]>
Commit | Line | Data |
---|---|---|
3397f9df EW |
1 | #!/usr/bin/env perl |
2 | use warnings; | |
3 | use strict; | |
4 | use vars qw/ $AUTHOR $VERSION | |
5 | $SVN_URL $SVN_INFO $SVN_WC | |
6 | $GIT_SVN_INDEX $GIT_SVN | |
7 | $GIT_DIR $REV_DIR/; | |
8 | $AUTHOR = 'Eric Wong <normalperson@yhbt.net>'; | |
9 | $VERSION = '0.9.0'; | |
10 | $GIT_DIR = $ENV{GIT_DIR} || "$ENV{PWD}/.git"; | |
11 | $GIT_SVN = $ENV{GIT_SVN_ID} || 'git-svn'; | |
12 | $GIT_SVN_INDEX = "$GIT_DIR/$GIT_SVN/index"; | |
13 | $ENV{GIT_DIR} ||= $GIT_DIR; | |
14 | $SVN_URL = undef; | |
15 | $REV_DIR = "$GIT_DIR/$GIT_SVN/revs"; | |
16 | $SVN_WC = "$GIT_DIR/$GIT_SVN/tree"; | |
17 | ||
18 | # make sure the svn binary gives consistent output between locales and TZs: | |
19 | $ENV{TZ} = 'UTC'; | |
20 | $ENV{LC_ALL} = 'C'; | |
21 | ||
22 | # If SVN:: library support is added, please make the dependencies | |
23 | # optional and preserve the capability to use the command-line client. | |
24 | # See what I do with XML::Simple to make the dependency optional. | |
25 | use Carp qw/croak/; | |
26 | use IO::File qw//; | |
27 | use File::Basename qw/dirname basename/; | |
28 | use File::Path qw/mkpath/; | |
29 | use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/; | |
30 | use File::Spec qw//; | |
31 | my $sha1 = qr/[a-f\d]{40}/; | |
32 | my $sha1_short = qr/[a-f\d]{6,40}/; | |
72942938 EW |
33 | my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit, |
34 | $_find_copies_harder, $_l); | |
3397f9df EW |
35 | |
36 | GetOptions( 'revision|r=s' => \$_revision, | |
37 | 'no-ignore-externals' => \$_no_ignore_ext, | |
38 | 'stdin|' => \$_stdin, | |
39 | 'edit|e' => \$_edit, | |
40 | 'rmdir' => \$_rmdir, | |
41 | 'help|H|h' => \$_help, | |
72942938 EW |
42 | 'find-copies-harder' => \$_find_copies_harder, |
43 | 'l=i' => \$_l, | |
a18b6327 | 44 | 'no-stop-on-copy' => \$_no_stop_copy ); |
3397f9df EW |
45 | my %cmd = ( |
46 | fetch => [ \&fetch, "Download new revisions from SVN" ], | |
47 | init => [ \&init, "Initialize and fetch (import)"], | |
48 | commit => [ \&commit, "Commit git revisions to SVN" ], | |
49 | rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)" ], | |
50 | help => [ \&usage, "Show help" ], | |
51 | ); | |
52 | my $cmd; | |
53 | for (my $i = 0; $i < @ARGV; $i++) { | |
54 | if (defined $cmd{$ARGV[$i]}) { | |
55 | $cmd = $ARGV[$i]; | |
56 | splice @ARGV, $i, 1; | |
57 | last; | |
58 | } | |
59 | }; | |
60 | ||
61 | # we may be called as git-svn-(command), or git-svn(command). | |
62 | foreach (keys %cmd) { | |
63 | if (/git\-svn\-?($_)(?:\.\w+)?$/) { | |
64 | $cmd = $1; | |
65 | last; | |
66 | } | |
67 | } | |
68 | usage(0) if $_help; | |
69 | usage(1) unless (defined $cmd); | |
70 | svn_check_ignore_externals(); | |
71 | $cmd{$cmd}->[0]->(@ARGV); | |
72 | exit 0; | |
73 | ||
74 | ####################### primary functions ###################### | |
75 | sub usage { | |
76 | my $exit = shift || 0; | |
77 | my $fd = $exit ? \*STDERR : \*STDOUT; | |
78 | print $fd <<""; | |
79 | git-svn - bidirectional operations between a single Subversion tree and git | |
80 | Usage: $0 <command> [options] [arguments]\n | |
81 | Available commands: | |
82 | ||
83 | foreach (sort keys %cmd) { | |
84 | print $fd ' ',pack('A10',$_),$cmd{$_}->[1],"\n"; | |
85 | } | |
86 | print $fd <<""; | |
87 | \nGIT_SVN_ID may be set in the environment to an arbitrary identifier if | |
88 | you're tracking multiple SVN branches/repositories in one git repository | |
89 | and want to keep them separate. | |
