1#!/usr/bin/env perl
2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3# License: GPL v2 or later
4use warnings;
5use strict;
6use vars qw/ $AUTHOR $VERSION
7 $SVN_URL $SVN_INFO $SVN_WC $SVN_UUID
8 $GIT_SVN_INDEX $GIT_SVN
9 $GIT_DIR $REV_DIR/;
10$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
11$VERSION = '1.1.0-pre';
12
13use Cwd qw/abs_path/;
14$GIT_DIR = abs_path($ENV{GIT_DIR} || '.git');
15$ENV{GIT_DIR} = $GIT_DIR;
16
17# make sure the svn binary gives consistent output between locales and TZs:
18$ENV{TZ} = 'UTC';
19$ENV{LC_ALL} = 'C';
20
21# If SVN:: library support is added, please make the dependencies
22# optional and preserve the capability to use the command-line client.
23# use eval { require SVN::... } to make it lazy load
24# We don't use any modules not in the standard Perl distribution:
25use Carp qw/croak/;
26use IO::File qw//;
27use File::Basename qw/dirname basename/;
28use File::Path qw/mkpath/;
29use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
30use File::Spec qw//;
31use POSIX qw/strftime/;
32my $sha1 = qr/[a-f\d]{40}/;
33my $sha1_short = qr/[a-f\d]{4,40}/;
34my ($_revision,$_stdin,$_no_ignore_ext,$_no_stop_copy,$_help,$_rmdir,$_edit,
35 $_find_copies_harder, $_l, $_version, $_upgrade, $_authors);
36my (@_branch_from, %tree_map, %users);
37my $_svn_co_url_revs;
38
39my %fc_opts = ( 'no-ignore-externals' => \$_no_ignore_ext,
40 'branch|b=s' => \@_branch_from,
41 'authors-file|A=s' => \$_authors );
42
43# yes, 'native' sets "\n". Patches to fix this for non-*nix systems welcome:
44my %EOL = ( CR => "\015", LF => "\012", CRLF => "\015\012", native => "\012" );
45
46my %cmd = (
47 fetch => [ \&fetch, "Download new revisions from SVN",
48 { 'revision|r=s' => \$_revision, %fc_opts } ],
49 init => [ \&init, "Initialize a repo for tracking" .
50 " (requires URL argument)", { } ],
51 commit => [ \&commit, "Commit git revisions to SVN",
52 { 'stdin|' => \$_stdin,
53 'edit|e' => \$_edit,
54 'rmdir' => \$_rmdir,
55 'find-copies-harder' => \$_find_copies_harder,
56 'l=i' => \$_l,
57 %fc_opts,
58 } ],
59 'show-ignore' => [ \&show_ignore, "Show svn:ignore listings", { } ],
60 rebuild => [ \&rebuild, "Rebuild git-svn metadata (after git clone)",
61 { 'no-ignore-externals' => \$_no_ignore_ext,
62 'upgrade' => \$_upgrade } ],
63);
64my $cmd;
65for (my $i = 0; $i < @ARGV; $i++) {
66 if (defined $cmd{$ARGV[$i]}) {
67 $cmd = $ARGV[$i];
68 splice @ARGV, $i, 1;
69 last;
70 }
71};
72
73my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
74
75# convert GetOpt::Long specs for use by git-repo-config
76foreach my $o (keys %opts) {
77 my $v = $opts{$o};
78 my ($key) = ($o =~ /^([a-z\-]+)/);
79 $key =~ s/-//g;
80 my $arg = 'git-repo-config';
81 $arg .= ' --int' if ($o =~ /=i$/);
82 $arg .= ' --bool' if ($o !~ /=[sfi]$/);
83 if (ref $v eq 'ARRAY') {
84 chomp(my @tmp = `$arg --get-all svn.$key`);
85 @$v = @tmp if @tmp;
86 } else {
87 chomp(my $tmp = `$arg --get svn.$key`);
88 if ($tmp && !($arg =~ / --bool / && $tmp eq 'false')) {
89 $$v = $tmp;
90 }
91 }
92}
93
94GetOptions(%opts, 'help|H|h' => \$_help,
95 'version|V' => \$_version,
96 'id|i=s' => \$GIT_SVN) or exit 1;
97
98$GIT_SVN ||= $ENV{GIT_SVN_ID} || 'git-svn';
99$GIT_SVN_INDEX = "$GIT_DIR/$GIT_SVN/index";
100$SVN_URL = undef;
101$REV_DIR = "$GIT_DIR/$GIT_SVN/revs";
102$SVN_WC = "$GIT_DIR/$GIT_SVN/tree";
103
104usage(0) if $_help;
105version() if $_version;
106usage(1) unless defined $cmd;
107load_authors() if $_authors;
108svn_compat_check();
109$cmd{$cmd}->[0]->(@ARGV);
110exit 0;
111
112####################### primary functions ######################
113sub usage {
114 my $exit = shift || 0;
115 my $fd = $exit ? \*STDERR : \*STDOUT;
116 print $fd <<"";
117git-svn - bidirectional operations between a single Subversion tree and git
118Usage: $0 <command> [options] [arguments]\n
119
120 print $fd "Available commands:\n" unless $cmd;
121
122 foreach (sort keys %cmd) {
123 next if $cmd && $cmd ne $_;
124 print $fd ' ',pack('A13',$_),$cmd{$_}->[1],"\n";
125 foreach (keys %{$cmd{$_}->[2]}) {
126 # prints out arguments as they should be passed:
127 my $x = s#=s$## ? '<arg>' : s#=i$## ? '<num>' : '';
128 print $fd ' ' x 17, join(', ', map { length $_ > 1 ?
129 "--$_" : "-$_" }
130 split /\|/,$_)," $x\n";
131 }
132 }
133 print $fd <<"";
134\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
135arbitrary identifier if you're tracking multiple SVN branches/repositories in
136one git repository and want to keep them separate. See git-svn(1) for more
137information.
