 1; | |
1053 | $self->rev_map_set($log_entry->{svm_revision}, $commit, | |
1054 | 0, $self->svm_uuid); | |
1055 | } | |
1056 | print " = $commit ($self->{ref_id})\n" unless $::_q > 1; | |
1057 | if (--$_gc_nr == 0) { | |
1058 | $_gc_nr = $_gc_period; | |
1059 | gc(); | |
1060 | } | |
1061 | return $commit; | |
1062 | } | |
1063 | ||
1064 | sub match_paths { | |
1065 | my ($self, $paths, $r) = @_; | |
5578ed74 MS |
1066 | return 1 if $self->path eq ''; |
1067 | if (my $path = $paths->{"/".$self->path}) { | |
29499c0b MS |
1068 | return ($path->{action} eq 'D') ? 0 : 1; |
1069 | } | |
5578ed74 | 1070 | $self->{path_regex} ||= qr{^/\Q@{[$self->path]}\E/}; |
29499c0b MS |
1071 | if (grep /$self->{path_regex}/, keys %$paths) { |
1072 | return 1; | |
1073 | } | |
1074 | my $c = ''; | |
5578ed74 | 1075 | foreach (split m#/#, $self->path) { |
29499c0b MS |
1076 | $c .= "/$_"; |
1077 | next unless ($paths->{$c} && | |
1078 | ($paths->{$c}->{action} =~ /^[AR]$/)); | |
5578ed74 | 1079 | if ($self->ra->check_path($self->path, $r) == |
29499c0b MS |
1080 | $SVN::Node::dir) { |
1081 | return 1; | |
1082 | } | |
1083 | } | |
1084 | return 0; | |
1085 | } | |
1086 | ||
1087 | sub find_parent_branch { | |
1088 | my ($self, $paths, $rev) = @_; | |
1089 | return undef unless $self->follow_parent; | |
1090 | unless (defined $paths) { | |
1091 | my $err_handler = $SVN::Error::handler; | |
1092 | $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs; | |
5578ed74 | 1093 | $self->ra->get_log([$self->path], $rev, $rev, 0, 1, 1, |
29499c0b MS |
1094 | sub { $paths = $_[0] }); |
1095 | $SVN::Error::handler = $err_handler; | |
1096 | } | |
1097 | return undef unless defined $paths; | |
1098 | ||
1099 | # look for a parent from another branch: | |
5578ed74 | 1100 | my @b_path_components = split m#/#, $self->path; |
29499c0b MS |
1101 | my @a_path_components; |
1102 | my $i; | |
1103 | while (@b_path_components) { | |
1104 | $i = $paths->{'/'.join('/', @b_path_components)}; | |
1105 | last if $i && defined $i->{copyfrom_path}; | |
1106 | unshift(@a_path_components, pop(@b_path_components)); | |
1107 | } | |
1108 | return undef unless defined $i && defined $i->{copyfrom_path}; | |
1109 | my $branch_from = $i->{copyfrom_path}; | |
1110 | if (@a_path_components) { | |
1111 | print STDERR "branch_from: $branch_from => "; | |
1112 | $branch_from .= '/'.join('/', @a_path_components); | |
1113 | print STDERR $branch_from, "\n"; | |
1114 | } | |
1115 | my $r = $i->{copyfrom_rev}; | |
1116 | my $repos_root = $self->ra->{repos_root}; | |
b1ea6c38 | 1117 | my $url = $self->ra->url; |
705b49cb | 1118 | my $new_url = canonicalize_url( add_path_to_url( $url, $branch_from ) ); |
29499c0b MS |
1119 | print STDERR "Found possible branch point: ", |
1120 | "$new_url => ", $self->full_url, ", $r\n" | |
1121 | unless $::_q > 1; | |
1122 | $branch_from =~ s#^/##; | |
1123 | my $gs = $self->other_gs($new_url, $url, | |
1124 | $branch_from, $r, $self->{ref_id}); | |
1125 | my ($r0, $parent) = $gs->find_rev_before($r, 1); | |
1126 | { | |
1127 | my ($base, $head); | |
1128 | if (!defined $r0 || !defined $parent) { | |
1129 | ($base, $head) = parse_revision_argument(0, $r); | |
1130 | } else { | |
1131 | if ($r0 < $r) { | |
6a8d999e | 1132 | $gs->ra->get_log([$gs->path], $r0 + 1, $r, 1, |
29499c0b MS |
1133 | 0, 1, sub { $base = $_[1] - 1 }); |
1134 | } | |
1135 | } | |
1136 | if (defined $base && $base <= $r) { | |
1137 | $gs->fetch($base, $r); | |
1138 | } | |
1139 | ($r0, $parent) = $gs->find_rev_before($r, 1); | |
1140 | } | |
1141 | if (defined $r0 && defined $parent) { | |
1142 | print STDERR "Found branch parent: ($self->{ref_id}) $parent\n" | |
1143 | unless $::_q > 1; | |
1144 | my $ed; | |
1145 | if ($self->ra->can_do_switch) { | |
1146 | $self->assert_index_clean($parent); | |
1147 | print STDERR "Following parent with do_switch\n" | |
1148 | unless $::_q > 1; | |
1149 | # do_switch works with svn/trunk >= r22312, but that | |
1150 | # is not included with SVN 1.4.3 (the latest version | |
1151 | # at the moment), so we can't rely on it | |
1152 | $self->{last_rev} = $r0; | |
1153 | $self->{last_commit} = $parent; | |
6a8d999e | 1154 | $ed = Git::SVN::Fetcher->new($self, $gs->path); |
29499c0b MS |
1155 | $gs->ra->gs_do_switch($r0, $rev, $gs, |
1156 | $self->full_url, $ed) | |
1157 | or die "SVN connection failed somewhere...\n"; | |
1158 | } elsif ($self->ra->trees_match($new_url, $r0, | |
1159 | $self->full_url, $rev)) { | |
1160 | print STDERR "Trees match:\n", | |
1161 | " $new_url\@$r0\n", | |
1162 | " ${\$self->full_url}\@$rev\n", | |
1163 | "Following parent with no changes\n" | |
1164 | unless $::_q > 1; | |
1165 | $self->tmp_index_do(sub { | |
1166 | command_noisy('read-tree', $parent); | |
1167 | }); | |
1168 | $self->{last_commit} = $parent; | |
1169 | } else { | |
1170 | print STDERR "Following parent with do_update\n" | |
1171 | unless $::_q > 1; | |
1172 | $ed = Git::SVN::Fetcher->new($self); | |
1173 | $self->ra->gs_do_update($rev, $rev, $self, $ed) | |
1174 | or die "SVN connection failed somewhere...\n"; | |
1175 | } | |
1176 | print STDERR "Successfully followed parent\n" unless $::_q > 1; | |
abfef3bb | 1177 | return $self->make_log_entry($rev, [$parent], $ed, $r0, $branch_from); |
29499c0b MS |
1178 | } |
1179 | return undef; | |
1180 | } | |
1181 | ||
1182 | sub do_fetch { | |
1183 | my ($self, $paths, $rev) = @_; | |
1184 | my $ed; | |
1185 | my ($last_rev, @parents); | |
1186 | if (my $lc = $self->last_commit) { | |
1187 | # we can have a branch that was deleted, then re-added | |
1188 | # under the same name but copied from another path, in | |
1189 | # which case we'll have multiple parents (we don't | |
01689909 | 1190 | # want to break the original ref or lose copypath info): |
29499c0b MS |
1191 | if (my $log_entry = $self->find_parent_branch($paths, $rev)) { |
1192 | push @{$log_entry->{parents}}, $lc; | |
1193 | return $log_entry; | |
1194 | } | |
1195 | $ed = Git::SVN::Fetcher->new($self); | |
1196 | $last_rev = $self->{last_rev}; | |
1197 | $ed->{c} = $lc; | |
1198 | @parents = ($lc); | |
1199 | } else { | |
1200 | $last_rev = $rev; | |
1201 | if (my $log_entry = $self->find_parent_branch($paths, $rev)) { | |
1202 | return $log_entry; | |
1203 | } | |
1204 | $ed = Git::SVN::Fetcher->new($self); | |
1205 | } | |
1206 | unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) { | |
1207 | die "SVN connection failed somewhere...\n"; | |
1208 | } | |
9ee13a93 | 1209 | $self->make_log_entry($rev, \@parents, $ed, $last_rev, $self->path); |
29499c0b MS |
1210 | } |
1211 | ||
1212 | sub mkemptydirs { | |
1213 | my ($self, $r) = @_; | |
1214 | ||
82625748 DG |
1215 | # add/remove/collect a paths table |
1216 | # | |
1217 | # Paths are split into a tree of nodes, stored as a hash of hashes. | |
1218 | # | |
1219 | # Each node contains a 'path' entry for the path (if any) associated | |
1220 | # with that node and a 'children' entry for any nodes under that | |
