![[tokkee]](http://tokkee.org/images/avatar.png){kind=avatar}
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use 5.008;
5 use warnings;
6 use strict;
7 use vars qw/ $AUTHOR $VERSION
8 $sha1 $sha1_short $_revision $_repository
9 $_q $_authors $_authors_prog %users/;
10 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
11 $VERSION = '@@GIT_VERSION@@';
13 # From which subdir have we been invoked?
14 my $cmd_dir_prefix = eval {
15 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
16 } || '';
18 my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
19 $ENV{GIT_DIR} ||= '.git';
20 $Git::SVN::default_repo_id = 'svn';
21 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
22 $Git::SVN::Ra::_log_window_size = 100;
23 $Git::SVN::_minimize_url = 'unset';
25 if (! exists $ENV{SVN_SSH}) {
26 if (exists $ENV{GIT_SSH}) {
27 $ENV{SVN_SSH} = $ENV{GIT_SSH};
28 if ($^O eq 'msys') {
29 $ENV{SVN_SSH} =~ s/\\/\\\\/g;
30 $ENV{SVN_SSH} =~ s/(.*)/"$1"/;
31 }
32 }
33 }
35 $Git::SVN::Log::TZ = $ENV{TZ};
36 $ENV{TZ} = 'UTC';
37 $| = 1; # unbuffer STDOUT
39 sub fatal (@) { print STDERR "@_\n"; exit 1 }
40 sub _req_svn {
41 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
42 require SVN::Ra;
43 require SVN::Delta;
44 if ($SVN::Core::VERSION lt '1.1.0') {
45 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
46 }
47 }
48 my $can_compress = eval { require Compress::Zlib; 1};
49 push @Git::SVN::Ra::ISA, 'SVN::Ra';
50 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
51 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
52 use Carp qw/croak/;
53 use Digest::MD5;
54 use IO::File qw//;
55 use File::Basename qw/dirname basename/;
56 use File::Path qw/mkpath/;
57 use File::Spec;
58 use File::Find;
59 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
60 use IPC::Open3;
61 use Git;
63 BEGIN {
64 # import functions from Git into our packages, en masse
65 no strict 'refs';
66 foreach (qw/command command_oneline command_noisy command_output_pipe
67 command_input_pipe command_close_pipe
68 command_bidi_pipe command_close_bidi_pipe/) {
69 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
70 Git::SVN::Migration Git::SVN::Log Git::SVN),
71 __PACKAGE__) {
72 *{"${package}::$_"} = \&{"Git::$_"};
73 }
74 }
75 }
77 my ($SVN);
79 $sha1 = qr/[a-f\d]{40}/;
80 $sha1_short = qr/[a-f\d]{4,40}/;
81 my ($_stdin, $_help, $_edit,
82 $_message, $_file, $_branch_dest,
83 $_template, $_shared,
84 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
85 $_merge, $_strategy, $_dry_run, $_local,
86 $_prefix, $_no_checkout, $_url, $_verbose,
87 $_git_format, $_commit_url, $_tag, $_merge_info);
88 $Git::SVN::_follow_parent = 1;
89 $_q ||= 0;
90 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
91 'config-dir=s' => \$Git::SVN::Ra::config_dir,
92 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
93 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
94 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
95 'authors-file|A=s' => \$_authors,
96 'authors-prog=s' => \$_authors_prog,
97 'repack:i' => \$Git::SVN::_repack,
98 'noMetadata' => \$Git::SVN::_no_metadata,
99 'useSvmProps' => \$Git::SVN::_use_svm_props,
100 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
101 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
102 'no-checkout' => \$_no_checkout,
103 'quiet|q+' => \$_q,
104 'repack-flags|repack-args|repack-opts=s' =>
105 \$Git::SVN::_repack_flags,
106 'use-log-author' => \$Git::SVN::_use_log_author,
107 'add-author-from' => \$Git::SVN::_add_author_from,
108 'localtime' => \$Git::SVN::_localtime,
109 %remote_opts );
111 my ($_trunk, @_tags, @_branches, $_stdlayout);
112 my %icv;
113 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
114 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
115 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
116 'stdlayout|s' => \$_stdlayout,
117 'minimize-url|m!' => \$Git::SVN::_minimize_url,
118 'no-metadata' => sub { $icv{noMetadata} = 1 },
119 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
120 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
121 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
122 'rewrite-uuid=s' => sub { $icv{rewriteUUID} = $_[1] },
123 %remote_opts );
124 my %cmt_opts = ( 'edit|e' => \$_edit,
125 'rmdir' => \$SVN::Git::Editor::_rmdir,
126 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
127 'l=i' => \$SVN::Git::Editor::_rename_limit,
128 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
129 );
131 my %cmd = (
132 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
133 { 'revision|r=s' => \$_revision,
134 'fetch-all|all' => \$_fetch_all,
135 'parent|p' => \$_fetch_parent,
136 %fc_opts } ],
137 clone => [ \&cmd_clone, "Initialize and fetch revisions",
138 { 'revision|r=s' => \$_revision,
139 %fc_opts, %init_opts } ],
140 init => [ \&cmd_init, "Initialize a repo for tracking" .
141 " (requires URL argument)",
142 \%init_opts ],
143 'multi-init' => [ \&cmd_multi_init,
144 "Deprecated alias for ".
145 "'$0 init -T<trunk> -b<branches> -t<tags>'",
146 \%init_opts ],
147 dcommit => [ \&cmd_dcommit,
148 'Commit several diffs to merge with upstream',
149 { 'merge|m|M' => \$_merge,
150 'strategy|s=s' => \$_strategy,
151 'verbose|v' => \$_verbose,
152 'dry-run|n' => \$_dry_run,
153 'fetch-all|all' => \$_fetch_all,
154 'commit-url=s' => \$_commit_url,
155 'revision|r=i' => \$_revision,
156 'no-rebase' => \$_no_rebase,
157 'mergeinfo=s' => \$_merge_info,
158 %cmt_opts, %fc_opts } ],
159 branch => [ \&cmd_branch,
160 'Create a branch in the SVN repository',
161 { 'message|m=s' => \$_message,
162 'destination|d=s' => \$_branch_dest,
163 'dry-run|n' => \$_dry_run,
164 'tag|t' => \$_tag,
165 'username=s' => \$Git::SVN::Prompt::_username,
166 'commit-url=s' => \$_commit_url } ],
167 tag => [ sub { $_tag = 1; cmd_branch(@_) },
168 'Create a tag in the SVN repository',
169 { 'message|m=s' => \$_message,
170 'destination|d=s' => \$_branch_dest,
171 'dry-run|n' => \$_dry_run,
172 'username=s' => \$Git::SVN::Prompt::_username,
173 'commit-url=s' => \$_commit_url } ],
174 'set-tree' => [ \&cmd_set_tree,
175 "Set an SVN repository to a git tree-ish",
176 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
177 'create-ignore' => [ \&cmd_create_ignore,
178 'Create a .gitignore per svn:ignore',
179 { 'revision|r=i' => \$_revision
180 } ],
181 'mkdirs' => [ \&cmd_mkdirs ,
182 "recreate empty directories after a checkout",
183 { 'revision|r=i' => \$_revision } ],
184 'propget' => [ \&cmd_propget,
185 'Print the value of a property on a file or directory',
186 { 'revision|r=i' => \$_revision } ],
187 'proplist' => [ \&cmd_proplist,
188 'List all properties of a file or directory',
189 { 'revision|r=i' => \$_revision } ],
190 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
191 { 'revision|r=i' => \$_revision
192 } ],
193 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
194 { 'revision|r=i' => \$_revision
195 } ],
196 'multi-fetch' => [ \&cmd_multi_fetch,
197 "Deprecated alias for $0 fetch --all",
198 { 'revision|r=s' => \$_revision, %fc_opts } ],
199 'migrate' => [ sub { },
200 # no-op, we automatically run this anyways,
201 'Migrate configuration/metadata/layout from
202 previous versions of git-svn',
203 { 'minimize' => \$Git::SVN::Migration::_minimize,
204 %remote_opts } ],
205 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
206 { 'limit=i' => \$Git::SVN::Log::limit,
207 'revision|r=s' => \$_revision,
208 'verbose|v' => \$Git::SVN::Log::verbose,
209 'incremental' => \$Git::SVN::Log::incremental,
210 'oneline' => \$Git::SVN::Log::oneline,
211 'show-commit' => \$Git::SVN::Log::show_commit,
212 'non-recursive' => \$Git::SVN::Log::non_recursive,
213 'authors-file|A=s' => \$_authors,
214 'color' => \$Git::SVN::Log::color,
215 'pager=s' => \$Git::SVN::Log::pager
216 } ],
217 'find-rev' => [ \&cmd_find_rev,
218 "Translate between SVN revision numbers and tree-ish",
219 {} ],
220 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
221 { 'merge|m|M' => \$_merge,
222 'verbose|v' => \$_verbose,
223 'strategy|s=s' => \$_strategy,
224 'local|l' => \$_local,
225 'fetch-all|all' => \$_fetch_all,
226 'dry-run|n' => \$_dry_run,
227 %fc_opts } ],
228 'commit-diff' => [ \&cmd_commit_diff,
229 'Commit a diff between two trees',
230 { 'message|m=s' => \$_message,
231 'file|F=s' => \$_file,
232 'revision|r=s' => \$_revision,
233 %cmt_opts } ],
234 'info' => [ \&cmd_info,
235 "Show info about the latest SVN revision
236 on the current branch",
237 { 'url' => \$_url, } ],
238 'blame' => [ \&Git::SVN::Log::cmd_blame,
239 "Show what revision and author last modified each line of a file",
240 { 'git-format' => \$_git_format } ],
241 'reset' => [ \&cmd_reset,
242 "Undo fetches back to the specified SVN revision",
243 { 'revision|r=s' => \$_revision,
244 'parent|p' => \$_fetch_parent } ],
245 'gc' => [ \&cmd_gc,
246 "Compress unhandled.log files in .git/svn and remove " .
247 "index files in .git/svn",
248 {} ],
249 );
251 my $cmd;
252 for (my $i = 0; $i < @ARGV; $i++) {
253 if (defined $cmd{$ARGV[$i]}) {
254 $cmd = $ARGV[$i];
255 splice @ARGV, $i, 1;
256 last;
257 } elsif ($ARGV[$i] eq 'help') {
258 $cmd = $ARGV[$i+1];
259 usage(0);
260 }
261 };
263 # make sure we're always running at the top-level working directory
264 unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
265 unless (-d $ENV{GIT_DIR}) {
266 if ($git_dir_user_set) {
267 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
268 "but it is not a directory\n";
269 }
270 my $git_dir = delete $ENV{GIT_DIR};
271 my $cdup = undef;
272 git_cmd_try {
273 $cdup = command_oneline(qw/rev-parse --show-cdup/);
274 $git_dir = '.' unless ($cdup);
275 chomp $cdup if ($cdup);
276 $cdup = "." unless ($cdup && length $cdup);
277 } "Already at toplevel, but $git_dir not found\n";
278 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
279 unless (-d $git_dir) {
280 die "$git_dir still not found after going to ",
281 "'$cdup'\n";
282 }
283 $ENV{GIT_DIR} = $git_dir;
284 }
285 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
286 }
288 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
290 read_git_config(\%opts);
291 if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
292 Getopt::Long::Configure('pass_through');
293 }
294 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
295 'minimize-connections' => \$Git::SVN::Migration::_minimize,
296 'id|i=s' => \$Git::SVN::default_ref_id,
297 'svn-remote|remote|R=s' => sub {
298 $Git::SVN::no_reuse_existing = 1;
299 $Git::SVN::default_repo_id = $_[1] });
300 exit 1 if (!$rv && $cmd && $cmd ne 'log');
302 usage(0) if $_help;
303 version() if $_version;
304 usage(1) unless defined $cmd;
305 load_authors() if $_authors;
306 if (defined $_authors_prog) {
307 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
308 }
310 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
311 Git::SVN::Migration::migration_check();
312 }
313 Git::SVN::init_vars();
314 eval {
315 Git::SVN::verify_remotes_sanity();
316 $cmd{$cmd}->[0]->(@ARGV);
317 };
318 fatal $@ if $@;
319 post_fetch_checkout();
320 exit 0;
322 ####################### primary functions ######################
323 sub usage {
324 my $exit = shift || 0;
325 my $fd = $exit ? \*STDERR : \*STDOUT;
326 print $fd <<"";
327 git-svn - bidirectional operations between a single Subversion tree and git
328 Usage: git svn <command> [options] [arguments]\n
330 print $fd "Available commands:\n" unless $cmd;
332 foreach (sort keys %cmd) {
333 next if $cmd && $cmd ne $_;
334 next if /^multi-/; # don't show deprecated commands
335 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
336 foreach (sort keys %{$cmd{$_}->[2]}) {
337 # mixed-case options are for .git/config only
338 next if /[A-Z]/ && /^[a-z]+$/i;
339 # prints out arguments as they should be passed:
340 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
341 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
342 "--$_" : "-$_" }
343 split /\|/,$_)," $x\n";
344 }
345 }
346 print $fd <<"";
347 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
348 arbitrary identifier if you're tracking multiple SVN branches/repositories in
349 one git repository and want to keep them separate. See git-svn(1) for more
350 information.
352 exit $exit;
353 }
355 sub version {
356 ::_req_svn();
357 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
358 exit 0;
359 }
361 sub do_git_init_db {
362 unless (-d $ENV{GIT_DIR}) {
363 my @init_db = ('init');
364 push @init_db, "--template=$_template" if defined $_template;
365 if (defined $_shared) {
366 if ($_shared =~ /[a-z]/) {
367 push @init_db, "--shared=$_shared";
368 } else {
369 push @init_db, "--shared";
370 }
371 }
372 command_noisy(@init_db);
373 $_repository = Git->repository(Repository => ".git");
374 }
375 my $set;
376 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
377 foreach my $i (keys %icv) {
378 die "'$set' and '$i' cannot both be set\n" if $set;
379 next unless defined $icv{$i};
380 command_noisy('config', "$pfx.$i", $icv{$i});
381 $set = $i;
382 }
383 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
384 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
385 if defined $$ignore_regex;
386 }
388 sub init_subdir {
389 my $repo_path = shift or return;
390 mkpath([$repo_path]) unless -d $repo_path;
391 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
392 $ENV{GIT_DIR} = '.git';
393 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
394 }
396 sub cmd_clone {
397 my ($url, $path) = @_;
398 if (!defined $path &&
399 (defined $_trunk || @_branches || @_tags ||
400 defined $_stdlayout) &&
401 $url !~ m#^[a-z\+]+://#) {
402 $path = $url;
403 }
404 $path = basename($url) if !defined $path || !length $path;
405 my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
406 cmd_init($url, $path);
407 command_oneline('config', 'svn.authorsfile', $authors_absolute)
408 if $_authors;
409 Git::SVN::fetch_all($Git::SVN::default_repo_id);
410 }
412 sub cmd_init {
413 if (defined $_stdlayout) {
414 $_trunk = 'trunk' if (!defined $_trunk);
415 @_tags = 'tags' if (! @_tags);
416 @_branches = 'branches' if (! @_branches);
417 }
418 if (defined $_trunk || @_branches || @_tags) {
419 return cmd_multi_init(@_);
420 }
421 my $url = shift or die "SVN repository location required ",
422 "as a command-line argument\n";
423 $url = canonicalize_url($url);
424 init_subdir(@_);
425 do_git_init_db();
427 if ($Git::SVN::_minimize_url eq 'unset') {
428 $Git::SVN::_minimize_url = 0;
429 }
431 Git::SVN->init($url);
432 }
434 sub cmd_fetch {
435 if (grep /^\d+=./, @_) {
436 die "'<rev>=<commit>' fetch arguments are ",
437 "no longer supported.\n";
438 }
439 my ($remote) = @_;
440 if (@_ > 1) {
441 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
442 }
443 $Git::SVN::no_reuse_existing = undef;
444 if ($_fetch_parent) {
445 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
446 unless ($gs) {
447 die "Unable to determine upstream SVN information from ",
448 "working tree history\n";
449 }
450 # just fetch, don't checkout.
451 $_no_checkout = 'true';
452 $_fetch_all ? $gs->fetch_all : $gs->fetch;
453 } elsif ($_fetch_all) {
454 cmd_multi_fetch();
455 } else {
456 $remote ||= $Git::SVN::default_repo_id;
457 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
458 }
459 }
461 sub cmd_set_tree {
462 my (@commits) = @_;
463 if ($_stdin || !@commits) {
464 print "Reading from stdin...\n";
465 @commits = ();
466 while (<STDIN>) {
467 if (/\b($sha1_short)\b/o) {
468 unshift @commits, $1;
469 }
470 }
471 }
472 my @revs;
473 foreach my $c (@commits) {
474 my @tmp = command('rev-parse',$c);
475 if (scalar @tmp == 1) {
476 push @revs, $tmp[0];
477 } elsif (scalar @tmp > 1) {
478 push @revs, reverse(command('rev-list',@tmp));
479 } else {
480 fatal "Failed to rev-parse $c";
481 }
482 }
483 my $gs = Git::SVN->new;
484 my ($r_last, $cmt_last) = $gs->last_rev_commit;
485 $gs->fetch;
486 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
487 fatal "There are new revisions that were fetched ",
488 "and need to be merged (or acknowledged) ",
489 "before committing.\nlast rev: $r_last\n",
490 " current: $gs->{last_rev}";
491 }
492 $gs->set_tree($_) foreach @revs;
493 print "Done committing ",scalar @revs," revisions to SVN\n";
494 unlink $gs->{index};
495 }
497 sub cmd_dcommit {
498 my $head = shift;
499 command_noisy(qw/update-index --refresh/);
500 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
501 'Cannot dcommit with a dirty index. Commit your changes first, '
502 . "or stash them with `git stash'.\n";
503 $head ||= 'HEAD';
505 my $old_head;
506 if ($head ne 'HEAD') {
507 $old_head = eval {
508 command_oneline([qw/symbolic-ref -q HEAD/])
509 };
510 if ($old_head) {
511 $old_head =~ s{^refs/heads/}{};
512 } else {
513 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
514 }
515 command(['checkout', $head], STDERR => 0);
516 }
518 my @refs;
519 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
520 unless ($gs) {
521 die "Unable to determine upstream SVN information from ",
522 "$head history.\nPerhaps the repository is empty.";
523 }
525 if (defined $_commit_url) {
526 $url = $_commit_url;
527 } else {
528 $url = eval { command_oneline('config', '--get',
529 "svn-remote.$gs->{repo_id}.commiturl") };
530 if (!$url) {
531 $url = $gs->full_url
532 }
533 }
535 my $last_rev = $_revision if defined $_revision;
536 if ($url) {
537 print "Committing to $url ...\n";
538 }
539 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
540 if ($_no_rebase && scalar(@$linear_refs) > 1) {
541 warn "Attempting to commit more than one change while ",
542 "--no-rebase is enabled.\n",
543 "If these changes depend on each other, re-running ",
544 "without --no-rebase may be required."
545 }
546 my $expect_url = $url;
547 Git::SVN::remove_username($expect_url);
548 while (1) {
549 my $d = shift @$linear_refs or last;
550 unless (defined $last_rev) {
551 (undef, $last_rev, undef) = cmt_metadata("$d~1");
552 unless (defined $last_rev) {
553 fatal "Unable to extract revision information ",
554 "from commit $d~1";
555 }
556 }
557 if ($_dry_run) {
558 print "diff-tree $d~1 $d\n";
559 } else {
560 my $cmt_rev;
561 my %ed_opts = ( r => $last_rev,
562 log => get_commit_entry($d)->{log},
563 ra => Git::SVN::Ra->new($url),
564 config => SVN::Core::config_get_config(
565 $Git::SVN::Ra::config_dir
566 ),
567 tree_a => "$d~1",
568 tree_b => $d,
569 editor_cb => sub {
570 print "Committed r$_[0]\n";
571 $cmt_rev = $_[0];
572 },
573 mergeinfo => $_merge_info,
574 svn_path => '');
575 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
576 print "No changes\n$d~1 == $d\n";
577 } elsif ($parents->{$d} && @{$parents->{$d}}) {
578 $gs->{inject_parents_dcommit}->{$cmt_rev} =
579 $parents->{$d};
580 }
581 $_fetch_all ? $gs->fetch_all : $gs->fetch;
582 $last_rev = $cmt_rev;
583 next if $_no_rebase;
585 # we always want to rebase against the current HEAD,
586 # not any head that was passed to us
587 my @diff = command('diff-tree', $d,
588 $gs->refname, '--');
589 my @finish;
590 if (@diff) {
591 @finish = rebase_cmd();
592 print STDERR "W: $d and ", $gs->refname,
593 " differ, using @finish:\n",
594 join("\n", @diff), "\n";
595 } else {
596 print "No changes between current HEAD and ",
597 $gs->refname,
598 "\nResetting to the latest ",
599 $gs->refname, "\n";
600 @finish = qw/reset --mixed/;
601 }
602 command_noisy(@finish, $gs->refname);
603 if (@diff) {
604 @refs = ();
605 my ($url_, $rev_, $uuid_, $gs_) =
606 working_head_info('HEAD', \@refs);
607 my ($linear_refs_, $parents_) =
608 linearize_history($gs_, \@refs);
609 if (scalar(@$linear_refs) !=
610 scalar(@$linear_refs_)) {
611 fatal "# of revisions changed ",
612 "\nbefore:\n",
613 join("\n", @$linear_refs),
614 "\n\nafter:\n",
615 join("\n", @$linear_refs_), "\n",
616 'If you are attempting to commit ',
617 "merges, try running:\n\t",
618 'git rebase --interactive',
619 '--preserve-merges ',
620 $gs->refname,
621 "\nBefore dcommitting";
622 }
623 if ($url_ ne $expect_url) {
624 if ($url_ eq $gs->metadata_url) {
625 print
626 "Accepting rewritten URL:",
627 " $url_\n";
628 } else {
629 fatal
630 "URL mismatch after rebase:",
631 " $url_ != $expect_url";
632 }
633 }
634 if ($uuid_ ne $uuid) {
635 fatal "uuid mismatch after rebase: ",
636 "$uuid_ != $uuid";
637 }
638 # remap parents
639 my (%p, @l, $i);
640 for ($i = 0; $i < scalar @$linear_refs; $i++) {
641 my $new = $linear_refs_->[$i] or next;
642 $p{$new} =
643 $parents->{$linear_refs->[$i]};
644 push @l, $new;
645 }
646 $parents = \%p;
647 $linear_refs = \@l;
648 }
649 }
650 }
652 if ($old_head) {
653 my $new_head = command_oneline(qw/rev-parse HEAD/);
654 my $new_is_symbolic = eval {
655 command_oneline(qw/symbolic-ref -q HEAD/);
656 };
657 if ($new_is_symbolic) {
658 print "dcommitted the branch ", $head, "\n";
659 } else {
660 print "dcommitted on a detached HEAD because you gave ",
661 "a revision argument.\n",
662 "The rewritten commit is: ", $new_head, "\n";
663 }
664 command(['checkout', $old_head], STDERR => 0);
665 }
667 unlink $gs->{index};
668 }
670 sub cmd_branch {
671 my ($branch_name, $head) = @_;
673 unless (defined $branch_name && length $branch_name) {
674 die(($_tag ? "tag" : "branch") . " name required\n");
675 }
676 $head ||= 'HEAD';
678 my (undef, $rev, undef, $gs) = working_head_info($head);
679 my $src = $gs->full_url;
681 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
682 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
683 my $glob;
684 if ($#{$allglobs} == 0) {
685 $glob = $allglobs->[0];
686 } else {
687 unless(defined $_branch_dest) {
688 die "Multiple ",
689 $_tag ? "tag" : "branch",
690 " paths defined for Subversion repository.\n",
691 "You must specify where you want to create the ",
692 $_tag ? "tag" : "branch",
693 " with the --destination argument.\n";
694 }
695 foreach my $g (@{$allglobs}) {
696 # SVN::Git::Editor could probably be moved to Git.pm..