90 | ||
91 | exit $exit; | |
92 | } | |
93 | ||
94 | sub rebuild { | |
95 | $SVN_URL = shift or undef; | |
96 | my $repo_uuid; | |
97 | my $newest_rev = 0; | |
98 | ||
99 | my $pid = open(my $rev_list,'-|'); | |
100 | defined $pid or croak $!; | |
101 | if ($pid == 0) { | |
102 | exec("git-rev-list","$GIT_SVN-HEAD") or croak $!; | |
103 | } | |
104 | my $first; | |
105 | while (<$rev_list>) { | |
106 | chomp; | |
107 | my $c = $_; | |
108 | croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o; | |
109 | my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`); | |
110 | next if (!@commit); # skip merges | |
111 | my $id = $commit[$#commit]; | |
112 | my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+) | |
113 | \s([a-f\d\-]+)$/x); | |
114 | if (!$rev || !$uuid || !$url) { | |
115 | # some of the original repositories I made had | |
116 | # indentifiers like this: | |
117 | ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+) | |
118 | \@([a-f\d\-]+)/x); | |
119 | if (!$rev || !$uuid) { | |
120 | croak "Unable to extract revision or UUID from ", | |
121 | "$c, $id\n"; | |
122 | } | |
123 | } | |
124 | print "r$rev = $c\n"; | |
125 | unless (defined $first) { | |
126 | if (!$SVN_URL && !$url) { | |
127 | croak "SVN repository location required: $url\n"; | |
128 | } | |
129 | $SVN_URL ||= $url; | |
130 | $repo_uuid = setup_git_svn(); | |
131 | $first = $rev; | |
132 | } | |
133 | if ($uuid ne $repo_uuid) { | |
134 | croak "Repository UUIDs do not match!\ngot: $uuid\n", | |
135 | "expected: $repo_uuid\n"; | |
136 | } | |
137 | assert_revision_eq_or_unknown($rev, $c); | |
138 | sys('git-update-ref',"$GIT_SVN/revs/$rev",$c); | |
139 | $newest_rev = $rev if ($rev > $newest_rev); | |
140 | } | |
141 | close $rev_list or croak $?; | |
142 | if (!chdir $SVN_WC) { | |
143 | my @svn_co = ('svn','co',"-r$first"); | |
144 | push @svn_co, '--ignore-externals' unless $_no_ignore_ext; | |
145 | sys(@svn_co, $SVN_URL, $SVN_WC); | |
146 | chdir $SVN_WC or croak $!; | |
147 | } | |
148 | ||
149 | $pid = fork; | |
150 | defined $pid or croak $!; | |
151 | if ($pid == 0) { | |
152 | my @svn_up = qw(svn up); | |
153 | push @svn_up, '--ignore-externals' unless $_no_ignore_ext; | |
154 | sys(@svn_up,"-r$newest_rev"); | |
155 | $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX; | |
156 | git_addremove(); | |
157 | exec('git-write-tree'); | |
158 | } | |
159 | waitpid $pid, 0; | |
160 | } | |
161 | ||
162 | sub init { | |
163 | $SVN_URL = shift or croak "SVN repository location required\n"; | |
164 | unless (-d $GIT_DIR) { | |
165 | sys('git-init-db'); | |
166 | } | |
167 | setup_git_svn(); | |
168 | } | |
169 | ||
170 | sub fetch { | |
171 | my (@parents) = @_; | |
172 | $SVN_URL ||= file_to_s("$GIT_DIR/$GIT_SVN/info/url"); | |
173 | my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL); | |
defc6492 EW |
174 | unless ($_revision) { |
175 | $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD'; | |
3397f9df | 176 | } |
defc6492 | 177 | push @log_args, "-r$_revision"; |
3397f9df EW |
178 | push @log_args, '--stop-on-copy' unless $_no_stop_copy; |
179 | ||
180 | eval { require XML::Simple or croak $! }; | |
181 | my $svn_log = $@ ? svn_log_raw(@log_args) : svn_log_xml(@log_args); | |
defc6492 | 182 | @$svn_log = sort { $a->{revision} <=> $b->{revision} } @$svn_log; |
3397f9df EW |
183 | |
184 | my $base = shift @$svn_log or croak "No base revision!\n"; | |
185 | my $last_commit = undef; | |
186 | unless (-d $SVN_WC) { | |
187 | my @svn_co = ('svn','co',"-r$base->{revision}"); | |