1221 | # location. | |
1222 | # | |
1223 | # Removing a path requires a hash lookup for each component then | |
1224 | # dropping that node (and anything under it), which is substantially | |
1225 | # faster than a grep slice into a single hash of paths for large | |
1226 | # numbers of paths. | |
1227 | # | |
1228 | # For a large (200K) number of empty_dir directives this reduces | |
1229 | # scanning time to 3 seconds vs 10 minutes for grep+delete on a single | |
1230 | # hash of paths. | |
1231 | sub add_path { | |
1232 | my ($paths_table, $path) = @_; | |
1233 | my $node_ref; | |
1234 | ||
1235 | foreach my $x (split('/', $path)) { | |
1236 | if (!exists($paths_table->{$x})) { | |
1237 | $paths_table->{$x} = { children => {} }; | |
1238 | } | |
1239 | ||
1240 | $node_ref = $paths_table->{$x}; | |
1241 | $paths_table = $paths_table->{$x}->{children}; | |
1242 | } | |
1243 | ||
1244 | $node_ref->{path} = $path; | |
1245 | } | |
1246 | ||
1247 | sub remove_path { | |
1248 | my ($paths_table, $path) = @_; | |
1249 | my $nodes_ref; | |
1250 | my $node_name; | |
1251 | ||
1252 | foreach my $x (split('/', $path)) { | |
1253 | if (!exists($paths_table->{$x})) { | |
1254 | return; | |
1255 | } | |
1256 | ||
1257 | $nodes_ref = $paths_table; | |
1258 | $node_name = $x; | |
1259 | ||
1260 | $paths_table = $paths_table->{$x}->{children}; | |
1261 | } | |
1262 | ||
1263 | delete($nodes_ref->{$node_name}); | |
1264 | } | |
1265 | ||
1266 | sub collect_paths { | |
1267 | my ($paths_table, $paths_ref) = @_; | |
1268 | ||
1269 | foreach my $v (values %$paths_table) { | |
1270 | my $p = $v->{path}; | |
1271 | my $c = $v->{children}; | |
1272 | ||
1273 | collect_paths($c, $paths_ref); | |
1274 | ||
1275 | if (defined($p)) { | |
1276 | push(@$paths_ref, $p); | |
1277 | } | |
1278 | } | |
1279 | } | |
1280 | ||
29499c0b | 1281 | sub scan { |
82625748 | 1282 | my ($r, $paths_table, $line) = @_; |
29499c0b MS |
1283 | if (defined $r && $line =~ /^r(\d+)$/) { |
1284 | return 0 if $1 > $r; | |
1285 | } elsif ($line =~ /^ \+empty_dir: (.+)$/) { | |
82625748 | 1286 | add_path($paths_table, $1); |
29499c0b | 1287 | } elsif ($line =~ /^ \-empty_dir: (.+)$/) { |
82625748 | 1288 | remove_path($paths_table, $1); |
29499c0b MS |
1289 | } |
1290 | 1; # continue | |
1291 | }; | |
1292 | ||
82625748 DG |
1293 | my @empty_dirs; |
1294 | my %paths_table; | |
1295 | ||
29499c0b MS |
1296 | my $gz_file = "$self->{dir}/unhandled.log.gz"; |
1297 | if (-f $gz_file) { | |
1298 | if (!can_compress()) { | |
1299 | warn "Compress::Zlib could not be found; ", | |
1300 | "empty directories in $gz_file will not be read\n"; | |
1301 | } else { | |
1302 | my $gz = Compress::Zlib::gzopen($gz_file, "rb") or | |
1303 | die "Unable to open $gz_file: $!\n"; | |
1304 | my $line; | |
1305 | while ($gz->gzreadline($line) > 0) { | |
82625748 | 1306 | scan($r, \%paths_table, $line) or last; |
29499c0b MS |
1307 | } |
1308 | $gz->gzclose; | |
1309 | } | |
1310 | } | |
1311 | ||
1312 | if (open my $fh, '<', "$self->{dir}/unhandled.log") { | |
1313 | binmode $fh or croak "binmode: $!"; | |
1314 | while (<$fh>) { | |
82625748 | 1315 | scan($r, \%paths_table, $_) or last; |
29499c0b MS |
1316 | } |
1317 | close $fh; | |
1318 | } | |
1319 | ||
82625748 | 1320 | collect_paths(\%paths_table, \@empty_dirs); |
5578ed74 | 1321 | my $strip = qr/\A\Q@{[$self->path]}\E(?:\/|$)/; |
82625748 | 1322 | foreach my $d (sort @empty_dirs) { |
29499c0b MS |
1323 | $d = uri_decode($d); |
1324 | $d =~ s/$strip//; | |
1325 | next unless length($d); | |
1326 | next if -d $d; | |
1327 | if (-e $d) { | |
1328 | warn "$d exists but is not a directory\n"; | |
1329 | } else { | |
1330 | print "creating empty directory: $d\n"; | |
1331 | mkpath([$d]); | |
1332 | } | |
1333 | } | |
1334 | } | |
1335 | ||
1336 | sub get_untracked { | |
1337 | my ($self, $ed) = @_; | |
1338 | my @out; | |
1339 | my $h = $ed->{empty}; | |
1340 | foreach (sort keys %$h) { | |
1341 | my $act = $h->{$_} ? '+empty_dir' : '-empty_dir'; | |
1342 | push @out, " $act: " . uri_encode($_); | |
1343 | warn "W: $act: $_\n"; | |
1344 | } | |
1345 | foreach my $t (qw/dir_prop file_prop/) { | |
1346 | $h = $ed->{$t} or next; | |
1347 | foreach my $path (sort keys %$h) { | |
1348 | my $ppath = $path eq '' ? '.' : $path; | |
1349 | foreach my $prop (sort keys %{$h->{$path}}) { | |
1350 | next if $SKIP_PROP{$prop}; | |
1351 | my $v = $h->{$path}->{$prop}; | |
1352 | my $t_ppath_prop = "$t: " . | |
1353 | uri_encode($ppath) . ' ' . | |
1354 | uri_encode($prop); | |
1355 | if (defined $v) { | |
1356 | push @out, " +$t_ppath_prop " . | |
1357 | uri_encode($v); | |
1358 | } else { | |
1359 | push @out, " -$t_ppath_prop"; | |
1360 | } | |
1361 | } | |
1362 | } | |
1363 | } | |
1364 | foreach my $t (qw/absent_file absent_directory/) { | |
1365 | $h = $ed->{$t} or next; | |
1366 | foreach my $parent (sort keys %$h) { | |
1367 | foreach my $path (sort @{$h->{$parent}}) { | |
1368 | push @out, " $t: " . | |
1369 | uri_encode("$parent/$path"); | |
1370 | warn "W: $t: $parent/$path ", | |
1371 | "Insufficient permissions?\n"; | |
1372 | } | |
1373 | } | |
1374 | } | |
1375 | \@out; | |
1376 | } | |
1377 | ||
29499c0b MS |
1378 | # parse_svn_date(DATE) |
1379 | # -------------------- | |
1380 | # Given a date (in UTC) from Subversion, return a string in the format | |
1381 | # "<TZ Offset> <local date/time>" that Git will use. | |
1382 | # | |
1383 | # By default the parsed date will be in UTC; if $Git::SVN::_localtime | |
1384 | # is true we'll convert it to the local timezone instead. | |
1385 | sub parse_svn_date { | |
1386 | my $date = shift || return '+0000 1970-01-01 00:00:00'; | |
1387 | my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T | |
784f4b6f | 1388 | (\d\d?)\:(\d\d)\:(\d\d)\.\d*Z$/x) or |
29499c0b MS |
1389 | croak "Unable to parse date: $date\n"; |
1390 | my $parsed_date; # Set next. | |
1391 | ||
1392 | if ($Git::SVN::_localtime) { | |
1393 | # Translate the Subversion datetime to an epoch time. | |
1394 | # Begin by switching ourselves to $date's timezone, UTC. | |
1395 | my $old_env_TZ = $ENV{TZ}; | |
1396 | $ENV{TZ} = 'UTC'; | |
1397 | ||
1398 | my $epoch_in_UTC = | |
45c956b3 | 1399 | Time::Local::timelocal($S, $M, $H, $d, $m - 1, $Y - 1900); |
29499c0b MS |
1400 | |
1401 | # Determine our local timezone (including DST) at the | |
1402 | # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the | |
1403 | # value of TZ, if any, at the time we were run. | |
1404 | if (defined $Git::SVN::Log::TZ) { | |
1405 | $ENV{TZ} = $Git::SVN::Log::TZ; | |
1406 | } else { | |
1407 | delete $ENV{TZ}; | |
1408 | } | |
1409 | ||
68868ff5 | 1410 | my $our_TZ = get_tz_offset(); |
29499c0b MS |
1411 | |
1412 | # This converts $epoch_in_UTC into our local timezone. | |
1413 | my ($sec, $min, $hour, $mday, $mon, $year, | |
1414 | $wday, $yday, $isdst) = localtime($epoch_in_UTC); | |
1415 | ||
1416 | $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d', | |