697 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
698 if ($_branch_dest =~ /$re/) {
699 $glob = $g;
700 last;
701 }
702 }
703 unless (defined $glob) {
704 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
705 foreach my $g (@{$allglobs}) {
706 $g->{path}->{left} =~ /$dest_re/ or next;
707 if (defined $glob) {
708 die "Ambiguous destination: ",
709 $_branch_dest, "\nmatches both '",
710 $glob->{path}->{left}, "' and '",
711 $g->{path}->{left}, "'\n";
712 }
713 $glob = $g;
714 }
715 unless (defined $glob) {
716 die "Unknown ",
717 $_tag ? "tag" : "branch",
718 " destination $_branch_dest\n";
719 }
720 }
721 }
722 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
723 my $url;
724 if (defined $_commit_url) {
725 $url = $_commit_url;
726 } else {
727 $url = eval { command_oneline('config', '--get',
728 "svn-remote.$gs->{repo_id}.commiturl") };
729 if (!$url) {
730 $url = $remote->{url};
731 }
732 }
733 my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
735 if ($dst =~ /^https:/ && $src =~ /^http:/) {
736 $src=~s/^http:/https:/;
737 }
739 ::_req_svn();
741 my $ctx = SVN::Client->new(
742 auth => Git::SVN::Ra::_auth_providers(),
743 log_msg => sub {
744 ${ $_[0] } = defined $_message
745 ? $_message
746 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
747 . $branch_name;
748 },
749 );
751 eval {
752 $ctx->ls($dst, 'HEAD', 0);
753 } and die "branch ${branch_name} already exists\n";
755 print "Copying ${src} at r${rev} to ${dst}...\n";
756 $ctx->copy($src, $rev, $dst)
757 unless $_dry_run;
759 $gs->fetch_all;
760 }
762 sub cmd_find_rev {
763 my $revision_or_hash = shift or die "SVN or git revision required ",
764 "as a command-line argument\n";
765 my $result;
766 if ($revision_or_hash =~ /^r\d+$/) {
767 my $head = shift;
768 $head ||= 'HEAD';
769 my @refs;
770 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
771 unless ($gs) {
772 die "Unable to determine upstream SVN information from ",
773 "$head history\n";
774 }
775 my $desired_revision = substr($revision_or_hash, 1);
776 $result = $gs->rev_map_get($desired_revision, $uuid);
777 } else {
778 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
779 $result = $rev;
780 }
781 print "$result\n" if $result;
782 }
784 sub cmd_rebase {
785 command_noisy(qw/update-index --refresh/);
786 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
787 unless ($gs) {
788 die "Unable to determine upstream SVN information from ",
789 "working tree history\n";
790 }
791 if ($_dry_run) {
792 print "Remote Branch: " . $gs->refname . "\n";
793 print "SVN URL: " . $url . "\n";
794 return;
795 }
796 if (command(qw/diff-index HEAD --/)) {
797 print STDERR "Cannot rebase with uncommited changes:\n";
798 command_noisy('status');
799 exit 1;
800 }
801 unless ($_local) {
802 # rebase will checkout for us, so no need to do it explicitly
803 $_no_checkout = 'true';
804 $_fetch_all ? $gs->fetch_all : $gs->fetch;
805 }
806 command_noisy(rebase_cmd(), $gs->refname);
807 $gs->mkemptydirs;
808 }
810 sub cmd_show_ignore {
811 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
812 $gs ||= Git::SVN->new;
813 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
814 $gs->prop_walk($gs->{path}, $r, sub {
815 my ($gs, $path, $props) = @_;
816 print STDOUT "\n# $path\n";
817 my $s = $props->{'svn:ignore'} or return;
818 $s =~ s/[\r\n]+/\n/g;
819 $s =~ s/^\n+//;
820 chomp $s;
821 $s =~ s#^#$path#gm;
822 print STDOUT "$s\n";
823 });
824 }
826 sub cmd_show_externals {
827 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
828 $gs ||= Git::SVN->new;
829 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
830 $gs->prop_walk($gs->{path}, $r, sub {
831 my ($gs, $path, $props) = @_;
832 print STDOUT "\n# $path\n";
833 my $s = $props->{'svn:externals'} or return;
834 $s =~ s/[\r\n]+/\n/g;
835 chomp $s;
836 $s =~ s#^#$path#gm;
837 print STDOUT "$s\n";
838 });
839 }
841 sub cmd_create_ignore {
842 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
843 $gs ||= Git::SVN->new;
844 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
845 $gs->prop_walk($gs->{path}, $r, sub {
846 my ($gs, $path, $props) = @_;
847 # $path is of the form /path/to/dir/
848 $path = '.' . $path;
849 # SVN can have attributes on empty directories,
850 # which git won't track
851 mkpath([$path]) unless -d $path;
852 my $ignore = $path . '.gitignore';
853 my $s = $props->{'svn:ignore'} or return;
854 open(GITIGNORE, '>', $ignore)
855 or fatal("Failed to open `$ignore' for writing: $!");
856 $s =~ s/[\r\n]+/\n/g;
857 $s =~ s/^\n+//;
858 chomp $s;
859 # Prefix all patterns so that the ignore doesn't apply
860 # to sub-directories.
861 $s =~ s#^#/#gm;
862 print GITIGNORE "$s\n";
863 close(GITIGNORE)
864 or fatal("Failed to close `$ignore': $!");
865 command_noisy('add', '-f', $ignore);
866 });
867 }
869 sub cmd_mkdirs {
870 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
871 $gs ||= Git::SVN->new;
872 $gs->mkemptydirs($_revision);
873 }
875 sub canonicalize_path {
876 my ($path) = @_;
877 my $dot_slash_added = 0;
878 if (substr($path, 0, 1) ne "/") {
879 $path = "./" . $path;
880 $dot_slash_added = 1;
881 }
882 # File::Spec->canonpath doesn't collapse x/../y into y (for a
883 # good reason), so let's do this manually.
884 $path =~ s#/+#/#g;
885 $path =~ s#/\.(?:/|$)#/#g;
886 $path =~ s#/[^/]+/\.\.##g;
887 $path =~ s#/$##g;
888 $path =~ s#^\./## if $dot_slash_added;
889 $path =~ s#^/##;
890 $path =~ s#^\.$##;
891 return $path;
892 }
894 sub canonicalize_url {
895 my ($url) = @_;
896 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
897 return $url;
898 }
900 # get_svnprops(PATH)
901 # ------------------
902 # Helper for cmd_propget and cmd_proplist below.
903 sub get_svnprops {
904 my $path = shift;
905 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
906 $gs ||= Git::SVN->new;
908 # prefix THE PATH by the sub-directory from which the user
909 # invoked us.
910 $path = $cmd_dir_prefix . $path;
911 fatal("No such file or directory: $path") unless -e $path;
912 my $is_dir = -d $path ? 1 : 0;
913 $path = $gs->{path} . '/' . $path;
915 # canonicalize the path (otherwise libsvn will abort or fail to
916 # find the file)
917 $path = canonicalize_path($path);
919 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
920 my $props;
921 if ($is_dir) {
922 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
923 }
924 else {
925 (undef, $props) = $gs->ra->get_file($path, $r, undef);
926 }
927 return $props;
928 }
930 # cmd_propget (PROP, PATH)
931 # ------------------------
932 # Print the SVN property PROP for PATH.
933 sub cmd_propget {
934 my ($prop, $path) = @_;
935 $path = '.' if not defined $path;
936 usage(1) if not defined $prop;
937 my $props = get_svnprops($path);
938 if (not defined $props->{$prop}) {
939 fatal("`$path' does not have a `$prop' SVN property.");
940 }
941 print $props->{$prop} . "\n";
942 }
944 # cmd_proplist (PATH)
945 # -------------------
946 # Print the list of SVN properties for PATH.
947 sub cmd_proplist {
948 my $path = shift;
949 $path = '.' if not defined $path;
950 my $props = get_svnprops($path);
951 print "Properties on '$path':\n";
952 foreach (sort keys %{$props}) {
953 print " $_\n";
954 }
955 }
957 sub cmd_multi_init {
958 my $url = shift;
959 unless (defined $_trunk || @_branches || @_tags) {
960 usage(1);
961 }
963 $_prefix = '' unless defined $_prefix;
964 if (defined $url) {
965 $url = canonicalize_url($url);
966 init_subdir(@_);
967 }
968 do_git_init_db();
969 if (defined $_trunk) {
970 $_trunk =~ s#^/+##;
971 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
972 # try both old-style and new-style lookups:
973 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
974 unless ($gs_trunk) {
975 my ($trunk_url, $trunk_path) =
976 complete_svn_url($url, $_trunk);
977 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
978 undef, $trunk_ref);
979 }
980 }
981 return unless @_branches || @_tags;
982 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
983 foreach my $path (@_branches) {
984 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
985 }
986 foreach my $path (@_tags) {
987 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
988 }
989 }
991 sub cmd_multi_fetch {
992 $Git::SVN::no_reuse_existing = undef;
993 my $remotes = Git::SVN::read_all_remotes();
994 foreach my $repo_id (sort keys %$remotes) {
995 if ($remotes->{$repo_id}->{url}) {
996 Git::SVN::fetch_all($repo_id, $remotes);
997 }
998 }
999 }
1001 # this command is special because it requires no metadata
1002 sub cmd_commit_diff {
1003 my ($ta, $tb, $url) = @_;
1004 my $usage = "Usage: $0 commit-diff -r<revision> ".
1005 "<tree-ish> <tree-ish> [<URL>]";
1006 fatal($usage) if (!defined $ta || !defined $tb);
1007 my $svn_path = '';
1008 if (!defined $url) {
1009 my $gs = eval { Git::SVN->new };
1010 if (!$gs) {
1011 fatal("Needed URL or usable git-svn --id in ",
1012 "the command-line\n", $usage);
1013 }
1014 $url = $gs->{url};
1015 $svn_path = $gs->{path};
1016 }
1017 unless (defined $_revision) {
1018 fatal("-r|--revision is a required argument\n", $usage);
1019 }
1020 if (defined $_message && defined $_file) {
1021 fatal("Both --message/-m and --file/-F specified ",
1022 "for the commit message.\n",
1023 "I have no idea what you mean");
1024 }
1025 if (defined $_file) {
1026 $_message = file_to_s($_file);
1027 } else {
1028 $_message ||= get_commit_entry($tb)->{log};
1029 }
1030 my $ra ||= Git::SVN::Ra->new($url);
1031 my $r = $_revision;
1032 if ($r eq 'HEAD') {
1033 $r = $ra->get_latest_revnum;
1034 } elsif ($r !~ /^\d+$/) {
1035 die "revision argument: $r not understood by git-svn\n";
1036 }
1037 my %ed_opts = ( r => $r,
1038 log => $_message,
1039 ra => $ra,
1040 tree_a => $ta,
1041 tree_b => $tb,
1042 editor_cb => sub { print "Committed r$_[0]\n" },
1043 svn_path => $svn_path );
1044 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1045 print "No changes\n$ta == $tb\n";
1046 }
1047 }
1049 sub escape_uri_only {
1050 my ($uri) = @_;
1051 my @tmp;
1052 foreach (split m{/}, $uri) {
1053 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1054 push @tmp, $_;
1055 }
1056 join('/', @tmp);
1057 }
1059 sub escape_url {
1060 my ($url) = @_;
1061 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1062 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1063 $url = "$scheme://$domain$uri";
1064 }
1065 $url;
1066 }
1068 sub cmd_info {
1069 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1070 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1071 if (exists $_[1]) {
1072 die "Too many arguments specified\n";
1073 }
1075 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1077 if (!$file_type && !$diff_status) {
1078 print STDERR "svn: '$path' is not under version control\n";
1079 exit 1;
1080 }
1082 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1083 unless ($gs) {
1084 die "Unable to determine upstream SVN information from ",
1085 "working tree history\n";
1086 }
1088 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1089 $path = "." if $path eq "";
1091 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1093 if ($_url) {
1094 print escape_url($full_url), "\n";
1095 return;
1096 }
1098 my $result = "Path: $path\n";
1099 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1100 $result .= "URL: " . escape_url($full_url) . "\n";
1102 eval {
1103 my $repos_root = $gs->repos_root;
1104 Git::SVN::remove_username($repos_root);
1105 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1106 };
1107 if ($@) {
1108 $result .= "Repository Root: (offline)\n";
1109 }
1110 ::_req_svn();
1111 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1112 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1113 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1115 $result .= "Node Kind: " .
1116 ($file_type eq "dir" ? "directory" : "file") . "\n";
1118 my $schedule = $diff_status eq "A"
1119 ? "add"
1120 : ($diff_status eq "D" ? "delete" : "normal");
1121 $result .= "Schedule: $schedule\n";
1123 if ($diff_status eq "A") {
1124 print $result, "\n";
1125 return;
1126 }
1128 my ($lc_author, $lc_rev, $lc_date_utc);
1129 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1130 my $log = command_output_pipe(@args);
1131 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1132 while (<$log>) {
1133 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1134 $lc_author = $1;
1135 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1136 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1137 (undef, $lc_rev, undef) = ::extract_metadata($1);
1138 }
1139 }
1140 close $log;
1142 Git::SVN::Log::set_local_timezone();
1144 $result .= "Last Changed Author: $lc_author\n";
1145 $result .= "Last Changed Rev: $lc_rev\n";
1146 $result .= "Last Changed Date: " .
1147 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1149 if ($file_type ne "dir") {
1150 my $text_last_updated_date =
1151 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1152 $result .=
1153 "Text Last Updated: " .
1154 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1155 "\n";
1156 my $checksum;
1157 if ($diff_status eq "D") {
1158 my ($fh, $ctx) =
1159 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1160 if ($file_type eq "link") {
1161 my $file_name = <$fh>;
1162 $checksum = md5sum("link $file_name");
1163 } else {
1164 $checksum = md5sum($fh);
1165 }
1166 command_close_pipe($fh, $ctx);
1167 } elsif ($file_type eq "link") {
1168 my $file_name =
1169 command(qw(cat-file blob), "HEAD:$path");
1170 $checksum =
1171 md5sum("link " . $file_name);
1172 } else {
1173 open FILE, "<", $path or die $!;
1174 $checksum = md5sum(\*FILE);
1175 close FILE or die $!;
1176 }
1177 $result .= "Checksum: " . $checksum . "\n";
1178 }
1180 print $result, "\n";
1181 }
1183 sub cmd_reset {
1184 my $target = shift || $_revision or die "SVN revision required\n";
1185 $target = $1 if $target =~ /^r(\d+)$/;
1186 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1187 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1188 unless ($gs) {
1189 die "Unable to determine upstream SVN information from ".
1190 "history\n";
1191 }
1192 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1193 die "Cannot find SVN revision $target\n" unless defined($c);
1194 $gs->rev_map_set($r, $c, 'reset', $uuid);
1195 print "r$r = $c ($gs->{ref_id})\n";
1196 }
1198 sub cmd_gc {
1199 if (!$can_compress) {
1200 warn "Compress::Zlib could not be found; unhandled.log " .
1201 "files will not be compressed.\n";
1202 }
1203 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1204 }
1206 ########################### utility functions #########################
1208 sub rebase_cmd {
1209 my @cmd = qw/rebase/;
1210 push @cmd, '-v' if $_verbose;
1211 push @cmd, qw/--merge/ if $_merge;
1212 push @cmd, "--strategy=$_strategy" if $_strategy;
1213 @cmd;
1214 }
1216 sub post_fetch_checkout {
1217 return if $_no_checkout;
1218 my $gs = $Git::SVN::_head or return;
1219 return if verify_ref('refs/heads/master^0');
1221 # look for "trunk" ref if it exists
1222 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1223 my $fetch = $remote->{fetch};
1224 if ($fetch) {
1225 foreach my $p (keys %$fetch) {
1226 basename($fetch->{$p}) eq 'trunk' or next;
1227 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1228 last;
1229 }
1230 }
1232 my $valid_head = verify_ref('HEAD^0');
1233 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1234 return if ($valid_head || !verify_ref('HEAD^0'));
1236 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1237 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1238 return if -f $index;
1240 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1241 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1242 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1243 print STDERR "Checked out HEAD:\n ",
1244 $gs->full_url, " r", $gs->last_rev, "\n";
1245 $gs->mkemptydirs($gs->last_rev);
1246 }
1248 sub complete_svn_url {
1249 my ($url, $path) = @_;
1250 $path =~ s#/+$##;
1251 if ($path !~ m#^[a-z\+]+://#) {
1252 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1253 fatal("E: '$path' is not a complete URL ",
1254 "and a separate URL is not specified");
1255 }
1256 return ($url, $path);
1257 }
1258 return ($path, '');
1259 }
1261 sub complete_url_ls_init {
1262 my ($ra, $repo_path, $switch, $pfx) = @_;
1263 unless ($repo_path) {
1264 print STDERR "W: $switch not specified\n";
1265 return;
1266 }
1267 $repo_path =~ s#/+$##;
1268 if ($repo_path =~ m#^[a-z\+]+://#) {
1269 $ra = Git::SVN::Ra->new($repo_path);
1270 $repo_path = '';
1271 } else {
1272 $repo_path =~ s#^/+##;
1273 unless ($ra) {
1274 fatal("E: '$repo_path' is not a complete URL ",
1275 "and a separate URL is not specified");
1276 }
1277 }
1278 my $url = $ra->{url};
1279 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1280 my $k = "svn-remote.$gs->{repo_id}.url";
1281 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1282 if ($orig_url && ($orig_url ne $gs->{url})) {
1283 die "$k already set: $orig_url\n",
1284 "wanted to set to: $gs->{url}\n";
1285 }
1286 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1287 my $remote_path = "$gs->{path}/$repo_path";
1288 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1289 $remote_path =~ s#/+#/#g;
1290 $remote_path =~ s#^/##g;
1291 $remote_path .= "/*" if $remote_path !~ /\*/;
1292 my ($n) = ($switch =~ /^--(\w+)/);
1293 if (length $pfx && $pfx !~ m#/$#) {
1294 die "--prefix='$pfx' must have a trailing slash '/'\n";
1295 }
1296 command_noisy('config',
1297 '--add',
1298 "svn-remote.$gs->{repo_id}.$n",
1299 "$remote_path:refs/remotes/$pfx*" .
1300 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1301 }
1303 sub verify_ref {
1304 my ($ref) = @_;
1305 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1306 { STDERR => 0 }); };
1307 }
1309 sub get_tree_from_treeish {
1310 my ($treeish) = @_;
1311 # $treeish can be a symbolic ref, too:
1312 my $type = command_oneline(qw/cat-file -t/, $treeish);
1313 my $expected;
1314 while ($type eq 'tag') {
1315 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1316 }
1317 if ($type eq 'commit') {
1318 $expected = (grep /^tree /, command(qw/cat-file commit/,
1319 $treeish))[0];
1320 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1321 die "Unable to get tree from $treeish\n" unless $expected;
1322 } elsif ($type eq 'tree') {
1323 $expected = $treeish;
1324 } else {
1325 die "$treeish is a $type, expected tree, tag or commit\n";
1326 }
1327 return $expected;
1328 }
1330 sub get_commit_entry {
1331 my ($treeish) = shift;
1332 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1333 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1334 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1335 open my $log_fh, '>', $commit_editmsg or croak $!;
1337 my $type = command_oneline(qw/cat-file -t/, $treeish);
1338 if ($type eq 'commit' || $type eq 'tag') {
1339 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1340 $type, $treeish);
1341 my $in_msg = 0;
1342 my $author;
1343 my $saw_from = 0;
1344 my $msgbuf = "";
1345 while (<$msg_fh>) {
1346 if (!$in_msg) {
1347 $in_msg = 1 if (/^\s*$/);
1348 $author = $1 if (/^author (.*>)/);
1349 } elsif (/^git-svn-id: /) {
1350 # skip this for now, we regenerate the
1351 # correct one on re-fetch anyways
1352 # TODO: set *:merge properties or like...
1353 } else {
1354 if (/^From:/ || /^Signed-off-by:/) {
1355 $saw_from = 1;
1356 }
1357 $msgbuf .= $_;
1358 }
1359 }
1360 $msgbuf =~ s/\s+$//s;
1361 if ($Git::SVN::_add_author_from && defined($author)
1362 && !$saw_from) {
1363 $msgbuf .= "\n\nFrom: $author";
1364 }
1365 print $log_fh $msgbuf or croak $!;
1366 command_close_pipe($msg_fh, $ctx);
1367 }
1368 close $log_fh or croak $!;
1370 if ($_edit || ($type eq 'tree')) {
1371 chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1372 system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1373 }
1374 rename $commit_editmsg, $commit_msg or croak $!;
1375 {
1376 require Encode;
1377 # SVN requires messages to be UTF-8 when entering the repo
1378 local $/;
1379 open $log_fh, '<', $commit_msg or croak $!;
1380 binmode $log_fh;
1381 chomp($log_entry{log} = <$log_fh>);
1383 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1384 my $msg = $log_entry{log};
1386 eval { $msg = Encode::decode($enc, $msg, 1) };
1387 if ($@) {
1388 die "Could not decode as $enc:\n", $msg,
1389 "\nPerhaps you need to set i18n.commitencoding\n";
1390 }
1392 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1393 die "Could not encode as UTF-8:\n$msg\n" if $@;
1395 $log_entry{log} = $msg;
1397 close $log_fh or croak $!;
1398 }
1399 unlink $commit_msg;
1400 \%log_entry;
1401 }
1403 sub s_to_file {
1404 my ($str, $file, $mode) = @_;
1405 open my $fd,'>',$file or croak $!;
1406 print $fd $str,"\n" or croak $!;
1407 close $fd or croak $!;
1408 chmod ($mode &~ umask, $file) if (defined $mode);
1409 }
1411 sub file_to_s {
1412 my $file = shift;
1413 open my $fd,'<',$file or croak "$!: file: $file\n";
1414 local $/;
1415 my $ret = <$fd>;
1416 close $fd or croak $!;
1417 $ret =~ s/\s*$//s;
1418 return $ret;
1419 }
1421 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1422 sub load_authors {
1423 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1424 my $log = $cmd eq 'log';
1425 while (<$authors>) {
1426 chomp;
1427 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1428 my ($user, $name, $email) = ($1, $2, $3);
1429 if ($log) {
1430 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1431 } else {
1432 $users{$user} = [$name, $email];
1433 }
1434 }
1435 close $authors or croak $!;
1436 }
1438 # convert GetOpt::Long specs for use by git-config
1439 sub read_git_config {
1440 my $opts = shift;
1441 my @config_only;
1442 foreach my $o (keys %$opts) {
1443 # if we have mixedCase and a long option-only, then
1444 # it's a config-only variable that we don't need for
1445 # the command-line.
