]>
Commit | Line | Data |
---|---|---|
d3968363 ML |
1 | #!/usr/bin/perl -w |
2 | # | |
3 | # This tool is copyright (c) 2005, Martin Langhoff. | |
4 | # It is released under the Gnu Public License, version 2. | |
5 | # | |
6 | # The basic idea is to walk the output of tla abrowse, | |
7 | # fetch the changesets and apply them. | |
8 | # | |
9 | =head1 Invocation | |
10 | ||
215a7ad1 | 11 | git-archimport -i <archive>/<branch> [<archive>/<branch>] |
d3968363 ML |
12 | [ <archive>/<branch> ] |
13 | ||
14 | The script expects you to provide the key roots where it can start the | |
15 | import from an 'initial import' or 'tag' type of Arch commit. It will | |
16 | then follow all the branching and tagging within the provided roots. | |
17 | ||
18 | It will die if it sees branches that have different roots. | |
19 | ||
20 | =head2 TODO | |
21 | ||
22 | - keep track of merged patches, and mark a git merge when it happens | |
23 | - smarter rules to parse the archive history "up" and "down" | |
24 | - be able to continue an import where we left off | |
25 | - audit shell-escaping of filenames | |
26 | ||
27 | =head1 Devel tricks | |
28 | ||
29 | Add print in front of the shell commands invoked via backticks. | |
30 | ||
31 | =cut | |
32 | ||
33 | use strict; | |
34 | use warnings; | |
35 | use Getopt::Std; | |
36 | use File::Spec; | |
37 | use File::Temp qw(tempfile); | |
38 | use File::Path qw(mkpath); | |
39 | use File::Basename qw(basename dirname); | |
40 | use String::ShellQuote; | |
41 | use Time::Local; | |
42 | use IO::Socket; | |
43 | use IO::Pipe; | |
44 | use POSIX qw(strftime dup2); | |
45 | use Data::Dumper qw/ Dumper /; | |
46 | use IPC::Open2; | |
47 | ||
48 | $SIG{'PIPE'}="IGNORE"; | |
49 | $ENV{'TZ'}="UTC"; | |
50 | ||
51 | our($opt_h,$opt_v, $opt_T, | |
3292ae47 | 52 | $opt_C,$opt_t); |
d3968363 ML |
53 | |
54 | sub usage() { | |
55 | print STDERR <<END; | |
56 | Usage: ${\basename $0} # fetch/update GIT from Arch | |
3292ae47 | 57 | [ -h ] [ -v ] [ -T ] |
d3968363 ML |
58 | [ -C GIT_repository ] [ -t tempdir ] |
59 | repository/arch-branch [ repository/arch-branch] ... | |
60 | END | |
61 | exit(1); | |
62 | } | |
63 | ||
b779d5f0 | 64 | getopts("ThviC:t:") or usage(); |
d3968363 ML |
65 | usage if $opt_h; |
66 | ||
67 | @ARGV >= 1 or usage(); | |
68 | my @arch_roots = @ARGV; | |
69 | ||
70 | my $tmp = $opt_t; | |
71 | $tmp ||= '/tmp'; | |
72 | $tmp .= '/git-archimport/'; | |
73 | ||
74 | my $git_tree = $opt_C; | |
75 | $git_tree ||= "."; | |
76 | ||
77 | ||
78 | my @psets = (); # the collection | |
b779d5f0 ML |
79 | my %psets = (); # the collection, by name |
80 | ||
81 | my %rptags = (); # my reverse private tags | |
82 | # to map a SHA1 to a commitid | |
d3968363 ML |
83 | |
84 | foreach my $root (@arch_roots) { | |
85 | my ($arepo, $abranch) = split(m!/!, $root); | |
86 | open ABROWSE, "tla abrowse -f -A $arepo --desc --merges $abranch |" | |
87 | or die "Problems with tla abrowse: $!"; | |
88 | ||
89 | my %ps = (); # the current one | |
90 | my $mode = ''; | |
91 | my $lastseen = ''; | |
92 | ||
93 | while (<ABROWSE>) { | |
94 | chomp; | |
95 | ||
96 | # first record padded w 8 spaces | |
97 | if (s/^\s{8}\b//) { | |
98 | ||
99 | # store the record we just captured | |
100 | if (%ps) { | |
101 | my %temp = %ps; # break references | |
102 | push (@psets, \%temp); | |
b779d5f0 | 103 | $psets{$temp{id}} = \%temp; |