1417 | $our_TZ, $year + 1900, $mon + 1, | |
1418 | $mday, $hour, $min, $sec); | |
1419 | ||
1420 | # Reset us to the timezone in effect when we entered | |
1421 | # this routine. | |
1422 | if (defined $old_env_TZ) { | |
1423 | $ENV{TZ} = $old_env_TZ; | |
1424 | } else { | |
1425 | delete $ENV{TZ}; | |
1426 | } | |
1427 | } else { | |
1428 | $parsed_date = "+0000 $Y-$m-$d $H:$M:$S"; | |
1429 | } | |
1430 | ||
1431 | return $parsed_date; | |
1432 | } | |
1433 | ||
1434 | sub other_gs { | |
1435 | my ($self, $new_url, $url, | |
1436 | $branch_from, $r, $old_ref_id) = @_; | |
1437 | my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from); | |
1438 | unless ($gs) { | |
1439 | my $ref_id = $old_ref_id; | |
1440 | $ref_id =~ s/\@\d+-*$//; | |
1441 | $ref_id .= "\@$r"; | |
1442 | # just grow a tail if we're not unique enough :x | |
1443 | $ref_id .= '-' while find_ref($ref_id); | |
1444 | my ($u, $p, $repo_id) = ($new_url, '', $ref_id); | |
1445 | if ($u =~ s#^\Q$url\E(/|$)##) { | |
1446 | $p = $u; | |
1447 | $u = $url; | |
1448 | $repo_id = $self->{repo_id}; | |
1449 | } | |
1450 | while (1) { | |
1451 | # It is possible to tag two different subdirectories at | |
1452 | # the same revision. If the url for an existing ref | |
1453 | # does not match, we must either find a ref with a | |
1454 | # matching url or create a new ref by growing a tail. | |
1455 | $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1); | |
1456 | my (undef, $max_commit) = $gs->rev_map_max(1); | |
1457 | last if (!$max_commit); | |
1458 | my ($url) = ::cmt_metadata($max_commit); | |
1459 | last if ($url eq $gs->metadata_url); | |
1460 | $ref_id .= '-'; | |
1461 | } | |
1462 | print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1; | |
1463 | } | |
1464 | $gs | |
1465 | } | |
1466 | ||
1467 | sub call_authors_prog { | |
1468 | my ($orig_author) = @_; | |
1469 | $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author); | |
1470 | my $author = `$::_authors_prog $orig_author`; | |
1471 | if ($? != 0) { | |
1472 | die "$::_authors_prog failed with exit code $?\n" | |
1473 | } | |
1474 | if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) { | |
1475 | my ($name, $email) = ($1, $2); | |
1476 | $email = undef if length $2 == 0; | |
1477 | return [$name, $email]; | |
1478 | } else { | |
1479 | die "Author: $orig_author: $::_authors_prog returned " | |
1480 | . "invalid author format: $author\n"; | |
1481 | } | |
1482 | } | |
1483 | ||
1484 | sub check_author { | |
1485 | my ($author) = @_; | |
1486 | if (!defined $author || length $author == 0) { | |
1487 | $author = '(no author)'; | |
1488 | } | |
1489 | if (!defined $::users{$author}) { | |
1490 | if (defined $::_authors_prog) { | |
1491 | $::users{$author} = call_authors_prog($author); | |
1492 | } elsif (defined $::_authors) { | |
1493 | die "Author: $author not defined in $::_authors file\n"; | |
1494 | } | |
1495 | } | |
1496 | $author; | |
1497 | } | |
1498 | ||
1499 | sub find_extra_svk_parents { | |
4ae9a7b9 | 1500 | my ($self, $tickets, $parents) = @_; |
29499c0b MS |
1501 | # aha! svk:merge property changed... |
1502 | my @tickets = split "\n", $tickets; | |
1503 | my @known_parents; | |
1504 | for my $ticket ( @tickets ) { | |
1505 | my ($uuid, $path, $rev) = split /:/, $ticket; | |
1506 | if ( $uuid eq $self->ra_uuid ) { | |
d2fd119c | 1507 | my $repos_root = $self->url; |
29499c0b MS |
1508 | my $branch_from = $path; |
1509 | $branch_from =~ s{^/}{}; | |
d2fd119c MS |
1510 | my $gs = $self->other_gs(add_path_to_url( $repos_root, $branch_from ), |
1511 | $repos_root, | |
29499c0b MS |
1512 | $branch_from, |
1513 | $rev, | |
1514 | $self->{ref_id}); | |
1515 | if ( my $commit = $gs->rev_map_get($rev, $uuid) ) { | |
1516 | # wahey! we found it, but it might be | |
1517 | # an old one (!) | |
1518 | push @known_parents, [ $rev, $commit ]; | |
1519 | } | |
1520 | } | |
1521 | } | |
1522 | # Ordering matters; highest-numbered commit merge tickets | |
1523 | # first, as they may account for later merge ticket additions | |
1524 | # or changes. | |
1525 | @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents; | |
1526 | for my $parent ( @known_parents ) { | |
1527 | my @cmd = ('rev-list', $parent, map { "^$_" } @$parents ); | |
1528 | my ($msg_fh, $ctx) = command_output_pipe(@cmd); | |
1529 | my $new; | |
1530 | while ( <$msg_fh> ) { | |
1531 | $new=1;last; | |
1532 | } | |
1533 | command_close_pipe($msg_fh, $ctx); | |
1534 | if ( $new ) { | |
1535 | print STDERR | |
1536 | "Found merge parent (svk:merge ticket): $parent\n"; | |
1537 | push @$parents, $parent; | |
1538 | } | |
1539 | } | |
1540 | } | |
1541 | ||
1542 | sub lookup_svn_merge { | |
1543 | my $uuid = shift; | |
1544 | my $url = shift; | |
abfef3bb JSO |
1545 | my $source = shift; |
1546 | my $revs = shift; | |
29499c0b | 1547 | |
29499c0b MS |
1548 | my $path = $source; |
1549 | $path =~ s{^/}{}; | |
1550 | my $gs = Git::SVN->find_by_url($url.$source, $url, $path); | |
1551 | if ( !$gs ) { | |
1552 | warn "Couldn't find revmap for $url$source\n"; | |
1553 | return; | |
1554 | } | |
1555 | my @ranges = split ",", $revs; | |
1556 | my ($tip, $tip_commit); | |
1557 | my @merged_commit_ranges; | |
1558 | # find the tip | |
1559 | for my $range ( @ranges ) { | |
47543d16 JP |
1560 | if ($range =~ /[*]$/) { |
1561 | warn "W: Ignoring partial merge in svn:mergeinfo " | |
1562 | ."dirprop: $source:$range\n"; | |
1563 | next; | |
1564 | } | |
29499c0b MS |
1565 | my ($bottom, $top) = split "-", $range; |
1566 | $top ||= $bottom; | |
1567 | my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top ); | |
1568 | my $top_commit = $gs->find_rev_before( $top, 1, $bottom ); | |
1569 | ||
1570 | unless ($top_commit and $bottom_commit) { | |
eae6cf5a | 1571 | warn "W: unknown path/rev in svn:mergeinfo " |
29499c0b MS |
1572 | ."dirprop: $source:$range\n"; |
1573 | next; | |
1574 | } | |
1575 | ||
1576 | if (scalar(command('rev-parse', "$bottom_commit^@"))) { | |
1577 | push @merged_commit_ranges, | |
1578 | "$bottom_commit^..$top_commit"; | |
1579 | } else { | |
1580 | push @merged_commit_ranges, "$top_commit"; | |
1581 | } | |
1582 | ||
1583 | if ( !defined $tip or $top > $tip ) { | |
1584 | $tip = $top; | |
1585 | $tip_commit = $top_commit; | |
1586 | } | |
1587 | } | |
1588 | return ($tip_commit, @merged_commit_ranges); | |
1589 | } | |
1590 | ||
1591 | sub _rev_list { | |
1592 | my ($msg_fh, $ctx) = command_output_pipe( | |
1593 | "rev-list", @_, | |
1594 | ); | |
1595 | my @rv; | |
1596 | while ( <$msg_fh> ) { | |
1597 | chomp; | |
1598 | push @rv, $_; | |
1599 | } | |
1600 | command_close_pipe($msg_fh, $ctx); | |
1601 | @rv; | |
1602 | } | |
1603 | ||
d0b34f24 | 1604 | sub check_cherry_pick2 { |
29499c0b MS |
1605 | my $base = shift; |
1606 | my $tip = shift; | |
1607 | my $parents = shift; | |
1608 | my @ranges = @_; | |
1609 | my %commits = map { $_ => 1 } | |
1610 | _rev_list("--no-merges", $tip, "--not", $base, @$parents, "--"); | |
1611 | for my $range ( @ranges ) { | |
