?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
clone: Supply the right commit hash to post-checkout when -b is used
[git/jnareb-git/bp-gitweb.git] / git-svn.perl
blobeb4b75aff347670b1817fb1c2584cff11a08fad5
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@@';
11
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 } || '';
16
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';
23
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 }
32
33 $Git::SVN::Log::TZ = $ENV{TZ};
34 $ENV{TZ} = 'UTC';
35 $| = 1; # unbuffer STDOUT
36
37 sub fatal (@) { print STDERR "@_\n"; exit 1 }
38 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
39 require SVN::Ra;
40 require SVN::Delta;
41 if ($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;
58
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 }
72
73 my ($SVN);
74
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 $_q ||= 0;
86 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
87 'config-dir=s' => \$Git::SVN::Ra::config_dir,
88 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
89 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
90 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
91 'authors-file|A=s' => \$_authors,
92 'authors-prog=s' => \$_authors_prog,
93 'repack:i' => \$Git::SVN::_repack,
94 'noMetadata' => \$Git::SVN::_no_metadata,
95 'useSvmProps' => \$Git::SVN::_use_svm_props,
96 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
97 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
98 'no-checkout' => \$_no_checkout,
99 'quiet|q+' => \$_q,
100 'repack-flags|repack-args|repack-opts=s' =>
101 \$Git::SVN::_repack_flags,
102 'use-log-author' => \$Git::SVN::_use_log_author,
103 'add-author-from' => \$Git::SVN::_add_author_from,
104 'localtime' => \$Git::SVN::_localtime,
105 %remote_opts );
106
107 my ($_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 );
125
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 );
237
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 };
249
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 => $ENV{GIT_DIR});
273 }
274
275 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
276
277 read_repo_config(\%opts);
278 if ($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');
288
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 }
296
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;
308
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
316
317 print $fd "Available commands:\n" unless $cmd;
318
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.
338
339 exit $exit;
340 }
341
342 sub version {
343 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
344 exit 0;
345 }
346
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 }
374
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 => $ENV{GIT_DIR});
381 }
382
383 sub cmd_clone {
384 my ($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 }
396
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();
411
412 if ($Git::SVN::_minimize_url eq 'unset') {
413 $Git::SVN::_minimize_url = 0;
414 }
415
416 Git::SVN->init($url);
417 }
418
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 }
444
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 }
480
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';
487
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', $head], STDERR => 0);
499 }
500
501 my @refs;
502 my ($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 }
507
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 }
517
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;
566
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 $linear_refs = \@l;
630 }
631 }
632 }
633
634 if ($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 }
648
649 unlink $gs->{index};
650 }
651
652 sub cmd_branch {
653 my ($branch_name, $head) = @_;
654
655 unless (defined $branch_name && length $branch_name) {
656 die(($_tag ? "tag" : "branch") . " name required\n");
657 }
658 $head ||= 'HEAD';
659
660 my ($src, $rev, undef, $gs) = working_head_info($head);
661
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 || ());
705
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 );
715
716 eval {
717 $ctx->ls($dst, 'HEAD', 0);
718 } and die "branch ${branch_name} already exists\n";
719
720 print "Copying ${src} at r${rev} to ${dst}...\n";
721 $ctx->copy($src, $rev, $dst)
722 unless $_dry_run;
723
724 $gs->fetch_all;
725 }
726
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 $result;
747 }
748
749 sub cmd_rebase {
750 command_noisy(qw/update-index --refresh/);
751 my ($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(), $gs->refname);
772 }
773
774 sub cmd_show_ignore {
775 my ($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 "$s\n";
787 });
788 }
789
790 sub cmd_show_externals {
791 my ($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 "$s\n";
802 });
803 }
804
805 sub cmd_create_ignore {
806 my ($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', $ignore);
830 });
831 }
832
833 sub canonicalize_path {
834 my ($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 $path;
850 }
851
852 sub canonicalize_url {
853 my ($url) = @_;
854 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
855 return $url;
856 }
857
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;
865
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;
872
873 # canonicalize the path (otherwise libsvn will abort or fail to
874 # find the file)
875 $path = canonicalize_path($path);
876
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 $props;
886 }
887
888 # cmd_propget (PROP, PATH)
889 # ------------------------
890 # Print the SVN property PROP for PATH.