d3968363 ML |
104 | %ps = (); |
105 | } | |
106 | ||
107 | my ($id, $type) = split(m/\s{3}/, $_); | |
108 | $ps{id} = $id; | |
109 | $ps{repo} = $arepo; | |
110 | ||
111 | # deal with types | |
112 | if ($type =~ m/^\(simple changeset\)/) { | |
113 | $ps{type} = 's'; | |
114 | } elsif ($type eq '(initial import)') { | |
115 | $ps{type} = 'i'; | |
116 | } elsif ($type =~ m/^\(tag revision of (.+)\)/) { | |
117 | $ps{type} = 't'; | |
118 | $ps{tag} = $1; | |
119 | } else { | |
120 | warn "Unknown type $type"; | |
121 | } | |
122 | $lastseen = 'id'; | |
123 | } | |
124 | ||
125 | if (s/^\s{10}//) { | |
126 | # 10 leading spaces or more | |
127 | # indicate commit metadata | |
128 | ||
129 | # date & author | |
130 | if ($lastseen eq 'id' && m/^\d{4}-\d{2}-\d{2}/) { | |
131 | ||
132 | my ($date, $authoremail) = split(m/\s{2,}/, $_); | |
133 | $ps{date} = $date; | |
134 | $ps{date} =~ s/\bGMT$//; # strip off trailign GMT | |
135 | if ($ps{date} =~ m/\b\w+$/) { | |
136 | warn 'Arch dates not in GMT?! - imported dates will be wrong'; | |
137 | } | |
138 | ||
139 | $authoremail =~ m/^(.+)\s(\S+)$/; | |
140 | $ps{author} = $1; | |
141 | $ps{email} = $2; | |
142 | ||
143 | $lastseen = 'date'; | |
144 | ||
145 | } elsif ($lastseen eq 'date') { | |
146 | # the only hint is position | |
147 | # subject is after date | |
148 | $ps{subj} = $_; | |
149 | $lastseen = 'subj'; | |
150 | ||
151 | } elsif ($lastseen eq 'subj' && $_ eq 'merges in:') { | |
152 | $ps{merges} = []; | |
153 | $lastseen = 'merges'; | |
154 | ||
155 | } elsif ($lastseen eq 'merges' && s/^\s{2}//) { | |
156 | push (@{$ps{merges}}, $_); | |
157 | } else { | |
158 | warn 'more metadata after merges!?'; | |
159 | } | |
160 | ||
161 | } | |
162 | } | |
163 | ||
164 | if (%ps) { | |
165 | my %temp = %ps; # break references | |
b779d5f0 ML |
166 | push (@psets, \%temp); |
167 | $psets{ $temp{id} } = \%temp; | |
d3968363 ML |
168 | %ps = (); |
169 | } | |
170 | close ABROWSE; | |
171 | } # end foreach $root | |
172 | ||
173 | ## Order patches by time | |
174 | @psets = sort {$a->{date}.$b->{id} cmp $b->{date}.$b->{id}} @psets; | |
175 | ||
176 | #print Dumper \@psets; | |
177 | ||
178 | ## | |
179 | ## TODO cleanup irrelevant patches | |
180 | ## and put an initial import | |
181 | ## or a full tag | |
3292ae47 ML |
182 | my $import = 0; |
183 | unless (-d '.git') { # initial import | |
d3968363 ML |
184 | if ($psets[0]{type} eq 'i' || $psets[0]{type} eq 't') { |
185 | print "Starting import from $psets[0]{id}\n"; | |
3292ae47 ML |
186 | `git-init-db`; |
187 | die $! if $?; | |
188 | $import = 1; | |
d3968363 ML |
189 | } else { |
190 | die "Need to start from an import or a tag -- cannot use $psets[0]{id}"; | |
191 | } | |
b779d5f0 ML |
192 | } else { # progressing an import |
193 | # load the rptags | |
194 | opendir(DIR, ".git/archimport/tags") | |
195 | || die "can't opendir: $!"; | |
196 | while (my $file = readdir(DIR)) { | |
197 | # skip non-interesting-files | |
198 | next unless -f ".git/archimport/tags/$file"; | |
199 | next if $file =~ m/--base-0$/; # don't care for base-0 | |
200 | my $sha = ptag($file); | |
201 | chomp $sha; | |
202 | # reconvert the 3rd '--' sequence from the end | |
203 | # into a slash | |
204 | # $file = reverse $file; | |
205 | # $file =~ s!^(.+?--.+?--.+?--.+?)--(.+)$!$1/$2!; | |
206 | # $file = reverse $file; | |