1612 | delete @commits{_rev_list($range, "--")}; | |
1613 | } | |
1614 | for my $commit (keys %commits) { | |
1615 | if (has_no_changes($commit)) { | |
1616 | delete $commits{$commit}; | |
1617 | } | |
1618 | } | |
d0b34f24 EW |
1619 | my @k = (keys %commits); |
1620 | return (scalar @k, $k[0]); | |
29499c0b MS |
1621 | } |
1622 | ||
1623 | sub has_no_changes { | |
1624 | my $commit = shift; | |
1625 | ||
1626 | my @revs = split / /, command_oneline( | |
1627 | qw(rev-list --parents -1 -m), $commit); | |
1628 | ||
1629 | # Commits with no parents, e.g. the start of a partial branch, | |
1630 | # have changes by definition. | |
1631 | return 1 if (@revs < 2); | |
1632 | ||
1633 | # Commits with multiple parents, e.g a merge, have no changes | |
1634 | # by definition. | |
1635 | return 0 if (@revs > 2); | |
1636 | ||
1637 | return (command_oneline("rev-parse", "$commit^{tree}") eq | |
1638 | command_oneline("rev-parse", "$commit~1^{tree}")); | |
1639 | } | |
1640 | ||
1641 | sub tie_for_persistent_memoization { | |
1642 | my $hash = shift; | |
1643 | my $path = shift; | |
1644 | ||
47092c10 EW |
1645 | unless ($memo_backend) { |
1646 | if (eval { require Git::SVN::Memoize::YAML; 1}) { | |
1647 | $memo_backend = 1; | |
1648 | } else { | |
1649 | require Memoize::Storable; | |
1650 | $memo_backend = -1; | |
1651 | } | |
1652 | } | |
1653 | ||
1654 | if ($memo_backend > 0) { | |
29499c0b MS |
1655 | tie %$hash => 'Git::SVN::Memoize::YAML', "$path.yaml"; |
1656 | } else { | |
1657 | tie %$hash => 'Memoize::Storable', "$path.db", 'nstore'; | |
1658 | } | |
1659 | } | |
1660 | ||
1661 | # The GIT_DIR environment variable is not always set until after the command | |
1662 | # line arguments are processed, so we can't memoize in a BEGIN block. | |
1663 | { | |
1664 | my $memoized = 0; | |
1665 | ||
1666 | sub memoize_svn_mergeinfo_functions { | |
1667 | return if $memoized; | |
1668 | $memoized = 1; | |
1669 | ||
1670 | my $cache_path = "$ENV{GIT_DIR}/svn/.caches/"; | |
1671 | mkpath([$cache_path]) unless -d $cache_path; | |
1672 | ||
1673 | my %lookup_svn_merge_cache; | |
d0b34f24 | 1674 | my %check_cherry_pick2_cache; |
29499c0b MS |
1675 | my %has_no_changes_cache; |
1676 | ||
1677 | tie_for_persistent_memoization(\%lookup_svn_merge_cache, | |
1678 | "$cache_path/lookup_svn_merge"); | |
1679 | memoize 'lookup_svn_merge', | |
1680 | SCALAR_CACHE => 'FAULT', | |
1681 | LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache], | |
1682 | ; | |
1683 | ||
d0b34f24 EW |
1684 | tie_for_persistent_memoization(\%check_cherry_pick2_cache, |
1685 | "$cache_path/check_cherry_pick2"); | |
1686 | memoize 'check_cherry_pick2', | |
29499c0b | 1687 | SCALAR_CACHE => 'FAULT', |
d0b34f24 | 1688 | LIST_CACHE => ['HASH' => \%check_cherry_pick2_cache], |
29499c0b MS |
1689 | ; |
1690 | ||
1691 | tie_for_persistent_memoization(\%has_no_changes_cache, | |
1692 | "$cache_path/has_no_changes"); | |
1693 | memoize 'has_no_changes', | |
1694 | SCALAR_CACHE => ['HASH' => \%has_no_changes_cache], | |
1695 | LIST_CACHE => 'FAULT', | |
1696 | ; | |
1697 | } | |
1698 | ||
1699 | sub unmemoize_svn_mergeinfo_functions { | |
1700 | return if not $memoized; | |
1701 | $memoized = 0; | |
1702 | ||
1703 | Memoize::unmemoize 'lookup_svn_merge'; | |
d0b34f24 | 1704 | Memoize::unmemoize 'check_cherry_pick2'; |
29499c0b MS |
1705 | Memoize::unmemoize 'has_no_changes'; |
1706 | } | |
1707 | ||
61b472ed PB |
1708 | sub clear_memoized_mergeinfo_caches { |
1709 | die "Only call this method in non-memoized context" if ($memoized); | |
1710 | ||
1711 | my $cache_path = "$ENV{GIT_DIR}/svn/.caches/"; | |
1712 | return unless -d $cache_path; | |
1713 | ||
1714 | for my $cache_file (("$cache_path/lookup_svn_merge", | |
d0b34f24 EW |
1715 | "$cache_path/check_cherry_pick", # old |
1716 | "$cache_path/check_cherry_pick2", | |
61b472ed PB |
1717 | "$cache_path/has_no_changes")) { |
1718 | for my $suffix (qw(yaml db)) { | |
1719 | my $file = "$cache_file.$suffix"; | |
1720 | next unless -e $file; | |
1721 | unlink($file) or die "unlink($file) failed: $!\n"; | |
1722 | } | |
1723 | } | |
1724 | } | |
1725 | ||
1726 | ||
29499c0b MS |
1727 | Memoize::memoize 'Git::SVN::repos_root'; |
1728 | } | |
1729 | ||
1730 | END { | |
1731 | # Force cache writeout explicitly instead of waiting for | |
1732 | # global destruction to avoid segfault in Storable: | |
1733 | # http://rt.cpan.org/Public/Bug/Display.html?id=36087 | |
1734 | unmemoize_svn_mergeinfo_functions(); | |
1735 | } | |
1736 | ||
1737 | sub parents_exclude { | |
1738 | my $parents = shift; | |
1739 | my @commits = @_; | |
1740 | return unless @commits; | |
1741 | ||
1742 | my @excluded; | |
1743 | my $excluded; | |
1744 | do { | |
1745 | my @cmd = ('rev-list', "-1", @commits, "--not", @$parents ); | |
1746 | $excluded = command_oneline(@cmd); | |
1747 | if ( $excluded ) { | |
1748 | my @new; | |
1749 | my $found; | |
1750 | for my $commit ( @commits ) { | |
1751 | if ( $commit eq $excluded ) { | |
1752 | push @excluded, $commit; | |
1753 | $found++; | |
29499c0b MS |
1754 | } |
1755 | else { | |
1756 | push @new, $commit; | |
1757 | } | |
1758 | } | |
1759 | die "saw commit '$excluded' in rev-list output, " | |
1760 | ."but we didn't ask for that commit (wanted: @commits --not @$parents)" | |
1761 | unless $found; | |
1762 | @commits = @new; | |
1763 | } | |
1764 | } | |
1765 | while ($excluded and @commits); | |
1766 | ||
1767 | return @excluded; | |
1768 | } | |
1769 | ||
abfef3bb JSO |
1770 | # Compute what's new in svn:mergeinfo. |
1771 | sub mergeinfo_changes { | |
1772 | my ($self, $old_path, $old_rev, $path, $rev, $mergeinfo_prop) = @_; | |
1773 | my %minfo = map {split ":", $_ } split "\n", $mergeinfo_prop; | |
1774 | my $old_minfo = {}; | |
1775 | ||
2b6c613f EW |
1776 | my $ra = $self->ra; |
1777 | # Give up if $old_path isn't in the repo. | |
1778 | # This is probably a merge on a subtree. | |
1779 | if ($ra->check_path($old_path, $old_rev) != $SVN::Node::dir) { | |
1780 | warn "W: ignoring svn:mergeinfo on $old_path, ", | |
1781 | "directory didn't exist in r$old_rev\n"; | |
1782 | return {}; | |
1783 | } | |
1784 | my (undef, undef, $props) = $ra->get_dir($old_path, $old_rev); | |
54b95346 EW |
1785 | if (defined $props->{"svn:mergeinfo"}) { |
1786 | my %omi = map {split ":", $_ } split "\n", | |
1787 | $props->{"svn:mergeinfo"}; | |
1788 | $old_minfo = \%omi; | |
1789 | } | |
abfef3bb JSO |
1790 | |
1791 | my %changes = (); | |
1792 | foreach my $p (keys %minfo) { | |
1793 | my $a = $old_minfo->{$p} || ""; | |
1794 | my $b = $minfo{$p}; | |
1795 | # Omit merged branches whose ranges lists are unchanged. | |
1796 | next if $a eq $b; | |
1797 | # Remove any common range list prefix. | |
1798 | ($a ^ $b) =~ /^[\0]*/; | |
1799 | my $common_prefix = rindex $b, ",", $+[0] - 1; | |
1800 | $changes{$p} = substr $b, $common_prefix + 1; | |
1801 | } | |
1802 | print STDERR "Checking svn:mergeinfo changes since r$old_rev: ", | |