891 sub cmd_propget {
892 my ($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 }
901
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 }
914
915 sub cmd_multi_init {
916 my $url = shift;
917 unless (defined $_trunk || @_branches || @_tags) {
918 usage(1);
919 }
920
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 }
947
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, $remotes);
953 }
954 }
955 }
956
957 # this command is special because it requires no metadata
958 sub cmd_commit_diff {
959 my ($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 == $tb\n";
1002 }
1003 }
1004
1005 sub escape_uri_only {
1006 my ($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, $_;
1011 }
1012 join('/', @tmp);
1013 }
1014
1015 sub escape_url {
1016 my ($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 }
1023
1024 sub cmd_info {
1025 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1026 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1027 if (exists $_[1]) {
1028 die "Too many arguments specified\n";
1029 }
1030
1031 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1032
1033 if (!$file_type && !$diff_status) {
1034 print STDERR "svn: '$path' is not under version control\n";
1035 exit 1;
1036 }
1037
1038 my ($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 }
1043
1044 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1045 $path = "." if $path eq "";
1046
1047 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1048
1049 if ($_url) {
1050 print escape_url($full_url), "\n";
1051 return;
1052 }
1053
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";
1057
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";
1069
1070 $result .= "Node Kind: " .
1071 ($file_type eq "dir" ? "directory" : "file") . "\n";
1072
1073 my $schedule = $diff_status eq "A"
1074 ? "add"
1075 : ($diff_status eq "D" ? "delete" : "normal");
1076 $result .= "Schedule: $schedule\n";
1077
1078 if ($diff_status eq "A") {
1079 print $result, "\n";
1080 return;
1081 }
1082
1083 my ($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;
1096
1097 Git::SVN::Log::set_local_timezone();
1098
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";
1103
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 }
1134
1135 print $result, "\n";
1136 }
1137
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 }
1151
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 }
1159
1160 ########################### utility functions #########################
1161
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 }
1169
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');
1174
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 }
1185
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'));
1189
1190 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1191 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1192 return if -f $index;
1193
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", $gs->last_rev, "\n";
1199 }
1200
1201 sub complete_svn_url {
1202 my ($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 }
1213
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 }
1255
1256 sub verify_ref {
1257 my ($ref) = @_;
1258 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1259 { STDERR => 0 }); };
1260 }
1261
1262 sub get_tree_from_treeish {
1263 my ($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 $expected;
1281 }
1282
1283 sub get_commit_entry {
1284 my ($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 $!;
1289
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 $!;
1322
1323 if ($_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>);
1336
1337 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1338 my $msg = $log_entry{log};
1339
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 }
1345
1346 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1347 die "Could not encode as UTF-8:\n$msg\n" if $@;
1348
1349 $log_entry{log} = $msg;
1350
1351 close $log_fh or croak $!;
1352 }
1353 unlink $commit_msg;
1354 \%log_entry;
1355 }
1356
1357 sub s_to_file {
1358 my ($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 }
1364
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 }
1374
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 }
1391
1392 # convert GetOpt::Long specs for use by git-config
1393 sub read_repo_config {
1394 return unless -d $ENV{GIT_DIR};
1395 my $opts = shift;
1396 my @config_only;
1397 foreach my $o (keys %$opts) {
1398 # if we have mixedCase and a long option-only, then
1399 # it's a config-only variable that we don't need for
1400 # the command-line.