138
139 exit $exit;
140}
141
142sub version {
143 print "git-svn version $VERSION\n";
144 exit 0;
145}
146
147sub rebuild {
148 $SVN_URL = shift or undef;
149 my $newest_rev = 0;
150 if ($_upgrade) {
151 sys('git-update-ref',"refs/remotes/$GIT_SVN","$GIT_SVN-HEAD");
152 } else {
153 check_upgrade_needed();
154 }
155
156 my $pid = open(my $rev_list,'-|');
157 defined $pid or croak $!;
158 if ($pid == 0) {
159 exec("git-rev-list","refs/remotes/$GIT_SVN") or croak $!;
160 }
161 my $latest;
162 while (<$rev_list>) {
163 chomp;
164 my $c = $_;
165 croak "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
166 my @commit = grep(/^git-svn-id: /,`git-cat-file commit $c`);
167 next if (!@commit); # skip merges
168 my $id = $commit[$#commit];
169 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
170 \s([a-f\d\-]+)$/x);
171 if (!$rev || !$uuid || !$url) {
172 # some of the original repositories I made had
173 # indentifiers like this:
174 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)
175 \@([a-f\d\-]+)/x);
176 if (!$rev || !$uuid) {
177 croak "Unable to extract revision or UUID from ",
178 "$c, $id\n";
179 }
180 }
181
182 # if we merged or otherwise started elsewhere, this is
183 # how we break out of it
184 next if (defined $SVN_UUID && ($uuid ne $SVN_UUID));
185 next if (defined $SVN_URL && defined $url && ($url ne $SVN_URL));
186
187 print "r$rev = $c\n";
188 unless (defined $latest) {
189 if (!$SVN_URL && !$url) {
190 croak "SVN repository location required: $url\n";
191 }
192 $SVN_URL ||= $url;
193 $SVN_UUID ||= $uuid;
194 setup_git_svn();
195 $latest = $rev;
196 }
197 assert_revision_eq_or_unknown($rev, $c);
198 sys('git-update-ref',"$GIT_SVN/revs/$rev",$c);
199 $newest_rev = $rev if ($rev > $newest_rev);
200 }
201 close $rev_list or croak $?;
202 if (!chdir $SVN_WC) {
203 svn_cmd_checkout($SVN_URL, $latest, $SVN_WC);
204 chdir $SVN_WC or croak $!;
205 }
206
207 $pid = fork;
208 defined $pid or croak $!;
209 if ($pid == 0) {
210 my @svn_up = qw(svn up);
211 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
212 sys(@svn_up,"-r$newest_rev");
213 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
214 index_changes();
215 exec('git-write-tree');
216 }
217 waitpid $pid, 0;
218
219 if ($_upgrade) {
220 print STDERR <<"";
221Keeping deprecated refs/head/$GIT_SVN-HEAD for now. Please remove it
222when you have upgraded your tools and habits to use refs/remotes/$GIT_SVN
223
224 }
225}
226
227sub init {
228 $SVN_URL = shift or die "SVN repository location required " .
229 "as a command-line argument\n";
230 unless (-d $GIT_DIR) {
231 sys('git-init-db');
232 }
233 setup_git_svn();
234}
235
236sub fetch {
237 my (@parents) = @_;
238 check_upgrade_needed();
239 $SVN_URL ||= file_to_s("$GIT_DIR/$GIT_SVN/info/url");
240 my @log_args = -d $SVN_WC ? ($SVN_WC) : ($SVN_URL);
241 unless ($_revision) {
242 $_revision = -d $SVN_WC ? 'BASE:HEAD' : '0:HEAD';
243 }
244 push @log_args, "-r$_revision";
245 push @log_args, '--stop-on-copy' unless $_no_stop_copy;
246
247 my $svn_log = svn_log_raw(@log_args);
248
249 my $base = next_log_entry($svn_log) or croak "No base revision!\n";
250 my $last_commit = undef;
251 unless (-d $SVN_WC) {
252 svn_cmd_checkout($SVN_URL,$base->{revision},$SVN_WC);
253 chdir $SVN_WC or croak $!;
254 read_uuid();
255 $last_commit = git_commit($base, @parents);
256 assert_tree($last_commit);
257 } else {
258 chdir $SVN_WC or croak $!;
259 read_uuid();
260 $last_commit = file_to_s("$REV_DIR/$base->{revision}");
261 }
262 my @svn_up = qw(svn up);
263 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
264 my $last = $base;
265 while (my $log_msg = next_log_entry($svn_log)) {
266 assert_tree($last_commit);
267 if ($last->{revision} >= $log_msg->{revision}) {
268 croak "Out of order: last >= current: ",
269 "$last->{revision} >= $log_msg->{revision}\n";
270 }
271 # Revert is needed for cases like:
272 # https://svn.musicpd.org/Jamming/trunk (r166:167), but
273 # I can't seem to reproduce something like that on a test...
