![[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 warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision $_repository
8 $_q $_authors $_authors_prog %users/;
9 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10 $VERSION = '@@GIT_VERSION@@';
12 # From which subdir have we been invoked?
13 my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15 } || '';
17 my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
18 $ENV{GIT_DIR} ||= '.git';
19 $Git::SVN::default_repo_id = 'svn';
20 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
21 $Git::SVN::Ra::_log_window_size = 100;
22 $Git::SVN::_minimize_url = 'unset';
24 if (! exists $ENV{SVN_SSH}) {
25 if (exists $ENV{GIT_SSH}) {
26 $ENV{SVN_SSH} = $ENV{GIT_SSH};
27 if ($^O eq 'msys') {
28 $ENV{SVN_SSH} =~ s/\\/\\\\/g;
29 }
30 }
31 }
33 $Git::SVN::Log::TZ = $ENV{TZ};
34 $ENV{TZ} = 'UTC';
35 $| = 1; # unbuffer STDOUT
37 sub fatal (@) { print STDERR "@_\n"; exit 1 }
38 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
39 require SVN::Ra;
40 require SVN::Delta;
41 if ($SVN::Core::VERSION lt '1.1.0') {
42 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
43 }
44 my $can_compress = eval { require Compress::Zlib; 1};
45 push @Git::SVN::Ra::ISA, 'SVN::Ra';
46 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
47 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
48 use Carp qw/croak/;
49 use Digest::MD5;
50 use IO::File qw//;
51 use File::Basename qw/dirname basename/;
52 use File::Path qw/mkpath/;
53 use File::Spec;
54 use File::Find;
55 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
56 use IPC::Open3;
57 use Git;
59 BEGIN {
60 # import functions from Git into our packages, en masse
61 no strict 'refs';
62 foreach (qw/command command_oneline command_noisy command_output_pipe
63 command_input_pipe command_close_pipe
64 command_bidi_pipe command_close_bidi_pipe/) {
65 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
66 Git::SVN::Migration Git::SVN::Log Git::SVN),
67 __PACKAGE__) {
68 *{"${package}::$_"} = \&{"Git::$_"};
69 }
70 }
71 }
73 my ($SVN);
75 $sha1 = qr/[a-f\d]{40}/;
76 $sha1_short = qr/[a-f\d]{4,40}/;
77 my ($_stdin, $_help, $_edit,
78 $_message, $_file, $_branch_dest,
79 $_template, $_shared,
80 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
81 $_merge, $_strategy, $_dry_run, $_local,
82 $_prefix, $_no_checkout, $_url, $_verbose,
83 $_git_format, $_commit_url, $_tag);
84 $Git::SVN::_follow_parent = 1;
85 $_q ||= 0;
86 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
87 'config-dir=s' => \$Git::SVN::Ra::config_dir,
88 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
89 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
90 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
91 'authors-file|A=s' => \$_authors,
92 'authors-prog=s' => \$_authors_prog,
93 'repack:i' => \$Git::SVN::_repack,
94 'noMetadata' => \$Git::SVN::_no_metadata,
95 'useSvmProps' => \$Git::SVN::_use_svm_props,
96 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
97 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
98 'no-checkout' => \$_no_checkout,
99 'quiet|q+' => \$_q,
100 'repack-flags|repack-args|repack-opts=s' =>
101 \$Git::SVN::_repack_flags,
102 'use-log-author' => \$Git::SVN::_use_log_author,
103 'add-author-from' => \$Git::SVN::_add_author_from,
104 'localtime' => \$Git::SVN::_localtime,
105 %remote_opts );
107 my ($_trunk, @_tags, @_branches, $_stdlayout);
108 my %icv;
109 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
110 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
111 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
112 'stdlayout|s' => \$_stdlayout,
113 'minimize-url|m!' => \$Git::SVN::_minimize_url,
114 'no-metadata' => sub { $icv{noMetadata} = 1 },
115 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
116 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
117 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
118 %remote_opts );
119 my %cmt_opts = ( 'edit|e' => \$_edit,
120 'rmdir' => \$SVN::Git::Editor::_rmdir,
121 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
122 'l=i' => \$SVN::Git::Editor::_rename_limit,
123 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
124 );
126 my %cmd = (
127 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
128 { 'revision|r=s' => \$_revision,
129 'fetch-all|all' => \$_fetch_all,
130 'parent|p' => \$_fetch_parent,
131 %fc_opts } ],
132 clone => [ \&cmd_clone, "Initialize and fetch revisions",
133 { 'revision|r=s' => \$_revision,
134 %fc_opts, %init_opts } ],
135 init => [ \&cmd_init, "Initialize a repo for tracking" .
136 " (requires URL argument)",
137 \%init_opts ],
138 'multi-init' => [ \&cmd_multi_init,
139 "Deprecated alias for ".
140 "'$0 init -T<trunk> -b<branches> -t<tags>'",
141 \%init_opts ],
142 dcommit => [ \&cmd_dcommit,
143 'Commit several diffs to merge with upstream',
144 { 'merge|m|M' => \$_merge,
145 'strategy|s=s' => \$_strategy,
146 'verbose|v' => \$_verbose,
147 'dry-run|n' => \$_dry_run,
148 'fetch-all|all' => \$_fetch_all,
149 'commit-url=s' => \$_commit_url,
150 'revision|r=i' => \$_revision,
151 'no-rebase' => \$_no_rebase,
152 %cmt_opts, %fc_opts } ],
153 branch => [ \&cmd_branch,
154 'Create a branch in the SVN repository',
155 { 'message|m=s' => \$_message,
156 'destination|d=s' => \$_branch_dest,
157 'dry-run|n' => \$_dry_run,
158 'tag|t' => \$_tag } ],
159 tag => [ sub { $_tag = 1; cmd_branch(@_) },
160 'Create a tag in the SVN repository',
161 { 'message|m=s' => \$_message,
162 'destination|d=s' => \$_branch_dest,
163 'dry-run|n' => \$_dry_run } ],
164 'set-tree' => [ \&cmd_set_tree,
165 "Set an SVN repository to a git tree-ish",
166 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
167 'create-ignore' => [ \&cmd_create_ignore,
168 'Create a .gitignore per svn:ignore',
169 { 'revision|r=i' => \$_revision
170 } ],
171 'propget' => [ \&cmd_propget,
172 'Print the value of a property on a file or directory',
173 { 'revision|r=i' => \$_revision } ],
174 'proplist' => [ \&cmd_proplist,
175 'List all properties of a file or directory',
176 { 'revision|r=i' => \$_revision } ],
177 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
178 { 'revision|r=i' => \$_revision
179 } ],
180 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
181 { 'revision|r=i' => \$_revision
182 } ],
183 'multi-fetch' => [ \&cmd_multi_fetch,
184 "Deprecated alias for $0 fetch --all",
185 { 'revision|r=s' => \$_revision, %fc_opts } ],
186 'migrate' => [ sub { },
187 # no-op, we automatically run this anyways,
188 'Migrate configuration/metadata/layout from
189 previous versions of git-svn',
190 { 'minimize' => \$Git::SVN::Migration::_minimize,
191 %remote_opts } ],
192 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
193 { 'limit=i' => \$Git::SVN::Log::limit,
194 'revision|r=s' => \$_revision,
195 'verbose|v' => \$Git::SVN::Log::verbose,
196 'incremental' => \$Git::SVN::Log::incremental,
197 'oneline' => \$Git::SVN::Log::oneline,
198 'show-commit' => \$Git::SVN::Log::show_commit,
199 'non-recursive' => \$Git::SVN::Log::non_recursive,
200 'authors-file|A=s' => \$_authors,
201 'color' => \$Git::SVN::Log::color,
202 'pager=s' => \$Git::SVN::Log::pager
203 } ],
204 'find-rev' => [ \&cmd_find_rev,
205 "Translate between SVN revision numbers and tree-ish",
206 {} ],
207 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
208 { 'merge|m|M' => \$_merge,
209 'verbose|v' => \$_verbose,
210 'strategy|s=s' => \$_strategy,
211 'local|l' => \$_local,
212 'fetch-all|all' => \$_fetch_all,
213 'dry-run|n' => \$_dry_run,
214 %fc_opts } ],
215 'commit-diff' => [ \&cmd_commit_diff,
216 'Commit a diff between two trees',
217 { 'message|m=s' => \$_message,
218 'file|F=s' => \$_file,
219 'revision|r=s' => \$_revision,
220 %cmt_opts } ],
221 'info' => [ \&cmd_info,
222 "Show info about the latest SVN revision
223 on the current branch",
224 { 'url' => \$_url, } ],
225 'blame' => [ \&Git::SVN::Log::cmd_blame,
226 "Show what revision and author last modified each line of a file",
227 { 'git-format' => \$_git_format } ],
228 'reset' => [ \&cmd_reset,
229 "Undo fetches back to the specified SVN revision",
230 { 'revision|r=s' => \$_revision,
231 'parent|p' => \$_fetch_parent } ],
232 'gc' => [ \&cmd_gc,
233 "Compress unhandled.log files in .git/svn and remove " .
234 "index files in .git/svn",
235 {} ],
236 );
238 my $cmd;
239 for (my $i = 0; $i < @ARGV; $i++) {
240 if (defined $cmd{$ARGV[$i]}) {
241 $cmd = $ARGV[$i];
242 splice @ARGV, $i, 1;
243 last;
244 } elsif ($ARGV[$i] eq 'help') {
245 $cmd = $ARGV[$i+1];
246 usage(0);
247 }
248 };
250 # make sure we're always running at the top-level working directory
251 unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
252 unless (-d $ENV{GIT_DIR}) {
253 if ($git_dir_user_set) {
254 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
255 "but it is not a directory\n";
256 }
257 my $git_dir = delete $ENV{GIT_DIR};
258 my $cdup = undef;
259 git_cmd_try {
260 $cdup = command_oneline(qw/rev-parse --show-cdup/);
261 $git_dir = '.' unless ($cdup);
262 chomp $cdup if ($cdup);
263 $cdup = "." unless ($cdup && length $cdup);
264 } "Already at toplevel, but $git_dir not found\n";
265 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
266 unless (-d $git_dir) {
267 die "$git_dir still not found after going to ",
268 "'$cdup'\n";
269 }
270 $ENV{GIT_DIR} = $git_dir;
271 }
272 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
273 }
275 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
277 read_git_config(\%opts);
278 if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
279 Getopt::Long::Configure('pass_through');
280 }
281 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
282 'minimize-connections' => \$Git::SVN::Migration::_minimize,
283 'id|i=s' => \$Git::SVN::default_ref_id,
284 'svn-remote|remote|R=s' => sub {
285 $Git::SVN::no_reuse_existing = 1;
286 $Git::SVN::default_repo_id = $_[1] });
287 exit 1 if (!$rv && $cmd && $cmd ne 'log');
289 usage(0) if $_help;
290 version() if $_version;
291 usage(1) unless defined $cmd;
292 load_authors() if $_authors;
293 if (defined $_authors_prog) {
294 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
295 }
297 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
298 Git::SVN::Migration::migration_check();
299 }
300 Git::SVN::init_vars();
301 eval {
302 Git::SVN::verify_remotes_sanity();
303 $cmd{$cmd}->[0]->(@ARGV);
304 };
305 fatal $@ if $@;
306 post_fetch_checkout();
307 exit 0;
309 ####################### primary functions ######################
310 sub usage {
311 my $exit = shift || 0;
312 my $fd = $exit ? \*STDERR : \*STDOUT;
313 print $fd <<"";
314 git-svn - bidirectional operations between a single Subversion tree and git
315 Usage: git svn <command> [options] [arguments]\n
317 print $fd "Available commands:\n" unless $cmd;
319 foreach (sort keys %cmd) {
320 next if $cmd && $cmd ne $_;
321 next if /^multi-/; # don't show deprecated commands
322 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
323 foreach (sort keys %{$cmd{$_}->[2]}) {
324 # mixed-case options are for .git/config only
325 next if /[A-Z]/ && /^[a-z]+$/i;
326 # prints out arguments as they should be passed:
327 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
328 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
329 "--$_" : "-$_" }
330 split /\|/,$_)," $x\n";
331 }
332 }
333 print $fd <<"";
334 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
335 arbitrary identifier if you're tracking multiple SVN branches/repositories in
336 one git repository and want to keep them separate. See git-svn(1) for more
337 information.
339 exit $exit;
340 }
342 sub version {
343 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
344 exit 0;
345 }
347 sub do_git_init_db {
348 unless (-d $ENV{GIT_DIR}) {
349 my @init_db = ('init');
350 push @init_db, "--template=$_template" if defined $_template;
351 if (defined $_shared) {
352 if ($_shared =~ /[a-z]/) {
353 push @init_db, "--shared=$_shared";
354 } else {
355 push @init_db, "--shared";
356 }
357 }
358 command_noisy(@init_db);
359 $_repository = Git->repository(Repository => ".git");
360 }
361 command_noisy('config', 'core.autocrlf', 'false');
362 my $set;
363 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
364 foreach my $i (keys %icv) {
365 die "'$set' and '$i' cannot both be set\n" if $set;
366 next unless defined $icv{$i};
367 command_noisy('config', "$pfx.$i", $icv{$i});
368 $set = $i;
369 }
370 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
371 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
372 if defined $$ignore_regex;
373 }
375 sub init_subdir {
376 my $repo_path = shift or return;
377 mkpath([$repo_path]) unless -d $repo_path;
378 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
379 $ENV{GIT_DIR} = '.git';
380 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
381 }
383 sub cmd_clone {
384 my ($url, $path) = @_;
385 if (!defined $path &&
386 (defined $_trunk || @_branches || @_tags ||
387 defined $_stdlayout) &&
388 $url !~ m#^[a-z\+]+://#) {
389 $path = $url;
390 }
391 $path = basename($url) if !defined $path || !length $path;
392 cmd_init($url, $path);
393 Git::SVN::fetch_all($Git::SVN::default_repo_id);
394 command_oneline('config', 'svn.authorsfile', $_authors) if $_authors;
395 }
397 sub cmd_init {
398 if (defined $_stdlayout) {
399 $_trunk = 'trunk' if (!defined $_trunk);
400 @_tags = 'tags' if (! @_tags);
401 @_branches = 'branches' if (! @_branches);
402 }
403 if (defined $_trunk || @_branches || @_tags) {
404 return cmd_multi_init(@_);
405 }
406 my $url = shift or die "SVN repository location required ",
407 "as a command-line argument\n";
408 $url = canonicalize_url($url);
409 init_subdir(@_);
410 do_git_init_db();
412 if ($Git::SVN::_minimize_url eq 'unset') {
413 $Git::SVN::_minimize_url = 0;
414 }
416 Git::SVN->init($url);
417 }
419 sub cmd_fetch {
420 if (grep /^\d+=./, @_) {
421 die "'<rev>=<commit>' fetch arguments are ",
422 "no longer supported.\n";
423 }
424 my ($remote) = @_;
425 if (@_ > 1) {
426 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
427 }
428 if ($_fetch_parent) {
429 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
430 unless ($gs) {
431 die "Unable to determine upstream SVN information from ",
432 "working tree history\n";
433 }
434 # just fetch, don't checkout.
435 $_no_checkout = 'true';
436 $_fetch_all ? $gs->fetch_all : $gs->fetch;
437 } elsif ($_fetch_all) {
438 cmd_multi_fetch();
439 } else {
440 $remote ||= $Git::SVN::default_repo_id;
441 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
442 }
443 }
445 sub cmd_set_tree {
446 my (@commits) = @_;
447 if ($_stdin || !@commits) {
448 print "Reading from stdin...\n";
449 @commits = ();
450 while (<STDIN>) {
451 if (/\b($sha1_short)\b/o) {
452 unshift @commits, $1;
453 }
454 }
455 }
456 my @revs;
457 foreach my $c (@commits) {
458 my @tmp = command('rev-parse',$c);
459 if (scalar @tmp == 1) {
460 push @revs, $tmp[0];
461 } elsif (scalar @tmp > 1) {
462 push @revs, reverse(command('rev-list',@tmp));
463 } else {
464 fatal "Failed to rev-parse $c";
465 }
466 }
467 my $gs = Git::SVN->new;
468 my ($r_last, $cmt_last) = $gs->last_rev_commit;
469 $gs->fetch;
470 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
471 fatal "There are new revisions that were fetched ",
472 "and need to be merged (or acknowledged) ",
473 "before committing.\nlast rev: $r_last\n",
474 " current: $gs->{last_rev}";
475 }
476 $gs->set_tree($_) foreach @revs;
477 print "Done committing ",scalar @revs," revisions to SVN\n";
478 unlink $gs->{index};
479 }
481 sub cmd_dcommit {
482 my $head = shift;
483 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
484 'Cannot dcommit with a dirty index. Commit your changes first, '
485 . "or stash them with `git stash'.\n";
486 $head ||= 'HEAD';
488 my $old_head;
489 if ($head ne 'HEAD') {
490 $old_head = eval {
491 command_oneline([qw/symbolic-ref -q HEAD/])
492 };
493 if ($old_head) {
494 $old_head =~ s{^refs/heads/}{};
495 } else {
496 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
497 }
498 command(['checkout', $head], STDERR => 0);
499 }
501 my @refs;
502 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
503 unless ($gs) {
504 die "Unable to determine upstream SVN information from ",
505 "$head history.\nPerhaps the repository is empty.";
506 }
508 if (defined $_commit_url) {
509 $url = $_commit_url;
510 } else {
511 $url = eval { command_oneline('config', '--get',
512 "svn-remote.$gs->{repo_id}.commiturl") };
513 if (!$url) {
514 $url = $gs->full_url
515 }
516 }
518 my $last_rev = $_revision if defined $_revision;
519 if ($url) {
520 print "Committing to $url ...\n";
521 }
522 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
523 if ($_no_rebase && scalar(@$linear_refs) > 1) {
524 warn "Attempting to commit more than one change while ",
525 "--no-rebase is enabled.\n",
526 "If these changes depend on each other, re-running ",
527 "without --no-rebase may be required."
528 }
529 my $expect_url = $url;
530 Git::SVN::remove_username($expect_url);
531 while (1) {
532 my $d = shift @$linear_refs or last;
533 unless (defined $last_rev) {
534 (undef, $last_rev, undef) = cmt_metadata("$d~1");
535 unless (defined $last_rev) {
536 fatal "Unable to extract revision information ",
537 "from commit $d~1";
538 }
539 }
540 if ($_dry_run) {
541 print "diff-tree $d~1 $d\n";
542 } else {
543 my $cmt_rev;
544 my %ed_opts = ( r => $last_rev,
545 log => get_commit_entry($d)->{log},
546 ra => Git::SVN::Ra->new($url),
547 config => SVN::Core::config_get_config(
548 $Git::SVN::Ra::config_dir
549 ),
550 tree_a => "$d~1",
551 tree_b => $d,
552 editor_cb => sub {
553 print "Committed r$_[0]\n";
554 $cmt_rev = $_[0];
555 },
556 svn_path => '');
557 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
558 print "No changes\n$d~1 == $d\n";
559 } elsif ($parents->{$d} && @{$parents->{$d}}) {
560 $gs->{inject_parents_dcommit}->{$cmt_rev} =
561 $parents->{$d};
562 }
563 $_fetch_all ? $gs->fetch_all : $gs->fetch;
564 $last_rev = $cmt_rev;
565 next if $_no_rebase;
567 # we always want to rebase against the current HEAD,
568 # not any head that was passed to us
569 my @diff = command('diff-tree', $d,
570 $gs->refname, '--');
571 my @finish;
572 if (@diff) {
573 @finish = rebase_cmd();
574 print STDERR "W: $d and ", $gs->refname,
575 " differ, using @finish:\n",
576 join("\n", @diff), "\n";
577 } else {
578 print "No changes between current HEAD and ",
579 $gs->refname,
580 "\nResetting to the latest ",
581 $gs->refname, "\n";
582 @finish = qw/reset --mixed/;
583 }
584 command_noisy(@finish, $gs->refname);
585 if (@diff) {
586 @refs = ();
587 my ($url_, $rev_, $uuid_, $gs_) =
588 working_head_info('HEAD', \@refs);
589 my ($linear_refs_, $parents_) =
590 linearize_history($gs_, \@refs);
591 if (scalar(@$linear_refs) !=
592 scalar(@$linear_refs_)) {
593 fatal "# of revisions changed ",
594 "\nbefore:\n",
595 join("\n", @$linear_refs),
596 "\n\nafter:\n",
597 join("\n", @$linear_refs_), "\n",
598 'If you are attempting to commit ',
599 "merges, try running:\n\t",
600 'git rebase --interactive',
601 '--preserve-merges ',
602 $gs->refname,
603 "\nBefore dcommitting";
604 }
605 if ($url_ ne $expect_url) {
606 if ($url_ eq $gs->metadata_url) {
607 print
608 "Accepting rewritten URL:",
609 " $url_\n";
610 } else {
611 fatal
612 "URL mismatch after rebase:",
613 " $url_ != $expect_url";
614 }
615 }
616 if ($uuid_ ne $uuid) {
617 fatal "uuid mismatch after rebase: ",
618 "$uuid_ != $uuid";
619 }
620 # remap parents
621 my (%p, @l, $i);
622 for ($i = 0; $i < scalar @$linear_refs; $i++) {
623 my $new = $linear_refs_->[$i] or next;
624 $p{$new} =
625 $parents->{$linear_refs->[$i]};
626 push @l, $new;
627 }
628 $parents = \%p;
629 $linear_refs = \@l;
630 }
631 }
632 }
634 if ($old_head) {
635 my $new_head = command_oneline(qw/rev-parse HEAD/);
636 my $new_is_symbolic = eval {
637 command_oneline(qw/symbolic-ref -q HEAD/);
638 };
639 if ($new_is_symbolic) {
640 print "dcommitted the branch ", $head, "\n";
641 } else {
642 print "dcommitted on a detached HEAD because you gave ",
643 "a revision argument.\n",
644 "The rewritten commit is: ", $new_head, "\n";
645 }
646 command(['checkout', $old_head], STDERR => 0);
647 }
649 unlink $gs->{index};
650 }
652 sub cmd_branch {
653 my ($branch_name, $head) = @_;
655 unless (defined $branch_name && length $branch_name) {
656 die(($_tag ? "tag" : "branch") . " name required\n");
657 }
658 $head ||= 'HEAD';
660 my ($src, $rev, undef, $gs) = working_head_info($head);
662 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
663 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
664 my $glob;
665 if ($#{$allglobs} == 0) {
666 $glob = $allglobs->[0];
667 } else {
668 unless(defined $_branch_dest) {
669 die "Multiple ",
670 $_tag ? "tag" : "branch",
671 " paths defined for Subversion repository.\n",
672 "You must specify where you want to create the ",
673 $_tag ? "tag" : "branch",
674 " with the --destination argument.\n";
675 }
676 foreach my $g (@{$allglobs}) {
677 # SVN::Git::Editor could probably be moved to Git.pm..