1446 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1447 my $v = $opts->{$o};
1448 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1449 $key =~ s/-//g;
1450 my $arg = 'git config';
1451 $arg .= ' --int' if ($o =~ /[:=]i$/);
1452 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1453 if (ref $v eq 'ARRAY') {
1454 chomp(my @tmp = `$arg --get-all svn.$key`);
1455 @$v = @tmp if @tmp;
1456 } else {
1457 chomp(my $tmp = `$arg --get svn.$key`);
1458 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1459 $$v = $tmp;
1460 }
1461 }
1462 }
1463 delete @$opts{@config_only} if @config_only;
1464 }
1466 sub extract_metadata {
1467 my $id = shift or return (undef, undef, undef);
1468 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1469 \s([a-f\d\-]+)$/ix);
1470 if (!defined $rev || !$uuid || !$url) {
1471 # some of the original repositories I made had
1472 # identifiers like this:
1473 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1474 }
1475 return ($url, $rev, $uuid);
1476 }
1478 sub cmt_metadata {
1479 return extract_metadata((grep(/^git-svn-id: /,
1480 command(qw/cat-file commit/, shift)))[-1]);
1481 }
1483 sub cmt_sha2rev_batch {
1484 my %s2r;
1485 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1486 my $list = shift;
1488 foreach my $sha (@{$list}) {
1489 my $first = 1;
1490 my $size = 0;
1491 print $out $sha, "\n";
1493 while (my $line = <$in>) {
1494 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1495 last;
1496 } elsif ($first &&
1497 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1498 $first = 0;
1499 $size = $1;
1500 next;
1501 } elsif ($line =~ /^(git-svn-id: )/) {
1502 my (undef, $rev, undef) =
1503 extract_metadata($line);
1504 $s2r{$sha} = $rev;
1505 }
1507 $size -= length($line);
1508 last if ($size == 0);
1509 }
1510 }
1512 command_close_bidi_pipe($pid, $in, $out, $ctx);
1514 return \%s2r;
1515 }
1517 sub working_head_info {
1518 my ($head, $refs) = @_;
1519 my @args = qw/log --no-color --no-decorate --first-parent
1520 --pretty=medium/;
1521 my ($fh, $ctx) = command_output_pipe(@args, $head);
1522 my $hash;
1523 my %max;
1524 while (<$fh>) {
1525 if ( m{^commit ($::sha1)$} ) {
1526 unshift @$refs, $hash if $hash and $refs;
1527 $hash = $1;
1528 next;
1529 }
1530 next unless s{^\s*(git-svn-id:)}{$1};
1531 my ($url, $rev, $uuid) = extract_metadata($_);
1532 if (defined $url && defined $rev) {
1533 next if $max{$url} and $max{$url} < $rev;
1534 if (my $gs = Git::SVN->find_by_url($url)) {
1535 my $c = $gs->rev_map_get($rev, $uuid);
1536 if ($c && $c eq $hash) {
1537 close $fh; # break the pipe
1538 return ($url, $rev, $uuid, $gs);
1539 } else {
1540 $max{$url} ||= $gs->rev_map_max;
1541 }
1542 }
1543 }
1544 }
1545 command_close_pipe($fh, $ctx);
1546 (undef, undef, undef, undef);
1547 }
1549 sub read_commit_parents {
1550 my ($parents, $c) = @_;
1551 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1552 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1553 @{$parents->{$c}} = split(/ /, $p);
1554 }
1556 sub linearize_history {
1557 my ($gs, $refs) = @_;
1558 my %parents;
1559 foreach my $c (@$refs) {
1560 read_commit_parents(\%parents, $c);
1561 }
1563 my @linear_refs;
1564 my %skip = ();
1565 my $last_svn_commit = $gs->last_commit;
1566 foreach my $c (reverse @$refs) {
1567 next if $c eq $last_svn_commit;
1568 last if $skip{$c};
1570 unshift @linear_refs, $c;
1571 $skip{$c} = 1;
1573 # we only want the first parent to diff against for linear
1574 # history, we save the rest to inject when we finalize the
1575 # svn commit
1576 my $fp_a = verify_ref("$c~1");
1577 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1578 if (!$fp_a || !$fp_b) {
1579 die "Commit $c\n",
1580 "has no parent commit, and therefore ",
1581 "nothing to diff against.\n",
1582 "You should be working from a repository ",
1583 "originally created by git-svn\n";
1584 }
1585 if ($fp_a ne $fp_b) {
1586 die "$c~1 = $fp_a, however parsing commit $c ",
1587 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1588 }
1590 foreach my $p (@{$parents{$c}}) {
1591 $skip{$p} = 1;
1592 }
1593 }
1594 (\@linear_refs, \%parents);
1595 }
1597 sub find_file_type_and_diff_status {
1598 my ($path) = @_;
1599 return ('dir', '') if $path eq '';
1601 my $diff_output =
1602 command_oneline(qw(diff --cached --name-status --), $path) || "";
1603 my $diff_status = (split(' ', $diff_output))[0] || "";
1605 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1607 return (undef, undef) if !$diff_status && !$ls_tree;
1609 if ($diff_status eq "A") {
1610 return ("link", $diff_status) if -l $path;
1611 return ("dir", $diff_status) if -d $path;
1612 return ("file", $diff_status);
1613 }
1615 my $mode = (split(' ', $ls_tree))[0] || "";
1617 return ("link", $diff_status) if $mode eq "120000";
1618 return ("dir", $diff_status) if $mode eq "040000";
1619 return ("file", $diff_status);
1620 }
1622 sub md5sum {
1623 my $arg = shift;
1624 my $ref = ref $arg;
1625 my $md5 = Digest::MD5->new();
1626 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1627 $md5->addfile($arg) or croak $!;
1628 } elsif ($ref eq 'SCALAR') {
1629 $md5->add($$arg) or croak $!;
1630 } elsif (!$ref) {
1631 $md5->add($arg) or croak $!;
1632 } else {
1633 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1634 }
1635 return $md5->hexdigest();
1636 }
1638 sub gc_directory {
1639 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1640 my $out_filename = $_ . ".gz";
1641 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1642 binmode $in_fh;
1643 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1644 die "Unable to open $out_filename: $!\n";
1646 my $res;
1647 while ($res = sysread($in_fh, my $str, 1024)) {
1648 $gz->gzwrite($str) or
1649 die "Unable to write: ".$gz->gzerror()."!\n";
1650 }
1651 unlink $_ or die "unlink $File::Find::name: $!\n";
1652 } elsif (-f $_ && basename($_) eq "index") {
1653 unlink $_ or die "unlink $_: $!\n";
1654 }
1655 }
1657 package Git::SVN;
1658 use strict;
1659 use warnings;
1660 use Fcntl qw/:DEFAULT :seek/;
1661 use constant rev_map_fmt => 'NH40';
1662 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1663 $_repack $_repack_flags $_use_svm_props $_head
1664 $_use_svnsync_props $no_reuse_existing $_minimize_url
1665 $_use_log_author $_add_author_from $_localtime/;
1666 use Carp qw/croak/;
1667 use File::Path qw/mkpath/;
1668 use File::Copy qw/copy/;
1669 use IPC::Open3;
1670 use Memoize; # core since 5.8.0, Jul 2002
1671 use Memoize::Storable;
1673 my ($_gc_nr, $_gc_period);
1675 # properties that we do not log:
1676 my %SKIP_PROP;
1677 BEGIN {
1678 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1679 svn:special svn:executable
1680 svn:entry:committed-rev
1681 svn:entry:last-author
1682 svn:entry:uuid
1683 svn:entry:committed-date/;
1685 # some options are read globally, but can be overridden locally
1686 # per [svn-remote "..."] section. Command-line options will *NOT*
1687 # override options set in an [svn-remote "..."] section
1688 no strict 'refs';
1689 for my $option (qw/follow_parent no_metadata use_svm_props
1690 use_svnsync_props/) {
1691 my $key = $option;
1692 $key =~ tr/_//d;
1693 my $prop = "-$option";
1694 *$option = sub {
1695 my ($self) = @_;
1696 return $self->{$prop} if exists $self->{$prop};
1697 my $k = "svn-remote.$self->{repo_id}.$key";
1698 eval { command_oneline(qw/config --get/, $k) };
1699 if ($@) {
1700 $self->{$prop} = ${"Git::SVN::_$option"};
1701 } else {
1702 my $v = command_oneline(qw/config --bool/,$k);
1703 $self->{$prop} = $v eq 'false' ? 0 : 1;
1704 }
1705 return $self->{$prop};
1706 }
1707 }
1708 }
1711 my (%LOCKFILES, %INDEX_FILES);
1712 END {
1713 unlink keys %LOCKFILES if %LOCKFILES;
1714 unlink keys %INDEX_FILES if %INDEX_FILES;
1715 }
1717 sub resolve_local_globs {
1718 my ($url, $fetch, $glob_spec) = @_;
1719 return unless defined $glob_spec;
1720 my $ref = $glob_spec->{ref};
1721 my $path = $glob_spec->{path};
1722 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1723 next unless m#^$ref->{regex}$#;
1724 my $p = $1;
1725 my $pathname = desanitize_refname($path->full_path($p));
1726 my $refname = desanitize_refname($ref->full_path($p));
1727 if (my $existing = $fetch->{$pathname}) {
1728 if ($existing ne $refname) {
1729 die "Refspec conflict:\n",
1730 "existing: $existing\n",
1731 " globbed: $refname\n";
1732 }
1733 my $u = (::cmt_metadata("$refname"))[0];
1734 $u =~ s!^\Q$url\E(/|$)!! or die
1735 "$refname: '$url' not found in '$u'\n";
1736 if ($pathname ne $u) {
1737 warn "W: Refspec glob conflict ",
1738 "(ref: $refname):\n",
1739 "expected path: $pathname\n",
1740 " real path: $u\n",
1741 "Continuing ahead with $u\n";
1742 next;
1743 }
1744 } else {
1745 $fetch->{$pathname} = $refname;
1746 }
1747 }
1748 }
1750 sub parse_revision_argument {
1751 my ($base, $head) = @_;
1752 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1753 return ($base, $head);
1754 }
1755 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1756 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1757 return ($head, $head) if ($::_revision eq 'HEAD');
1758 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1759 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1760 die "revision argument: $::_revision not understood by git-svn\n";
1761 }
1763 sub fetch_all {
1764 my ($repo_id, $remotes) = @_;
1765 if (ref $repo_id) {
1766 my $gs = $repo_id;
1767 $repo_id = undef;
1768 $repo_id = $gs->{repo_id};
1769 }
1770 $remotes ||= read_all_remotes();
1771 my $remote = $remotes->{$repo_id} or
1772 die "[svn-remote \"$repo_id\"] unknown\n";
1773 my $fetch = $remote->{fetch};
1774 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1775 my (@gs, @globs);
1776 my $ra = Git::SVN::Ra->new($url);
1777 my $uuid = $ra->get_uuid;
1778 my $head = $ra->get_latest_revnum;
1780 # ignore errors, $head revision may not even exist anymore
1781 eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
1782 warn "W: $@\n" if $@;
1784 my $base = defined $fetch ? $head : 0;
1786 # read the max revs for wildcard expansion (branches/*, tags/*)
1787 foreach my $t (qw/branches tags/) {
1788 defined $remote->{$t} or next;
1789 push @globs, @{$remote->{$t}};
1791 my $max_rev = eval { tmp_config(qw/--int --get/,
1792 "svn-remote.$repo_id.${t}-maxRev") };
1793 if (defined $max_rev && ($max_rev < $base)) {
1794 $base = $max_rev;
1795 } elsif (!defined $max_rev) {
1796 $base = 0;
1797 }
1798 }
1800 if ($fetch) {
1801 foreach my $p (sort keys %$fetch) {
1802 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1803 my $lr = $gs->rev_map_max;
1804 if (defined $lr) {
1805 $base = $lr if ($lr < $base);
1806 }
1807 push @gs, $gs;
1808 }
1809 }
1811 ($base, $head) = parse_revision_argument($base, $head);
1812 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1813 }
1815 sub read_all_remotes {
1816 my $r = {};
1817 my $use_svm_props = eval { command_oneline(qw/config --bool
1818 svn.useSvmProps/) };
1819 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1820 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1821 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1822 if (m!^(.+)\.fetch=$svn_refspec$!) {
1823 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1824 die("svn-remote.$remote: remote ref '$remote_ref' "
1825 . "must start with 'refs/'\n")
1826 unless $remote_ref =~ m{^refs/};
1827 $local_ref = uri_decode($local_ref);
1828 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1829 $r->{$remote}->{svm} = {} if $use_svm_props;
1830 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1831 $r->{$1}->{svm} = {};
1832 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1833 $r->{$1}->{url} = $2;
1834 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1835 my ($remote, $t, $local_ref, $remote_ref) =
1836 ($1, $2, $3, $4);
1837 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1838 . "must start with 'refs/'\n")
1839 unless $remote_ref =~ m{^refs/};
1840 $local_ref = uri_decode($local_ref);
1841 my $rs = {
1842 t => $t,
1843 remote => $remote,
1844 path => Git::SVN::GlobSpec->new($local_ref, 1),
1845 ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
1846 if (length($rs->{ref}->{right}) != 0) {
1847 die "The '*' glob character must be the last ",
1848 "character of '$remote_ref'\n";
1849 }
1850 push @{ $r->{$remote}->{$t} }, $rs;
1851 }
1852 }
1854 map {
1855 if (defined $r->{$_}->{svm}) {
1856 my $svm;
1857 eval {
1858 my $section = "svn-remote.$_";
1859 $svm = {
1860 source => tmp_config('--get',
1861 "$section.svm-source"),
1862 replace => tmp_config('--get',
1863 "$section.svm-replace"),
1864 }
1865 };
1866 $r->{$_}->{svm} = $svm;
1867 }
1868 } keys %$r;
1870 $r;
1871 }
1873 sub init_vars {
1874 $_gc_nr = $_gc_period = 1000;
1875 if (defined $_repack || defined $_repack_flags) {
1876 warn "Repack options are obsolete; they have no effect.\n";
1877 }
1878 }
1880 sub verify_remotes_sanity {
1881 return unless -d $ENV{GIT_DIR};
1882 my %seen;
1883 foreach (command(qw/config -l/)) {
1884 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1885 if ($seen{$1}) {
1886 die "Remote ref refs/remote/$1 is tracked by",
1887 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1888 "Please resolve this ambiguity in ",
1889 "your git configuration file before ",
1890 "continuing\n";
1891 }
1892 $seen{$1} = $_;
1893 }
1894 }
1895 }
1897 sub find_existing_remote {
1898 my ($url, $remotes) = @_;
1899 return undef if $no_reuse_existing;
1900 my $existing;
1901 foreach my $repo_id (keys %$remotes) {
1902 my $u = $remotes->{$repo_id}->{url} or next;
1903 next if $u ne $url;
1904 $existing = $repo_id;
1905 last;
1906 }
1907 $existing;
1908 }
1910 sub init_remote_config {
1911 my ($self, $url, $no_write) = @_;
1912 $url =~ s!/+$!!; # strip trailing slash
1913 my $r = read_all_remotes();
1914 my $existing = find_existing_remote($url, $r);
1915 if ($existing) {
1916 unless ($no_write) {
1917 print STDERR "Using existing ",
1918 "[svn-remote \"$existing\"]\n";
1919 }
1920 $self->{repo_id} = $existing;
1921 } elsif ($_minimize_url) {
1922 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1923 $existing = find_existing_remote($min_url, $r);
1924 if ($existing) {
1925 unless ($no_write) {
1926 print STDERR "Using existing ",
1927 "[svn-remote \"$existing\"]\n";
1928 }
1929 $self->{repo_id} = $existing;
1930 }
1931 if ($min_url ne $url) {
1932 unless ($no_write) {
1933 print STDERR "Using higher level of URL: ",
1934 "$url => $min_url\n";
1935 }
1936 my $old_path = $self->{path};
1937 $self->{path} = $url;
1938 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1939 if (length $old_path) {
1940 $self->{path} .= "/$old_path";
1941 }
1942 $url = $min_url;
1943 }
1944 }
1945 my $orig_url;
1946 if (!$existing) {
1947 # verify that we aren't overwriting anything:
1948 $orig_url = eval {
1949 command_oneline('config', '--get',
1950 "svn-remote.$self->{repo_id}.url")
1951 };
1952 if ($orig_url && ($orig_url ne $url)) {
1953 die "svn-remote.$self->{repo_id}.url already set: ",
1954 "$orig_url\nwanted to set to: $url\n";
1955 }
1956 }
1957 my ($xrepo_id, $xpath) = find_ref($self->refname);
1958 if (!$no_write && defined $xpath) {
1959 die "svn-remote.$xrepo_id.fetch already set to track ",
1960 "$xpath:", $self->refname, "\n";
1961 }
1962 unless ($no_write) {
1963 command_noisy('config',
1964 "svn-remote.$self->{repo_id}.url", $url);
1965 $self->{path} =~ s{^/}{};
1966 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1967 command_noisy('config', '--add',
1968 "svn-remote.$self->{repo_id}.fetch",
1969 "$self->{path}:".$self->refname);
1970 }
1971 $self->{url} = $url;
1972 }
1974 sub find_by_url { # repos_root and, path are optional
1975 my ($class, $full_url, $repos_root, $path) = @_;
1977 return undef unless defined $full_url;
1978 remove_username($full_url);
1979 remove_username($repos_root) if defined $repos_root;
1980 my $remotes = read_all_remotes();
1981 if (defined $full_url && defined $repos_root && !defined $path) {
1982 $path = $full_url;
1983 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1984 }
1985 foreach my $repo_id (keys %$remotes) {
1986 my $u = $remotes->{$repo_id}->{url} or next;
1987 remove_username($u);
1988 next if defined $repos_root && $repos_root ne $u;
1990 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1991 foreach my $t (qw/branches tags/) {
1992 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1993 resolve_local_globs($u, $fetch, $globspec);
1994 }
1995 }
1996 my $p = $path;
1997 my $rwr = rewrite_root({repo_id => $repo_id});
1998 my $svm = $remotes->{$repo_id}->{svm}
1999 if defined $remotes->{$repo_id}->{svm};
2000 unless (defined $p) {
2001 $p = $full_url;
2002 my $z = $u;
2003 my $prefix = '';
2004 if ($rwr) {
2005 $z = $rwr;
2006 remove_username($z);
2007 } elsif (defined $svm) {
2008 $z = $svm->{source};
2009 $prefix = $svm->{replace};
2010 $prefix =~ s#^\Q$u\E(?:/|$)##;
2011 $prefix =~ s#/$##;
2012 }
2013 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2014 }
2015 foreach my $f (keys %$fetch) {
2016 next if $f ne $p;
2017 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2018 }
2019 }
2020 undef;
2021 }
2023 sub init {
2024 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2025 my $self = _new($class, $repo_id, $ref_id, $path);
2026 if (defined $url) {
2027 $self->init_remote_config($url, $no_write);
2028 }
2029 $self;
2030 }
2032 sub find_ref {
2033 my ($ref_id) = @_;
2034 foreach (command(qw/config -l/)) {
2035 next unless m!^svn-remote\.(.+)\.fetch=
2036 \s*(.*?)\s*:\s*(.+?)\s*$!x;
2037 my ($repo_id, $path, $ref) = ($1, $2, $3);
2038 if ($ref eq $ref_id) {
2039 $path = '' if ($path =~ m#^\./?#);
2040 return ($repo_id, $path);
2041 }
2042 }
2043 (undef, undef, undef);
2044 }
2046 sub new {
2047 my ($class, $ref_id, $repo_id, $path) = @_;
2048 if (defined $ref_id && !defined $repo_id && !defined $path) {
2049 ($repo_id, $path) = find_ref($ref_id);
2050 if (!defined $repo_id) {
2051 die "Could not find a \"svn-remote.*.fetch\" key ",
2052 "in the repository configuration matching: ",
2053 "$ref_id\n";
2054 }
2055 }
2056 my $self = _new($class, $repo_id, $ref_id, $path);
2057 if (!defined $self->{path} || !length $self->{path}) {
2058 my $fetch = command_oneline('config', '--get',
2059 "svn-remote.$repo_id.fetch",
2060 ":$ref_id\$") or
2061 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2062 "\":$ref_id\$\" in config\n";
2063 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2064 }
2065 $self->{path} =~ s{/+}{/}g;
2066 $self->{path} =~ s{\A/}{};
2067 $self->{path} =~ s{/\z}{};
2068 $self->{url} = command_oneline('config', '--get',
2069 "svn-remote.$repo_id.url") or
2070 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2071 $self->rebuild;
2072 $self;
2073 }
2075 sub refname {
2076 my ($refname) = $_[0]->{ref_id} ;
2078 # It cannot end with a slash /, we'll throw up on this because
2079 # SVN can't have directories with a slash in their name, either:
2080 if ($refname =~ m{/$}) {
2081 die "ref: '$refname' ends with a trailing slash, this is ",
2082 "not permitted by git nor Subversion\n";
2083 }
2085 # It cannot have ASCII control character space, tilde ~, caret ^,
2086 # colon :, question-mark ?, asterisk *, space, or open bracket [
2087 # anywhere.
2088 #
2089 # Additionally, % must be escaped because it is used for escaping
2090 # and we want our escaped refname to be reversible
2091 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2093 # no slash-separated component can begin with a dot .
2094 # /.* becomes /%2E*
2095 $refname =~ s{/\.}{/%2E}g;
2097 # It cannot have two consecutive dots .. anywhere
2098 # .. becomes %2E%2E
2099 $refname =~ s{\.\.}{%2E%2E}g;
2101 # trailing dots and .lock are not allowed
2102 # .$ becomes %2E and .lock becomes %2Elock
2103 $refname =~ s{\.(?=$|lock$)}{%2E};
2105 # the sequence @{ is used to access the reflog
2106 # @{ becomes %40{
2107 $refname =~ s{\@\{}{%40\{}g;
2109 return $refname;
2110 }
2112 sub desanitize_refname {
2113 my ($refname) = @_;
2114 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2115 return $refname;
2116 }
2118 sub svm_uuid {
2119 my ($self) = @_;
2120 return $self->{svm}->{uuid} if $self->svm;
2121 $self->ra;
2122 unless ($self->{svm}) {
2123 die "SVM UUID not cached, and reading remotely failed\n";
2124 }
2125 $self->{svm}->{uuid};
2126 }
2128 sub svm {
2129 my ($self) = @_;
2130 return $self->{svm} if $self->{svm};
2131 my $svm;
2132 # see if we have it in our config, first:
2133 eval {
2134 my $section = "svn-remote.$self->{repo_id}";
2135 $svm = {
2136 source => tmp_config('--get', "$section.svm-source"),
2137 uuid => tmp_config('--get', "$section.svm-uuid"),
2138 replace => tmp_config('--get', "$section.svm-replace"),
2139 }
2140 };
2141 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2142 $self->{svm} = $svm;
2143 }
2144 $self->{svm};
2145 }
2147 sub _set_svm_vars {
2148 my ($self, $ra) = @_;
2149 return $ra if $self->svm;
2151 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2152 "(svm:source, svm:uuid) ",
2153 "from the following URLs:\n" );
2154 sub read_svm_props {
2155 my ($self, $ra, $path, $r) = @_;
2156 my $props = ($ra->get_dir($path, $r))[2];
2157 my $src = $props->{'svm:source'};
2158 my $uuid = $props->{'svm:uuid'};
2159 return undef if (!$src || !$uuid);
2161 chomp($src, $uuid);
2163 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2164 or die "doesn't look right - svm:uuid is '$uuid'\n";
2166 # the '!' is used to mark the repos_root!/relative/path
2167 $src =~ s{/?!/?}{/};
2168 $src =~ s{/+$}{}; # no trailing slashes please
2169 # username is of no interest
2170 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2172 my $replace = $ra->{url};
2173 $replace .= "/$path" if length $path;
2175 my $section = "svn-remote.$self->{repo_id}";
2176 tmp_config("$section.svm-source", $src);
2177 tmp_config("$section.svm-replace", $replace);
2178 tmp_config("$section.svm-uuid", $uuid);
2179 $self->{svm} = {
2180 source => $src,
2181 uuid => $uuid,
2182 replace => $replace
2183 };
2184 }
2186 my $r = $ra->get_latest_revnum;
2187 my $path = $self->{path};
2188 my %tried;
2189 while (length $path) {
2190 unless ($tried{"$self->{url}/$path"}) {
2191 return $ra if $self->read_svm_props($ra, $path, $r);
2192 $tried{"$self->{url}/$path"} = 1;
2193 }
2194 $path =~ s#/?[^/]+$##;
2195 }
2196 die "Path: '$path' should be ''\n" if $path ne '';
2197 return $ra if $self->read_svm_props($ra, $path, $r);
2198 $tried{"$self->{url}/$path"} = 1;
2200 if ($ra->{repos_root} eq $self->{url}) {
2201 die @err, (map { " $_\n" } keys %tried), "\n";
2202 }
2204 # nope, make sure we're connected to the repository root:
2205 my $ok;
2206 my @tried_b;
2207 $path = $ra->{svn_path};
2208 $ra = Git::SVN::Ra->new($ra->{repos_root});
2209 while (length $path) {
2210 unless ($tried{"$ra->{url}/$path"}) {
2211 $ok = $self->read_svm_props($ra, $path, $r);
2212 last if $ok;
2213 $tried{"$ra->{url}/$path"} = 1;
2214 }
2215 $path =~ s#/?[^/]+$##;
2216 }
2217 die "Path: '$path' should be ''\n" if $path ne '';
2218 $ok ||= $self->read_svm_props($ra, $path, $r);
2219 $tried{"$ra->{url}/$path"} = 1;
2220 if (!$ok) {
2221 die @err, (map { " $_\n" } keys %tried), "\n";
2222 }
2223 Git::SVN::Ra->new($self->{url});
2224 }
2226 sub svnsync {
2227 my ($self) = @_;
2228 return $self->{svnsync} if $self->{svnsync};
2230 if ($self->no_metadata) {
2231 die "Can't have both 'noMetadata' and ",
2232 "'useSvnsyncProps' options set!\n";
2233 }
2234 if ($self->rewrite_root) {
2235 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2236 "options set!\n";
2237 }
2238 if ($self->rewrite_uuid) {
2239 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2240 "options set!\n";
2241 }
2243 my $svnsync;
2244 # see if we have it in our config, first:
2245 eval {
2246 my $section = "svn-remote.$self->{repo_id}";
2248 my $url = tmp_config('--get', "$section.svnsync-url");
2249 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2250 die "doesn't look right - svn:sync-from-url is '$url'\n";
2252 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2253 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2254 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2256 $svnsync = { url => $url, uuid => $uuid }
2257 };
2258 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2259 return $self->{svnsync} = $svnsync;
2260 }
2262 my $err = "useSvnsyncProps set, but failed to read " .
2263 "svnsync property: svn:sync-from-";
2264 my $rp = $self->ra->rev_proplist(0);
2266 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2267 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2268 die "doesn't look right - svn:sync-from-url is '$url'\n";
2270 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2271 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2272 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2274 my $section = "svn-remote.$self->{repo_id}";
2275 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2276 tmp_config('--add', "$section.svnsync-url", $url);
2277 return $self->{svnsync} = { url => $url, uuid => $uuid };
2278 }
2280 # this allows us to memoize our SVN::Ra UUID locally and avoid a
2281 # remote lookup (useful for 'git svn log').
2282 sub ra_uuid {
2283 my ($self) = @_;
2284 unless ($self->{ra_uuid}) {
2285 my $key = "svn-remote.$self->{repo_id}.uuid";
2286 my $uuid = eval { tmp_config('--get', $key) };
2287 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2288 $self->{ra_uuid} = $uuid;
2289 } else {
2290 die "ra_uuid called without URL\n" unless $self->{url};
2291 $self->{ra_uuid} = $self->ra->get_uuid;
2292 tmp_config('--add', $key, $self->{ra_uuid});
2293 }
2294 }
2295 $self->{ra_uuid};
2296 }
2298 sub _set_repos_root {
2299 my ($self, $repos_root) = @_;
2300 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2301 $repos_root ||= $self->ra->{repos_root};
2302 tmp_config($k, $repos_root);
2303 $repos_root;
2304 }
2306 sub repos_root {
2307 my ($self) = @_;
2308 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2309 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2310 }
2312 sub ra {
2313 my ($self) = shift;
2314 my $ra = Git::SVN::Ra->new($self->{url});
2315 $self->_set_repos_root($ra->{repos_root});
2316 if ($self->use_svm_props && !$self->{svm}) {
2317 if ($self->no_metadata) {
2318 die "Can't have both 'noMetadata' and ",
2319 "'useSvmProps' options set!\n";
2320 } elsif ($self->use_svnsync_props) {
2321 die "Can't have both 'useSvnsyncProps' and ",
2322 "'useSvmProps' options set!\n";
2323 }
2324 $ra = $self->_set_svm_vars($ra);
2325 $self->{-want_revprops} = 1;
2326 }
2327 $ra;
2328 }
2330 # prop_walk(PATH, REV, SUB)
2331 # -------------------------
2332 # Recursively traverse PATH at revision REV and invoke SUB for each
2333 # directory that contains a SVN property. SUB will be invoked as
2334 # follows: &SUB(gs, path, props); where `gs' is this instance of
2335 # Git::SVN, `path' the path to the directory where the properties
2336 # `props' were found. The `path' will be relative to point of checkout,
2337 # that is, if url://repo/trunk is the current Git branch, and that
2338 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
2339 # as `path' (note the trailing `/').