1401 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1402 my $v = $opts->{$o};
1403 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1404 $key =~ s/-//g;
1405 my $arg = 'git config';
1406 $arg .= ' --int' if ($o =~ /[:=]i$/);
1407 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1408 if (ref $v eq 'ARRAY') {
1409 chomp(my @tmp = `$arg --get-all svn.$key`);
1410 @$v = @tmp if @tmp;
1411 } else {
1412 chomp(my $tmp = `$arg --get svn.$key`);
1413 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1414 $$v = $tmp;
1415 }
1416 }
1417 }
1418 delete @$opts{@config_only} if @config_only;
1419 }
1420
1421 sub extract_metadata {
1422 my $id = shift or return (undef, undef, undef);
1423 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1424 \s([a-f\d\-]+)$/ix);
1425 if (!defined $rev || !$uuid || !$url) {
1426 # some of the original repositories I made had
1427 # identifiers like this:
1428 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1429 }
1430 return ($url, $rev, $uuid);
1431 }
1432
1433 sub cmt_metadata {
1434 return extract_metadata((grep(/^git-svn-id: /,
1435 command(qw/cat-file commit/, shift)))[-1]);
1436 }
1437
1438 sub cmt_sha2rev_batch {
1439 my %s2r;
1440 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1441 my $list = shift;
1442
1443 foreach my $sha (@{$list}) {
1444 my $first = 1;
1445 my $size = 0;
1446 print $out $sha, "\n";
1447
1448 while (my $line = <$in>) {
1449 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1450 last;
1451 } elsif ($first &&
1452 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1453 $first = 0;
1454 $size = $1;
1455 next;
1456 } elsif ($line =~ /^(git-svn-id: )/) {
1457 my (undef, $rev, undef) =
1458 extract_metadata($line);
1459 $s2r{$sha} = $rev;
1460 }
1461
1462 $size -= length($line);
1463 last if ($size == 0);
1464 }
1465 }
1466
1467 command_close_bidi_pipe($pid, $in, $out, $ctx);
1468
1469 return \%s2r;
1470 }
1471
1472 sub working_head_info {
1473 my ($head, $refs) = @_;
1474 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1475 my ($fh, $ctx) = command_output_pipe(@args, $head);
1476 my $hash;
1477 my %max;
1478 while (<$fh>) {
1479 if ( m{^commit ($::sha1)$} ) {
1480 unshift @$refs, $hash if $hash and $refs;
1481 $hash = $1;
1482 next;
1483 }
1484 next unless s{^\s*(git-svn-id:)}{$1};
1485 my ($url, $rev, $uuid) = extract_metadata($_);
1486 if (defined $url && defined $rev) {
1487 next if $max{$url} and $max{$url} < $rev;
1488 if (my $gs = Git::SVN->find_by_url($url)) {
1489 my $c = $gs->rev_map_get($rev, $uuid);
1490 if ($c && $c eq $hash) {
1491 close $fh; # break the pipe
1492 return ($url, $rev, $uuid, $gs);
1493 } else {
1494 $max{$url} ||= $gs->rev_map_max;
1495 }
1496 }
1497 }
1498 }
1499 command_close_pipe($fh, $ctx);
1500 (undef, undef, undef, undef);
1501 }
1502
1503 sub read_commit_parents {
1504 my ($parents, $c) = @_;
1505 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1506 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1507 @{$parents->{$c}} = split(/ /, $p);
1508 }
1509
1510 sub linearize_history {
1511 my ($gs, $refs) = @_;
1512 my %parents;
1513 foreach my $c (@$refs) {
1514 read_commit_parents(\%parents, $c);
1515 }
1516
1517 my @linear_refs;
1518 my %skip = ();
1519 my $last_svn_commit = $gs->last_commit;
1520 foreach my $c (reverse @$refs) {
1521 next if $c eq $last_svn_commit;
1522 last if $skip{$c};
1523
1524 unshift @linear_refs, $c;
1525 $skip{$c} = 1;
1526
1527 # we only want the first parent to diff against for linear
1528 # history, we save the rest to inject when we finalize the
1529 # svn commit
1530 my $fp_a = verify_ref("$c~1");
1531 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1532 if (!$fp_a || !$fp_b) {
1533 die "Commit $c\n",
1534 "has no parent commit, and therefore ",
1535 "nothing to diff against.\n",
1536 "You should be working from a repository ",