678 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
679 if ($_branch_dest =~ /$re/) {
680 $glob = $g;
681 last;
682 }
683 }
684 unless (defined $glob) {
685 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
686 foreach my $g (@{$allglobs}) {
687 $g->{path}->{left} =~ /$dest_re/ or next;
688 if (defined $glob) {
689 die "Ambiguous destination: ",
690 $_branch_dest, "\nmatches both '",
691 $glob->{path}->{left}, "' and '",
692 $g->{path}->{left}, "'\n";
693 }
694 $glob = $g;
695 }
696 unless (defined $glob) {
697 die "Unknown ",
698 $_tag ? "tag" : "branch",
699 " destination $_branch_dest\n";
700 }
701 }
702 }
703 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
704 my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
706 my $ctx = SVN::Client->new(
707 auth => Git::SVN::Ra::_auth_providers(),
708 log_msg => sub {
709 ${ $_[0] } = defined $_message
710 ? $_message
711 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
712 . $branch_name;
713 },
714 );
716 eval {
717 $ctx->ls($dst, 'HEAD', 0);
718 } and die "branch ${branch_name} already exists\n";
720 print "Copying ${src} at r${rev} to ${dst}...\n";
721 $ctx->copy($src, $rev, $dst)
722 unless $_dry_run;
724 $gs->fetch_all;
725 }
727 sub cmd_find_rev {
728 my $revision_or_hash = shift or die "SVN or git revision required ",
729 "as a command-line argument\n";
730 my $result;
731 if ($revision_or_hash =~ /^r\d+$/) {
732 my $head = shift;
733 $head ||= 'HEAD';
734 my @refs;
735 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
736 unless ($gs) {
737 die "Unable to determine upstream SVN information from ",
738 "$head history\n";
739 }
740 my $desired_revision = substr($revision_or_hash, 1);
741 $result = $gs->rev_map_get($desired_revision, $uuid);
742 } else {
743 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
744 $result = $rev;
745 }
746 print "$result\n" if $result;
747 }
749 sub cmd_rebase {
750 command_noisy(qw/update-index --refresh/);
751 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
752 unless ($gs) {
753 die "Unable to determine upstream SVN information from ",
754 "working tree history\n";
755 }
756 if ($_dry_run) {
757 print "Remote Branch: " . $gs->refname . "\n";
758 print "SVN URL: " . $url . "\n";
759 return;
760 }
761 if (command(qw/diff-index HEAD --/)) {
762 print STDERR "Cannot rebase with uncommited changes:\n";
763 command_noisy('status');
764 exit 1;
765 }
766 unless ($_local) {
767 # rebase will checkout for us, so no need to do it explicitly
768 $_no_checkout = 'true';
769 $_fetch_all ? $gs->fetch_all : $gs->fetch;
770 }
771 command_noisy(rebase_cmd(), $gs->refname);
772 }
774 sub cmd_show_ignore {
775 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
776 $gs ||= Git::SVN->new;
777 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
778 $gs->prop_walk($gs->{path}, $r, sub {
779 my ($gs, $path, $props) = @_;
780 print STDOUT "\n# $path\n";
781 my $s = $props->{'svn:ignore'} or return;
782 $s =~ s/[\r\n]+/\n/g;
783 $s =~ s/^\n+//;
784 chomp $s;
785 $s =~ s#^#$path#gm;
786 print STDOUT "$s\n";
787 });
788 }
790 sub cmd_show_externals {
791 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
792 $gs ||= Git::SVN->new;
793 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
794 $gs->prop_walk($gs->{path}, $r, sub {
795 my ($gs, $path, $props) = @_;
796 print STDOUT "\n# $path\n";
797 my $s = $props->{'svn:externals'} or return;
798 $s =~ s/[\r\n]+/\n/g;
799 chomp $s;
800 $s =~ s#^#$path#gm;
801 print STDOUT "$s\n";
802 });
803 }
805 sub cmd_create_ignore {
806 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
807 $gs ||= Git::SVN->new;
808 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
809 $gs->prop_walk($gs->{path}, $r, sub {
810 my ($gs, $path, $props) = @_;
811 # $path is of the form /path/to/dir/
812 $path = '.' . $path;
813 # SVN can have attributes on empty directories,
814 # which git won't track
815 mkpath([$path]) unless -d $path;
816 my $ignore = $path . '.gitignore';
817 my $s = $props->{'svn:ignore'} or return;
818 open(GITIGNORE, '>', $ignore)
819 or fatal("Failed to open `$ignore' for writing: $!");
820 $s =~ s/[\r\n]+/\n/g;
821 $s =~ s/^\n+//;
822 chomp $s;
823 # Prefix all patterns so that the ignore doesn't apply
824 # to sub-directories.
825 $s =~ s#^#/#gm;
826 print GITIGNORE "$s\n";
827 close(GITIGNORE)
828 or fatal("Failed to close `$ignore': $!");
829 command_noisy('add', '-f', $ignore);
830 });
831 }
833 sub canonicalize_path {
834 my ($path) = @_;
835 my $dot_slash_added = 0;
836 if (substr($path, 0, 1) ne "/") {
837 $path = "./" . $path;
838 $dot_slash_added = 1;
839 }
840 # File::Spec->canonpath doesn't collapse x/../y into y (for a
841 # good reason), so let's do this manually.
842 $path =~ s#/+#/#g;
843 $path =~ s#/\.(?:/|$)#/#g;
844 $path =~ s#/[^/]+/\.\.##g;
845 $path =~ s#/$##g;
846 $path =~ s#^\./## if $dot_slash_added;
847 $path =~ s#^/##;
848 $path =~ s#^\.$##;
849 return $path;
850 }
852 sub canonicalize_url {
853 my ($url) = @_;
854 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
855 return $url;
856 }
858 # get_svnprops(PATH)
859 # ------------------
860 # Helper for cmd_propget and cmd_proplist below.
861 sub get_svnprops {
862 my $path = shift;
863 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
864 $gs ||= Git::SVN->new;
866 # prefix THE PATH by the sub-directory from which the user
867 # invoked us.
868 $path = $cmd_dir_prefix . $path;
869 fatal("No such file or directory: $path") unless -e $path;
870 my $is_dir = -d $path ? 1 : 0;
871 $path = $gs->{path} . '/' . $path;
873 # canonicalize the path (otherwise libsvn will abort or fail to
874 # find the file)
875 $path = canonicalize_path($path);
877 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
878 my $props;
879 if ($is_dir) {
880 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
881 }
882 else {
883 (undef, $props) = $gs->ra->get_file($path, $r, undef);
884 }
885 return $props;
886 }
888 # cmd_propget (PROP, PATH)
889 # ------------------------
890 # Print the SVN property PROP for PATH.
891 sub cmd_propget {
892 my ($prop, $path) = @_;
893 $path = '.' if not defined $path;
894 usage(1) if not defined $prop;
895 my $props = get_svnprops($path);
896 if (not defined $props->{$prop}) {
897 fatal("`$path' does not have a `$prop' SVN property.");
898 }
899 print $props->{$prop} . "\n";
900 }
902 # cmd_proplist (PATH)
903 # -------------------
904 # Print the list of SVN properties for PATH.
905 sub cmd_proplist {
906 my $path = shift;
907 $path = '.' if not defined $path;
908 my $props = get_svnprops($path);
909 print "Properties on '$path':\n";
910 foreach (sort keys %{$props}) {
911 print " $_\n";
912 }
913 }
915 sub cmd_multi_init {
916 my $url = shift;
917 unless (defined $_trunk || @_branches || @_tags) {
918 usage(1);
919 }
921 $_prefix = '' unless defined $_prefix;
922 if (defined $url) {
923 $url = canonicalize_url($url);
924 init_subdir(@_);
925 }
926 do_git_init_db();
927 if (defined $_trunk) {
928 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
929 # try both old-style and new-style lookups:
930 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
931 unless ($gs_trunk) {
932 my ($trunk_url, $trunk_path) =
933 complete_svn_url($url, $_trunk);
934 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
935 undef, $trunk_ref);
936 }
937 }
938 return unless @_branches || @_tags;
939 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
940 foreach my $path (@_branches) {
941 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
942 }
943 foreach my $path (@_tags) {
944 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
945 }
946 }
948 sub cmd_multi_fetch {
949 my $remotes = Git::SVN::read_all_remotes();
950 foreach my $repo_id (sort keys %$remotes) {
951 if ($remotes->{$repo_id}->{url}) {
952 Git::SVN::fetch_all($repo_id, $remotes);
953 }
954 }
955 }
957 # this command is special because it requires no metadata
958 sub cmd_commit_diff {
959 my ($ta, $tb, $url) = @_;
960 my $usage = "Usage: $0 commit-diff -r<revision> ".
961 "<tree-ish> <tree-ish> [<URL>]";
962 fatal($usage) if (!defined $ta || !defined $tb);
963 my $svn_path = '';
964 if (!defined $url) {
965 my $gs = eval { Git::SVN->new };
966 if (!$gs) {
967 fatal("Needed URL or usable git-svn --id in ",
968 "the command-line\n", $usage);
969 }
970 $url = $gs->{url};
971 $svn_path = $gs->{path};
972 }
973 unless (defined $_revision) {
974 fatal("-r|--revision is a required argument\n", $usage);
975 }
976 if (defined $_message && defined $_file) {
977 fatal("Both --message/-m and --file/-F specified ",
978 "for the commit message.\n",
979 "I have no idea what you mean");
980 }
981 if (defined $_file) {
982 $_message = file_to_s($_file);
983 } else {
984 $_message ||= get_commit_entry($tb)->{log};
985 }
986 my $ra ||= Git::SVN::Ra->new($url);
987 my $r = $_revision;
988 if ($r eq 'HEAD') {
989 $r = $ra->get_latest_revnum;
990 } elsif ($r !~ /^\d+$/) {
991 die "revision argument: $r not understood by git-svn\n";
992 }
993 my %ed_opts = ( r => $r,
994 log => $_message,
995 ra => $ra,
996 tree_a => $ta,
997 tree_b => $tb,
998 editor_cb => sub { print "Committed r$_[0]\n" },
999 svn_path => $svn_path );
1000 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1001 print "No changes\n$ta == $tb\n";
1002 }
1003 }
1005 sub escape_uri_only {
1006 my ($uri) = @_;
1007 my @tmp;
1008 foreach (split m{/}, $uri) {
1009 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1010 push @tmp, $_;
1011 }
1012 join('/', @tmp);
1013 }
1015 sub escape_url {
1016 my ($url) = @_;
1017 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1018 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1019 $url = "$scheme://$domain$uri";
1020 }
1021 $url;
1022 }
1024 sub cmd_info {
1025 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1026 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1027 if (exists $_[1]) {
1028 die "Too many arguments specified\n";
1029 }
1031 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1033 if (!$file_type && !$diff_status) {
1034 print STDERR "svn: '$path' is not under version control\n";
1035 exit 1;
1036 }
1038 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1039 unless ($gs) {
1040 die "Unable to determine upstream SVN information from ",
1041 "working tree history\n";
1042 }
1044 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1045 $path = "." if $path eq "";
1047 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1049 if ($_url) {
1050 print escape_url($full_url), "\n";
1051 return;
1052 }
1054 my $result = "Path: $path\n";
1055 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1056 $result .= "URL: " . escape_url($full_url) . "\n";
1058 eval {
1059 my $repos_root = $gs->repos_root;
1060 Git::SVN::remove_username($repos_root);
1061 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1062 };
1063 if ($@) {
1064 $result .= "Repository Root: (offline)\n";
1065 }
1066 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1067 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1068 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1070 $result .= "Node Kind: " .
1071 ($file_type eq "dir" ? "directory" : "file") . "\n";
1073 my $schedule = $diff_status eq "A"
1074 ? "add"
1075 : ($diff_status eq "D" ? "delete" : "normal");
1076 $result .= "Schedule: $schedule\n";
1078 if ($diff_status eq "A") {
1079 print $result, "\n";
1080 return;
1081 }
1083 my ($lc_author, $lc_rev, $lc_date_utc);
1084 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1085 my $log = command_output_pipe(@args);
1086 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1087 while (<$log>) {
1088 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1089 $lc_author = $1;
1090 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1091 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1092 (undef, $lc_rev, undef) = ::extract_metadata($1);
1093 }
1094 }
1095 close $log;
1097 Git::SVN::Log::set_local_timezone();
1099 $result .= "Last Changed Author: $lc_author\n";
1100 $result .= "Last Changed Rev: $lc_rev\n";
1101 $result .= "Last Changed Date: " .
1102 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1104 if ($file_type ne "dir") {
1105 my $text_last_updated_date =
1106 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1107 $result .=
1108 "Text Last Updated: " .
1109 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1110 "\n";
1111 my $checksum;
1112 if ($diff_status eq "D") {
1113 my ($fh, $ctx) =
1114 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1115 if ($file_type eq "link") {
1116 my $file_name = <$fh>;
1117 $checksum = md5sum("link $file_name");
1118 } else {
1119 $checksum = md5sum($fh);
1120 }
1121 command_close_pipe($fh, $ctx);
1122 } elsif ($file_type eq "link") {
1123 my $file_name =
1124 command(qw(cat-file blob), "HEAD:$path");
1125 $checksum =
1126 md5sum("link " . $file_name);
1127 } else {
1128 open FILE, "<", $path or die $!;
1129 $checksum = md5sum(\*FILE);
1130 close FILE or die $!;
1131 }
1132 $result .= "Checksum: " . $checksum . "\n";
1133 }
1135 print $result, "\n";
1136 }
1138 sub cmd_reset {
1139 my $target = shift || $_revision or die "SVN revision required\n";
1140 $target = $1 if $target =~ /^r(\d+)$/;
1141 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1142 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1143 unless ($gs) {
1144 die "Unable to determine upstream SVN information from ".
1145 "history\n";
1146 }
1147 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1148 $gs->rev_map_set($r, $c, 'reset', $uuid);
1149 print "r$r = $c ($gs->{ref_id})\n";
1150 }
1152 sub cmd_gc {
1153 if (!$can_compress) {
1154 warn "Compress::Zlib could not be found; unhandled.log " .
1155 "files will not be compressed.\n";
1156 }
1157 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1158 }
1160 ########################### utility functions #########################
1162 sub rebase_cmd {
1163 my @cmd = qw/rebase/;
1164 push @cmd, '-v' if $_verbose;
1165 push @cmd, qw/--merge/ if $_merge;
1166 push @cmd, "--strategy=$_strategy" if $_strategy;
1167 @cmd;
1168 }
1170 sub post_fetch_checkout {
1171 return if $_no_checkout;
1172 my $gs = $Git::SVN::_head or return;
1173 return if verify_ref('refs/heads/master^0');
1175 # look for "trunk" ref if it exists
1176 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1177 my $fetch = $remote->{fetch};
1178 if ($fetch) {
1179 foreach my $p (keys %$fetch) {
1180 basename($fetch->{$p}) eq 'trunk' or next;
1181 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1182 last;
1183 }
1184 }
1186 my $valid_head = verify_ref('HEAD^0');
1187 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1188 return if ($valid_head || !verify_ref('HEAD^0'));
1190 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1191 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1192 return if -f $index;
1194 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1195 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1196 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1197 print STDERR "Checked out HEAD:\n ",
1198 $gs->full_url, " r", $gs->last_rev, "\n";
1199 }
1201 sub complete_svn_url {
1202 my ($url, $path) = @_;
1203 $path =~ s#/+$##;
1204 if ($path !~ m#^[a-z\+]+://#) {
1205 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1206 fatal("E: '$path' is not a complete URL ",
1207 "and a separate URL is not specified");
1208 }
1209 return ($url, $path);
1210 }
1211 return ($path, '');
1212 }
1214 sub complete_url_ls_init {
1215 my ($ra, $repo_path, $switch, $pfx) = @_;
1216 unless ($repo_path) {
1217 print STDERR "W: $switch not specified\n";
1218 return;
1219 }
1220 $repo_path =~ s#/+$##;
1221 if ($repo_path =~ m#^[a-z\+]+://#) {
1222 $ra = Git::SVN::Ra->new($repo_path);
1223 $repo_path = '';
1224 } else {
1225 $repo_path =~ s#^/+##;
1226 unless ($ra) {
1227 fatal("E: '$repo_path' is not a complete URL ",
1228 "and a separate URL is not specified");
1229 }
1230 }
1231 my $url = $ra->{url};
1232 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1233 my $k = "svn-remote.$gs->{repo_id}.url";
1234 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1235 if ($orig_url && ($orig_url ne $gs->{url})) {
1236 die "$k already set: $orig_url\n",
1237 "wanted to set to: $gs->{url}\n";
1238 }
1239 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1240 my $remote_path = "$gs->{path}/$repo_path";
1241 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1242 $remote_path =~ s#/+#/#g;
1243 $remote_path =~ s#^/##g;
1244 $remote_path .= "/*" if $remote_path !~ /\*/;
1245 my ($n) = ($switch =~ /^--(\w+)/);
1246 if (length $pfx && $pfx !~ m#/$#) {
1247 die "--prefix='$pfx' must have a trailing slash '/'\n";
1248 }
1249 command_noisy('config',
1250 '--add',
1251 "svn-remote.$gs->{repo_id}.$n",
1252 "$remote_path:refs/remotes/$pfx*" .
1253 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1254 }
1256 sub verify_ref {
1257 my ($ref) = @_;
1258 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1259 { STDERR => 0 }); };
1260 }
1262 sub get_tree_from_treeish {
1263 my ($treeish) = @_;
1264 # $treeish can be a symbolic ref, too:
1265 my $type = command_oneline(qw/cat-file -t/, $treeish);
1266 my $expected;
1267 while ($type eq 'tag') {
1268 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1269 }
1270 if ($type eq 'commit') {
1271 $expected = (grep /^tree /, command(qw/cat-file commit/,
1272 $treeish))[0];
1273 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1274 die "Unable to get tree from $treeish\n" unless $expected;
1275 } elsif ($type eq 'tree') {
1276 $expected = $treeish;
1277 } else {
1278 die "$treeish is a $type, expected tree, tag or commit\n";
1279 }
1280 return $expected;
1281 }
1283 sub get_commit_entry {
1284 my ($treeish) = shift;
1285 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1286 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1287 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1288 open my $log_fh, '>', $commit_editmsg or croak $!;
1290 my $type = command_oneline(qw/cat-file -t/, $treeish);
1291 if ($type eq 'commit' || $type eq 'tag') {
1292 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1293 $type, $treeish);
1294 my $in_msg = 0;
1295 my $author;
1296 my $saw_from = 0;
1297 my $msgbuf = "";
1298 while (<$msg_fh>) {
1299 if (!$in_msg) {
1300 $in_msg = 1 if (/^\s*$/);
1301 $author = $1 if (/^author (.*>)/);
1302 } elsif (/^git-svn-id: /) {
1303 # skip this for now, we regenerate the
1304 # correct one on re-fetch anyways
1305 # TODO: set *:merge properties or like...
1306 } else {
1307 if (/^From:/ || /^Signed-off-by:/) {
1308 $saw_from = 1;
1309 }
1310 $msgbuf .= $_;
1311 }
1312 }
1313 $msgbuf =~ s/\s+$//s;
1314 if ($Git::SVN::_add_author_from && defined($author)
1315 && !$saw_from) {
1316 $msgbuf .= "\n\nFrom: $author";
1317 }
1318 print $log_fh $msgbuf or croak $!;
1319 command_close_pipe($msg_fh, $ctx);
1320 }
1321 close $log_fh or croak $!;
1323 if ($_edit || ($type eq 'tree')) {
1324 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1325 # TODO: strip out spaces, comments, like git-commit.sh
1326 system($editor, $commit_editmsg);
1327 }
1328 rename $commit_editmsg, $commit_msg or croak $!;
1329 {
1330 require Encode;
1331 # SVN requires messages to be UTF-8 when entering the repo
1332 local $/;
1333 open $log_fh, '<', $commit_msg or croak $!;
1334 binmode $log_fh;
1335 chomp($log_entry{log} = <$log_fh>);
1337 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1338 my $msg = $log_entry{log};
1340 eval { $msg = Encode::decode($enc, $msg, 1) };
1341 if ($@) {
1342 die "Could not decode as $enc:\n", $msg,
1343 "\nPerhaps you need to set i18n.commitencoding\n";
1344 }
1346 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1347 die "Could not encode as UTF-8:\n$msg\n" if $@;
1349 $log_entry{log} = $msg;
1351 close $log_fh or croak $!;
1352 }
1353 unlink $commit_msg;
1354 \%log_entry;
1355 }
1357 sub s_to_file {
1358 my ($str, $file, $mode) = @_;
1359 open my $fd,'>',$file or croak $!;
1360 print $fd $str,"\n" or croak $!;
1361 close $fd or croak $!;
1362 chmod ($mode &~ umask, $file) if (defined $mode);
1363 }
1365 sub file_to_s {
1366 my $file = shift;
1367 open my $fd,'<',$file or croak "$!: file: $file\n";
1368 local $/;
1369 my $ret = <$fd>;
1370 close $fd or croak $!;
1371 $ret =~ s/\s*$//s;
1372 return $ret;
1373 }
1375 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1376 sub load_authors {
1377 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1378 my $log = $cmd eq 'log';
1379 while (<$authors>) {
1380 chomp;
1381 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1382 my ($user, $name, $email) = ($1, $2, $3);
1383 if ($log) {
1384 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1385 } else {
1386 $users{$user} = [$name, $email];
1387 }
1388 }
1389 close $authors or croak $!;
1390 }
1392 # convert GetOpt::Long specs for use by git-config
1393 sub read_git_config {
1394 my $opts = shift;
1395 my @config_only;
1396 foreach my $o (keys %$opts) {
1397 # if we have mixedCase and a long option-only, then
1398 # it's a config-only variable that we don't need for
1399 # the command-line.