188 | push @svn_co,'--ignore-externals' unless $_no_ignore_ext; | |
189 | sys(@svn_co, $SVN_URL, $SVN_WC); | |
190 | chdir $SVN_WC or croak $!; | |
191 | $last_commit = git_commit($base, @parents); | |
192 | unless (-f "$GIT_DIR/refs/heads/master") { | |
193 | sys(qw(git-update-ref refs/heads/master),$last_commit); | |
194 | } | |
195 | assert_svn_wc_clean($base->{revision}, $last_commit); | |
196 | } else { | |
197 | chdir $SVN_WC or croak $!; | |
198 | $last_commit = file_to_s("$REV_DIR/$base->{revision}"); | |
199 | } | |
200 | my @svn_up = qw(svn up); | |
201 | push @svn_up, '--ignore-externals' unless $_no_ignore_ext; | |
202 | my $last_rev = $base->{revision}; | |
203 | foreach my $log_msg (@$svn_log) { | |
204 | assert_svn_wc_clean($last_rev, $last_commit); | |
205 | $last_rev = $log_msg->{revision}; | |
206 | sys(@svn_up,"-r$last_rev"); | |
207 | $last_commit = git_commit($log_msg, $last_commit, @parents); | |
208 | } | |
209 | assert_svn_wc_clean($last_rev, $last_commit); | |
210 | return pop @$svn_log; | |
211 | } | |
212 | ||
213 | sub commit { | |
214 | my (@commits) = @_; | |
215 | if ($_stdin || !@commits) { | |
216 | print "Reading from stdin...\n"; | |
217 | @commits = (); | |
218 | while (<STDIN>) { | |
219 | if (/^([a-f\d]{6,40})\b/) { | |
220 | unshift @commits, $1; | |
221 | } | |
222 | } | |
223 | } | |
224 | my @revs; | |
225 | foreach (@commits) { | |
226 | push @revs, (safe_qx('git-rev-parse',$_)); | |
227 | } | |
228 | chomp @revs; | |
229 | ||
230 | fetch(); | |
231 | chdir $SVN_WC or croak $!; | |
232 | my $svn_current_rev = svn_info('.')->{'Last Changed Rev'}; | |
233 | foreach my $c (@revs) { | |
234 | print "Committing $c\n"; | |
235 | svn_checkout_tree($svn_current_rev, $c); | |
236 | $svn_current_rev = svn_commit_tree($svn_current_rev, $c); | |
237 | } | |
238 | print "Done committing ",scalar @revs," revisions to SVN\n"; | |
239 | ||
240 | } | |
241 | ||
242 | ########################### utility functions ######################### | |
243 | ||
244 | sub setup_git_svn { | |
245 | defined $SVN_URL or croak "SVN repository location required\n"; | |
246 | unless (-d $GIT_DIR) { | |
247 | croak "GIT_DIR=$GIT_DIR does not exist!\n"; | |
248 | } | |
249 | mkpath(["$GIT_DIR/$GIT_SVN"]); | |
250 | mkpath(["$GIT_DIR/$GIT_SVN/info"]); | |
251 | mkpath([$REV_DIR]); | |
252 | s_to_file($SVN_URL,"$GIT_DIR/$GIT_SVN/info/url"); | |
253 | my $uuid = svn_info($SVN_URL)->{'Repository UUID'} or | |
254 | croak "Repository UUID unreadable\n"; | |
255 | s_to_file($uuid,"$GIT_DIR/$GIT_SVN/info/uuid"); | |
256 | ||
257 | open my $fd, '>>', "$GIT_DIR/$GIT_SVN/info/exclude" or croak $!; | |
258 | print $fd '.svn',"\n"; | |
259 | close $fd or croak $!; | |
260 | return $uuid; | |
261 | } | |
262 | ||
263 | sub assert_svn_wc_clean { | |
264 | my ($svn_rev, $commit) = @_; | |
265 | croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/); | |
266 | croak "$commit is not a sha1!\n" unless ($commit =~ /^$sha1$/o); | |
267 | my $svn_info = svn_info('.'); | |
268 | if ($svn_rev != $svn_info->{'Last Changed Rev'}) { | |
269 | croak "Expected r$svn_rev, got r", | |
270 | $svn_info->{'Last Changed Rev'},"\n"; | |
271 | } | |
272 | my @status = grep(!/^Performing status on external/,(`svn status`)); | |
273 | @status = grep(!/^\s*$/,@status); | |
274 | if (scalar @status) { | |
275 | print STDERR "Tree ($SVN_WC) is not clean:\n"; | |
276 | print STDERR $_ foreach @status; | |
277 | croak; | |
278 | } | |
