?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
git-svn: avoid tracking change-less revisions
[git/jnareb-git.git] / git-svn.perl
blob6ff3a8c5c56fc03a2b9851981bf174cb67bdc95c
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $SVN_URL
8 $GIT_SVN_INDEX $GIT_SVN
9 $GIT_DIR $GIT_SVN_DIR $REVDB
10 $_follow_parent $sha1 $sha1_short $_revision
11 $_cp_remote $_upgrade $_rmdir $_q $_cp_similarity
12 $_find_copies_harder $_l $_authors %users/;
13 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
14 $VERSION = '@@GIT_VERSION@@';
15
16 $ENV{GIT_DIR} ||= '.git';
17 $Git::SVN::default_repo_id = 'git-svn';
18 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
19
20 my $LC_ALL = $ENV{LC_ALL};
21 $Git::SVN::Log::TZ = $ENV{TZ};
22 # make sure the svn binary gives consistent output between locales and TZs:
23 $ENV{TZ} = 'UTC';
24 $ENV{LC_ALL} = 'C';
25 $| = 1; # unbuffer STDOUT
26
27 sub fatal (@) { print STDERR @_; exit 1 }
28 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
29 require SVN::Ra;
30 require SVN::Delta;
31 if ($SVN::Core::VERSION lt '1.1.0') {
32 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)\n";
33 }
34 push @Git::SVN::Ra::ISA, 'SVN::Ra';
35 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
36 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
37 use Carp qw/croak/;
38 use IO::File qw//;
39 use File::Basename qw/dirname basename/;
40 use File::Path qw/mkpath/;
41 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev pass_through/;
42 use IPC::Open3;
43 use Git;
44
45 BEGIN {
46 my $s;
47 foreach (qw/command command_oneline command_noisy command_output_pipe
48 command_input_pipe command_close_pipe/) {
49 $s .= "*SVN::Git::Editor::$_ = *SVN::Git::Fetcher::$_ = ".
50 "*Git::SVN::Migration::$_ = ".
51 "*Git::SVN::Log::$_ = *Git::SVN::$_ = *$_ = *Git::$_; ";
52 }
53 eval $s;
54 }
55
56 my ($SVN);
57
58 my $_optimize_commits = 1 unless $ENV{GIT_SVN_NO_OPTIMIZE_COMMITS};
59 $sha1 = qr/[a-f\d]{40}/;
60 $sha1_short = qr/[a-f\d]{4,40}/;
61 my ($_stdin, $_help, $_edit,
62 $_repack, $_repack_nr, $_repack_flags,
63 $_message, $_file, $_no_metadata,
64 $_template, $_shared,
65 $_version, $_upgrade,
66 $_merge, $_strategy, $_dry_run,
67 $_prefix);
68
69 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
70 'config-dir=s' => \$Git::SVN::Ra::config_dir,
71 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
72 my %fc_opts = ( 'follow-parent|follow' => \$_follow_parent,
73 'authors-file|A=s' => \$_authors,
74 'repack:i' => \$_repack,
75 'no-metadata' => \$_no_metadata,
76 'quiet|q' => \$_q,
77 'repack-flags|repack-args|repack-opts=s' => \$_repack_flags,
78 %remote_opts );
79
80 my ($_trunk, $_tags, $_branches);
81 my %multi_opts = ( 'trunk|T=s' => \$_trunk,
82 'tags|t=s' => \$_tags,
83 'branches|b=s' => \$_branches );
84 my %init_opts = ( 'template=s' => \$_template, 'shared' => \$_shared );
85 my %cmt_opts = ( 'edit|e' => \$_edit,
86 'rmdir' => \$_rmdir,
87 'find-copies-harder' => \$_find_copies_harder,
88 'l=i' => \$_l,
89 'copy-similarity|C=i'=> \$_cp_similarity
90 );
91
92 my %cmd = (
93 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
94 { 'revision|r=s' => \$_revision, %fc_opts } ],
95 init => [ \&cmd_init, "Initialize a repo for tracking" .
96 " (requires URL argument)",
97 \%init_opts ],
98 dcommit => [ \&cmd_dcommit,
99 'Commit several diffs to merge with upstream',
100 { 'merge|m|M' => \$_merge,
101 'strategy|s=s' => \$_strategy,
102 'dry-run|n' => \$_dry_run,
103 %cmt_opts, %fc_opts } ],
104 'set-tree' => [ \&cmd_set_tree,
105 "Set an SVN repository to a git tree-ish",
106 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
107 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
108 { 'revision|r=i' => \$_revision } ],
109 rebuild => [ \&cmd_rebuild, "Rebuild git-svn metadata (after git clone)",
110 { 'copy-remote|remote=s' => \$_cp_remote,
111 'upgrade' => \$_upgrade } ],
112 'multi-init' => [ \&cmd_multi_init,
113 'Initialize multiple trees (like git-svnimport)',
114 { %multi_opts, %init_opts, %remote_opts,
115 'revision|r=i' => \$_revision,
116 'prefix=s' => \$_prefix,
117 } ],
118 'multi-fetch' => [ \&cmd_multi_fetch,
119 'Fetch multiple trees (like git-svnimport)',
120 \%fc_opts ],
121 'migrate' => [ sub { },
122 # no-op, we automatically run this anyways,
123 'Migrate configuration/metadata/layout from
124 previous versions of git-svn',
125 \%remote_opts ],
126 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
127 { 'limit=i' => \$Git::SVN::Log::limit,
128 'revision|r=s' => \$_revision,
129 'verbose|v' => \$Git::SVN::Log::verbose,
130 'incremental' => \$Git::SVN::Log::incremental,
131 'oneline' => \$Git::SVN::Log::oneline,
132 'show-commit' => \$Git::SVN::Log::show_commit,
133 'non-recursive' => \$Git::SVN::Log::non_recursive,
134 'authors-file|A=s' => \$_authors,
135 'color' => \$Git::SVN::Log::color,
136 'pager=s' => \$Git::SVN::Log::pager,
137 } ],
138 'commit-diff' => [ \&cmd_commit_diff,
139 'Commit a diff between two trees',
140 { 'message|m=s' => \$_message,
141 'file|F=s' => \$_file,
142 'revision|r=s' => \$_revision,
143 %cmt_opts } ],
144 );
145
146 my $cmd;
147 for (my $i = 0; $i < @ARGV; $i++) {
148 if (defined $cmd{$ARGV[$i]}) {
149 $cmd = $ARGV[$i];
150 splice @ARGV, $i, 1;
151 last;
152 }
153 };
154
155 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
156
157 read_repo_config(\%opts);
158 my $rv = GetOptions(%opts, 'help|H|h' => \$_help,
159 'version|V' => \$_version,
160 'minimize-connections' =>
161 \$Git::SVN::Migration::_minimize,
162 'id|i=s' => \$Git::SVN::default_ref_id);
163 exit 1 if (!$rv && $cmd ne 'log');
164
165 usage(0) if $_help;
166 version() if $_version;
167 usage(1) unless defined $cmd;
168 load_authors() if $_authors;
169 unless ($cmd =~ /^(?:init|rebuild|multi-init|commit-diff)$/) {
170 Git::SVN::Migration::migration_check();
171 }
172 eval {
173 Git::SVN::verify_remotes_sanity();
174 $cmd{$cmd}->[0]->(@ARGV);
175 };
176 fatal $@ if $@;
177 exit 0;
178
179 ####################### primary functions ######################
180 sub usage {
181 my $exit = shift || 0;
182 my $fd = $exit ? \*STDERR : \*STDOUT;
183 print $fd <<"";
184 git-svn - bidirectional operations between a single Subversion tree and git
185 Usage: $0 <command> [options] [arguments]\n
186
187 print $fd "Available commands:\n" unless $cmd;
188
189 foreach (sort keys %cmd) {
190 next if $cmd && $cmd ne $_;
191 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
192 foreach (keys %{$cmd{$_}->[2]}) {
193 # prints out arguments as they should be passed:
194 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
195 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
196 "--$_" : "-$_" }
197 split /\|/,$_)," $x\n";
198 }
199 }
200 print $fd <<"";
201 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
202 arbitrary identifier if you're tracking multiple SVN branches/repositories in
203 one git repository and want to keep them separate. See git-svn(1) for more
204 information.
205
206 exit $exit;
207 }
208
209 sub version {
210 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
211 exit 0;
212 }
213
214 sub cmd_rebuild {
215 my $url = shift;
216 my $gs = $url ? Git::SVN->init($url)
217 : eval { Git::SVN->new };
218 $gs ||= Git::SVN->_new;
219 if (!verify_ref($gs->refname.'^0')) {
220 $gs->copy_remote_ref;
221 }
222
223 my ($rev_list, $ctx) = command_output_pipe("rev-list", $gs->refname);
224 my $latest;
225 my $svn_uuid;
226 while (<$rev_list>) {
227 chomp;
228 my $c = $_;
229 fatal "Non-SHA1: $c\n" unless $c =~ /^$sha1$/o;
230 my ($url, $rev, $uuid) = cmt_metadata($c);
231
232 # ignore merges (from set-tree)
233 next if (!defined $rev || !$uuid);
234
235 # if we merged or otherwise started elsewhere, this is
236 # how we break out of it
237 if ((defined $svn_uuid && ($uuid ne $svn_uuid)) ||
238 ($gs->{url} && $url && ($url ne $gs->{url}))) {
239 next;
240 }
241
242 unless (defined $latest) {
243 if (!$gs->{url} && !$url) {
244 fatal "SVN repository location required\n";
245 }
246 $gs = Git::SVN->init($url);
247 $latest = $rev;
248 }
249 $gs->rev_db_set($rev, $c);
250 print "r$rev = $c\n";
251 }
252 command_close_pipe($rev_list, $ctx);
253 }
254
255 sub do_git_init_db {
256 unless (-d $ENV{GIT_DIR}) {
257 my @init_db = ('init');
258 push @init_db, "--template=$_template" if defined $_template;
259 push @init_db, "--shared" if defined $_shared;
260 command_noisy(@init_db);
261 }
262 }
263
264 sub cmd_init {
265 my $url = shift or die "SVN repository location required " .