1537 "originally created by git-svn\n";
1538 }
1539 if ($fp_a ne $fp_b) {
1540 die "$c~1 = $fp_a, however parsing commit $c ",
1541 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1542 }
1543
1544 foreach my $p (@{$parents{$c}}) {
1545 $skip{$p} = 1;
1546 }
1547 }
1548 (\@linear_refs, \%parents);
1549 }
1550
1551 sub find_file_type_and_diff_status {
1552 my ($path) = @_;
1553 return ('dir', '') if $path eq '';
1554
1555 my $diff_output =
1556 command_oneline(qw(diff --cached --name-status --), $path) || "";
1557 my $diff_status = (split(' ', $diff_output))[0] || "";
1558
1559 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1560
1561 return (undef, undef) if !$diff_status && !$ls_tree;
1562
1563 if ($diff_status eq "A") {
1564 return ("link", $diff_status) if -l $path;
1565 return ("dir", $diff_status) if -d $path;
1566 return ("file", $diff_status);
1567 }
1568
1569 my $mode = (split(' ', $ls_tree))[0] || "";
1570
1571 return ("link", $diff_status) if $mode eq "120000";
1572 return ("dir", $diff_status) if $mode eq "040000";
1573 return ("file", $diff_status);
1574 }
1575
1576 sub md5sum {
1577 my $arg = shift;
1578 my $ref = ref $arg;
1579 my $md5 = Digest::MD5->new();
1580 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1581 $md5->addfile($arg) or croak $!;
1582 } elsif ($ref eq 'SCALAR') {
1583 $md5->add($$arg) or croak $!;
1584 } elsif (!$ref) {
1585 $md5->add($arg) or croak $!;
1586 } else {
1587 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1588 }
1589 return $md5->hexdigest();
1590 }
1591
1592 sub gc_directory {
1593 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1594 my $out_filename = $_ . ".gz";
1595 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1596 binmode $in_fh;
1597 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1598 die "Unable to open $out_filename: $!\n";
1599
1600 my $res;
1601 while ($res = sysread($in_fh, my $str, 1024)) {
1602 $gz->gzwrite($str) or
1603 die "Unable to write: ".$gz->gzerror()."!\n";
1604 }
1605 unlink $_ or die "unlink $File::Find::name: $!\n";
1606 } elsif (-f $_ && basename($_) eq "index") {
1607 unlink $_ or die "unlink $_: $!\n";
1608 }
1609 }
1610
1611 package Git::SVN;
1612 use strict;
1613 use warnings;
1614 use Fcntl qw/:DEFAULT :seek/;
1615 use constant rev_map_fmt => 'NH40';
1616 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1617 $_repack $_repack_flags $_use_svm_props $_head
1618 $_use_svnsync_props $no_reuse_existing $_minimize_url
1619 $_use_log_author $_add_author_from $_localtime/;
1620 use Carp qw/croak/;
1621 use File::Path qw/mkpath/;
1622 use File::Copy qw/copy/;
1623 use IPC::Open3;
1624
1625 my ($_gc_nr, $_gc_period);
1626
1627 # properties that we do not log:
1628 my %SKIP_PROP;
1629 BEGIN {
1630 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1631 svn:special svn:executable
1632 svn:entry:committed-rev
1633 svn:entry:last-author
1634 svn:entry:uuid
1635 svn:entry:committed-date/;
1636
1637 # some options are read globally, but can be overridden locally
1638 # per [svn-remote "..."] section. Command-line options will *NOT*
1639 # override options set in an [svn-remote "..."] section
1640 no strict 'refs';
1641 for my $option (qw/follow_parent no_metadata use_svm_props
1642 use_svnsync_props/) {
1643 my $key = $option;
1644 $key =~ tr/_//d;
1645 my $prop = "-$option";
1646 *$option = sub {
1647 my ($self) = @_;
1648 return $self->{$prop} if exists $self->{$prop};
1649 my $k = "svn-remote.$self->{repo_id}.$key";
1650 eval { command_oneline(qw/config --get/, $k) };
1651 if ($@) {
1652 $self->{$prop} = ${"Git::SVN::_$option"};
1653 } else {
1654 my $v = command_oneline(qw/config --bool/,$k);
1655 $self->{$prop} = $v eq 'false' ? 0 : 1;
1656 }
1657 return $self->{$prop};
1658 }
1659 }
1660 }
1661
1662
1663 my (%LOCKFILES, %INDEX_FILES);
1664 END {
1665 unlink keys %LOCKFILES if %LOCKFILES;