207 | $rptags{$sha} = $file; | |
208 | } | |
209 | closedir DIR; | |
d3968363 ML |
210 | } |
211 | ||
3292ae47 | 212 | # process patchsets |
d3968363 ML |
213 | foreach my $ps (@psets) { |
214 | ||
215 | $ps->{branch} = branchname($ps->{id}); | |
216 | ||
217 | # | |
218 | # ensure we have a clean state | |
219 | # | |
220 | if (`git diff-files`) { | |
221 | die "Unclean tree when about to process $ps->{id} " . | |
222 | " - did we fail to commit cleanly before?"; | |
223 | } | |
224 | die $! if $?; | |
225 | ||
3292ae47 ML |
226 | # |
227 | # skip commits already in repo | |
228 | # | |
229 | if (ptag($ps->{id})) { | |
230 | $opt_v && print "Skipping already imported: $ps->{id}\n"; | |
231 | next; | |
232 | } | |
233 | ||
d3968363 ML |
234 | # |
235 | # create the branch if needed | |
236 | # | |
3292ae47 ML |
237 | if ($ps->{type} eq 'i' && !$import) { |
238 | die "Should not have more than one 'Initial import' per GIT import: $ps->{id}"; | |
d3968363 ML |
239 | } |
240 | ||
3292ae47 | 241 | unless ($import) { # skip for import |
d3968363 ML |
242 | if ( -e ".git/refs/heads/$ps->{branch}") { |
243 | # we know about this branch | |
244 | `git checkout $ps->{branch}`; | |
245 | } else { | |
246 | # new branch! we need to verify a few things | |
247 | die "Branch on a non-tag!" unless $ps->{type} eq 't'; | |
248 | my $branchpoint = ptag($ps->{tag}); | |
249 | die "Tagging from unknown id unsupported: $ps->{tag}" | |
250 | unless $branchpoint; | |
251 | ||
252 | # find where we are supposed to branch from | |
253 | `git checkout -b $ps->{branch} $branchpoint`; | |
52586ecb ML |
254 | |
255 | # If we trust Arch with the fact that this is just | |
256 | # a tag, and it does not affect the state of the tree | |
257 | # then we just tag and move on | |
258 | tag($ps->{id}, $branchpoint); | |
259 | ptag($ps->{id}, $branchpoint); | |
260 | print " * Tagged $ps->{id} at $branchpoint\n"; | |
261 | next; | |
d3968363 ML |
262 | } |
263 | die $! if $?; | |
264 | } | |
265 | ||
d3968363 ML |
266 | # |
267 | # Apply the import/changeset/merge into the working tree | |
268 | # | |
269 | if ($ps->{type} eq 'i' || $ps->{type} eq 't') { | |
d3968363 | 270 | apply_import($ps) or die $!; |
3292ae47 | 271 | $import=0; |
d3968363 ML |
272 | } elsif ($ps->{type} eq 's') { |
273 | apply_cset($ps); | |
274 | } | |
275 | ||
276 | # | |
277 | # prepare update git's index, based on what arch knows | |
278 | # about the pset, resolve parents, etc | |
279 | # | |
280 | my $tree; | |
281 | ||
282 | my $commitlog = `tla cat-archive-log -A $ps->{repo} $ps->{id}`; | |
283 | die "Error in cat-archive-log: $!" if $?; | |
284 | ||
285 | # parselog will git-add/rm files | |
286 | # and generally prepare things for the commit | |
287 | # NOTE: parselog will shell-quote filenames! | |
288 | my ($sum, $msg, $add, $del, $mod, $ren) = parselog($commitlog); | |
289 | my $logmessage = "$sum\n$msg"; | |
290 | ||
291 | ||
292 | # imports don't give us good info | |
293 | # on added files. Shame on them | |
294 | if ($ps->{type} eq 'i' || $ps->{type} eq 't') { | |
215a7ad1 JH |
295 | `find . -type f -print0 | grep -zv '^./.git' | xargs -0 -l100 git-update-index --add`; |
296 | `git-ls-files --deleted -z | xargs --no-run-if-empty -0 -l100 git-update-index --remove`; | |
d3968363 ML |
297 | } |
298 | ||
299 | if (@$add) { | |
300 | while (@$add) { | |