266 "as a command-line argument\n";
267 if (my $repo_path = shift) {
268 unless (-d $repo_path) {
269 mkpath([$repo_path]);
270 }
271 chdir $repo_path or croak $!;
272 $ENV{GIT_DIR} = $repo_path . "/.git";
273 }
274 do_git_init_db();
275
276 Git::SVN->init($url);
277 }
278
279 sub cmd_fetch {
280 if (@_) {
281 die "Additional fetch arguments are no longer supported.\n",
282 "Use --follow-parent if you have moved/copied directories
283 instead.\n";
284 }
285 my $gs = Git::SVN->new;
286 $gs->fetch;
287 if ($gs->{last_commit} && !verify_ref('refs/heads/master^0')) {
288 command_noisy(qw(update-ref refs/heads/master),
289 $gs->{last_commit});
290 }
291 }
292
293 sub cmd_set_tree {
294 my (@commits) = @_;
295 if ($_stdin || !@commits) {
296 print "Reading from stdin...\n";
297 @commits = ();
298 while (<STDIN>) {
299 if (/\b($sha1_short)\b/o) {
300 unshift @commits, $1;
301 }
302 }
303 }
304 my @revs;
305 foreach my $c (@commits) {
306 my @tmp = command('rev-parse',$c);
307 if (scalar @tmp == 1) {
308 push @revs, $tmp[0];
309 } elsif (scalar @tmp > 1) {
310 push @revs, reverse(command('rev-list',@tmp));
311 } else {
312 fatal "Failed to rev-parse $c\n";
313 }
314 }
315 my $gs = Git::SVN->new;
316 my ($r_last, $cmt_last) = $gs->last_rev_commit;
317 $gs->fetch;
318 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
319 fatal "There are new revisions that were fetched ",
320 "and need to be merged (or acknowledged) ",
321 "before committing.\nlast rev: $r_last\n",
322 " current: $gs->{last_rev}\n";
323 }
324 $gs->set_tree($_) foreach @revs;
325 print "Done committing ",scalar @revs," revisions to SVN\n";
326 }
327
328 sub cmd_dcommit {
329 my $head = shift;
330 my $gs = Git::SVN->new;
331 $head ||= 'HEAD';
332 my @refs = command(qw/rev-list --no-merges/, $gs->refname."..$head");
333 my $last_rev;
334 foreach my $d (reverse @refs) {
335 if (!verify_ref("$d~1")) {
336 fatal "Commit $d\n",
337 "has no parent commit, and therefore ",
338 "nothing to diff against.\n",
339 "You should be working from a repository ",
340 "originally created by git-svn\n";
341 }
342 unless (defined $last_rev) {
343 (undef, $last_rev, undef) = cmt_metadata("$d~1");
344 unless (defined $last_rev) {
345 fatal "Unable to extract revision information ",
346 "from commit $d~1\n";
347 }
348 }
349 if ($_dry_run) {
350 print "diff-tree $d~1 $d\n";
351 } else {
352 my $log = get_commit_entry($d)->{log};
353 my $ra = $gs->ra;
354 my $pool = SVN::Pool->new;
355 my %ed_opts = ( r => $last_rev,
356 ra => $ra->dup,
357 svn_path => $ra->{svn_path} );
358 my $ed = SVN::Git::Editor->new(\%ed_opts,
359 $ra->get_commit_editor($log,
360 sub { print "Committed r$_[0]\n";
361 $last_rev = $_[0]; }),
362 $pool);
363 my $mods = $ed->apply_diff("$d~1", $d);
364 if (@$mods == 0) {
365 print "No changes\n$d~1 == $d\n";
366 }
367 }
368 }
369 return if $_dry_run;
370 $gs->fetch;
371 # we always want to rebase against the current HEAD, not any
372 # head that was passed to us
373 my @diff = command('diff-tree', 'HEAD', $gs->refname, '--');
374 my @finish;
375 if (@diff) {
376 @finish = qw/rebase/;
377 push @finish, qw/--merge/ if $_merge;
378 push @finish, "--strategy=$_strategy" if $_strategy;
379 print STDERR "W: HEAD and ", $gs->refname, " differ, ",
380 "using @finish:\n", "@diff";
381 } else {
382 print "No changes between current HEAD and ",
383 $gs->refname, "\nResetting to the latest ",
384 $gs->refname, "\n";
385 @finish = qw/reset --mixed/;
386 }
387 command_noisy(@finish, $gs->refname);
388 }
389
390 sub cmd_show_ignore {
391 my $gs = Git::SVN->new;
392 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
393 $gs->traverse_ignore(\*STDOUT, '', $r);
394 }
395
396 sub cmd_multi_init {
397 my $url = shift;
398 unless (defined $_trunk || defined $_branches || defined $_tags) {
399 usage(1);
400 }
401 do_git_init_db();
402 $_prefix = '' unless defined $_prefix;
403 $url =~ s#/+$## if defined $url;
404 if (defined $_trunk) {
405 my $trunk_ref = $_prefix . 'trunk';
406 # try both old-style and new-style lookups:
407 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
408 unless ($gs_trunk) {
409 my ($trunk_url, $trunk_path) =
410 complete_svn_url($url, $_trunk);
411 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
412 undef, $trunk_ref);
413 }
414 }
415 return unless defined $_branches || defined $_tags;
416 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
417 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
418 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
419 }
420
421 sub cmd_multi_fetch {
422 my @gs;
423 foreach (command(qw/config -l/)) {
424 next unless m!^svn-remote\.(.+)\.fetch=
425 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
426 my ($repo_id, $path, $ref_id) = ($1, $2, $3);
427 push @gs, Git::SVN->new($ref_id, $repo_id, $path);
428 }
429 foreach (@gs) {
430 $_->fetch;
431 }
432 }
433
434 # this command is special because it requires no metadata
435 sub cmd_commit_diff {
436 my ($ta, $tb, $url) = @_;
437 my $usage = "Usage: $0 commit-diff -r<revision> ".
438 "<tree-ish> <tree-ish> [<URL>]\n";
439 fatal($usage) if (!defined $ta || !defined $tb);
440 if (!defined $url) {
441 my $gs = eval { Git::SVN->new };
442 if (!$gs) {
443 fatal("Needed URL or usable git-svn --id in ",
444 "the command-line\n", $usage);
445 }
446 $url = $gs->{url};
447 }
448 unless (defined $_revision) {
449 fatal("-r|--revision is a required argument\n", $usage);
450 }
451 if (defined $_message && defined $_file) {
452 fatal("Both --message/-m and --file/-F specified ",
453 "for the commit message.\n",
454 "I have no idea what you mean\n");
455 }
456 if (defined $_file) {
457 $_message = file_to_s($_file);
458 } else {
459 $_message ||= get_commit_entry($tb)->{log};
460 }
461 my $ra ||= Git::SVN::Ra->new($url);
462 my $r = $_revision;
463 if ($r eq 'HEAD') {
464 $r = $ra->get_latest_revnum;
465 } elsif ($r !~ /^\d+$/) {
466 die "revision argument: $r not understood by git-svn\n";
467 }
468 my $pool = SVN::Pool->new;
469 my %ed_opts = ( r => $r,
470 ra => $ra->dup,
471 svn_path => $ra->{svn_path} );
472 my $ed = SVN::Git::Editor->new(\%ed_opts,
473 $ra->get_commit_editor($_message,
474 sub { print "Committed r$_[0]\n" }),
475 $pool);
476 my $mods = $ed->apply_diff($ta, $tb);
477 if (@$mods == 0) {
478 print "No changes\n$ta == $tb\n";
479 }
480 $pool->clear;
481 }
482
483 ########################### utility functions #########################
484
485 sub complete_svn_url {
486 my ($url, $path) = @_;
487 $path =~ s#/+$##;
488 if ($path !~ m#^[a-z\+]+://#) {
489 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
490 fatal("E: '$path' is not a complete URL ",
491 "and a separate URL is not specified\n");
492 }
493 return ($url, $path);
494 }
495 return ($path, '');
496 }
497
498 sub complete_url_ls_init {
499 my ($ra, $repo_path, $switch, $pfx) = @_;
500 unless ($repo_path) {
501 print STDERR "W: $switch not specified\n";
502 return;
503 }
504 $repo_path =~ s#/+$##;
505 if ($repo_path =~ m#^[a-z\+]+://#) {
506 $ra = Git::SVN::Ra->new($repo_path);
507 $repo_path = '';
508 } else {
509 $repo_path =~ s#^/+##;
510 unless ($ra) {
511 fatal("E: '$repo_path' is not a complete URL ",
512 "and a separate URL is not specified\n");
513 }
514 }
515 my $r = defined $_revision ? $_revision : $ra->get_latest_revnum;
516 my ($dirent, undef, undef) = $ra->get_dir($repo_path, $r);
517 my $url = $ra->{url};
518 foreach my $d (sort keys %$dirent) {
519 next if ($dirent->{$d}->kind != $SVN::Node::dir);
520 my $path = "$repo_path/$d";
521 my $ref = "$pfx$d";
522 my $gs = eval { Git::SVN->new($ref) };
523 # don't try to init already existing refs
524 unless ($gs) {
525 print "init $url/$path => $ref\n";
526 Git::SVN->init($url, $path, undef, $ref);
527 }
528 }
529 }
530
531 sub verify_ref {
532 my ($ref) = @_;
533 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
534 { STDERR => 0 }); };
535 }
536
537 sub get_tree_from_treeish {
538 my ($treeish) = @_;
539 # $treeish can be a symbolic ref, too:
540 my $type = command_oneline(qw/cat-file -t/, $treeish);
541 my $expected;
542 while ($type eq 'tag') {
543 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
544 }
545 if ($type eq 'commit') {
546 $expected = (grep /^tree /, command(qw/cat-file commit/,
547 $treeish))[0];
548 ($expected) = ($expected =~ /^tree ($sha1)$/o);
549 die "Unable to get tree from $treeish\n" unless $expected;
550 } elsif ($type eq 'tree') {
551 $expected = $treeish;
552 } else {
553 die "$treeish is a $type, expected tree, tag or commit\n";
554 }
555 return $expected;
556 }
557
558 sub get_commit_entry {
559 my ($treeish) = shift;
560 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
561 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
562 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
563 open my $log_fh, '>', $commit_editmsg or croak $!;
564
565 my $type = command_oneline(qw/cat-file -t/, $treeish);
566 if ($type eq 'commit' || $type eq 'tag') {
567 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
568 $type, $treeish);
569 my $in_msg = 0;
570 while (<$msg_fh>) {
571 if (!$in_msg) {
572 $in_msg = 1 if (/^\s*$/);
573 } elsif (/^git-svn-id: /) {
574 # skip this for now, we regenerate the
575 # correct one on re-fetch anyways
576 # TODO: set *:merge properties or like...
577 } else {
578 print $log_fh $_ or croak $!;
579 }
580 }
581 command_close_pipe($msg_fh, $ctx);
582 }
583 close $log_fh or croak $!;
584
585 if ($_edit || ($type eq 'tree')) {
586 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
587 # TODO: strip out spaces, comments, like git-commit.sh
588 system($editor, $commit_editmsg);
589 }
590 rename $commit_editmsg, $commit_msg or croak $!;
591 open $log_fh, '<', $commit_msg or croak $!;
592 { local $/; chomp($log_entry{log} = <$log_fh>); }
593 close $log_fh or croak $!;
594 unlink $commit_msg;
595 \%log_entry;
596 }
597
598 sub s_to_file {
599 my ($str, $file, $mode) = @_;
600 open my $fd,'>',$file or croak $!;
601 print $fd $str,"\n" or croak $!;
602 close $fd or croak $!;
603 chmod ($mode &~ umask, $file) if (defined $mode);
604 }
605
606 sub file_to_s {
607 my $file = shift;
608 open my $fd,'<',$file or croak "$!: file: $file\n";
609 local $/;
610 my $ret = <$fd>;
611 close $fd or croak $!;
612 $ret =~ s/\s*$//s;
613 return $ret;
614 }
615
616 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
617 sub load_authors {
618 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
619 my $log = $cmd eq 'log';
620 while (<$authors>) {
621 chomp;
622 next unless /^(\S+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
623 my ($user, $name, $email) = ($1, $2, $3);
624 if ($log) {
625 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
626 } else {
627 $users{$user} = [$name, $email];
628 }
629 }
630 close $authors or croak $!;
631 }
632
633 # convert GetOpt::Long specs for use by git-config
634 sub read_repo_config {
635 return unless -d $ENV{GIT_DIR};
636 my $opts = shift;
637 foreach my $o (keys %$opts) {
638 my $v = $opts->{$o};
639 my ($key) = ($o =~ /^([a-z\-]+)/);
640 $key =~ s/-//g;
641 my $arg = 'git-config';
642 $arg .= ' --int' if ($o =~ /[:=]i$/);
643 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
644 if (ref $v eq 'ARRAY') {
645 chomp(my @tmp = `$arg --get-all svn.$key`);
646 @$v = @tmp if @tmp;
647 } else {
648 chomp(my $tmp = `$arg --get svn.$key`);
649 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
650 $$v = $tmp;
651 }
652 }
653 }
654 }
655
656 sub extract_metadata {
657 my $id = shift or return (undef, undef, undef);
658 my ($url, $rev, $uuid) = ($id =~ /^git-svn-id:\s(\S+?)\@(\d+)
659 \s([a-f\d\-]+)$/x);
660 if (!defined $rev || !$uuid || !$url) {
661 # some of the original repositories I made had
662 # identifiers like this:
663 ($rev, $uuid) = ($id =~/^git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
664 }
665 return ($url, $rev, $uuid);
666 }
667
668 sub cmt_metadata {
669 return extract_metadata((grep(/^git-svn-id: /,
670 command(qw/cat-file commit/, shift)))[-1]);
671 }
672
673 sub get_commit_time {
674 my $cmt = shift;
675 my $fh = command_output_pipe(qw/rev-list --pretty=raw -n1/, $cmt);
676 while (<$fh>) {
677 /^committer\s(?:.+) (\d+) ([\-\+]?\d+)$/ or next;
678 my ($s, $tz) = ($1, $2);
679 if ($tz =~ s/^\+//) {
680 $s += tz_to_s_offset($tz);
681 } elsif ($tz =~ s/^\-//) {
682 $s -= tz_to_s_offset($tz);
683 }
684 close $fh;
685 return $s;
686 }
687 die "Can't get commit time for commit: $cmt\n";
688 }
689
690 sub tz_to_s_offset {
691 my ($tz) = @_;
692 $tz =~ s/(\d\d)$//;
693 return ($1 * 60) + ($tz * 3600);
694 }
695
696 package Git::SVN;
697 use strict;
698 use warnings;
699 use vars qw/$default_repo_id $default_ref_id/;
700 use Carp qw/croak/;
701 use File::Path qw/mkpath/;
702 use IPC::Open3;
703
704 # properties that we do not log:
705 my %SKIP_PROP;
706 BEGIN {
707 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
708 svn:special svn:executable
709 svn:entry:committed-rev
710 svn:entry:last-author
711 svn:entry:uuid
712 svn:entry:committed-date/;
713 }
714
715 sub read_all_remotes {
716 my $r = {};
717 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
718 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
719 $r->{$1}->{fetch}->{$2} = $3;
720 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
721 $r->{$1}->{url} = $2;
722 }
723 }
724 $r;
725 }
726
727 sub verify_remotes_sanity {
728 return unless -d $ENV{GIT_DIR};
729 my %seen;
730 foreach (command(qw/config -l/)) {
731 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
732 if ($seen{$1}) {
733 die "Remote ref refs/remote/$1 is tracked by",
734 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
735 "Please resolve this ambiguity in ",
736 "your git configuration file before ",
737 "continuing\n";
738 }
739 $seen{$1} = $_;
740 }
741 }
742 }
743
744 # we allow more chars than remotes2config.sh...