279 | my ($tree_a) = grep(/^tree $sha1$/o,`git-cat-file commit $commit`); | |
280 | $tree_a =~ s/^tree //; | |
281 | chomp $tree_a; | |
282 | chomp(my $tree_b = `GIT_INDEX_FILE=$GIT_SVN_INDEX git-write-tree`); | |
283 | if ($tree_a ne $tree_b) { | |
284 | croak "$svn_rev != $commit, $tree_a != $tree_b\n"; | |
285 | } | |
286 | } | |
287 | ||
288 | sub parse_diff_tree { | |
289 | my $diff_fh = shift; | |
290 | local $/ = "\0"; | |
291 | my $state = 'meta'; | |
292 | my @mods; | |
293 | while (<$diff_fh>) { | |
294 | chomp $_; # this gets rid of the trailing "\0" | |
295 | print $_,"\n"; | |
296 | if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s | |
297 | $sha1\s($sha1)\s([MTCRAD])\d*$/xo) { | |
298 | push @mods, { mode_a => $1, mode_b => $2, | |
299 | sha1_b => $3, chg => $4 }; | |
300 | if ($4 =~ /^(?:C|R)$/) { | |
301 | $state = 'file_a'; | |
302 | } else { | |
303 | $state = 'file_b'; | |
304 | } | |
305 | } elsif ($state eq 'file_a') { | |
306 | my $x = $mods[$#mods] or croak __LINE__,": Empty array\n"; | |
307 | if ($x->{chg} !~ /^(?:C|R)$/) { | |
308 | croak __LINE__,": Error parsing $_, $x->{chg}\n"; | |
309 | } | |
310 | $x->{file_a} = $_; | |
311 | $state = 'file_b'; | |
312 | } elsif ($state eq 'file_b') { | |
313 | my $x = $mods[$#mods] or croak __LINE__,": Empty array\n"; | |
314 | if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) { | |
315 | croak __LINE__,": Error parsing $_, $x->{chg}\n"; | |
316 | } | |
317 | if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) { | |
318 | croak __LINE__,": Error parsing $_, $x->{chg}\n"; | |
319 | } | |
320 | $x->{file_b} = $_; | |
321 | $state = 'meta'; | |
322 | } else { | |
323 | croak __LINE__,": Error parsing $_\n"; | |
324 | } | |
325 | } | |
326 | close $diff_fh or croak $!; | |
327 | return \@mods; | |
328 | } | |
329 | ||
330 | sub svn_check_prop_executable { | |
331 | my $m = shift; | |
332 | if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) { | |
333 | sys(qw(svn propset svn:executable 1), $m->{file_b}); | |
334 | } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) { | |
335 | sys(qw(svn propdel svn:executable), $m->{file_b}); | |
336 | } | |
337 | } | |
338 | ||
339 | sub svn_ensure_parent_path { | |
340 | my $dir_b = dirname(shift); | |
341 | svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir); | |
342 | mkpath([$dir_b]) unless (-d $dir_b); | |
343 | sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn"); | |
344 | } | |
345 | ||
346 | sub svn_checkout_tree { | |
347 | my ($svn_rev, $commit) = @_; | |
348 | my $from = file_to_s("$REV_DIR/$svn_rev"); | |
349 | assert_svn_wc_clean($svn_rev,$from); | |
350 | print "diff-tree '$from' '$commit'\n"; | |
351 | my $pid = open my $diff_fh, '-|'; | |
352 | defined $pid or croak $!; | |
353 | if ($pid == 0) { | |
72942938 EW |
354 | my @diff_tree = qw(git-diff-tree -z -r -C); |
355 | push @diff_tree, '--find-copies-harder' if $_find_copies_harder; | |
356 | push @diff_tree, "-l$_l" if defined $_l; | |
357 | exec(@diff_tree, $from, $commit) or croak $!; | |
3397f9df EW |
358 | } |
359 | my $mods = parse_diff_tree($diff_fh); | |
360 | unless (@$mods) { | |
361 | # git can do empty commits, SVN doesn't allow it... | |
362 | return $svn_rev; | |
363 | } | |
364 | my %rm; | |
365 | foreach my $m (@$mods) { | |
366 | if ($m->{chg} eq 'C') { | |
367 | svn_ensure_parent_path( $m->{file_b} ); | |
368 | sys(qw(svn cp), $m->{file_a}, $m->{file_b}); | |
369 | blob_to_file( $m->{sha1_b}, $m->{file_b}); | |
370 | svn_check_prop_executable($m); | |