1803 | scalar(keys %minfo), " sources, ", | |
1804 | scalar(keys %changes), " changed\n"; | |
1805 | ||
1806 | return \%changes; | |
1807 | } | |
29499c0b MS |
1808 | |
1809 | # note: this function should only be called if the various dirprops | |
1810 | # have actually changed | |
1811 | sub find_extra_svn_parents { | |
4ae9a7b9 | 1812 | my ($self, $mergeinfo, $parents) = @_; |
29499c0b MS |
1813 | # aha! svk:merge property changed... |
1814 | ||
1815 | memoize_svn_mergeinfo_functions(); | |
1816 | ||
1817 | # We first search for merged tips which are not in our | |
1818 | # history. Then, we figure out which git revisions are in | |
1819 | # that tip, but not this revision. If all of those revisions | |
1820 | # are now marked as merge, we can add the tip as a parent. | |
abfef3bb | 1821 | my @merges = sort keys %$mergeinfo; |
29499c0b | 1822 | my @merge_tips; |
06ee19e8 | 1823 | my $url = $self->url; |
29499c0b | 1824 | my $uuid = $self->ra_uuid; |
f271fad2 | 1825 | my @all_ranges; |
29499c0b MS |
1826 | for my $merge ( @merges ) { |
1827 | my ($tip_commit, @ranges) = | |
abfef3bb JSO |
1828 | lookup_svn_merge( $uuid, $url, |
1829 | $merge, $mergeinfo->{$merge} ); | |
29499c0b MS |
1830 | unless (!$tip_commit or |
1831 | grep { $_ eq $tip_commit } @$parents ) { | |
1832 | push @merge_tips, $tip_commit; | |
f271fad2 | 1833 | push @all_ranges, @ranges; |
29499c0b MS |
1834 | } else { |
1835 | push @merge_tips, undef; | |
1836 | } | |
1837 | } | |
1838 | ||
1839 | my %excluded = map { $_ => 1 } | |
1840 | parents_exclude($parents, grep { defined } @merge_tips); | |
1841 | ||
1842 | # check merge tips for new parents | |
1843 | my @new_parents; | |
1844 | for my $merge_tip ( @merge_tips ) { | |
abfef3bb | 1845 | my $merge = shift @merges; |
29499c0b | 1846 | next unless $merge_tip and $excluded{$merge_tip}; |
abfef3bb | 1847 | my $spec = "$merge:$mergeinfo->{$merge}"; |
29499c0b | 1848 | |
29499c0b MS |
1849 | # check out 'new' tips |
1850 | my $merge_base; | |
1851 | eval { | |
1852 | $merge_base = command_oneline( | |
1853 | "merge-base", | |
1854 | @$parents, $merge_tip, | |
1855 | ); | |
1856 | }; | |
1857 | if ($@) { | |
1858 | die "An error occurred during merge-base" | |
1859 | unless $@->isa("Git::Error::Command"); | |
1860 | ||
1861 | warn "W: Cannot find common ancestor between ". | |
1862 | "@$parents and $merge_tip. Ignoring merge info.\n"; | |
1863 | next; | |
1864 | } | |
1865 | ||
1866 | # double check that there are no missing non-merge commits | |
d0b34f24 | 1867 | my ($ninc, $ifirst) = check_cherry_pick2( |
29499c0b MS |
1868 | $merge_base, $merge_tip, |
1869 | $parents, | |
f271fad2 | 1870 | @all_ranges, |
29499c0b MS |
1871 | ); |
1872 | ||
d0b34f24 | 1873 | if ($ninc) { |
da0bc948 | 1874 | warn "W: svn cherry-pick ignored ($spec) - missing " . |
d0b34f24 | 1875 | "$ninc commit(s) (eg $ifirst)\n"; |
29499c0b | 1876 | } else { |
da0bc948 | 1877 | warn "Found merge parent ($spec): ", $merge_tip, "\n"; |
29499c0b MS |
1878 | push @new_parents, $merge_tip; |
1879 | } | |
1880 | } | |
1881 | ||
1882 | # cater for merges which merge commits from multiple branches | |
1883 | if ( @new_parents > 1 ) { | |
1884 | for ( my $i = 0; $i <= $#new_parents; $i++ ) { | |
1885 | for ( my $j = 0; $j <= $#new_parents; $j++ ) { | |
1886 | next if $i == $j; | |
1887 | next unless $new_parents[$i]; | |
1888 | next unless $new_parents[$j]; | |
1889 | my $revs = command_oneline( | |
1890 | "rev-list", "-1", | |
1891 | "$new_parents[$i]..$new_parents[$j]", | |
1892 | ); | |
1893 | if ( !$revs ) { | |
1894 | undef($new_parents[$j]); | |
1895 | } | |
1896 | } | |
1897 | } | |
1898 | } | |
1899 | push @$parents, grep { defined } @new_parents; | |
1900 | } | |
1901 | ||
1902 | sub make_log_entry { | |
abfef3bb | 1903 | my ($self, $rev, $parents, $ed, $parent_rev, $parent_path) = @_; |
29499c0b MS |
1904 | my $untracked = $self->get_untracked($ed); |
1905 | ||
1906 | my @parents = @$parents; | |
9ee13a93 JSO |
1907 | my $props = $ed->{dir_prop}{$self->path}; |
1908 | if ( $props->{"svk:merge"} ) { | |
4ae9a7b9 | 1909 | $self->find_extra_svk_parents($props->{"svk:merge"}, \@parents); |
9ee13a93 JSO |
1910 | } |
1911 | if ( $props->{"svn:mergeinfo"} ) { | |
1912 | my $mi_changes = $self->mergeinfo_changes | |
1913 | ($parent_path, $parent_rev, | |
1914 | $self->path, $rev, | |
1915 | $props->{"svn:mergeinfo"}); | |
4ae9a7b9 | 1916 | $self->find_extra_svn_parents($mi_changes, \@parents); |
29499c0b MS |
1917 | } |
1918 | ||
1919 | open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!; | |
1920 | print $un "r$rev\n" or croak $!; | |
1921 | print $un $_, "\n" foreach @$untracked; | |
1922 | my %log_entry = ( parents => \@parents, revision => $rev, | |
1923 | log => ''); | |
1924 | ||
1925 | my $headrev; | |
1926 | my $logged = delete $self->{logged_rev_props}; | |
1927 | if (!$logged || $self->{-want_revprops}) { | |
1928 | my $rp = $self->ra->rev_proplist($rev); | |
1929 | foreach (sort keys %$rp) { | |
1930 | my $v = $rp->{$_}; | |
1931 | if (/^svn:(author|date|log)$/) { | |
1932 | $log_entry{$1} = $v; | |
1933 | } elsif ($_ eq 'svm:headrev') { | |
1934 | $headrev = $v; | |
1935 | } else { | |
1936 | print $un " rev_prop: ", uri_encode($_), ' ', | |
1937 | uri_encode($v), "\n"; | |
1938 | } | |
1939 | } | |
1940 | } else { | |
1941 | map { $log_entry{$_} = $logged->{$_} } keys %$logged; | |
1942 | } | |
1943 | close $un or croak $!; | |
1944 | ||
1945 | $log_entry{date} = parse_svn_date($log_entry{date}); | |
1946 | $log_entry{log} .= "\n"; | |
1947 | my $author = $log_entry{author} = check_author($log_entry{author}); | |
1948 | my ($name, $email) = defined $::users{$author} ? @{$::users{$author}} | |
1949 | : ($author, undef); | |
1950 | ||
1951 | my ($commit_name, $commit_email) = ($name, $email); | |
1952 | if ($_use_log_author) { | |
1953 | my $name_field; | |
1954 | if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) { | |
1955 | $name_field = $1; | |
1956 | } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) { | |
1957 | $name_field = $1; | |
1958 | } | |
1959 | if (!defined $name_field) { | |
1960 | if (!defined $email) { | |
1961 | $email = $name; | |
1962 | } | |
1963 | } elsif ($name_field =~ /(.*?)\s+<(.*)>/) { | |
1964 | ($name, $email) = ($1, $2); | |
1965 | } elsif ($name_field =~ /(.*)@/) { | |
1966 | ($name, $email) = ($1, $name_field); | |
1967 | } else { | |
1968 | ($name, $email) = ($name_field, $name_field); | |
1969 | } | |
1970 | } | |
1971 | if (defined $headrev && $self->use_svm_props) { | |
1972 | if ($self->rewrite_root) { | |
1973 | die "Can't have both 'useSvmProps' and 'rewriteRoot' ", | |
1974 | "options set!\n"; | |
1975 | } | |
1976 | if ($self->rewrite_uuid) { | |
1977 | die "Can't have both 'useSvmProps' and 'rewriteUUID' ", | |
1978 | "options set!\n"; | |
1979 | } | |
1980 | my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i; | |
1981 | # we don't want "SVM: initializing mirror for junk" ... | |