274 sys(qw/svn revert -R ./);
275 assert_svn_wc_clean($last->{revision});
276 sys(@svn_up,"-r$log_msg->{revision}");
277 $last_commit = git_commit($log_msg, $last_commit, @parents);
278 $last = $log_msg;
279 }
280 unless (-e "$GIT_DIR/refs/heads/master") {
281 sys(qw(git-update-ref refs/heads/master),$last_commit);
282 }
283 return $last;
284}
285
286sub commit {
287 my (@commits) = @_;
288 check_upgrade_needed();
289 if ($_stdin || !@commits) {
290 print "Reading from stdin...\n";
291 @commits = ();
292 while (<STDIN>) {
293 if (/\b($sha1_short)\b/o) {
294 unshift @commits, $1;
295 }
296 }
297 }
298 my @revs;
299 foreach my $c (@commits) {
300 chomp(my @tmp = safe_qx('git-rev-parse',$c));
301 if (scalar @tmp == 1) {
302 push @revs, $tmp[0];
303 } elsif (scalar @tmp > 1) {
304 push @revs, reverse (safe_qx('git-rev-list',@tmp));
305 } else {
306 die "Failed to rev-parse $c\n";
307 }
308 }
309 chomp @revs;
310
311 fetch();
312 chdir $SVN_WC or croak $!;
313 my $info = svn_info('.');
314 read_uuid($info);
315 my $svn_current_rev = $info->{'Last Changed Rev'};
316 foreach my $c (@revs) {
317 my $mods = svn_checkout_tree($svn_current_rev, $c);
318 if (scalar @$mods == 0) {
319 print "Skipping, no changes detected\n";
320 next;
321 }
322 $svn_current_rev = svn_commit_tree($svn_current_rev, $c);
323 }
324 print "Done committing ",scalar @revs," revisions to SVN\n";
325}
326
327sub show_ignore {
328 require File::Find or die $!;
329 my $exclude_file = "$GIT_DIR/info/exclude";
330 open my $fh, '<', $exclude_file or croak $!;
331 chomp(my @excludes = (<$fh>));
332 close $fh or croak $!;
333
334 $SVN_URL ||= file_to_s("$GIT_DIR/$GIT_SVN/info/url");
335 chdir $SVN_WC or croak $!;
336 my %ign;
337 File::Find::find({wanted=>sub{if(lstat $_ && -d _ && -d "$_/.svn"){
338 s#^\./##;
339 @{$ign{$_}} = safe_qx(qw(svn propget svn:ignore),$_);
340 }}, no_chdir=>1},'.');
341
342 print "\n# /\n";
343 foreach (@{$ign{'.'}}) { print '/',$_ if /\S/ }
344 delete $ign{'.'};
345 foreach my $i (sort keys %ign) {
346 print "\n# ",$i,"\n";
347 foreach (@{$ign{$i}}) { print '/',$i,'/',$_ if /\S/ }
348 }
349}
350
351########################### utility functions #########################
352
353sub read_uuid {
354 return if $SVN_UUID;
355 my $info = shift || svn_info('.');
356 $SVN_UUID = $info->{'Repository UUID'} or
357 croak "Repository UUID unreadable\n";
358 s_to_file($SVN_UUID,"$GIT_DIR/$GIT_SVN/info/uuid");
359}
360
361sub setup_git_svn {
362 defined $SVN_URL or croak "SVN repository location required\n";
363 unless (-d $GIT_DIR) {
364 croak "GIT_DIR=$GIT_DIR does not exist!\n";
365 }
366 mkpath(["$GIT_DIR/$GIT_SVN"]);
367 mkpath(["$GIT_DIR/$GIT_SVN/info"]);
368 mkpath([$REV_DIR]);
369 s_to_file($SVN_URL,"$GIT_DIR/$GIT_SVN/info/url");
370
371 open my $fd, '>>', "$GIT_DIR/$GIT_SVN/info/exclude" or croak $!;
372 print $fd '.svn',"\n";
373 close $fd or croak $!;
374}
375
376sub assert_svn_wc_clean {
377 my ($svn_rev) = @_;
378 croak "$svn_rev is not an integer!\n" unless ($svn_rev =~ /^\d+$/);
379 my $lcr = svn_info('.')->{'Last Changed Rev'};
380 if ($svn_rev != $lcr) {
381 print STDERR "Checking for copy-tree ... ";
382 my @diff = grep(/^Index: /,(safe_qx(qw(svn diff),
383 "-r$lcr:$svn_rev")));
384 if (@diff) {
385 croak "Nope! Expected r$svn_rev, got r$lcr\n";
386 } else {
387 print STDERR "OK!\n";
388 }
389 }
390 my @status = grep(!/^Performing status on external/,(`svn status`));
391 @status = grep(!/^\s*$/,@status);
392 if (scalar @status) {
393 print STDERR "Tree ($SVN_WC) is not clean:\n";
394 print STDERR $_ foreach @status;
395 croak;
396 }
397}
398
399sub assert_tree {
400 my ($treeish) = @_;
401 croak "Not a sha1: $treeish\n" unless $treeish =~ /^$sha1$/o;
402 chomp(my $type = `git-cat-file -t $treeish`);
403 my $expected;
404 while ($type eq 'tag') {
405 chomp(($treeish, $type) = `git-cat-file tag $treeish`);
406 }
407 if ($type eq 'commit') {
408 $expected = (grep /^tree /,`git-cat-file commit $treeish`)[0];
409 ($expected) = ($expected =~ /^tree ($sha1)$/);
410 die "Unable to get tree from $treeish\n" unless $expected;
411 } elsif ($type eq 'tree') {
412 $expected = $treeish;
413 } else {
414 die "$treeish is a $type, expected tree, tag or commit\n";
415 }
416
417 my $old_index = $ENV{GIT_INDEX_FILE};
418 my $tmpindex = $GIT_SVN_INDEX.'.assert-tmp';
419 if (-e $tmpindex) {
420 unlink $tmpindex or croak $!;
421 }
422 $ENV{GIT_INDEX_FILE} = $tmpindex;
423 index_changes(1);
424 chomp(my $tree = `git-write-tree`);
425 if ($old_index) {
426 $ENV{GIT_INDEX_FILE} = $old_index;
427 } else {
428 delete $ENV{GIT_INDEX_FILE};
429 }
430 if ($tree ne $expected) {