2340 sub prop_walk {
2341 my ($self, $path, $rev, $sub) = @_;
2343 $path =~ s#^/##;
2344 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2345 $path =~ s#^/*#/#g;
2346 my $p = $path;
2347 # Strip the irrelevant part of the path.
2348 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2349 # Ensure the path is terminated by a `/'.
2350 $p =~ s#/*$#/#;
2352 # The properties contain all the internal SVN stuff nobody
2353 # (usually) cares about.
2354 my $interesting_props = 0;
2355 foreach (keys %{$props}) {
2356 # If it doesn't start with `svn:', it must be a
2357 # user-defined property.
2358 ++$interesting_props and next if $_ !~ /^svn:/;
2359 # FIXME: Fragile, if SVN adds new public properties,
2360 # this needs to be updated.
2361 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2362 |eol-style|mime-type
2363 |externals|needs-lock)$/x;
2364 }
2365 &$sub($self, $p, $props) if $interesting_props;
2367 foreach (sort keys %$dirent) {
2368 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2369 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2370 }
2371 }
2373 sub last_rev { ($_[0]->last_rev_commit)[0] }
2374 sub last_commit { ($_[0]->last_rev_commit)[1] }
2376 # returns the newest SVN revision number and newest commit SHA1
2377 sub last_rev_commit {
2378 my ($self) = @_;
2379 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2380 return ($self->{last_rev}, $self->{last_commit});
2381 }
2382 my $c = ::verify_ref($self->refname.'^0');
2383 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2384 my $rev = (::cmt_metadata($c))[1];
2385 if (defined $rev) {
2386 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2387 return ($rev, $c);
2388 }
2389 }
2390 my $map_path = $self->map_path;
2391 unless (-e $map_path) {
2392 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2393 return (undef, undef);
2394 }
2395 my ($rev, $commit) = $self->rev_map_max(1);
2396 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2397 return ($rev, $commit);
2398 }
2400 sub get_fetch_range {
2401 my ($self, $min, $max) = @_;
2402 $max ||= $self->ra->get_latest_revnum;
2403 $min ||= $self->rev_map_max;
2404 (++$min, $max);
2405 }
2407 sub tmp_config {
2408 my (@args) = @_;
2409 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2410 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2411 if (! -f $config && -f $old_def_config) {
2412 rename $old_def_config, $config or
2413 die "Failed rename $old_def_config => $config: $!\n";
2414 }
2415 my $old_config = $ENV{GIT_CONFIG};
2416 $ENV{GIT_CONFIG} = $config;
2417 $@ = undef;
2418 my @ret = eval {
2419 unless (-f $config) {
2420 mkfile($config);
2421 open my $fh, '>', $config or
2422 die "Can't open $config: $!\n";
2423 print $fh "; This file is used internally by ",
2424 "git-svn\n" or die
2425 "Couldn't write to $config: $!\n";
2426 print $fh "; You should not have to edit it\n" or
2427 die "Couldn't write to $config: $!\n";
2428 close $fh or die "Couldn't close $config: $!\n";
2429 }
2430 command('config', @args);
2431 };
2432 my $err = $@;
2433 if (defined $old_config) {
2434 $ENV{GIT_CONFIG} = $old_config;
2435 } else {
2436 delete $ENV{GIT_CONFIG};
2437 }
2438 die $err if $err;
2439 wantarray ? @ret : $ret[0];
2440 }
2442 sub tmp_index_do {
2443 my ($self, $sub) = @_;
2444 my $old_index = $ENV{GIT_INDEX_FILE};
2445 $ENV{GIT_INDEX_FILE} = $self->{index};
2446 $@ = undef;
2447 my @ret = eval {
2448 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2449 mkpath([$dir]) unless -d $dir;
2450 &$sub;
2451 };
2452 my $err = $@;
2453 if (defined $old_index) {
2454 $ENV{GIT_INDEX_FILE} = $old_index;
2455 } else {
2456 delete $ENV{GIT_INDEX_FILE};
2457 }
2458 die $err if $err;
2459 wantarray ? @ret : $ret[0];
2460 }
2462 sub assert_index_clean {
2463 my ($self, $treeish) = @_;
2465 $self->tmp_index_do(sub {
2466 command_noisy('read-tree', $treeish) unless -e $self->{index};
2467 my $x = command_oneline('write-tree');
2468 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2469 /^tree ($::sha1)/mo);
2470 return if $y eq $x;
2472 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2473 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2474 command_noisy('read-tree', $treeish);
2475 $x = command_oneline('write-tree');
2476 if ($y ne $x) {
2477 ::fatal "trees ($treeish) $y != $x\n",
2478 "Something is seriously wrong...";
2479 }
2480 });
2481 }
2483 sub get_commit_parents {
2484 my ($self, $log_entry) = @_;
2485 my (%seen, @ret, @tmp);
2486 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2487 if (my $ip = $self->{inject_parents}) {
2488 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2489 push @tmp, $commit;
2490 }
2491 }
2492 if (my $cur = ::verify_ref($self->refname.'^0')) {
2493 push @tmp, $cur;
2494 }
2495 if (my $ipd = $self->{inject_parents_dcommit}) {
2496 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2497 push @tmp, @$commit;
2498 }
2499 }
2500 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2501 while (my $p = shift @tmp) {
2502 next if $seen{$p};
2503 $seen{$p} = 1;
2504 push @ret, $p;
2505 }
2506 @ret;
2507 }
2509 sub rewrite_root {
2510 my ($self) = @_;
2511 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2512 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2513 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2514 if ($rwr) {
2515 $rwr =~ s#/+$##;
2516 if ($rwr !~ m#^[a-z\+]+://#) {
2517 die "$rwr is not a valid URL (key: $k)\n";
2518 }
2519 }
2520 $self->{-rewrite_root} = $rwr;
2521 }
2523 sub rewrite_uuid {
2524 my ($self) = @_;
2525 return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2526 my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2527 my $rwid = eval { command_oneline(qw/config --get/, $k) };
2528 if ($rwid) {
2529 $rwid =~ s#/+$##;
2530 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2531 die "$rwid is not a valid UUID (key: $k)\n";
2532 }
2533 }
2534 $self->{-rewrite_uuid} = $rwid;
2535 }
2537 sub metadata_url {
2538 my ($self) = @_;
2539 ($self->rewrite_root || $self->{url}) .
2540 (length $self->{path} ? '/' . $self->{path} : '');
2541 }
2543 sub full_url {
2544 my ($self) = @_;
2545 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2546 }
2549 sub set_commit_header_env {
2550 my ($log_entry) = @_;
2551 my %env;
2552 foreach my $ned (qw/NAME EMAIL DATE/) {
2553 foreach my $ac (qw/AUTHOR COMMITTER/) {
2554 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2555 }
2556 }
2558 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2559 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2560 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2562 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2563 ? $log_entry->{commit_name}
2564 : $log_entry->{name};
2565 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2566 ? $log_entry->{commit_email}
2567 : $log_entry->{email};
2568 \%env;
2569 }
2571 sub restore_commit_header_env {
2572 my ($env) = @_;
2573 foreach my $ned (qw/NAME EMAIL DATE/) {
2574 foreach my $ac (qw/AUTHOR COMMITTER/) {
2575 my $k = "GIT_${ac}_${ned}";
2576 if (defined $env->{$k}) {
2577 $ENV{$k} = $env->{$k};
2578 } else {
2579 delete $ENV{$k};
2580 }
2581 }
2582 }
2583 }
2585 sub gc {
2586 command_noisy('gc', '--auto');
2587 };
2589 sub do_git_commit {
2590 my ($self, $log_entry) = @_;
2591 my $lr = $self->last_rev;
2592 if (defined $lr && $lr >= $log_entry->{revision}) {
2593 die "Last fetched revision of ", $self->refname,
2594 " was r$lr, but we are about to fetch: ",
2595 "r$log_entry->{revision}!\n";
2596 }
2597 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2598 croak "$log_entry->{revision} = $c already exists! ",
2599 "Why are we refetching it?\n";
2600 }
2601 my $old_env = set_commit_header_env($log_entry);
2602 my $tree = $log_entry->{tree};
2603 if (!defined $tree) {
2604 $tree = $self->tmp_index_do(sub {
2605 command_oneline('write-tree') });
2606 }
2607 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2609 my @exec = ('git', 'commit-tree', $tree);
2610 foreach ($self->get_commit_parents($log_entry)) {
2611 push @exec, '-p', $_;
2612 }
2613 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2614 or croak $!;
2615 binmode $msg_fh;
2617 # we always get UTF-8 from SVN, but we may want our commits in
2618 # a different encoding.
2619 if (my $enc = Git::config('i18n.commitencoding')) {
2620 require Encode;
2621 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2622 }
2623 print $msg_fh $log_entry->{log} or croak $!;
2624 restore_commit_header_env($old_env);
2625 unless ($self->no_metadata) {
2626 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2627 or croak $!;
2628 }
2629 $msg_fh->flush == 0 or croak $!;
2630 close $msg_fh or croak $!;
2631 chomp(my $commit = do { local $/; <$out_fh> });
2632 close $out_fh or croak $!;
2633 waitpid $pid, 0;
2634 croak $? if $?;
2635 if ($commit !~ /^$::sha1$/o) {
2636 die "Failed to commit, invalid sha1: $commit\n";
2637 }
2639 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2641 $self->{last_rev} = $log_entry->{revision};
2642 $self->{last_commit} = $commit;
2643 print "r$log_entry->{revision}" unless $::_q > 1;
2644 if (defined $log_entry->{svm_revision}) {
2645 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2646 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2647 0, $self->svm_uuid);
2648 }
2649 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2650 if (--$_gc_nr == 0) {
2651 $_gc_nr = $_gc_period;
2652 gc();
2653 }
2654 return $commit;
2655 }
2657 sub match_paths {
2658 my ($self, $paths, $r) = @_;
2659 return 1 if $self->{path} eq '';
2660 if (my $path = $paths->{"/$self->{path}"}) {
2661 return ($path->{action} eq 'D') ? 0 : 1;
2662 }
2663 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2664 if (grep /$self->{path_regex}/, keys %$paths) {
2665 return 1;
2666 }
2667 my $c = '';
2668 foreach (split m#/#, $self->{path}) {
2669 $c .= "/$_";
2670 next unless ($paths->{$c} &&
2671 ($paths->{$c}->{action} =~ /^[AR]$/));
2672 if ($self->ra->check_path($self->{path}, $r) ==
2673 $SVN::Node::dir) {
2674 return 1;
2675 }
2676 }
2677 return 0;
2678 }
2680 sub find_parent_branch {
2681 my ($self, $paths, $rev) = @_;
2682 return undef unless $self->follow_parent;
2683 unless (defined $paths) {
2684 my $err_handler = $SVN::Error::handler;
2685 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2686 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2687 sub { $paths = $_[0] });
2688 $SVN::Error::handler = $err_handler;
2689 }
2690 return undef unless defined $paths;
2692 # look for a parent from another branch:
2693 my @b_path_components = split m#/#, $self->{path};
2694 my @a_path_components;
2695 my $i;
2696 while (@b_path_components) {
2697 $i = $paths->{'/'.join('/', @b_path_components)};
2698 last if $i && defined $i->{copyfrom_path};
2699 unshift(@a_path_components, pop(@b_path_components));
2700 }
2701 return undef unless defined $i && defined $i->{copyfrom_path};
2702 my $branch_from = $i->{copyfrom_path};
2703 if (@a_path_components) {
2704 print STDERR "branch_from: $branch_from => ";
2705 $branch_from .= '/'.join('/', @a_path_components);
2706 print STDERR $branch_from, "\n";
2707 }
2708 my $r = $i->{copyfrom_rev};
2709 my $repos_root = $self->ra->{repos_root};
2710 my $url = $self->ra->{url};
2711 my $new_url = $url . $branch_from;
2712 print STDERR "Found possible branch point: ",
2713 "$new_url => ", $self->full_url, ", $r\n"
2714 unless $::_q > 1;
2715 $branch_from =~ s#^/##;
2716 my $gs = $self->other_gs($new_url, $url,
2717 $branch_from, $r, $self->{ref_id});
2718 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2719 {
2720 my ($base, $head);
2721 if (!defined $r0 || !defined $parent) {
2722 ($base, $head) = parse_revision_argument(0, $r);
2723 } else {
2724 if ($r0 < $r) {
2725 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2726 0, 1, sub { $base = $_[1] - 1 });
2727 }
2728 }
2729 if (defined $base && $base <= $r) {
2730 $gs->fetch($base, $r);
2731 }
2732 ($r0, $parent) = $gs->find_rev_before($r, 1);
2733 }
2734 if (defined $r0 && defined $parent) {
2735 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2736 unless $::_q > 1;
2737 my $ed;
2738 if ($self->ra->can_do_switch) {
2739 $self->assert_index_clean($parent);
2740 print STDERR "Following parent with do_switch\n"
2741 unless $::_q > 1;
2742 # do_switch works with svn/trunk >= r22312, but that
2743 # is not included with SVN 1.4.3 (the latest version
2744 # at the moment), so we can't rely on it
2745 $self->{last_rev} = $r0;
2746 $self->{last_commit} = $parent;
2747 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2748 $gs->ra->gs_do_switch($r0, $rev, $gs,
2749 $self->full_url, $ed)
2750 or die "SVN connection failed somewhere...\n";
2751 } elsif ($self->ra->trees_match($new_url, $r0,
2752 $self->full_url, $rev)) {
2753 print STDERR "Trees match:\n",
2754 " $new_url\@$r0\n",
2755 " ${\$self->full_url}\@$rev\n",
2756 "Following parent with no changes\n"
2757 unless $::_q > 1;
2758 $self->tmp_index_do(sub {
2759 command_noisy('read-tree', $parent);
2760 });
2761 $self->{last_commit} = $parent;
2762 } else {
2763 print STDERR "Following parent with do_update\n"
2764 unless $::_q > 1;
2765 $ed = SVN::Git::Fetcher->new($self);
2766 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2767 or die "SVN connection failed somewhere...\n";
2768 }
2769 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2770 return $self->make_log_entry($rev, [$parent], $ed);
2771 }
2772 return undef;
2773 }
2775 sub do_fetch {
2776 my ($self, $paths, $rev) = @_;
2777 my $ed;
2778 my ($last_rev, @parents);
2779 if (my $lc = $self->last_commit) {
2780 # we can have a branch that was deleted, then re-added
2781 # under the same name but copied from another path, in
2782 # which case we'll have multiple parents (we don't
2783 # want to break the original ref, nor lose copypath info):
2784 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2785 push @{$log_entry->{parents}}, $lc;
2786 return $log_entry;
2787 }
2788 $ed = SVN::Git::Fetcher->new($self);
2789 $last_rev = $self->{last_rev};
2790 $ed->{c} = $lc;
2791 @parents = ($lc);
2792 } else {
2793 $last_rev = $rev;
2794 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2795 return $log_entry;
2796 }
2797 $ed = SVN::Git::Fetcher->new($self);
2798 }
2799 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2800 die "SVN connection failed somewhere...\n";
2801 }
2802 $self->make_log_entry($rev, \@parents, $ed);
2803 }
2805 sub mkemptydirs {
2806 my ($self, $r) = @_;
2808 sub scan {
2809 my ($r, $empty_dirs, $line) = @_;
2810 if (defined $r && $line =~ /^r(\d+)$/) {
2811 return 0 if $1 > $r;
2812 } elsif ($line =~ /^ \+empty_dir: (.+)$/) {
2813 $empty_dirs->{$1} = 1;
2814 } elsif ($line =~ /^ \-empty_dir: (.+)$/) {
2815 my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
2816 delete @$empty_dirs{@d};
2817 }
2818 1; # continue
2819 };
2821 my %empty_dirs = ();
2822 my $gz_file = "$self->{dir}/unhandled.log.gz";
2823 if (-f $gz_file) {
2824 if (!$can_compress) {
2825 warn "Compress::Zlib could not be found; ",
2826 "empty directories in $gz_file will not be read\n";
2827 } else {
2828 my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
2829 die "Unable to open $gz_file: $!\n";
2830 my $line;
2831 while ($gz->gzreadline($line) > 0) {
2832 scan($r, \%empty_dirs, $line) or last;
2833 }
2834 $gz->gzclose;
2835 }
2836 }
2838 if (open my $fh, '<', "$self->{dir}/unhandled.log") {
2839 binmode $fh or croak "binmode: $!";
2840 while (<$fh>) {
2841 scan($r, \%empty_dirs, $_) or last;
2842 }
2843 close $fh;
2844 }
2846 my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2847 foreach my $d (sort keys %empty_dirs) {
2848 $d = uri_decode($d);
2849 $d =~ s/$strip//;
2850 next unless length($d);
2851 next if -d $d;
2852 if (-e $d) {
2853 warn "$d exists but is not a directory\n";
2854 } else {
2855 print "creating empty directory: $d\n";
2856 mkpath([$d]);
2857 }
2858 }
2859 }
2861 sub get_untracked {
2862 my ($self, $ed) = @_;
2863 my @out;
2864 my $h = $ed->{empty};
2865 foreach (sort keys %$h) {
2866 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2867 push @out, " $act: " . uri_encode($_);
2868 warn "W: $act: $_\n";
2869 }
2870 foreach my $t (qw/dir_prop file_prop/) {
2871 $h = $ed->{$t} or next;
2872 foreach my $path (sort keys %$h) {
2873 my $ppath = $path eq '' ? '.' : $path;
2874 foreach my $prop (sort keys %{$h->{$path}}) {
2875 next if $SKIP_PROP{$prop};
2876 my $v = $h->{$path}->{$prop};
2877 my $t_ppath_prop = "$t: " .
2878 uri_encode($ppath) . ' ' .
2879 uri_encode($prop);
2880 if (defined $v) {
2881 push @out, " +$t_ppath_prop " .
2882 uri_encode($v);
2883 } else {
2884 push @out, " -$t_ppath_prop";
2885 }
2886 }
2887 }
2888 }
2889 foreach my $t (qw/absent_file absent_directory/) {
2890 $h = $ed->{$t} or next;
2891 foreach my $parent (sort keys %$h) {
2892 foreach my $path (sort @{$h->{$parent}}) {
2893 push @out, " $t: " .
2894 uri_encode("$parent/$path");
2895 warn "W: $t: $parent/$path ",
2896 "Insufficient permissions?\n";
2897 }
2898 }
2899 }
2900 \@out;
2901 }
2903 # parse_svn_date(DATE)
2904 # --------------------
2905 # Given a date (in UTC) from Subversion, return a string in the format
2906 # "<TZ Offset> <local date/time>" that Git will use.
2907 #
2908 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
2909 # is true we'll convert it to the local timezone instead.
2910 sub parse_svn_date {
2911 my $date = shift || return '+0000 1970-01-01 00:00:00';
2912 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2913 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2914 croak "Unable to parse date: $date\n";
2915 my $parsed_date; # Set next.
2917 if ($Git::SVN::_localtime) {
2918 # Translate the Subversion datetime to an epoch time.
2919 # Begin by switching ourselves to $date's timezone, UTC.
2920 my $old_env_TZ = $ENV{TZ};
2921 $ENV{TZ} = 'UTC';
2923 my $epoch_in_UTC =
2924 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2926 # Determine our local timezone (including DST) at the
2927 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2928 # value of TZ, if any, at the time we were run.
2929 if (defined $Git::SVN::Log::TZ) {
2930 $ENV{TZ} = $Git::SVN::Log::TZ;
2931 } else {
2932 delete $ENV{TZ};
2933 }
2935 my $our_TZ =
2936 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2938 # This converts $epoch_in_UTC into our local timezone.
2939 my ($sec, $min, $hour, $mday, $mon, $year,
2940 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2942 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2943 $our_TZ, $year + 1900, $mon + 1,
2944 $mday, $hour, $min, $sec);
2946 # Reset us to the timezone in effect when we entered
2947 # this routine.
2948 if (defined $old_env_TZ) {
2949 $ENV{TZ} = $old_env_TZ;
2950 } else {
2951 delete $ENV{TZ};
2952 }
2953 } else {
2954 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2955 }
2957 return $parsed_date;
2958 }
2960 sub other_gs {
2961 my ($self, $new_url, $url,
2962 $branch_from, $r, $old_ref_id) = @_;
2963 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2964 unless ($gs) {
2965 my $ref_id = $old_ref_id;
2966 $ref_id =~ s/\@\d+-*$//;
2967 $ref_id .= "\@$r";
2968 # just grow a tail if we're not unique enough :x
2969 $ref_id .= '-' while find_ref($ref_id);
2970 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2971 if ($u =~ s#^\Q$url\E(/|$)##) {
2972 $p = $u;
2973 $u = $url;
2974 $repo_id = $self->{repo_id};
2975 }
2976 while (1) {
2977 # It is possible to tag two different subdirectories at
2978 # the same revision. If the url for an existing ref
2979 # does not match, we must either find a ref with a
2980 # matching url or create a new ref by growing a tail.
2981 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2982 my (undef, $max_commit) = $gs->rev_map_max(1);
2983 last if (!$max_commit);
2984 my ($url) = ::cmt_metadata($max_commit);
2985 last if ($url eq $gs->full_url);
2986 $ref_id .= '-';
2987 }
2988 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2989 }
2990 $gs
2991 }
2993 sub call_authors_prog {
2994 my ($orig_author) = @_;
2995 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2996 my $author = `$::_authors_prog $orig_author`;
2997 if ($? != 0) {
2998 die "$::_authors_prog failed with exit code $?\n"
2999 }
3000 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
3001 my ($name, $email) = ($1, $2);
3002 $email = undef if length $2 == 0;
3003 return [$name, $email];
3004 } else {
3005 die "Author: $orig_author: $::_authors_prog returned "
3006 . "invalid author format: $author\n";
3007 }
3008 }
3010 sub check_author {
3011 my ($author) = @_;
3012 if (!defined $author || length $author == 0) {
3013 $author = '(no author)';
3014 }
3015 if (!defined $::users{$author}) {
3016 if (defined $::_authors_prog) {
3017 $::users{$author} = call_authors_prog($author);
3018 } elsif (defined $::_authors) {
3019 die "Author: $author not defined in $::_authors file\n";
3020 }
3021 }
3022 $author;
3023 }
3025 sub find_extra_svk_parents {
3026 my ($self, $ed, $tickets, $parents) = @_;
3027 # aha! svk:merge property changed...
3028 my @tickets = split "\n", $tickets;
3029 my @known_parents;
3030 for my $ticket ( @tickets ) {
3031 my ($uuid, $path, $rev) = split /:/, $ticket;
3032 if ( $uuid eq $self->ra_uuid ) {
3033 my $url = $self->{url};
3034 my $repos_root = $url;
3035 my $branch_from = $path;
3036 $branch_from =~ s{^/}{};
3037 my $gs = $self->other_gs($repos_root."/".$branch_from,
3038 $url,
3039 $branch_from,
3040 $rev,
3041 $self->{ref_id});
3042 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3043 # wahey! we found it, but it might be
3044 # an old one (!)
3045 push @known_parents, [ $rev, $commit ];
3046 }
3047 }
3048 }
3049 # Ordering matters; highest-numbered commit merge tickets
3050 # first, as they may account for later merge ticket additions
3051 # or changes.
3052 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3053 for my $parent ( @known_parents ) {
3054 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3055 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3056 my $new;
3057 while ( <$msg_fh> ) {
3058 $new=1;last;
3059 }
3060 command_close_pipe($msg_fh, $ctx);
3061 if ( $new ) {
3062 print STDERR
3063 "Found merge parent (svk:merge ticket): $parent\n";
3064 push @$parents, $parent;
3065 }
3066 }
3067 }
3069 sub lookup_svn_merge {
3070 my $uuid = shift;
3071 my $url = shift;
3072 my $merge = shift;
3074 my ($source, $revs) = split ":", $merge;
3075 my $path = $source;
3076 $path =~ s{^/}{};
3077 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3078 if ( !$gs ) {
3079 warn "Couldn't find revmap for $url$source\n";
3080 return;
3081 }
3082 my @ranges = split ",", $revs;
3083 my ($tip, $tip_commit);
3084 my @merged_commit_ranges;
3085 # find the tip
3086 for my $range ( @ranges ) {
3087 my ($bottom, $top) = split "-", $range;
3088 $top ||= $bottom;
3089 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3090 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3092 unless ($top_commit and $bottom_commit) {
3093 warn "W:unknown path/rev in svn:mergeinfo "
3094 ."dirprop: $source:$range\n";
3095 next;
3096 }
3098 push @merged_commit_ranges,
3099 "$bottom_commit^..$top_commit";
3101 if ( !defined $tip or $top > $tip ) {
3102 $tip = $top;
3103 $tip_commit = $top_commit;
3104 }
3105 }
3106 return ($tip_commit, @merged_commit_ranges);
3107 }
3109 sub _rev_list {
3110 my ($msg_fh, $ctx) = command_output_pipe(
3111 "rev-list", @_,
3112 );
3113 my @rv;
3114 while ( <$msg_fh> ) {
3115 chomp;
3116 push @rv, $_;
3117 }
3118 command_close_pipe($msg_fh, $ctx);
3119 @rv;
3120 }
3122 sub check_cherry_pick {
3123 my $base = shift;
3124 my $tip = shift;
3125 my $parents = shift;
3126 my @ranges = @_;
3127 my %commits = map { $_ => 1 }
3128 _rev_list("--no-merges", $tip, "--not", $base, @$parents);
3129 for my $range ( @ranges ) {
3130 delete @commits{_rev_list($range)};
3131 }
3132 for my $commit (keys %commits) {
3133 if (has_no_changes($commit)) {
3134 delete $commits{$commit};
3135 }
3136 }
3137 return (keys %commits);
3138 }
3140 sub has_no_changes {
3141 my $commit = shift;
3143 my @revs = split / /, command_oneline(
3144 qw(rev-list --parents -1 -m), $commit);
3146 # Commits with no parents, e.g. the start of a partial branch,
3147 # have changes by definition.