1400 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1401 my $v = $opts->{$o};
1402 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1403 $key =~ s/-//g;
1404 my $arg = 'git config';
1405 $arg .= ' --int' if ($o =~ /[:=]i$/);
1406 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1407 if (ref $v eq 'ARRAY') {
1408 chomp(my @tmp = `$arg --get-all svn.$key`);
1409 @$v = @tmp if @tmp;
1410 } else {
1411 chomp(my $tmp = `$arg --get svn.$key`);
1412 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1413 $$v = $tmp;
1414 }
1415 }
1416 }
1417 delete @$opts{@config_only} if @config_only;
1418 }
1420 sub extract_metadata {
1421 my $id = shift or return (undef, undef, undef);
1422 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1423 \s([a-f\d\-]+)$/ix);
1424 if (!defined $rev || !$uuid || !$url) {
1425 # some of the original repositories I made had
1426 # identifiers like this:
1427 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1428 }
1429 return ($url, $rev, $uuid);
1430 }
1432 sub cmt_metadata {
1433 return extract_metadata((grep(/^git-svn-id: /,
1434 command(qw/cat-file commit/, shift)))[-1]);
1435 }
1437 sub cmt_sha2rev_batch {
1438 my %s2r;
1439 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1440 my $list = shift;
1442 foreach my $sha (@{$list}) {
1443 my $first = 1;
1444 my $size = 0;
1445 print $out $sha, "\n";
1447 while (my $line = <$in>) {
1448 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1449 last;
1450 } elsif ($first &&
1451 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1452 $first = 0;
1453 $size = $1;
1454 next;
1455 } elsif ($line =~ /^(git-svn-id: )/) {
1456 my (undef, $rev, undef) =
1457 extract_metadata($line);
1458 $s2r{$sha} = $rev;
1459 }
1461 $size -= length($line);
1462 last if ($size == 0);
1463 }
1464 }
1466 command_close_bidi_pipe($pid, $in, $out, $ctx);
1468 return \%s2r;
1469 }
1471 sub working_head_info {
1472 my ($head, $refs) = @_;
1473 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1474 my ($fh, $ctx) = command_output_pipe(@args, $head);
1475 my $hash;
1476 my %max;
1477 while (<$fh>) {
1478 if ( m{^commit ($::sha1)$} ) {
1479 unshift @$refs, $hash if $hash and $refs;
1480 $hash = $1;
1481 next;
1482 }
1483 next unless s{^\s*(git-svn-id:)}{$1};
1484 my ($url, $rev, $uuid) = extract_metadata($_);
1485 if (defined $url && defined $rev) {
1486 next if $max{$url} and $max{$url} < $rev;
1487 if (my $gs = Git::SVN->find_by_url($url)) {
1488 my $c = $gs->rev_map_get($rev, $uuid);
1489 if ($c && $c eq $hash) {
1490 close $fh; # break the pipe
1491 return ($url, $rev, $uuid, $gs);
1492 } else {
1493 $max{$url} ||= $gs->rev_map_max;
1494 }
1495 }
1496 }
1497 }
1498 command_close_pipe($fh, $ctx);
1499 (undef, undef, undef, undef);
1500 }
1502 sub read_commit_parents {
1503 my ($parents, $c) = @_;
1504 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1505 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1506 @{$parents->{$c}} = split(/ /, $p);
1507 }
1509 sub linearize_history {
1510 my ($gs, $refs) = @_;
1511 my %parents;
1512 foreach my $c (@$refs) {
1513 read_commit_parents(\%parents, $c);
1514 }
1516 my @linear_refs;
1517 my %skip = ();
1518 my $last_svn_commit = $gs->last_commit;
1519 foreach my $c (reverse @$refs) {
1520 next if $c eq $last_svn_commit;
1521 last if $skip{$c};
1523 unshift @linear_refs, $c;
1524 $skip{$c} = 1;
1526 # we only want the first parent to diff against for linear
1527 # history, we save the rest to inject when we finalize the
1528 # svn commit
1529 my $fp_a = verify_ref("$c~1");
1530 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1531 if (!$fp_a || !$fp_b) {
1532 die "Commit $c\n",
1533 "has no parent commit, and therefore ",
1534 "nothing to diff against.\n",
1535 "You should be working from a repository ",
1536 "originally created by git-svn\n";
1537 }
1538 if ($fp_a ne $fp_b) {
1539 die "$c~1 = $fp_a, however parsing commit $c ",
1540 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1541 }
1543 foreach my $p (@{$parents{$c}}) {
1544 $skip{$p} = 1;
1545 }
1546 }
1547 (\@linear_refs, \%parents);
1548 }
1550 sub find_file_type_and_diff_status {
1551 my ($path) = @_;
1552 return ('dir', '') if $path eq '';
1554 my $diff_output =
1555 command_oneline(qw(diff --cached --name-status --), $path) || "";
1556 my $diff_status = (split(' ', $diff_output))[0] || "";
1558 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1560 return (undef, undef) if !$diff_status && !$ls_tree;
1562 if ($diff_status eq "A") {
1563 return ("link", $diff_status) if -l $path;
1564 return ("dir", $diff_status) if -d $path;
1565 return ("file", $diff_status);
1566 }
1568 my $mode = (split(' ', $ls_tree))[0] || "";
1570 return ("link", $diff_status) if $mode eq "120000";
1571 return ("dir", $diff_status) if $mode eq "040000";
1572 return ("file", $diff_status);
1573 }
1575 sub md5sum {
1576 my $arg = shift;
1577 my $ref = ref $arg;
1578 my $md5 = Digest::MD5->new();
1579 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1580 $md5->addfile($arg) or croak $!;
1581 } elsif ($ref eq 'SCALAR') {
1582 $md5->add($$arg) or croak $!;
1583 } elsif (!$ref) {
1584 $md5->add($arg) or croak $!;
1585 } else {
1586 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1587 }
1588 return $md5->hexdigest();
1589 }
1591 sub gc_directory {
1592 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1593 my $out_filename = $_ . ".gz";
1594 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1595 binmode $in_fh;
1596 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1597 die "Unable to open $out_filename: $!\n";
1599 my $res;
1600 while ($res = sysread($in_fh, my $str, 1024)) {
1601 $gz->gzwrite($str) or
1602 die "Unable to write: ".$gz->gzerror()."!\n";
1603 }
1604 unlink $_ or die "unlink $File::Find::name: $!\n";
1605 } elsif (-f $_ && basename($_) eq "index") {
1606 unlink $_ or die "unlink $_: $!\n";
1607 }
1608 }
1610 package Git::SVN;
1611 use strict;
1612 use warnings;
1613 use Fcntl qw/:DEFAULT :seek/;
1614 use constant rev_map_fmt => 'NH40';
1615 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1616 $_repack $_repack_flags $_use_svm_props $_head
1617 $_use_svnsync_props $no_reuse_existing $_minimize_url
1618 $_use_log_author $_add_author_from $_localtime/;
1619 use Carp qw/croak/;
1620 use File::Path qw/mkpath/;
1621 use File::Copy qw/copy/;
1622 use IPC::Open3;
1624 my ($_gc_nr, $_gc_period);
1626 # properties that we do not log:
1627 my %SKIP_PROP;
1628 BEGIN {
1629 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1630 svn:special svn:executable
1631 svn:entry:committed-rev
1632 svn:entry:last-author
1633 svn:entry:uuid
1634 svn:entry:committed-date/;
1636 # some options are read globally, but can be overridden locally
1637 # per [svn-remote "..."] section. Command-line options will *NOT*
1638 # override options set in an [svn-remote "..."] section
1639 no strict 'refs';
1640 for my $option (qw/follow_parent no_metadata use_svm_props
1641 use_svnsync_props/) {
1642 my $key = $option;
1643 $key =~ tr/_//d;
1644 my $prop = "-$option";
1645 *$option = sub {
1646 my ($self) = @_;
1647 return $self->{$prop} if exists $self->{$prop};
1648 my $k = "svn-remote.$self->{repo_id}.$key";
1649 eval { command_oneline(qw/config --get/, $k) };
1650 if ($@) {
1651 $self->{$prop} = ${"Git::SVN::_$option"};
1652 } else {
1653 my $v = command_oneline(qw/config --bool/,$k);
1654 $self->{$prop} = $v eq 'false' ? 0 : 1;
1655 }
1656 return $self->{$prop};
1657 }
1658 }
1659 }
1662 my (%LOCKFILES, %INDEX_FILES);
1663 END {
1664 unlink keys %LOCKFILES if %LOCKFILES;
1665 unlink keys %INDEX_FILES if %INDEX_FILES;
1666 }
1668 sub resolve_local_globs {
1669 my ($url, $fetch, $glob_spec) = @_;
1670 return unless defined $glob_spec;
1671 my $ref = $glob_spec->{ref};
1672 my $path = $glob_spec->{path};
1673 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1674 next unless m#^$ref->{regex}$#;
1675 my $p = $1;
1676 my $pathname = desanitize_refname($path->full_path($p));
1677 my $refname = desanitize_refname($ref->full_path($p));
1678 if (my $existing = $fetch->{$pathname}) {
1679 if ($existing ne $refname) {
1680 die "Refspec conflict:\n",
1681 "existing: $existing\n",
1682 " globbed: $refname\n";
1683 }
1684 my $u = (::cmt_metadata("$refname"))[0];
1685 $u =~ s!^\Q$url\E(/|$)!! or die
1686 "$refname: '$url' not found in '$u'\n";
1687 if ($pathname ne $u) {
1688 warn "W: Refspec glob conflict ",
1689 "(ref: $refname):\n",
1690 "expected path: $pathname\n",
1691 " real path: $u\n",
1692 "Continuing ahead with $u\n";
1693 next;
1694 }
1695 } else {
1696 $fetch->{$pathname} = $refname;
1697 }
1698 }
1699 }
1701 sub parse_revision_argument {
1702 my ($base, $head) = @_;
1703 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1704 return ($base, $head);
1705 }
1706 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1707 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1708 return ($head, $head) if ($::_revision eq 'HEAD');
1709 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1710 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1711 die "revision argument: $::_revision not understood by git-svn\n";
1712 }
1714 sub fetch_all {
1715 my ($repo_id, $remotes) = @_;
1716 if (ref $repo_id) {
1717 my $gs = $repo_id;
1718 $repo_id = undef;
1719 $repo_id = $gs->{repo_id};
1720 }
1721 $remotes ||= read_all_remotes();
1722 my $remote = $remotes->{$repo_id} or
1723 die "[svn-remote \"$repo_id\"] unknown\n";
1724 my $fetch = $remote->{fetch};
1725 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1726 my (@gs, @globs);
1727 my $ra = Git::SVN::Ra->new($url);
1728 my $uuid = $ra->get_uuid;
1729 my $head = $ra->get_latest_revnum;
1730 $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1731 my $base = defined $fetch ? $head : 0;
1733 # read the max revs for wildcard expansion (branches/*, tags/*)
1734 foreach my $t (qw/branches tags/) {
1735 defined $remote->{$t} or next;
1736 push @globs, @{$remote->{$t}};
1738 my $max_rev = eval { tmp_config(qw/--int --get/,
1739 "svn-remote.$repo_id.${t}-maxRev") };
1740 if (defined $max_rev && ($max_rev < $base)) {
1741 $base = $max_rev;
1742 } elsif (!defined $max_rev) {
1743 $base = 0;
1744 }
1745 }
1747 if ($fetch) {
1748 foreach my $p (sort keys %$fetch) {
1749 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1750 my $lr = $gs->rev_map_max;
1751 if (defined $lr) {
1752 $base = $lr if ($lr < $base);
1753 }
1754 push @gs, $gs;
1755 }
1756 }
1758 ($base, $head) = parse_revision_argument($base, $head);
1759 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1760 }
1762 sub read_all_remotes {
1763 my $r = {};
1764 my $use_svm_props = eval { command_oneline(qw/config --bool
1765 svn.useSvmProps/) };
1766 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1767 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1768 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1769 if (m!^(.+)\.fetch=$svn_refspec$!) {
1770 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1771 die("svn-remote.$remote: remote ref '$remote_ref' "
1772 . "must start with 'refs/'\n")
1773 unless $remote_ref =~ m{^refs/};
1774 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1775 $r->{$remote}->{svm} = {} if $use_svm_props;
1776 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1777 $r->{$1}->{svm} = {};
1778 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1779 $r->{$1}->{url} = $2;
1780 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1781 my ($remote, $t, $local_ref, $remote_ref) =
1782 ($1, $2, $3, $4);
1783 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1784 . "must start with 'refs/'\n")
1785 unless $remote_ref =~ m{^refs/};
1786 my $rs = {
1787 t => $t,
1788 remote => $remote,
1789 path => Git::SVN::GlobSpec->new($local_ref),
1790 ref => Git::SVN::GlobSpec->new($remote_ref) };
1791 if (length($rs->{ref}->{right}) != 0) {
1792 die "The '*' glob character must be the last ",
1793 "character of '$remote_ref'\n";
1794 }
1795 push @{ $r->{$remote}->{$t} }, $rs;
1796 }
1797 }
1799 map {
1800 if (defined $r->{$_}->{svm}) {
1801 my $svm;
1802 eval {
1803 my $section = "svn-remote.$_";
1804 $svm = {
1805 source => tmp_config('--get',
1806 "$section.svm-source"),
1807 replace => tmp_config('--get',
1808 "$section.svm-replace"),
1809 }
1810 };
1811 $r->{$_}->{svm} = $svm;
1812 }
1813 } keys %$r;
1815 $r;
1816 }
1818 sub init_vars {
1819 $_gc_nr = $_gc_period = 1000;
1820 if (defined $_repack || defined $_repack_flags) {
1821 warn "Repack options are obsolete; they have no effect.\n";
1822 }
1823 }
1825 sub verify_remotes_sanity {
1826 return unless -d $ENV{GIT_DIR};
1827 my %seen;
1828 foreach (command(qw/config -l/)) {
1829 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1830 if ($seen{$1}) {
1831 die "Remote ref refs/remote/$1 is tracked by",
1832 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1833 "Please resolve this ambiguity in ",
1834 "your git configuration file before ",
1835 "continuing\n";
1836 }
1837 $seen{$1} = $_;
1838 }
1839 }
1840 }
1842 sub find_existing_remote {
1843 my ($url, $remotes) = @_;
1844 return undef if $no_reuse_existing;
1845 my $existing;
1846 foreach my $repo_id (keys %$remotes) {
1847 my $u = $remotes->{$repo_id}->{url} or next;
1848 next if $u ne $url;
1849 $existing = $repo_id;
1850 last;
1851 }
1852 $existing;
1853 }
1855 sub init_remote_config {
1856 my ($self, $url, $no_write) = @_;
1857 $url =~ s!/+$!!; # strip trailing slash
1858 my $r = read_all_remotes();
1859 my $existing = find_existing_remote($url, $r);
1860 if ($existing) {
1861 unless ($no_write) {
1862 print STDERR "Using existing ",
1863 "[svn-remote \"$existing\"]\n";
1864 }
1865 $self->{repo_id} = $existing;
1866 } elsif ($_minimize_url) {
1867 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1868 $existing = find_existing_remote($min_url, $r);
1869 if ($existing) {
1870 unless ($no_write) {
1871 print STDERR "Using existing ",
1872 "[svn-remote \"$existing\"]\n";
1873 }
1874 $self->{repo_id} = $existing;
1875 }
1876 if ($min_url ne $url) {
1877 unless ($no_write) {
1878 print STDERR "Using higher level of URL: ",
1879 "$url => $min_url\n";
1880 }
1881 my $old_path = $self->{path};
1882 $self->{path} = $url;
1883 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1884 if (length $old_path) {
1885 $self->{path} .= "/$old_path";
1886 }
1887 $url = $min_url;
1888 }
1889 }
1890 my $orig_url;
1891 if (!$existing) {
1892 # verify that we aren't overwriting anything:
1893 $orig_url = eval {
1894 command_oneline('config', '--get',
1895 "svn-remote.$self->{repo_id}.url")
1896 };
1897 if ($orig_url && ($orig_url ne $url)) {
1898 die "svn-remote.$self->{repo_id}.url already set: ",
1899 "$orig_url\nwanted to set to: $url\n";
1900 }
1901 }
1902 my ($xrepo_id, $xpath) = find_ref($self->refname);
1903 if (!$no_write && defined $xpath) {
1904 die "svn-remote.$xrepo_id.fetch already set to track ",
1905 "$xpath:", $self->refname, "\n";
1906 }
1907 unless ($no_write) {
1908 command_noisy('config',
1909 "svn-remote.$self->{repo_id}.url", $url);
1910 $self->{path} =~ s{^/}{};
1911 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1912 command_noisy('config', '--add',
1913 "svn-remote.$self->{repo_id}.fetch",
1914 "$self->{path}:".$self->refname);
1915 }
1916 $self->{url} = $url;
1917 }
1919 sub find_by_url { # repos_root and, path are optional
1920 my ($class, $full_url, $repos_root, $path) = @_;
1922 return undef unless defined $full_url;
1923 remove_username($full_url);
1924 remove_username($repos_root) if defined $repos_root;
1925 my $remotes = read_all_remotes();
1926 if (defined $full_url && defined $repos_root && !defined $path) {
1927 $path = $full_url;
1928 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1929 }
1930 foreach my $repo_id (keys %$remotes) {
1931 my $u = $remotes->{$repo_id}->{url} or next;
1932 remove_username($u);
1933 next if defined $repos_root && $repos_root ne $u;
1935 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1936 foreach my $t (qw/branches tags/) {
1937 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1938 resolve_local_globs($u, $fetch, $globspec);
1939 }
1940 }
1941 my $p = $path;
1942 my $rwr = rewrite_root({repo_id => $repo_id});
1943 my $svm = $remotes->{$repo_id}->{svm}
1944 if defined $remotes->{$repo_id}->{svm};
1945 unless (defined $p) {
1946 $p = $full_url;
1947 my $z = $u;
1948 my $prefix = '';
1949 if ($rwr) {
1950 $z = $rwr;
1951 remove_username($z);
1952 } elsif (defined $svm) {
1953 $z = $svm->{source};
1954 $prefix = $svm->{replace};
1955 $prefix =~ s#^\Q$u\E(?:/|$)##;
1956 $prefix =~ s#/$##;
1957 }
1958 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1959 }
1960 foreach my $f (keys %$fetch) {
1961 next if $f ne $p;
1962 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1963 }
1964 }
1965 undef;
1966 }
1968 sub init {
1969 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1970 my $self = _new($class, $repo_id, $ref_id, $path);
1971 if (defined $url) {
1972 $self->init_remote_config($url, $no_write);
1973 }
1974 $self;
1975 }
1977 sub find_ref {
1978 my ($ref_id) = @_;
1979 foreach (command(qw/config -l/)) {
1980 next unless m!^svn-remote\.(.+)\.fetch=
1981 \s*(.*?)\s*:\s*(.+?)\s*$!x;
1982 my ($repo_id, $path, $ref) = ($1, $2, $3);
1983 if ($ref eq $ref_id) {
1984 $path = '' if ($path =~ m#^\./?#);
1985 return ($repo_id, $path);
1986 }
1987 }
1988 (undef, undef, undef);
1989 }
1991 sub new {
1992 my ($class, $ref_id, $repo_id, $path) = @_;
1993 if (defined $ref_id && !defined $repo_id && !defined $path) {
1994 ($repo_id, $path) = find_ref($ref_id);
1995 if (!defined $repo_id) {
1996 die "Could not find a \"svn-remote.*.fetch\" key ",
1997 "in the repository configuration matching: ",
1998 "$ref_id\n";
1999 }
2000 }
2001 my $self = _new($class, $repo_id, $ref_id, $path);
2002 if (!defined $self->{path} || !length $self->{path}) {
2003 my $fetch = command_oneline('config', '--get',
2004 "svn-remote.$repo_id.fetch",
2005 ":$ref_id\$") or
2006 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2007 "\":$ref_id\$\" in config\n";
2008 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2009 }
2010 $self->{url} = command_oneline('config', '--get',
2011 "svn-remote.$repo_id.url") or
2012 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2013 $self->rebuild;
2014 $self;
2015 }
2017 sub refname {
2018 my ($refname) = $_[0]->{ref_id} ;
2020 # It cannot end with a slash /, we'll throw up on this because
2021 # SVN can't have directories with a slash in their name, either:
2022 if ($refname =~ m{/$}) {
2023 die "ref: '$refname' ends with a trailing slash, this is ",
2024 "not permitted by git nor Subversion\n";
2025 }
2027 # It cannot have ASCII control character space, tilde ~, caret ^,
2028 # colon :, question-mark ?, asterisk *, space, or open bracket [
2029 # anywhere.
2030 #
2031 # Additionally, % must be escaped because it is used for escaping
2032 # and we want our escaped refname to be reversible
2033 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2035 # no slash-separated component can begin with a dot .
2036 # /.* becomes /%2E*
2037 $refname =~ s{/\.}{/%2E}g;
2039 # It cannot have two consecutive dots .. anywhere
2040 # .. becomes %2E%2E
2041 $refname =~ s{\.\.}{%2E%2E}g;
2043 return $refname;
2044 }
2046 sub desanitize_refname {
2047 my ($refname) = @_;
2048 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2049 return $refname;
2050 }
2052 sub svm_uuid {
2053 my ($self) = @_;
2054 return $self->{svm}->{uuid} if $self->svm;
2055 $self->ra;
2056 unless ($self->{svm}) {
2057 die "SVM UUID not cached, and reading remotely failed\n";
2058 }
2059 $self->{svm}->{uuid};
2060 }
2062 sub svm {
2063 my ($self) = @_;
2064 return $self->{svm} if $self->{svm};
2065 my $svm;
2066 # see if we have it in our config, first:
2067 eval {
2068 my $section = "svn-remote.$self->{repo_id}";
2069 $svm = {
2070 source => tmp_config('--get', "$section.svm-source"),
2071 uuid => tmp_config('--get', "$section.svm-uuid"),
2072 replace => tmp_config('--get', "$section.svm-replace"),
2073 }
2074 };
2075 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2076 $self->{svm} = $svm;
2077 }
2078 $self->{svm};
2079 }
2081 sub _set_svm_vars {
2082 my ($self, $ra) = @_;
2083 return $ra if $self->svm;
2085 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2086 "(svm:source, svm:uuid) ",
2087 "from the following URLs:\n" );
2088 sub read_svm_props {
2089 my ($self, $ra, $path, $r) = @_;
2090 my $props = ($ra->get_dir($path, $r))[2];
2091 my $src = $props->{'svm:source'};
2092 my $uuid = $props->{'svm:uuid'};
2093 return undef if (!$src || !$uuid);
2095 chomp($src, $uuid);
2097 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2098 or die "doesn't look right - svm:uuid is '$uuid'\n";
2100 # the '!' is used to mark the repos_root!/relative/path
2101 $src =~ s{/?!/?}{/};
2102 $src =~ s{/+$}{}; # no trailing slashes please
2103 # username is of no interest
2104 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2106 my $replace = $ra->{url};
2107 $replace .= "/$path" if length $path;
2109 my $section = "svn-remote.$self->{repo_id}";
2110 tmp_config("$section.svm-source", $src);
2111 tmp_config("$section.svm-replace", $replace);
2112 tmp_config("$section.svm-uuid", $uuid);
2113 $self->{svm} = {
2114 source => $src,
2115 uuid => $uuid,
2116 replace => $replace
2117 };
2118 }
2120 my $r = $ra->get_latest_revnum;
2121 my $path = $self->{path};
2122 my %tried;
2123 while (length $path) {
2124 unless ($tried{"$self->{url}/$path"}) {
2125 return $ra if $self->read_svm_props($ra, $path, $r);
2126 $tried{"$self->{url}/$path"} = 1;
2127 }
2128 $path =~ s#/?[^/]+$##;
2129 }
2130 die "Path: '$path' should be ''\n" if $path ne '';
2131 return $ra if $self->read_svm_props($ra, $path, $r);
2132 $tried{"$self->{url}/$path"} = 1;
2134 if ($ra->{repos_root} eq $self->{url}) {
2135 die @err, (map { " $_\n" } keys %tried), "\n";
2136 }
2138 # nope, make sure we're connected to the repository root:
2139 my $ok;
2140 my @tried_b;
2141 $path = $ra->{svn_path};
2142 $ra = Git::SVN::Ra->new($ra->{repos_root});
2143 while (length $path) {
2144 unless ($tried{"$ra->{url}/$path"}) {
2145 $ok = $self->read_svm_props($ra, $path, $r);
2146 last if $ok;
2147 $tried{"$ra->{url}/$path"} = 1;
2148 }
2149 $path =~ s#/?[^/]+$##;
2150 }
2151 die "Path: '$path' should be ''\n" if $path ne '';
2152 $ok ||= $self->read_svm_props($ra, $path, $r);
2153 $tried{"$ra->{url}/$path"} = 1;
2154 if (!$ok) {
2155 die @err, (map { " $_\n" } keys %tried), "\n";
2156 }
2157 Git::SVN::Ra->new($self->{url});
2158 }
2160 sub svnsync {
2161 my ($self) = @_;
2162 return $self->{svnsync} if $self->{svnsync};
2164 if ($self->no_metadata) {
2165 die "Can't have both 'noMetadata' and ",
2166 "'useSvnsyncProps' options set!\n";
2167 }
2168 if ($self->rewrite_root) {
2169 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2170 "options set!\n";
2171 }
2173 my $svnsync;
2174 # see if we have it in our config, first:
2175 eval {
2176 my $section = "svn-remote.$self->{repo_id}";
2178 my $url = tmp_config('--get', "$section.svnsync-url");
2179 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2180 die "doesn't look right - svn:sync-from-url is '$url'\n";
2182 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2183 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2184 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2186 $svnsync = { url => $url, uuid => $uuid }
2187 };
2188 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2189 return $self->{svnsync} = $svnsync;
2190 }
2192 my $err = "useSvnsyncProps set, but failed to read " .
2193 "svnsync property: svn:sync-from-";
2194 my $rp = $self->ra->rev_proplist(0);
2196 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2197 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2198 die "doesn't look right - svn:sync-from-url is '$url'\n";
2200 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2201 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2202 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2204 my $section = "svn-remote.$self->{repo_id}";
2205 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2206 tmp_config('--add', "$section.svnsync-url", $url);
2207 return $self->{svnsync} = { url => $url, uuid => $uuid };
2208 }
2210 # this allows us to memoize our SVN::Ra UUID locally and avoid a
2211 # remote lookup (useful for 'git svn log').