1666 unlink keys %INDEX_FILES if %INDEX_FILES;
1667 }
1668
1669 sub resolve_local_globs {
1670 my ($url, $fetch, $glob_spec) = @_;
1671 return unless defined $glob_spec;
1672 my $ref = $glob_spec->{ref};
1673 my $path = $glob_spec->{path};
1674 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1675 next unless m#^$ref->{regex}$#;
1676 my $p = $1;
1677 my $pathname = desanitize_refname($path->full_path($p));
1678 my $refname = desanitize_refname($ref->full_path($p));
1679 if (my $existing = $fetch->{$pathname}) {
1680 if ($existing ne $refname) {
1681 die "Refspec conflict:\n",
1682 "existing: $existing\n",
1683 " globbed: $refname\n";
1684 }
1685 my $u = (::cmt_metadata("$refname"))[0];
1686 $u =~ s!^\Q$url\E(/|$)!! or die
1687 "$refname: '$url' not found in '$u'\n";
1688 if ($pathname ne $u) {
1689 warn "W: Refspec glob conflict ",
1690 "(ref: $refname):\n",
1691 "expected path: $pathname\n",
1692 " real path: $u\n",
1693 "Continuing ahead with $u\n";
1694 next;
1695 }
1696 } else {
1697 $fetch->{$pathname} = $refname;
1698 }
1699 }
1700 }
1701
1702 sub parse_revision_argument {
1703 my ($base, $head) = @_;
1704 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1705 return ($base, $head);
1706 }
1707 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1708 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1709 return ($head, $head) if ($::_revision eq 'HEAD');
1710 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1711 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1712 die "revision argument: $::_revision not understood by git-svn\n";
1713 }
1714
1715 sub fetch_all {
1716 my ($repo_id, $remotes) = @_;
1717 if (ref $repo_id) {
1718 my $gs = $repo_id;
1719 $repo_id = undef;
1720 $repo_id = $gs->{repo_id};
1721 }
1722 $remotes ||= read_all_remotes();
1723 my $remote = $remotes->{$repo_id} or
1724 die "[svn-remote \"$repo_id\"] unknown\n";
1725 my $fetch = $remote->{fetch};
1726 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1727 my (@gs, @globs);
1728 my $ra = Git::SVN::Ra->new($url);
1729 my $uuid = $ra->get_uuid;
1730 my $head = $ra->get_latest_revnum;
1731 $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1732 my $base = defined $fetch ? $head : 0;
1733
1734 # read the max revs for wildcard expansion (branches/*, tags/*)
1735 foreach my $t (qw/branches tags/) {
1736 defined $remote->{$t} or next;
1737 push @globs, @{$remote->{$t}};
1738
1739 my $max_rev = eval { tmp_config(qw/--int --get/,
1740 "svn-remote.$repo_id.${t}-maxRev") };
1741 if (defined $max_rev && ($max_rev < $base)) {
1742 $base = $max_rev;
1743 } elsif (!defined $max_rev) {
1744 $base = 0;
1745 }
1746 }
1747
1748 if ($fetch) {
1749 foreach my $p (sort keys %$fetch) {
1750 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1751 my $lr = $gs->rev_map_max;
1752 if (defined $lr) {
1753 $base = $lr if ($lr < $base);
1754 }
1755 push @gs, $gs;
1756 }
1757 }
1758
1759 ($base, $head) = parse_revision_argument($base, $head);
1760 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1761 }
1762
1763 sub read_all_remotes {
1764 my $r = {};
1765 my $use_svm_props = eval { command_oneline(qw/config --bool
1766 svn.useSvmProps/) };
1767 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1768 my $svn_refspec = qr{\s*/?(.*?)\s*:\s*(.+?)\s*};
1769 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1770 if (m!^(.+)\.fetch=$svn_refspec$!) {
1771 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1772 die("svn-remote.$remote: remote ref '$remote_ref' "
1773 . "must start with 'refs/'\n")
1774 unless $remote_ref =~ m{^refs/};
1775 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1776 $r->{$remote}->{svm} = {} if $use_svm_props;
1777 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1778 $r->{$1}->{svm} = {};
1779 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1780 $r->{$1}->{url} = $2;