301 | my @slice = splice(@$add, 0, 100); | |
302 | my $slice = join(' ', @slice); | |
215a7ad1 JH |
303 | `git-update-index --add $slice`; |
304 | die "Error in git-update-index --add: $!" if $?; | |
d3968363 ML |
305 | } |
306 | } | |
307 | if (@$del) { | |
308 | foreach my $file (@$del) { | |
309 | unlink $file or die "Problems deleting $file : $!"; | |
310 | } | |
311 | while (@$del) { | |
312 | my @slice = splice(@$del, 0, 100); | |
313 | my $slice = join(' ', @slice); | |
215a7ad1 JH |
314 | `git-update-index --remove $slice`; |
315 | die "Error in git-update-index --remove: $!" if $?; | |
d3968363 ML |
316 | } |
317 | } | |
318 | if (@$ren) { # renamed | |
319 | if (@$ren % 2) { | |
320 | die "Odd number of entries in rename!?"; | |
321 | } | |
322 | ; | |
323 | while (@$ren) { | |
324 | my $from = pop @$ren; | |
325 | my $to = pop @$ren; | |
326 | ||
327 | unless (-d dirname($to)) { | |
328 | mkpath(dirname($to)); # will die on err | |
329 | } | |
330 | #print "moving $from $to"; | |
331 | `mv $from $to`; | |
332 | die "Error renaming $from $to : $!" if $?; | |
215a7ad1 JH |
333 | `git-update-index --remove $from`; |
334 | die "Error in git-update-index --remove: $!" if $?; | |
335 | `git-update-index --add $to`; | |
336 | die "Error in git-update-index --add: $!" if $?; | |
d3968363 ML |
337 | } |
338 | ||
339 | } | |
340 | if (@$mod) { # must be _after_ renames | |
341 | while (@$mod) { | |
342 | my @slice = splice(@$mod, 0, 100); | |
343 | my $slice = join(' ', @slice); | |
215a7ad1 JH |
344 | `git-update-index $slice`; |
345 | die "Error in git-update-index: $!" if $?; | |
d3968363 ML |
346 | } |
347 | } | |
348 | ||
215a7ad1 | 349 | # warn "errors when running git-update-index! $!"; |
d3968363 ML |
350 | $tree = `git-write-tree`; |
351 | die "cannot write tree $!" if $?; | |
352 | chomp $tree; | |
353 | ||
354 | ||
355 | # | |
356 | # Who's your daddy? | |
357 | # | |
358 | my @par; | |
359 | if ( -e ".git/refs/heads/$ps->{branch}") { | |
360 | if (open HEAD, "<.git/refs/heads/$ps->{branch}") { | |
361 | my $p = <HEAD>; | |
362 | close HEAD; | |
363 | chomp $p; | |
364 | push @par, '-p', $p; | |
365 | } else { | |
366 | if ($ps->{type} eq 's') { | |
367 | warn "Could not find the right head for the branch $ps->{branch}"; | |
368 | } | |
369 | } | |
370 | } | |
371 | ||
b779d5f0 ML |
372 | if ($ps->{merges}) { |
373 | push @par, find_parents($ps); | |
374 | } | |
d3968363 ML |
375 | my $par = join (' ', @par); |
376 | ||
377 | # | |
378 | # Commit, tag and clean state | |
379 | # | |
380 | $ENV{TZ} = 'GMT'; | |
381 | $ENV{GIT_AUTHOR_NAME} = $ps->{author}; | |
382 | $ENV{GIT_AUTHOR_EMAIL} = $ps->{email}; | |
383 | $ENV{GIT_AUTHOR_DATE} = $ps->{date}; | |
384 | $ENV{GIT_COMMITTER_NAME} = $ps->{author}; | |
385 | $ENV{GIT_COMMITTER_EMAIL} = $ps->{email}; | |
386 | $ENV{GIT_COMMITTER_DATE} = $ps->{date}; | |
387 | ||
388 | my ($pid, $commit_rh, $commit_wh); | |
389 | $commit_rh = 'commit_rh'; | |
390 | $commit_wh = 'commit_wh'; | |
391 | ||
392 | $pid = open2(*READER, *WRITER, "git-commit-tree $tree $par") | |
393 | or die $!; | |
394 | print WRITER $logmessage; # write | |
395 | close WRITER; | |
396 | my $commitid = <READER>; # read | |
397 | chomp $commitid; | |
398 | close READER; | |
399 | waitpid $pid,0; # close; | |
400 | ||
401 | if (length $commitid != 40) { | |