3148 return 1 if (@revs < 2);
3150 # Commits with multiple parents, e.g a merge, have no changes
3151 # by definition.
3152 return 0 if (@revs > 2);
3154 return (command_oneline("rev-parse", "$commit^{tree}") eq
3155 command_oneline("rev-parse", "$commit~1^{tree}"));
3156 }
3158 # The GIT_DIR environment variable is not always set until after the command
3159 # line arguments are processed, so we can't memoize in a BEGIN block.
3160 {
3161 my $memoized = 0;
3163 sub memoize_svn_mergeinfo_functions {
3164 return if $memoized;
3165 $memoized = 1;
3167 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3168 mkpath([$cache_path]) unless -d $cache_path;
3170 tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3171 "$cache_path/lookup_svn_merge.db", 'nstore';
3172 memoize 'lookup_svn_merge',
3173 SCALAR_CACHE => 'FAULT',
3174 LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3175 ;
3177 tie my %check_cherry_pick_cache => 'Memoize::Storable',
3178 "$cache_path/check_cherry_pick.db", 'nstore';
3179 memoize 'check_cherry_pick',
3180 SCALAR_CACHE => 'FAULT',
3181 LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3182 ;
3184 tie my %has_no_changes_cache => 'Memoize::Storable',
3185 "$cache_path/has_no_changes.db", 'nstore';
3186 memoize 'has_no_changes',
3187 SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3188 LIST_CACHE => 'FAULT',
3189 ;
3190 }
3192 sub unmemoize_svn_mergeinfo_functions {
3193 return if not $memoized;
3194 $memoized = 0;
3196 Memoize::unmemoize 'lookup_svn_merge';
3197 Memoize::unmemoize 'check_cherry_pick';
3198 Memoize::unmemoize 'has_no_changes';
3199 }
3200 }
3202 END {
3203 # Force cache writeout explicitly instead of waiting for
3204 # global destruction to avoid segfault in Storable:
3205 # http://rt.cpan.org/Public/Bug/Display.html?id=36087
3206 unmemoize_svn_mergeinfo_functions();
3207 }
3209 sub parents_exclude {
3210 my $parents = shift;
3211 my @commits = @_;
3212 return unless @commits;
3214 my @excluded;
3215 my $excluded;
3216 do {
3217 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3218 $excluded = command_oneline(@cmd);
3219 if ( $excluded ) {
3220 my @new;
3221 my $found;
3222 for my $commit ( @commits ) {
3223 if ( $commit eq $excluded ) {
3224 push @excluded, $commit;
3225 $found++;
3226 last;
3227 }
3228 else {
3229 push @new, $commit;
3230 }
3231 }
3232 die "saw commit '$excluded' in rev-list output, "
3233 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3234 unless $found;
3235 @commits = @new;
3236 }
3237 }
3238 while ($excluded and @commits);
3240 return @excluded;
3241 }
3244 # note: this function should only be called if the various dirprops
3245 # have actually changed
3246 sub find_extra_svn_parents {
3247 my ($self, $ed, $mergeinfo, $parents) = @_;
3248 # aha! svk:merge property changed...
3250 memoize_svn_mergeinfo_functions();
3252 # We first search for merged tips which are not in our
3253 # history. Then, we figure out which git revisions are in
3254 # that tip, but not this revision. If all of those revisions
3255 # are now marked as merge, we can add the tip as a parent.
3256 my @merges = split "\n", $mergeinfo;
3257 my @merge_tips;
3258 my $url = $self->{url};
3259 my $uuid = $self->ra_uuid;
3260 my %ranges;
3261 for my $merge ( @merges ) {
3262 my ($tip_commit, @ranges) =
3263 lookup_svn_merge( $uuid, $url, $merge );
3264 unless (!$tip_commit or
3265 grep { $_ eq $tip_commit } @$parents ) {
3266 push @merge_tips, $tip_commit;
3267 $ranges{$tip_commit} = \@ranges;
3268 } else {
3269 push @merge_tips, undef;
3270 }
3271 }
3273 my %excluded = map { $_ => 1 }
3274 parents_exclude($parents, grep { defined } @merge_tips);
3276 # check merge tips for new parents
3277 my @new_parents;
3278 for my $merge_tip ( @merge_tips ) {
3279 my $spec = shift @merges;
3280 next unless $merge_tip and $excluded{$merge_tip};
3282 my $ranges = $ranges{$merge_tip};
3284 # check out 'new' tips
3285 my $merge_base;
3286 eval {
3287 $merge_base = command_oneline(
3288 "merge-base",
3289 @$parents, $merge_tip,
3290 );
3291 };
3292 if ($@) {
3293 die "An error occurred during merge-base"
3294 unless $@->isa("Git::Error::Command");
3296 warn "W: Cannot find common ancestor between ".
3297 "@$parents and $merge_tip. Ignoring merge info.\n";
3298 next;
3299 }
3301 # double check that there are no missing non-merge commits
3302 my (@incomplete) = check_cherry_pick(
3303 $merge_base, $merge_tip,
3304 $parents,
3305 @$ranges,
3306 );
3308 if ( @incomplete ) {
3309 warn "W:svn cherry-pick ignored ($spec) - missing "
3310 .@incomplete." commit(s) (eg $incomplete[0])\n";
3311 } else {
3312 warn
3313 "Found merge parent (svn:mergeinfo prop): ",
3314 $merge_tip, "\n";
3315 push @new_parents, $merge_tip;
3316 }
3317 }
3319 # cater for merges which merge commits from multiple branches
3320 if ( @new_parents > 1 ) {
3321 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3322 for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3323 next if $i == $j;
3324 next unless $new_parents[$i];
3325 next unless $new_parents[$j];
3326 my $revs = command_oneline(
3327 "rev-list", "-1",
3328 "$new_parents[$i]..$new_parents[$j]",
3329 );
3330 if ( !$revs ) {
3331 undef($new_parents[$j]);
3332 }
3333 }
3334 }
3335 }
3336 push @$parents, grep { defined } @new_parents;
3337 }
3339 sub make_log_entry {
3340 my ($self, $rev, $parents, $ed) = @_;
3341 my $untracked = $self->get_untracked($ed);
3343 my @parents = @$parents;
3344 my $ps = $ed->{path_strip} || "";
3345 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3346 my $props = $ed->{dir_prop}{$path};
3347 if ( $props->{"svk:merge"} ) {
3348 $self->find_extra_svk_parents
3349 ($ed, $props->{"svk:merge"}, \@parents);
3350 }
3351 if ( $props->{"svn:mergeinfo"} ) {
3352 $self->find_extra_svn_parents
3353 ($ed,
3354 $props->{"svn:mergeinfo"},
3355 \@parents);
3356 }
3357 }
3359 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3360 print $un "r$rev\n" or croak $!;
3361 print $un $_, "\n" foreach @$untracked;
3362 my %log_entry = ( parents => \@parents, revision => $rev,
3363 log => '');
3365 my $headrev;
3366 my $logged = delete $self->{logged_rev_props};
3367 if (!$logged || $self->{-want_revprops}) {
3368 my $rp = $self->ra->rev_proplist($rev);
3369 foreach (sort keys %$rp) {
3370 my $v = $rp->{$_};
3371 if (/^svn:(author|date|log)$/) {
3372 $log_entry{$1} = $v;
3373 } elsif ($_ eq 'svm:headrev') {
3374 $headrev = $v;
3375 } else {
3376 print $un " rev_prop: ", uri_encode($_), ' ',
3377 uri_encode($v), "\n";
3378 }
3379 }
3380 } else {
3381 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3382 }
3383 close $un or croak $!;
3385 $log_entry{date} = parse_svn_date($log_entry{date});
3386 $log_entry{log} .= "\n";
3387 my $author = $log_entry{author} = check_author($log_entry{author});
3388 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3389 : ($author, undef);
3391 my ($commit_name, $commit_email) = ($name, $email);
3392 if ($_use_log_author) {
3393 my $name_field;
3394 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3395 $name_field = $1;
3396 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3397 $name_field = $1;
3398 }
3399 if (!defined $name_field) {
3400 if (!defined $email) {
3401 $email = $name;
3402 }
3403 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3404 ($name, $email) = ($1, $2);
3405 } elsif ($name_field =~ /(.*)@/) {
3406 ($name, $email) = ($1, $name_field);
3407 } else {
3408 ($name, $email) = ($name_field, $name_field);
3409 }
3410 }
3411 if (defined $headrev && $self->use_svm_props) {
3412 if ($self->rewrite_root) {
3413 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3414 "options set!\n";
3415 }
3416 if ($self->rewrite_uuid) {
3417 die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3418 "options set!\n";
3419 }
3420 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3421 # we don't want "SVM: initializing mirror for junk" ...
3422 return undef if $r == 0;
3423 my $svm = $self->svm;
3424 if ($uuid ne $svm->{uuid}) {
3425 die "UUID mismatch on SVM path:\n",
3426 "expected: $svm->{uuid}\n",
3427 " got: $uuid\n";
3428 }
3429 my $full_url = $self->full_url;
3430 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3431 die "Failed to replace '$svm->{replace}' with ",
3432 "'$svm->{source}' in $full_url\n";
3433 # throw away username for storing in records
3434 remove_username($full_url);
3435 $log_entry{metadata} = "$full_url\@$r $uuid";
3436 $log_entry{svm_revision} = $r;
3437 $email ||= "$author\@$uuid";
3438 $commit_email ||= "$author\@$uuid";
3439 } elsif ($self->use_svnsync_props) {
3440 my $full_url = $self->svnsync->{url};
3441 $full_url .= "/$self->{path}" if length $self->{path};
3442 remove_username($full_url);
3443 my $uuid = $self->svnsync->{uuid};
3444 $log_entry{metadata} = "$full_url\@$rev $uuid";
3445 $email ||= "$author\@$uuid";
3446 $commit_email ||= "$author\@$uuid";
3447 } else {
3448 my $url = $self->metadata_url;
3449 remove_username($url);
3450 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3451 $log_entry{metadata} = "$url\@$rev " . $uuid;
3452 $email ||= "$author\@" . $uuid;
3453 $commit_email ||= "$author\@" . $uuid;
3454 }
3455 $log_entry{name} = $name;
3456 $log_entry{email} = $email;
3457 $log_entry{commit_name} = $commit_name;
3458 $log_entry{commit_email} = $commit_email;
3459 \%log_entry;
3460 }
3462 sub fetch {
3463 my ($self, $min_rev, $max_rev, @parents) = @_;
3464 my ($last_rev, $last_commit) = $self->last_rev_commit;
3465 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3466 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3467 }
3469 sub set_tree_cb {
3470 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3471 $self->{inject_parents} = { $rev => $tree };
3472 $self->fetch(undef, undef);
3473 }
3475 sub set_tree {
3476 my ($self, $tree) = (shift, shift);
3477 my $log_entry = ::get_commit_entry($tree);
3478 unless ($self->{last_rev}) {
3479 ::fatal("Must have an existing revision to commit");
3480 }
3481 my %ed_opts = ( r => $self->{last_rev},
3482 log => $log_entry->{log},
3483 ra => $self->ra,
3484 tree_a => $self->{last_commit},
3485 tree_b => $tree,
3486 editor_cb => sub {
3487 $self->set_tree_cb($log_entry, $tree, @_) },
3488 svn_path => $self->{path} );
3489 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3490 print "No changes\nr$self->{last_rev} = $tree\n";
3491 }
3492 }
3494 sub rebuild_from_rev_db {
3495 my ($self, $path) = @_;
3496 my $r = -1;
3497 open my $fh, '<', $path or croak "open: $!";
3498 binmode $fh or croak "binmode: $!";
3499 while (<$fh>) {
3500 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3501 chomp($_);
3502 ++$r;
3503 next if $_ eq ('0' x 40);
3504 $self->rev_map_set($r, $_);
3505 print "r$r = $_\n";
3506 }
3507 close $fh or croak "close: $!";
3508 unlink $path or croak "unlink: $!";
3509 }
3511 sub rebuild {
3512 my ($self) = @_;
3513 my $map_path = $self->map_path;
3514 my $partial = (-e $map_path && ! -z $map_path);
3515 return unless ::verify_ref($self->refname.'^0');
3516 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3517 my $rev_db = $self->rev_db_path;
3518 $self->rebuild_from_rev_db($rev_db);
3519 if ($self->use_svm_props) {
3520 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3521 $self->rebuild_from_rev_db($svm_rev_db);
3522 }
3523 $self->unlink_rev_db_symlink;
3524 return;
3525 }
3526 print "Rebuilding $map_path ...\n" if (!$partial);
3527 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3528 (undef, undef));
3529 my ($log, $ctx) =
3530 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3531 ($head ? "$head.." : "") . $self->refname,
3532 '--');
3533 my $metadata_url = $self->metadata_url;
3534 remove_username($metadata_url);
3535 my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3536 my $c;
3537 while (<$log>) {
3538 if ( m{^commit ($::sha1)$} ) {
3539 $c = $1;
3540 next;
3541 }
3542 next unless s{^\s*(git-svn-id:)}{$1};
3543 my ($url, $rev, $uuid) = ::extract_metadata($_);
3544 remove_username($url);
3546 # ignore merges (from set-tree)
3547 next if (!defined $rev || !$uuid);
3549 # if we merged or otherwise started elsewhere, this is
3550 # how we break out of it
3551 if (($uuid ne $svn_uuid) ||
3552 ($metadata_url && $url && ($url ne $metadata_url))) {
3553 next;
3554 }
3555 if ($partial && $head) {
3556 print "Partial-rebuilding $map_path ...\n";
3557 print "Currently at $base_rev = $head\n";
3558 $head = undef;
3559 }
3561 $self->rev_map_set($rev, $c);
3562 print "r$rev = $c\n";
3563 }
3564 command_close_pipe($log, $ctx);
3565 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3566 my $rev_db_path = $self->rev_db_path;
3567 if (-f $self->rev_db_path) {
3568 unlink $self->rev_db_path or croak "unlink: $!";
3569 }
3570 $self->unlink_rev_db_symlink;
3571 }
3573 # rev_map:
3574 # Tie::File seems to be prone to offset errors if revisions get sparse,
3575 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
3576 # one of my favorite modules is out :< Next up would be one of the DBM
3577 # modules, but I'm not sure which is most portable...
3578 #
3579 # This is the replacement for the rev_db format, which was too big
3580 # and inefficient for large repositories with a lot of sparse history
3581 # (mainly tags)
3582 #
3583 # The format is this:
3584 # - 24 bytes for every record,
3585 # * 4 bytes for the integer representing an SVN revision number
3586 # * 20 bytes representing the sha1 of a git commit
3587 # - No empty padding records like the old format
3588 # (except the last record, which can be overwritten)
3589 # - new records are written append-only since SVN revision numbers
3590 # increase monotonically
3591 # - lookups on SVN revision number are done via a binary search
3592 # - Piping the file to xxd -c24 is a good way of dumping it for
3593 # viewing or editing (piped back through xxd -r), should the need
3594 # ever arise.
3595 # - The last record can be padding revision with an all-zero sha1
3596 # This is used to optimize fetch performance when using multiple
3597 # "fetch" directives in .git/config
3598 #
3599 # These files are disposable unless noMetadata or useSvmProps is set
3601 sub _rev_map_set {
3602 my ($fh, $rev, $commit) = @_;
3604 binmode $fh or croak "binmode: $!";
3605 my $size = (stat($fh))[7];
3606 ($size % 24) == 0 or croak "inconsistent size: $size";
3608 my $wr_offset = 0;
3609 if ($size > 0) {
3610 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3611 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3612 $read == 24 or croak "read only $read bytes (!= 24)";
3613 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3614 if ($last_commit eq ('0' x40)) {
3615 if ($size >= 48) {
3616 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3617 $read = sysread($fh, $buf, 24) or
3618 croak "read: $!";
3619 $read == 24 or
3620 croak "read only $read bytes (!= 24)";
3621 ($last_rev, $last_commit) =
3622 unpack(rev_map_fmt, $buf);
3623 if ($last_commit eq ('0' x40)) {
3624 croak "inconsistent .rev_map\n";
3625 }
3626 }
3627 if ($last_rev >= $rev) {
3628 croak "last_rev is higher!: $last_rev >= $rev";
3629 }
3630 $wr_offset = -24;
3631 }
3632 }
3633 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3634 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3635 croak "write: $!";
3636 }
3638 sub _rev_map_reset {
3639 my ($fh, $rev, $commit) = @_;
3640 my $c = _rev_map_get($fh, $rev);
3641 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3642 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3643 truncate $fh, $offset or croak "truncate: $!";
3644 }
3646 sub mkfile {
3647 my ($path) = @_;
3648 unless (-e $path) {
3649 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3650 mkpath([$dir]) unless -d $dir;
3651 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3652 close $fh or die "Couldn't close (create) $path: $!\n";
3653 }
3654 }
3656 sub rev_map_set {
3657 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3658 defined $commit or die "missing arg3\n";
3659 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3660 my $db = $self->map_path($uuid);
3661 my $db_lock = "$db.lock";
3662 my $sig;
3663 $update_ref ||= 0;
3664 if ($update_ref) {
3665 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3666 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3667 }
3668 mkfile($db);
3670 $LOCKFILES{$db_lock} = 1;
3671 my $sync;
3672 # both of these options make our .rev_db file very, very important
3673 # and we can't afford to lose it because rebuild() won't work
3674 if ($self->use_svm_props || $self->no_metadata) {
3675 $sync = 1;
3676 copy($db, $db_lock) or die "rev_map_set(@_): ",
3677 "Failed to copy: ",
3678 "$db => $db_lock ($!)\n";
3679 } else {
3680 rename $db, $db_lock or die "rev_map_set(@_): ",
3681 "Failed to rename: ",
3682 "$db => $db_lock ($!)\n";
3683 }
3685 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3686 or croak "Couldn't open $db_lock: $!\n";
3687 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3688 _rev_map_set($fh, $rev, $commit);
3689 if ($sync) {
3690 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3691 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3692 }
3693 close $fh or croak $!;
3694 if ($update_ref) {
3695 $_head = $self;
3696 my $note = "";
3697 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3698 command_noisy('update-ref', '-m', "r$rev$note",
3699 $self->refname, $commit);
3700 }
3701 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3702 "$db_lock => $db ($!)\n";
3703 delete $LOCKFILES{$db_lock};
3704 if ($update_ref) {
3705 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3706 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3707 kill $sig, $$ if defined $sig;
3708 }
3709 }
3711 # If want_commit, this will return an array of (rev, commit) where
3712 # commit _must_ be a valid commit in the archive.
3713 # Otherwise, it'll return the max revision (whether or not the
3714 # commit is valid or just a 0x40 placeholder).
3715 sub rev_map_max {
3716 my ($self, $want_commit) = @_;
3717 $self->rebuild;
3718 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3719 $want_commit ? ($r, $c) : $r;
3720 }
3722 sub rev_map_max_norebuild {
3723 my ($self, $want_commit) = @_;
3724 my $map_path = $self->map_path;
3725 stat $map_path or return $want_commit ? (0, undef) : 0;
3726 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3727 binmode $fh or croak "binmode: $!";
3728 my $size = (stat($fh))[7];
3729 ($size % 24) == 0 or croak "inconsistent size: $size";
3731 if ($size == 0) {
3732 close $fh or croak "close: $!";
3733 return $want_commit ? (0, undef) : 0;
3734 }
3736 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3737 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3738 my ($r, $c) = unpack(rev_map_fmt, $buf);
3739 if ($want_commit && $c eq ('0' x40)) {
3740 if ($size < 48) {
3741 return $want_commit ? (0, undef) : 0;
3742 }
3743 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3744 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3745 ($r, $c) = unpack(rev_map_fmt, $buf);
3746 if ($c eq ('0'x40)) {
3747 croak "Penultimate record is all-zeroes in $map_path";
3748 }
3749 }
3750 close $fh or croak "close: $!";
3751 $want_commit ? ($r, $c) : $r;
3752 }
3754 sub rev_map_get {
3755 my ($self, $rev, $uuid) = @_;
3756 my $map_path = $self->map_path($uuid);
3757 return undef unless -e $map_path;
3759 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3760 my $c = _rev_map_get($fh, $rev);
3761 close($fh) or croak "close: $!";
3762 $c
3763 }
3765 sub _rev_map_get {
3766 my ($fh, $rev) = @_;
3768 binmode $fh or croak "binmode: $!";
3769 my $size = (stat($fh))[7];
3770 ($size % 24) == 0 or croak "inconsistent size: $size";
3772 if ($size == 0) {
3773 return undef;
3774 }
3776 my ($l, $u) = (0, $size - 24);
3777 my ($r, $c, $buf);
3779 while ($l <= $u) {
3780 my $i = int(($l/24 + $u/24) / 2) * 24;
3781 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3782 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3783 my ($r, $c) = unpack(rev_map_fmt, $buf);
3785 if ($r < $rev) {
3786 $l = $i + 24;
3787 } elsif ($r > $rev) {
3788 $u = $i - 24;
3789 } else { # $r == $rev
3790 return $c eq ('0' x 40) ? undef : $c;
3791 }
3792 }
3793 undef;
3794 }
3796 # Finds the first svn revision that exists on (if $eq_ok is true) or
3797 # before $rev for the current branch. It will not search any lower
3798 # than $min_rev. Returns the git commit hash and svn revision number
3799 # if found, else (undef, undef).
3800 sub find_rev_before {
3801 my ($self, $rev, $eq_ok, $min_rev) = @_;
3802 --$rev unless $eq_ok;
3803 $min_rev ||= 1;
3804 my $max_rev = $self->rev_map_max;
3805 $rev = $max_rev if ($rev > $max_rev);
3806 while ($rev >= $min_rev) {
3807 if (my $c = $self->rev_map_get($rev)) {
3808 return ($rev, $c);
3809 }
3810 --$rev;
3811 }
3812 return (undef, undef);
3813 }
3815 # Finds the first svn revision that exists on (if $eq_ok is true) or
3816 # after $rev for the current branch. It will not search any higher
3817 # than $max_rev. Returns the git commit hash and svn revision number
3818 # if found, else (undef, undef).
3819 sub find_rev_after {
3820 my ($self, $rev, $eq_ok, $max_rev) = @_;
3821 ++$rev unless $eq_ok;
3822 $max_rev ||= $self->rev_map_max;
3823 while ($rev <= $max_rev) {
3824 if (my $c = $self->rev_map_get($rev)) {
3825 return ($rev, $c);
3826 }
3827 ++$rev;
3828 }
3829 return (undef, undef);
3830 }
3832 sub _new {
3833 my ($class, $repo_id, $ref_id, $path) = @_;
3834 unless (defined $repo_id && length $repo_id) {
3835 $repo_id = $Git::SVN::default_repo_id;
3836 }
3837 unless (defined $ref_id && length $ref_id) {
3838 $_prefix = '' unless defined($_prefix);
3839 $_[2] = $ref_id =
3840 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3841 }
3842 $_[1] = $repo_id;
3843 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3845 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3846 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3847 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3848 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3849 if (-d $old_dir && ! -d $dir) {
3850 $dir = $old_dir;
3851 }
3852 }
3854 $_[3] = $path = '' unless (defined $path);
3855 mkpath([$dir]);
3856 bless {
3857 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3858 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3859 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3860 }
3862 # for read-only access of old .rev_db formats
3863 sub unlink_rev_db_symlink {
3864 my ($self) = @_;
3865 my $link = $self->rev_db_path;
3866 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3867 if (-l $link) {
3868 unlink $link or croak "unlink: $link failed!";
3869 }
3870 }
3872 sub rev_db_path {
3873 my ($self, $uuid) = @_;
3874 my $db_path = $self->map_path($uuid);
3875 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3876 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3877 $db_path;
3878 }
3880 # the new replacement for .rev_db
3881 sub map_path {
3882 my ($self, $uuid) = @_;
3883 $uuid ||= $self->ra_uuid;
3884 "$self->{map_root}.$uuid";
3885 }
3887 sub uri_encode {
3888 my ($f) = @_;
3889 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3890 $f
3891 }
3893 sub uri_decode {
3894 my ($f) = @_;
3895 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3896 $f
3897 }
3899 sub remove_username {
3900 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3901 }
3903 package Git::SVN::Prompt;
3904 use strict;
3905 use warnings;
3906 require SVN::Core;
3907 use vars qw/$_no_auth_cache $_username/;
3909 sub simple {
3910 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3911 $may_save = undef if $_no_auth_cache;
3912 $default_username = $_username if defined $_username;
3913 if (defined $default_username && length $default_username) {
3914 if (defined $realm && length $realm) {
3915 print STDERR "Authentication realm: $realm\n";
3916 STDERR->flush;
3917 }
3918 $cred->username($default_username);
3919 } else {
3920 username($cred, $realm, $may_save, $pool);
3921 }
3922 $cred->password(_read_password("Password for '" .