2212 sub ra_uuid {
2213 my ($self) = @_;
2214 unless ($self->{ra_uuid}) {
2215 my $key = "svn-remote.$self->{repo_id}.uuid";
2216 my $uuid = eval { tmp_config('--get', $key) };
2217 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2218 $self->{ra_uuid} = $uuid;
2219 } else {
2220 die "ra_uuid called without URL\n" unless $self->{url};
2221 $self->{ra_uuid} = $self->ra->get_uuid;
2222 tmp_config('--add', $key, $self->{ra_uuid});
2223 }
2224 }
2225 $self->{ra_uuid};
2226 }
2228 sub _set_repos_root {
2229 my ($self, $repos_root) = @_;
2230 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2231 $repos_root ||= $self->ra->{repos_root};
2232 tmp_config($k, $repos_root);
2233 $repos_root;
2234 }
2236 sub repos_root {
2237 my ($self) = @_;
2238 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2239 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2240 }
2242 sub ra {
2243 my ($self) = shift;
2244 my $ra = Git::SVN::Ra->new($self->{url});
2245 $self->_set_repos_root($ra->{repos_root});
2246 if ($self->use_svm_props && !$self->{svm}) {
2247 if ($self->no_metadata) {
2248 die "Can't have both 'noMetadata' and ",
2249 "'useSvmProps' options set!\n";
2250 } elsif ($self->use_svnsync_props) {
2251 die "Can't have both 'useSvnsyncProps' and ",
2252 "'useSvmProps' options set!\n";
2253 }
2254 $ra = $self->_set_svm_vars($ra);
2255 $self->{-want_revprops} = 1;
2256 }
2257 $ra;
2258 }
2260 # prop_walk(PATH, REV, SUB)
2261 # -------------------------
2262 # Recursively traverse PATH at revision REV and invoke SUB for each
2263 # directory that contains a SVN property. SUB will be invoked as
2264 # follows: &SUB(gs, path, props); where `gs' is this instance of
2265 # Git::SVN, `path' the path to the directory where the properties
2266 # `props' were found. The `path' will be relative to point of checkout,
2267 # that is, if url://repo/trunk is the current Git branch, and that
2268 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
2269 # as `path' (note the trailing `/').
2270 sub prop_walk {
2271 my ($self, $path, $rev, $sub) = @_;
2273 $path =~ s#^/##;
2274 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2275 $path =~ s#^/*#/#g;
2276 my $p = $path;
2277 # Strip the irrelevant part of the path.
2278 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2279 # Ensure the path is terminated by a `/'.
2280 $p =~ s#/*$#/#;
2282 # The properties contain all the internal SVN stuff nobody
2283 # (usually) cares about.
2284 my $interesting_props = 0;
2285 foreach (keys %{$props}) {
2286 # If it doesn't start with `svn:', it must be a
2287 # user-defined property.
2288 ++$interesting_props and next if $_ !~ /^svn:/;
2289 # FIXME: Fragile, if SVN adds new public properties,
2290 # this needs to be updated.
2291 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2292 |eol-style|mime-type
2293 |externals|needs-lock)$/x;
2294 }
2295 &$sub($self, $p, $props) if $interesting_props;
2297 foreach (sort keys %$dirent) {
2298 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2299 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2300 }
2301 }
2303 sub last_rev { ($_[0]->last_rev_commit)[0] }
2304 sub last_commit { ($_[0]->last_rev_commit)[1] }
2306 # returns the newest SVN revision number and newest commit SHA1
2307 sub last_rev_commit {
2308 my ($self) = @_;
2309 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2310 return ($self->{last_rev}, $self->{last_commit});
2311 }
2312 my $c = ::verify_ref($self->refname.'^0');
2313 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2314 my $rev = (::cmt_metadata($c))[1];
2315 if (defined $rev) {
2316 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2317 return ($rev, $c);
2318 }
2319 }
2320 my $map_path = $self->map_path;
2321 unless (-e $map_path) {
2322 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2323 return (undef, undef);
2324 }
2325 my ($rev, $commit) = $self->rev_map_max(1);
2326 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2327 return ($rev, $commit);
2328 }
2330 sub get_fetch_range {
2331 my ($self, $min, $max) = @_;
2332 $max ||= $self->ra->get_latest_revnum;
2333 $min ||= $self->rev_map_max;
2334 (++$min, $max);
2335 }
2337 sub tmp_config {
2338 my (@args) = @_;
2339 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2340 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2341 if (! -f $config && -f $old_def_config) {
2342 rename $old_def_config, $config or
2343 die "Failed rename $old_def_config => $config: $!\n";
2344 }
2345 my $old_config = $ENV{GIT_CONFIG};
2346 $ENV{GIT_CONFIG} = $config;
2347 $@ = undef;
2348 my @ret = eval {
2349 unless (-f $config) {
2350 mkfile($config);
2351 open my $fh, '>', $config or
2352 die "Can't open $config: $!\n";
2353 print $fh "; This file is used internally by ",
2354 "git-svn\n" or die
2355 "Couldn't write to $config: $!\n";
2356 print $fh "; You should not have to edit it\n" or
2357 die "Couldn't write to $config: $!\n";
2358 close $fh or die "Couldn't close $config: $!\n";
2359 }
2360 command('config', @args);
2361 };
2362 my $err = $@;
2363 if (defined $old_config) {
2364 $ENV{GIT_CONFIG} = $old_config;
2365 } else {
2366 delete $ENV{GIT_CONFIG};
2367 }
2368 die $err if $err;
2369 wantarray ? @ret : $ret[0];
2370 }
2372 sub tmp_index_do {
2373 my ($self, $sub) = @_;
2374 my $old_index = $ENV{GIT_INDEX_FILE};
2375 $ENV{GIT_INDEX_FILE} = $self->{index};
2376 $@ = undef;
2377 my @ret = eval {
2378 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2379 mkpath([$dir]) unless -d $dir;
2380 &$sub;
2381 };
2382 my $err = $@;
2383 if (defined $old_index) {
2384 $ENV{GIT_INDEX_FILE} = $old_index;
2385 } else {
2386 delete $ENV{GIT_INDEX_FILE};
2387 }
2388 die $err if $err;
2389 wantarray ? @ret : $ret[0];
2390 }
2392 sub assert_index_clean {
2393 my ($self, $treeish) = @_;
2395 $self->tmp_index_do(sub {
2396 command_noisy('read-tree', $treeish) unless -e $self->{index};
2397 my $x = command_oneline('write-tree');
2398 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2399 /^tree ($::sha1)/mo);
2400 return if $y eq $x;
2402 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2403 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2404 command_noisy('read-tree', $treeish);
2405 $x = command_oneline('write-tree');
2406 if ($y ne $x) {
2407 ::fatal "trees ($treeish) $y != $x\n",
2408 "Something is seriously wrong...";
2409 }
2410 });
2411 }
2413 sub get_commit_parents {
2414 my ($self, $log_entry) = @_;
2415 my (%seen, @ret, @tmp);
2416 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2417 if (my $ip = $self->{inject_parents}) {
2418 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2419 push @tmp, $commit;
2420 }
2421 }
2422 if (my $cur = ::verify_ref($self->refname.'^0')) {
2423 push @tmp, $cur;
2424 }
2425 if (my $ipd = $self->{inject_parents_dcommit}) {
2426 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2427 push @tmp, @$commit;
2428 }
2429 }
2430 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2431 while (my $p = shift @tmp) {
2432 next if $seen{$p};
2433 $seen{$p} = 1;
2434 push @ret, $p;
2435 # MAXPARENT is defined to 16 in commit-tree.c:
2436 last if @ret >= 16;
2437 }
2438 if (@tmp) {
2439 die "r$log_entry->{revision}: No room for parents:\n\t",
2440 join("\n\t", @tmp), "\n";
2441 }
2442 @ret;
2443 }
2445 sub rewrite_root {
2446 my ($self) = @_;
2447 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2448 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2449 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2450 if ($rwr) {
2451 $rwr =~ s#/+$##;
2452 if ($rwr !~ m#^[a-z\+]+://#) {
2453 die "$rwr is not a valid URL (key: $k)\n";
2454 }
2455 }
2456 $self->{-rewrite_root} = $rwr;
2457 }
2459 sub metadata_url {
2460 my ($self) = @_;
2461 ($self->rewrite_root || $self->{url}) .
2462 (length $self->{path} ? '/' . $self->{path} : '');
2463 }
2465 sub full_url {
2466 my ($self) = @_;
2467 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2468 }
2471 sub set_commit_header_env {
2472 my ($log_entry) = @_;
2473 my %env;
2474 foreach my $ned (qw/NAME EMAIL DATE/) {
2475 foreach my $ac (qw/AUTHOR COMMITTER/) {
2476 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2477 }
2478 }
2480 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2481 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2482 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2484 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2485 ? $log_entry->{commit_name}
2486 : $log_entry->{name};
2487 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2488 ? $log_entry->{commit_email}
2489 : $log_entry->{email};
2490 \%env;
2491 }
2493 sub restore_commit_header_env {
2494 my ($env) = @_;
2495 foreach my $ned (qw/NAME EMAIL DATE/) {
2496 foreach my $ac (qw/AUTHOR COMMITTER/) {
2497 my $k = "GIT_${ac}_${ned}";
2498 if (defined $env->{$k}) {
2499 $ENV{$k} = $env->{$k};
2500 } else {
2501 delete $ENV{$k};
2502 }
2503 }
2504 }
2505 }
2507 sub gc {
2508 command_noisy('gc', '--auto');
2509 };
2511 sub do_git_commit {
2512 my ($self, $log_entry) = @_;
2513 my $lr = $self->last_rev;
2514 if (defined $lr && $lr >= $log_entry->{revision}) {
2515 die "Last fetched revision of ", $self->refname,
2516 " was r$lr, but we are about to fetch: ",
2517 "r$log_entry->{revision}!\n";
2518 }
2519 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2520 croak "$log_entry->{revision} = $c already exists! ",
2521 "Why are we refetching it?\n";
2522 }
2523 my $old_env = set_commit_header_env($log_entry);
2524 my $tree = $log_entry->{tree};
2525 if (!defined $tree) {
2526 $tree = $self->tmp_index_do(sub {
2527 command_oneline('write-tree') });
2528 }
2529 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2531 my @exec = ('git', 'commit-tree', $tree);
2532 foreach ($self->get_commit_parents($log_entry)) {
2533 push @exec, '-p', $_;
2534 }
2535 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2536 or croak $!;
2537 binmode $msg_fh;
2539 # we always get UTF-8 from SVN, but we may want our commits in
2540 # a different encoding.
2541 if (my $enc = Git::config('i18n.commitencoding')) {
2542 require Encode;
2543 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2544 }
2545 print $msg_fh $log_entry->{log} or croak $!;
2546 restore_commit_header_env($old_env);
2547 unless ($self->no_metadata) {
2548 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2549 or croak $!;
2550 }
2551 $msg_fh->flush == 0 or croak $!;
2552 close $msg_fh or croak $!;
2553 chomp(my $commit = do { local $/; <$out_fh> });
2554 close $out_fh or croak $!;
2555 waitpid $pid, 0;
2556 croak $? if $?;
2557 if ($commit !~ /^$::sha1$/o) {
2558 die "Failed to commit, invalid sha1: $commit\n";
2559 }
2561 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2563 $self->{last_rev} = $log_entry->{revision};
2564 $self->{last_commit} = $commit;
2565 print "r$log_entry->{revision}" unless $::_q > 1;
2566 if (defined $log_entry->{svm_revision}) {
2567 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2568 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2569 0, $self->svm_uuid);
2570 }
2571 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2572 if (--$_gc_nr == 0) {
2573 $_gc_nr = $_gc_period;
2574 gc();
2575 }
2576 return $commit;
2577 }
2579 sub match_paths {
2580 my ($self, $paths, $r) = @_;
2581 return 1 if $self->{path} eq '';
2582 if (my $path = $paths->{"/$self->{path}"}) {
2583 return ($path->{action} eq 'D') ? 0 : 1;
2584 }
2585 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2586 if (grep /$self->{path_regex}/, keys %$paths) {
2587 return 1;
2588 }
2589 my $c = '';
2590 foreach (split m#/#, $self->{path}) {
2591 $c .= "/$_";
2592 next unless ($paths->{$c} &&
2593 ($paths->{$c}->{action} =~ /^[AR]$/));
2594 if ($self->ra->check_path($self->{path}, $r) ==
2595 $SVN::Node::dir) {
2596 return 1;
2597 }
2598 }
2599 return 0;
2600 }
2602 sub find_parent_branch {
2603 my ($self, $paths, $rev) = @_;
2604 return undef unless $self->follow_parent;
2605 unless (defined $paths) {
2606 my $err_handler = $SVN::Error::handler;
2607 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2608 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2609 sub { $paths = $_[0] });
2610 $SVN::Error::handler = $err_handler;
2611 }
2612 return undef unless defined $paths;
2614 # look for a parent from another branch:
2615 my @b_path_components = split m#/#, $self->{path};
2616 my @a_path_components;
2617 my $i;
2618 while (@b_path_components) {
2619 $i = $paths->{'/'.join('/', @b_path_components)};
2620 last if $i && defined $i->{copyfrom_path};
2621 unshift(@a_path_components, pop(@b_path_components));
2622 }
2623 return undef unless defined $i && defined $i->{copyfrom_path};
2624 my $branch_from = $i->{copyfrom_path};
2625 if (@a_path_components) {
2626 print STDERR "branch_from: $branch_from => ";
2627 $branch_from .= '/'.join('/', @a_path_components);
2628 print STDERR $branch_from, "\n";
2629 }
2630 my $r = $i->{copyfrom_rev};
2631 my $repos_root = $self->ra->{repos_root};
2632 my $url = $self->ra->{url};
2633 my $new_url = $url . $branch_from;
2634 print STDERR "Found possible branch point: ",
2635 "$new_url => ", $self->full_url, ", $r\n"
2636 unless $::_q > 1;
2637 $branch_from =~ s#^/##;
2638 my $gs = $self->other_gs($new_url, $url,
2639 $branch_from, $r, $self->{ref_id});
2640 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2641 {
2642 my ($base, $head);
2643 if (!defined $r0 || !defined $parent) {
2644 ($base, $head) = parse_revision_argument(0, $r);
2645 } else {
2646 if ($r0 < $r) {
2647 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2648 0, 1, sub { $base = $_[1] - 1 });
2649 }
2650 }
2651 if (defined $base && $base <= $r) {
2652 $gs->fetch($base, $r);
2653 }
2654 ($r0, $parent) = $gs->find_rev_before($r, 1);
2655 }
2656 if (defined $r0 && defined $parent) {
2657 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2658 unless $::_q > 1;
2659 my $ed;
2660 if ($self->ra->can_do_switch) {
2661 $self->assert_index_clean($parent);
2662 print STDERR "Following parent with do_switch\n"
2663 unless $::_q > 1;
2664 # do_switch works with svn/trunk >= r22312, but that
2665 # is not included with SVN 1.4.3 (the latest version
2666 # at the moment), so we can't rely on it
2667 $self->{last_rev} = $r0;
2668 $self->{last_commit} = $parent;
2669 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2670 $gs->ra->gs_do_switch($r0, $rev, $gs,
2671 $self->full_url, $ed)
2672 or die "SVN connection failed somewhere...\n";
2673 } elsif ($self->ra->trees_match($new_url, $r0,
2674 $self->full_url, $rev)) {
2675 print STDERR "Trees match:\n",
2676 " $new_url\@$r0\n",
2677 " ${\$self->full_url}\@$rev\n",
2678 "Following parent with no changes\n"
2679 unless $::_q > 1;
2680 $self->tmp_index_do(sub {
2681 command_noisy('read-tree', $parent);
2682 });
2683 $self->{last_commit} = $parent;
2684 } else {
2685 print STDERR "Following parent with do_update\n"
2686 unless $::_q > 1;
2687 $ed = SVN::Git::Fetcher->new($self);
2688 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2689 or die "SVN connection failed somewhere...\n";
2690 }
2691 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2692 return $self->make_log_entry($rev, [$parent], $ed);
2693 }
2694 return undef;
2695 }
2697 sub do_fetch {
2698 my ($self, $paths, $rev) = @_;
2699 my $ed;
2700 my ($last_rev, @parents);
2701 if (my $lc = $self->last_commit) {
2702 # we can have a branch that was deleted, then re-added
2703 # under the same name but copied from another path, in
2704 # which case we'll have multiple parents (we don't
2705 # want to break the original ref, nor lose copypath info):
2706 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2707 push @{$log_entry->{parents}}, $lc;
2708 return $log_entry;
2709 }
2710 $ed = SVN::Git::Fetcher->new($self);
2711 $last_rev = $self->{last_rev};
2712 $ed->{c} = $lc;
2713 @parents = ($lc);
2714 } else {
2715 $last_rev = $rev;
2716 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2717 return $log_entry;
2718 }
2719 $ed = SVN::Git::Fetcher->new($self);
2720 }
2721 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2722 die "SVN connection failed somewhere...\n";
2723 }
2724 $self->make_log_entry($rev, \@parents, $ed);
2725 }
2727 sub get_untracked {
2728 my ($self, $ed) = @_;
2729 my @out;
2730 my $h = $ed->{empty};
2731 foreach (sort keys %$h) {
2732 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2733 push @out, " $act: " . uri_encode($_);
2734 warn "W: $act: $_\n";
2735 }
2736 foreach my $t (qw/dir_prop file_prop/) {
2737 $h = $ed->{$t} or next;
2738 foreach my $path (sort keys %$h) {
2739 my $ppath = $path eq '' ? '.' : $path;
2740 foreach my $prop (sort keys %{$h->{$path}}) {
2741 next if $SKIP_PROP{$prop};
2742 my $v = $h->{$path}->{$prop};
2743 my $t_ppath_prop = "$t: " .
2744 uri_encode($ppath) . ' ' .
2745 uri_encode($prop);
2746 if (defined $v) {
2747 push @out, " +$t_ppath_prop " .
2748 uri_encode($v);
2749 } else {
2750 push @out, " -$t_ppath_prop";
2751 }
2752 }
2753 }
2754 }
2755 foreach my $t (qw/absent_file absent_directory/) {
2756 $h = $ed->{$t} or next;
2757 foreach my $parent (sort keys %$h) {
2758 foreach my $path (sort @{$h->{$parent}}) {
2759 push @out, " $t: " .
2760 uri_encode("$parent/$path");
2761 warn "W: $t: $parent/$path ",
2762 "Insufficient permissions?\n";
2763 }
2764 }
2765 }
2766 \@out;
2767 }
2769 # parse_svn_date(DATE)
2770 # --------------------
2771 # Given a date (in UTC) from Subversion, return a string in the format
2772 # "<TZ Offset> <local date/time>" that Git will use.
2773 #
2774 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
2775 # is true we'll convert it to the local timezone instead.
2776 sub parse_svn_date {
2777 my $date = shift || return '+0000 1970-01-01 00:00:00';
2778 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2779 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2780 croak "Unable to parse date: $date\n";
2781 my $parsed_date; # Set next.
2783 if ($Git::SVN::_localtime) {
2784 # Translate the Subversion datetime to an epoch time.
2785 # Begin by switching ourselves to $date's timezone, UTC.
2786 my $old_env_TZ = $ENV{TZ};
2787 $ENV{TZ} = 'UTC';
2789 my $epoch_in_UTC =
2790 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2792 # Determine our local timezone (including DST) at the
2793 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2794 # value of TZ, if any, at the time we were run.
2795 if (defined $Git::SVN::Log::TZ) {
2796 $ENV{TZ} = $Git::SVN::Log::TZ;
2797 } else {
2798 delete $ENV{TZ};
2799 }
2801 my $our_TZ =
2802 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2804 # This converts $epoch_in_UTC into our local timezone.
2805 my ($sec, $min, $hour, $mday, $mon, $year,
2806 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2808 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2809 $our_TZ, $year + 1900, $mon + 1,
2810 $mday, $hour, $min, $sec);
2812 # Reset us to the timezone in effect when we entered
2813 # this routine.
2814 if (defined $old_env_TZ) {
2815 $ENV{TZ} = $old_env_TZ;
2816 } else {
2817 delete $ENV{TZ};
2818 }
2819 } else {
2820 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2821 }
2823 return $parsed_date;
2824 }
2826 sub other_gs {
2827 my ($self, $new_url, $url,
2828 $branch_from, $r, $old_ref_id) = @_;
2829 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2830 unless ($gs) {
2831 my $ref_id = $old_ref_id;
2832 $ref_id =~ s/\@\d+$//;
2833 $ref_id .= "\@$r";
2834 # just grow a tail if we're not unique enough :x
2835 $ref_id .= '-' while find_ref($ref_id);
2836 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2837 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2838 if ($u =~ s#^\Q$url\E(/|$)##) {
2839 $p = $u;
2840 $u = $url;
2841 $repo_id = $self->{repo_id};
2842 }
2843 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2844 }
2845 $gs
2846 }
2848 sub call_authors_prog {
2849 my ($orig_author) = @_;
2850 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2851 my $author = `$::_authors_prog $orig_author`;
2852 if ($? != 0) {
2853 die "$::_authors_prog failed with exit code $?\n"
2854 }
2855 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2856 my ($name, $email) = ($1, $2);
2857 $email = undef if length $2 == 0;
2858 return [$name, $email];
2859 } else {
2860 die "Author: $orig_author: $::_authors_prog returned "
2861 . "invalid author format: $author\n";
2862 }
2863 }
2865 sub check_author {
2866 my ($author) = @_;
2867 if (!defined $author || length $author == 0) {
2868 $author = '(no author)';
2869 }
2870 if (!defined $::users{$author}) {
2871 if (defined $::_authors_prog) {
2872 $::users{$author} = call_authors_prog($author);
2873 } elsif (defined $::_authors) {
2874 die "Author: $author not defined in $::_authors file\n";
2875 }
2876 }
2877 $author;
2878 }
2880 sub find_extra_svk_parents {
2881 my ($self, $ed, $tickets, $parents) = @_;
2882 # aha! svk:merge property changed...
2883 my @tickets = split "\n", $tickets;
2884 my @known_parents;
2885 for my $ticket ( @tickets ) {
2886 my ($uuid, $path, $rev) = split /:/, $ticket;
2887 if ( $uuid eq $self->ra_uuid ) {
2888 my $url = $self->rewrite_root || $self->{url};
2889 my $repos_root = $url;
2890 my $branch_from = $path;
2891 $branch_from =~ s{^/}{};
2892 my $gs = $self->other_gs($repos_root."/".$branch_from,
2893 $url,
2894 $branch_from,
2895 $rev,
2896 $self->{ref_id});
2897 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
2898 # wahey! we found it, but it might be
2899 # an old one (!)
2900 push @known_parents, $commit;
2901 }
2902 }
2903 }
2904 for my $parent ( @known_parents ) {
2905 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
2906 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
2907 my $new;
2908 while ( <$msg_fh> ) {
2909 $new=1;last;
2910 }
2911 command_close_pipe($msg_fh, $ctx);
2912 if ( $new ) {
2913 print STDERR
2914 "Found merge parent (svk:merge ticket): $parent\n";
2915 push @$parents, $parent;
2916 }
2917 }
2918 }
2920 # note: this function should only be called if the various dirprops
2921 # have actually changed
2922 sub find_extra_svn_parents {
2923 my ($self, $ed, $mergeinfo, $parents) = @_;
2924 # aha! svk:merge property changed...
2926 # We first search for merged tips which are not in our
2927 # history. Then, we figure out which git revisions are in
2928 # that tip, but not this revision. If all of those revisions
2929 # are now marked as merge, we can add the tip as a parent.