1982 | return undef if $r == 0; | |
1983 | my $svm = $self->svm; | |
1984 | if ($uuid ne $svm->{uuid}) { | |
1985 | die "UUID mismatch on SVM path:\n", | |
1986 | "expected: $svm->{uuid}\n", | |
1987 | " got: $uuid\n"; | |
1988 | } | |
1989 | my $full_url = $self->full_url; | |
1990 | $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or | |
1991 | die "Failed to replace '$svm->{replace}' with ", | |
1992 | "'$svm->{source}' in $full_url\n"; | |
1993 | # throw away username for storing in records | |
1994 | remove_username($full_url); | |
1995 | $log_entry{metadata} = "$full_url\@$r $uuid"; | |
1996 | $log_entry{svm_revision} = $r; | |
1997 | $email ||= "$author\@$uuid"; | |
1998 | $commit_email ||= "$author\@$uuid"; | |
1999 | } elsif ($self->use_svnsync_props) { | |
705b49cb MS |
2000 | my $full_url = canonicalize_url( |
2001 | add_path_to_url( $self->svnsync->{url}, $self->path ) | |
2002 | ); | |
29499c0b MS |
2003 | remove_username($full_url); |
2004 | my $uuid = $self->svnsync->{uuid}; | |
2005 | $log_entry{metadata} = "$full_url\@$rev $uuid"; | |
2006 | $email ||= "$author\@$uuid"; | |
2007 | $commit_email ||= "$author\@$uuid"; | |
2008 | } else { | |
2009 | my $url = $self->metadata_url; | |
2010 | remove_username($url); | |
2011 | my $uuid = $self->rewrite_uuid || $self->ra->get_uuid; | |
2012 | $log_entry{metadata} = "$url\@$rev " . $uuid; | |
2013 | $email ||= "$author\@" . $uuid; | |
2014 | $commit_email ||= "$author\@" . $uuid; | |
2015 | } | |
2016 | $log_entry{name} = $name; | |
2017 | $log_entry{email} = $email; | |
2018 | $log_entry{commit_name} = $commit_name; | |
2019 | $log_entry{commit_email} = $commit_email; | |
2020 | \%log_entry; | |
2021 | } | |
2022 | ||
2023 | sub fetch { | |
2024 | my ($self, $min_rev, $max_rev, @parents) = @_; | |
2025 | my ($last_rev, $last_commit) = $self->last_rev_commit; | |
2026 | my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev); | |
2027 | $self->ra->gs_fetch_loop_common($base, $head, [$self]); | |
2028 | } | |
2029 | ||
2030 | sub set_tree_cb { | |
2031 | my ($self, $log_entry, $tree, $rev, $date, $author) = @_; | |
2032 | $self->{inject_parents} = { $rev => $tree }; | |
2033 | $self->fetch(undef, undef); | |
2034 | } | |
2035 | ||
2036 | sub set_tree { | |
2037 | my ($self, $tree) = (shift, shift); | |
2038 | my $log_entry = ::get_commit_entry($tree); | |
2039 | unless ($self->{last_rev}) { | |
2040 | fatal("Must have an existing revision to commit"); | |
2041 | } | |
2042 | my %ed_opts = ( r => $self->{last_rev}, | |
2043 | log => $log_entry->{log}, | |
2044 | ra => $self->ra, | |
2045 | tree_a => $self->{last_commit}, | |
2046 | tree_b => $tree, | |
2047 | editor_cb => sub { | |
2048 | $self->set_tree_cb($log_entry, $tree, @_) }, | |
5578ed74 | 2049 | svn_path => $self->path ); |
29499c0b MS |
2050 | if (!Git::SVN::Editor->new(\%ed_opts)->apply_diff) { |
2051 | print "No changes\nr$self->{last_rev} = $tree\n"; | |
2052 | } | |
2053 | } | |
2054 | ||
2055 | sub rebuild_from_rev_db { | |
2056 | my ($self, $path) = @_; | |
2057 | my $r = -1; | |
2058 | open my $fh, '<', $path or croak "open: $!"; | |
2059 | binmode $fh or croak "binmode: $!"; | |
2060 | while (<$fh>) { | |
2061 | length($_) == 41 or croak "inconsistent size in ($_) != 41"; | |
2062 | chomp($_); | |
2063 | ++$r; | |
2064 | next if $_ eq ('0' x 40); | |
2065 | $self->rev_map_set($r, $_); | |
2066 | print "r$r = $_\n"; | |
2067 | } | |
2068 | close $fh or croak "close: $!"; | |
2069 | unlink $path or croak "unlink: $!"; | |
2070 | } | |
2071 | ||
ab0bcec9 | 2072 | #define a global associate map to record rebuild status |
2073 | my %rebuild_status; | |
2074 | #define a global associate map to record rebuild verify status | |
2075 | my %rebuild_verify_status; | |
2076 | ||
29499c0b MS |
2077 | sub rebuild { |
2078 | my ($self) = @_; | |
2079 | my $map_path = $self->map_path; | |
2080 | my $partial = (-e $map_path && ! -z $map_path); | |
ab0bcec9 | 2081 | my $verify_key = $self->refname.'^0'; |
2082 | if (!$rebuild_verify_status{$verify_key}) { | |
2083 | my $verify_result = ::verify_ref($verify_key); | |
2084 | if ($verify_result) { | |
2085 | $rebuild_verify_status{$verify_key} = 1; | |
2086 | } | |
2087 | } | |
2088 | if (!$rebuild_verify_status{$verify_key}) { | |
2089 | return; | |
2090 | } | |
29499c0b MS |
2091 | if (!$partial && ($self->use_svm_props || $self->no_metadata)) { |
2092 | my $rev_db = $self->rev_db_path; | |
2093 | $self->rebuild_from_rev_db($rev_db); | |
2094 | if ($self->use_svm_props) { | |
2095 | my $svm_rev_db = $self->rev_db_path($self->svm_uuid); | |
2096 | $self->rebuild_from_rev_db($svm_rev_db); | |
2097 | } | |
2098 | $self->unlink_rev_db_symlink; | |
2099 | return; | |
2100 | } | |
2101 | print "Rebuilding $map_path ...\n" if (!$partial); | |
2102 | my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) : | |
2103 | (undef, undef)); | |
ab0bcec9 | 2104 | my $key_value = ($head ? "$head.." : "") . $self->refname; |
2105 | if (exists $rebuild_status{$key_value}) { | |
2106 | print "Done rebuilding $map_path\n" if (!$partial || !$head); | |
2107 | my $rev_db_path = $self->rev_db_path; | |
2108 | if (-f $self->rev_db_path) { | |
2109 | unlink $self->rev_db_path or croak "unlink: $!"; | |
2110 | } | |
2111 | $self->unlink_rev_db_symlink; | |
2112 | return; | |
2113 | } | |
29499c0b | 2114 | my ($log, $ctx) = |
ab0bcec9 | 2115 | command_output_pipe(qw/rev-list --pretty=raw --reverse/, |
2116 | $key_value, | |
29499c0b | 2117 | '--'); |
ab0bcec9 | 2118 | $rebuild_status{$key_value} = 1; |
29499c0b MS |
2119 | my $metadata_url = $self->metadata_url; |
2120 | remove_username($metadata_url); | |
2121 | my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid; | |
2122 | my $c; | |
2123 | while (<$log>) { | |
2124 | if ( m{^commit ($::sha1)$} ) { | |
2125 | $c = $1; | |
2126 | next; | |
2127 | } | |
2128 | next unless s{^\s*(git-svn-id:)}{$1}; | |
2129 | my ($url, $rev, $uuid) = ::extract_metadata($_); | |
2130 | remove_username($url); | |
2131 | ||
2132 | # ignore merges (from set-tree) | |
2133 | next if (!defined $rev || !$uuid); | |
2134 | ||
2135 | # if we merged or otherwise started elsewhere, this is | |
2136 | # how we break out of it | |
2137 | if (($uuid ne $svn_uuid) || | |
2138 | ($metadata_url && $url && ($url ne $metadata_url))) { | |
2139 | next; | |
2140 | } | |
2141 | if ($partial && $head) { | |
2142 | print "Partial-rebuilding $map_path ...\n"; | |
2143 | print "Currently at $base_rev = $head\n"; | |
2144 | $head = undef; | |
2145 | } | |
2146 | ||
2147 | $self->rev_map_set($rev, $c); | |
2148 | print "r$rev = $c\n"; | |
2149 | } | |
2150 | command_close_pipe($log, $ctx); | |
2151 | print "Done rebuilding $map_path\n" if (!$partial || !$head); | |
2152 | my $rev_db_path = $self->rev_db_path; | |
2153 | if (-f $self->rev_db_path) { | |
2154 | unlink $self->rev_db_path or croak "unlink: $!"; | |
2155 | } | |
2156 | $self->unlink_rev_db_symlink; | |
2157 | } | |
2158 | ||