402 | die "Something went wrong with the commit! $! $commitid"; | |
403 | } | |
404 | # | |
405 | # Update the branch | |
406 | # | |
407 | open HEAD, ">.git/refs/heads/$ps->{branch}"; | |
408 | print HEAD $commitid; | |
409 | close HEAD; | |
410 | unlink ('.git/HEAD'); | |
411 | symlink("refs/heads/$ps->{branch}",".git/HEAD"); | |
412 | ||
413 | # tag accordingly | |
414 | ptag($ps->{id}, $commitid); # private tag | |
415 | if ($opt_T || $ps->{type} eq 't' || $ps->{type} eq 'i') { | |
416 | tag($ps->{id}, $commitid); | |
417 | } | |
418 | print " * Committed $ps->{id}\n"; | |
419 | print " + tree $tree\n"; | |
420 | print " + commit $commitid\n"; | |
b779d5f0 ML |
421 | $opt_v && print " + commit date is $ps->{date} \n"; |
422 | $opt_v && print " + parents: $par \n"; | |
d3968363 ML |
423 | } |
424 | ||
425 | sub branchname { | |
426 | my $id = shift; | |
427 | $id =~ s#^.+?/##; | |
428 | my @parts = split(m/--/, $id); | |
429 | return join('--', @parts[0..1]); | |
430 | } | |
431 | ||
432 | sub apply_import { | |
433 | my $ps = shift; | |
434 | my $bname = branchname($ps->{id}); | |
435 | ||
436 | `mkdir -p $tmp`; | |
437 | ||
438 | `tla get -s --no-pristine -A $ps->{repo} $ps->{id} $tmp/import`; | |
439 | die "Cannot get import: $!" if $?; | |
440 | `rsync -v --archive --delete --exclude '.git' --exclude '.arch-ids' --exclude '{arch}' $tmp/import/* ./`; | |
441 | die "Cannot rsync import:$!" if $?; | |
442 | ||
443 | `rm -fr $tmp/import`; | |
444 | die "Cannot remove tempdir: $!" if $?; | |
445 | ||
446 | ||
447 | return 1; | |
448 | } | |
449 | ||
450 | sub apply_cset { | |
451 | my $ps = shift; | |
452 | ||
453 | `mkdir -p $tmp`; | |
454 | ||
455 | # get the changeset | |
456 | `tla get-changeset -A $ps->{repo} $ps->{id} $tmp/changeset`; | |
457 | die "Cannot get changeset: $!" if $?; | |
458 | ||
459 | # apply patches | |
460 | if (`find $tmp/changeset/patches -type f -name '*.patch'`) { | |
461 | # this can be sped up considerably by doing | |
462 | # (find | xargs cat) | patch | |
463 | # but that cna get mucked up by patches | |
464 | # with missing trailing newlines or the standard | |
465 | # 'missing newline' flag in the patch - possibly | |
466 | # produced with an old/buggy diff. | |
467 | # slow and safe, we invoke patch once per patchfile | |
468 | `find $tmp/changeset/patches -type f -name '*.patch' -print0 | grep -zv '{arch}' | xargs -iFILE -0 --no-run-if-empty patch -p1 --forward -iFILE`; | |
469 | die "Problem applying patches! $!" if $?; | |
470 | } | |
471 | ||
472 | # apply changed binary files | |
473 | if (my @modified = `find $tmp/changeset/patches -type f -name '*.modified'`) { | |
474 | foreach my $mod (@modified) { | |
475 | chomp $mod; | |
476 | my $orig = $mod; | |
477 | $orig =~ s/\.modified$//; # lazy | |
478 | $orig =~ s!^\Q$tmp\E/changeset/patches/!!; | |
479 | #print "rsync -p '$mod' '$orig'"; | |
480 | `rsync -p $mod ./$orig`; | |
481 | die "Problem applying binary changes! $!" if $?; | |
482 | } | |
483 | } | |
484 | ||
485 | # bring in new files | |
486 | `rsync --archive --exclude '.git' --exclude '.arch-ids' --exclude '{arch}' $tmp/changeset/new-files-archive/* ./`; | |
487 | ||
488 | # deleted files are hinted from the commitlog processing | |
489 | ||
490 | `rm -fr $tmp/changeset`; | |
491 | } | |
492 | ||
493 | ||
494 | # =for reference | |
495 | # A log entry looks like | |