2930 my @merges = split "\n", $mergeinfo;
2931 my @merge_tips;
2932 my @merged_commit_ranges;
2933 my $url = $self->rewrite_root || $self->{url};
2934 for my $merge ( @merges ) {
2935 my ($source, $revs) = split ":", $merge;
2936 my $path = $source;
2937 $path =~ s{^/}{};
2938 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
2939 if ( !$gs ) {
2940 warn "Couldn't find revmap for $url$source\n";
2941 next;
2942 }
2943 my @ranges = split ",", $revs;
2944 my ($tip, $tip_commit);
2945 # find the tip
2946 for my $range ( @ranges ) {
2947 my ($bottom, $top) = split "-", $range;
2948 $top ||= $bottom;
2949 my $bottom_commit =
2950 $gs->rev_map_get($bottom, $self->ra_uuid) ||
2951 $gs->rev_map_get($bottom+1, $self->ra_uuid);
2952 my $top_commit;
2953 for (; !$top_commit && $top >= $bottom; --$top) {
2954 $top_commit =
2955 $gs->rev_map_get($top, $self->ra_uuid);
2956 }
2958 unless ($top_commit and $bottom_commit) {
2959 warn "W:unknown path/rev in svn:mergeinfo "
2960 ."dirprop: $source:$range\n";
2961 next;
2962 }
2964 push @merged_commit_ranges,
2965 "$bottom_commit..$top_commit";
2967 if ( !defined $tip or $top > $tip ) {
2968 $tip = $top;
2969 $tip_commit = $top_commit;
2970 }
2971 }
2972 unless (!$tip_commit or
2973 grep { $_ eq $tip_commit } @$parents ) {
2974 push @merge_tips, $tip_commit;
2975 } else {
2976 push @merge_tips, undef;
2977 }
2978 }
2979 for my $merge_tip ( @merge_tips ) {
2980 my $spec = shift @merges;
2981 next unless $merge_tip;
2982 my @cmd = ('rev-list', "-1", $merge_tip,
2983 "--not", @$parents );
2984 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
2985 my $new;
2986 while ( <$msg_fh> ) {
2987 $new=1;last;
2988 }
2989 command_close_pipe($msg_fh, $ctx);
2990 if ( $new ) {
2991 push @cmd, @merged_commit_ranges;
2992 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
2993 my $unmerged;
2994 while ( <$msg_fh> ) {
2995 $unmerged=1;last;
2996 }
2997 command_close_pipe($msg_fh, $ctx);
2998 if ( $unmerged ) {
2999 warn "W:svn cherry-pick ignored ($spec)\n";
3000 } else {
3001 warn
3002 "Found merge parent (svn:mergeinfo prop): ",
3003 $merge_tip, "\n";
3004 push @$parents, $merge_tip;
3005 }
3006 }
3007 }
3008 }
3010 sub make_log_entry {
3011 my ($self, $rev, $parents, $ed) = @_;
3012 my $untracked = $self->get_untracked($ed);
3014 my @parents = @$parents;
3015 my $ps = $ed->{path_strip} || "";
3016 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3017 my $props = $ed->{dir_prop}{$path};
3018 if ( $props->{"svk:merge"} ) {
3019 $self->find_extra_svk_parents
3020 ($ed, $props->{"svk:merge"}, \@parents);
3021 }
3022 if ( $props->{"svn:mergeinfo"} ) {
3023 $self->find_extra_svn_parents
3024 ($ed,
3025 $props->{"svn:mergeinfo"},
3026 \@parents);
3027 }
3028 }
3030 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3031 print $un "r$rev\n" or croak $!;
3032 print $un $_, "\n" foreach @$untracked;
3033 my %log_entry = ( parents => \@parents, revision => $rev,
3034 log => '');
3036 my $headrev;
3037 my $logged = delete $self->{logged_rev_props};
3038 if (!$logged || $self->{-want_revprops}) {
3039 my $rp = $self->ra->rev_proplist($rev);
3040 foreach (sort keys %$rp) {
3041 my $v = $rp->{$_};
3042 if (/^svn:(author|date|log)$/) {
3043 $log_entry{$1} = $v;
3044 } elsif ($_ eq 'svm:headrev') {
3045 $headrev = $v;
3046 } else {
3047 print $un " rev_prop: ", uri_encode($_), ' ',
3048 uri_encode($v), "\n";
3049 }
3050 }
3051 } else {
3052 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3053 }
3054 close $un or croak $!;
3056 $log_entry{date} = parse_svn_date($log_entry{date});
3057 $log_entry{log} .= "\n";
3058 my $author = $log_entry{author} = check_author($log_entry{author});
3059 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3060 : ($author, undef);
3062 my ($commit_name, $commit_email) = ($name, $email);
3063 if ($_use_log_author) {
3064 my $name_field;
3065 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3066 $name_field = $1;
3067 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3068 $name_field = $1;
3069 }
3070 if (!defined $name_field) {
3071 if (!defined $email) {
3072 $email = $name;
3073 }
3074 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3075 ($name, $email) = ($1, $2);
3076 } elsif ($name_field =~ /(.*)@/) {
3077 ($name, $email) = ($1, $name_field);
3078 } else {
3079 ($name, $email) = ($name_field, $name_field);
3080 }
3081 }
3082 if (defined $headrev && $self->use_svm_props) {
3083 if ($self->rewrite_root) {
3084 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3085 "options set!\n";
3086 }
3087 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3088 # we don't want "SVM: initializing mirror for junk" ...
3089 return undef if $r == 0;
3090 my $svm = $self->svm;
3091 if ($uuid ne $svm->{uuid}) {
3092 die "UUID mismatch on SVM path:\n",
3093 "expected: $svm->{uuid}\n",
3094 " got: $uuid\n";
3095 }
3096 my $full_url = $self->full_url;
3097 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3098 die "Failed to replace '$svm->{replace}' with ",
3099 "'$svm->{source}' in $full_url\n";
3100 # throw away username for storing in records
3101 remove_username($full_url);
3102 $log_entry{metadata} = "$full_url\@$r $uuid";
3103 $log_entry{svm_revision} = $r;
3104 $email ||= "$author\@$uuid";
3105 $commit_email ||= "$author\@$uuid";
3106 } elsif ($self->use_svnsync_props) {
3107 my $full_url = $self->svnsync->{url};
3108 $full_url .= "/$self->{path}" if length $self->{path};
3109 remove_username($full_url);
3110 my $uuid = $self->svnsync->{uuid};
3111 $log_entry{metadata} = "$full_url\@$rev $uuid";
3112 $email ||= "$author\@$uuid";
3113 $commit_email ||= "$author\@$uuid";
3114 } else {
3115 my $url = $self->metadata_url;
3116 remove_username($url);
3117 $log_entry{metadata} = "$url\@$rev " .
3118 $self->ra->get_uuid;
3119 $email ||= "$author\@" . $self->ra->get_uuid;
3120 $commit_email ||= "$author\@" . $self->ra->get_uuid;
3121 }
3122 $log_entry{name} = $name;
3123 $log_entry{email} = $email;
3124 $log_entry{commit_name} = $commit_name;
3125 $log_entry{commit_email} = $commit_email;
3126 \%log_entry;
3127 }
3129 sub fetch {
3130 my ($self, $min_rev, $max_rev, @parents) = @_;
3131 my ($last_rev, $last_commit) = $self->last_rev_commit;
3132 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3133 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3134 }
3136 sub set_tree_cb {
3137 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3138 $self->{inject_parents} = { $rev => $tree };
3139 $self->fetch(undef, undef);
3140 }
3142 sub set_tree {
3143 my ($self, $tree) = (shift, shift);
3144 my $log_entry = ::get_commit_entry($tree);
3145 unless ($self->{last_rev}) {
3146 ::fatal("Must have an existing revision to commit");
3147 }
3148 my %ed_opts = ( r => $self->{last_rev},
3149 log => $log_entry->{log},
3150 ra => $self->ra,
3151 tree_a => $self->{last_commit},
3152 tree_b => $tree,
3153 editor_cb => sub {
3154 $self->set_tree_cb($log_entry, $tree, @_) },
3155 svn_path => $self->{path} );
3156 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3157 print "No changes\nr$self->{last_rev} = $tree\n";
3158 }
3159 }
3161 sub rebuild_from_rev_db {
3162 my ($self, $path) = @_;
3163 my $r = -1;
3164 open my $fh, '<', $path or croak "open: $!";
3165 binmode $fh or croak "binmode: $!";
3166 while (<$fh>) {
3167 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3168 chomp($_);
3169 ++$r;
3170 next if $_ eq ('0' x 40);
3171 $self->rev_map_set($r, $_);
3172 print "r$r = $_\n";
3173 }
3174 close $fh or croak "close: $!";
3175 unlink $path or croak "unlink: $!";
3176 }
3178 sub rebuild {
3179 my ($self) = @_;
3180 my $map_path = $self->map_path;
3181 my $partial = (-e $map_path && ! -z $map_path);
3182 return unless ::verify_ref($self->refname.'^0');
3183 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3184 my $rev_db = $self->rev_db_path;
3185 $self->rebuild_from_rev_db($rev_db);
3186 if ($self->use_svm_props) {
3187 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3188 $self->rebuild_from_rev_db($svm_rev_db);
3189 }
3190 $self->unlink_rev_db_symlink;
3191 return;
3192 }
3193 print "Rebuilding $map_path ...\n" if (!$partial);
3194 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3195 (undef, undef));
3196 my ($log, $ctx) =
3197 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3198 ($head ? "$head.." : "") . $self->refname,
3199 '--');
3200 my $metadata_url = $self->metadata_url;
3201 remove_username($metadata_url);
3202 my $svn_uuid = $self->ra_uuid;
3203 my $c;
3204 while (<$log>) {
3205 if ( m{^commit ($::sha1)$} ) {
3206 $c = $1;
3207 next;
3208 }
3209 next unless s{^\s*(git-svn-id:)}{$1};
3210 my ($url, $rev, $uuid) = ::extract_metadata($_);
3211 remove_username($url);
3213 # ignore merges (from set-tree)
3214 next if (!defined $rev || !$uuid);
3216 # if we merged or otherwise started elsewhere, this is
3217 # how we break out of it
3218 if (($uuid ne $svn_uuid) ||
3219 ($metadata_url && $url && ($url ne $metadata_url))) {
3220 next;
3221 }
3222 if ($partial && $head) {
3223 print "Partial-rebuilding $map_path ...\n";
3224 print "Currently at $base_rev = $head\n";
3225 $head = undef;
3226 }
3228 $self->rev_map_set($rev, $c);
3229 print "r$rev = $c\n";
3230 }
3231 command_close_pipe($log, $ctx);
3232 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3233 my $rev_db_path = $self->rev_db_path;
3234 if (-f $self->rev_db_path) {
3235 unlink $self->rev_db_path or croak "unlink: $!";
3236 }
3237 $self->unlink_rev_db_symlink;
3238 }
3240 # rev_map:
3241 # Tie::File seems to be prone to offset errors if revisions get sparse,
3242 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
3243 # one of my favorite modules is out :< Next up would be one of the DBM
3244 # modules, but I'm not sure which is most portable...
3245 #
3246 # This is the replacement for the rev_db format, which was too big
3247 # and inefficient for large repositories with a lot of sparse history
3248 # (mainly tags)
3249 #
3250 # The format is this:
3251 # - 24 bytes for every record,
3252 # * 4 bytes for the integer representing an SVN revision number
3253 # * 20 bytes representing the sha1 of a git commit
3254 # - No empty padding records like the old format
3255 # (except the last record, which can be overwritten)
3256 # - new records are written append-only since SVN revision numbers
3257 # increase monotonically
3258 # - lookups on SVN revision number are done via a binary search
3259 # - Piping the file to xxd -c24 is a good way of dumping it for
3260 # viewing or editing (piped back through xxd -r), should the need
3261 # ever arise.
3262 # - The last record can be padding revision with an all-zero sha1
3263 # This is used to optimize fetch performance when using multiple
3264 # "fetch" directives in .git/config
3265 #
3266 # These files are disposable unless noMetadata or useSvmProps is set
3268 sub _rev_map_set {
3269 my ($fh, $rev, $commit) = @_;
3271 binmode $fh or croak "binmode: $!";
3272 my $size = (stat($fh))[7];
3273 ($size % 24) == 0 or croak "inconsistent size: $size";
3275 my $wr_offset = 0;
3276 if ($size > 0) {
3277 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3278 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3279 $read == 24 or croak "read only $read bytes (!= 24)";
3280 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3281 if ($last_commit eq ('0' x40)) {
3282 if ($size >= 48) {
3283 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3284 $read = sysread($fh, $buf, 24) or
3285 croak "read: $!";
3286 $read == 24 or
3287 croak "read only $read bytes (!= 24)";
3288 ($last_rev, $last_commit) =
3289 unpack(rev_map_fmt, $buf);
3290 if ($last_commit eq ('0' x40)) {
3291 croak "inconsistent .rev_map\n";
3292 }
3293 }
3294 if ($last_rev >= $rev) {
3295 croak "last_rev is higher!: $last_rev >= $rev";
3296 }
3297 $wr_offset = -24;
3298 }
3299 }
3300 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3301 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3302 croak "write: $!";
3303 }
3305 sub _rev_map_reset {
3306 my ($fh, $rev, $commit) = @_;
3307 my $c = _rev_map_get($fh, $rev);
3308 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3309 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3310 truncate $fh, $offset or croak "truncate: $!";
3311 }
3313 sub mkfile {
3314 my ($path) = @_;
3315 unless (-e $path) {
3316 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3317 mkpath([$dir]) unless -d $dir;
3318 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3319 close $fh or die "Couldn't close (create) $path: $!\n";
3320 }
3321 }
3323 sub rev_map_set {
3324 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3325 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3326 my $db = $self->map_path($uuid);
3327 my $db_lock = "$db.lock";
3328 my $sig;
3329 $update_ref ||= 0;
3330 if ($update_ref) {
3331 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3332 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3333 }
3334 mkfile($db);
3336 $LOCKFILES{$db_lock} = 1;
3337 my $sync;
3338 # both of these options make our .rev_db file very, very important
3339 # and we can't afford to lose it because rebuild() won't work
3340 if ($self->use_svm_props || $self->no_metadata) {
3341 $sync = 1;
3342 copy($db, $db_lock) or die "rev_map_set(@_): ",
3343 "Failed to copy: ",
3344 "$db => $db_lock ($!)\n";
3345 } else {
3346 rename $db, $db_lock or die "rev_map_set(@_): ",
3347 "Failed to rename: ",
3348 "$db => $db_lock ($!)\n";
3349 }
3351 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3352 or croak "Couldn't open $db_lock: $!\n";
3353 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3354 _rev_map_set($fh, $rev, $commit);
3355 if ($sync) {
3356 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3357 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3358 }
3359 close $fh or croak $!;
3360 if ($update_ref) {
3361 $_head = $self;
3362 my $note = "";
3363 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3364 command_noisy('update-ref', '-m', "r$rev$note",
3365 $self->refname, $commit);
3366 }
3367 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3368 "$db_lock => $db ($!)\n";
3369 delete $LOCKFILES{$db_lock};
3370 if ($update_ref) {
3371 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3372 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3373 kill $sig, $$ if defined $sig;
3374 }
3375 }
3377 # If want_commit, this will return an array of (rev, commit) where
3378 # commit _must_ be a valid commit in the archive.
3379 # Otherwise, it'll return the max revision (whether or not the
3380 # commit is valid or just a 0x40 placeholder).
3381 sub rev_map_max {
3382 my ($self, $want_commit) = @_;
3383 $self->rebuild;
3384 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3385 $want_commit ? ($r, $c) : $r;
3386 }
3388 sub rev_map_max_norebuild {
3389 my ($self, $want_commit) = @_;
3390 my $map_path = $self->map_path;
3391 stat $map_path or return $want_commit ? (0, undef) : 0;
3392 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3393 binmode $fh or croak "binmode: $!";
3394 my $size = (stat($fh))[7];
3395 ($size % 24) == 0 or croak "inconsistent size: $size";
3397 if ($size == 0) {
3398 close $fh or croak "close: $!";
3399 return $want_commit ? (0, undef) : 0;
3400 }
3402 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3403 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3404 my ($r, $c) = unpack(rev_map_fmt, $buf);
3405 if ($want_commit && $c eq ('0' x40)) {
3406 if ($size < 48) {
3407 return $want_commit ? (0, undef) : 0;
3408 }
3409 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3410 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3411 ($r, $c) = unpack(rev_map_fmt, $buf);
3412 if ($c eq ('0'x40)) {
3413 croak "Penultimate record is all-zeroes in $map_path";
3414 }
3415 }
3416 close $fh or croak "close: $!";
3417 $want_commit ? ($r, $c) : $r;
3418 }
3420 sub rev_map_get {
3421 my ($self, $rev, $uuid) = @_;
3422 my $map_path = $self->map_path($uuid);
3423 return undef unless -e $map_path;
3425 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3426 my $c = _rev_map_get($fh, $rev);
3427 close($fh) or croak "close: $!";
3428 $c
3429 }
3431 sub _rev_map_get {
3432 my ($fh, $rev) = @_;
3434 binmode $fh or croak "binmode: $!";
3435 my $size = (stat($fh))[7];
3436 ($size % 24) == 0 or croak "inconsistent size: $size";
3438 if ($size == 0) {
3439 return undef;
3440 }
3442 my ($l, $u) = (0, $size - 24);
3443 my ($r, $c, $buf);
3445 while ($l <= $u) {
3446 my $i = int(($l/24 + $u/24) / 2) * 24;
3447 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3448 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3449 my ($r, $c) = unpack(rev_map_fmt, $buf);
3451 if ($r < $rev) {
3452 $l = $i + 24;
3453 } elsif ($r > $rev) {
3454 $u = $i - 24;
3455 } else { # $r == $rev
3456 return $c eq ('0' x 40) ? undef : $c;
3457 }
3458 }
3459 undef;
3460 }
3462 # Finds the first svn revision that exists on (if $eq_ok is true) or
3463 # before $rev for the current branch. It will not search any lower
3464 # than $min_rev. Returns the git commit hash and svn revision number
3465 # if found, else (undef, undef).
3466 sub find_rev_before {
3467 my ($self, $rev, $eq_ok, $min_rev) = @_;
3468 --$rev unless $eq_ok;
3469 $min_rev ||= 1;
3470 my $max_rev = $self->rev_map_max;
3471 $rev = $max_rev if ($rev > $max_rev);
3472 while ($rev >= $min_rev) {
3473 if (my $c = $self->rev_map_get($rev)) {
3474 return ($rev, $c);
3475 }
3476 --$rev;
3477 }
3478 return (undef, undef);
3479 }
3481 # Finds the first svn revision that exists on (if $eq_ok is true) or
3482 # after $rev for the current branch. It will not search any higher
3483 # than $max_rev. Returns the git commit hash and svn revision number
3484 # if found, else (undef, undef).
3485 sub find_rev_after {
3486 my ($self, $rev, $eq_ok, $max_rev) = @_;
3487 ++$rev unless $eq_ok;
3488 $max_rev ||= $self->rev_map_max;
3489 while ($rev <= $max_rev) {
3490 if (my $c = $self->rev_map_get($rev)) {
3491 return ($rev, $c);
3492 }
3493 ++$rev;
3494 }
3495 return (undef, undef);
3496 }
3498 sub _new {
3499 my ($class, $repo_id, $ref_id, $path) = @_;
3500 unless (defined $repo_id && length $repo_id) {
3501 $repo_id = $Git::SVN::default_repo_id;
3502 }
3503 unless (defined $ref_id && length $ref_id) {
3504 $_prefix = '' unless defined($_prefix);
3505 $_[2] = $ref_id =
3506 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3507 }
3508 $_[1] = $repo_id;
3509 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3511 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3512 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3513 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3514 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3515 if (-d $old_dir && ! -d $dir) {
3516 $dir = $old_dir;
3517 }
3518 }
3520 $_[3] = $path = '' unless (defined $path);
3521 mkpath([$dir]);
3522 bless {
3523 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3524 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3525 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3526 }
3528 # for read-only access of old .rev_db formats
3529 sub unlink_rev_db_symlink {
3530 my ($self) = @_;
3531 my $link = $self->rev_db_path;
3532 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3533 if (-l $link) {
3534 unlink $link or croak "unlink: $link failed!";
3535 }
3536 }
3538 sub rev_db_path {
3539 my ($self, $uuid) = @_;
3540 my $db_path = $self->map_path($uuid);
3541 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3542 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3543 $db_path;
3544 }
3546 # the new replacement for .rev_db
3547 sub map_path {
3548 my ($self, $uuid) = @_;
3549 $uuid ||= $self->ra_uuid;
3550 "$self->{map_root}.$uuid";
3551 }
3553 sub uri_encode {
3554 my ($f) = @_;
3555 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3556 $f
3557 }
3559 sub remove_username {
3560 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3561 }
3563 package Git::SVN::Prompt;
3564 use strict;
3565 use warnings;
3566 require SVN::Core;
3567 use vars qw/$_no_auth_cache $_username/;
3569 sub simple {
3570 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3571 $may_save = undef if $_no_auth_cache;
3572 $default_username = $_username if defined $_username;
3573 if (defined $default_username && length $default_username) {
3574 if (defined $realm && length $realm) {
3575 print STDERR "Authentication realm: $realm\n";
3576 STDERR->flush;
3577 }
3578 $cred->username($default_username);
3579 } else {
3580 username($cred, $realm, $may_save, $pool);
3581 }
3582 $cred->password(_read_password("Password for '" .
3583 $cred->username . "': ", $realm));
3584 $cred->may_save($may_save);
3585 $SVN::_Core::SVN_NO_ERROR;
3586 }
3588 sub ssl_server_trust {
3589 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3590 $may_save = undef if $_no_auth_cache;
3591 print STDERR "Error validating server certificate for '$realm':\n";
3592 {
3593 no warnings 'once';
3594 # All variables SVN::Auth::SSL::* are used only once,
3595 # so we're shutting up Perl warnings about this.
3596 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3597 print STDERR " - The certificate is not issued ",
3598 "by a trusted authority. Use the\n",
3599 " fingerprint to validate ",
3600 "the certificate manually!\n";
3601 }
3602 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3603 print STDERR " - The certificate hostname ",
3604 "does not match.\n";
3605 }
3606 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3607 print STDERR " - The certificate is not yet valid.\n";
3608 }
3609 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3610 print STDERR " - The certificate has expired.\n";
3611 }
3612 if ($failures & $SVN::Auth::SSL::OTHER) {
3613 print STDERR " - The certificate has ",
3614 "an unknown error.\n";
3615 }
3616 } # no warnings 'once'
3617 printf STDERR
3618 "Certificate information:\n".
3619 " - Hostname: %s\n".
3620 " - Valid: from %s until %s\n".
3621 " - Issuer: %s\n".
3622 " - Fingerprint: %s\n",
3623 map $cert_info->$_, qw(hostname valid_from valid_until
3624 issuer_dname fingerprint);
3625 my $choice;
3626 prompt:
3627 print STDERR $may_save ?