3923 $cred->username . "': ", $realm));
3924 $cred->may_save($may_save);
3925 $SVN::_Core::SVN_NO_ERROR;
3926 }
3928 sub ssl_server_trust {
3929 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3930 $may_save = undef if $_no_auth_cache;
3931 print STDERR "Error validating server certificate for '$realm':\n";
3932 {
3933 no warnings 'once';
3934 # All variables SVN::Auth::SSL::* are used only once,
3935 # so we're shutting up Perl warnings about this.
3936 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3937 print STDERR " - The certificate is not issued ",
3938 "by a trusted authority. Use the\n",
3939 " fingerprint to validate ",
3940 "the certificate manually!\n";
3941 }
3942 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3943 print STDERR " - The certificate hostname ",
3944 "does not match.\n";
3945 }
3946 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3947 print STDERR " - The certificate is not yet valid.\n";
3948 }
3949 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3950 print STDERR " - The certificate has expired.\n";
3951 }
3952 if ($failures & $SVN::Auth::SSL::OTHER) {
3953 print STDERR " - The certificate has ",
3954 "an unknown error.\n";
3955 }
3956 } # no warnings 'once'
3957 printf STDERR
3958 "Certificate information:\n".
3959 " - Hostname: %s\n".
3960 " - Valid: from %s until %s\n".
3961 " - Issuer: %s\n".
3962 " - Fingerprint: %s\n",
3963 map $cert_info->$_, qw(hostname valid_from valid_until
3964 issuer_dname fingerprint);
3965 my $choice;
3966 prompt:
3967 print STDERR $may_save ?
3968 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3969 "(R)eject or accept (t)emporarily? ";
3970 STDERR->flush;
3971 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3972 if ($choice =~ /^t$/i) {
3973 $cred->may_save(undef);
3974 } elsif ($choice =~ /^r$/i) {
3975 return -1;
3976 } elsif ($may_save && $choice =~ /^p$/i) {
3977 $cred->may_save($may_save);
3978 } else {
3979 goto prompt;
3980 }
3981 $cred->accepted_failures($failures);
3982 $SVN::_Core::SVN_NO_ERROR;
3983 }
3985 sub ssl_client_cert {
3986 my ($cred, $realm, $may_save, $pool) = @_;
3987 $may_save = undef if $_no_auth_cache;
3988 print STDERR "Client certificate filename: ";
3989 STDERR->flush;
3990 chomp(my $filename = <STDIN>);
3991 $cred->cert_file($filename);
3992 $cred->may_save($may_save);
3993 $SVN::_Core::SVN_NO_ERROR;
3994 }
3996 sub ssl_client_cert_pw {
3997 my ($cred, $realm, $may_save, $pool) = @_;
3998 $may_save = undef if $_no_auth_cache;
3999 $cred->password(_read_password("Password: ", $realm));
4000 $cred->may_save($may_save);
4001 $SVN::_Core::SVN_NO_ERROR;
4002 }
4004 sub username {
4005 my ($cred, $realm, $may_save, $pool) = @_;
4006 $may_save = undef if $_no_auth_cache;
4007 if (defined $realm && length $realm) {
4008 print STDERR "Authentication realm: $realm\n";
4009 }
4010 my $username;
4011 if (defined $_username) {
4012 $username = $_username;
4013 } else {
4014 print STDERR "Username: ";
4015 STDERR->flush;
4016 chomp($username = <STDIN>);
4017 }
4018 $cred->username($username);
4019 $cred->may_save($may_save);
4020 $SVN::_Core::SVN_NO_ERROR;
4021 }
4023 sub _read_password {
4024 my ($prompt, $realm) = @_;
4025 my $password = '';
4026 if (exists $ENV{GIT_ASKPASS}) {
4027 open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
4028 $password = <PH>;
4029 $password =~ s/[\012\015]//; # \n\r
4030 close(PH);
4031 } else {
4032 print STDERR $prompt;
4033 STDERR->flush;
4034 require Term::ReadKey;
4035 Term::ReadKey::ReadMode('noecho');
4036 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
4037 last if $key =~ /[\012\015]/; # \n\r
4038 $password .= $key;
4039 }
4040 Term::ReadKey::ReadMode('restore');
4041 print STDERR "\n";
4042 STDERR->flush;
4043 }
4044 $password;
4045 }
4047 package SVN::Git::Fetcher;
4048 use vars qw/@ISA/;
4049 use strict;
4050 use warnings;
4051 use Carp qw/croak/;
4052 use IO::File qw//;
4053 use vars qw/$_ignore_regex/;
4055 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
4056 sub new {
4057 my ($class, $git_svn, $switch_path) = @_;
4058 my $self = SVN::Delta::Editor->new;
4059 bless $self, $class;
4060 if (exists $git_svn->{last_commit}) {
4061 $self->{c} = $git_svn->{last_commit};
4062 $self->{empty_symlinks} =
4063 _mark_empty_symlinks($git_svn, $switch_path);
4064 }
4065 $self->{ignore_regex} = eval { command_oneline('config', '--get',
4066 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
4067 $self->{empty} = {};
4068 $self->{dir_prop} = {};
4069 $self->{file_prop} = {};
4070 $self->{absent_dir} = {};
4071 $self->{absent_file} = {};
4072 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4073 $self->{pathnameencoding} = Git::config('svn.pathnameencoding');
4074 $self;
4075 }
4077 # this uses the Ra object, so it must be called before do_{switch,update},
4078 # not inside them (when the Git::SVN::Fetcher object is passed) to
4079 # do_{switch,update}
4080 sub _mark_empty_symlinks {
4081 my ($git_svn, $switch_path) = @_;
4082 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4083 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4085 my %ret;
4086 my ($rev, $cmt) = $git_svn->last_rev_commit;
4087 return {} unless ($rev && $cmt);
4089 # allow the warning to be printed for each revision we fetch to
4090 # ensure the user sees it. The user can also disable the workaround
4091 # on the repository even while git svn is running and the next
4092 # revision fetched will skip this expensive function.
4093 my $printed_warning;
4094 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4095 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4096 local $/ = "\0";
4097 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4098 $pfx .= '/' if length($pfx);
4099 while (<$ls>) {
4100 chomp;
4101 s/\A100644 blob $empty_blob\t//o or next;
4102 unless ($printed_warning) {
4103 print STDERR "Scanning for empty symlinks, ",
4104 "this may take a while if you have ",
4105 "many empty files\n",
4106 "You may disable this with `",
4107 "git config svn.brokenSymlinkWorkaround ",
4108 "false'.\n",
4109 "This may be done in a different ",
4110 "terminal without restarting ",
4111 "git svn\n";
4112 $printed_warning = 1;
4113 }
4114 my $path = $_;
4115 my (undef, $props) =
4116 $git_svn->ra->get_file($pfx.$path, $rev, undef);
4117 if ($props->{'svn:special'}) {
4118 $ret{$path} = 1;
4119 }
4120 }
4121 command_close_pipe($ls, $ctx);
4122 \%ret;
4123 }
4125 # returns true if a given path is inside a ".git" directory
4126 sub in_dot_git {
4127 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4128 }
4130 # return value: 0 -- don't ignore, 1 -- ignore
4131 sub is_path_ignored {
4132 my ($self, $path) = @_;
4133 return 1 if in_dot_git($path);
4134 return 1 if defined($self->{ignore_regex}) &&
4135 $path =~ m!$self->{ignore_regex}!;
4136 return 0 unless defined($_ignore_regex);
4137 return 1 if $path =~ m!$_ignore_regex!o;
4138 return 0;
4139 }
4141 sub set_path_strip {
4142 my ($self, $path) = @_;
4143 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4144 }
4146 sub open_root {
4147 { path => '' };
4148 }
4150 sub open_directory {
4151 my ($self, $path, $pb, $rev) = @_;
4152 { path => $path };
4153 }
4155 sub git_path {
4156 my ($self, $path) = @_;
4157 if (my $enc = $self->{pathnameencoding}) {
4158 require Encode;
4159 Encode::from_to($path, 'UTF-8', $enc);
4160 }
4161 if ($self->{path_strip}) {
4162 $path =~ s!$self->{path_strip}!! or
4163 die "Failed to strip path '$path' ($self->{path_strip})\n";
4164 }
4165 $path;
4166 }
4168 sub delete_entry {
4169 my ($self, $path, $rev, $pb) = @_;
4170 return undef if $self->is_path_ignored($path);
4172 my $gpath = $self->git_path($path);
4173 return undef if ($gpath eq '');
4175 # remove entire directories.
4176 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4177 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4178 if ($tree) {
4179 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4180 -r --name-only -z/,
4181 $tree);
4182 local $/ = "\0";
4183 while (<$ls>) {
4184 chomp;
4185 my $rmpath = "$gpath/$_";
4186 $self->{gii}->remove($rmpath);
4187 print "\tD\t$rmpath\n" unless $::_q;
4188 }
4189 print "\tD\t$gpath/\n" unless $::_q;
4190 command_close_pipe($ls, $ctx);
4191 } else {
4192 $self->{gii}->remove($gpath);
4193 print "\tD\t$gpath\n" unless $::_q;
4194 }
4195 $self->{empty}->{$path} = 0;
4196 undef;
4197 }
4199 sub open_file {
4200 my ($self, $path, $pb, $rev) = @_;
4201 my ($mode, $blob);
4203 goto out if $self->is_path_ignored($path);
4205 my $gpath = $self->git_path($path);
4206 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4207 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4208 unless (defined $mode && defined $blob) {
4209 die "$path was not found in commit $self->{c} (r$rev)\n";
4210 }
4211 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4212 $mode = '120000';
4213 }
4214 out:
4215 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4216 pool => SVN::Pool->new, action => 'M' };
4217 }
4219 sub add_file {
4220 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4221 my $mode;
4223 if (!$self->is_path_ignored($path)) {
4224 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4225 delete $self->{empty}->{$dir};
4226 $mode = '100644';
4227 }
4228 { path => $path, mode_a => $mode, mode_b => $mode,
4229 pool => SVN::Pool->new, action => 'A' };
4230 }
4232 sub add_directory {
4233 my ($self, $path, $cp_path, $cp_rev) = @_;
4234 goto out if $self->is_path_ignored($path);
4235 my $gpath = $self->git_path($path);
4236 if ($gpath eq '') {
4237 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4238 -r --name-only -z/,
4239 $self->{c});
4240 local $/ = "\0";
4241 while (<$ls>) {
4242 chomp;
4243 $self->{gii}->remove($_);
4244 print "\tD\t$_\n" unless $::_q;
4245 }
4246 command_close_pipe($ls, $ctx);
4247 $self->{empty}->{$path} = 0;
4248 }
4249 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4250 delete $self->{empty}->{$dir};
4251 $self->{empty}->{$path} = 1;
4252 out:
4253 { path => $path };
4254 }
4256 sub change_dir_prop {
4257 my ($self, $db, $prop, $value) = @_;
4258 return undef if $self->is_path_ignored($db->{path});
4259 $self->{dir_prop}->{$db->{path}} ||= {};
4260 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4261 undef;
4262 }
4264 sub absent_directory {
4265 my ($self, $path, $pb) = @_;
4266 return undef if $self->is_path_ignored($path);
4267 $self->{absent_dir}->{$pb->{path}} ||= [];
4268 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4269 undef;
4270 }
4272 sub absent_file {
4273 my ($self, $path, $pb) = @_;
4274 return undef if $self->is_path_ignored($path);
4275 $self->{absent_file}->{$pb->{path}} ||= [];
4276 push @{$self->{absent_file}->{$pb->{path}}}, $path;
4277 undef;
4278 }
4280 sub change_file_prop {
4281 my ($self, $fb, $prop, $value) = @_;
4282 return undef if $self->is_path_ignored($fb->{path});
4283 if ($prop eq 'svn:executable') {
4284 if ($fb->{mode_b} != 120000) {
4285 $fb->{mode_b} = defined $value ? 100755 : 100644;
4286 }
4287 } elsif ($prop eq 'svn:special') {
4288 $fb->{mode_b} = defined $value ? 120000 : 100644;
4289 } else {
4290 $self->{file_prop}->{$fb->{path}} ||= {};
4291 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4292 }
4293 undef;
4294 }
4296 sub apply_textdelta {
4297 my ($self, $fb, $exp) = @_;
4298 return undef if $self->is_path_ignored($fb->{path});
4299 my $fh = $::_repository->temp_acquire('svn_delta');
4300 # $fh gets auto-closed() by SVN::TxDelta::apply(),
4301 # (but $base does not,) so dup() it for reading in close_file
4302 open my $dup, '<&', $fh or croak $!;
4303 my $base = $::_repository->temp_acquire('git_blob');
4305 if ($fb->{blob}) {
4306 my ($base_is_link, $size);
4308 if ($fb->{mode_a} eq '120000' &&
4309 ! $self->{empty_symlinks}->{$fb->{path}}) {
4310 print $base 'link ' or die "print $!\n";
4311 $base_is_link = 1;
4312 }
4313 retry:
4314 $size = $::_repository->cat_blob($fb->{blob}, $base);
4315 die "Failed to read object $fb->{blob}" if ($size < 0);
4317 if (defined $exp) {
4318 seek $base, 0, 0 or croak $!;
4319 my $got = ::md5sum($base);
4320 if ($got ne $exp) {
4321 my $err = "Checksum mismatch: ".
4322 "$fb->{path} $fb->{blob}\n" .
4323 "expected: $exp\n" .
4324 " got: $got\n";
4325 if ($base_is_link) {
4326 warn $err,
4327 "Retrying... (possibly ",
4328 "a bad symlink from SVN)\n";
4329 $::_repository->temp_reset($base);
4330 $base_is_link = 0;
4331 goto retry;
4332 }
4333 die $err;
4334 }
4335 }
4336 }
4337 seek $base, 0, 0 or croak $!;
4338 $fb->{fh} = $fh;
4339 $fb->{base} = $base;
4340 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4341 }
4343 sub close_file {
4344 my ($self, $fb, $exp) = @_;
4345 return undef if $self->is_path_ignored($fb->{path});
4347 my $hash;
4348 my $path = $self->git_path($fb->{path});
4349 if (my $fh = $fb->{fh}) {
4350 if (defined $exp) {
4351 seek($fh, 0, 0) or croak $!;
4352 my $got = ::md5sum($fh);
4353 if ($got ne $exp) {
4354 die "Checksum mismatch: $path\n",
4355 "expected: $exp\n got: $got\n";
4356 }
4357 }
4358 if ($fb->{mode_b} == 120000) {
4359 sysseek($fh, 0, 0) or croak $!;
4360 my $rd = sysread($fh, my $buf, 5);
4362 if (!defined $rd) {
4363 croak "sysread: $!\n";
4364 } elsif ($rd == 0) {
4365 warn "$path has mode 120000",
4366 " but it points to nothing\n",
4367 "converting to an empty file with mode",
4368 " 100644\n";
4369 $fb->{mode_b} = '100644';
4370 } elsif ($buf ne 'link ') {
4371 warn "$path has mode 120000",
4372 " but is not a link\n";
4373 } else {
4374 my $tmp_fh = $::_repository->temp_acquire(
4375 'svn_hash');
4376 my $res;
4377 while ($res = sysread($fh, my $str, 1024)) {
4378 my $out = syswrite($tmp_fh, $str, $res);
4379 defined($out) && $out == $res
4380 or croak("write ",
4381 Git::temp_path($tmp_fh),
4382 ": $!\n");
4383 }
4384 defined $res or croak $!;
4386 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4387 Git::temp_release($tmp_fh, 1);
4388 }
4389 }
4391 $hash = $::_repository->hash_and_insert_object(
4392 Git::temp_path($fh));
4393 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4395 Git::temp_release($fb->{base}, 1);
4396 Git::temp_release($fh, 1);
4397 } else {
4398 $hash = $fb->{blob} or die "no blob information\n";
4399 }
4400 $fb->{pool}->clear;
4401 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4402 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4403 undef;
4404 }
4406 sub abort_edit {
4407 my $self = shift;
4408 $self->{nr} = $self->{gii}->{nr};
4409 delete $self->{gii};
4410 $self->SUPER::abort_edit(@_);
4411 }
4413 sub close_edit {
4414 my $self = shift;
4415 $self->{git_commit_ok} = 1;
4416 $self->{nr} = $self->{gii}->{nr};
4417 delete $self->{gii};
4418 $self->SUPER::close_edit(@_);
4419 }
4421 package SVN::Git::Editor;
4422 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4423 use strict;
4424 use warnings;
4425 use Carp qw/croak/;
4426 use IO::File;
4428 sub new {
4429 my ($class, $opts) = @_;
4430 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4431 die "$_ required!\n" unless (defined $opts->{$_});
4432 }
4434 my $pool = SVN::Pool->new;
4435 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4436 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4437 $opts->{r}, $mods);
4439 # $opts->{ra} functions should not be used after this:
4440 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4441 $opts->{editor_cb}, $pool);
4442 my $self = SVN::Delta::Editor->new(@ce, $pool);
4443 bless $self, $class;
4444 foreach (qw/svn_path r tree_a tree_b/) {
4445 $self->{$_} = $opts->{$_};
4446 }
4447 $self->{url} = $opts->{ra}->{url};
4448 $self->{mods} = $mods;
4449 $self->{types} = $types;
4450 $self->{pool} = $pool;
4451 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4452 $self->{rm} = { };
4453 $self->{path_prefix} = length $self->{svn_path} ?
4454 "$self->{svn_path}/" : '';
4455 $self->{config} = $opts->{config};
4456 $self->{mergeinfo} = $opts->{mergeinfo};
4457 return $self;
4458 }
4460 sub generate_diff {
4461 my ($tree_a, $tree_b) = @_;
4462 my @diff_tree = qw(diff-tree -z -r);
4463 if ($_cp_similarity) {
4464 push @diff_tree, "-C$_cp_similarity";
4465 } else {
4466 push @diff_tree, '-C';
4467 }
4468 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4469 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4470 push @diff_tree, $tree_a, $tree_b;
4471 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4472 local $/ = "\0";
4473 my $state = 'meta';
4474 my @mods;
4475 while (<$diff_fh>) {
4476 chomp $_; # this gets rid of the trailing "\0"
4477 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4478 ($::sha1)\s($::sha1)\s
4479 ([MTCRAD])\d*$/xo) {
4480 push @mods, { mode_a => $1, mode_b => $2,
4481 sha1_a => $3, sha1_b => $4,
4482 chg => $5 };
4483 if ($5 =~ /^(?:C|R)$/) {
4484 $state = 'file_a';
4485 } else {
4486 $state = 'file_b';
4487 }
4488 } elsif ($state eq 'file_a') {
4489 my $x = $mods[$#mods] or croak "Empty array\n";
4490 if ($x->{chg} !~ /^(?:C|R)$/) {
4491 croak "Error parsing $_, $x->{chg}\n";
4492 }
4493 $x->{file_a} = $_;
4494 $state = 'file_b';
4495 } elsif ($state eq 'file_b') {
4496 my $x = $mods[$#mods] or croak "Empty array\n";
4497 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4498 croak "Error parsing $_, $x->{chg}\n";
4499 }
4500 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4501 croak "Error parsing $_, $x->{chg}\n";
4502 }
4503 $x->{file_b} = $_;
4504 $state = 'meta';
4505 } else {
4506 croak "Error parsing $_\n";
4507 }
4508 }
4509 command_close_pipe($diff_fh, $ctx);
4510 \@mods;
4511 }
4513 sub check_diff_paths {
4514 my ($ra, $pfx, $rev, $mods) = @_;
4515 my %types;
4516 $pfx .= '/' if length $pfx;
4518 sub type_diff_paths {
4519 my ($ra, $types, $path, $rev) = @_;
4520 my @p = split m#/+#, $path;
4521 my $c = shift @p;
4522 unless (defined $types->{$c}) {
4523 $types->{$c} = $ra->check_path($c, $rev);
4524 }
4525 while (@p) {
4526 $c .= '/' . shift @p;
4527 next if defined $types->{$c};
4528 $types->{$c} = $ra->check_path($c, $rev);
4529 }
4530 }
4532 foreach my $m (@$mods) {
4533 foreach my $f (qw/file_a file_b/) {
4534 next unless defined $m->{$f};
4535 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4536 if (length $pfx.$dir && ! defined $types{$dir}) {
4537 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4538 }
4539 }
4540 }
4541 \%types;
4542 }
4544 sub split_path {
4545 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4546 }
4548 sub repo_path {
4549 my ($self, $path) = @_;
4550 if (my $enc = $self->{pathnameencoding}) {
4551 require Encode;
4552 Encode::from_to($path, $enc, 'UTF-8');
4553 }
4554 $self->{path_prefix}.(defined $path ? $path : '');
4555 }
4557 sub url_path {
4558 my ($self, $path) = @_;
4559 if ($self->{url} =~ m#^https?://#) {
4560 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4561 }
4562 $self->{url} . '/' . $self->repo_path($path);
4563 }
4565 sub rmdirs {
4566 my ($self) = @_;
4567 my $rm = $self->{rm};
4568 delete $rm->{''}; # we never delete the url we're tracking
4569 return unless %$rm;
4571 foreach (keys %$rm) {
4572 my @d = split m#/#, $_;
4573 my $c = shift @d;
4574 $rm->{$c} = 1;
4575 while (@d) {
4576 $c .= '/' . shift @d;
4577 $rm->{$c} = 1;
4578 }
4579 }
4580 delete $rm->{$self->{svn_path}};
4581 delete $rm->{''}; # we never delete the url we're tracking
4582 return unless %$rm;
4584 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4585 $self->{tree_b});
4586 local $/ = "\0";
4587 while (<$fh>) {
4588 chomp;
4589 my @dn = split m#/#, $_;
4590 while (pop @dn) {
4591 delete $rm->{join '/', @dn};
4592 }
4593 unless (%$rm) {
4594 close $fh;
4595 return;
4596 }
4597 }
4598 command_close_pipe($fh, $ctx);
4600 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4601 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4602 $self->close_directory($bat->{$d}, $p);
4603 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4604 print "\tD+\t$d/\n" unless $::_q;
4605 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4606 delete $bat->{$d};
4607 }
4608 }
4610 sub open_or_add_dir {
4611 my ($self, $full_path, $baton) = @_;
4612 my $t = $self->{types}->{$full_path};
4613 if (!defined $t) {
4614 die "$full_path not known in r$self->{r} or we have a bug!\n";
4615 }
4616 {
4617 no warnings 'once';
4618 # SVN::Node::none and SVN::Node::file are used only once,
4619 # so we're shutting up Perl's warnings about them.
4620 if ($t == $SVN::Node::none) {
4621 return $self->add_directory($full_path, $baton,
4622 undef, -1, $self->{pool});
4623 } elsif ($t == $SVN::Node::dir) {
4624 return $self->open_directory($full_path, $baton,
4625 $self->{r}, $self->{pool});
4626 } # no warnings 'once'
4627 print STDERR "$full_path already exists in repository at ",
4628 "r$self->{r} and it is not a directory (",
4629 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4630 } # no warnings 'once'
4631 exit 1;
4632 }
4634 sub ensure_path {
4635 my ($self, $path) = @_;
4636 my $bat = $self->{bat};
4637 my $repo_path = $self->repo_path($path);
4638 return $bat->{''} unless (length $repo_path);
4639 my @p = split m#/+#, $repo_path;
4640 my $c = shift @p;
4641 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4642 while (@p) {
4643 my $c0 = $c;
4644 $c .= '/' . shift @p;
4645 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4646 }
4647 return $bat->{$c};
4648 }
4650 # Subroutine to convert a globbing pattern to a regular expression.
4651 # From perl cookbook.
4652 sub glob2pat {
4653 my $globstr = shift;
4654 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4655 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4656 return '^' . $globstr . '$';
4657 }
4659 sub check_autoprop {
4660 my ($self, $pattern, $properties, $file, $fbat) = @_;
4661 # Convert the globbing pattern to a regular expression.
4662 my $regex = glob2pat($pattern);
4663 # Check if the pattern matches the file name.
4664 if($file =~ m/($regex)/) {
4665 # Parse the list of properties to set.
4666 my @props = split(/;/, $properties);
4667 foreach my $prop (@props) {
4668 # Parse 'name=value' syntax and set the property.
4669 if ($prop =~ /([^=]+)=(.*)/) {
4670 my ($n,$v) = ($1,$2);
4671 for ($n, $v) {
4672 s/^\s+//; s/\s+$//;
4673 }
4674 $self->change_file_prop($fbat, $n, $v);
4675 }
4676 }
4677 }
4678 }
4680 sub apply_autoprops {
4681 my ($self, $file, $fbat) = @_;
4682 my $conf_t = ${$self->{config}}{'config'};
4683 no warnings 'once';
4684 # Check [miscellany]/enable-auto-props in svn configuration.