496 | # Revision: moodle-org--moodle--1.3.3--patch-15 | |
497 | # Archive: arch-eduforge@catalyst.net.nz--2004 | |
498 | # Creator: Penny Leach <penny@catalyst.net.nz> | |
499 | # Date: Wed May 25 14:15:34 NZST 2005 | |
500 | # Standard-date: 2005-05-25 02:15:34 GMT | |
501 | # New-files: lang/de/.arch-ids/block_glossary_random.php.id | |
502 | # lang/de/.arch-ids/block_html.php.id | |
503 | # New-directories: lang/de/help/questionnaire | |
504 | # lang/de/help/questionnaire/.arch-ids | |
505 | # Renamed-files: .arch-ids/db_sears.sql.id db/.arch-ids/db_sears.sql.id | |
506 | # db_sears.sql db/db_sears.sql | |
507 | # Removed-files: lang/be/docs/.arch-ids/release.html.id | |
508 | # lang/be/docs/.arch-ids/releaseold.html.id | |
509 | # Modified-files: admin/cron.php admin/delete.php | |
510 | # admin/editor.html backup/lib.php backup/restore.php | |
511 | # New-patches: arch-eduforge@catalyst.net.nz--2004/moodle-org--moodle--1.3.3--patch-15 | |
512 | # Summary: Updating to latest from MOODLE_14_STABLE (1.4.5+) | |
513 | # Keywords: | |
514 | # | |
515 | # Updating yadda tadda tadda madda | |
516 | sub parselog { | |
517 | my $log = shift; | |
518 | #print $log; | |
519 | ||
520 | my (@add, @del, @mod, @ren, @kw, $sum, $msg ); | |
521 | ||
522 | if ($log =~ m/(?:\n|^)New-files:(.*?)(?=\n\w)/s ) { | |
523 | my $files = $1; | |
524 | @add = split(m/\s+/s, $files); | |
525 | } | |
526 | ||
527 | if ($log =~ m/(?:\n|^)Removed-files:(.*?)(?=\n\w)/s ) { | |
528 | my $files = $1; | |
529 | @del = split(m/\s+/s, $files); | |
530 | } | |
531 | ||
532 | if ($log =~ m/(?:\n|^)Modified-files:(.*?)(?=\n\w)/s ) { | |
533 | my $files = $1; | |
534 | @mod = split(m/\s+/s, $files); | |
535 | } | |
536 | ||
537 | if ($log =~ m/(?:\n|^)Renamed-files:(.*?)(?=\n\w)/s ) { | |
538 | my $files = $1; | |
539 | @ren = split(m/\s+/s, $files); | |
540 | } | |
541 | ||
542 | $sum =''; | |
543 | if ($log =~ m/^Summary:(.+?)$/m ) { | |
544 | $sum = $1; | |
545 | $sum =~ s/^\s+//; | |
546 | $sum =~ s/\s+$//; | |
547 | } | |
548 | ||
549 | $msg = ''; | |
550 | if ($log =~ m/\n\n(.+)$/s) { | |
551 | $msg = $1; | |
552 | $msg =~ s/^\s+//; | |
553 | $msg =~ s/\s+$//; | |
554 | } | |
555 | ||
556 | ||
557 | # cleanup the arrays | |
558 | foreach my $ref ( (\@add, \@del, \@mod, \@ren) ) { | |
559 | my @tmp = (); | |
560 | while (my $t = pop @$ref) { | |
561 | next unless length ($t); | |
562 | next if $t =~ m!\{arch\}/!; | |
563 | next if $t =~ m!\.arch-ids/!; | |
564 | next if $t =~ m!\.arch-inventory$!; | |
565 | push (@tmp, shell_quote($t)); | |
566 | } | |
567 | @$ref = @tmp; | |
568 | } | |
569 | ||
570 | #print Dumper [$sum, $msg, \@add, \@del, \@mod, \@ren]; | |
571 | return ($sum, $msg, \@add, \@del, \@mod, \@ren); | |
572 | } | |
573 | ||
574 | # write/read a tag | |
575 | sub tag { | |
576 | my ($tag, $commit) = @_; | |
577 | $tag =~ s|/|--|g; | |
578 | $tag = shell_quote($tag); | |
579 | ||
580 | if ($commit) { | |
581 | open(C,">.git/refs/tags/$tag") | |
582 | or die "Cannot create tag $tag: $!\n"; | |
583 | print C "$commit\n" | |
584 | or die "Cannot write tag $tag: $!\n"; | |
585 | close(C) | |
586 | or die "Cannot write tag $tag: $!\n"; | |
b779d5f0 | 587 | print " * Created tag ' $tag' on '$commit'\n" if $opt_v; |
d3968363 ML |
588 | } else { # read |
589 | open(C,"<.git/refs/tags/$tag") | |
590 | or die "Cannot read tag $tag: $!\n"; | |
591 | $commit = <C>; | |