745 sub sanitize_remote_name {
746 my ($name) = @_;
747 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
748 $name;
749 }
750
751 sub find_existing_remote {
752 my ($url, $remotes) = @_;
753 my $existing;
754 foreach my $repo_id (keys %$remotes) {
755 my $u = $remotes->{$repo_id}->{url} or next;
756 next if $u ne $url;
757 $existing = $repo_id;
758 last;
759 }
760 $existing;
761 }
762
763 sub init_remote_config {
764 my ($self, $url) = @_;
765 $url =~ s!/+$!!; # strip trailing slash
766 my $r = read_all_remotes();
767 my $existing = find_existing_remote($url, $r);
768 if ($existing) {
769 print STDERR "Using existing ",
770 "[svn-remote \"$existing\"]\n";
771 $self->{repo_id} = $existing;
772 } else {
773 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
774 $existing = find_existing_remote($min_url, $r);
775 if ($existing) {
776 print STDERR "Using existing ",
777 "[svn-remote \"$existing\"]\n";
778 $self->{repo_id} = $existing;
779 }
780 if ($min_url ne $url) {
781 print STDERR "Using higher level of URL: ",
782 "$url => $min_url\n";
783 my $old_path = $self->{path};
784 $self->{path} = $url;
785 $self->{path} =~ s!^\Q$min_url\E/*!!;
786 if (length $old_path) {
787 $self->{path} .= "/$old_path";
788 }
789 $url = $min_url;
790 }
791 }
792 my $orig_url;
793 if (!$existing) {
794 # verify that we aren't overwriting anything:
795 $orig_url = eval {
796 command_oneline('config', '--get',
797 "svn-remote.$self->{repo_id}.url")
798 };
799 if ($orig_url && ($orig_url ne $url)) {
800 die "svn-remote.$self->{repo_id}.url already set: ",
801 "$orig_url\nwanted to set to: $url\n";
802 }
803 }
804 my ($xrepo_id, $xpath) = find_ref($self->refname);
805 if (defined $xpath) {
806 die "svn-remote.$xrepo_id.fetch already set to track ",
807 "$xpath:refs/remotes/", $self->refname, "\n";
808 }
809 command_noisy('config',
810 "svn-remote.$self->{repo_id}.url", $url);
811 command_noisy('config', '--add',
812 "svn-remote.$self->{repo_id}.fetch",
813 "$self->{path}:".$self->refname);
814 $self->{url} = $url;
815 }
816
817 sub init {
818 my ($class, $url, $path, $repo_id, $ref_id) = @_;
819 my $self = _new($class, $repo_id, $ref_id, $path);
820 if (defined $url) {
821 $self->init_remote_config($url);
822 }
823 $self;
824 }
825
826 sub find_ref {
827 my ($ref_id) = @_;
828 foreach (command(qw/config -l/)) {
829 next unless m!^svn-remote\.(.+)\.fetch=
830 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
831 my ($repo_id, $path, $ref) = ($1, $2, $3);
832 if ($ref eq $ref_id) {
833 $path = '' if ($path =~ m#^\./?#);
834 return ($repo_id, $path);
835 }
836 }
837 (undef, undef, undef);
838 }
839
840 sub new {
841 my ($class, $ref_id, $repo_id, $path) = @_;
842 if (defined $ref_id && !defined $repo_id && !defined $path) {
843 ($repo_id, $path) = find_ref($ref_id);
844 if (!defined $repo_id) {
845 die "Could not find a \"svn-remote.*.fetch\" key ",
846 "in the repository configuration matching: ",
847 "refs/remotes/$ref_id\n";
848 }
849 }
850 my $self = _new($class, $repo_id, $ref_id, $path);
851 if (!defined $self->{path} || !length $self->{path}) {
852 my $fetch = command_oneline('config', '--get',
853 "svn-remote.$repo_id.fetch",
854 ":refs/remotes/$ref_id\$") or
855 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
856 "\":refs/remotes/$ref_id\$\" in config\n";
857 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
858 }
859 $self->{url} = command_oneline('config', '--get',
860 "svn-remote.$repo_id.url") or
861 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
862 $self;
863 }
864
865 sub refname { "refs/remotes/$_[0]->{ref_id}" }
866
867 sub ra {
868 my ($self) = shift;
869 $self->{ra} ||= Git::SVN::Ra->new($self->{url});
870 }
871
872 sub rel_path {
873 my ($self) = @_;
874 my $repos_root = $self->ra->{repos_root};
875 return $self->{path} if ($self->{url} eq $repos_root);
876 my $url = $self->{url} .
877 (length $self->{path} ? "/$self->{path}" : $self->{path});
878 $url =~ s!^\Q$repos_root\E/*!!g;
879 $url;
880 }
881
882 sub copy_remote_ref {
883 my ($self) = @_;
884 my $origin = $::_cp_remote ? $::_cp_remote : 'origin';
885 my $ref = $self->refname;
886 if (command('ls-remote', $origin, $ref)) {
887 command_noisy('fetch', $origin, "$ref:$ref");
888 } elsif ($::_cp_remote && !$::_upgrade) {
889 die "Unable to find remote reference: $ref on $origin\n";
890 }
891 }
892
893 sub traverse_ignore {
894 my ($self, $fh, $path, $r) = @_;
895 $path =~ s#^/+##g;
896 my ($dirent, undef, $props) = $self->ra->get_dir($path, $r);
897 my $p = $path;
898 $p =~ s#^\Q$self->{ra}->{svn_path}\E/##;
899 print $fh length $p ? "\n# $p\n" : "\n# /\n";
900 if (my $s = $props->{'svn:ignore'}) {
901 $s =~ s/[\r\n]+/\n/g;
902 chomp $s;
903 if (length $p == 0) {
904 $s =~ s#\n#\n/$p#g;
905 print $fh "/$s\n";
906 } else {
907 $s =~ s#\n#\n/$p/#g;
908 print $fh "/$p/$s\n";
909 }
910 }
911 foreach (sort keys %$dirent) {
912 next if $dirent->{$_}->kind != $SVN::Node::dir;
913 $self->traverse_ignore($fh, "$path/$_", $r);
914 }
915 }
916
917 # returns the newest SVN revision number and newest commit SHA1
918 sub last_rev_commit {
919 my ($self) = @_;
920 if (defined $self->{last_rev} && defined $self->{last_commit}) {
921 return ($self->{last_rev}, $self->{last_commit});
922 }
923 my $c = ::verify_ref($self->refname.'^0');
924 if ($c) {
925 my $rev = (::cmt_metadata($c))[1];
926 if (defined $rev) {
927 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
928 return ($rev, $c);
929 }
930 }
931 my $offset = -41; # from tail
932 my $rl;
933 open my $fh, '<', $self->{db_path} or
934 croak "$self->{db_path} not readable: $!\n";
935 seek $fh, $offset, 2;
936 $rl = readline $fh;
937 defined $rl or return (undef, undef);
938 chomp $rl;
939 while ($c ne $rl && tell $fh != 0) {
940 $offset -= 41;
941 seek $fh, $offset, 2;
942 $rl = readline $fh;
943 defined $rl or return (undef, undef);
944 chomp $rl;
945 }
946 my $rev = tell $fh;
947 croak $! if ($rev < 0);
948 $rev = ($rev - 41) / 41;
949 close $fh or croak $!;
950 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
951 return ($rev, $c);
952 }
953
954 sub parse_revision {
955 my ($self, $base) = @_;
956 my $head = $self->ra->get_latest_revnum;
957 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
958 return ($base + 1, $head) if (defined $base);
959 return (0, $head);
960 }
961 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
962 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
963 if ($::_revision =~ /^BASE:(\d+)$/) {
964 return ($base + 1, $1) if (defined $base);
965 return (0, $head);
966 }
967 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
968 die "revision argument: $::_revision not understood by git-svn\n",
969 "Try using the command-line svn client instead\n";
970 }
971
972 sub tmp_index_do {
973 my ($self, $sub) = @_;
974 my $old_index = $ENV{GIT_INDEX_FILE};
975 $ENV{GIT_INDEX_FILE} = $self->{index};
976 my @ret = &$sub;
977 if ($old_index) {
978 $ENV{GIT_INDEX_FILE} = $old_index;
979 } else {
980 delete $ENV{GIT_INDEX_FILE};
981 }
982 wantarray ? @ret : $ret[0];
983 }
984
985 sub assert_index_clean {
986 my ($self, $treeish) = @_;
987
988 $self->tmp_index_do(sub {
989 command_noisy('read-tree', $treeish) unless -e $self->{index};
990 my $x = command_oneline('write-tree');
991 my ($y) = (command(qw/cat-file commit/, $treeish) =~
992 /^tree ($::sha1)/mo);
993 if ($y ne $x) {
994 unlink $self->{index} or croak $!;
995 command_noisy('read-tree', $treeish);
996 }
997 $x = command_oneline('write-tree');
998 if ($y ne $x) {
999 ::fatal "trees ($treeish) $y != $x\n",
1000 "Something is seriously wrong...\n";
1001 }
1002 });
1003 }
1004
1005 sub get_commit_parents {
1006 my ($self, $log_entry, @parents) = @_;
1007 my (%seen, @ret, @tmp);
1008 # commit parents can be conditionally bound to a particular
1009 # svn revision via: "svn_revno=commit_sha1", filter them out here:
1010 foreach my $p (@parents) {
1011 next unless defined $p;
1012 if ($p =~ /^(\d+)=($::sha1_short)$/o) {
1013 push @tmp, $2 if $1 == $log_entry->{revision};
1014 } else {
1015 push @tmp, $p if $p =~ /^$::sha1_short$/o;
1016 }
1017 }
1018 if (my $cur = ::verify_ref($self->refname.'^0')) {
1019 push @tmp, $cur;
1020 }
1021 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
1022 while (my $p = shift @tmp) {
1023 next if $seen{$p};
1024 $seen{$p} = 1;
1025 push @ret, $p;
1026 # MAXPARENT is defined to 16 in commit-tree.c:
1027 last if @ret >= 16;
1028 }
1029 if (@tmp) {
1030 die "r$log_entry->{revision}: No room for parents:\n\t",
1031 join("\n\t", @tmp), "\n";
1032 }
1033 @ret;
1034 }
1035
1036 sub full_url {
1037 my ($self) = @_;
1038 $self->ra->{url} . (length $self->{path} ? '/' . $self->{path} : '');
1039 }
1040
1041 sub do_git_commit {
1042 my ($self, $log_entry, @parents) = @_;
1043 if (my $c = $self->rev_db_get($log_entry->{revision})) {
1044 croak "$log_entry->{revision} = $c already exists! ",
1045 "Why are we refetching it?\n";
1046 }
1047 my $author = $log_entry->{author};
1048 my ($name, $email) = (defined $::users{$author} ? @{$::users{$author}}
1049 : ($author, "$author\@".$self->ra->uuid));
1050 $ENV{GIT_AUTHOR_NAME} = $ENV{GIT_COMMITTER_NAME} = $name;
1051 $ENV{GIT_AUTHOR_EMAIL} = $ENV{GIT_COMMITTER_EMAIL} = $email;
1052 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
1053
1054 my $tree = $log_entry->{tree};
1055 if (!defined $tree) {
1056 $tree = $self->tmp_index_do(sub {
1057 command_oneline('write-tree') });
1058 }
1059 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
1060
1061 my @exec = ('git-commit-tree', $tree);
1062 foreach ($self->get_commit_parents($log_entry, @parents)) {
1063 push @exec, '-p', $_;
1064 }
1065 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
1066 or croak $!;
1067 print $msg_fh $log_entry->{log} or croak $!;
1068 print $msg_fh "\ngit-svn-id: ", $self->full_url, '@',
1069 $log_entry->{revision}, ' ',
1070 $self->ra->uuid, "\n" or croak $!;
1071 $msg_fh->flush == 0 or croak $!;
1072 close $msg_fh or croak $!;
1073 chomp(my $commit = do { local $/; <$out_fh> });
1074 close $out_fh or croak $!;
1075 waitpid $pid, 0;
1076 croak $? if $?;
1077 if ($commit !~ /^$::sha1$/o) {
1078 die "Failed to commit, invalid sha1: $commit\n";
1079 }
1080
1081 command_noisy('update-ref',$self->refname, $commit);
1082 $self->rev_db_set($log_entry->{revision}, $commit);
1083
1084 $self->{last_rev} = $log_entry->{revision};
1085 $self->{last_commit} = $commit;