4685 if (SVN::_Core::svn_config_get_bool(
4686 $conf_t,
4687 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4688 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4689 0)) {
4690 # Auto-props are enabled. Enumerate them to look for matches.
4691 my $callback = sub {
4692 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4693 };
4694 SVN::_Core::svn_config_enumerate(
4695 $conf_t,
4696 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4697 $callback);
4698 }
4699 }
4701 sub A {
4702 my ($self, $m) = @_;
4703 my ($dir, $file) = split_path($m->{file_b});
4704 my $pbat = $self->ensure_path($dir);
4705 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4706 undef, -1);
4707 print "\tA\t$m->{file_b}\n" unless $::_q;
4708 $self->apply_autoprops($file, $fbat);
4709 $self->chg_file($fbat, $m);
4710 $self->close_file($fbat,undef,$self->{pool});
4711 }
4713 sub C {
4714 my ($self, $m) = @_;
4715 my ($dir, $file) = split_path($m->{file_b});
4716 my $pbat = $self->ensure_path($dir);
4717 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4718 $self->url_path($m->{file_a}), $self->{r});
4719 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4720 $self->chg_file($fbat, $m);
4721 $self->close_file($fbat,undef,$self->{pool});
4722 }
4724 sub delete_entry {
4725 my ($self, $path, $pbat) = @_;
4726 my $rpath = $self->repo_path($path);
4727 my ($dir, $file) = split_path($rpath);
4728 $self->{rm}->{$dir} = 1;
4729 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4730 }
4732 sub R {
4733 my ($self, $m) = @_;
4734 my ($dir, $file) = split_path($m->{file_b});
4735 my $pbat = $self->ensure_path($dir);
4736 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4737 $self->url_path($m->{file_a}), $self->{r});
4738 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4739 $self->apply_autoprops($file, $fbat);
4740 $self->chg_file($fbat, $m);
4741 $self->close_file($fbat,undef,$self->{pool});
4743 ($dir, $file) = split_path($m->{file_a});
4744 $pbat = $self->ensure_path($dir);
4745 $self->delete_entry($m->{file_a}, $pbat);
4746 }
4748 sub M {
4749 my ($self, $m) = @_;
4750 my ($dir, $file) = split_path($m->{file_b});
4751 my $pbat = $self->ensure_path($dir);
4752 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4753 $pbat,$self->{r},$self->{pool});
4754 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4755 $self->chg_file($fbat, $m);
4756 $self->close_file($fbat,undef,$self->{pool});
4757 }
4759 sub T { shift->M(@_) }
4761 sub change_file_prop {
4762 my ($self, $fbat, $pname, $pval) = @_;
4763 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4764 }
4766 sub change_dir_prop {
4767 my ($self, $pbat, $pname, $pval) = @_;
4768 $self->SUPER::change_dir_prop($pbat, $pname, $pval, $self->{pool});
4769 }
4771 sub _chg_file_get_blob ($$$$) {
4772 my ($self, $fbat, $m, $which) = @_;
4773 my $fh = $::_repository->temp_acquire("git_blob_$which");
4774 if ($m->{"mode_$which"} =~ /^120/) {
4775 print $fh 'link ' or croak $!;
4776 $self->change_file_prop($fbat,'svn:special','*');
4777 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4778 $self->change_file_prop($fbat,'svn:special',undef);
4779 }
4780 my $blob = $m->{"sha1_$which"};
4781 return ($fh,) if ($blob =~ /^0{40}$/);
4782 my $size = $::_repository->cat_blob($blob, $fh);
4783 croak "Failed to read object $blob" if ($size < 0);
4784 $fh->flush == 0 or croak $!;
4785 seek $fh, 0, 0 or croak $!;
4787 my $exp = ::md5sum($fh);
4788 seek $fh, 0, 0 or croak $!;
4789 return ($fh, $exp);
4790 }
4792 sub chg_file {
4793 my ($self, $fbat, $m) = @_;
4794 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4795 $self->change_file_prop($fbat,'svn:executable','*');
4796 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4797 $self->change_file_prop($fbat,'svn:executable',undef);
4798 }
4799 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4800 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4801 my $pool = SVN::Pool->new;
4802 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4803 if (-s $fh_a) {
4804 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4805 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4806 if (defined $res) {
4807 die "Unexpected result from send_txstream: $res\n",
4808 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4809 }
4810 } else {
4811 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4812 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4813 if ($got ne $exp_b);
4814 }
4815 Git::temp_release($fh_b, 1);
4816 Git::temp_release($fh_a, 1);
4817 $pool->clear;
4818 }
4820 sub D {
4821 my ($self, $m) = @_;
4822 my ($dir, $file) = split_path($m->{file_b});
4823 my $pbat = $self->ensure_path($dir);
4824 print "\tD\t$m->{file_b}\n" unless $::_q;
4825 $self->delete_entry($m->{file_b}, $pbat);
4826 }
4828 sub close_edit {
4829 my ($self) = @_;
4830 my ($p,$bat) = ($self->{pool}, $self->{bat});
4831 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4832 next if $_ eq '';
4833 $self->close_directory($bat->{$_}, $p);
4834 }
4835 $self->close_directory($bat->{''}, $p);
4836 $self->SUPER::close_edit($p);
4837 $p->clear;
4838 }
4840 sub abort_edit {
4841 my ($self) = @_;
4842 $self->SUPER::abort_edit($self->{pool});
4843 }
4845 sub DESTROY {
4846 my $self = shift;
4847 $self->SUPER::DESTROY(@_);
4848 $self->{pool}->clear;
4849 }
4851 # this drives the editor
4852 sub apply_diff {
4853 my ($self) = @_;
4854 my $mods = $self->{mods};
4855 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4856 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4857 my $f = $m->{chg};
4858 if (defined $o{$f}) {
4859 $self->$f($m);
4860 } else {
4861 fatal("Invalid change type: $f");
4862 }
4863 }
4865 if (defined($self->{mergeinfo})) {
4866 $self->change_dir_prop($self->{bat}{''}, "svn:mergeinfo",
4867 $self->{mergeinfo});
4868 }
4869 $self->rmdirs if $_rmdir;
4870 if (@$mods == 0) {
4871 $self->abort_edit;
4872 } else {
4873 $self->close_edit;
4874 }
4875 return scalar @$mods;
4876 }
4878 package Git::SVN::Ra;
4879 use vars qw/@ISA $config_dir $_log_window_size/;
4880 use strict;
4881 use warnings;
4882 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4884 BEGIN {
4885 # enforce temporary pool usage for some simple functions
4886 no strict 'refs';
4887 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4888 get_file/) {
4889 my $SUPER = "SUPER::$f";
4890 *$f = sub {
4891 my $self = shift;
4892 my $pool = SVN::Pool->new;
4893 my @ret = $self->$SUPER(@_,$pool);
4894 $pool->clear;
4895 wantarray ? @ret : $ret[0];
4896 };
4897 }
4898 }
4900 sub _auth_providers () {
4901 [
4902 SVN::Client::get_simple_provider(),
4903 SVN::Client::get_ssl_server_trust_file_provider(),
4904 SVN::Client::get_simple_prompt_provider(
4905 \&Git::SVN::Prompt::simple, 2),
4906 SVN::Client::get_ssl_client_cert_file_provider(),
4907 SVN::Client::get_ssl_client_cert_prompt_provider(
4908 \&Git::SVN::Prompt::ssl_client_cert, 2),
4909 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4910 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4911 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4912 SVN::Client::get_username_provider(),
4913 SVN::Client::get_ssl_server_trust_prompt_provider(
4914 \&Git::SVN::Prompt::ssl_server_trust),
4915 SVN::Client::get_username_prompt_provider(
4916 \&Git::SVN::Prompt::username, 2)
4917 ]
4918 }
4920 sub escape_uri_only {
4921 my ($uri) = @_;
4922 my @tmp;
4923 foreach (split m{/}, $uri) {
4924 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4925 push @tmp, $_;
4926 }
4927 join('/', @tmp);
4928 }
4930 sub escape_url {
4931 my ($url) = @_;
4932 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4933 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4934 $url = "$scheme://$domain$uri";
4935 }
4936 $url;
4937 }
4939 sub new {
4940 my ($class, $url) = @_;
4941 $url =~ s!/+$!!;
4942 return $RA if ($RA && $RA->{url} eq $url);
4944 ::_req_svn();
4946 SVN::_Core::svn_config_ensure($config_dir, undef);
4947 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4948 my $config = SVN::Core::config_get_config($config_dir);
4949 $RA = undef;
4950 my $dont_store_passwords = 1;
4951 my $conf_t = ${$config}{'config'};
4952 {
4953 no warnings 'once';
4954 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4955 # produces warnings that variables are used only once.
4956 # I had not found the better way to shut them up, so
4957 # the warnings of type 'once' are disabled in this block.
4958 if (SVN::_Core::svn_config_get_bool($conf_t,
4959 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4960 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4961 1) == 0) {
4962 SVN::_Core::svn_auth_set_parameter($baton,
4963 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4964 bless (\$dont_store_passwords, "_p_void"));
4965 }
4966 if (SVN::_Core::svn_config_get_bool($conf_t,
4967 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4968 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4969 1) == 0) {
4970 $Git::SVN::Prompt::_no_auth_cache = 1;
4971 }
4972 } # no warnings 'once'
4973 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4974 config => $config,
4975 pool => SVN::Pool->new,
4976 auth_provider_callbacks => $callbacks);
4977 $self->{url} = $url;
4978 $self->{svn_path} = $url;
4979 $self->{repos_root} = $self->get_repos_root;
4980 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4981 $self->{cache} = { check_path => { r => 0, data => {} },
4982 get_dir => { r => 0, data => {} } };
4983 $RA = bless $self, $class;
4984 }
4986 sub check_path {
4987 my ($self, $path, $r) = @_;
4988 my $cache = $self->{cache}->{check_path};
4989 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4990 return $cache->{data}->{$path};
4991 }
4992 my $pool = SVN::Pool->new;
4993 my $t = $self->SUPER::check_path($path, $r, $pool);
4994 $pool->clear;
4995 if ($r != $cache->{r}) {
4996 %{$cache->{data}} = ();
4997 $cache->{r} = $r;
4998 }
4999 $cache->{data}->{$path} = $t;
5000 }
5002 sub get_dir {
5003 my ($self, $dir, $r) = @_;
5004 my $cache = $self->{cache}->{get_dir};
5005 if ($r == $cache->{r}) {
5006 if (my $x = $cache->{data}->{$dir}) {
5007 return wantarray ? @$x : $x->[0];
5008 }
5009 }
5010 my $pool = SVN::Pool->new;
5011 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
5012 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
5013 $pool->clear;
5014 if ($r != $cache->{r}) {
5015 %{$cache->{data}} = ();
5016 $cache->{r} = $r;
5017 }
5018 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
5019 wantarray ? (\%dirents, $r, $props) : \%dirents;
5020 }
5022 sub DESTROY {
5023 # do not call the real DESTROY since we store ourselves in $RA
5024 }
5026 # get_log(paths, start, end, limit,
5027 # discover_changed_paths, strict_node_history, receiver)
5028 sub get_log {
5029 my ($self, @args) = @_;
5030 my $pool = SVN::Pool->new;
5032 # svn_log_changed_path_t objects passed to get_log are likely to be
5033 # overwritten even if only the refs are copied to an external variable,
5034 # so we should dup the structures in their entirety. Using an
5035 # externally passed pool (instead of our temporary and quickly cleared
5036 # pool in Git::SVN::Ra) does not help matters at all...
5037 my $receiver = pop @args;
5038 my $prefix = "/".$self->{svn_path};
5039 $prefix =~ s#/+($)##;
5040 my $prefix_regex = qr#^\Q$prefix\E#;
5041 push(@args, sub {
5042 my ($paths) = $_[0];
5043 return &$receiver(@_) unless $paths;
5044 $_[0] = ();
5045 foreach my $p (keys %$paths) {
5046 my $i = $paths->{$p};
5047 # Make path relative to our url, not repos_root
5048 $p =~ s/$prefix_regex//;
5049 my %s = map { $_ => $i->$_; }
5050 qw/copyfrom_path copyfrom_rev action/;
5051 if ($s{'copyfrom_path'}) {
5052 $s{'copyfrom_path'} =~ s/$prefix_regex//;
5053 }
5054 $_[0]{$p} = \%s;
5055 }
5056 &$receiver(@_);
5057 });
5060 # the limit parameter was not supported in SVN 1.1.x, so we
5061 # drop it. Therefore, the receiver callback passed to it
5062 # is made aware of this limitation by being wrapped if
5063 # the limit passed to is being wrapped.
5064 if ($SVN::Core::VERSION le '1.2.0') {
5065 my $limit = splice(@args, 3, 1);
5066 if ($limit > 0) {
5067 my $receiver = pop @args;
5068 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
5069 }
5070 }
5071 my $ret = $self->SUPER::get_log(@args, $pool);
5072 $pool->clear;
5073 $ret;
5074 }
5076 sub trees_match {
5077 my ($self, $url1, $rev1, $url2, $rev2) = @_;
5078 my $ctx = SVN::Client->new(auth => _auth_providers);
5079 my $out = IO::File->new_tmpfile;
5081 # older SVN (1.1.x) doesn't take $pool as the last parameter for
5082 # $ctx->diff(), so we'll create a default one
5083 my $pool = SVN::Pool->new_default_sub;
5085 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5086 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5087 $out->flush;
5088 my $ret = (($out->stat)[7] == 0);
5089 close $out or croak $!;
5091 $ret;
5092 }
5094 sub get_commit_editor {
5095 my ($self, $log, $cb, $pool) = @_;
5096 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
5097 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5098 }
5100 sub gs_do_update {
5101 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5102 my $new = ($rev_a == $rev_b);
5103 my $path = $gs->{path};
5105 if ($new && -e $gs->{index}) {
5106 unlink $gs->{index} or die
5107 "Couldn't unlink index: $gs->{index}: $!\n";
5108 }
5109 my $pool = SVN::Pool->new;
5110 $editor->set_path_strip($path);
5111 my (@pc) = split m#/#, $path;
5112 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5113 1, $editor, $pool);
5114 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5116 # Since we can't rely on svn_ra_reparent being available, we'll
5117 # just have to do some magic with set_path to make it so
5118 # we only want a partial path.
5119 my $sp = '';
5120 my $final = join('/', @pc);
5121 while (@pc) {
5122 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5123 $sp .= '/' if length $sp;
5124 $sp .= shift @pc;
5125 }
5126 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5128 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5130 $reporter->finish_report($pool);
5131 $pool->clear;
5132 $editor->{git_commit_ok};
5133 }
5135 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5136 # svn_ra_reparent didn't work before 1.4)
5137 sub gs_do_switch {
5138 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5139 my $path = $gs->{path};
5140 my $pool = SVN::Pool->new;
5142 my $full_url = $self->{url};
5143 my $old_url = $full_url;
5144 $full_url .= '/' . $path if length $path;
5145 my ($ra, $reparented);
5147 if ($old_url =~ m#^svn(\+ssh)?://# ||
5148 ($full_url =~ m#^https?://# &&
5149 escape_url($full_url) ne $full_url)) {
5150 $_[0] = undef;
5151 $self = undef;
5152 $RA = undef;
5153 $ra = Git::SVN::Ra->new($full_url);
5154 $ra_invalid = 1;
5155 } elsif ($old_url ne $full_url) {
5156 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5157 $self->{url} = $full_url;
5158 $reparented = 1;
5159 }
5161 $ra ||= $self;
5162 $url_b = escape_url($url_b);
5163 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5164 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5165 $reporter->set_path('', $rev_a, 0, @lock, $pool);
5166 $reporter->finish_report($pool);
5168 if ($reparented) {
5169 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5170 $self->{url} = $old_url;
5171 }
5173 $pool->clear;
5174 $editor->{git_commit_ok};
5175 }
5177 sub longest_common_path {
5178 my ($gsv, $globs) = @_;
5179 my %common;
5180 my $common_max = scalar @$gsv;
5182 foreach my $gs (@$gsv) {
5183 my @tmp = split m#/#, $gs->{path};
5184 my $p = '';
5185 foreach (@tmp) {
5186 $p .= length($p) ? "/$_" : $_;
5187 $common{$p} ||= 0;
5188 $common{$p}++;
5189 }
5190 }
5191 $globs ||= [];
5192 $common_max += scalar @$globs;
5193 foreach my $glob (@$globs) {
5194 my @tmp = split m#/#, $glob->{path}->{left};
5195 my $p = '';
5196 foreach (@tmp) {
5197 $p .= length($p) ? "/$_" : $_;
5198 $common{$p} ||= 0;
5199 $common{$p}++;
5200 }
5201 }
5203 my $longest_path = '';
5204 foreach (sort {length $b <=> length $a} keys %common) {
5205 if ($common{$_} == $common_max) {
5206 $longest_path = $_;
5207 last;
5208 }
5209 }
5210 $longest_path;
5211 }
5213 sub gs_fetch_loop_common {
5214 my ($self, $base, $head, $gsv, $globs) = @_;
5215 return if ($base > $head);
5216 my $inc = $_log_window_size;
5217 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5218 my $longest_path = longest_common_path($gsv, $globs);
5219 my $ra_url = $self->{url};
5220 my $find_trailing_edge;
5221 while (1) {
5222 my %revs;
5223 my $err;
5224 my $err_handler = $SVN::Error::handler;
5225 $SVN::Error::handler = sub {
5226 ($err) = @_;
5227 skip_unknown_revs($err);
5228 };
5229 sub _cb {
5230 my ($paths, $r, $author, $date, $log) = @_;
5231 [ $paths,
5232 { author => $author, date => $date, log => $log } ];
5233 }
5234 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5235 sub { $revs{$_[1]} = _cb(@_) });
5236 if ($err) {
5237 print "Checked through r$max\r";
5238 } else {
5239 $find_trailing_edge = 1;
5240 }
5241 if ($err and $find_trailing_edge) {
5242 print STDERR "Path '$longest_path' ",
5243 "was probably deleted:\n",
5244 $err->expanded_message,
5245 "\nWill attempt to follow ",
5246 "revisions r$min .. r$max ",
5247 "committed before the deletion\n";
5248 my $hi = $max;
5249 while (--$hi >= $min) {
5250 my $ok;
5251 $self->get_log([$longest_path], $min, $hi,
5252 0, 1, 1, sub {
5253 $ok = $_[1];
5254 $revs{$_[1]} = _cb(@_) });
5255 if ($ok) {
5256 print STDERR "r$min .. r$ok OK\n";
5257 last;
5258 }
5259 }
5260 $find_trailing_edge = 0;
5261 }
5262 $SVN::Error::handler = $err_handler;
5264 my %exists = map { $_->{path} => $_ } @$gsv;
5265 foreach my $r (sort {$a <=> $b} keys %revs) {
5266 my ($paths, $logged) = @{$revs{$r}};
5268 foreach my $gs ($self->match_globs(\%exists, $paths,
5269 $globs, $r)) {
5270 if ($gs->rev_map_max >= $r) {
5271 next;
5272 }
5273 next unless $gs->match_paths($paths, $r);
5274 $gs->{logged_rev_props} = $logged;
5275 if (my $last_commit = $gs->last_commit) {
5276 $gs->assert_index_clean($last_commit);
5277 }
5278 my $log_entry = $gs->do_fetch($paths, $r);
5279 if ($log_entry) {
5280 $gs->do_git_commit($log_entry);
5281 }
5282 $INDEX_FILES{$gs->{index}} = 1;
5283 }
5284 foreach my $g (@$globs) {
5285 my $k = "svn-remote.$g->{remote}." .
5286 "$g->{t}-maxRev";
5287 Git::SVN::tmp_config($k, $r);
5288 }
5289 if ($ra_invalid) {
5290 $_[0] = undef;
5291 $self = undef;
5292 $RA = undef;
5293 $self = Git::SVN::Ra->new($ra_url);
5294 $ra_invalid = undef;
5295 }
5296 }
5297 # pre-fill the .rev_db since it'll eventually get filled in
5298 # with '0' x40 if something new gets committed
5299 foreach my $gs (@$gsv) {
5300 next if $gs->rev_map_max >= $max;
5301 next if defined $gs->rev_map_get($max);
5302 $gs->rev_map_set($max, 0 x40);
5303 }
5304 foreach my $g (@$globs) {
5305 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5306 Git::SVN::tmp_config($k, $max);
5307 }
5308 last if $max >= $head;
5309 $min = $max + 1;
5310 $max += $inc;
5311 $max = $head if ($max > $head);
5312 }
5313 Git::SVN::gc();
5314 }
5316 sub get_dir_globbed {
5317 my ($self, $left, $depth, $r) = @_;
5319 my @x = eval { $self->get_dir($left, $r) };
5320 return unless scalar @x == 3;
5321 my $dirents = $x[0];
5322 my @finalents;
5323 foreach my $de (keys %$dirents) {
5324 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5325 if ($depth > 1) {
5326 my @args = ("$left/$de", $depth - 1, $r);
5327 foreach my $dir ($self->get_dir_globbed(@args)) {
5328 push @finalents, "$de/$dir";
5329 }
5330 } else {
5331 push @finalents, $de;
5332 }
5333 }
5334 @finalents;
5335 }
5337 sub match_globs {
5338 my ($self, $exists, $paths, $globs, $r) = @_;
5340 sub get_dir_check {
5341 my ($self, $exists, $g, $r) = @_;
5343 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5344 $g->{path}->{depth},
5345 $r);
5347 foreach my $de (@dirs) {
5348 my $p = $g->{path}->full_path($de);
5349 next if $exists->{$p};
5350 next if (length $g->{path}->{right} &&
5351 ($self->check_path($p, $r) !=
5352 $SVN::Node::dir));
5353 next unless $p =~ /$g->{path}->{regex}/;
5354 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5355 $g->{ref}->full_path($de), 1);
5356 }
5357 }
5358 foreach my $g (@$globs) {
5359 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5360 if ($path->{action} =~ /^[AR]$/) {
5361 get_dir_check($self, $exists, $g, $r);
5362 }
5363 }
5364 foreach (keys %$paths) {
5365 if (/$g->{path}->{left_regex}/ &&
5366 !/$g->{path}->{regex}/) {
5367 next if $paths->{$_}->{action} !~ /^[AR]$/;
5368 get_dir_check($self, $exists, $g, $r);
5369 }
5370 next unless /$g->{path}->{regex}/;
5371 my $p = $1;
5372 my $pathname = $g->{path}->full_path($p);
5373 next if $exists->{$pathname};
5374 next if ($self->check_path($pathname, $r) !=
5375 $SVN::Node::dir);
5376 $exists->{$pathname} = Git::SVN->init(
5377 $self->{url}, $pathname, undef,
5378 $g->{ref}->full_path($p), 1);
5379 }
5380 my $c = '';
5381 foreach (split m#/#, $g->{path}->{left}) {
5382 $c .= "/$_";
5383 next unless ($paths->{$c} &&
5384 ($paths->{$c}->{action} =~ /^[AR]$/));
5385 get_dir_check($self, $exists, $g, $r);
5386 }
5387 }
5388 values %$exists;
5389 }
5391 sub minimize_url {
5392 my ($self) = @_;
5393 return $self->{url} if ($self->{url} eq $self->{repos_root});
5394 my $url = $self->{repos_root};
5395 my @components = split(m!/!, $self->{svn_path});
5396 my $c = '';
5397 do {
5398 $url .= "/$c" if length $c;
5399 eval {
5400 my $ra = (ref $self)->new($url);
5401 my $latest = $ra->get_latest_revnum;
5402 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5403 };
5404 } while ($@ && ($c = shift @components));
5405 $url;
5406 }
5408 sub can_do_switch {
5409 my $self = shift;
5410 unless (defined $can_do_switch) {
5411 my $pool = SVN::Pool->new;
5412 my $rep = eval {
5413 $self->do_switch(1, '', 0, $self->{url},
5414 SVN::Delta::Editor->new, $pool);
5415 };
5416 if ($@) {
5417 $can_do_switch = 0;
5418 } else {
5419 $rep->abort_report($pool);
5420 $can_do_switch = 1;
5421 }
5422 $pool->clear;
5423 }
5424 $can_do_switch;
5425 }
5427 sub skip_unknown_revs {
5428 my ($err) = @_;
5429 my $errno = $err->apr_err();
5430 # Maybe the branch we're tracking didn't
5431 # exist when the repo started, so it's
5432 # not an error if it doesn't, just continue
5433 #
5434 # Wonderfully consistent library, eh?
5435 # 160013 - svn:// and file://
5436 # 175002 - http(s)://
5437 # 175007 - http(s):// (this repo required authorization, too...)
5438 # More codes may be discovered later...