1781 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1782 my ($remote, $t, $local_ref, $remote_ref) =
1783 ($1, $2, $3, $4);
1784 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1785 . "must start with 'refs/'\n")
1786 unless $remote_ref =~ m{^refs/};
1787 my $rs = {
1788 t => $t,
1789 remote => $remote,
1790 path => Git::SVN::GlobSpec->new($local_ref),
1791 ref => Git::SVN::GlobSpec->new($remote_ref) };
1792 if (length($rs->{ref}->{right}) != 0) {
1793 die "The '*' glob character must be the last ",
1794 "character of '$remote_ref'\n";
1795 }
1796 push @{ $r->{$remote}->{$t} }, $rs;
1797 }
1798 }
1799
1800 map {
1801 if (defined $r->{$_}->{svm}) {
1802 my $svm;
1803 eval {
1804 my $section = "svn-remote.$_";
1805 $svm = {
1806 source => tmp_config('--get',
1807 "$section.svm-source"),
1808 replace => tmp_config('--get',
1809 "$section.svm-replace"),
1810 }
1811 };
1812 $r->{$_}->{svm} = $svm;
1813 }
1814 } keys %$r;
1815
1816 $r;
1817 }
1818
1819 sub init_vars {
1820 $_gc_nr = $_gc_period = 1000;
1821 if (defined $_repack || defined $_repack_flags) {
1822 warn "Repack options are obsolete; they have no effect.\n";
1823 }
1824 }
1825
1826 sub verify_remotes_sanity {
1827 return unless -d $ENV{GIT_DIR};
1828 my %seen;
1829 foreach (command(qw/config -l/)) {
1830 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1831 if ($seen{$1}) {
1832 die "Remote ref refs/remote/$1 is tracked by",
1833 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1834 "Please resolve this ambiguity in ",
1835 "your git configuration file before ",
1836 "continuing\n";
1837 }
1838 $seen{$1} = $_;
1839 }
1840 }
1841 }
1842
1843 sub find_existing_remote {
1844 my ($url, $remotes) = @_;
1845 return undef if $no_reuse_existing;
1846 my $existing;
1847 foreach my $repo_id (keys %$remotes) {
1848 my $u = $remotes->{$repo_id}->{url} or next;
1849 next if $u ne $url;
1850 $existing = $repo_id;
1851 last;
1852 }
1853 $existing;
1854 }
1855
1856 sub init_remote_config {
1857 my ($self, $url, $no_write) = @_;
1858 $url =~ s!/+$!!; # strip trailing slash
1859 my $r = read_all_remotes();
1860 my $existing = find_existing_remote($url, $r);
1861 if ($existing) {
1862 unless ($no_write) {
1863 print STDERR "Using existing ",
1864 "[svn-remote \"$existing\"]\n";
1865 }
1866 $self->{repo_id} = $existing;
1867 } elsif ($_minimize_url) {
1868 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1869 $existing = find_existing_remote($min_url, $r);
1870 if ($existing) {
1871 unless ($no_write) {
1872 print STDERR "Using existing ",
1873 "[svn-remote \"$existing\"]\n";
1874 }
1875 $self->{repo_id} = $existing;
1876 }
1877 if ($min_url ne $url) {
1878 unless ($no_write) {
1879 print STDERR "Using higher level of URL: ",
1880 "$url => $min_url\n";
1881 }
1882 my $old_path = $self->{path};
1883 $self->{path} = $url;
1884 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1885 if (length $old_path) {
1886 $self->{path} .= "/$old_path";
1887 }
1888 $url = $min_url;
1889 }
1890 }
1891 my $orig_url;
1892 if (!$existing) {
1893 # verify that we aren't overwriting anything:
1894 $orig_url = eval {
1895 command_oneline('config', '--get',
1896 "svn-remote.$self->{repo_id}.url")
1897 };
1898 if ($orig_url && ($orig_url ne $url)) {
1899 die "svn-remote.$self->{repo_id}.url already set: ",
1900 "$orig_url\nwanted to set to: $url\n";
1901 }
1902 }
1903 my ($xrepo_id, $xpath) = find_ref($self->refname);
1904 if (!$no_write && defined $xpath) {
1905 die "svn-remote.$xrepo_id.fetch already set to track ",
1906 "$xpath:", $self->refname, "\n";
1907 }
1908 unless ($no_write) {
1909 command_noisy('config',
1910 "svn-remote.$self->{repo_id}.url", $url);
1911 $self->{path} =~ s{^/}{};
1912 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1913 command_noisy('config', '--add',
1914 "svn-remote.$self->{repo_id}.fetch",
1915 "$self->{path}:".$self->refname);