2159 | # rev_map: | |
2160 | # Tie::File seems to be prone to offset errors if revisions get sparse, | |
2161 | # it's not that fast, either. Tie::File is also not in Perl 5.6. So | |
2162 | # one of my favorite modules is out :< Next up would be one of the DBM | |
2163 | # modules, but I'm not sure which is most portable... | |
2164 | # | |
2165 | # This is the replacement for the rev_db format, which was too big | |
2166 | # and inefficient for large repositories with a lot of sparse history | |
2167 | # (mainly tags) | |
2168 | # | |
2169 | # The format is this: | |
2170 | # - 24 bytes for every record, | |
2171 | # * 4 bytes for the integer representing an SVN revision number | |
2172 | # * 20 bytes representing the sha1 of a git commit | |
2173 | # - No empty padding records like the old format | |
2174 | # (except the last record, which can be overwritten) | |
2175 | # - new records are written append-only since SVN revision numbers | |
2176 | # increase monotonically | |
2177 | # - lookups on SVN revision number are done via a binary search | |
2178 | # - Piping the file to xxd -c24 is a good way of dumping it for | |
2179 | # viewing or editing (piped back through xxd -r), should the need | |
2180 | # ever arise. | |
2181 | # - The last record can be padding revision with an all-zero sha1 | |
2182 | # This is used to optimize fetch performance when using multiple | |
2183 | # "fetch" directives in .git/config | |
2184 | # | |
2185 | # These files are disposable unless noMetadata or useSvmProps is set | |
2186 | ||
2187 | sub _rev_map_set { | |
2188 | my ($fh, $rev, $commit) = @_; | |
2189 | ||
2190 | binmode $fh or croak "binmode: $!"; | |
2191 | my $size = (stat($fh))[7]; | |
2192 | ($size % 24) == 0 or croak "inconsistent size: $size"; | |
2193 | ||
2194 | my $wr_offset = 0; | |
2195 | if ($size > 0) { | |
2196 | sysseek($fh, -24, SEEK_END) or croak "seek: $!"; | |
2197 | my $read = sysread($fh, my $buf, 24) or croak "read: $!"; | |
2198 | $read == 24 or croak "read only $read bytes (!= 24)"; | |
2199 | my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf); | |
2200 | if ($last_commit eq ('0' x40)) { | |
2201 | if ($size >= 48) { | |
2202 | sysseek($fh, -48, SEEK_END) or croak "seek: $!"; | |
2203 | $read = sysread($fh, $buf, 24) or | |
2204 | croak "read: $!"; | |
2205 | $read == 24 or | |
2206 | croak "read only $read bytes (!= 24)"; | |
2207 | ($last_rev, $last_commit) = | |
2208 | unpack(rev_map_fmt, $buf); | |
2209 | if ($last_commit eq ('0' x40)) { | |
2210 | croak "inconsistent .rev_map\n"; | |
2211 | } | |
2212 | } | |
2213 | if ($last_rev >= $rev) { | |
2214 | croak "last_rev is higher!: $last_rev >= $rev"; | |
2215 | } | |
2216 | $wr_offset = -24; | |
2217 | } | |
2218 | } | |
2219 | sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!"; | |
2220 | syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or | |
2221 | croak "write: $!"; | |
2222 | } | |
2223 | ||
2224 | sub _rev_map_reset { | |
2225 | my ($fh, $rev, $commit) = @_; | |
2226 | my $c = _rev_map_get($fh, $rev); | |
2227 | $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n"; | |
2228 | my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!"; | |
2229 | truncate $fh, $offset or croak "truncate: $!"; | |
2230 | } | |
2231 | ||
2232 | sub mkfile { | |
2233 | my ($path) = @_; | |
2234 | unless (-e $path) { | |
2235 | my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#); | |
2236 | mkpath([$dir]) unless -d $dir; | |
2237 | open my $fh, '>>', $path or die "Couldn't create $path: $!\n"; | |
2238 | close $fh or die "Couldn't close (create) $path: $!\n"; | |
2239 | } | |
2240 | } | |
2241 | ||
2242 | sub rev_map_set { | |
2243 | my ($self, $rev, $commit, $update_ref, $uuid) = @_; | |
2244 | defined $commit or die "missing arg3\n"; | |
2245 | length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n"; | |
2246 | my $db = $self->map_path($uuid); | |
2247 | my $db_lock = "$db.lock"; | |
2248 | my $sigmask; | |
2249 | $update_ref ||= 0; | |
2250 | if ($update_ref) { | |
2251 | $sigmask = POSIX::SigSet->new(); | |
2252 | my $signew = POSIX::SigSet->new(SIGINT, SIGHUP, SIGTERM, | |
2253 | SIGALRM, SIGUSR1, SIGUSR2); | |
2254 | sigprocmask(SIG_BLOCK, $signew, $sigmask) or | |
2255 | croak "Can't block signals: $!"; | |
2256 | } | |
2257 | mkfile($db); | |
2258 | ||
2259 | $LOCKFILES{$db_lock} = 1; | |
2260 | my $sync; | |
2261 | # both of these options make our .rev_db file very, very important | |
2262 | # and we can't afford to lose it because rebuild() won't work | |
2263 | if ($self->use_svm_props || $self->no_metadata) { | |
47092c10 | 2264 | require File::Copy; |
29499c0b | 2265 | $sync = 1; |
47092c10 | 2266 | File::Copy::copy($db, $db_lock) or die "rev_map_set(@_): ", |
29499c0b MS |
2267 | "Failed to copy: ", |
2268 | "$db => $db_lock ($!)\n"; | |
2269 | } else { | |
2270 | rename $db, $db_lock or die "rev_map_set(@_): ", | |
2271 | "Failed to rename: ", | |
2272 | "$db => $db_lock ($!)\n"; | |
2273 | } | |
2274 | ||
2275 | sysopen(my $fh, $db_lock, O_RDWR | O_CREAT) | |
2276 | or croak "Couldn't open $db_lock: $!\n"; | |
61b472ed PB |
2277 | if ($update_ref eq 'reset') { |
2278 | clear_memoized_mergeinfo_caches(); | |
2279 | _rev_map_reset($fh, $rev, $commit); | |
2280 | } else { | |
2281 | _rev_map_set($fh, $rev, $commit); | |
2282 | } | |
2283 | ||
29499c0b MS |
2284 | if ($sync) { |
2285 | $fh->flush or die "Couldn't flush $db_lock: $!\n"; | |
2286 | $fh->sync or die "Couldn't sync $db_lock: $!\n"; | |
2287 | } | |
2288 | close $fh or croak $!; | |
2289 | if ($update_ref) { | |
2290 | $_head = $self; | |
2291 | my $note = ""; | |
2292 | $note = " ($update_ref)" if ($update_ref !~ /^\d*$/); | |
2293 | command_noisy('update-ref', '-m', "r$rev$note", | |
2294 | $self->refname, $commit); | |
2295 | } | |
2296 | rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ", | |
2297 | "$db_lock => $db ($!)\n"; | |
2298 | delete $LOCKFILES{$db_lock}; | |
2299 | if ($update_ref) { | |
2300 | sigprocmask(SIG_SETMASK, $sigmask) or | |
2301 | croak "Can't restore signal mask: $!"; | |
2302 | } | |
2303 | } | |
2304 | ||
2305 | # If want_commit, this will return an array of (rev, commit) where | |
2306 | # commit _must_ be a valid commit in the archive. | |
2307 | # Otherwise, it'll return the max revision (whether or not the | |
2308 | # commit is valid or just a 0x40 placeholder). | |
2309 | sub rev_map_max { | |
2310 | my ($self, $want_commit) = @_; | |
2311 | $self->rebuild; | |
2312 | my ($r, $c) = $self->rev_map_max_norebuild($want_commit); | |
2313 | $want_commit ? ($r, $c) : $r; | |
2314 | } | |
2315 | ||
2316 | sub rev_map_max_norebuild { | |
2317 | my ($self, $want_commit) = @_; | |
2318 | my $map_path = $self->map_path; | |
2319 | stat $map_path or return $want_commit ? (0, undef) : 0; | |
2320 | sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!"; | |
2321 | binmode $fh or croak "binmode: $!"; | |
2322 | my $size = (stat($fh))[7]; | |