431 croak "Tree mismatch, Got: $tree, Expected: $expected\n";
432 }
433 unlink $tmpindex;
434}
435
436sub parse_diff_tree {
437 my $diff_fh = shift;
438 local $/ = "\0";
439 my $state = 'meta';
440 my @mods;
441 while (<$diff_fh>) {
442 chomp $_; # this gets rid of the trailing "\0"
443 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
444 $sha1\s($sha1)\s([MTCRAD])\d*$/xo) {
445 push @mods, { mode_a => $1, mode_b => $2,
446 sha1_b => $3, chg => $4 };
447 if ($4 =~ /^(?:C|R)$/) {
448 $state = 'file_a';
449 } else {
450 $state = 'file_b';
451 }
452 } elsif ($state eq 'file_a') {
453 my $x = $mods[$#mods] or croak "Empty array\n";
454 if ($x->{chg} !~ /^(?:C|R)$/) {
455 croak "Error parsing $_, $x->{chg}\n";
456 }
457 $x->{file_a} = $_;
458 $state = 'file_b';
459 } elsif ($state eq 'file_b') {
460 my $x = $mods[$#mods] or croak "Empty array\n";
461 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
462 croak "Error parsing $_, $x->{chg}\n";
463 }
464 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
465 croak "Error parsing $_, $x->{chg}\n";
466 }
467 $x->{file_b} = $_;
468 $state = 'meta';
469 } else {
470 croak "Error parsing $_\n";
471 }
472 }
473 close $diff_fh or croak $!;
474
475 return \@mods;
476}
477
478sub svn_check_prop_executable {
479 my $m = shift;
480 return if -l $m->{file_b};
481 if ($m->{mode_b} =~ /755$/) {
482 chmod((0755 &~ umask),$m->{file_b}) or croak $!;
483 if ($m->{mode_a} !~ /755$/) {
484 sys(qw(svn propset svn:executable 1), $m->{file_b});
485 }
486 -x $m->{file_b} or croak "$m->{file_b} is not executable!\n";
487 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
488 sys(qw(svn propdel svn:executable), $m->{file_b});
489 chmod((0644 &~ umask),$m->{file_b}) or croak $!;
490 -x $m->{file_b} and croak "$m->{file_b} is executable!\n";
491 }
492}
493
494sub svn_ensure_parent_path {
495 my $dir_b = dirname(shift);
496 svn_ensure_parent_path($dir_b) if ($dir_b ne File::Spec->curdir);
497 mkpath([$dir_b]) unless (-d $dir_b);
498 sys(qw(svn add -N), $dir_b) unless (-d "$dir_b/.svn");
499}
500
501sub precommit_check {
502 my $mods = shift;
503 my (%rm_file, %rmdir_check, %added_check);
504
505 my %o = ( D => 0, R => 1, C => 2, A => 3, M => 3, T => 3 );
506 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
507 if ($m->{chg} eq 'R') {
508 if (-d $m->{file_b}) {
509 err_dir_to_file("$m->{file_a} => $m->{file_b}");
510 }
511 # dir/$file => dir/file/$file
512 my $dirname = dirname($m->{file_b});
513 while ($dirname ne File::Spec->curdir) {
514 if ($dirname ne $m->{file_a}) {
515 $dirname = dirname($dirname);
516 next;
517 }
518 err_file_to_dir("$m->{file_a} => $m->{file_b}");
519 }
520 # baz/zzz => baz (baz is a file)
521 $dirname = dirname($m->{file_a});
522 while ($dirname ne File::Spec->curdir) {
523 if ($dirname ne $m->{file_b}) {
524 $dirname = dirname($dirname);
525 next;
526 }
527 err_dir_to_file("$m->{file_a} => $m->{file_b}");
528 }
529 }
530 if ($m->{chg} =~ /^(D|R)$/) {
531 my $t = $1 eq 'D' ? 'file_b' : 'file_a';
532 $rm_file{ $m->{$t} } = 1;
533 my $dirname = dirname( $m->{$t} );
534 my $basename = basename( $m->{$t} );
535 $rmdir_check{$dirname}->{$basename} = 1;
536 } elsif ($m->{chg} =~ /^(?:A|C)$/) {
537 if (-d $m->{file_b}) {
538 err_dir_to_file($m->{file_b});
539 }
540 my $dirname = dirname( $m->{file_b} );
541 my $basename = basename( $m->{file_b} );
542 $added_check{$dirname}->{$basename} = 1;
543 while ($dirname ne File::Spec->curdir) {
544 if ($rm_file{$dirname}) {
545 err_file_to_dir($m->{file_b});
546 }
547 $dirname = dirname $dirname;
548 }
549 }
550 }
551 return (\%rmdir_check, \%added_check);
552
553 sub err_dir_to_file {
554 my $file = shift;
555 print STDERR "Node change from directory to file ",
556 "is not supported by Subversion: ",$file,"\n";
557 exit 1;
558 }
559 sub err_file_to_dir {
560 my $file = shift;
561 print STDERR "Node change from file to directory ",
562 "is not supported by Subversion: ",$file,"\n";
563 exit 1;
564 }
565}
566
567sub svn_checkout_tree {
568 my ($svn_rev, $treeish) = @_;
569 my $from = file_to_s("$REV_DIR/$svn_rev");
570 assert_svn_wc_clean($svn_rev);
571 assert_tree($from);
572 print "diff-tree $from $treeish\n";
573 my $pid = open my $diff_fh, '-|';
574 defined $pid or croak $!;
575 if ($pid == 0) {
576 my @diff_tree = qw(git-diff-tree -z -r -C);
577 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
578 push @diff_tree, "-l$_l" if defined $_l;
579 exec(@diff_tree, $from, $treeish) or croak $!;
580 }
581 my $mods = parse_diff_tree($diff_fh);
582 unless (@$mods) {
583 # git can do empty commits, but SVN doesn't allow it...