592 | chomp $commit; | |
593 | die "Error reading tag $tag: $!\n" unless length $commit == 40; | |
594 | close(C) | |
595 | or die "Cannot read tag $tag: $!\n"; | |
596 | return $commit; | |
597 | } | |
598 | } | |
599 | ||
600 | # write/read a private tag | |
601 | # reads fail softly if the tag isn't there | |
602 | sub ptag { | |
603 | my ($tag, $commit) = @_; | |
604 | $tag =~ s|/|--|g; | |
605 | $tag = shell_quote($tag); | |
606 | ||
607 | unless (-d '.git/archimport/tags') { | |
608 | mkpath('.git/archimport/tags'); | |
609 | } | |
610 | ||
611 | if ($commit) { # write | |
612 | open(C,">.git/archimport/tags/$tag") | |
613 | or die "Cannot create tag $tag: $!\n"; | |
614 | print C "$commit\n" | |
615 | or die "Cannot write tag $tag: $!\n"; | |
616 | close(C) | |
617 | or die "Cannot write tag $tag: $!\n"; | |
b779d5f0 ML |
618 | $rptags{$commit} = $tag |
619 | unless $tag =~ m/--base-0$/; | |
d3968363 ML |
620 | } else { # read |
621 | # if the tag isn't there, return 0 | |
622 | unless ( -s ".git/archimport/tags/$tag") { | |
d3968363 ML |
623 | return 0; |
624 | } | |
625 | open(C,"<.git/archimport/tags/$tag") | |
626 | or die "Cannot read tag $tag: $!\n"; | |
627 | $commit = <C>; | |
628 | chomp $commit; | |
629 | die "Error reading tag $tag: $!\n" unless length $commit == 40; | |
630 | close(C) | |
631 | or die "Cannot read tag $tag: $!\n"; | |
b779d5f0 ML |
632 | unless (defined $rptags{$commit}) { |
633 | $rptags{$commit} = $tag; | |
634 | } | |
d3968363 ML |
635 | return $commit; |
636 | } | |
637 | } | |
b779d5f0 ML |
638 | |
639 | sub find_parents { | |
640 | # | |
641 | # Identify what branches are merging into me | |
642 | # and whether we are fully merged | |
643 | # git-merge-base <headsha> <headsha> should tell | |
644 | # me what the base of the merge should be | |
645 | # | |
646 | my $ps = shift; | |
647 | ||
648 | my %branches; # holds an arrayref per branch | |
649 | # the arrayref contains a list of | |
650 | # merged patches between the base | |
651 | # of the merge and the current head | |
652 | ||
653 | my @parents; # parents found for this commit | |
654 | ||
655 | # simple loop to split the merges | |
656 | # per branch | |
657 | foreach my $merge (@{$ps->{merges}}) { | |
658 | my $branch = branchname($merge); | |
659 | unless (defined $branches{$branch} ){ | |
660 | $branches{$branch} = []; | |
661 | } | |
662 | push @{$branches{$branch}}, $merge; | |
663 | } | |
664 | ||
665 | # | |
666 | # foreach branch find a merge base and walk it to the | |
667 | # head where we are, collecting the merged patchsets that | |
668 | # Arch has recorded. Keep that in @have | |
669 | # Compare that with the commits on the other branch | |
670 | # between merge-base and the tip of the branch (@need) | |
671 | # and see if we have a series of consecutive patches | |
672 | # starting from the merge base. The tip of the series | |
673 | # of consecutive patches merged is our new parent for | |
674 | # that branch. | |
675 | # | |
676 | foreach my $branch (keys %branches) { | |
677 | my $mergebase = `git-merge-base $branch $ps->{branch}`; | |
678 | die "Cannot find merge base for $branch and $ps->{branch}" if $?; | |
679 | chomp $mergebase; | |
680 | ||
681 | # now walk up to the mergepoint collecting what patches we have | |
682 | my $branchtip = git_rev_parse($ps->{branch}); | |