3628 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3629 "(R)eject or accept (t)emporarily? ";
3630 STDERR->flush;
3631 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3632 if ($choice =~ /^t$/i) {
3633 $cred->may_save(undef);
3634 } elsif ($choice =~ /^r$/i) {
3635 return -1;
3636 } elsif ($may_save && $choice =~ /^p$/i) {
3637 $cred->may_save($may_save);
3638 } else {
3639 goto prompt;
3640 }
3641 $cred->accepted_failures($failures);
3642 $SVN::_Core::SVN_NO_ERROR;
3643 }
3645 sub ssl_client_cert {
3646 my ($cred, $realm, $may_save, $pool) = @_;
3647 $may_save = undef if $_no_auth_cache;
3648 print STDERR "Client certificate filename: ";
3649 STDERR->flush;
3650 chomp(my $filename = <STDIN>);
3651 $cred->cert_file($filename);
3652 $cred->may_save($may_save);
3653 $SVN::_Core::SVN_NO_ERROR;
3654 }
3656 sub ssl_client_cert_pw {
3657 my ($cred, $realm, $may_save, $pool) = @_;
3658 $may_save = undef if $_no_auth_cache;
3659 $cred->password(_read_password("Password: ", $realm));
3660 $cred->may_save($may_save);
3661 $SVN::_Core::SVN_NO_ERROR;
3662 }
3664 sub username {
3665 my ($cred, $realm, $may_save, $pool) = @_;
3666 $may_save = undef if $_no_auth_cache;
3667 if (defined $realm && length $realm) {
3668 print STDERR "Authentication realm: $realm\n";
3669 }
3670 my $username;
3671 if (defined $_username) {
3672 $username = $_username;
3673 } else {
3674 print STDERR "Username: ";
3675 STDERR->flush;
3676 chomp($username = <STDIN>);
3677 }
3678 $cred->username($username);
3679 $cred->may_save($may_save);
3680 $SVN::_Core::SVN_NO_ERROR;
3681 }
3683 sub _read_password {
3684 my ($prompt, $realm) = @_;
3685 print STDERR $prompt;
3686 STDERR->flush;
3687 require Term::ReadKey;
3688 Term::ReadKey::ReadMode('noecho');
3689 my $password = '';
3690 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3691 last if $key =~ /[\012\015]/; # \n\r
3692 $password .= $key;
3693 }
3694 Term::ReadKey::ReadMode('restore');
3695 print STDERR "\n";
3696 STDERR->flush;
3697 $password;
3698 }
3700 package SVN::Git::Fetcher;
3701 use vars qw/@ISA/;
3702 use strict;
3703 use warnings;
3704 use Carp qw/croak/;
3705 use File::Temp qw/tempfile/;
3706 use IO::File qw//;
3707 use vars qw/$_ignore_regex/;
3709 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
3710 sub new {
3711 my ($class, $git_svn, $switch_path) = @_;
3712 my $self = SVN::Delta::Editor->new;
3713 bless $self, $class;
3714 if (exists $git_svn->{last_commit}) {
3715 $self->{c} = $git_svn->{last_commit};
3716 $self->{empty_symlinks} =
3717 _mark_empty_symlinks($git_svn, $switch_path);
3718 }
3719 $self->{ignore_regex} = eval { command_oneline('config', '--get',
3720 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3721 $self->{empty} = {};
3722 $self->{dir_prop} = {};
3723 $self->{file_prop} = {};
3724 $self->{absent_dir} = {};
3725 $self->{absent_file} = {};
3726 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3727 $self;
3728 }
3730 # this uses the Ra object, so it must be called before do_{switch,update},
3731 # not inside them (when the Git::SVN::Fetcher object is passed) to
3732 # do_{switch,update}
3733 sub _mark_empty_symlinks {
3734 my ($git_svn, $switch_path) = @_;
3735 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3736 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3738 my %ret;
3739 my ($rev, $cmt) = $git_svn->last_rev_commit;
3740 return {} unless ($rev && $cmt);
3742 # allow the warning to be printed for each revision we fetch to
3743 # ensure the user sees it. The user can also disable the workaround
3744 # on the repository even while git svn is running and the next
3745 # revision fetched will skip this expensive function.
3746 my $printed_warning;
3747 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3748 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3749 local $/ = "\0";
3750 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3751 $pfx .= '/' if length($pfx);
3752 while (<$ls>) {
3753 chomp;
3754 s/\A100644 blob $empty_blob\t//o or next;
3755 unless ($printed_warning) {
3756 print STDERR "Scanning for empty symlinks, ",
3757 "this may take a while if you have ",
3758 "many empty files\n",
3759 "You may disable this with `",
3760 "git config svn.brokenSymlinkWorkaround ",
3761 "false'.\n",
3762 "This may be done in a different ",
3763 "terminal without restarting ",
3764 "git svn\n";
3765 $printed_warning = 1;
3766 }
3767 my $path = $_;
3768 my (undef, $props) =
3769 $git_svn->ra->get_file($pfx.$path, $rev, undef);
3770 if ($props->{'svn:special'}) {
3771 $ret{$path} = 1;
3772 }
3773 }
3774 command_close_pipe($ls, $ctx);
3775 \%ret;
3776 }
3778 # returns true if a given path is inside a ".git" directory
3779 sub in_dot_git {
3780 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3781 }
3783 # return value: 0 -- don't ignore, 1 -- ignore
3784 sub is_path_ignored {
3785 my ($self, $path) = @_;
3786 return 1 if in_dot_git($path);
3787 return 1 if defined($self->{ignore_regex}) &&
3788 $path =~ m!$self->{ignore_regex}!;
3789 return 0 unless defined($_ignore_regex);
3790 return 1 if $path =~ m!$_ignore_regex!o;
3791 return 0;
3792 }
3794 sub set_path_strip {
3795 my ($self, $path) = @_;
3796 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3797 }
3799 sub open_root {
3800 { path => '' };
3801 }
3803 sub open_directory {
3804 my ($self, $path, $pb, $rev) = @_;
3805 { path => $path };
3806 }
3808 sub git_path {
3809 my ($self, $path) = @_;
3810 if ($self->{path_strip}) {
3811 $path =~ s!$self->{path_strip}!! or
3812 die "Failed to strip path '$path' ($self->{path_strip})\n";
3813 }
3814 $path;
3815 }
3817 sub delete_entry {
3818 my ($self, $path, $rev, $pb) = @_;
3819 return undef if $self->is_path_ignored($path);
3821 my $gpath = $self->git_path($path);
3822 return undef if ($gpath eq '');
3824 # remove entire directories.
3825 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3826 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3827 if ($tree) {
3828 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3829 -r --name-only -z/,
3830 $tree);
3831 local $/ = "\0";
3832 while (<$ls>) {
3833 chomp;
3834 my $rmpath = "$gpath/$_";
3835 $self->{gii}->remove($rmpath);
3836 print "\tD\t$rmpath\n" unless $::_q;
3837 }
3838 print "\tD\t$gpath/\n" unless $::_q;
3839 command_close_pipe($ls, $ctx);
3840 $self->{empty}->{$path} = 0
3841 } else {
3842 $self->{gii}->remove($gpath);
3843 print "\tD\t$gpath\n" unless $::_q;
3844 }
3845 undef;
3846 }
3848 sub open_file {
3849 my ($self, $path, $pb, $rev) = @_;
3850 my ($mode, $blob);
3852 goto out if $self->is_path_ignored($path);
3854 my $gpath = $self->git_path($path);
3855 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3856 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3857 unless (defined $mode && defined $blob) {
3858 die "$path was not found in commit $self->{c} (r$rev)\n";
3859 }
3860 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3861 $mode = '120000';
3862 }
3863 out:
3864 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3865 pool => SVN::Pool->new, action => 'M' };
3866 }
3868 sub add_file {
3869 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3870 my $mode;
3872 if (!$self->is_path_ignored($path)) {
3873 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3874 delete $self->{empty}->{$dir};
3875 $mode = '100644';
3876 }
3877 { path => $path, mode_a => $mode, mode_b => $mode,
3878 pool => SVN::Pool->new, action => 'A' };
3879 }
3881 sub add_directory {
3882 my ($self, $path, $cp_path, $cp_rev) = @_;
3883 goto out if $self->is_path_ignored($path);
3884 my $gpath = $self->git_path($path);
3885 if ($gpath eq '') {
3886 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3887 -r --name-only -z/,
3888 $self->{c});
3889 local $/ = "\0";
3890 while (<$ls>) {
3891 chomp;
3892 $self->{gii}->remove($_);
3893 print "\tD\t$_\n" unless $::_q;
3894 }
3895 command_close_pipe($ls, $ctx);
3896 $self->{empty}->{$path} = 0;
3897 }
3898 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3899 delete $self->{empty}->{$dir};
3900 $self->{empty}->{$path} = 1;
3901 out:
3902 { path => $path };
3903 }
3905 sub change_dir_prop {
3906 my ($self, $db, $prop, $value) = @_;
3907 return undef if $self->is_path_ignored($db->{path});
3908 $self->{dir_prop}->{$db->{path}} ||= {};
3909 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3910 undef;
3911 }
3913 sub absent_directory {
3914 my ($self, $path, $pb) = @_;
3915 return undef if $self->is_path_ignored($path);
3916 $self->{absent_dir}->{$pb->{path}} ||= [];
3917 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3918 undef;
3919 }
3921 sub absent_file {
3922 my ($self, $path, $pb) = @_;
3923 return undef if $self->is_path_ignored($path);
3924 $self->{absent_file}->{$pb->{path}} ||= [];
3925 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3926 undef;
3927 }
3929 sub change_file_prop {
3930 my ($self, $fb, $prop, $value) = @_;
3931 return undef if $self->is_path_ignored($fb->{path});
3932 if ($prop eq 'svn:executable') {
3933 if ($fb->{mode_b} != 120000) {
3934 $fb->{mode_b} = defined $value ? 100755 : 100644;
3935 }
3936 } elsif ($prop eq 'svn:special') {
3937 $fb->{mode_b} = defined $value ? 120000 : 100644;
3938 } else {
3939 $self->{file_prop}->{$fb->{path}} ||= {};
3940 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3941 }
3942 undef;
3943 }
3945 sub apply_textdelta {
3946 my ($self, $fb, $exp) = @_;
3947 return undef if $self->is_path_ignored($fb->{path});
3948 my $fh = $::_repository->temp_acquire('svn_delta');
3949 # $fh gets auto-closed() by SVN::TxDelta::apply(),
3950 # (but $base does not,) so dup() it for reading in close_file
3951 open my $dup, '<&', $fh or croak $!;
3952 my $base = $::_repository->temp_acquire('git_blob');
3954 if ($fb->{blob}) {
3955 my ($base_is_link, $size);
3957 if ($fb->{mode_a} eq '120000' &&
3958 ! $self->{empty_symlinks}->{$fb->{path}}) {
3959 print $base 'link ' or die "print $!\n";
3960 $base_is_link = 1;
3961 }
3962 retry:
3963 $size = $::_repository->cat_blob($fb->{blob}, $base);
3964 die "Failed to read object $fb->{blob}" if ($size < 0);
3966 if (defined $exp) {
3967 seek $base, 0, 0 or croak $!;
3968 my $got = ::md5sum($base);
3969 if ($got ne $exp) {
3970 my $err = "Checksum mismatch: ".
3971 "$fb->{path} $fb->{blob}\n" .
3972 "expected: $exp\n" .
3973 " got: $got\n";
3974 if ($base_is_link) {
3975 warn $err,
3976 "Retrying... (possibly ",
3977 "a bad symlink from SVN)\n";
3978 $::_repository->temp_reset($base);
3979 $base_is_link = 0;
3980 goto retry;
3981 }
3982 die $err;
3983 }
3984 }
3985 }
3986 seek $base, 0, 0 or croak $!;
3987 $fb->{fh} = $fh;
3988 $fb->{base} = $base;
3989 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3990 }
3992 sub close_file {
3993 my ($self, $fb, $exp) = @_;
3994 return undef if $self->is_path_ignored($fb->{path});
3996 my $hash;
3997 my $path = $self->git_path($fb->{path});
3998 if (my $fh = $fb->{fh}) {
3999 if (defined $exp) {
4000 seek($fh, 0, 0) or croak $!;
4001 my $got = ::md5sum($fh);
4002 if ($got ne $exp) {
4003 die "Checksum mismatch: $path\n",
4004 "expected: $exp\n got: $got\n";
4005 }
4006 }
4007 if ($fb->{mode_b} == 120000) {
4008 sysseek($fh, 0, 0) or croak $!;
4009 my $rd = sysread($fh, my $buf, 5);
4011 if (!defined $rd) {
4012 croak "sysread: $!\n";
4013 } elsif ($rd == 0) {
4014 warn "$path has mode 120000",
4015 " but it points to nothing\n",
4016 "converting to an empty file with mode",
4017 " 100644\n";
4018 $fb->{mode_b} = '100644';
4019 } elsif ($buf ne 'link ') {
4020 warn "$path has mode 120000",
4021 " but is not a link\n";
4022 } else {
4023 my $tmp_fh = $::_repository->temp_acquire(
4024 'svn_hash');
4025 my $res;
4026 while ($res = sysread($fh, my $str, 1024)) {
4027 my $out = syswrite($tmp_fh, $str, $res);
4028 defined($out) && $out == $res
4029 or croak("write ",
4030 Git::temp_path($tmp_fh),
4031 ": $!\n");
4032 }
4033 defined $res or croak $!;
4035 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4036 Git::temp_release($tmp_fh, 1);
4037 }
4038 }
4040 $hash = $::_repository->hash_and_insert_object(
4041 Git::temp_path($fh));
4042 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4044 Git::temp_release($fb->{base}, 1);
4045 Git::temp_release($fh, 1);
4046 } else {
4047 $hash = $fb->{blob} or die "no blob information\n";
4048 }
4049 $fb->{pool}->clear;
4050 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4051 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4052 undef;
4053 }
4055 sub abort_edit {
4056 my $self = shift;
4057 $self->{nr} = $self->{gii}->{nr};
4058 delete $self->{gii};
4059 $self->SUPER::abort_edit(@_);
4060 }
4062 sub close_edit {
4063 my $self = shift;
4064 $self->{git_commit_ok} = 1;
4065 $self->{nr} = $self->{gii}->{nr};
4066 delete $self->{gii};
4067 $self->SUPER::close_edit(@_);
4068 }
4070 package SVN::Git::Editor;
4071 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4072 use strict;
4073 use warnings;
4074 use Carp qw/croak/;
4075 use IO::File;
4077 sub new {
4078 my ($class, $opts) = @_;
4079 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4080 die "$_ required!\n" unless (defined $opts->{$_});
4081 }
4083 my $pool = SVN::Pool->new;
4084 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4085 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4086 $opts->{r}, $mods);
4088 # $opts->{ra} functions should not be used after this:
4089 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4090 $opts->{editor_cb}, $pool);
4091 my $self = SVN::Delta::Editor->new(@ce, $pool);
4092 bless $self, $class;
4093 foreach (qw/svn_path r tree_a tree_b/) {
4094 $self->{$_} = $opts->{$_};
4095 }
4096 $self->{url} = $opts->{ra}->{url};
4097 $self->{mods} = $mods;
4098 $self->{types} = $types;
4099 $self->{pool} = $pool;
4100 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4101 $self->{rm} = { };
4102 $self->{path_prefix} = length $self->{svn_path} ?
4103 "$self->{svn_path}/" : '';
4104 $self->{config} = $opts->{config};
4105 return $self;
4106 }
4108 sub generate_diff {
4109 my ($tree_a, $tree_b) = @_;
4110 my @diff_tree = qw(diff-tree -z -r);
4111 if ($_cp_similarity) {
4112 push @diff_tree, "-C$_cp_similarity";
4113 } else {
4114 push @diff_tree, '-C';
4115 }
4116 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4117 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4118 push @diff_tree, $tree_a, $tree_b;
4119 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4120 local $/ = "\0";
4121 my $state = 'meta';
4122 my @mods;
4123 while (<$diff_fh>) {
4124 chomp $_; # this gets rid of the trailing "\0"
4125 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4126 ($::sha1)\s($::sha1)\s
4127 ([MTCRAD])\d*$/xo) {
4128 push @mods, { mode_a => $1, mode_b => $2,
4129 sha1_a => $3, sha1_b => $4,
4130 chg => $5 };
4131 if ($5 =~ /^(?:C|R)$/) {
4132 $state = 'file_a';
4133 } else {
4134 $state = 'file_b';
4135 }
4136 } elsif ($state eq 'file_a') {
4137 my $x = $mods[$#mods] or croak "Empty array\n";
4138 if ($x->{chg} !~ /^(?:C|R)$/) {
4139 croak "Error parsing $_, $x->{chg}\n";
4140 }
4141 $x->{file_a} = $_;
4142 $state = 'file_b';
4143 } elsif ($state eq 'file_b') {
4144 my $x = $mods[$#mods] or croak "Empty array\n";
4145 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4146 croak "Error parsing $_, $x->{chg}\n";
4147 }
4148 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4149 croak "Error parsing $_, $x->{chg}\n";
4150 }
4151 $x->{file_b} = $_;
4152 $state = 'meta';
4153 } else {
4154 croak "Error parsing $_\n";
4155 }
4156 }
4157 command_close_pipe($diff_fh, $ctx);
4158 \@mods;
4159 }
4161 sub check_diff_paths {
4162 my ($ra, $pfx, $rev, $mods) = @_;
4163 my %types;
4164 $pfx .= '/' if length $pfx;
4166 sub type_diff_paths {
4167 my ($ra, $types, $path, $rev) = @_;
4168 my @p = split m#/+#, $path;
4169 my $c = shift @p;
4170 unless (defined $types->{$c}) {
4171 $types->{$c} = $ra->check_path($c, $rev);
4172 }
4173 while (@p) {
4174 $c .= '/' . shift @p;
4175 next if defined $types->{$c};
4176 $types->{$c} = $ra->check_path($c, $rev);
4177 }
4178 }
4180 foreach my $m (@$mods) {
4181 foreach my $f (qw/file_a file_b/) {
4182 next unless defined $m->{$f};
4183 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4184 if (length $pfx.$dir && ! defined $types{$dir}) {
4185 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4186 }
4187 }
4188 }
4189 \%types;
4190 }
4192 sub split_path {
4193 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4194 }
4196 sub repo_path {
4197 my ($self, $path) = @_;
4198 $self->{path_prefix}.(defined $path ? $path : '');
4199 }
4201 sub url_path {
4202 my ($self, $path) = @_;
4203 if ($self->{url} =~ m#^https?://#) {
4204 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4205 }
4206 $self->{url} . '/' . $self->repo_path($path);
4207 }
4209 sub rmdirs {
4210 my ($self) = @_;
4211 my $rm = $self->{rm};
4212 delete $rm->{''}; # we never delete the url we're tracking
4213 return unless %$rm;
4215 foreach (keys %$rm) {
4216 my @d = split m#/#, $_;
4217 my $c = shift @d;
4218 $rm->{$c} = 1;
4219 while (@d) {
4220 $c .= '/' . shift @d;
4221 $rm->{$c} = 1;
4222 }
4223 }
4224 delete $rm->{$self->{svn_path}};
4225 delete $rm->{''}; # we never delete the url we're tracking
4226 return unless %$rm;
4228 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4229 $self->{tree_b});
4230 local $/ = "\0";
4231 while (<$fh>) {
4232 chomp;
4233 my @dn = split m#/#, $_;
4234 while (pop @dn) {
4235 delete $rm->{join '/', @dn};
4236 }
4237 unless (%$rm) {
4238 close $fh;
4239 return;
4240 }
4241 }
4242 command_close_pipe($fh, $ctx);
4244 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4245 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4246 $self->close_directory($bat->{$d}, $p);
4247 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4248 print "\tD+\t$d/\n" unless $::_q;
4249 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4250 delete $bat->{$d};
4251 }
4252 }
4254 sub open_or_add_dir {
4255 my ($self, $full_path, $baton) = @_;
4256 my $t = $self->{types}->{$full_path};
4257 if (!defined $t) {
4258 die "$full_path not known in r$self->{r} or we have a bug!\n";
4259 }
4260 {
4261 no warnings 'once';
4262 # SVN::Node::none and SVN::Node::file are used only once,
4263 # so we're shutting up Perl's warnings about them.
4264 if ($t == $SVN::Node::none) {
4265 return $self->add_directory($full_path, $baton,
4266 undef, -1, $self->{pool});
4267 } elsif ($t == $SVN::Node::dir) {
4268 return $self->open_directory($full_path, $baton,
4269 $self->{r}, $self->{pool});
4270 } # no warnings 'once'
4271 print STDERR "$full_path already exists in repository at ",
4272 "r$self->{r} and it is not a directory (",
4273 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4274 } # no warnings 'once'
4275 exit 1;
4276 }
4278 sub ensure_path {
4279 my ($self, $path) = @_;
4280 my $bat = $self->{bat};
4281 my $repo_path = $self->repo_path($path);
4282 return $bat->{''} unless (length $repo_path);
4283 my @p = split m#/+#, $repo_path;
4284 my $c = shift @p;
4285 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4286 while (@p) {
4287 my $c0 = $c;
4288 $c .= '/' . shift @p;
4289 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4290 }
4291 return $bat->{$c};
4292 }
4294 # Subroutine to convert a globbing pattern to a regular expression.
4295 # From perl cookbook.
4296 sub glob2pat {
4297 my $globstr = shift;
4298 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4299 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4300 return '^' . $globstr . '$';
4301 }
4303 sub check_autoprop {
4304 my ($self, $pattern, $properties, $file, $fbat) = @_;
4305 # Convert the globbing pattern to a regular expression.
4306 my $regex = glob2pat($pattern);
4307 # Check if the pattern matches the file name.
4308 if($file =~ m/($regex)/) {
4309 # Parse the list of properties to set.
4310 my @props = split(/;/, $properties);
4311 foreach my $prop (@props) {
4312 # Parse 'name=value' syntax and set the property.
4313 if ($prop =~ /([^=]+)=(.*)/) {
4314 my ($n,$v) = ($1,$2);
4315 for ($n, $v) {
4316 s/^\s+//; s/\s+$//;
4317 }
4318 $self->change_file_prop($fbat, $n, $v);
4319 }
4320 }
4321 }
4322 }
4324 sub apply_autoprops {
4325 my ($self, $file, $fbat) = @_;
4326 my $conf_t = ${$self->{config}}{'config'};
4327 no warnings 'once';
4328 # Check [miscellany]/enable-auto-props in svn configuration.
4329 if (SVN::_Core::svn_config_get_bool(
4330 $conf_t,
4331 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4332 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4333 0)) {
4334 # Auto-props are enabled. Enumerate them to look for matches.