1086 print "r$log_entry->{revision} = $commit\n";
1087 return $commit;
1088 }
1089
1090 sub revisions_eq {
1091 my ($self, $r0, $r1) = @_;
1092 return 1 if $r0 == $r1;
1093 my $nr = 0;
1094 $self->ra->get_log([$self->{path}], $r0, $r1,
1095 0, 0, 1, sub { $nr++ });
1096 return 0 if ($nr > 1);
1097 return 1;
1098 }
1099
1100 sub find_parent_branch {
1101 my ($self, $paths, $rev) = @_;
1102 return undef unless $::_follow_parent;
1103
1104 # look for a parent from another branch:
1105 my @b_path_components = split m#/#, $self->rel_path;
1106 my @a_path_components;
1107 my $i;
1108 while (@b_path_components) {
1109 $i = $paths->{'/'.join('/', @b_path_components)};
1110 last if $i;
1111 unshift(@a_path_components, pop(@b_path_components));
1112 }
1113 goto not_found unless defined $i;
1114 my $branch_from = $i->copyfrom_path or goto not_found;
1115 if (@a_path_components) {
1116 print STDERR "branch_from: $branch_from => ";
1117 $branch_from .= '/'.join('/', @a_path_components);
1118 print STDERR $branch_from, "\n";
1119 }
1120 my $r = $i->copyfrom_rev;
1121 my $repos_root = $self->ra->{repos_root};
1122 my $url = $self->ra->{url};
1123 my $new_url = $repos_root . $branch_from;
1124 print STDERR "Found possible branch point: ",
1125 "$new_url => ", $self->full_url, ", $r\n";
1126 $branch_from =~ s#^/##;
1127 my $remotes = read_all_remotes();
1128 my $gs;
1129 foreach my $repo_id (keys %$remotes) {
1130 my $u = $remotes->{$repo_id}->{url} or next;
1131 next if $url ne $u;
1132 my $fetch = $remotes->{$repo_id}->{fetch};
1133 foreach my $f (keys %$fetch) {
1134 next if $f ne $branch_from;
1135 $gs = Git::SVN->new($fetch->{$f}, $repo_id, $f);
1136 last;
1137 }
1138 last if $gs;
1139 }
1140 unless ($gs) {
1141 my $ref_id = $branch_from;
1142 $ref_id .= "\@$r" if find_ref($ref_id);
1143 # just grow a tail if we're not unique enough :x
1144 $ref_id .= '-' while find_ref($ref_id);
1145 $gs = Git::SVN->init($new_url, '', $ref_id, $ref_id);
1146 }
1147 my ($r0, $parent) = $gs->find_rev_before($r, 1);
1148 if ($::_follow_parent && (!defined $r0 || !defined $parent)) {
1149 $gs->ra->get_log([$gs->{path}], 0, $r, 0, 1, 1, sub {
1150 my ($paths, $rev) = @_;
1151 my $log_entry = eval { $gs->do_fetch($paths, $rev) };
1152 $gs->do_git_commit($log_entry) if $log_entry;
1153 });
1154 ($r0, $parent) = $gs->last_rev_commit;
1155 }
1156 if (defined $r0 && defined $parent && $gs->revisions_eq($r0, $r)) {
1157 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
1158 $self->assert_index_clean($parent);
1159 my $ed;
1160 if ($self->ra->can_do_switch) {
1161 print STDERR "Following parent with do_switch\n";
1162 # do_switch works with svn/trunk >= r22312, but that
1163 # is not included with SVN 1.4.2 (the latest version
1164 # at the moment), so we can't rely on it
1165 $self->{last_commit} = $parent;
1166 $ed = SVN::Git::Fetcher->new($self);
1167 $gs->ra->gs_do_switch($r0, $rev, $gs->{path}, 1,
1168 $self->full_url, $ed)
1169 or die "SVN connection failed somewhere...\n";
1170 } else {
1171 print STDERR "Following parent with do_update\n";
1172 $ed = SVN::Git::Fetcher->new($self);
1173 $self->ra->gs_do_update($rev, $rev, $self->{path},
1174 1, $ed)
1175 or die "SVN connection failed somewhere...\n";
1176 }
1177 return $self->make_log_entry($rev, [$parent], $ed);
1178 }
1179 not_found:
1180 print STDERR "Branch parent for path: '/",
1181 $self->rel_path, "' not found\n";
1182 return undef unless $paths;
1183 foreach my $x (sort keys %$paths) {
1184 my $p = $paths->{$x};
1185 print STDERR ' ', $p->action, ' ', $x;
1186 if (my $cp_from = $p->copyfrom_path) {
1187 print STDERR "(from $cp_from:", $p->copyfrom_rev, ')';
1188 }
1189 print STDERR "\n";
1190 }
1191 return undef;
1192 }
1193
1194 sub do_fetch {
1195 my ($self, $paths, $rev) = @_;
1196 my $ed;
1197 my ($last_rev, @parents);
1198 if ($self->{last_commit}) {
1199 $ed = SVN::Git::Fetcher->new($self);
1200 $last_rev = $self->{last_rev};
1201 $ed->{c} = $self->{last_commit};
1202 @parents = ($self->{last_commit});
1203 } else {
1204 $last_rev = $rev;
1205 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
1206 return $log_entry;
1207 }
1208 $ed = SVN::Git::Fetcher->new($self);
1209 }
1210 unless ($self->ra->gs_do_update($last_rev, $rev,
1211 $self->{path}, 1, $ed)) {
1212 die "SVN connection failed somewhere...\n";
1213 }
1214 $self->make_log_entry($rev, \@parents, $ed);
1215 }
1216
1217 sub get_untracked {
1218 my ($self, $ed) = @_;
1219 my @out;
1220 my $h = $ed->{empty};
1221 foreach (sort keys %$h) {
1222 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
1223 push @out, " $act: " . uri_encode($_);
1224 warn "W: $act: $_\n";
1225 }
1226 foreach my $t (qw/dir_prop file_prop/) {
1227 $h = $ed->{$t} or next;
1228 foreach my $path (sort keys %$h) {
1229 my $ppath = $path eq '' ? '.' : $path;
1230 foreach my $prop (sort keys %{$h->{$path}}) {
1231 next if $SKIP_PROP{$prop};
1232 my $v = $h->{$path}->{$prop};
1233 my $t_ppath_prop = "$t: " .
1234 uri_encode($ppath) . ' ' .
1235 uri_encode($prop);
1236 if (defined $v) {
1237 push @out, " +$t_ppath_prop " .
1238 uri_encode($v);
1239 } else {
1240 push @out, " -$t_ppath_prop";
1241 }
1242 }
1243 }
1244 }
1245 foreach my $t (qw/absent_file absent_directory/) {
1246 $h = $ed->{$t} or next;
1247 foreach my $parent (sort keys %$h) {
1248 foreach my $path (sort @{$h->{$parent}}) {
1249 push @out, " $t: " .
1250 uri_encode("$parent/$path");
1251 warn "W: $t: $parent/$path ",
1252 "Insufficient permissions?\n";
1253 }
1254 }
1255 }
1256 \@out;
1257 }
1258
1259 sub parse_svn_date {
1260 my $date = shift || return '+0000 1970-01-01 00:00:00';
1261 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
1262 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
1263 croak "Unable to parse date: $date\n";
1264 "+0000 $Y-$m-$d $H:$M:$S";
1265 }
1266
1267 sub check_author {
1268 my ($author) = @_;
1269 if (!defined $author || length $author == 0) {
1270 $author = '(no author)';
1271 }
1272 if (defined $::_authors && ! defined $::users{$author}) {
1273 die "Author: $author not defined in $::_authors file\n";
1274 }
1275 $author;
1276 }
1277
1278 sub make_log_entry {
1279 my ($self, $rev, $parents, $ed) = @_;
1280 my $untracked = $self->get_untracked($ed);
1281
1282 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
1283 print $un "r$rev\n" or croak $!;
1284 print $un $_, "\n" foreach @$untracked;
1285 my %log_entry = ( parents => $parents || [], revision => $rev,
1286 log => '');
1287 my $rp = $self->ra->rev_proplist($rev);
1288 foreach (sort keys %$rp) {
1289 my $v = $rp->{$_};
1290 if (/^svn:(author|date|log)$/) {
1291 $log_entry{$1} = $v;
1292 } else {
1293 print $un " rev_prop: ", uri_encode($_), ' ',
1294 uri_encode($v), "\n";
1295 }
1296 }
1297 close $un or croak $!;
1298
1299 delete $rp->{'svn:date'}; # this is the only revprop for r0
1300 return undef if ($ed->{nr} == 0 && scalar @$untracked == 0 &&
1301 scalar keys %$rp == 0);
1302
1303 $log_entry{date} = parse_svn_date($log_entry{date});
1304 $log_entry{author} = check_author($log_entry{author});
1305 $log_entry{log} .= "\n";
1306 \%log_entry;
1307 }
1308
1309 sub fetch {
1310 my ($self, @parents) = @_;
1311 my ($last_rev, $last_commit) = $self->last_rev_commit;
1312 my ($base, $head) = $self->parse_revision($last_rev);
1313 return if ($base > $head);
1314 if (defined $last_commit) {
1315 $self->assert_index_clean($last_commit);
1316 }
1317 my $inc = 1000;
1318 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
1319 my $err_handler = $SVN::Error::handler;
1320 $SVN::Error::handler = \&skip_unknown_revs;
1321 while (1) {
1322 my @revs;
1323 $self->ra->get_log([$self->{path}], $min, $max, 0, 1, 1, sub {
1324 my ($paths, $rev, $author, $date, $log) = @_;
1325 push @revs, [ $paths, $rev ] });
1326 foreach (@revs) {
1327 if (my $log_entry = $self->do_fetch(@$_)) {
1328 $self->do_git_commit($log_entry, @parents);
1329 }
1330 }
1331 last if $max >= $head;
1332 $min = $max + 1;
1333 $max += $inc;
1334 $max = $head if ($max > $head);
1335 }
1336 $SVN::Error::handler = $err_handler;
1337 }
1338
1339 sub set_tree_cb {
1340 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
1341 # TODO: enable and test optimized commits:
1342 if (0 && $rev == ($self->{last_rev} + 1)) {
1343 $log_entry->{revision} = $rev;
1344 $log_entry->{author} = $author;
1345 $self->do_git_commit($log_entry, "$rev=$tree");
1346 } else {
1347 $self->fetch("$rev=$tree");
1348 }
1349 }
1350
1351 sub set_tree {
1352 my ($self, $tree) = (shift, shift);
1353 my $log_entry = ::get_commit_entry($tree);
1354 unless ($self->{last_rev}) {
1355 fatal("Must have an existing revision to commit\n");
1356 }
1357 my $pool = SVN::Pool->new;
1358 my $ed = SVN::Git::Editor->new({ r => $self->{last_rev},
1359 ra => $self->ra->dup,
1360 svn_path => $self->ra->{svn_path}
1361 },
1362 $self->ra->get_commit_editor(
1363 $log_entry->{log}, sub {
1364 $self->set_tree_cb($log_entry,
1365 $tree, @_);
1366 }),
1367 $pool);
1368 my $mods = $ed->apply_diff($self->{last_commit}, $tree);
1369 if (@$mods == 0) {
1370 print "No changes\nr$self->{last_rev} = $tree\n";
1371 }
1372 $pool->clear;
1373 }
1374
1375 sub skip_unknown_revs {
1376 my ($err) = @_;
1377 my $errno = $err->apr_err();
1378 # Maybe the branch we're tracking didn't
1379 # exist when the repo started, so it's
1380 # not an error if it doesn't, just continue
1381 #
1382 # Wonderfully consistent library, eh?
1383 # 160013 - svn:// and file://
1384 # 175002 - http(s)://
1385 # 175007 - http(s):// (this repo required authorization, too...)
1386 # More codes may be discovered later...
1387 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
1388 return;
1389 }
1390 croak "Error from SVN, ($errno): ", $err->expanded_message,"\n";
1391 }
1392
1393 # rev_db:
1394 # Tie::File seems to be prone to offset errors if revisions get sparse,
1395 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
1396 # one of my favorite modules is out :< Next up would be one of the DBM
1397 # modules, but I'm not sure which is most portable... So I'll just
1398 # go with something that's plain-text, but still capable of
1399 # being randomly accessed. So here's my ultra-simple fixed-width
1400 # database. All records are 40 characters + "\n", so it's easy to seek
1401 # to a revision: (41 * rev) is the byte offset.
1402 # A record of 40 0s denotes an empty revision.
1403 # And yes, it's still pretty fast (faster than Tie::File).