584 return $mods;
585 }
586 my ($rm, $add) = precommit_check($mods);
587
588 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
589 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
590 if ($m->{chg} eq 'C') {
591 svn_ensure_parent_path( $m->{file_b} );
592 sys(qw(svn cp), $m->{file_a}, $m->{file_b});
593 apply_mod_line_blob($m);
594 svn_check_prop_executable($m);
595 } elsif ($m->{chg} eq 'D') {
596 sys(qw(svn rm --force), $m->{file_b});
597 } elsif ($m->{chg} eq 'R') {
598 svn_ensure_parent_path( $m->{file_b} );
599 sys(qw(svn mv --force), $m->{file_a}, $m->{file_b});
600 apply_mod_line_blob($m);
601 svn_check_prop_executable($m);
602 } elsif ($m->{chg} eq 'M') {
603 apply_mod_line_blob($m);
604 svn_check_prop_executable($m);
605 } elsif ($m->{chg} eq 'T') {
606 sys(qw(svn rm --force),$m->{file_b});
607 apply_mod_line_blob($m);
608 sys(qw(svn add --force), $m->{file_b});
609 svn_check_prop_executable($m);
610 } elsif ($m->{chg} eq 'A') {
611 svn_ensure_parent_path( $m->{file_b} );
612 apply_mod_line_blob($m);
613 sys(qw(svn add --force), $m->{file_b});
614 svn_check_prop_executable($m);
615 } else {
616 croak "Invalid chg: $m->{chg}\n";
617 }
618 }
619
620 assert_tree($treeish);
621 if ($_rmdir) { # remove empty directories
622 handle_rmdir($rm, $add);
623 }
624 assert_tree($treeish);
625 return $mods;
626}
627
628# svn ls doesn't work with respect to the current working tree, but what's
629# in the repository. There's not even an option for it... *sigh*
630# (added files don't show up and removed files remain in the ls listing)
631sub svn_ls_current {
632 my ($dir, $rm, $add) = @_;
633 chomp(my @ls = safe_qx('svn','ls',$dir));
634 my @ret = ();
635 foreach (@ls) {
636 s#/$##; # trailing slashes are evil
637 push @ret, $_ unless $rm->{$dir}->{$_};
638 }
639 if (exists $add->{$dir}) {
640 push @ret, keys %{$add->{$dir}};
641 }
642 return \@ret;
643}
644
645sub handle_rmdir {
646 my ($rm, $add) = @_;
647
648 foreach my $dir (sort {length $b <=> length $a} keys %$rm) {
649 my $ls = svn_ls_current($dir, $rm, $add);
650 next if (scalar @$ls);
651 sys(qw(svn rm --force),$dir);
652
653 my $dn = dirname $dir;
654 $rm->{ $dn }->{ basename $dir } = 1;
655 $ls = svn_ls_current($dn, $rm, $add);
656 while (scalar @$ls == 0 && $dn ne File::Spec->curdir) {
657 sys(qw(svn rm --force),$dn);
658 $dir = basename $dn;
659 $dn = dirname $dn;
660 $rm->{ $dn }->{ $dir } = 1;
661 $ls = svn_ls_current($dn, $rm, $add);
662 }
663 }
664}
665
666sub svn_commit_tree {
667 my ($svn_rev, $commit) = @_;
668 my $commit_msg = "$GIT_DIR/$GIT_SVN/.svn-commit.tmp.$$";
669 my %log_msg = ( msg => '' );
670 open my $msg, '>', $commit_msg or croak $!;
671
672 chomp(my $type = `git-cat-file -t $commit`);
673 if ($type eq 'commit') {
674 my $pid = open my $msg_fh, '-|';
675 defined $pid or croak $!;
676
677 if ($pid == 0) {
678 exec(qw(git-cat-file commit), $commit) or croak $!;
679 }
680 my $in_msg = 0;
681 while (<$msg_fh>) {
682 if (!$in_msg) {
683 $in_msg = 1 if (/^\s*$/);
684 } elsif (/^git-svn-id: /) {
685 # skip this, we regenerate the correct one
686 # on re-fetch anyways
687 } else {
688 print $msg $_ or croak $!;
689 }
690 }
691 close $msg_fh or croak $!;
692 }
693 close $msg or croak $!;
694
695 if ($_edit || ($type eq 'tree')) {
696 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
697 system($editor, $commit_msg);
698 }
699
700 # file_to_s removes all trailing newlines, so just use chomp() here:
701 open $msg, '<', $commit_msg or croak $!;
702 { local $/; chomp($log_msg{msg} = <$msg>); }
703 close $msg or croak $!;
704
705 my ($oneline) = ($log_msg{msg} =~ /([^\n\r]+)/);
706 print "Committing $commit: $oneline\n";
707
708 my @ci_output = safe_qx(qw(svn commit -F),$commit_msg);
709 my ($committed) = grep(/^Committed revision \d+\./,@ci_output);
710 unlink $commit_msg;
711 defined $committed or croak
712 "Commit output failed to parse committed revision!\n",
713 join("\n",@ci_output),"\n";
714 my ($rev_committed) = ($committed =~ /^Committed revision (\d+)\./);
715
716 my @svn_up = qw(svn up);
717 push @svn_up, '--ignore-externals' unless $_no_ignore_ext;
718 if ($rev_committed == ($svn_rev + 1)) {
719 push @svn_up, "-r$rev_committed";
720 sys(@svn_up);
721 my $info = svn_info('.');
722 my $date = $info->{'Last Changed Date'} or die "Missing date\n";
723 if ($info->{'Last Changed Rev'} != $rev_committed) {
724 croak "$info->{'Last Changed Rev'} != $rev_committed\n"
725 }
726 my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
727 /(\d{4})\-(\d\d)\-(\d\d)\s
728 (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
729 or croak "Failed to parse date: $date\n";
730 $log_msg{date} = "$tz $Y-$m-$d $H:$M:$S";
731 $log_msg{author} = $info->{'Last Changed Author'};
732 $log_msg{revision} = $rev_committed;
733 $log_msg{msg} .= "\n";
734 my $parent = file_to_s("$REV_DIR/$svn_rev");
735 git_commit(\%log_msg, $parent, $commit);
736 return $rev_committed;
737 }
738 # resync immediately
739 push @svn_up, "-r$svn_rev";
740 sys(@svn_up);
741 return fetch("$rev_committed=$commit")->{revision};
742}
743
744# read the entire log into a temporary file (which is removed ASAP)
745# and store the file handle + parser state
746sub svn_log_raw {
747 my (@log_args) = @_;
748 my $log_fh = IO::File->new_tmpfile or croak $!;
749 my $pid = fork;
750 defined $pid or croak $!;
751 if (!$pid) {
752 open STDOUT, '>&', $log_fh or croak $!;
753 exec (qw(svn log), @log_args) or croak $!