683 | my @ancestors = `git-rev-list --merge-order $branchtip ^$mergebase`; | |
684 | my %have; # collected merges this branch has | |
685 | foreach my $merge (@{$ps->{merges}}) { | |
686 | $have{$merge} = 1; | |
687 | } | |
688 | my %ancestorshave; | |
689 | foreach my $par (@ancestors) { | |
690 | $par = commitid2pset($par); | |
691 | if (defined $par->{merges}) { | |
692 | foreach my $merge (@{$par->{merges}}) { | |
693 | $ancestorshave{$merge}=1; | |
694 | } | |
695 | } | |
696 | } | |
697 | # print "++++ Merges in $ps->{id} are....\n"; | |
698 | # my @have = sort keys %have; print Dumper(\@have); | |
699 | ||
700 | # merge what we have with what ancestors have | |
701 | %have = (%have, %ancestorshave); | |
702 | ||
703 | # see what the remote branch has - these are the merges we | |
704 | # will want to have in a consecutive series from the mergebase | |
705 | my $otherbranchtip = git_rev_parse($branch); | |
706 | my @needraw = `git-rev-list --merge-order $otherbranchtip ^$mergebase`; | |
707 | my @need; | |
708 | foreach my $needps (@needraw) { # get the psets | |
709 | $needps = commitid2pset($needps); | |
710 | # git-rev-list will also | |
711 | # list commits merged in via earlier | |
712 | # merges. we are only interested in commits | |
713 | # from the branch we're looking at | |
714 | if ($branch eq $needps->{branch}) { | |
715 | push @need, $needps->{id}; | |
716 | } | |
717 | } | |
718 | ||
719 | # print "++++ Merges from $branch we want are....\n"; | |
720 | # print Dumper(\@need); | |
721 | ||
722 | my $newparent; | |
723 | while (my $needed_commit = pop @need) { | |
724 | if ($have{$needed_commit}) { | |
725 | $newparent = $needed_commit; | |
726 | } else { | |
727 | last; # break out of the while | |
728 | } | |
729 | } | |
730 | if ($newparent) { | |
731 | push @parents, $newparent; | |
732 | } | |
733 | ||
734 | ||
735 | } # end foreach branch | |
736 | ||
737 | # prune redundant parents | |
738 | my %parents; | |
739 | foreach my $p (@parents) { | |
740 | $parents{$p} = 1; | |
741 | } | |
742 | foreach my $p (@parents) { | |
743 | next unless exists $psets{$p}{merges}; | |
744 | next unless ref $psets{$p}{merges}; | |
745 | my @merges = @{$psets{$p}{merges}}; | |
746 | foreach my $merge (@merges) { | |
747 | if ($parents{$merge}) { | |
748 | delete $parents{$merge}; | |
749 | } | |
750 | } | |
751 | } | |
752 | @parents = keys %parents; | |
753 | @parents = map { " -p " . ptag($_) } @parents; | |
754 | return @parents; | |
755 | } | |
756 | ||
757 | sub git_rev_parse { | |
758 | my $name = shift; | |
759 | my $val = `git-rev-parse $name`; | |
760 | die "Error: git-rev-parse $name" if $?; | |
761 | chomp $val; | |
762 | return $val; | |
763 | } | |
764 | ||
765 | # resolve a SHA1 to a known patchset | |
766 | sub commitid2pset { | |
767 | my $commitid = shift; | |
768 | chomp $commitid; | |
769 | my $name = $rptags{$commitid} | |
770 | || die "Cannot find reverse tag mapping for $commitid"; | |
771 | # the keys in %rptag are slightly munged; unmunge | |
772 | # reconvert the 3rd '--' sequence from the end | |
773 | # into a slash | |
774 | $name = reverse $name; | |
775 | $name =~ s!^(.+?--.+?--.+?--.+?)--(.+)$!$1/$2!; | |
776 | $name = reverse $name; | |
777 | my $ps = $psets{$name} | |
778 | || (print Dumper(sort keys %psets)) && die "Cannot find patchset for $name"; | |
779 | return $ps; | |
780 | } |