1404
1405 sub rev_db_set {
1406 my ($self, $rev, $commit) = @_;
1407 length $commit == 40 or croak "arg3 must be a full SHA1 hexsum\n";
1408 open my $fh, '+<', $self->{db_path} or croak $!;
1409 my $offset = $rev * 41;
1410 # assume that append is the common case:
1411 seek $fh, 0, 2 or croak $!;
1412 my $pos = tell $fh;
1413 if ($pos < $offset) {
1414 print $fh (('0' x 40),"\n") x (($offset - $pos) / 41)
1415 or croak $!;
1416 }
1417 seek $fh, $offset, 0 or croak $!;
1418 print $fh $commit,"\n" or croak $!;
1419 close $fh or croak $!;
1420 }
1421
1422 sub rev_db_get {
1423 my ($self, $rev) = @_;
1424 my $ret;
1425 my $offset = $rev * 41;
1426 open my $fh, '<', $self->{db_path} or croak $!;
1427 if (seek $fh, $offset, 0) {
1428 $ret = readline $fh;
1429 if (defined $ret) {
1430 chomp $ret;
1431 $ret = undef if ($ret =~ /^0{40}$/);
1432 }
1433 }
1434 close $fh or croak $!;
1435 $ret;
1436 }
1437
1438 sub find_rev_before {
1439 my ($self, $rev, $eq_ok) = @_;
1440 --$rev unless $eq_ok;
1441 while ($rev > 0) {
1442 if (my $c = $self->rev_db_get($rev)) {
1443 return ($rev, $c);
1444 }
1445 --$rev;
1446 }
1447 return (undef, undef);
1448 }
1449
1450 sub _new {
1451 my ($class, $repo_id, $ref_id, $path) = @_;
1452 unless (defined $repo_id && length $repo_id) {
1453 $repo_id = $Git::SVN::default_repo_id;
1454 }
1455 unless (defined $ref_id && length $ref_id) {
1456 $_[2] = $ref_id = $Git::SVN::default_ref_id;
1457 }
1458 $_[1] = $repo_id = sanitize_remote_name($repo_id);
1459 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
1460 $_[3] = $path = '' unless (defined $path);
1461 mkpath([$dir]);
1462 unless (-f "$dir/.rev_db") {
1463 open my $fh, '>>', "$dir/.rev_db" or croak $!;
1464 close $fh or croak $!;
1465 }
1466 bless { ref_id => $ref_id, dir => $dir, index => "$dir/index",
1467 path => $path,
1468 db_path => "$dir/.rev_db", repo_id => $repo_id }, $class;
1469 }
1470
1471 sub uri_encode {
1472 my ($f) = @_;
1473 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1474 $f
1475 }
1476
1477 package Git::SVN::Prompt;
1478 use strict;
1479 use warnings;
1480 require SVN::Core;
1481 use vars qw/$_no_auth_cache $_username/;
1482
1483 sub simple {
1484 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
1485 $may_save = undef if $_no_auth_cache;
1486 $default_username = $_username if defined $_username;
1487 if (defined $default_username && length $default_username) {
1488 if (defined $realm && length $realm) {
1489 print STDERR "Authentication realm: $realm\n";
1490 STDERR->flush;
1491 }
1492 $cred->username($default_username);
1493 } else {
1494 username($cred, $realm, $may_save, $pool);
1495 }
1496 $cred->password(_read_password("Password for '" .
1497 $cred->username . "': ", $realm));
1498 $cred->may_save($may_save);
1499 $SVN::_Core::SVN_NO_ERROR;
1500 }
1501
1502 sub ssl_server_trust {
1503 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
1504 $may_save = undef if $_no_auth_cache;
1505 print STDERR "Error validating server certificate for '$realm':\n";
1506 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
1507 print STDERR " - The certificate is not issued by a trusted ",
1508 "authority. Use the\n",
1509 " fingerprint to validate the certificate manually!\n";
1510 }
1511 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
1512 print STDERR " - The certificate hostname does not match.\n";
1513 }
1514 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
1515 print STDERR " - The certificate is not yet valid.\n";
1516 }
1517 if ($failures & $SVN::Auth::SSL::EXPIRED) {
1518 print STDERR " - The certificate has expired.\n";
1519 }
1520 if ($failures & $SVN::Auth::SSL::OTHER) {
1521 print STDERR " - The certificate has an unknown error.\n";
1522 }
1523 printf STDERR
1524 "Certificate information:\n".
1525 " - Hostname: %s\n".
1526 " - Valid: from %s until %s\n".
1527 " - Issuer: %s\n".
1528 " - Fingerprint: %s\n",
1529 map $cert_info->$_, qw(hostname valid_from valid_until
1530 issuer_dname fingerprint);
1531 my $choice;
1532 prompt:
1533 print STDERR $may_save ?
1534 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
1535 "(R)eject or accept (t)emporarily? ";
1536 STDERR->flush;
1537 $choice = lc(substr(<STDIN> || 'R', 0, 1));
1538 if ($choice =~ /^t$/i) {
1539 $cred->may_save(undef);
1540 } elsif ($choice =~ /^r$/i) {
1541 return -1;
1542 } elsif ($may_save && $choice =~ /^p$/i) {
1543 $cred->may_save($may_save);
1544 } else {
1545 goto prompt;
1546 }
1547 $cred->accepted_failures($failures);
1548 $SVN::_Core::SVN_NO_ERROR;
1549 }
1550
1551 sub ssl_client_cert {
1552 my ($cred, $realm, $may_save, $pool) = @_;
1553 $may_save = undef if $_no_auth_cache;
1554 print STDERR "Client certificate filename: ";
1555 STDERR->flush;
1556 chomp(my $filename = <STDIN>);
1557 $cred->cert_file($filename);
1558 $cred->may_save($may_save);
1559 $SVN::_Core::SVN_NO_ERROR;
1560 }
1561
1562 sub ssl_client_cert_pw {
1563 my ($cred, $realm, $may_save, $pool) = @_;
1564 $may_save = undef if $_no_auth_cache;
1565 $cred->password(_read_password("Password: ", $realm));
1566 $cred->may_save($may_save);
1567 $SVN::_Core::SVN_NO_ERROR;
1568 }
1569
1570 sub username {
1571 my ($cred, $realm, $may_save, $pool) = @_;
1572 $may_save = undef if $_no_auth_cache;
1573 if (defined $realm && length $realm) {
1574 print STDERR "Authentication realm: $realm\n";
1575 }
1576 my $username;
1577 if (defined $_username) {
1578 $username = $_username;
1579 } else {
1580 print STDERR "Username: ";
1581 STDERR->flush;
1582 chomp($username = <STDIN>);
1583 }
1584 $cred->username($username);
1585 $cred->may_save($may_save);
1586 $SVN::_Core::SVN_NO_ERROR;
1587 }
1588
1589 sub _read_password {
1590 my ($prompt, $realm) = @_;
1591 print STDERR $prompt;
1592 STDERR->flush;
1593 require Term::ReadKey;
1594 Term::ReadKey::ReadMode('noecho');
1595 my $password = '';
1596 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
1597 last if $key =~ /[\012\015]/; # \n\r
1598 $password .= $key;
1599 }
1600 Term::ReadKey::ReadMode('restore');
1601 print STDERR "\n";
1602 STDERR->flush;
1603 $password;
1604 }
1605
1606 package main;
1607
1608 sub uri_encode {
1609 my ($f) = @_;
1610 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
1611 $f
1612 }
1613
1614 sub uri_decode {
1615 my ($f) = @_;
1616 $f =~ tr/+/ /;
1617 $f =~ s/%([A-F0-9]{2})/chr hex($1)/ge;
1618 $f
1619 }
1620
1621 {
1622 my $kill_stupid_warnings = $SVN::Node::none.$SVN::Node::file.
1623 $SVN::Node::dir.$SVN::Node::unknown.
1624 $SVN::Node::none.$SVN::Node::file.
1625 $SVN::Node::dir.$SVN::Node::unknown.
1626 $SVN::Auth::SSL::CNMISMATCH.
1627 $SVN::Auth::SSL::NOTYETVALID.
1628 $SVN::Auth::SSL::EXPIRED.
1629 $SVN::Auth::SSL::UNKNOWNCA.
1630 $SVN::Auth::SSL::OTHER;
1631 }
1632
1633 package SVN::Git::Fetcher;
1634 use vars qw/@ISA/;
1635 use strict;
1636 use warnings;
1637 use Carp qw/croak/;
1638 use IO::File qw//;
1639
1640 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
1641 sub new {
1642 my ($class, $git_svn) = @_;
1643 my $self = SVN::Delta::Editor->new;
1644 bless $self, $class;
1645 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
1646 $self->{empty} = {};
1647 $self->{dir_prop} = {};
1648 $self->{file_prop} = {};
1649 $self->{absent_dir} = {};
1650 $self->{absent_file} = {};
1651 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
1652 require Digest::MD5;
1653 $self;
1654 }
1655
1656 sub set_path_strip {
1657 my ($self, $path) = @_;
1658 $self->{path_strip} = qr/^\Q$path\E\/?/;
1659 }
1660
1661 sub open_root {
1662 { path => '' };
1663 }
1664
1665 sub open_directory {
1666 my ($self, $path, $pb, $rev) = @_;
1667 { path => $path };
1668 }
1669
1670 sub git_path {
1671 my ($self, $path) = @_;
1672 $path =~ s!$self->{path_strip}!! if $self->{path_strip};
1673 $path;
1674 }
1675
1676 sub delete_entry {
1677 my ($self, $path, $rev, $pb) = @_;
1678
1679 my $gpath = $self->git_path($path);
1680 # remove entire directories.