371 | } elsif ($m->{chg} eq 'D') { | |
372 | $rm{dirname $m->{file_b}}->{basename $m->{file_b}} = 1; | |
373 | sys(qw(svn rm --force), $m->{file_b}); | |
374 | } elsif ($m->{chg} eq 'R') { | |
375 | svn_ensure_parent_path( $m->{file_b} ); | |
376 | sys(qw(svn mv --force), $m->{file_a}, $m->{file_b}); | |
377 | blob_to_file( $m->{sha1_b}, $m->{file_b}); | |
378 | svn_check_prop_executable($m); | |
379 | $rm{dirname $m->{file_a}}->{basename $m->{file_a}} = 1; | |
380 | } elsif ($m->{chg} eq 'M') { | |
381 | if ($m->{mode_b} =~ /^120/ && $m->{mode_a} =~ /^120/) { | |
382 | unlink $m->{file_b} or croak $!; | |
383 | blob_to_symlink($m->{sha1_b}, $m->{file_b}); | |
384 | } else { | |
385 | blob_to_file($m->{sha1_b}, $m->{file_b}); | |
386 | } | |
387 | svn_check_prop_executable($m); | |
388 | } elsif ($m->{chg} eq 'T') { | |
389 | sys(qw(svn rm --force),$m->{file_b}); | |
390 | if ($m->{mode_b} =~ /^120/ && $m->{mode_a} =~ /^100/) { | |
391 | blob_to_symlink($m->{sha1_b}, $m->{file_b}); | |
392 | } else { | |
393 | blob_to_file($m->{sha1_b}, $m->{file_b}); | |
394 | } | |
395 | svn_check_prop_executable($m); | |
396 | sys(qw(svn add --force), $m->{file_b}); | |
397 | } elsif ($m->{chg} eq 'A') { | |
398 | svn_ensure_parent_path( $m->{file_b} ); | |
399 | blob_to_file( $m->{sha1_b}, $m->{file_b}); | |
400 | if ($m->{mode_b} =~ /755$/) { | |
401 | chmod 0755, $m->{file_b}; | |
402 | } | |
403 | sys(qw(svn add --force), $m->{file_b}); | |
404 | } else { | |
405 | croak "Invalid chg: $m->{chg}\n"; | |
406 | } | |
407 | } | |
408 | if ($_rmdir) { | |
409 | my $old_index = $ENV{GIT_INDEX_FILE}; | |
410 | $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX; | |
411 | foreach my $dir (keys %rm) { | |
412 | my $files = $rm{$dir}; | |
413 | my @files; | |
414 | foreach (safe_qx('svn','ls',$dir)) { | |
415 | chomp; | |
416 | push @files, $_ unless $files->{$_}; | |
417 | } | |
418 | sys(qw(svn rm),$dir) unless @files; | |
419 | } | |
420 | if ($old_index) { | |
421 | $ENV{GIT_INDEX_FILE} = $old_index; | |
422 | } else { | |
423 | delete $ENV{GIT_INDEX_FILE}; | |
424 | } | |
425 | } | |
426 | } | |
427 | ||
428 | sub svn_commit_tree { | |
429 | my ($svn_rev, $commit) = @_; | |
430 | my $commit_msg = "$GIT_DIR/$GIT_SVN/.svn-commit.tmp.$$"; | |
431 | open my $msg, '>', $commit_msg or croak $!; | |
432 | ||
433 | chomp(my $type = `git-cat-file -t $commit`); | |
434 | if ($type eq 'commit') { | |
435 | my $pid = open my $msg_fh, '-|'; | |
436 | defined $pid or croak $!; | |
437 | ||
438 | if ($pid == 0) { | |
439 | exec(qw(git-cat-file commit), $commit) or croak $!; | |
440 | } | |
441 | my $in_msg = 0; | |
442 | while (<$msg_fh>) { | |
443 | if (!$in_msg) { | |
444 | $in_msg = 1 if (/^\s*$/); | |
445 | } else { | |
446 | print $msg $_ or croak $!; | |
447 | } | |
448 | } | |
449 | close $msg_fh or croak $!; | |
450 | } | |
451 | close $msg or croak $!; | |
452 | ||
453 | if ($_edit || ($type eq 'tree')) { | |
454 | my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi'; | |
455 | system($editor, $commit_msg); | |
456 | } | |
457 | my @ci_output = safe_qx(qw(svn commit -F),$commit_msg); | |
458 | my ($committed) = grep(/^Committed revision \d+\./,@ci_output); | |
459 | unlink $commit_msg; | |
460 | defined $committed or croak | |
461 | "Commit output failed to parse committed revision!\n", | |
462 | join("\n",@ci_output),"\n"; | |
463 | my ($rev_committed) = ($committed =~ /^Committed revision (\d+)\./); | |
464 | ||
465 | # resync immediately | |