5439 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5440 my $err_key = $err->expanded_message;
5441 # revision numbers change every time, filter them out
5442 $err_key =~ s/\d+/\0/g;
5443 $err_key = "$errno\0$err_key";
5444 unless ($ignored_err{$err_key}) {
5445 warn "W: Ignoring error from SVN, path probably ",
5446 "does not exist: ($errno): ",
5447 $err->expanded_message,"\n";
5448 warn "W: Do not be alarmed at the above message ",
5449 "git-svn is just searching aggressively for ",
5450 "old history.\n",
5451 "This may take a while on large repositories\n";
5452 $ignored_err{$err_key} = 1;
5453 }
5454 return;
5455 }
5456 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5457 }
5459 package Git::SVN::Log;
5460 use strict;
5461 use warnings;
5462 use POSIX qw/strftime/;
5463 use Time::Local;
5464 use constant commit_log_separator => ('-' x 72) . "\n";
5465 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5466 %rusers $show_commit $incremental/;
5467 my $l_fmt;
5469 sub cmt_showable {
5470 my ($c) = @_;
5471 return 1 if defined $c->{r};
5473 # big commit message got truncated by the 16k pretty buffer in rev-list
5474 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5475 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5476 @{$c->{l}} = ();
5477 my @log = command(qw/cat-file commit/, $c->{c});
5479 # shift off the headers
5480 shift @log while ($log[0] ne '');
5481 shift @log;
5483 # TODO: make $c->{l} not have a trailing newline in the future
5484 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5486 (undef, $c->{r}, undef) = ::extract_metadata(
5487 (grep(/^git-svn-id: /, @log))[-1]);
5488 }
5489 return defined $c->{r};
5490 }
5492 sub log_use_color {
5493 return $color || Git->repository->get_colorbool('color.diff');
5494 }
5496 sub git_svn_log_cmd {
5497 my ($r_min, $r_max, @args) = @_;
5498 my $head = 'HEAD';
5499 my (@files, @log_opts);
5500 foreach my $x (@args) {
5501 if ($x eq '--' || @files) {
5502 push @files, $x;
5503 } else {
5504 if (::verify_ref("$x^0")) {
5505 $head = $x;
5506 } else {
5507 push @log_opts, $x;
5508 }
5509 }
5510 }
5512 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5513 $gs ||= Git::SVN->_new;
5514 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5515 $gs->refname);
5516 push @cmd, '-r' unless $non_recursive;
5517 push @cmd, qw/--raw --name-status/ if $verbose;
5518 push @cmd, '--color' if log_use_color();
5519 push @cmd, @log_opts;
5520 if (defined $r_max && $r_max == $r_min) {
5521 push @cmd, '--max-count=1';
5522 if (my $c = $gs->rev_map_get($r_max)) {
5523 push @cmd, $c;
5524 }
5525 } elsif (defined $r_max) {
5526 if ($r_max < $r_min) {
5527 ($r_min, $r_max) = ($r_max, $r_min);
5528 }
5529 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5530 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5531 # If there are no commits in the range, both $c_max and $c_min
5532 # will be undefined. If there is at least 1 commit in the
5533 # range, both will be defined.
5534 return () if !defined $c_min || !defined $c_max;
5535 if ($c_min eq $c_max) {
5536 push @cmd, '--max-count=1', $c_min;
5537 } else {
5538 push @cmd, '--boundary', "$c_min..$c_max";
5539 }
5540 }
5541 return (@cmd, @files);
5542 }
5544 # adapted from pager.c
5545 sub config_pager {
5546 if (! -t *STDOUT) {
5547 $ENV{GIT_PAGER_IN_USE} = 'false';
5548 $pager = undef;
5549 return;
5550 }
5551 chomp($pager = command_oneline(qw(var GIT_PAGER)));
5552 if ($pager eq 'cat') {
5553 $pager = undef;
5554 }
5555 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5556 }
5558 sub run_pager {
5559 return unless defined $pager;
5560 pipe my ($rfd, $wfd) or return;
5561 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5562 if (!$pid) {
5563 open STDOUT, '>&', $wfd or
5564 ::fatal "Can't redirect to stdout: $!";
5565 return;
5566 }
5567 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5568 $ENV{LESS} ||= 'FRSX';
5569 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5570 }
5572 sub format_svn_date {
5573 # some systmes don't handle or mishandle %z, so be creative.
5574 my $t = shift || time;
5575 my $gm = timelocal(gmtime($t));
5576 my $sign = qw( + + - )[ $t <=> $gm ];
5577 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5578 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5579 }
5581 sub parse_git_date {
5582 my ($t, $tz) = @_;
5583 # Date::Parse isn't in the standard Perl distro :(
5584 if ($tz =~ s/^\+//) {
5585 $t += tz_to_s_offset($tz);
5586 } elsif ($tz =~ s/^\-//) {
5587 $t -= tz_to_s_offset($tz);
5588 }
5589 return $t;
5590 }
5592 sub set_local_timezone {
5593 if (defined $TZ) {
5594 $ENV{TZ} = $TZ;
5595 } else {
5596 delete $ENV{TZ};
5597 }
5598 }
5600 sub tz_to_s_offset {
5601 my ($tz) = @_;
5602 $tz =~ s/(\d\d)$//;
5603 return ($1 * 60) + ($tz * 3600);
5604 }
5606 sub get_author_info {
5607 my ($dest, $author, $t, $tz) = @_;
5608 $author =~ s/(?:^\s*|\s*$)//g;
5609 $dest->{a_raw} = $author;
5610 my $au;
5611 if ($::_authors) {
5612 $au = $rusers{$author} || undef;
5613 }
5614 if (!$au) {
5615 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5616 }
5617 $dest->{t} = $t;
5618 $dest->{tz} = $tz;
5619 $dest->{a} = $au;
5620 $dest->{t_utc} = parse_git_date($t, $tz);
5621 }
5623 sub process_commit {
5624 my ($c, $r_min, $r_max, $defer) = @_;
5625 if (defined $r_min && defined $r_max) {
5626 if ($r_min == $c->{r} && $r_min == $r_max) {
5627 show_commit($c);
5628 return 0;
5629 }
5630 return 1 if $r_min == $r_max;
5631 if ($r_min < $r_max) {
5632 # we need to reverse the print order
5633 return 0 if (defined $limit && --$limit < 0);
5634 push @$defer, $c;
5635 return 1;
5636 }
5637 if ($r_min != $r_max) {
5638 return 1 if ($r_min < $c->{r});
5639 return 1 if ($r_max > $c->{r});
5640 }
5641 }
5642 return 0 if (defined $limit && --$limit < 0);
5643 show_commit($c);
5644 return 1;
5645 }
5647 sub show_commit {
5648 my $c = shift;
5649 if ($oneline) {
5650 my $x = "\n";
5651 if (my $l = $c->{l}) {
5652 while ($l->[0] =~ /^\s*$/) { shift @$l }
5653 $x = $l->[0];
5654 }
5655 $l_fmt ||= 'A' . length($c->{r});
5656 print 'r',pack($l_fmt, $c->{r}),' | ';
5657 print "$c->{c} | " if $show_commit;
5658 print $x;
5659 } else {
5660 show_commit_normal($c);
5661 }
5662 }
5664 sub show_commit_changed_paths {
5665 my ($c) = @_;
5666 return unless $c->{changed};
5667 print "Changed paths:\n", @{$c->{changed}};
5668 }
5670 sub show_commit_normal {
5671 my ($c) = @_;
5672 print commit_log_separator, "r$c->{r} | ";
5673 print "$c->{c} | " if $show_commit;
5674 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5675 my $nr_line = 0;
5677 if (my $l = $c->{l}) {
5678 while ($l->[$#$l] eq "\n" && $#$l > 0
5679 && $l->[($#$l - 1)] eq "\n") {
5680 pop @$l;
5681 }
5682 $nr_line = scalar @$l;
5683 if (!$nr_line) {
5684 print "1 line\n\n\n";
5685 } else {
5686 if ($nr_line == 1) {
5687 $nr_line = '1 line';
5688 } else {
5689 $nr_line .= ' lines';
5690 }
5691 print $nr_line, "\n";
5692 show_commit_changed_paths($c);
5693 print "\n";
5694 print $_ foreach @$l;
5695 }
5696 } else {
5697 print "1 line\n";
5698 show_commit_changed_paths($c);
5699 print "\n";
5701 }
5702 foreach my $x (qw/raw stat diff/) {
5703 if ($c->{$x}) {
5704 print "\n";
5705 print $_ foreach @{$c->{$x}}
5706 }
5707 }
5708 }
5710 sub cmd_show_log {
5711 my (@args) = @_;
5712 my ($r_min, $r_max);
5713 my $r_last = -1; # prevent dupes
5714 set_local_timezone();
5715 if (defined $::_revision) {
5716 if ($::_revision =~ /^(\d+):(\d+)$/) {
5717 ($r_min, $r_max) = ($1, $2);
5718 } elsif ($::_revision =~ /^\d+$/) {
5719 $r_min = $r_max = $::_revision;
5720 } else {
5721 ::fatal "-r$::_revision is not supported, use ",
5722 "standard 'git log' arguments instead";
5723 }
5724 }
5726 config_pager();
5727 @args = git_svn_log_cmd($r_min, $r_max, @args);
5728 if (!@args) {
5729 print commit_log_separator unless $incremental || $oneline;
5730 return;
5731 }
5732 my $log = command_output_pipe(@args);
5733 run_pager();
5734 my (@k, $c, $d, $stat);
5735 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5736 while (<$log>) {
5737 if (/^${esc_color}commit -?($::sha1_short)/o) {
5738 my $cmt = $1;
5739 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5740 $r_last = $c->{r};
5741 process_commit($c, $r_min, $r_max, \@k) or
5742 goto out;
5743 }
5744 $d = undef;
5745 $c = { c => $cmt };
5746 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5747 get_author_info($c, $1, $2, $3);
5748 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5749 # ignore
5750 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5751 push @{$c->{raw}}, $_;
5752 } elsif (/^${esc_color}[ACRMDT]\t/) {
5753 # we could add $SVN->{svn_path} here, but that requires
5754 # remote access at the moment (repo_path_split)...
5755 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5756 push @{$c->{changed}}, $_;
5757 } elsif (/^${esc_color}diff /o) {
5758 $d = 1;
5759 push @{$c->{diff}}, $_;
5760 } elsif ($d) {
5761 push @{$c->{diff}}, $_;
5762 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5763 $esc_color*[\+\-]*$esc_color$/x) {
5764 $stat = 1;
5765 push @{$c->{stat}}, $_;
5766 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5767 push @{$c->{stat}}, $_;
5768 $stat = undef;
5769 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5770 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5771 } elsif (s/^${esc_color} //o) {
5772 push @{$c->{l}}, $_;
5773 }
5774 }
5775 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5776 $r_last = $c->{r};
5777 process_commit($c, $r_min, $r_max, \@k);
5778 }
5779 if (@k) {
5780 ($r_min, $r_max) = ($r_max, $r_min);
5781 process_commit($_, $r_min, $r_max) foreach reverse @k;
5782 }
5783 out:
5784 close $log;
5785 print commit_log_separator unless $incremental || $oneline;
5786 }
5788 sub cmd_blame {
5789 my $path = pop;
5791 config_pager();
5792 run_pager();
5794 my ($fh, $ctx, $rev);
5796 if ($_git_format) {
5797 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5798 while (my $line = <$fh>) {
5799 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5800 # Uncommitted edits show up as a rev ID of
5801 # all zeros, which we can't look up with
5802 # cmt_metadata
5803 if ($1 !~ /^0+$/) {
5804 (undef, $rev, undef) =
5805 ::cmt_metadata($1);
5806 $rev = '0' if (!$rev);
5807 } else {
5808 $rev = '0';
5809 }
5810 $rev = sprintf('%-10s', $rev);
5811 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5812 }
5813 print $line;
5814 }
5815 } else {
5816 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5817 '--', $path);
5818 my ($sha1);
5819 my %authors;
5820 my @buffer;
5821 my %dsha; #distinct sha keys
5823 while (my $line = <$fh>) {
5824 push @buffer, $line;
5825 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5826 $dsha{$1} = 1;
5827 }
5828 }
5830 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5832 foreach my $line (@buffer) {
5833 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5834 $rev = $s2r->{$1};
5835 $rev = '0' if (!$rev)
5836 }
5837 elsif ($line =~ /^author (.*)/) {
5838 $authors{$rev} = $1;
5839 $authors{$rev} =~ s/\s/_/g;
5840 }
5841 elsif ($line =~ /^\t(.*)$/) {
5842 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5843 }
5844 }
5845 }
5846 command_close_pipe($fh, $ctx);
5847 }
5849 package Git::SVN::Migration;
5850 # these version numbers do NOT correspond to actual version numbers
5851 # of git nor git-svn. They are just relative.
5852 #
5853 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5854 #
5855 # v1 layout: .git/$id/info/url, refs/remotes/$id
5856 #
5857 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5858 #
5859 # v3 layout: .git/svn/$id, refs/remotes/$id
5860 # - info/url may remain for backwards compatibility
5861 # - this is what we migrate up to this layout automatically,
5862 # - this will be used by git svn init on single branches
5863 # v3.1 layout (auto migrated):
5864 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5865 # for backwards compatibility
5866 #
5867 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5868 # - this is only created for newly multi-init-ed
5869 # repositories. Similar in spirit to the
5870 # --use-separate-remotes option in git-clone (now default)
5871 # - we do not automatically migrate to this (following
5872 # the example set by core git)
5873 #
5874 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
5875 # - newer, more-efficient format that uses 24-bytes per record
5876 # with no filler space.
5877 # - use xxd -c24 < .rev_map.$UUID to view and debug
5878 # - This is a one-way migration, repositories updated to the
5879 # new format will not be able to use old git-svn without
5880 # rebuilding the .rev_db. Rebuilding the rev_db is not
5881 # possible if noMetadata or useSvmProps are set; but should
5882 # be no problem for users that use the (sensible) defaults.
5883 use strict;
5884 use warnings;
5885 use Carp qw/croak/;
5886 use File::Path qw/mkpath/;
5887 use File::Basename qw/dirname basename/;
5888 use vars qw/$_minimize/;
5890 sub migrate_from_v0 {
5891 my $git_dir = $ENV{GIT_DIR};
5892 return undef unless -d $git_dir;
5893 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5894 my $migrated = 0;
5895 while (<$fh>) {
5896 chomp;
5897 my ($id, $orig_ref) = ($_, $_);
5898 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5899 next unless -f "$git_dir/$id/info/url";
5900 my $new_ref = "refs/remotes/$id";
5901 if (::verify_ref("$new_ref^0")) {
5902 print STDERR "W: $orig_ref is probably an old ",
5903 "branch used by an ancient version of ",
5904 "git-svn.\n",
5905 "However, $new_ref also exists.\n",
5906 "We will not be able ",
5907 "to use this branch until this ",
5908 "ambiguity is resolved.\n";
5909 next;
5910 }
5911 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5912 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5913 command_noisy('update-ref', $new_ref, $orig_ref);
5914 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5915 $migrated++;
5916 }
5917 command_close_pipe($fh, $ctx);
5918 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5919 $migrated;
5920 }
5922 sub migrate_from_v1 {
5923 my $git_dir = $ENV{GIT_DIR};
5924 my $migrated = 0;
5925 return $migrated unless -d $git_dir;
5926 my $svn_dir = "$git_dir/svn";
5928 # just in case somebody used 'svn' as their $id at some point...
5929 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5931 print STDERR "Migrating from a git-svn v1 layout...\n";
5932 mkpath([$svn_dir]);
5933 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5934 "$svn_dir\n\t(required for this version ",
5935 "($::VERSION) of git-svn) does not exist.\n";
5936 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5937 while (<$fh>) {
5938 my $x = $_;
5939 next unless $x =~ s#^refs/remotes/##;
5940 chomp $x;
5941 next unless -f "$git_dir/$x/info/url";
5942 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5943 next unless $u;
5944 my $dn = dirname("$git_dir/svn/$x");
5945 mkpath([$dn]) unless -d $dn;
5946 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5947 mkpath(["$git_dir/svn/svn"]);
5948 print STDERR " - $git_dir/$x/info => ",
5949 "$git_dir/svn/$x/info\n";
5950 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5951 croak "$!: $x";
5952 # don't worry too much about these, they probably
5953 # don't exist with repos this old (save for index,
5954 # and we can easily regenerate that)
5955 foreach my $f (qw/unhandled.log index .rev_db/) {
5956 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5957 }
5958 } else {
5959 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5960 rename "$git_dir/$x", "$git_dir/svn/$x" or
5961 croak "$!: $x";
5962 }
5963 $migrated++;
5964 }
5965 command_close_pipe($fh, $ctx);
5966 print STDERR "Done migrating from a git-svn v1 layout\n";
5967 $migrated;
5968 }
5970 sub read_old_urls {
5971 my ($l_map, $pfx, $path) = @_;
5972 my @dir;
5973 foreach (<$path/*>) {
5974 if (-r "$_/info/url") {
5975 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5976 my $ref_id = $pfx . basename $_;
5977 my $url = ::file_to_s("$_/info/url");
5978 $l_map->{$ref_id} = $url;
5979 } elsif (-d $_) {
5980 push @dir, $_;
5981 }
5982 }
5983 foreach (@dir) {
5984 my $x = $_;
5985 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5986 read_old_urls($l_map, $x, $_);
5987 }
5988 }
5990 sub migrate_from_v2 {
5991 my @cfg = command(qw/config -l/);
5992 return if grep /^svn-remote\..+\.url=/, @cfg;
5993 my %l_map;
5994 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5995 my $migrated = 0;
5997 foreach my $ref_id (sort keys %l_map) {
5998 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5999 if ($@) {
6000 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
6001 }
6002 $migrated++;
6003 }
6004 $migrated;
6005 }
6007 sub minimize_connections {
6008 my $r = Git::SVN::read_all_remotes();
6009 my $new_urls = {};
6010 my $root_repos = {};
6011 foreach my $repo_id (keys %$r) {
6012 my $url = $r->{$repo_id}->{url} or next;
6013 my $fetch = $r->{$repo_id}->{fetch} or next;
6014 my $ra = Git::SVN::Ra->new($url);
6016 # skip existing cases where we already connect to the root
6017 if (($ra->{url} eq $ra->{repos_root}) ||
6018 ($ra->{repos_root} eq $repo_id)) {
6019 $root_repos->{$ra->{url}} = $repo_id;
6020 next;
6021 }
6023 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
6024 my $root_path = $ra->{url};
6025 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
6026 foreach my $path (keys %$fetch) {
6027 my $ref_id = $fetch->{$path};
6028 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
6030 # make sure we can read when connecting to
6031 # a higher level of a repository
6032 my ($last_rev, undef) = $gs->last_rev_commit;
6033 if (!defined $last_rev) {
6034 $last_rev = eval {
6035 $root_ra->get_latest_revnum;
6036 };
6037 next if $@;
6038 }
6039 my $new = $root_path;
6040 $new .= length $path ? "/$path" : '';
6041 eval {
6042 $root_ra->get_log([$new], $last_rev, $last_rev,
6043 0, 0, 1, sub { });
6044 };
6045 next if $@;
6046 $new_urls->{$ra->{repos_root}}->{$new} =
6047 { ref_id => $ref_id,
6048 old_repo_id => $repo_id,
6049 old_path => $path };
6050 }
6051 }
6053 my @emptied;
6054 foreach my $url (keys %$new_urls) {
6055 # see if we can re-use an existing [svn-remote "repo_id"]
6056 # instead of creating a(n ugly) new section:
6057 my $repo_id = $root_repos->{$url} || $url;
6059 my $fetch = $new_urls->{$url};
6060 foreach my $path (keys %$fetch) {
6061 my $x = $fetch->{$path};
6062 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
6063 my $pfx = "svn-remote.$x->{old_repo_id}";
6065 my $old_fetch = quotemeta("$x->{old_path}:".
6066 "$x->{ref_id}");
6067 command_noisy(qw/config --unset/,
6068 "$pfx.fetch", '^'. $old_fetch . '$');
6069 delete $r->{$x->{old_repo_id}}->
6070 {fetch}->{$x->{old_path}};
6071 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6072 command_noisy(qw/config --unset/,
6073 "$pfx.url");
6074 push @emptied, $x->{old_repo_id}
6075 }
6076 }
6077 }
6078 if (@emptied) {
6079 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6080 print STDERR <<EOF;
6081 The following [svn-remote] sections in your config file ($file) are empty
6082 and can be safely removed:
6083 EOF
6084 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6085 }
6086 }
6088 sub migration_check {
6089 migrate_from_v0();
6090 migrate_from_v1();
6091 migrate_from_v2();
6092 minimize_connections() if $_minimize;
6093 }
6095 package Git::IndexInfo;
6096 use strict;
6097 use warnings;
6098 use Git qw/command_input_pipe command_close_pipe/;
6100 sub new {
6101 my ($class) = @_;
6102 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6103 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6104 }
6106 sub remove {
6107 my ($self, $path) = @_;
6108 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6109 return ++$self->{nr};
6110 }
6111 undef;
6112 }
6114 sub update {
6115 my ($self, $mode, $hash, $path) = @_;
6116 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6117 return ++$self->{nr};
6118 }
6119 undef;
6120 }
6122 sub DESTROY {
6123 my ($self) = @_;
6124 command_close_pipe($self->{gui}, $self->{ctx});
6125 }
6127 package Git::SVN::GlobSpec;
6128 use strict;
6129 use warnings;
6131 sub new {
6132 my ($class, $glob, $pattern_ok) = @_;
6133 my $re = $glob;
6134 $re =~ s!/+$!!g; # no need for trailing slashes
6135 my (@left, @right, @patterns);
6136 my $state = "left";
6137 my $die_msg = "Only one set of wildcard directories " .
6138 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6139 for my $part (split(m|/|, $glob)) {
6140 if ($part =~ /\*/ && $part ne "*") {
6141 die "Invalid pattern in '$glob': $part\n";
6142 } elsif ($pattern_ok && $part =~ /[{}]/ &&
6143 $part !~ /^\{[^{}]+\}/) {
6144 die "Invalid pattern in '$glob': $part\n";
6145 }
6146 if ($part eq "*") {
6147 die $die_msg if $state eq "right";
6148 $state = "pattern";
6149 push(@patterns, "[^/]*");
6150 } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6151 die $die_msg if $state eq "right";
6152 $state = "pattern";
6153 my $p = quotemeta($1);
6154 $p =~ s/\\,/|/g;
6155 push(@patterns, "(?:$p)");
6156 } else {
6157 if ($state eq "left") {
6158 push(@left, $part);
6159 } else {
6160 push(@right, $part);
6161 $state = "right";
6162 }
6163 }
6164 }
6165 my $depth = @patterns;
6166 if ($depth == 0) {
6167 die "One '*' is needed in glob: '$glob'\n";
6168 }
6169 my $left = join('/', @left);
6170 my $right = join('/', @right);
6171 $re = join('/', @patterns);
6172 $re = join('\/',
6173 grep(length, quotemeta($left), "($re)", quotemeta($right)));
6174 my $left_re = qr/^\/\Q$left\E(\/|$)/;
6175 bless { left => $left, right => $right, left_regex => $left_re,
6176 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6177 }
6179 sub full_path {
6180 my ($self, $path) = @_;
6181 return (length $self->{left} ? "$self->{left}/" : '') .
6182 $path . (length $self->{right} ? "/$self->{right}" : '');
6183 }
6185 __END__
6187 Data structures:
6190 $remotes = { # returned by read_all_remotes()
6191 'svn' => {
6192 # svn-remote.svn.url=https://svn.musicpd.org
6193 url => 'https://svn.musicpd.org',
6194 # svn-remote.svn.fetch=mpd/trunk:trunk
6195 fetch => {
6196 'mpd/trunk' => 'trunk',
6197 },
6198 # svn-remote.svn.tags=mpd/tags/*:tags/*
6199 tags => {
6200 path => {
6201 left => 'mpd/tags',
6202 right => '',
6203 regex => qr!mpd/tags/([^/]+)$!,
6204 glob => 'tags/*',
6205 },
6206 ref => {
6207 left => 'tags',
6208 right => '',
6209 regex => qr!tags/([^/]+)$!,
6210 glob => 'tags/*',
6211 },
6212 }
6213 }
6214 };
6216 $log_entry hashref as returned by libsvn_log_entry()
6217 {
6218 log => 'whitespace-formatted log entry
6219 ', # trailing newline is preserved
6220 revision => '8', # integer
6221 date => '2004-02-24T17:01:44.108345Z', # commit date
6222 author => 'committer name'
6223 };
6226 # this is generated by generate_diff();
6227 @mods = array of diff-index line hashes, each element represents one line
6228 of diff-index output
6230 diff-index line ($m hash)
6231 {
6232 mode_a => first column of diff-index output, no leading ':',
6233 mode_b => second column of diff-index output,
6234 sha1_b => sha1sum of the final blob,
6235 chg => change type [MCRADT],
6236 file_a => original file name of a file (iff chg is 'C' or 'R')
6237 file_b => new/current file name of a file (any chg)
6238 }
6239 ;
6241 # retval of read_url_paths{,_all}();
6242 $l_map = {
6243 # repository root url
6244 'https://svn.musicpd.org' => {
6245 # repository path # GIT_SVN_ID
6246 'mpd/trunk' => 'trunk',
6247 'mpd/tags/0.11.5' => 'tags/0.11.5',
6248 },
6249 }
6251 Notes:
6252 I don't trust the each() function on unless I created %hash myself
6253 because the internal iterator may not have started at base.