1681 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
1682 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
1683 -r --name-only -z/,
1684 $self->{c}, '--', $gpath);
1685 local $/ = "\0";
1686 while (<$ls>) {
1687 chomp;
1688 $self->{gii}->remove($_);
1689 print "\tD\t$_\n" unless $self->{q};
1690 }
1691 print "\tD\t$gpath/\n" unless $self->{q};
1692 command_close_pipe($ls, $ctx);
1693 $self->{empty}->{$path} = 0
1694 } else {
1695 $self->{gii}->remove($gpath);
1696 print "\tD\t$gpath\n" unless $self->{q};
1697 }
1698 undef;
1699 }
1700
1701 sub open_file {
1702 my ($self, $path, $pb, $rev) = @_;
1703 my $gpath = $self->git_path($path);
1704 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
1705 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
1706 unless (defined $mode && defined $blob) {
1707 die "$path was not found in commit $self->{c} (r$rev)\n";
1708 }
1709 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
1710 pool => SVN::Pool->new, action => 'M' };
1711 }
1712
1713 sub add_file {
1714 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
1715 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1716 delete $self->{empty}->{$dir};
1717 { path => $path, mode_a => 100644, mode_b => 100644,
1718 pool => SVN::Pool->new, action => 'A' };
1719 }
1720
1721 sub add_directory {
1722 my ($self, $path, $cp_path, $cp_rev) = @_;
1723 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
1724 delete $self->{empty}->{$dir};
1725 $self->{empty}->{$path} = 1;
1726 { path => $path };
1727 }
1728
1729 sub change_dir_prop {
1730 my ($self, $db, $prop, $value) = @_;
1731 $self->{dir_prop}->{$db->{path}} ||= {};
1732 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
1733 undef;
1734 }
1735
1736 sub absent_directory {
1737 my ($self, $path, $pb) = @_;
1738 $self->{absent_dir}->{$pb->{path}} ||= [];
1739 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
1740 undef;
1741 }
1742
1743 sub absent_file {
1744 my ($self, $path, $pb) = @_;
1745 $self->{absent_file}->{$pb->{path}} ||= [];
1746 push @{$self->{absent_file}->{$pb->{path}}}, $path;
1747 undef;
1748 }
1749
1750 sub change_file_prop {
1751 my ($self, $fb, $prop, $value) = @_;
1752 if ($prop eq 'svn:executable') {
1753 if ($fb->{mode_b} != 120000) {
1754 $fb->{mode_b} = defined $value ? 100755 : 100644;
1755 }
1756 } elsif ($prop eq 'svn:special') {
1757 $fb->{mode_b} = defined $value ? 120000 : 100644;
1758 } else {
1759 $self->{file_prop}->{$fb->{path}} ||= {};
1760 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
1761 }
1762 undef;
1763 }
1764
1765 sub apply_textdelta {
1766 my ($self, $fb, $exp) = @_;
1767 my $fh = IO::File->new_tmpfile;
1768 $fh->autoflush(1);
1769 # $fh gets auto-closed() by SVN::TxDelta::apply(),
1770 # (but $base does not,) so dup() it for reading in close_file
1771 open my $dup, '<&', $fh or croak $!;
1772 my $base = IO::File->new_tmpfile;
1773 $base->autoflush(1);
1774 if ($fb->{blob}) {
1775 defined (my $pid = fork) or croak $!;
1776 if (!$pid) {
1777 open STDOUT, '>&', $base or croak $!;
1778 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
1779 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
1780 }
1781 waitpid $pid, 0;
1782 croak $? if $?;
1783
1784 if (defined $exp) {
1785 seek $base, 0, 0 or croak $!;
1786 my $md5 = Digest::MD5->new;
1787 $md5->addfile($base);
1788 my $got = $md5->hexdigest;
1789 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
1790 "expected: $exp\n",
1791 " got: $got\n" if ($got ne $exp);
1792 }
1793 }
1794 seek $base, 0, 0 or croak $!;
1795 $fb->{fh} = $dup;
1796 $fb->{base} = $base;
1797 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
1798 }
1799
1800 sub close_file {
1801 my ($self, $fb, $exp) = @_;
1802 my $hash;
1803 my $path = $self->git_path($fb->{path});
1804 if (my $fh = $fb->{fh}) {
1805 seek($fh, 0, 0) or croak $!;
1806 my $md5 = Digest::MD5->new;
1807 $md5->addfile($fh);
1808 my $got = $md5->hexdigest;
1809 die "Checksum mismatch: $path\n",
1810 "expected: $exp\n got: $got\n" if ($got ne $exp);
1811 seek($fh, 0, 0) or croak $!;
1812 if ($fb->{mode_b} == 120000) {
1813 read($fh, my $buf, 5) == 5 or croak $!;
1814 $buf eq 'link ' or die "$path has mode 120000",
1815 "but is not a link\n";
1816 }
1817 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
1818 if (!$pid) {
1819 open STDIN, '<&', $fh or croak $!;
1820 exec qw/git-hash-object -w --stdin/ or croak $!;
1821 }
1822 chomp($hash = do { local $/; <$out> });
1823 close $out or croak $!;
1824 close $fh or croak $!;
1825 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
1826 close $fb->{base} or croak $!;
1827 } else {
1828 $hash = $fb->{blob} or die "no blob information\n";
1829 }
1830 $fb->{pool}->clear;
1831 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
1832 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $self->{q};
1833 undef;
1834 }
1835
1836 sub abort_edit {
1837 my $self = shift;
1838 $self->{nr} = $self->{gii}->{nr};
1839 delete $self->{gii};
1840 $self->SUPER::abort_edit(@_);
1841 }
1842
1843 sub close_edit {
1844 my $self = shift;
1845 $self->{git_commit_ok} = 1;
1846 $self->{nr} = $self->{gii}->{nr};
1847 delete $self->{gii};
1848 $self->SUPER::close_edit(@_);
1849 }
1850
1851 package SVN::Git::Editor;
1852 use vars qw/@ISA/;
1853 use strict;
1854 use warnings;
1855 use Carp qw/croak/;
1856 use IO::File;
1857
1858 sub new {
1859 my $class = shift;
1860 my $git_svn = shift;
1861 my $self = SVN::Delta::Editor->new(@_);
1862 bless $self, $class;
1863 foreach (qw/svn_path r ra/) {
1864 die "$_ required!\n" unless (defined $git_svn->{$_});
1865 $self->{$_} = $git_svn->{$_};
1866 }
1867 $self->{pool} = SVN::Pool->new;
1868 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
1869 $self->{rm} = { };
1870 require Digest::MD5;
1871 return $self;
1872 }
1873
1874 sub split_path {
1875 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
1876 }
1877
1878 sub repo_path {
1879 (defined $_[1] && length $_[1]) ? $_[1] : ''
1880 }
1881
1882 sub url_path {
1883 my ($self, $path) = @_;
1884 $self->{ra}->{url} . '/' . $self->repo_path($path);
1885 }
1886
1887 sub rmdirs {
1888 my ($self, $tree_b) = @_;
1889 my $rm = $self->{rm};
1890 delete $rm->{''}; # we never delete the url we're tracking
1891 return unless %$rm;
1892
1893 foreach (keys %$rm) {
1894 my @d = split m#/#, $_;
1895 my $c = shift @d;
1896 $rm->{$c} = 1;
1897 while (@d) {
1898 $c .= '/' . shift @d;
1899 $rm->{$c} = 1;
1900 }
1901 }
1902 delete $rm->{$self->{svn_path}};
1903 delete $rm->{''}; # we never delete the url we're tracking
1904 return unless %$rm;
1905
1906 my ($fh, $ctx) = command_output_pipe(
1907 qw/ls-tree --name-only -r -z/, $tree_b);
1908 local $/ = "\0";
1909 while (<$fh>) {
1910 chomp;
1911 my @dn = split m#/#, $_;
1912 while (pop @dn) {
1913 delete $rm->{join '/', @dn};
1914 }
1915 unless (%$rm) {
1916 close $fh;
1917 return;
1918 }
1919 }
1920 command_close_pipe($fh, $ctx);
1921
1922 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
1923 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
1924 $self->close_directory($bat->{$d}, $p);
1925 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
1926 print "\tD+\t$d/\n" unless $::_q;
1927 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
1928 delete $bat->{$d};
1929 }
1930 }
1931
1932 sub open_or_add_dir {
1933 my ($self, $full_path, $baton) = @_;
1934 my $t = $self->{ra}->check_path($full_path, $self->{r});
1935 if ($t == $SVN::Node::none) {
1936 return $self->add_directory($full_path, $baton,
1937 undef, -1, $self->{pool});
1938 } elsif ($t == $SVN::Node::dir) {
1939 return $self->open_directory($full_path, $baton,
1940 $self->{r}, $self->{pool});
1941 }
1942 print STDERR "$full_path already exists in repository at ",
1943 "r$self->{r} and it is not a directory (",
1944 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
1945 exit 1;
1946 }
1947
1948 sub ensure_path {
1949 my ($self, $path) = @_;
1950 my $bat = $self->{bat};
1951 $path = $self->repo_path($path);
1952 return $bat->{''} unless (length $path);
1953 my @p = split m#/+#, $path;
1954 my $c = shift @p;
1955 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
1956 while (@p) {
1957 my $c0 = $c;
1958 $c .= '/' . shift @p;
1959 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
1960 }
1961 return $bat->{$c};
1962 }
1963
1964 sub A {
1965 my ($self, $m) = @_;
1966 my ($dir, $file) = split_path($m->{file_b});
1967 my $pbat = $self->ensure_path($dir);
1968 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
1969 undef, -1);
1970 print "\tA\t$m->{file_b}\n" unless $::_q;
1971 $self->chg_file($fbat, $m);
1972 $self->close_file($fbat,undef,$self->{pool});
1973 }
1974
1975 sub C {
1976 my ($self, $m) = @_;
1977 my ($dir, $file) = split_path($m->{file_b});
1978 my $pbat = $self->ensure_path($dir);
1979 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
1980 $self->url_path($m->{file_a}), $self->{r});
1981 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
1982 $self->chg_file($fbat, $m);
1983 $self->close_file($fbat,undef,$self->{pool});
1984 }
1985
1986 sub delete_entry {
1987 my ($self, $path, $pbat) = @_;
1988 my $rpath = $self->repo_path($path);
1989 my ($dir, $file) = split_path($rpath);
1990 $self->{rm}->{$dir} = 1;
1991 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
1992 }
1993
1994 sub R {
1995 my ($self, $m) = @_;
1996 my ($dir, $file) = split_path($m->{file_b});
1997 my $pbat = $self->ensure_path($dir);
1998 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
1999 $self->url_path($m->{file_a}), $self->{r});
2000 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
2001 $self->chg_file($fbat, $m);
2002 $self->close_file($fbat,undef,$self->{pool});
2003
2004 ($dir, $file) = split_path($m->{file_a});
2005 $pbat = $self->ensure_path($dir);
2006 $self->delete_entry($m->{file_a}, $pbat);
2007 }
2008
2009 sub M {
2010 my ($self, $m) = @_;
2011 my ($dir, $file) = split_path($m->{file_b});
2012 my $pbat = $self->ensure_path($dir);
2013 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
2014 $pbat,$self->{r},$self->{pool});
2015 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
2016 $self->chg_file($fbat, $m);
2017 $self->close_file($fbat,undef,$self->{pool});
2018 }
2019
2020 sub T { shift->M(@_) }
2021
2022 sub change_file_prop {
2023 my ($self, $fbat, $pname, $pval) = @_;
2024 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
2025 }
2026
2027 sub chg_file {
2028 my ($self, $fbat, $m) = @_;
2029 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
2030 $self->change_file_prop($fbat,'svn:executable','*');
2031 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
2032 $self->change_file_prop($fbat,'svn:executable',undef);
2033 }
2034 my $fh = IO::File->new_tmpfile or croak $!;
2035 if ($m->{mode_b} =~ /^120/) {
2036 print $fh 'link ' or croak $!;
2037 $self->change_file_prop($fbat,'svn:special','*');
2038 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
2039 $self->change_file_prop($fbat,'svn:special',undef);
2040 }
2041 defined(my $pid = fork) or croak $!;
2042 if (!$pid) {
2043 open STDOUT, '>&', $fh or croak $!;
2044 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
2045 }
2046 waitpid $pid, 0;
2047 croak $? if $?;
2048 $fh->flush == 0 or croak $!;
2049 seek $fh, 0, 0 or croak $!;
2050
2051 my $md5 = Digest::MD5->new;
2052 $md5->addfile($fh) or croak $!;
2053 seek $fh, 0, 0 or croak $!;
2054
2055 my $exp = $md5->hexdigest;
2056 my $pool = SVN::Pool->new;
2057 my $atd = $self->apply_textdelta($fbat, undef, $pool);
2058 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
2059 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
2060 $pool->clear;
2061
2062 close $fh or croak $!;
2063 }
2064
2065 sub D {
2066 my ($self, $m) = @_;
2067 my ($dir, $file) = split_path($m->{file_b});
2068 my $pbat = $self->ensure_path($dir);
2069 print "\tD\t$m->{file_b}\n" unless $::_q;
2070 $self->delete_entry($m->{file_b}, $pbat);
2071 }
2072
2073 sub close_edit {
2074 my ($self) = @_;
2075 my ($p,$bat) = ($self->{pool}, $self->{bat});
2076 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
2077 $self->close_directory($bat->{$_}, $p);
2078 }
2079 $self->SUPER::close_edit($p);
2080 $p->clear;
2081 }
2082
2083 sub abort_edit {
2084 my ($self) = @_;
2085 $self->SUPER::abort_edit($self->{pool});
2086 $self->{pool}->clear;
2087 }
2088
2089 # this drives the editor
2090 sub apply_diff {
2091 my ($self, $tree_a, $tree_b) = @_;
2092 my @diff_tree = qw(diff-tree -z -r);
2093 if ($::_cp_similarity) {
2094 push @diff_tree, "-C$::_cp_similarity";
2095 } else {
2096 push @diff_tree, '-C';
2097 }
2098 push @diff_tree, '--find-copies-harder' if $::_find_copies_harder;
2099 push @diff_tree, "-l$::_l" if defined $::_l;
2100 push @diff_tree, $tree_a, $tree_b;
2101 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
2102 my $nl = $/;
2103 local $/ = "\0";
2104 my $state = 'meta';
2105 my @mods;
2106 while (<$diff_fh>) {
2107 chomp $_; # this gets rid of the trailing "\0"
2108 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
2109 $::sha1\s($::sha1)\s
2110 ([MTCRAD])\d*$/xo) {
2111 push @mods, { mode_a => $1, mode_b => $2,
2112 sha1_b => $3, chg => $4 };