466 | my @svn_up = (qw(svn up), "-r$svn_rev"); | |
467 | push @svn_up, '--ignore-externals' unless $_no_ignore_ext; | |
468 | sys(@svn_up); | |
469 | return fetch("$rev_committed=$commit")->{revision}; | |
470 | } | |
471 | ||
472 | sub svn_log_xml { | |
473 | my (@log_args) = @_; | |
474 | my $log_fh = IO::File->new_tmpfile or croak $!; | |
475 | ||
476 | my $pid = fork; | |
477 | defined $pid or croak $!; | |
478 | ||
479 | if ($pid == 0) { | |
480 | open STDOUT, '>&', $log_fh or croak $!; | |
481 | exec (qw(svn log --xml), @log_args) or croak $! | |
482 | } | |
483 | ||
484 | waitpid $pid, 0; | |
485 | croak $? if $?; | |
486 | ||
487 | seek $log_fh, 0, 0; | |
488 | my @svn_log; | |
489 | my $log = XML::Simple::XMLin( $log_fh, | |
490 | ForceArray => ['path','revision','logentry'], | |
491 | KeepRoot => 0, | |
492 | KeyAttr => { logentry => '+revision', | |
493 | paths => '+path' }, | |
494 | )->{logentry}; | |
495 | foreach my $r (sort {$a <=> $b} keys %$log) { | |
496 | my $log_msg = $log->{$r}; | |
497 | my ($Y,$m,$d,$H,$M,$S) = ($log_msg->{date} =~ | |
498 | /(\d{4})\-(\d\d)\-(\d\d)T | |
499 | (\d\d)\:(\d\d)\:(\d\d)\.\d+Z$/x) | |
500 | or croak "Failed to parse date: ", | |
501 | $log->{$r}->{date}; | |
502 | $log_msg->{date} = "+0000 $Y-$m-$d $H:$M:$S"; | |
503 | ||
504 | # XML::Simple can't handle <msg></msg> as a string: | |
505 | if (ref $log_msg->{msg} eq 'HASH') { | |
506 | $log_msg->{msg} = "\n"; | |
507 | } else { | |
508 | $log_msg->{msg} .= "\n"; | |
509 | } | |
510 | push @svn_log, $log->{$r}; | |
511 | } | |
512 | return \@svn_log; | |
513 | } | |
514 | ||
515 | sub svn_log_raw { | |
516 | my (@log_args) = @_; | |
517 | my $pid = open my $log_fh,'-|'; | |
518 | defined $pid or croak $!; | |
519 | ||
520 | if ($pid == 0) { | |
521 | exec (qw(svn log), @log_args) or croak $! | |
522 | } | |
523 | ||
524 | my @svn_log; | |
525 | my $state; | |
526 | while (<$log_fh>) { | |
527 | chomp; | |
528 | if (/^\-{72}$/) { | |
529 | $state = 'rev'; | |
530 | ||
531 | # if we have an empty log message, put something there: | |
532 | if (@svn_log) { | |
1c6bbbf3 | 533 | $svn_log[$#svn_log]->{msg} ||= "\n"; |
3397f9df EW |
534 | } |
535 | next; | |
536 | } | |
537 | if ($state eq 'rev' && s/^r(\d+)\s*\|\s*//) { | |
538 | my $rev = $1; | |
539 | my ($author, $date) = split(/\s*\|\s*/, $_, 2); | |
540 | my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~ | |
541 | /(\d{4})\-(\d\d)\-(\d\d)\s | |
542 | (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x) | |
543 | or croak "Failed to parse date: $date\n"; | |
544 | my %log_msg = ( revision => $rev, | |
545 | date => "$tz $Y-$m-$d $H:$M:$S", | |
546 | author => $author, | |
547 | msg => '' ); | |
1c6bbbf3 | 548 | push @svn_log, \%log_msg; |
3397f9df EW |
549 | $state = 'msg_start'; |
550 | next; | |
551 | } | |
552 | # skip the first blank line of the message: | |
553 | if ($state eq 'msg_start' && /^$/) { | |
554 | $state = 'msg'; | |
555 | } elsif ($state eq 'msg') { | |
1c6bbbf3 | 556 | $svn_log[$#svn_log]->{msg} .= $_."\n"; |
3397f9df EW |
557 | } |
558 | } | |
559 | close $log_fh or croak $?; | |
560 | return \@svn_log; | |
561 | } | |
562 | ||
563 | sub svn_info { | |
564 | my $url = shift || $SVN_URL; | |
565 | ||
566 | my $pid = open my $info_fh, '-|'; | |
567 | defined $pid or croak $!; | |
568 | ||
569 | if ($pid == 0) { | |
570 | exec(qw(svn info),$url) or croak $!; | |
571 | } | |
572 | ||
573 | my $ret = {}; | |
574 | # only single-lines seem to exist in svn info output | |