2323 | ($size % 24) == 0 or croak "inconsistent size: $size"; | |
2324 | ||
2325 | if ($size == 0) { | |
2326 | close $fh or croak "close: $!"; | |
2327 | return $want_commit ? (0, undef) : 0; | |
2328 | } | |
2329 | ||
2330 | sysseek($fh, -24, SEEK_END) or croak "seek: $!"; | |
2331 | sysread($fh, my $buf, 24) == 24 or croak "read: $!"; | |
2332 | my ($r, $c) = unpack(rev_map_fmt, $buf); | |
2333 | if ($want_commit && $c eq ('0' x40)) { | |
2334 | if ($size < 48) { | |
2335 | return $want_commit ? (0, undef) : 0; | |
2336 | } | |
2337 | sysseek($fh, -48, SEEK_END) or croak "seek: $!"; | |
2338 | sysread($fh, $buf, 24) == 24 or croak "read: $!"; | |
2339 | ($r, $c) = unpack(rev_map_fmt, $buf); | |
2340 | if ($c eq ('0'x40)) { | |
2341 | croak "Penultimate record is all-zeroes in $map_path"; | |
2342 | } | |
2343 | } | |
2344 | close $fh or croak "close: $!"; | |
2345 | $want_commit ? ($r, $c) : $r; | |
2346 | } | |
2347 | ||
2348 | sub rev_map_get { | |
2349 | my ($self, $rev, $uuid) = @_; | |
2350 | my $map_path = $self->map_path($uuid); | |
2351 | return undef unless -e $map_path; | |
2352 | ||
2353 | sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!"; | |
2354 | my $c = _rev_map_get($fh, $rev); | |
2355 | close($fh) or croak "close: $!"; | |
2356 | $c | |
2357 | } | |
2358 | ||
2359 | sub _rev_map_get { | |
2360 | my ($fh, $rev) = @_; | |
2361 | ||
2362 | binmode $fh or croak "binmode: $!"; | |
2363 | my $size = (stat($fh))[7]; | |
2364 | ($size % 24) == 0 or croak "inconsistent size: $size"; | |
2365 | ||
2366 | if ($size == 0) { | |
2367 | return undef; | |
2368 | } | |
2369 | ||
2370 | my ($l, $u) = (0, $size - 24); | |
2371 | my ($r, $c, $buf); | |
2372 | ||
2373 | while ($l <= $u) { | |
2374 | my $i = int(($l/24 + $u/24) / 2) * 24; | |
2375 | sysseek($fh, $i, SEEK_SET) or croak "seek: $!"; | |
2376 | sysread($fh, my $buf, 24) == 24 or croak "read: $!"; | |
2377 | my ($r, $c) = unpack(rev_map_fmt, $buf); | |
2378 | ||
2379 | if ($r < $rev) { | |
2380 | $l = $i + 24; | |
2381 | } elsif ($r > $rev) { | |
2382 | $u = $i - 24; | |
2383 | } else { # $r == $rev | |
2384 | return $c eq ('0' x 40) ? undef : $c; | |
2385 | } | |
2386 | } | |
2387 | undef; | |
2388 | } | |
2389 | ||
2390 | # Finds the first svn revision that exists on (if $eq_ok is true) or | |
2391 | # before $rev for the current branch. It will not search any lower | |
2392 | # than $min_rev. Returns the git commit hash and svn revision number | |
2393 | # if found, else (undef, undef). | |
2394 | sub find_rev_before { | |
2395 | my ($self, $rev, $eq_ok, $min_rev) = @_; | |
2396 | --$rev unless $eq_ok; | |
2397 | $min_rev ||= 1; | |
2398 | my $max_rev = $self->rev_map_max; | |
2399 | $rev = $max_rev if ($rev > $max_rev); | |
2400 | while ($rev >= $min_rev) { | |
2401 | if (my $c = $self->rev_map_get($rev)) { | |
2402 | return ($rev, $c); | |
2403 | } | |
2404 | --$rev; | |
2405 | } | |
2406 | return (undef, undef); | |
2407 | } | |
2408 | ||
2409 | # Finds the first svn revision that exists on (if $eq_ok is true) or | |
2410 | # after $rev for the current branch. It will not search any higher | |
2411 | # than $max_rev. Returns the git commit hash and svn revision number | |
2412 | # if found, else (undef, undef). | |
2413 | sub find_rev_after { | |
2414 | my ($self, $rev, $eq_ok, $max_rev) = @_; | |
2415 | ++$rev unless $eq_ok; | |
2416 | $max_rev ||= $self->rev_map_max; | |
2417 | while ($rev <= $max_rev) { | |
2418 | if (my $c = $self->rev_map_get($rev)) { | |
2419 | return ($rev, $c); | |
2420 | } | |
2421 | ++$rev; | |
2422 | } | |
2423 | return (undef, undef); | |
2424 | } | |
2425 | ||
2426 | sub _new { | |
2427 | my ($class, $repo_id, $ref_id, $path) = @_; | |
2428 | unless (defined $repo_id && length $repo_id) { | |
2429 | $repo_id = $default_repo_id; | |
2430 | } | |
2431 | unless (defined $ref_id && length $ref_id) { | |
2432 | # Access the prefix option from the git-svn main program if it's loaded. | |
2433 | my $prefix = defined &::opt_prefix ? ::opt_prefix() : ""; | |
2434 | $_[2] = $ref_id = | |
2435 | "refs/remotes/$prefix$default_ref_id"; | |
2436 | } | |
2437 | $_[1] = $repo_id; | |
2438 | my $dir = "$ENV{GIT_DIR}/svn/$ref_id"; | |
2439 | ||
2440 | # Older repos imported by us used $GIT_DIR/svn/foo instead of | |
2441 | # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo | |
9a2bb059 | 2442 | if ($ref_id =~ m{^refs/remotes/(.+)}) { |
29499c0b MS |
2443 | my $old_dir = "$ENV{GIT_DIR}/svn/$1"; |
2444 | if (-d $old_dir && ! -d $dir) { | |
2445 | $dir = $old_dir; | |
2446 | } | |
2447 | } | |
2448 | ||
2449 | $_[3] = $path = '' unless (defined $path); | |
2450 | mkpath([$dir]); | |
5578ed74 | 2451 | my $obj = bless { |
29499c0b | 2452 | ref_id => $ref_id, dir => $dir, index => "$dir/index", |
5578ed74 | 2453 | config => "$ENV{GIT_DIR}/svn/config", |
29499c0b | 2454 | map_root => "$dir/.rev_map", repo_id => $repo_id }, $class; |
5578ed74 MS |
2455 | |
2456 | # Ensure it gets canonicalized | |
2457 | $obj->path($path); | |
2458 | ||
2459 | return $obj; | |
2460 | } | |
2461 | ||
2462 | sub path { | |
2463 | my $self = shift; | |
2464 | ||
2465 | if (@_) { | |
2466 | my $path = shift; | |
52de6fa2 | 2467 | $self->{_path} = canonicalize_path($path); |
5578ed74 MS |
2468 | return; |
2469 | } | |
2470 | ||
52de6fa2 | 2471 | return $self->{_path}; |
29499c0b MS |
2472 | } |
2473 | ||
06ee19e8 MS |
2474 | sub url { |
2475 | my $self = shift; | |
2476 | ||
2477 | if (@_) { | |
2478 | my $url = shift; | |
565e56c2 | 2479 | $self->{url} = canonicalize_url($url); |
06ee19e8 MS |
2480 | return; |
2481 | } | |
2482 | ||
2483 | return $self->{url}; | |
29499c0b MS |
2484 | } |
2485 | ||
2486 | # for read-only access of old .rev_db formats | |
2487 | sub unlink_rev_db_symlink { | |
2488 | my ($self) = @_; | |
2489 | my $link = $self->rev_db_path; | |
2490 | $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link"; | |
2491 | if (-l $link) { | |
2492 | unlink $link or croak "unlink: $link failed!"; | |
2493 | } | |
2494 | } | |
2495 | ||
2496 | sub rev_db_path { | |
2497 | my ($self, $uuid) = @_; | |
2498 | my $db_path = $self->map_path($uuid); | |
2499 | $db_path =~ s{/\.rev_map\.}{/\.rev_db\.} | |
2500 | or croak "map_path: $db_path does not contain '/.rev_map.' !"; | |
2501 | $db_path; | |
2502 | } | |
2503 | ||
2504 | # the new replacement for .rev_db | |
2505 | sub map_path { | |
2506 | my ($self, $uuid) = @_; | |
2507 | $uuid ||= $self->ra_uuid; | |
2508 | "$self->{map_root}.$uuid"; | |
2509 | } | |
2510 | ||
2511 | sub uri_encode { | |
2512 | my ($f) = @_; | |
1b67bef2 | 2513 | $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#sprintf("%%%02X",ord($1))#eg; |
29499c0b MS |
2514 | $f |
2515 | } | |
2516 | ||
2517 | sub uri_decode { | |
2518 | my ($f) = @_; | |
2519 | $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg; | |
2520 | $f | |
2521 | } | |
2522 | ||
2523 | sub remove_username { | |
2524 | $_[0] =~ s{^([^:]*://)[^@]+@}{$1}; | |
2525 | } | |
2526 | ||
2527 | 1; |