4335 my $callback = sub {
4336 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4337 };
4338 SVN::_Core::svn_config_enumerate(
4339 $conf_t,
4340 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4341 $callback);
4342 }
4343 }
4345 sub A {
4346 my ($self, $m) = @_;
4347 my ($dir, $file) = split_path($m->{file_b});
4348 my $pbat = $self->ensure_path($dir);
4349 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4350 undef, -1);
4351 print "\tA\t$m->{file_b}\n" unless $::_q;
4352 $self->apply_autoprops($file, $fbat);
4353 $self->chg_file($fbat, $m);
4354 $self->close_file($fbat,undef,$self->{pool});
4355 }
4357 sub C {
4358 my ($self, $m) = @_;
4359 my ($dir, $file) = split_path($m->{file_b});
4360 my $pbat = $self->ensure_path($dir);
4361 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4362 $self->url_path($m->{file_a}), $self->{r});
4363 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4364 $self->chg_file($fbat, $m);
4365 $self->close_file($fbat,undef,$self->{pool});
4366 }
4368 sub delete_entry {
4369 my ($self, $path, $pbat) = @_;
4370 my $rpath = $self->repo_path($path);
4371 my ($dir, $file) = split_path($rpath);
4372 $self->{rm}->{$dir} = 1;
4373 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4374 }
4376 sub R {
4377 my ($self, $m) = @_;
4378 my ($dir, $file) = split_path($m->{file_b});
4379 my $pbat = $self->ensure_path($dir);
4380 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4381 $self->url_path($m->{file_a}), $self->{r});
4382 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4383 $self->apply_autoprops($file, $fbat);
4384 $self->chg_file($fbat, $m);
4385 $self->close_file($fbat,undef,$self->{pool});
4387 ($dir, $file) = split_path($m->{file_a});
4388 $pbat = $self->ensure_path($dir);
4389 $self->delete_entry($m->{file_a}, $pbat);
4390 }
4392 sub M {
4393 my ($self, $m) = @_;
4394 my ($dir, $file) = split_path($m->{file_b});
4395 my $pbat = $self->ensure_path($dir);
4396 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4397 $pbat,$self->{r},$self->{pool});
4398 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4399 $self->chg_file($fbat, $m);
4400 $self->close_file($fbat,undef,$self->{pool});
4401 }
4403 sub T { shift->M(@_) }
4405 sub change_file_prop {
4406 my ($self, $fbat, $pname, $pval) = @_;
4407 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4408 }
4410 sub _chg_file_get_blob ($$$$) {
4411 my ($self, $fbat, $m, $which) = @_;
4412 my $fh = $::_repository->temp_acquire("git_blob_$which");
4413 if ($m->{"mode_$which"} =~ /^120/) {
4414 print $fh 'link ' or croak $!;
4415 $self->change_file_prop($fbat,'svn:special','*');
4416 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4417 $self->change_file_prop($fbat,'svn:special',undef);
4418 }
4419 my $blob = $m->{"sha1_$which"};
4420 return ($fh,) if ($blob =~ /^0{40}$/);
4421 my $size = $::_repository->cat_blob($blob, $fh);
4422 croak "Failed to read object $blob" if ($size < 0);
4423 $fh->flush == 0 or croak $!;
4424 seek $fh, 0, 0 or croak $!;
4426 my $exp = ::md5sum($fh);
4427 seek $fh, 0, 0 or croak $!;
4428 return ($fh, $exp);
4429 }
4431 sub chg_file {
4432 my ($self, $fbat, $m) = @_;
4433 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4434 $self->change_file_prop($fbat,'svn:executable','*');
4435 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4436 $self->change_file_prop($fbat,'svn:executable',undef);
4437 }
4438 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4439 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4440 my $pool = SVN::Pool->new;
4441 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4442 if (-s $fh_a) {
4443 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4444 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4445 if (defined $res) {
4446 die "Unexpected result from send_txstream: $res\n",
4447 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4448 }
4449 } else {
4450 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4451 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4452 if ($got ne $exp_b);
4453 }
4454 Git::temp_release($fh_b, 1);
4455 Git::temp_release($fh_a, 1);
4456 $pool->clear;
4457 }
4459 sub D {
4460 my ($self, $m) = @_;
4461 my ($dir, $file) = split_path($m->{file_b});
4462 my $pbat = $self->ensure_path($dir);
4463 print "\tD\t$m->{file_b}\n" unless $::_q;
4464 $self->delete_entry($m->{file_b}, $pbat);
4465 }
4467 sub close_edit {
4468 my ($self) = @_;
4469 my ($p,$bat) = ($self->{pool}, $self->{bat});
4470 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4471 next if $_ eq '';
4472 $self->close_directory($bat->{$_}, $p);
4473 }
4474 $self->close_directory($bat->{''}, $p);
4475 $self->SUPER::close_edit($p);
4476 $p->clear;
4477 }
4479 sub abort_edit {
4480 my ($self) = @_;
4481 $self->SUPER::abort_edit($self->{pool});
4482 }
4484 sub DESTROY {
4485 my $self = shift;
4486 $self->SUPER::DESTROY(@_);
4487 $self->{pool}->clear;
4488 }
4490 # this drives the editor
4491 sub apply_diff {
4492 my ($self) = @_;
4493 my $mods = $self->{mods};
4494 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4495 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4496 my $f = $m->{chg};
4497 if (defined $o{$f}) {
4498 $self->$f($m);
4499 } else {
4500 fatal("Invalid change type: $f");
4501 }
4502 }
4503 $self->rmdirs if $_rmdir;
4504 if (@$mods == 0) {
4505 $self->abort_edit;
4506 } else {
4507 $self->close_edit;
4508 }
4509 return scalar @$mods;
4510 }
4512 package Git::SVN::Ra;
4513 use vars qw/@ISA $config_dir $_log_window_size/;
4514 use strict;
4515 use warnings;
4516 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4518 BEGIN {
4519 # enforce temporary pool usage for some simple functions
4520 no strict 'refs';
4521 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4522 get_file/) {
4523 my $SUPER = "SUPER::$f";
4524 *$f = sub {
4525 my $self = shift;
4526 my $pool = SVN::Pool->new;
4527 my @ret = $self->$SUPER(@_,$pool);
4528 $pool->clear;
4529 wantarray ? @ret : $ret[0];
4530 };
4531 }
4532 }
4534 sub _auth_providers () {
4535 [
4536 SVN::Client::get_simple_provider(),
4537 SVN::Client::get_ssl_server_trust_file_provider(),
4538 SVN::Client::get_simple_prompt_provider(
4539 \&Git::SVN::Prompt::simple, 2),
4540 SVN::Client::get_ssl_client_cert_file_provider(),
4541 SVN::Client::get_ssl_client_cert_prompt_provider(
4542 \&Git::SVN::Prompt::ssl_client_cert, 2),
4543 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4544 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4545 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4546 SVN::Client::get_username_provider(),
4547 SVN::Client::get_ssl_server_trust_prompt_provider(
4548 \&Git::SVN::Prompt::ssl_server_trust),
4549 SVN::Client::get_username_prompt_provider(
4550 \&Git::SVN::Prompt::username, 2)
4551 ]
4552 }
4554 sub escape_uri_only {
4555 my ($uri) = @_;
4556 my @tmp;
4557 foreach (split m{/}, $uri) {
4558 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4559 push @tmp, $_;
4560 }
4561 join('/', @tmp);
4562 }
4564 sub escape_url {
4565 my ($url) = @_;
4566 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4567 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4568 $url = "$scheme://$domain$uri";
4569 }
4570 $url;
4571 }
4573 sub new {
4574 my ($class, $url) = @_;
4575 $url =~ s!/+$!!;
4576 return $RA if ($RA && $RA->{url} eq $url);
4578 SVN::_Core::svn_config_ensure($config_dir, undef);
4579 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4580 my $config = SVN::Core::config_get_config($config_dir);
4581 $RA = undef;
4582 my $dont_store_passwords = 1;
4583 my $conf_t = ${$config}{'config'};
4584 {
4585 no warnings 'once';
4586 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4587 # produces warnings that variables are used only once.
4588 # I had not found the better way to shut them up, so
4589 # the warnings of type 'once' are disabled in this block.
4590 if (SVN::_Core::svn_config_get_bool($conf_t,
4591 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4592 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4593 1) == 0) {
4594 SVN::_Core::svn_auth_set_parameter($baton,
4595 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4596 bless (\$dont_store_passwords, "_p_void"));
4597 }
4598 if (SVN::_Core::svn_config_get_bool($conf_t,
4599 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4600 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4601 1) == 0) {
4602 $Git::SVN::Prompt::_no_auth_cache = 1;
4603 }
4604 } # no warnings 'once'
4605 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4606 config => $config,
4607 pool => SVN::Pool->new,
4608 auth_provider_callbacks => $callbacks);
4609 $self->{url} = $url;
4610 $self->{svn_path} = $url;
4611 $self->{repos_root} = $self->get_repos_root;
4612 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4613 $self->{cache} = { check_path => { r => 0, data => {} },
4614 get_dir => { r => 0, data => {} } };
4615 $RA = bless $self, $class;
4616 }
4618 sub check_path {
4619 my ($self, $path, $r) = @_;
4620 my $cache = $self->{cache}->{check_path};
4621 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4622 return $cache->{data}->{$path};
4623 }
4624 my $pool = SVN::Pool->new;
4625 my $t = $self->SUPER::check_path($path, $r, $pool);
4626 $pool->clear;
4627 if ($r != $cache->{r}) {
4628 %{$cache->{data}} = ();
4629 $cache->{r} = $r;
4630 }
4631 $cache->{data}->{$path} = $t;
4632 }
4634 sub get_dir {
4635 my ($self, $dir, $r) = @_;
4636 my $cache = $self->{cache}->{get_dir};
4637 if ($r == $cache->{r}) {
4638 if (my $x = $cache->{data}->{$dir}) {
4639 return wantarray ? @$x : $x->[0];
4640 }
4641 }
4642 my $pool = SVN::Pool->new;
4643 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4644 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4645 $pool->clear;
4646 if ($r != $cache->{r}) {
4647 %{$cache->{data}} = ();
4648 $cache->{r} = $r;
4649 }
4650 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4651 wantarray ? (\%dirents, $r, $props) : \%dirents;
4652 }
4654 sub DESTROY {
4655 # do not call the real DESTROY since we store ourselves in $RA
4656 }
4658 # get_log(paths, start, end, limit,
4659 # discover_changed_paths, strict_node_history, receiver)
4660 sub get_log {
4661 my ($self, @args) = @_;
4662 my $pool = SVN::Pool->new;
4664 # svn_log_changed_path_t objects passed to get_log are likely to be
4665 # overwritten even if only the refs are copied to an external variable,
4666 # so we should dup the structures in their entirety. Using an
4667 # externally passed pool (instead of our temporary and quickly cleared
4668 # pool in Git::SVN::Ra) does not help matters at all...
4669 my $receiver = pop @args;
4670 my $prefix = "/".$self->{svn_path};
4671 $prefix =~ s#/+($)##;
4672 my $prefix_regex = qr#^\Q$prefix\E#;
4673 push(@args, sub {
4674 my ($paths) = $_[0];
4675 return &$receiver(@_) unless $paths;
4676 $_[0] = ();
4677 foreach my $p (keys %$paths) {
4678 my $i = $paths->{$p};
4679 # Make path relative to our url, not repos_root
4680 $p =~ s/$prefix_regex//;
4681 my %s = map { $_ => $i->$_; }
4682 qw/copyfrom_path copyfrom_rev action/;
4683 if ($s{'copyfrom_path'}) {
4684 $s{'copyfrom_path'} =~ s/$prefix_regex//;
4685 }
4686 $_[0]{$p} = \%s;
4687 }
4688 &$receiver(@_);
4689 });
4692 # the limit parameter was not supported in SVN 1.1.x, so we
4693 # drop it. Therefore, the receiver callback passed to it
4694 # is made aware of this limitation by being wrapped if
4695 # the limit passed to is being wrapped.
4696 if ($SVN::Core::VERSION le '1.2.0') {
4697 my $limit = splice(@args, 3, 1);
4698 if ($limit > 0) {
4699 my $receiver = pop @args;
4700 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4701 }
4702 }
4703 my $ret = $self->SUPER::get_log(@args, $pool);
4704 $pool->clear;
4705 $ret;
4706 }
4708 sub trees_match {
4709 my ($self, $url1, $rev1, $url2, $rev2) = @_;
4710 my $ctx = SVN::Client->new(auth => _auth_providers);
4711 my $out = IO::File->new_tmpfile;
4713 # older SVN (1.1.x) doesn't take $pool as the last parameter for
4714 # $ctx->diff(), so we'll create a default one
4715 my $pool = SVN::Pool->new_default_sub;
4717 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4718 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4719 $out->flush;
4720 my $ret = (($out->stat)[7] == 0);
4721 close $out or croak $!;
4723 $ret;
4724 }
4726 sub get_commit_editor {
4727 my ($self, $log, $cb, $pool) = @_;
4728 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4729 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4730 }
4732 sub gs_do_update {
4733 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4734 my $new = ($rev_a == $rev_b);
4735 my $path = $gs->{path};
4737 if ($new && -e $gs->{index}) {
4738 unlink $gs->{index} or die
4739 "Couldn't unlink index: $gs->{index}: $!\n";
4740 }
4741 my $pool = SVN::Pool->new;
4742 $editor->set_path_strip($path);
4743 my (@pc) = split m#/#, $path;
4744 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4745 1, $editor, $pool);
4746 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4748 # Since we can't rely on svn_ra_reparent being available, we'll
4749 # just have to do some magic with set_path to make it so
4750 # we only want a partial path.
4751 my $sp = '';
4752 my $final = join('/', @pc);
4753 while (@pc) {
4754 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4755 $sp .= '/' if length $sp;
4756 $sp .= shift @pc;
4757 }
4758 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4760 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4762 $reporter->finish_report($pool);
4763 $pool->clear;
4764 $editor->{git_commit_ok};
4765 }
4767 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4768 # svn_ra_reparent didn't work before 1.4)
4769 sub gs_do_switch {
4770 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4771 my $path = $gs->{path};
4772 my $pool = SVN::Pool->new;
4774 my $full_url = $self->{url};
4775 my $old_url = $full_url;
4776 $full_url .= '/' . $path if length $path;
4777 my ($ra, $reparented);
4779 if ($old_url =~ m#^svn(\+ssh)?://# ||
4780 ($full_url =~ m#^https?://# &&
4781 escape_url($full_url) ne $full_url)) {
4782 $_[0] = undef;
4783 $self = undef;
4784 $RA = undef;
4785 $ra = Git::SVN::Ra->new($full_url);
4786 $ra_invalid = 1;
4787 } elsif ($old_url ne $full_url) {
4788 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4789 $self->{url} = $full_url;
4790 $reparented = 1;
4791 }
4793 $ra ||= $self;
4794 $url_b = escape_url($url_b);
4795 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4796 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4797 $reporter->set_path('', $rev_a, 0, @lock, $pool);
4798 $reporter->finish_report($pool);
4800 if ($reparented) {
4801 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4802 $self->{url} = $old_url;
4803 }
4805 $pool->clear;
4806 $editor->{git_commit_ok};
4807 }
4809 sub longest_common_path {
4810 my ($gsv, $globs) = @_;
4811 my %common;
4812 my $common_max = scalar @$gsv;
4814 foreach my $gs (@$gsv) {
4815 my @tmp = split m#/#, $gs->{path};
4816 my $p = '';
4817 foreach (@tmp) {
4818 $p .= length($p) ? "/$_" : $_;
4819 $common{$p} ||= 0;
4820 $common{$p}++;
4821 }
4822 }
4823 $globs ||= [];
4824 $common_max += scalar @$globs;
4825 foreach my $glob (@$globs) {
4826 my @tmp = split m#/#, $glob->{path}->{left};
4827 my $p = '';
4828 foreach (@tmp) {
4829 $p .= length($p) ? "/$_" : $_;
4830 $common{$p} ||= 0;
4831 $common{$p}++;
4832 }
4833 }
4835 my $longest_path = '';
4836 foreach (sort {length $b <=> length $a} keys %common) {
4837 if ($common{$_} == $common_max) {
4838 $longest_path = $_;
4839 last;
4840 }
4841 }
4842 $longest_path;
4843 }
4845 sub gs_fetch_loop_common {
4846 my ($self, $base, $head, $gsv, $globs) = @_;
4847 return if ($base > $head);
4848 my $inc = $_log_window_size;
4849 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4850 my $longest_path = longest_common_path($gsv, $globs);
4851 my $ra_url = $self->{url};
4852 my $find_trailing_edge;
4853 while (1) {
4854 my %revs;
4855 my $err;
4856 my $err_handler = $SVN::Error::handler;
4857 $SVN::Error::handler = sub {
4858 ($err) = @_;
4859 skip_unknown_revs($err);
4860 };
4861 sub _cb {
4862 my ($paths, $r, $author, $date, $log) = @_;
4863 [ $paths,
4864 { author => $author, date => $date, log => $log } ];
4865 }
4866 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4867 sub { $revs{$_[1]} = _cb(@_) });
4868 if ($err) {
4869 print "Checked through r$max\r";
4870 } else {
4871 $find_trailing_edge = 1;
4872 }
4873 if ($err and $find_trailing_edge) {
4874 print STDERR "Path '$longest_path' ",
4875 "was probably deleted:\n",
4876 $err->expanded_message,
4877 "\nWill attempt to follow ",
4878 "revisions r$min .. r$max ",
4879 "committed before the deletion\n";
4880 my $hi = $max;
4881 while (--$hi >= $min) {
4882 my $ok;
4883 $self->get_log([$longest_path], $min, $hi,
4884 0, 1, 1, sub {
4885 $ok = $_[1];
4886 $revs{$_[1]} = _cb(@_) });
4887 if ($ok) {
4888 print STDERR "r$min .. r$ok OK\n";
4889 last;
4890 }
4891 }
4892 $find_trailing_edge = 0;
4893 }
4894 $SVN::Error::handler = $err_handler;
4896 my %exists = map { $_->{path} => $_ } @$gsv;
4897 foreach my $r (sort {$a <=> $b} keys %revs) {
4898 my ($paths, $logged) = @{$revs{$r}};
4900 foreach my $gs ($self->match_globs(\%exists, $paths,
4901 $globs, $r)) {
4902 if ($gs->rev_map_max >= $r) {
4903 next;
4904 }
4905 next unless $gs->match_paths($paths, $r);
4906 $gs->{logged_rev_props} = $logged;
4907 if (my $last_commit = $gs->last_commit) {
4908 $gs->assert_index_clean($last_commit);
4909 }
4910 my $log_entry = $gs->do_fetch($paths, $r);
4911 if ($log_entry) {
4912 $gs->do_git_commit($log_entry);
4913 }
4914 $INDEX_FILES{$gs->{index}} = 1;
4915 }
4916 foreach my $g (@$globs) {
4917 my $k = "svn-remote.$g->{remote}." .
4918 "$g->{t}-maxRev";
4919 Git::SVN::tmp_config($k, $r);
4920 }
4921 if ($ra_invalid) {
4922 $_[0] = undef;
4923 $self = undef;
4924 $RA = undef;
4925 $self = Git::SVN::Ra->new($ra_url);
4926 $ra_invalid = undef;
4927 }
4928 }
4929 # pre-fill the .rev_db since it'll eventually get filled in
4930 # with '0' x40 if something new gets committed
4931 foreach my $gs (@$gsv) {
4932 next if $gs->rev_map_max >= $max;
4933 next if defined $gs->rev_map_get($max);
4934 $gs->rev_map_set($max, 0 x40);
4935 }
4936 foreach my $g (@$globs) {
4937 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4938 Git::SVN::tmp_config($k, $max);
4939 }
4940 last if $max >= $head;
4941 $min = $max + 1;
4942 $max += $inc;
4943 $max = $head if ($max > $head);
4944 }
4945 Git::SVN::gc();
4946 }
4948 sub get_dir_globbed {
4949 my ($self, $left, $depth, $r) = @_;
4951 my @x = eval { $self->get_dir($left, $r) };
4952 return unless scalar @x == 3;
4953 my $dirents = $x[0];
4954 my @finalents;
4955 foreach my $de (keys %$dirents) {
4956 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4957 if ($depth > 1) {
4958 my @args = ("$left/$de", $depth - 1, $r);
4959 foreach my $dir ($self->get_dir_globbed(@args)) {
4960 push @finalents, "$de/$dir";
4961 }
4962 } else {
4963 push @finalents, $de;
4964 }
4965 }
4966 @finalents;
4967 }
4969 sub match_globs {
4970 my ($self, $exists, $paths, $globs, $r) = @_;
4972 sub get_dir_check {
4973 my ($self, $exists, $g, $r) = @_;
4975 my @dirs = $self->get_dir_globbed($g->{path}->{left},
4976 $g->{path}->{depth},
4977 $r);
4979 foreach my $de (@dirs) {
4980 my $p = $g->{path}->full_path($de);
4981 next if $exists->{$p};
4982 next if (length $g->{path}->{right} &&
4983 ($self->check_path($p, $r) !=
4984 $SVN::Node::dir));
4985 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4986 $g->{ref}->full_path($de), 1);
4987 }
4988 }
4989 foreach my $g (@$globs) {
4990 if (my $path = $paths->{"/$g->{path}->{left}"}) {
4991 if ($path->{action} =~ /^[AR]$/) {
4992 get_dir_check($self, $exists, $g, $r);
4993 }
4994 }
4995 foreach (keys %$paths) {
4996 if (/$g->{path}->{left_regex}/ &&
4997 !/$g->{path}->{regex}/) {
4998 next if $paths->{$_}->{action} !~ /^[AR]$/;
4999 get_dir_check($self, $exists, $g, $r);
5000 }
5001 next unless /$g->{path}->{regex}/;
5002 my $p = $1;
5003 my $pathname = $g->{path}->full_path($p);
5004 next if $exists->{$pathname};
5005 next if ($self->check_path($pathname, $r) !=
5006 $SVN::Node::dir);
5007 $exists->{$pathname} = Git::SVN->init(
5008 $self->{url}, $pathname, undef,
5009 $g->{ref}->full_path($p), 1);
5010 }
5011 my $c = '';
5012 foreach (split m#/#, $g->{path}->{left}) {
5013 $c .= "/$_";
5014 next unless ($paths->{$c} &&
5015 ($paths->{$c}->{action} =~ /^[AR]$/));
5016 get_dir_check($self, $exists, $g, $r);
5017 }
5018 }
5019 values %$exists;
5020 }
5022 sub minimize_url {
5023 my ($self) = @_;
5024 return $self->{url} if ($self->{url} eq $self->{repos_root});
5025 my $url = $self->{repos_root};
5026 my @components = split(m!/!, $self->{svn_path});
5027 my $c = '';
5028 do {
5029 $url .= "/$c" if length $c;
5030 eval {
5031 my $ra = (ref $self)->new($url);
5032 my $latest = $ra->get_latest_revnum;
5033 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5034 };
5035 } while ($@ && ($c = shift @components));
5036 $url;
5037 }
5039 sub can_do_switch {
5040 my $self = shift;
5041 unless (defined $can_do_switch) {
5042 my $pool = SVN::Pool->new;
5043 my $rep = eval {
5044 $self->do_switch(1, '', 0, $self->{url},
5045 SVN::Delta::Editor->new, $pool);
5046 };
5047 if ($@) {
5048 $can_do_switch = 0;
5049 } else {
5050 $rep->abort_report($pool);
5051 $can_do_switch = 1;
5052 }
5053 $pool->clear;
5054 }
5055 $can_do_switch;
5056 }
5058 sub skip_unknown_revs {
5059 my ($err) = @_;
5060 my $errno = $err->apr_err();
5061 # Maybe the branch we're tracking didn't
5062 # exist when the repo started, so it's
5063 # not an error if it doesn't, just continue
5064 #
5065 # Wonderfully consistent library, eh?
5066 # 160013 - svn:// and file://
5067 # 175002 - http(s)://
5068 # 175007 - http(s):// (this repo required authorization, too...)
5069 # More codes may be discovered later...