2113 if ($4 =~ /^(?:C|R)$/) {
2114 $state = 'file_a';
2115 } else {
2116 $state = 'file_b';
2117 }
2118 } elsif ($state eq 'file_a') {
2119 my $x = $mods[$#mods] or croak "Empty array\n";
2120 if ($x->{chg} !~ /^(?:C|R)$/) {
2121 croak "Error parsing $_, $x->{chg}\n";
2122 }
2123 $x->{file_a} = $_;
2124 $state = 'file_b';
2125 } elsif ($state eq 'file_b') {
2126 my $x = $mods[$#mods] or croak "Empty array\n";
2127 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
2128 croak "Error parsing $_, $x->{chg}\n";
2129 }
2130 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
2131 croak "Error parsing $_, $x->{chg}\n";
2132 }
2133 $x->{file_b} = $_;
2134 $state = 'meta';
2135 } else {
2136 croak "Error parsing $_\n";
2137 }
2138 }
2139 command_close_pipe($diff_fh, $ctx);
2140 $/ = $nl;
2141
2142 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
2143 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @mods) {
2144 my $f = $m->{chg};
2145 if (defined $o{$f}) {
2146 $self->$f($m);
2147 } else {
2148 fatal("Invalid change type: $f\n");
2149 }
2150 }
2151 $self->rmdirs($tree_b) if $::_rmdir;
2152 if (@mods == 0) {
2153 $self->abort_edit;
2154 } else {
2155 $self->close_edit;
2156 }
2157 \@mods;
2158 }
2159
2160 package Git::SVN::Ra;
2161 use vars qw/@ISA $config_dir/;
2162 use strict;
2163 use warnings;
2164 my ($can_do_switch);
2165 my %RA;
2166
2167 BEGIN {
2168 # enforce temporary pool usage for some simple functions
2169 my $e;
2170 foreach (qw/get_latest_revnum rev_proplist get_file
2171 check_path get_dir get_uuid get_repos_root/) {
2172 $e .= "sub $_ {
2173 my \$self = shift;
2174 my \$pool = SVN::Pool->new;
2175 my \@ret = \$self->SUPER::$_(\@_,\$pool);
2176 \$pool->clear;
2177 wantarray ? \@ret : \$ret[0]; }\n";
2178 }
2179 eval $e;
2180 }
2181
2182 sub new {
2183 my ($class, $url) = @_;
2184 $url =~ s!/+$!!;
2185 return $RA{$url} if $RA{$url};
2186
2187 SVN::_Core::svn_config_ensure($config_dir, undef);
2188 my ($baton, $callbacks) = SVN::Core::auth_open_helper([
2189 SVN::Client::get_simple_provider(),
2190 SVN::Client::get_ssl_server_trust_file_provider(),
2191 SVN::Client::get_simple_prompt_provider(
2192 \&Git::SVN::Prompt::simple, 2),
2193 SVN::Client::get_ssl_client_cert_prompt_provider(
2194 \&Git::SVN::Prompt::ssl_client_cert, 2),
2195 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
2196 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
2197 SVN::Client::get_username_provider(),
2198 SVN::Client::get_ssl_server_trust_prompt_provider(
2199 \&Git::SVN::Prompt::ssl_server_trust),
2200 SVN::Client::get_username_prompt_provider(
2201 \&Git::SVN::Prompt::username, 2),
2202 ]);
2203 my $config = SVN::Core::config_get_config($config_dir);
2204 my $self = SVN::Ra->new(url => $url, auth => $baton,
2205 config => $config,
2206 pool => SVN::Pool->new,
2207 auth_provider_callbacks => $callbacks);
2208 $self->{svn_path} = $url;
2209 $self->{repos_root} = $self->get_repos_root;
2210 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E/*##;
2211 $RA{$url} = bless $self, $class;
2212 }
2213
2214 sub DESTROY {
2215 # do not call the real DESTROY since we store ourselves in %RA
2216 }
2217
2218 sub dup {
2219 my ($self) = @_;
2220 my $dup = SVN::Ra->new(pool => SVN::Pool->new,
2221 map { $_ => $self->{$_} } qw/config url
2222 auth auth_provider_callbacks repos_root svn_path/);
2223 bless $dup, ref $self;
2224 }
2225
2226 sub get_log {
2227 my ($self, @args) = @_;
2228 my $pool = SVN::Pool->new;
2229 $args[4]-- if $args[4] && ! $::_follow_parent;
2230 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
2231 my $ret = $self->SUPER::get_log(@args, $pool);
2232 $pool->clear;
2233 $ret;
2234 }
2235
2236 sub get_commit_editor {
2237 my ($self, $log, $cb, $pool) = @_;
2238 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
2239 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
2240 }
2241
2242 sub uuid {
2243 my ($self) = @_;
2244 $self->{uuid} ||= $self->get_uuid;
2245 }
2246
2247 sub gs_do_update {
2248 my ($self, $rev_a, $rev_b, $path, $recurse, $editor) = @_;
2249 my $pool = SVN::Pool->new;
2250 $editor->set_path_strip($path);
2251 my $reporter = $self->do_update($rev_b, $path, $recurse,
2252 $editor, $pool);
2253 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2254 my $new = ($rev_a == $rev_b);
2255 $reporter->set_path('', $rev_a, $new, @lock, $pool);
2256 $reporter->finish_report($pool);
2257 $pool->clear;
2258 $editor->{git_commit_ok};
2259 }
2260
2261 sub gs_do_switch {
2262 my ($self, $rev_a, $rev_b, $path, $recurse, $url_b, $editor) = @_;
2263 my $pool = SVN::Pool->new;
2264 $editor->set_path_strip($path);
2265 my $reporter = $self->do_switch($rev_b, $path, $recurse,
2266 $url_b, $editor, $pool);
2267 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
2268 $reporter->set_path('', $rev_a, 0, @lock, $pool);
2269 $reporter->finish_report($pool);
2270 $pool->clear;
2271 $editor->{git_commit_ok};
2272 }
2273
2274 sub minimize_url {
2275 my ($self) = @_;
2276 return $self->{url} if ($self->{url} eq $self->{repos_root});
2277 my $url = $self->{repos_root};
2278 my @components = split(m!/!, $self->{svn_path});
2279 my $c = '';
2280 do {
2281 $url .= "/$c" if length $c;
2282 eval { (ref $self)->new($url)->get_latest_revnum };
2283 } while ($@ && ($c = shift @components));
2284 $url;
2285 }
2286
2287 sub can_do_switch {
2288 my $self = shift;
2289 unless (defined $can_do_switch) {
2290 my $pool = SVN::Pool->new;
2291 my $rep = eval {
2292 $self->do_switch(1, '', 0, $self->{url},
2293 SVN::Delta::Editor->new, $pool);
2294 };
2295 if ($@) {
2296 $can_do_switch = 0;
2297 } else {
2298 $rep->abort_report($pool);
2299 $can_do_switch = 1;
2300 }
2301 $pool->clear;
2302 }
2303 $can_do_switch;
2304 }
2305
2306 package Git::SVN::Log;
2307 use strict;
2308 use warnings;
2309 use POSIX qw/strftime/;
2310 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
2311 %rusers $show_commit $incremental/;
2312 my $l_fmt;
2313
2314 sub cmt_showable {
2315 my ($c) = @_;
2316 return 1 if defined $c->{r};
2317 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
2318 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
2319 my @log = command(qw/cat-file commit/, $c->{c});
2320 shift @log while ($log[0] ne "\n");
2321 shift @log;
2322 @{$c->{l}} = grep !/^git-svn-id: /, @log;
2323
2324 (undef, $c->{r}, undef) = ::extract_metadata(
2325 (grep(/^git-svn-id: /, @log))[-1]);
2326 }
2327 return defined $c->{r};
2328 }
2329
2330 sub log_use_color {
2331 return 1 if $color;
2332 my ($dc, $dcvar);
2333 $dcvar = 'color.diff';
2334 $dc = `git-config --get $dcvar`;
2335 if ($dc eq '') {
2336 # nothing at all; fallback to "diff.color"
2337 $dcvar = 'diff.color';
2338 $dc = `git-config --get $dcvar`;
2339 }
2340 chomp($dc);
2341 if ($dc eq 'auto') {
2342 my $pc;
2343 $pc = `git-config --get color.pager`;
2344 if ($pc eq '') {
2345 # does not have it -- fallback to pager.color
2346 $pc = `git-config --bool --get pager.color`;
2347 }
2348 else {
2349 $pc = `git-config --bool --get color.pager`;
2350 if ($?) {
2351 $pc = 'false';
2352 }
2353 }
2354 chomp($pc);
2355 if (-t *STDOUT || (defined $pager && $pc eq 'true')) {
2356 return ($ENV{TERM} && $ENV{TERM} ne 'dumb');
2357 }
2358 return 0;
2359 }
2360 return 0 if $dc eq 'never';
2361 return 1 if $dc eq 'always';
2362 chomp($dc = `git-config --bool --get $dcvar`);
2363 return ($dc eq 'true');
2364 }
2365
2366 sub git_svn_log_cmd {
2367 my ($r_min, $r_max) = @_;
2368 my $gs = Git::SVN->_new;
2369 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
2370 $gs->refname);
2371 push @cmd, '-r' unless $non_recursive;
2372 push @cmd, qw/--raw --name-status/ if $verbose;
2373 push @cmd, '--color' if log_use_color();
2374 return @cmd unless defined $r_max;
2375 if ($r_max == $r_min) {
2376 push @cmd, '--max-count=1';
2377 if (my $c = $gs->rev_db_get($r_max)) {
2378 push @cmd, $c;
2379 }
2380 } else {
2381 my ($c_min, $c_max);
2382 $c_max = $gs->rev_db_get($r_max);
2383 $c_min = $gs->rev_db_get($r_min);
2384 if (defined $c_min && defined $c_max) {
2385 if ($r_max > $r_max) {
2386 push @cmd, "$c_min..$c_max";
2387 } else {
2388 push @cmd, "$c_max..$c_min";
2389 }
2390 } elsif ($r_max > $r_min) {
2391 push @cmd, $c_max;
2392 } else {
2393 push @cmd, $c_min;
2394 }
2395 }
2396 return @cmd;
2397 }
2398
2399 # adapted from pager.c
2400 sub config_pager {
2401 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
2402 if (!defined $pager) {
2403 $pager = 'less';
2404 } elsif (length $pager == 0 || $pager eq 'cat') {
2405 $pager = undef;
2406 }
2407 }
2408
2409 sub run_pager {
2410 return unless -t *STDOUT;
2411 pipe my $rfd, my $wfd or return;
2412 defined(my $pid = fork) or ::fatal "Can't fork: $!\n";
2413 if (!$pid) {
2414 open STDOUT, '>&', $wfd or
2415 ::fatal "Can't redirect to stdout: $!\n";
2416 return;
2417 }
2418 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!\n";
2419 $ENV{LESS} ||= 'FRSX';
2420 exec $pager or ::fatal "Can't run pager: $! ($pager)\n";
2421 }
2422
2423 sub get_author_info {
2424 my ($dest, $author, $t, $tz) = @_;
2425 $author =~ s/(?:^\s*|\s*$)//g;
2426 $dest->{a_raw} = $author;
2427 my $au;
2428 if ($::_authors) {
2429 $au = $rusers{$author} || undef;
2430 }
2431 if (!$au) {
2432 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
2433 }
2434 $dest->{t} = $t;
2435 $dest->{tz} = $tz;
2436 $dest->{a} = $au;
2437 # Date::Parse isn't in the standard Perl distro :(
2438 if ($tz =~ s/^\+//) {
2439 $t += ::tz_to_s_offset($tz);
2440 } elsif ($tz =~ s/^\-//) {
2441 $t -= ::tz_to_s_offset($tz);
2442 }
2443 $dest->{t_utc} = $t;
2444 }
2445
2446 sub process_commit {
2447 my ($c, $r_min, $r_max, $defer) = @_;
2448 if (defined $r_min && defined $r_max) {
2449 if ($r_min == $c->{r} && $r_min == $r_max) {
2450 show_commit($c);
2451 return 0;
2452 }
2453 return 1 if $r_min == $r_max;
2454 if ($r_min < $r_max) {
2455 # we need to reverse the print order
2456 return 0 if (defined $limit && --$limit < 0);
2457 push @$defer, $c;
2458 return 1;
2459 }
2460 if ($r_min != $r_max) {
2461 return 1 if ($r_min < $c->{r});
2462 return 1 if ($r_max > $c->{r});
2463 }
2464 }
2465 return 0 if (defined $limit && --$limit < 0);
2466 show_commit($c);
2467 return 1;
2468 }
2469
2470 sub show_commit {
2471 my $c = shift;
2472 if ($oneline) {
2473 my $x = "\n";
2474 if (my $l = $c->{l}) {
2475 while ($l->[0] =~ /^\s*$/) { shift @$l }
2476 $x = $l->[0];
2477 }
2478 $l_fmt ||= 'A' . length($c->{r});
2479 print 'r',pack($l_fmt, $c->{r}),' | ';
2480 print "$c->{c} | " if $show_commit;
2481 print $x;
2482 } else {
2483 show_commit_normal($c);
2484 }
2485 }
2486
2487 sub show_commit_changed_paths {
2488 my ($c) = @_;
2489 return unless $c->{changed};
2490 print "Changed paths:\n", @{$c->{changed}};
2491 }
2492
2493 sub show_commit_normal {
2494 my ($c) = @_;
2495 print '-' x72, "\nr$c->{r} | ";
2496 print "$c->{c} | " if $show_commit;
2497 print "$c->{a} | ", strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)",
2498 localtime($c->{t_utc})), ' | ';
2499 my $nr_line = 0;
2500
2501 if (my $l = $c->{l}) {
2502 while ($l->[$#$l] eq "\n" && $#$l > 0
2503 && $l->[($#$l - 1)] eq "\n") {
2504 pop @$l;
2505 }
2506 $nr_line = scalar @$l;
2507 if (!$nr_line) {
2508 print "1 line\n\n\n";
2509 } else {
2510 if ($nr_line == 1) {
2511 $nr_line = '1 line';
2512 } else {
2513 $nr_line .= ' lines';
2514 }
2515 print $nr_line, "\n";
2516 show_commit_changed_paths($c);
2517 print "\n";
2518 print $_ foreach @$l;
2519 }
2520 } else {
2521 print "1 line\n";
2522 show_commit_changed_paths($c);
2523 print "\n";
2524
2525 }
2526 foreach my $x (qw/raw diff/) {
2527 if ($c->{$x}) {
2528 print "\n";
2529 print $_ foreach @{$c->{$x}}
2530 }
2531 }
2532 }
2533
2534 sub cmd_show_log {
2535 my (@args) = @_;
2536 my ($r_min, $r_max);
2537 my $r_last = -1; # prevent dupes
2538 if (defined $TZ) {
2539 $ENV{TZ} = $TZ;
2540 } else {
2541 delete $ENV{TZ};
2542 }
2543 if (defined $::_revision) {
2544 if ($::_revision =~ /^(\d+):(\d+)$/) {
2545 ($r_min, $r_max) = ($1, $2);
2546 } elsif ($::_revision =~ /^\d+$/) {
2547 $r_min = $r_max = $::_revision;
2548 } else {
2549 ::fatal "-r$::_revision is not supported, use ",
2550 "standard \'git log\' arguments instead\n";
2551 }
2552 }
2553
2554 config_pager();
2555 @args = (git_svn_log_cmd($r_min, $r_max), @args);
2556 my $log = command_output_pipe(@args);
2557 run_pager();
2558 my (@k, $c, $d);
2559 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
2560 while (<$log>) {
2561 if (/^${esc_color}commit ($::sha1_short)/o) {
2562 my $cmt = $1;
2563 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
2564 $r_last = $c->{r};
2565 process_commit($c, $r_min, $r_max, \@k) or
2566 goto out;
2567 }
2568 $d = undef;
2569 $c = { c => $cmt };
2570 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
2571 get_author_info($c, $1, $2, $3);
2572 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
2573 # ignore
2574 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
2575 push @{$c->{raw}}, $_;
2576 } elsif (/^${esc_color}[ACRMDT]\t/) {
2577 # we could add $SVN->{svn_path} here, but that requires
2578 # remote access at the moment (repo_path_split)...