754 }
755 waitpid $pid, 0;
756 croak if $?;
757 seek $log_fh, 0, 0 or croak $!;
758 return { state => 'sep', fh => $log_fh };
759}
760
761sub next_log_entry {
762 my $log = shift; # retval of svn_log_raw()
763 my $ret = undef;
764 my $fh = $log->{fh};
765
766 while (<$fh>) {
767 chomp;
768 if (/^\-{72}$/) {
769 if ($log->{state} eq 'msg') {
770 if ($ret->{lines}) {
771 $ret->{msg} .= $_."\n";
772 unless(--$ret->{lines}) {
773 $log->{state} = 'sep';
774 }
775 } else {
776 croak "Log parse error at: $_\n",
777 $ret->{revision},
778 "\n";
779 }
780 next;
781 }
782 if ($log->{state} ne 'sep') {
783 croak "Log parse error at: $_\n",
784 "state: $log->{state}\n",
785 $ret->{revision},
786 "\n";
787 }
788 $log->{state} = 'rev';
789
790 # if we have an empty log message, put something there:
791 if ($ret) {
792 $ret->{msg} ||= "\n";
793 delete $ret->{lines};
794 return $ret;
795 }
796 next;
797 }
798 if ($log->{state} eq 'rev' && s/^r(\d+)\s*\|\s*//) {
799 my $rev = $1;
800 my ($author, $date, $lines) = split(/\s*\|\s*/, $_, 3);
801 ($lines) = ($lines =~ /(\d+)/);
802 my ($Y,$m,$d,$H,$M,$S,$tz) = ($date =~
803 /(\d{4})\-(\d\d)\-(\d\d)\s
804 (\d\d)\:(\d\d)\:(\d\d)\s([\-\+]\d+)/x)
805 or croak "Failed to parse date: $date\n";
806 $ret = { revision => $rev,
807 date => "$tz $Y-$m-$d $H:$M:$S",
808 author => $author,
809 lines => $lines,
810 msg => '' };
811 if (defined $_authors && ! defined $users{$author}) {
812 die "Author: $author not defined in ",
813 "$_authors file\n";
814 }
815 $log->{state} = 'msg_start';
816 next;
817 }
818 # skip the first blank line of the message:
819 if ($log->{state} eq 'msg_start' && /^$/) {
820 $log->{state} = 'msg';
821 } elsif ($log->{state} eq 'msg') {
822 if ($ret->{lines}) {
823 $ret->{msg} .= $_."\n";
824 unless (--$ret->{lines}) {
825 $log->{state} = 'sep';
826 }
827 } else {
828 croak "Log parse error at: $_\n",
829 $ret->{revision},"\n";
830 }
831 }
832 }
833 return $ret;
834}
835
836sub svn_info {
837 my $url = shift || $SVN_URL;
838
839 my $pid = open my $info_fh, '-|';
840 defined $pid or croak $!;
841
842 if ($pid == 0) {
843 exec(qw(svn info),$url) or croak $!;
844 }
845
846 my $ret = {};
847 # only single-lines seem to exist in svn info output
848 while (<$info_fh>) {
849 chomp $_;
850 if (m#^([^:]+)\s*:\s*(\S.*)$#) {
851 $ret->{$1} = $2;
852 push @{$ret->{-order}}, $1;
853 }
854 }
855 close $info_fh or croak $!;
856 return $ret;
857}
858
859sub sys { system(@_) == 0 or croak $? }
860
861sub eol_cp {
862 my ($from, $to) = @_;
863 my $es = safe_qx(qw/svn propget svn:eol-style/, $to);
864 open my $rfd, '<', $from or croak $!;
865 binmode $rfd or croak $!;
866 open my $wfd, '>', $to or croak $!;
867 binmode $wfd or croak $!;
868
869 my $eol = $EOL{$es} or undef;
870 if ($eol) {
871 print "$eol: $from => $to\n";
872 }
873 my $buf;
874 while (1) {
875 my ($r, $w, $t);
876 defined($r = sysread($rfd, $buf, 4096)) or croak $!;
877 return unless $r;
878 $buf =~ s/(?:\015|\012|\015\012)/$eol/gs if $eol;
879 for ($w = 0; $w < $r; $w += $t) {
880 $t = syswrite($wfd, $buf, $r - $w, $w) or croak $!;
881 }
882 }
883}
884
885sub do_update_index {
886 my ($z_cmd, $cmd, $no_text_base) = @_;
887
888 my $z = open my $p, '-|';
889 defined $z or croak $!;
890 unless ($z) { exec @$z_cmd or croak $! }
891
892 my $pid = open my $ui, '|-';
893 defined $pid or croak $!;
894 unless ($pid) {
895 exec('git-update-index',"--$cmd",'-z','--stdin') or croak $!;
896 }
897 local $/ = "\0";
898 while (my $x = <$p>) {
899 chomp $x;
900 if (!$no_text_base && lstat $x && ! -l _ &&
901 safe_qx(qw/svn propget svn:keywords/,$x)) {
902 my $mode = -x _ ? 0755 : 0644;
903 my ($v,$d,$f) = File::Spec->splitpath($x);
904 my $tb = File::Spec->catfile($d, '.svn', 'tmp',
905 'text-base',"$f.svn-base");
906 $tb =~ s#^/##;
907 unless (-f $tb) {
908 $tb = File::Spec->catfile($d, '.svn',
909 'text-base',"$f.svn-base");
910 $tb =~ s#^/##;
911 }
912 unlink $x or croak $!;
913 eol_cp($tb, $x);
914 chmod(($mode &~ umask), $x) or croak $!;
915 }
916 print $ui $x,"\0";
917 }
918 close $ui or croak $!;
919}
920
921sub index_changes {
922 my $no_text_base = shift;
923 do_update_index([qw/git-diff-files --name-only -z/],
924 'remove',
925 $no_text_base);
926 do_update_index([qw/git-ls-files -z --others/,
927 "--exclude-from=$GIT_DIR/$GIT_SVN/info/exclude"],
928 'add',
929 $no_text_base);
930}
931
932sub s_to_file {
933 my ($str, $file, $mode) = @_;
934 open my $fd,'>',$file or croak $!;
935 print $fd $str,"\n" or croak $!;
936 close $fd or croak $!;
937 chmod ($mode &~ umask, $file) if (defined $mode);