1916 }
1917 $self->{url} = $url;
1918 }
1919
1920 sub find_by_url { # repos_root and, path are optional
1921 my ($class, $full_url, $repos_root, $path) = @_;
1922
1923 return undef unless defined $full_url;
1924 remove_username($full_url);
1925 remove_username($repos_root) if defined $repos_root;
1926 my $remotes = read_all_remotes();
1927 if (defined $full_url && defined $repos_root && !defined $path) {
1928 $path = $full_url;
1929 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1930 }
1931 foreach my $repo_id (keys %$remotes) {
1932 my $u = $remotes->{$repo_id}->{url} or next;
1933 remove_username($u);
1934 next if defined $repos_root && $repos_root ne $u;
1935
1936 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1937 foreach my $t (qw/branches tags/) {
1938 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1939 resolve_local_globs($u, $fetch, $globspec);
1940 }
1941 }
1942 my $p = $path;
1943 my $rwr = rewrite_root({repo_id => $repo_id});
1944 my $svm = $remotes->{$repo_id}->{svm}
1945 if defined $remotes->{$repo_id}->{svm};
1946 unless (defined $p) {
1947 $p = $full_url;
1948 my $z = $u;
1949 my $prefix = '';
1950 if ($rwr) {
1951 $z = $rwr;
1952 remove_username($z);
1953 } elsif (defined $svm) {
1954 $z = $svm->{source};
1955 $prefix = $svm->{replace};
1956 $prefix =~ s#^\Q$u\E(?:/|$)##;
1957 $prefix =~ s#/$##;
1958 }
1959 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1960 }
1961 foreach my $f (keys %$fetch) {
1962 next if $f ne $p;
1963 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1964 }
1965 }
1966 undef;
1967 }
1968
1969 sub init {
1970 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1971 my $self = _new($class, $repo_id, $ref_id, $path);
1972 if (defined $url) {
1973 $self->init_remote_config($url, $no_write);
1974 }
1975 $self;
1976 }
1977
1978 sub find_ref {
1979 my ($ref_id) = @_;
1980 foreach (command(qw/config -l/)) {
1981 next unless m!^svn-remote\.(.+)\.fetch=
1982 \s*/?(.*?)\s*:\s*(.+?)\s*$!x;
1983 my ($repo_id, $path, $ref) = ($1, $2, $3);
1984 if ($ref eq $ref_id) {
1985 $path = '' if ($path =~ m#^\./?#);
1986 return ($repo_id, $path);
1987 }
1988 }
1989 (undef, undef, undef);
1990 }
1991
1992 sub new {
1993 my ($class, $ref_id, $repo_id, $path) = @_;
1994 if (defined $ref_id && !defined $repo_id && !defined $path) {
1995 ($repo_id, $path) = find_ref($ref_id);
1996 if (!defined $repo_id) {
1997 die "Could not find a \"svn-remote.*.fetch\" key ",
1998 "in the repository configuration matching: ",
1999 "$ref_id\n";
2000 }
2001 }
2002 my $self = _new($class, $repo_id, $ref_id, $path);
2003 if (!defined $self->{path} || !length $self->{path}) {
2004 my $fetch = command_oneline('config', '--get',
2005 "svn-remote.$repo_id.fetch",
2006 ":$ref_id\$") or
2007 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2008 "\":$ref_id\$\" in config\n";
2009 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2010 }
2011 $self->{url} = command_oneline('config', '--get',
2012 "svn-remote.$repo_id.url") or
2013 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2014 $self->rebuild;
2015 $self;
2016 }
2017
2018 sub refname {
2019 my ($refname) = $_[0]->{ref_id} ;
2020
2021 # It cannot end with a slash /, we'll throw up on this because
2022 # SVN can't have directories with a slash in their name, either:
2023 if ($refname =~ m{/$}) {
2024 die "ref: '$refname' ends with a trailing slash, this is ",
2025 "not permitted by git nor Subversion\n";
2026 }
2027
2028 # It cannot have ASCII control character space, tilde ~, caret ^,
2029 # colon :, question-mark ?, asterisk *, space, or open bracket [
2030 # anywhere.
2031 #
2032 # Additionally, % must be escaped because it is used for escaping
2033 # and we want our escaped refname to be reversible
2034 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2035
2036 # no slash-separated component can begin with a dot .