2579 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
2580 push @{$c->{changed}}, $_;
2581 } elsif (/^${esc_color}diff /o) {
2582 $d = 1;
2583 push @{$c->{diff}}, $_;
2584 } elsif ($d) {
2585 push @{$c->{diff}}, $_;
2586 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
2587 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
2588 } elsif (s/^${esc_color} //o) {
2589 push @{$c->{l}}, $_;
2590 }
2591 }
2592 if ($c && defined $c->{r} && $c->{r} != $r_last) {
2593 $r_last = $c->{r};
2594 process_commit($c, $r_min, $r_max, \@k);
2595 }
2596 if (@k) {
2597 my $swap = $r_max;
2598 $r_max = $r_min;
2599 $r_min = $swap;
2600 process_commit($_, $r_min, $r_max) foreach reverse @k;
2601 }
2602 out:
2603 close $log;
2604 print '-' x72,"\n" unless $incremental || $oneline;
2605 }
2606
2607 package Git::SVN::Migration;
2608 # these version numbers do NOT correspond to actual version numbers
2609 # of git nor git-svn. They are just relative.
2610 #
2611 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
2612 #
2613 # v1 layout: .git/$id/info/url, refs/remotes/$id
2614 #
2615 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
2616 #
2617 # v3 layout: .git/svn/$id, refs/remotes/$id
2618 # - info/url may remain for backwards compatibility
2619 # - this is what we migrate up to this layout automatically,
2620 # - this will be used by git svn init on single branches
2621 #
2622 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
2623 # - this is only created for newly multi-init-ed
2624 # repositories. Similar in spirit to the
2625 # --use-separate-remotes option in git-clone (now default)
2626 # - we do not automatically migrate to this (following
2627 # the example set by core git)
2628 use strict;
2629 use warnings;
2630 use Carp qw/croak/;
2631 use File::Path qw/mkpath/;
2632 use File::Basename qw/dirname basename/;
2633 use vars qw/$_minimize/;
2634
2635 sub migrate_from_v0 {
2636 my $git_dir = $ENV{GIT_DIR};
2637 return undef unless -d $git_dir;
2638 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2639 my $migrated = 0;
2640 while (<$fh>) {
2641 chomp;
2642 my ($id, $orig_ref) = ($_, $_);
2643 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
2644 next unless -f "$git_dir/$id/info/url";
2645 my $new_ref = "refs/remotes/$id";
2646 if (::verify_ref("$new_ref^0")) {
2647 print STDERR "W: $orig_ref is probably an old ",
2648 "branch used by an ancient version of ",
2649 "git-svn.\n",
2650 "However, $new_ref also exists.\n",
2651 "We will not be able ",
2652 "to use this branch until this ",
2653 "ambiguity is resolved.\n";
2654 next;
2655 }
2656 print STDERR "Migrating from v0 layout...\n" if !$migrated;
2657 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
2658 command_noisy('update-ref', $new_ref, $orig_ref);
2659 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
2660 $migrated++;
2661 }
2662 command_close_pipe($fh, $ctx);
2663 print STDERR "Done migrating from v0 layout...\n" if $migrated;
2664 $migrated;
2665 }
2666
2667 sub migrate_from_v1 {
2668 my $git_dir = $ENV{GIT_DIR};
2669 my $migrated = 0;
2670 return $migrated unless -d $git_dir;
2671 my $svn_dir = "$git_dir/svn";
2672
2673 # just in case somebody used 'svn' as their $id at some point...
2674 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
2675
2676 print STDERR "Migrating from a git-svn v1 layout...\n";
2677 mkpath([$svn_dir]);
2678 print STDERR "Data from a previous version of git-svn exists, but\n\t",
2679 "$svn_dir\n\t(required for this version ",
2680 "($::VERSION) of git-svn) does not. exist\n";
2681 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
2682 while (<$fh>) {
2683 my $x = $_;
2684 next unless $x =~ s#^refs/remotes/##;
2685 chomp $x;
2686 next unless -f "$git_dir/$x/info/url";
2687 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
2688 next unless $u;
2689 my $dn = dirname("$git_dir/svn/$x");
2690 mkpath([$dn]) unless -d $dn;
2691 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
2692 mkpath(["$git_dir/svn/svn"]);
2693 print STDERR " - $git_dir/$x/info => ",
2694 "$git_dir/svn/$x/info\n";
2695 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
2696 croak "$!: $x";
2697 # don't worry too much about these, they probably
2698 # don't exist with repos this old (save for index,
2699 # and we can easily regenerate that)
2700 foreach my $f (qw/unhandled.log index .rev_db/) {
2701 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
2702 }
2703 } else {
2704 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
2705 rename "$git_dir/$x", "$git_dir/svn/$x" or
2706 croak "$!: $x";
2707 }
2708 $migrated++;
2709 }
2710 command_close_pipe($fh, $ctx);
2711 print STDERR "Done migrating from a git-svn v1 layout\n";
2712 $migrated;
2713 }
2714
2715 sub read_old_urls {
2716 my ($l_map, $pfx, $path) = @_;
2717 my @dir;
2718 foreach (<$path/*>) {
2719 if (-r "$_/info/url") {
2720 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
2721 my $ref_id = $pfx . basename $_;
2722 my $url = ::file_to_s("$_/info/url");
2723 $l_map->{$ref_id} = $url;
2724 } elsif (-d $_) {
2725 push @dir, $_;
2726 }
2727 }
2728 foreach (@dir) {
2729 my $x = $_;
2730 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
2731 read_old_urls($l_map, $x, $_);
2732 }
2733 }
2734
2735 sub migrate_from_v2 {
2736 my @cfg = command(qw/config -l/);
2737 return if grep /^svn-remote\..+\.url=/, @cfg;
2738 my %l_map;
2739 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
2740 my $migrated = 0;
2741
2742 foreach my $ref_id (sort keys %l_map) {
2743 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
2744 $migrated++;
2745 }
2746 $migrated;
2747 }
2748
2749 sub minimize_connections {
2750 my $r = Git::SVN::read_all_remotes();
2751 my $new_urls = {};
2752 my $root_repos = {};
2753 foreach my $repo_id (keys %$r) {
2754 my $url = $r->{$repo_id}->{url} or next;
2755 my $fetch = $r->{$repo_id}->{fetch} or next;
2756 my $ra = Git::SVN::Ra->new($url);
2757
2758 # skip existing cases where we already connect to the root
2759 if (($ra->{url} eq $ra->{repos_root}) ||
2760 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
2761 $repo_id)) {
2762 $root_repos->{$ra->{url}} = $repo_id;
2763 next;
2764 }
2765
2766 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
2767 my $root_path = $ra->{url};
2768 $root_path =~ s#^\Q$ra->{repos_root}\E/*##;
2769 foreach my $path (keys %$fetch) {
2770 my $ref_id = $fetch->{$path};
2771 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
2772
2773 # make sure we can read when connecting to
2774 # a higher level of a repository
2775 my ($last_rev, undef) = $gs->last_rev_commit;
2776 if (!defined $last_rev) {
2777 $last_rev = eval {
2778 $root_ra->get_latest_revnum;
2779 };
2780 next if $@;
2781 }
2782 my $new = $root_path;
2783 $new .= length $path ? "/$path" : '';
2784 eval {
2785 $root_ra->get_log([$new], $last_rev, $last_rev,
2786 0, 0, 1, sub { });
2787 };
2788 next if $@;
2789 $new_urls->{$ra->{repos_root}}->{$new} =
2790 { ref_id => $ref_id,
2791 old_repo_id => $repo_id,
2792 old_path => $path };
2793 }
2794 }
2795
2796 my @emptied;
2797 foreach my $url (keys %$new_urls) {
2798 # see if we can re-use an existing [svn-remote "repo_id"]
2799 # instead of creating a(n ugly) new section:
2800 my $repo_id = $root_repos->{$url} ||
2801 Git::SVN::sanitize_remote_name($url);
2802
2803 my $fetch = $new_urls->{$url};
2804 foreach my $path (keys %$fetch) {
2805 my $x = $fetch->{$path};
2806 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
2807 my $pfx = "svn-remote.$x->{old_repo_id}";
2808
2809 my $old_fetch = quotemeta("$x->{old_path}:".
2810 "refs/remotes/$x->{ref_id}");
2811 command_noisy(qw/config --unset/,
2812 "$pfx.fetch", '^'. $old_fetch . '$');
2813 delete $r->{$x->{old_repo_id}}->
2814 {fetch}->{$x->{old_path}};
2815 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
2816 command_noisy(qw/config --unset/,
2817 "$pfx.url");
2818 push @emptied, $x->{old_repo_id}
2819 }
2820 }
2821 }
2822 if (@emptied) {
2823 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
2824 "$ENV{GIT_DIR}/config";
2825 print STDERR <<EOF;
2826 The following [svn-remote] sections in your config file ($file) are empty
2827 and can be safely removed:
2828 EOF
2829 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
2830 }
2831 }
2832
2833 sub migration_check {
2834 migrate_from_v0();
2835 migrate_from_v1();
2836 migrate_from_v2();
2837 minimize_connections() if $_minimize;
2838 }
2839
2840 package Git::IndexInfo;
2841 use strict;
2842 use warnings;
2843 use Git qw/command_input_pipe command_close_pipe/;
2844
2845 sub new {
2846 my ($class) = @_;
2847 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
2848 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
2849 }
2850
2851 sub remove {
2852 my ($self, $path) = @_;
2853 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
2854 return ++$self->{nr};
2855 }
2856 undef;
2857 }
2858
2859 sub update {
2860 my ($self, $mode, $hash, $path) = @_;
2861 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
2862 return ++$self->{nr};
2863 }
2864 undef;
2865 }
2866
2867 sub DESTROY {
2868 my ($self) = @_;
2869 command_close_pipe($self->{gui}, $self->{ctx});
2870 }
2871
2872 __END__
2873
2874 Data structures:
2875
2876 $log_entry hashref as returned by libsvn_log_entry()
2877 {
2878 log => 'whitespace-formatted log entry
2879 ', # trailing newline is preserved
2880 revision => '8', # integer
2881 date => '2004-02-24T17:01:44.108345Z', # commit date
2882 author => 'committer name'
2883 };
2884
2885 @mods = array of diff-index line hashes, each element represents one line
2886 of diff-index output
2887
2888 diff-index line ($m hash)
2889 {
2890 mode_a => first column of diff-index output, no leading ':',
2891 mode_b => second column of diff-index output,
2892 sha1_b => sha1sum of the final blob,
2893 chg => change type [MCRADT],
2894 file_a => original file name of a file (iff chg is 'C' or 'R')
2895 file_b => new/current file name of a file (any chg)
2896 }
2897 ;
2898
2899 # retval of read_url_paths{,_all}();
2900 $l_map = {
2901 # repository root url
2902 'https://svn.musicpd.org' => {
2903 # repository path # GIT_SVN_ID
2904 'mpd/trunk' => 'trunk',
2905 'mpd/tags/0.11.5' => 'tags/0.11.5',
2906 },
2907 }
2908
2909 Notes:
2910 I don't trust the each() function on unless I created %hash myself
2911 because the internal iterator may not have started at base.
git with Jakub Narębski modifications