938}
939
940sub file_to_s {
941 my $file = shift;
942 open my $fd,'<',$file or croak "$!: file: $file\n";
943 local $/;
944 my $ret = <$fd>;
945 close $fd or croak $!;
946 $ret =~ s/\s*$//s;
947 return $ret;
948}
949
950sub assert_revision_unknown {
951 my $revno = shift;
952 if (-f "$REV_DIR/$revno") {
953 croak "$REV_DIR/$revno already exists! ",
954 "Why are we refetching it?";
955 }
956}
957
958sub trees_eq {
959 my ($x, $y) = @_;
960 my @x = safe_qx('git-cat-file','commit',$x);
961 my @y = safe_qx('git-cat-file','commit',$y);
962 if (($y[0] ne $x[0]) || $x[0] !~ /^tree $sha1\n$/
963 || $y[0] !~ /^tree $sha1\n$/) {
964 print STDERR "Trees not equal: $y[0] != $x[0]\n";
965 return 0
966 }
967 return 1;
968}
969
970sub assert_revision_eq_or_unknown {
971 my ($revno, $commit) = @_;
972 if (-f "$REV_DIR/$revno") {
973 my $current = file_to_s("$REV_DIR/$revno");
974 if (($commit ne $current) && !trees_eq($commit, $current)) {
975 croak "$REV_DIR/$revno already exists!\n",
976 "current: $current\nexpected: $commit\n";
977 }
978 return;
979 }
980}
981
982sub git_commit {
983 my ($log_msg, @parents) = @_;
984 assert_revision_unknown($log_msg->{revision});
985 my $out_fh = IO::File->new_tmpfile or croak $!;
986
987 map_tree_joins() if (@_branch_from && !%tree_map);
988
989 # commit parents can be conditionally bound to a particular
990 # svn revision via: "svn_revno=commit_sha1", filter them out here:
991 my @exec_parents;
992 foreach my $p (@parents) {
993 next unless defined $p;
994 if ($p =~ /^(\d+)=($sha1_short)$/o) {
995 if ($1 == $log_msg->{revision}) {
996 push @exec_parents, $2;
997 }
998 } else {
999 push @exec_parents, $p if $p =~ /$sha1_short/o;
1000 }
1001 }
1002
1003 my $pid = fork;
1004 defined $pid or croak $!;
1005 if ($pid == 0) {
1006 $ENV{GIT_INDEX_FILE} = $GIT_SVN_INDEX;
1007 index_changes();
1008 chomp(my $tree = `git-write-tree`);
1009 croak if $?;
1010 if (exists $tree_map{$tree}) {
1011 my %seen_parent = map { $_ => 1 } @exec_parents;
1012 foreach (@{$tree_map{$tree}}) {
1013 # MAXPARENT is defined to 16 in commit-tree.c:
1014 if ($seen_parent{$_} || @exec_parents > 16) {
1015 next;
1016 }
1017 push @exec_parents, $_;
1018 $seen_parent{$_} = 1;
1019 }
1020 }
1021 my $msg_fh = IO::File->new_tmpfile or croak $!;
1022 print $msg_fh $log_msg->{msg}, "\ngit-svn-id: ",
1023 "$SVN_URL\@$log_msg->{revision}",
1024 " $SVN_UUID\n" or croak $!;
1025 $msg_fh->flush == 0 or croak $!;
1026 seek $msg_fh, 0, 0 or croak $!;
1027
1028 set_commit_env($log_msg);
1029
1030 my @exec = ('git-commit-tree',$tree);
1031 push @exec, '-p', $_ foreach @exec_parents;
1032 open STDIN, '<&', $msg_fh or croak $!;
1033 open STDOUT, '>&', $out_fh or croak $!;
1034 exec @exec or croak $!;
1035 }
1036 waitpid($pid,0);
1037 croak if $?;
1038
1039 $out_fh->flush == 0 or croak $!;
1040 seek $out_fh, 0, 0 or croak $!;
1041 chomp(my $commit = do { local $/; <$out_fh> });
1042 if ($commit !~ /^$sha1$/o) {
1043 croak "Failed to commit, invalid sha1: $commit\n";
1044 }
1045 my @update_ref = ('git-update-ref',"refs/remotes/$GIT_SVN",$commit);
1046 if (my $primary_parent = shift @exec_parents) {
1047 $pid = fork;
1048 defined $pid or croak $!;
1049 if (!$pid) {
1050 close STDERR;
1051 close STDOUT;
1052 exec 'git-rev-parse','--verify',
1053 "refs/remotes/$GIT_SVN^0";
1054 }
1055 waitpid $pid, 0;
1056 push @update_ref, $primary_parent unless $?;
1057 }
1058 sys(@update_ref);
1059 sys('git-update-ref',"$GIT_SVN/revs/$log_msg->{revision}",$commit);
1060 print "r$log_msg->{revision} = $commit\n";
1061 return $commit;
1062}
1063
1064sub set_commit_env {
1065 my ($log_msg) = @_;
1066 my $author = $log_msg->{author};
1067 my ($name,$email) = defined $users{$author} ? @{$users{$author}}
1068 : ($author,"$author\@$SVN_UUID");
1069 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1070 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1071 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_msg->{date};
1072}
1073
1074sub apply_mod_line_blob {
1075 my $m = shift;
1076 if ($m->{mode_b} =~ /^120/) {
1077 blob_to_symlink($m->{sha1_b}, $m->{file_b});
1078 } else {
1079 blob_to_file($m->{sha1_b}, $m->{file_b});
1080 }
1081}
1082
1083sub blob_to_symlink {
1084 my ($blob, $link) = @_;
1085 defined $link or croak "\$link not defined!\n";
1086 croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
1087 if (-l $link || -f _) {
1088 unlink $link or croak $!;
1089 }
1090
1091 my $dest = `git-cat-file blob $blob`; # no newline, so no chomp
1092 symlink $dest, $link or croak $!;
1093}
1094
1095sub blob_to_file {