5070 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5071 my $err_key = $err->expanded_message;
5072 # revision numbers change every time, filter them out
5073 $err_key =~ s/\d+/\0/g;
5074 $err_key = "$errno\0$err_key";
5075 unless ($ignored_err{$err_key}) {
5076 warn "W: Ignoring error from SVN, path probably ",
5077 "does not exist: ($errno): ",
5078 $err->expanded_message,"\n";
5079 warn "W: Do not be alarmed at the above message ",
5080 "git-svn is just searching aggressively for ",
5081 "old history.\n",
5082 "This may take a while on large repositories\n";
5083 $ignored_err{$err_key} = 1;
5084 }
5085 return;
5086 }
5087 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5088 }
5090 package Git::SVN::Log;
5091 use strict;
5092 use warnings;
5093 use POSIX qw/strftime/;
5094 use Time::Local;
5095 use constant commit_log_separator => ('-' x 72) . "\n";
5096 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5097 %rusers $show_commit $incremental/;
5098 my $l_fmt;
5100 sub cmt_showable {
5101 my ($c) = @_;
5102 return 1 if defined $c->{r};
5104 # big commit message got truncated by the 16k pretty buffer in rev-list
5105 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5106 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5107 @{$c->{l}} = ();
5108 my @log = command(qw/cat-file commit/, $c->{c});
5110 # shift off the headers
5111 shift @log while ($log[0] ne '');
5112 shift @log;
5114 # TODO: make $c->{l} not have a trailing newline in the future
5115 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5117 (undef, $c->{r}, undef) = ::extract_metadata(
5118 (grep(/^git-svn-id: /, @log))[-1]);
5119 }
5120 return defined $c->{r};
5121 }
5123 sub log_use_color {
5124 return $color || Git->repository->get_colorbool('color.diff');
5125 }
5127 sub git_svn_log_cmd {
5128 my ($r_min, $r_max, @args) = @_;
5129 my $head = 'HEAD';
5130 my (@files, @log_opts);
5131 foreach my $x (@args) {
5132 if ($x eq '--' || @files) {
5133 push @files, $x;
5134 } else {
5135 if (::verify_ref("$x^0")) {
5136 $head = $x;
5137 } else {
5138 push @log_opts, $x;
5139 }
5140 }
5141 }
5143 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5144 $gs ||= Git::SVN->_new;
5145 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5146 $gs->refname);
5147 push @cmd, '-r' unless $non_recursive;
5148 push @cmd, qw/--raw --name-status/ if $verbose;
5149 push @cmd, '--color' if log_use_color();
5150 push @cmd, @log_opts;
5151 if (defined $r_max && $r_max == $r_min) {
5152 push @cmd, '--max-count=1';
5153 if (my $c = $gs->rev_map_get($r_max)) {
5154 push @cmd, $c;
5155 }
5156 } elsif (defined $r_max) {
5157 if ($r_max < $r_min) {
5158 ($r_min, $r_max) = ($r_max, $r_min);
5159 }
5160 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5161 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5162 # If there are no commits in the range, both $c_max and $c_min
5163 # will be undefined. If there is at least 1 commit in the
5164 # range, both will be defined.
5165 return () if !defined $c_min || !defined $c_max;
5166 if ($c_min eq $c_max) {
5167 push @cmd, '--max-count=1', $c_min;
5168 } else {
5169 push @cmd, '--boundary', "$c_min..$c_max";
5170 }
5171 }
5172 return (@cmd, @files);
5173 }
5175 # adapted from pager.c
5176 sub config_pager {
5177 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
5178 if (!defined $pager) {
5179 $pager = 'less';
5180 } elsif (length $pager == 0 || $pager eq 'cat') {
5181 $pager = undef;
5182 }
5183 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5184 }
5186 sub run_pager {
5187 return unless -t *STDOUT && defined $pager;
5188 pipe my ($rfd, $wfd) or return;
5189 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5190 if (!$pid) {
5191 open STDOUT, '>&', $wfd or
5192 ::fatal "Can't redirect to stdout: $!";
5193 return;
5194 }
5195 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5196 $ENV{LESS} ||= 'FRSX';
5197 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5198 }
5200 sub format_svn_date {
5201 # some systmes don't handle or mishandle %z, so be creative.
5202 my $t = shift || time;
5203 my $gm = timelocal(gmtime($t));
5204 my $sign = qw( + + - )[ $t <=> $gm ];
5205 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5206 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5207 }
5209 sub parse_git_date {
5210 my ($t, $tz) = @_;
5211 # Date::Parse isn't in the standard Perl distro :(
5212 if ($tz =~ s/^\+//) {
5213 $t += tz_to_s_offset($tz);
5214 } elsif ($tz =~ s/^\-//) {
5215 $t -= tz_to_s_offset($tz);
5216 }
5217 return $t;
5218 }
5220 sub set_local_timezone {
5221 if (defined $TZ) {
5222 $ENV{TZ} = $TZ;
5223 } else {
5224 delete $ENV{TZ};
5225 }
5226 }
5228 sub tz_to_s_offset {
5229 my ($tz) = @_;
5230 $tz =~ s/(\d\d)$//;
5231 return ($1 * 60) + ($tz * 3600);
5232 }
5234 sub get_author_info {
5235 my ($dest, $author, $t, $tz) = @_;
5236 $author =~ s/(?:^\s*|\s*$)//g;
5237 $dest->{a_raw} = $author;
5238 my $au;
5239 if ($::_authors) {
5240 $au = $rusers{$author} || undef;
5241 }
5242 if (!$au) {
5243 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5244 }
5245 $dest->{t} = $t;
5246 $dest->{tz} = $tz;
5247 $dest->{a} = $au;
5248 $dest->{t_utc} = parse_git_date($t, $tz);
5249 }
5251 sub process_commit {
5252 my ($c, $r_min, $r_max, $defer) = @_;
5253 if (defined $r_min && defined $r_max) {
5254 if ($r_min == $c->{r} && $r_min == $r_max) {
5255 show_commit($c);
5256 return 0;
5257 }
5258 return 1 if $r_min == $r_max;
5259 if ($r_min < $r_max) {
5260 # we need to reverse the print order
5261 return 0 if (defined $limit && --$limit < 0);
5262 push @$defer, $c;
5263 return 1;
5264 }
5265 if ($r_min != $r_max) {
5266 return 1 if ($r_min < $c->{r});
5267 return 1 if ($r_max > $c->{r});
5268 }
5269 }
5270 return 0 if (defined $limit && --$limit < 0);
5271 show_commit($c);
5272 return 1;
5273 }
5275 sub show_commit {
5276 my $c = shift;
5277 if ($oneline) {
5278 my $x = "\n";
5279 if (my $l = $c->{l}) {
5280 while ($l->[0] =~ /^\s*$/) { shift @$l }
5281 $x = $l->[0];
5282 }
5283 $l_fmt ||= 'A' . length($c->{r});
5284 print 'r',pack($l_fmt, $c->{r}),' | ';
5285 print "$c->{c} | " if $show_commit;
5286 print $x;
5287 } else {
5288 show_commit_normal($c);
5289 }
5290 }
5292 sub show_commit_changed_paths {
5293 my ($c) = @_;
5294 return unless $c->{changed};
5295 print "Changed paths:\n", @{$c->{changed}};
5296 }
5298 sub show_commit_normal {
5299 my ($c) = @_;
5300 print commit_log_separator, "r$c->{r} | ";
5301 print "$c->{c} | " if $show_commit;
5302 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5303 my $nr_line = 0;
5305 if (my $l = $c->{l}) {
5306 while ($l->[$#$l] eq "\n" && $#$l > 0
5307 && $l->[($#$l - 1)] eq "\n") {
5308 pop @$l;
5309 }
5310 $nr_line = scalar @$l;
5311 if (!$nr_line) {
5312 print "1 line\n\n\n";
5313 } else {
5314 if ($nr_line == 1) {
5315 $nr_line = '1 line';
5316 } else {
5317 $nr_line .= ' lines';
5318 }
5319 print $nr_line, "\n";
5320 show_commit_changed_paths($c);
5321 print "\n";
5322 print $_ foreach @$l;
5323 }
5324 } else {
5325 print "1 line\n";
5326 show_commit_changed_paths($c);
5327 print "\n";
5329 }
5330 foreach my $x (qw/raw stat diff/) {
5331 if ($c->{$x}) {
5332 print "\n";
5333 print $_ foreach @{$c->{$x}}
5334 }
5335 }
5336 }
5338 sub cmd_show_log {
5339 my (@args) = @_;
5340 my ($r_min, $r_max);
5341 my $r_last = -1; # prevent dupes
5342 set_local_timezone();
5343 if (defined $::_revision) {
5344 if ($::_revision =~ /^(\d+):(\d+)$/) {
5345 ($r_min, $r_max) = ($1, $2);
5346 } elsif ($::_revision =~ /^\d+$/) {
5347 $r_min = $r_max = $::_revision;
5348 } else {
5349 ::fatal "-r$::_revision is not supported, use ",
5350 "standard 'git log' arguments instead";
5351 }
5352 }
5354 config_pager();
5355 @args = git_svn_log_cmd($r_min, $r_max, @args);
5356 if (!@args) {
5357 print commit_log_separator unless $incremental || $oneline;
5358 return;
5359 }
5360 my $log = command_output_pipe(@args);
5361 run_pager();
5362 my (@k, $c, $d, $stat);
5363 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5364 while (<$log>) {
5365 if (/^${esc_color}commit -?($::sha1_short)/o) {
5366 my $cmt = $1;
5367 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5368 $r_last = $c->{r};
5369 process_commit($c, $r_min, $r_max, \@k) or
5370 goto out;
5371 }
5372 $d = undef;
5373 $c = { c => $cmt };
5374 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5375 get_author_info($c, $1, $2, $3);
5376 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5377 # ignore
5378 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5379 push @{$c->{raw}}, $_;
5380 } elsif (/^${esc_color}[ACRMDT]\t/) {
5381 # we could add $SVN->{svn_path} here, but that requires
5382 # remote access at the moment (repo_path_split)...
5383 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5384 push @{$c->{changed}}, $_;
5385 } elsif (/^${esc_color}diff /o) {
5386 $d = 1;
5387 push @{$c->{diff}}, $_;
5388 } elsif ($d) {
5389 push @{$c->{diff}}, $_;
5390 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5391 $esc_color*[\+\-]*$esc_color$/x) {
5392 $stat = 1;
5393 push @{$c->{stat}}, $_;
5394 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5395 push @{$c->{stat}}, $_;
5396 $stat = undef;
5397 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5398 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5399 } elsif (s/^${esc_color} //o) {
5400 push @{$c->{l}}, $_;
5401 }
5402 }
5403 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5404 $r_last = $c->{r};
5405 process_commit($c, $r_min, $r_max, \@k);
5406 }
5407 if (@k) {
5408 ($r_min, $r_max) = ($r_max, $r_min);
5409 process_commit($_, $r_min, $r_max) foreach reverse @k;
5410 }
5411 out:
5412 close $log;
5413 print commit_log_separator unless $incremental || $oneline;
5414 }
5416 sub cmd_blame {
5417 my $path = pop;
5419 config_pager();
5420 run_pager();
5422 my ($fh, $ctx, $rev);
5424 if ($_git_format) {
5425 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5426 while (my $line = <$fh>) {
5427 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5428 # Uncommitted edits show up as a rev ID of
5429 # all zeros, which we can't look up with
5430 # cmt_metadata
5431 if ($1 !~ /^0+$/) {
5432 (undef, $rev, undef) =
5433 ::cmt_metadata($1);
5434 $rev = '0' if (!$rev);
5435 } else {
5436 $rev = '0';
5437 }
5438 $rev = sprintf('%-10s', $rev);
5439 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5440 }
5441 print $line;
5442 }
5443 } else {
5444 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5445 '--', $path);
5446 my ($sha1);
5447 my %authors;
5448 my @buffer;
5449 my %dsha; #distinct sha keys
5451 while (my $line = <$fh>) {
5452 push @buffer, $line;
5453 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5454 $dsha{$1} = 1;
5455 }
5456 }
5458 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5460 foreach my $line (@buffer) {
5461 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5462 $rev = $s2r->{$1};
5463 $rev = '0' if (!$rev)
5464 }
5465 elsif ($line =~ /^author (.*)/) {
5466 $authors{$rev} = $1;
5467 $authors{$rev} =~ s/\s/_/g;
5468 }
5469 elsif ($line =~ /^\t(.*)$/) {
5470 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5471 }
5472 }
5473 }
5474 command_close_pipe($fh, $ctx);
5475 }
5477 package Git::SVN::Migration;
5478 # these version numbers do NOT correspond to actual version numbers
5479 # of git nor git-svn. They are just relative.
5480 #
5481 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5482 #
5483 # v1 layout: .git/$id/info/url, refs/remotes/$id
5484 #
5485 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5486 #
5487 # v3 layout: .git/svn/$id, refs/remotes/$id
5488 # - info/url may remain for backwards compatibility
5489 # - this is what we migrate up to this layout automatically,
5490 # - this will be used by git svn init on single branches
5491 # v3.1 layout (auto migrated):
5492 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5493 # for backwards compatibility
5494 #
5495 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5496 # - this is only created for newly multi-init-ed
5497 # repositories. Similar in spirit to the
5498 # --use-separate-remotes option in git-clone (now default)
5499 # - we do not automatically migrate to this (following
5500 # the example set by core git)
5501 #
5502 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
5503 # - newer, more-efficient format that uses 24-bytes per record
5504 # with no filler space.
5505 # - use xxd -c24 < .rev_map.$UUID to view and debug
5506 # - This is a one-way migration, repositories updated to the
5507 # new format will not be able to use old git-svn without
5508 # rebuilding the .rev_db. Rebuilding the rev_db is not
5509 # possible if noMetadata or useSvmProps are set; but should
5510 # be no problem for users that use the (sensible) defaults.
5511 use strict;
5512 use warnings;
5513 use Carp qw/croak/;
5514 use File::Path qw/mkpath/;
5515 use File::Basename qw/dirname basename/;
5516 use vars qw/$_minimize/;
5518 sub migrate_from_v0 {
5519 my $git_dir = $ENV{GIT_DIR};
5520 return undef unless -d $git_dir;
5521 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5522 my $migrated = 0;
5523 while (<$fh>) {
5524 chomp;
5525 my ($id, $orig_ref) = ($_, $_);
5526 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5527 next unless -f "$git_dir/$id/info/url";
5528 my $new_ref = "refs/remotes/$id";
5529 if (::verify_ref("$new_ref^0")) {
5530 print STDERR "W: $orig_ref is probably an old ",
5531 "branch used by an ancient version of ",
5532 "git-svn.\n",
5533 "However, $new_ref also exists.\n",
5534 "We will not be able ",
5535 "to use this branch until this ",
5536 "ambiguity is resolved.\n";
5537 next;
5538 }
5539 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5540 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5541 command_noisy('update-ref', $new_ref, $orig_ref);
5542 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5543 $migrated++;
5544 }
5545 command_close_pipe($fh, $ctx);
5546 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5547 $migrated;
5548 }
5550 sub migrate_from_v1 {
5551 my $git_dir = $ENV{GIT_DIR};
5552 my $migrated = 0;
5553 return $migrated unless -d $git_dir;
5554 my $svn_dir = "$git_dir/svn";
5556 # just in case somebody used 'svn' as their $id at some point...
5557 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5559 print STDERR "Migrating from a git-svn v1 layout...\n";
5560 mkpath([$svn_dir]);
5561 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5562 "$svn_dir\n\t(required for this version ",
5563 "($::VERSION) of git-svn) does not exist.\n";
5564 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5565 while (<$fh>) {
5566 my $x = $_;
5567 next unless $x =~ s#^refs/remotes/##;
5568 chomp $x;
5569 next unless -f "$git_dir/$x/info/url";
5570 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5571 next unless $u;
5572 my $dn = dirname("$git_dir/svn/$x");
5573 mkpath([$dn]) unless -d $dn;
5574 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5575 mkpath(["$git_dir/svn/svn"]);
5576 print STDERR " - $git_dir/$x/info => ",
5577 "$git_dir/svn/$x/info\n";
5578 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5579 croak "$!: $x";
5580 # don't worry too much about these, they probably
5581 # don't exist with repos this old (save for index,
5582 # and we can easily regenerate that)
5583 foreach my $f (qw/unhandled.log index .rev_db/) {
5584 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5585 }
5586 } else {
5587 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5588 rename "$git_dir/$x", "$git_dir/svn/$x" or
5589 croak "$!: $x";
5590 }
5591 $migrated++;
5592 }
5593 command_close_pipe($fh, $ctx);
5594 print STDERR "Done migrating from a git-svn v1 layout\n";
5595 $migrated;
5596 }
5598 sub read_old_urls {
5599 my ($l_map, $pfx, $path) = @_;
5600 my @dir;
5601 foreach (<$path/*>) {
5602 if (-r "$_/info/url") {
5603 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5604 my $ref_id = $pfx . basename $_;
5605 my $url = ::file_to_s("$_/info/url");
5606 $l_map->{$ref_id} = $url;
5607 } elsif (-d $_) {
5608 push @dir, $_;
5609 }
5610 }
5611 foreach (@dir) {
5612 my $x = $_;
5613 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5614 read_old_urls($l_map, $x, $_);
5615 }
5616 }
5618 sub migrate_from_v2 {
5619 my @cfg = command(qw/config -l/);
5620 return if grep /^svn-remote\..+\.url=/, @cfg;
5621 my %l_map;
5622 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5623 my $migrated = 0;
5625 foreach my $ref_id (sort keys %l_map) {
5626 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5627 if ($@) {
5628 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5629 }
5630 $migrated++;
5631 }
5632 $migrated;
5633 }
5635 sub minimize_connections {
5636 my $r = Git::SVN::read_all_remotes();
5637 my $new_urls = {};
5638 my $root_repos = {};
5639 foreach my $repo_id (keys %$r) {
5640 my $url = $r->{$repo_id}->{url} or next;
5641 my $fetch = $r->{$repo_id}->{fetch} or next;
5642 my $ra = Git::SVN::Ra->new($url);
5644 # skip existing cases where we already connect to the root
5645 if (($ra->{url} eq $ra->{repos_root}) ||
5646 ($ra->{repos_root} eq $repo_id)) {
5647 $root_repos->{$ra->{url}} = $repo_id;
5648 next;
5649 }
5651 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5652 my $root_path = $ra->{url};
5653 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5654 foreach my $path (keys %$fetch) {
5655 my $ref_id = $fetch->{$path};
5656 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5658 # make sure we can read when connecting to
5659 # a higher level of a repository
5660 my ($last_rev, undef) = $gs->last_rev_commit;
5661 if (!defined $last_rev) {
5662 $last_rev = eval {
5663 $root_ra->get_latest_revnum;
5664 };
5665 next if $@;
5666 }
5667 my $new = $root_path;
5668 $new .= length $path ? "/$path" : '';
5669 eval {
5670 $root_ra->get_log([$new], $last_rev, $last_rev,
5671 0, 0, 1, sub { });
5672 };
5673 next if $@;
5674 $new_urls->{$ra->{repos_root}}->{$new} =
5675 { ref_id => $ref_id,
5676 old_repo_id => $repo_id,
5677 old_path => $path };
5678 }
5679 }
5681 my @emptied;
5682 foreach my $url (keys %$new_urls) {
5683 # see if we can re-use an existing [svn-remote "repo_id"]
5684 # instead of creating a(n ugly) new section:
5685 my $repo_id = $root_repos->{$url} || $url;
5687 my $fetch = $new_urls->{$url};
5688 foreach my $path (keys %$fetch) {
5689 my $x = $fetch->{$path};
5690 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5691 my $pfx = "svn-remote.$x->{old_repo_id}";
5693 my $old_fetch = quotemeta("$x->{old_path}:".
5694 "$x->{ref_id}");
5695 command_noisy(qw/config --unset/,
5696 "$pfx.fetch", '^'. $old_fetch . '$');
5697 delete $r->{$x->{old_repo_id}}->
5698 {fetch}->{$x->{old_path}};
5699 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5700 command_noisy(qw/config --unset/,
5701 "$pfx.url");
5702 push @emptied, $x->{old_repo_id}
5703 }
5704 }
5705 }
5706 if (@emptied) {
5707 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5708 print STDERR <<EOF;
5709 The following [svn-remote] sections in your config file ($file) are empty
5710 and can be safely removed:
5711 EOF
5712 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5713 }
5714 }
5716 sub migration_check {
5717 migrate_from_v0();
5718 migrate_from_v1();
5719 migrate_from_v2();
5720 minimize_connections() if $_minimize;
5721 }
5723 package Git::IndexInfo;
5724 use strict;
5725 use warnings;
5726 use Git qw/command_input_pipe command_close_pipe/;
5728 sub new {
5729 my ($class) = @_;
5730 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5731 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5732 }
5734 sub remove {
5735 my ($self, $path) = @_;
5736 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5737 return ++$self->{nr};
5738 }
5739 undef;
5740 }
5742 sub update {
5743 my ($self, $mode, $hash, $path) = @_;
5744 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5745 return ++$self->{nr};
5746 }
5747 undef;
5748 }
5750 sub DESTROY {
5751 my ($self) = @_;
5752 command_close_pipe($self->{gui}, $self->{ctx});
5753 }
5755 package Git::SVN::GlobSpec;
5756 use strict;
5757 use warnings;
5759 sub new {
5760 my ($class, $glob) = @_;
5761 my $re = $glob;
5762 $re =~ s!/+$!!g; # no need for trailing slashes
5763 $re =~ m!^([^*]*)(\*(?:/\*)*)(.*)$!;
5764 my $temp = $re;
5765 my ($left, $right) = ($1, $3);
5766 $re = $2;
5767 my $depth = $re =~ tr/*/*/;
5768 if ($depth != $temp =~ tr/*/*/) {
5769 die "Only one set of wildcard directories " .
5770 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5771 }
5772 if ($depth == 0) {
5773 die "One '*' is needed for glob: '$glob'\n";
5774 }
5775 $re =~ s!\*!\[^/\]*!g;
5776 $re = quotemeta($left) . "($re)" . quotemeta($right);
5777 if (length $left && !($left =~ s!/+$!!g)) {
5778 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5779 }
5780 if (length $right && !($right =~ s!^/+!!g)) {
5781 die "Missing leading '/' on right side of: '$glob' ($right)\n";
5782 }
5783 my $left_re = qr/^\/\Q$left\E(\/|$)/;
5784 bless { left => $left, right => $right, left_regex => $left_re,
5785 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5786 }
5788 sub full_path {
5789 my ($self, $path) = @_;
5790 return (length $self->{left} ? "$self->{left}/" : '') .
5791 $path . (length $self->{right} ? "/$self->{right}" : '');
5792 }
5794 __END__
5796 Data structures:
5799 $remotes = { # returned by read_all_remotes()
5800 'svn' => {
5801 # svn-remote.svn.url=https://svn.musicpd.org
5802 url => 'https://svn.musicpd.org',
5803 # svn-remote.svn.fetch=mpd/trunk:trunk
5804 fetch => {
5805 'mpd/trunk' => 'trunk',
5806 },
5807 # svn-remote.svn.tags=mpd/tags/*:tags/*
5808 tags => {
5809 path => {
5810 left => 'mpd/tags',
5811 right => '',
5812 regex => qr!mpd/tags/([^/]+)$!,
5813 glob => 'tags/*',
5814 },
5815 ref => {
5816 left => 'tags',
5817 right => '',
5818 regex => qr!tags/([^/]+)$!,
5819 glob => 'tags/*',
5820 },
5821 }
5822 }
5823 };
5825 $log_entry hashref as returned by libsvn_log_entry()
5826 {
5827 log => 'whitespace-formatted log entry
5828 ', # trailing newline is preserved
5829 revision => '8', # integer
5830 date => '2004-02-24T17:01:44.108345Z', # commit date
5831 author => 'committer name'
5832 };
5835 # this is generated by generate_diff();
5836 @mods = array of diff-index line hashes, each element represents one line
5837 of diff-index output
5839 diff-index line ($m hash)
5840 {
5841 mode_a => first column of diff-index output, no leading ':',
5842 mode_b => second column of diff-index output,
5843 sha1_b => sha1sum of the final blob,
5844 chg => change type [MCRADT],
5845 file_a => original file name of a file (iff chg is 'C' or 'R')
5846 file_b => new/current file name of a file (any chg)
5847 }
5848 ;
5850 # retval of read_url_paths{,_all}();
5851 $l_map = {
5852 # repository root url
5853 'https://svn.musicpd.org' => {
5854 # repository path # GIT_SVN_ID
5855 'mpd/trunk' => 'trunk',
5856 'mpd/tags/0.11.5' => 'tags/0.11.5',
5857 },
5858 }
5860 Notes:
5861 I don't trust the each() function on unless I created %hash myself
5862 because the internal iterator may not have started at base.