575 | while (<$info_fh>) { | |
576 | chomp $_; | |
577 | if (m#^([^:]+)\s*:\s*(\S*)$#) { | |
578 | $ret->{$1} = $2; | |
579 | push @{$ret->{-order}}, $1; | |
580 | } | |
581 | } | |
582 | close $info_fh or croak $!; | |
583 | return $ret; | |
584 | } | |
585 | ||
586 | sub sys { system(@_) == 0 or croak $? } | |
587 | ||
588 | sub git_addremove { | |
08703215 EW |
589 | system( "git-diff-files --name-only -z ". |
590 | " | git-update-index --remove -z --stdin; ". | |
591 | "git-ls-files -z --others ". | |
3397f9df | 592 | "'--exclude-from=$GIT_DIR/$GIT_SVN/info/exclude'". |
08703215 EW |
593 | " | git-update-index --add -z --stdin; " |
594 | ) == 0 or croak $? | |
3397f9df EW |
595 | } |
596 | ||
597 | sub s_to_file { | |
598 | my ($str, $file, $mode) = @_; | |
599 | open my $fd,'>',$file or croak $!; | |
600 | print $fd $str,"\n" or croak $!; | |
601 | close $fd or croak $!; | |
602 | chmod ($mode &~ umask, $file) if (defined $mode); | |
603 | } | |
604 | ||
605 | sub file_to_s { | |
606 | my $file = shift; | |
607 | open my $fd,'<',$file or croak "$!: file: $file\n"; | |
608 | local $/; | |
609 | my $ret = <$fd>; | |
610 | close $fd or croak $!; | |
611 | $ret =~ s/\s*$//s; | |
612 | return $ret; | |
613 | } | |
614 | ||
615 | sub assert_revision_unknown { | |
616 | my $revno = shift; | |
617 | if (-f "$REV_DIR/$revno") { | |
618 | croak "$REV_DIR/$revno already exists! ", | |
619 | "Why are we refetching it?"; | |
620 | } | |
621 | } | |
622 | ||
623 | sub assert_revision_eq_or_unknown { | |
624 | my ($revno, $commit) = @_; | |
625 | if (-f "$REV_DIR/$revno") { | |
626 | my $current = file_to_s("$REV_DIR/$revno"); | |
627 | if ($commit ne $current) { | |
628 | croak "$REV_DIR/$revno already exists!\n", | |
629 | "current: $current\nexpected: $commit\n"; | |
630 | } | |
631 | return; | |
632 | } | |
633 | } | |
634 | ||
635 | sub git_commit { | |
636 | my ($log_msg, @parents) = @_; | |
637 | assert_revision_unknown($log_msg->{revision}); | |
638 | my $out_fh = IO::File->new_tmpfile or croak $!; | |
639 | my $info = svn_info('.'); | |
640 | my $uuid = $info->{'Repository UUID'}; | |
641 | defined $uuid or croak "Unable to get Repository UUID\n"; | |
642 | ||
643 | # commit parents can be conditionally bound to a particular | |
644 | # svn revision via: "svn_revno=commit_sha1", filter them out here: | |
645 | my @exec_parents; | |
646 | foreach my $p (@parents) { | |
647 | next unless defined $p; | |
648 | if ($p =~ /^(\d+)=($sha1_short)$/o) { | |
649 | if ($1 == $log_msg->{revision}) { | |
650 | push @exec_parents, $2; | |
651 | } | |
652 | } else { | |
653 | push @exec_parents, $p if $p =~ /$sha1_short/o; | |
654 | } | |
655 | } | |
656 | ||
657 | my $pid = fork; | |
658 | defined $pid or croak $!; | |
659 | if ($pid == 0) { | |
660 | $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX; | |
661 | git_addremove(); | |
662 | chomp(my $tree = `git-write-tree`); | |
663 | croak if $?; | |
664 | my $msg_fh = IO::File->new_tmpfile or croak $!; | |
665 | print $msg_fh $log_msg->{msg}, "\ngit-svn-id: ", | |
666 | "$SVN_URL\@$log_msg->{revision}", | |
667 | " $uuid\n" or croak $!; | |
668 | $msg_fh->flush == 0 or croak $!; | |
669 | seek $msg_fh, 0, 0 or croak $!; | |
670 | ||
671 | $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = | |
672 | $log_msg->{author}; | |
673 | $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = | |
674 | $log_msg->{author}."\@$uuid"; | |
675 | $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = | |
676 | $log_msg->{date}; | |