1096 my ($blob, $file) = @_;
1097 defined $file or croak "\$file not defined!\n";
1098 croak "Not a sha1: $blob\n" unless $blob =~ /^$sha1$/o;
1099 if (-l $file || -f _) {
1100 unlink $file or croak $!;
1101 }
1102
1103 open my $blob_fh, '>', $file or croak "$!: $file\n";
1104 my $pid = fork;
1105 defined $pid or croak $!;
1106
1107 if ($pid == 0) {
1108 open STDOUT, '>&', $blob_fh or croak $!;
1109 exec('git-cat-file','blob',$blob);
1110 }
1111 waitpid $pid, 0;
1112 croak $? if $?;
1113
1114 close $blob_fh or croak $!;
1115}
1116
1117sub safe_qx {
1118 my $pid = open my $child, '-|';
1119 defined $pid or croak $!;
1120 if ($pid == 0) {
1121 exec(@_) or croak $?;
1122 }
1123 my @ret = (<$child>);
1124 close $child or croak $?;
1125 die $? if $?; # just in case close didn't error out
1126 return wantarray ? @ret : join('',@ret);
1127}
1128
1129sub svn_compat_check {
1130 my @co_help = safe_qx(qw(svn co -h));
1131 unless (grep /ignore-externals/,@co_help) {
1132 print STDERR "W: Installed svn version does not support ",
1133 "--ignore-externals\n";
1134 $_no_ignore_ext = 1;
1135 }
1136 if (grep /usage: checkout URL\[\@REV\]/,@co_help) {
1137 $_svn_co_url_revs = 1;
1138 }
1139
1140 # I really, really hope nobody hits this...
1141 unless (grep /stop-on-copy/, (safe_qx(qw(svn log -h)))) {
1142 print STDERR <<'';
1143W: The installed svn version does not support the --stop-on-copy flag in
1144 the log command.
1145 Lets hope the directory you're tracking is not a branch or tag
1146 and was never moved within the repository...
1147
1148 $_no_stop_copy = 1;
1149 }
1150}
1151
1152# *sigh*, new versions of svn won't honor -r<rev> without URL@<rev>,
1153# (and they won't honor URL@<rev> without -r<rev>, too!)
1154sub svn_cmd_checkout {
1155 my ($url, $rev, $dir) = @_;
1156 my @cmd = ('svn','co', "-r$rev");
1157 push @cmd, '--ignore-externals' unless $_no_ignore_ext;
1158 $url .= "\@$rev" if $_svn_co_url_revs;
1159 sys(@cmd, $url, $dir);
1160}
1161
1162sub check_upgrade_needed {
1163 my $old = eval {
1164 my $pid = open my $child, '-|';
1165 defined $pid or croak $!;
1166 if ($pid == 0) {
1167 close STDERR;
1168 exec('git-rev-parse',"$GIT_SVN-HEAD") or croak $?;
1169 }
1170 my @ret = (<$child>);
1171 close $child or croak $?;
1172 die $? if $?; # just in case close didn't error out
1173 return wantarray ? @ret : join('',@ret);
1174 };
1175 return unless $old;
1176 my $head = eval { safe_qx('git-rev-parse',"refs/remotes/$GIT_SVN") };
1177 if ($@ || !$head) {
1178 print STDERR "Please run: $0 rebuild --upgrade\n";
1179 exit 1;
1180 }
1181}
1182
1183# fills %tree_map with a reverse mapping of trees to commits. Useful
1184# for finding parents to commit on.
1185sub map_tree_joins {
1186 foreach my $br (@_branch_from) {
1187 my $pid = open my $pipe, '-|';
1188 defined $pid or croak $!;
1189 if ($pid == 0) {
1190 exec(qw(git-rev-list --pretty=raw), $br) or croak $?;
1191 }
1192 while (<$pipe>) {
1193 if (/^commit ($sha1)$/o) {
1194 my $commit = $1;
1195 my ($tree) = (<$pipe> =~ /^tree ($sha1)$/o);
1196 unless (defined $tree) {
1197 die "Failed to parse commit $commit\n";
1198 }
1199 push @{$tree_map{$tree}}, $commit;
1200 }
1201 }
1202 close $pipe or croak $?;
1203 }
1204}
1205
1206# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1207sub load_authors {
1208 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1209 while (<$authors>) {
1210 chomp;
1211 next unless /^(\S+?)\s*=\s*(.+?)\s*<(.+)>\s*$/;
1212 my ($user, $name, $email) = ($1, $2, $3);
1213 $users{$user} = [$name, $email];
1214 }
1215 close $authors or croak $!;
1216}
1217
1218__END__
1219
1220Data structures:
1221
1222$svn_log hashref (as returned by svn_log_raw)
1223{
1224 fh => file handle of the log file,
1225 state => state of the log file parser (sep/msg/rev/msg_start...)
1226}
1227
1228$log_msg hashref as returned by next_log_entry($svn_log)
1229{
1230 msg => 'whitespace-formatted log entry
1231', # trailing newline is preserved
1232 revision => '8', # integer
1233 date => '2004-02-24T17:01:44.108345Z', # commit date
1234 author => 'committer name'
1235};
1236
1237
1238@mods = array of diff-index line hashes, each element represents one line
1239 of diff-index output
1240
1241diff-index line ($m hash)
1242{
1243 mode_a => first column of diff-index output, no leading ':',
1244 mode_b => second column of diff-index output,
1245 sha1_b => sha1sum of the final blob,
1246 chg => change type [MCRADT],
1247 file_a => original file name of a file (iff chg is 'C' or 'R')
1248 file_b => new/current file name of a file (any chg)
1249}
1250;