677 | my @exec = ('git-commit-tree',$tree); | |
678 | push @exec, '-p', $_ foreach @exec_parents; | |
679 | open STDIN, '<&', $msg_fh or croak $!; | |
680 | open STDOUT, '>&', $out_fh or croak $!; | |
681 | exec @exec or croak $!; | |
682 | } | |
683 | waitpid($pid,0); | |
684 | croak if $?; | |
685 | ||
686 | $out_fh->flush == 0 or croak $!; | |
687 | seek $out_fh, 0, 0 or croak $!; | |
688 | chomp(my $commit = do { local $/; <$out_fh> }); | |
689 | if ($commit !~ /^$sha1$/o) { | |
690 | croak "Failed to commit, invalid sha1: $commit\n"; | |
691 | } | |
692 | my @update_ref = ('git-update-ref',"refs/heads/$GIT_SVN-HEAD",$commit); | |
693 | if (my $primary_parent = shift @exec_parents) { | |
694 | push @update_ref, $primary_parent; | |
695 | } | |
696 | sys(@update_ref); | |
697 | sys('git-update-ref',"$GIT_SVN/revs/$log_msg->{revision}",$commit); | |
698 | print "r$log_msg->{revision} = $commit\n"; | |
699 | return $commit; | |
700 | } | |
701 | ||
702 | sub blob_to_symlink { | |
703 | my ($blob, $link) = @_; | |
704 | defined $link or croak "\$link not defined!\n"; | |
705 | croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o; | |
706 | my $dest = `git-cat-file blob $blob`; # no newline, so no chomp | |
707 | symlink $dest, $link or croak $!; | |
708 | } | |
709 | ||
710 | sub blob_to_file { | |
711 | my ($blob, $file) = @_; | |
712 | defined $file or croak "\$file not defined!\n"; | |
713 | croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o; | |
714 | open my $blob_fh, '>', $file or croak "$!: $file\n"; | |
715 | my $pid = fork; | |
716 | defined $pid or croak $!; | |
717 | ||
718 | if ($pid == 0) { | |
719 | open STDOUT, '>&', $blob_fh or croak $!; | |
720 | exec('git-cat-file','blob',$blob); | |
721 | } | |
722 | waitpid $pid, 0; | |
723 | croak $? if $?; | |
724 | ||
725 | close $blob_fh or croak $!; | |
726 | } | |
727 | ||
728 | sub safe_qx { | |
729 | my $pid = open my $child, '-|'; | |
730 | defined $pid or croak $!; | |
731 | if ($pid == 0) { | |
732 | exec(@_) or croak $?; | |
733 | } | |
734 | my @ret = (<$child>); | |
735 | close $child or croak $?; | |
736 | die $? if $?; # just in case close didn't error out | |
737 | return wantarray ? @ret : join('',@ret); | |
738 | } | |
739 | ||
740 | sub svn_check_ignore_externals { | |
741 | return if $_no_ignore_ext; | |
742 | unless (grep /ignore-externals/,(safe_qx(qw(svn co -h)))) { | |
743 | print STDERR "W: Installed svn version does not support ", | |
744 | "--ignore-externals\n"; | |
745 | $_no_ignore_ext = 1; | |
746 | } | |
747 | } | |
748 | __END__ | |
749 | ||
750 | Data structures: | |
751 | ||
752 | @svn_log = array of log_msg hashes | |
753 | ||
754 | $log_msg hash | |
755 | { | |
756 | msg => 'whitespace-formatted log entry | |
757 | ', # trailing newline is preserved | |
758 | revision => '8', # integer | |
759 | date => '2004-02-24T17:01:44.108345Z', # commit date | |
760 | author => 'committer name' | |
761 | }; | |
762 | ||
763 | ||
764 | @mods = array of diff-index line hashes, each element represents one line | |
765 | of diff-index output | |
766 | ||
767 | diff-index line ($m hash) | |
768 | { | |
769 | mode_a => first column of diff-index output, no leading ':', | |
770 | mode_b => second column of diff-index output, | |
771 | sha1_b => sha1sum of the final blob, | |
772 | chg => change type [MCRAD], | |
773 | file_a => original file name of a file (iff chg is 'C' or 'R') | |
774 | file_b => new/current file name of a file (any chg) | |
775 | } | |
776 | ; |