2037 # /.* becomes /%2E*
2038 $refname =~ s{/\.}{/%2E}g;
2039
2040 # It cannot have two consecutive dots .. anywhere
2041 # .. becomes %2E%2E
2042 $refname =~ s{\.\.}{%2E%2E}g;
2043
2044 return $refname;
2045 }
2046
2047 sub desanitize_refname {
2048 my ($refname) = @_;
2049 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2050 return $refname;
2051 }
2052
2053 sub svm_uuid {
2054 my ($self) = @_;
2055 return $self->{svm}->{uuid} if $self->svm;
2056 $self->ra;
2057 unless ($self->{svm}) {
2058 die "SVM UUID not cached, and reading remotely failed\n";
2059 }
2060 $self->{svm}->{uuid};
2061 }
2062
2063 sub svm {
2064 my ($self) = @_;
2065 return $self->{svm} if $self->{svm};
2066 my $svm;
2067 # see if we have it in our config, first:
2068 eval {
2069 my $section = "svn-remote.$self->{repo_id}";
2070 $svm = {
2071 source => tmp_config('--get', "$section.svm-source"),
2072 uuid => tmp_config('--get', "$section.svm-uuid"),
2073 replace => tmp_config('--get', "$section.svm-replace"),
2074 }
2075 };
2076 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2077 $self->{svm} = $svm;
2078 }
2079 $self->{svm};
2080 }
2081
2082 sub _set_svm_vars {
2083 my ($self, $ra) = @_;
2084 return $ra if $self->svm;
2085
2086 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2087 "(svm:source, svm:uuid) ",
2088 "from the following URLs:\n" );
2089 sub read_svm_props {
2090 my ($self, $ra, $path, $r) = @_;
2091 my $props = ($ra->get_dir($path, $r))[2];
2092 my $src = $props->{'svm:source'};
2093 my $uuid = $props->{'svm:uuid'};
2094 return undef if (!$src || !$uuid);
2095
2096 chomp($src, $uuid);
2097
2098 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2099 or die "doesn't look right - svm:uuid is '$uuid'\n";
2100
2101 # the '!' is used to mark the repos_root!/relative/path
2102 $src =~ s{/?!/?}{/};
2103 $src =~ s{/+$}{}; # no trailing slashes please
2104 # username is of no interest
2105 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2106
2107 my $replace = $ra->{url};
2108 $replace .= "/$path" if length $path;
2109
2110 my $section = "svn-remote.$self->{repo_id}";
2111 tmp_config("$section.svm-source", $src);
2112 tmp_config("$section.svm-replace", $replace);
2113 tmp_config("$section.svm-uuid", $uuid);
2114 $self->{svm} = {
2115 source => $src,
2116 uuid => $uuid,
2117 replace => $replace
2118 };
2119 }
2120
2121 my $r = $ra->get_latest_revnum;
2122 my $path = $self->{path};
2123 my %tried;
2124 while (length $path) {
2125 unless ($tried{"$self->{url}/$path"}) {
2126 return $ra if $self->read_svm_props($ra, $path, $r);
2127 $tried{"$self->{url}/$path"} = 1;
2128 }
2129 $path =~ s#/?[^/]+$##;
2130 }
2131 die "Path: '$path' should be ''\n" if $path ne '';
2132 return $ra if $self->read_svm_props($ra, $path, $r);
2133 $tried{"$self->{url}/$path"} = 1;
2134
2135 if ($ra->{repos_root} eq $self->{url}) {
2136 die @err, (map { " $_\n" } keys %tried), "\n";
2137 }
2138
2139 # nope, make sure we're connected to the repository root:
2140 my $ok;
2141 my @tried_b;
2142 $path = $ra->{svn_path};
2143 $ra = Git::SVN::Ra->new($ra->{repos_root});
2144 while (length $path) {
2145 unless ($tried{"$ra->{url}/$path"}) {
2146 $ok = $self->read_svm_props($ra, $path, $r);
2147 last if $ok;
2148 $tried{"$ra->{url}/$path"} = 1;
2149 }
2150 $path =~ s#/?[^/]+$##;
2151 }
2152 die "Path: '$path' should be ''\n" if $path ne '';
2153 $ok ||= $self->read_svm_props($ra, $path, $r);
2154 $tried{"$ra->{url}/$path"} = 1;
2155 if (!$ok) {
2156 die @err, (map { " $_\n" } keys %tried), "\n";
2157 }
2158 Git::SVN::Ra->new($self->{url});
2159 }
2160
2161 sub svnsync {
2162 my ($self) = @_;
2163 return $self->{svnsync} if $self->{svnsync};
2164
2165 if ($self->no_metadata) {
2166 die "Can't have both 'noMetadata' and ",
2167 "'useSvnsyncProps' options set!\n";
2168 }
2169 if ($self->rewrite_root) {
2170 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2171 "options set!\n";
2172 }
2173
2174 my $svnsync;
2175 # see if we have it in our config, first:
2176 eval {
2177 my $section = "svn-remote.$self->{repo_id}";
2178
2179 my $url = tmp_config('--get', "$section.svnsync-url");
2180 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2181 die "doesn't look right - svn:sync-from-url is '$url'\n";
2182
2183 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2184 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2185 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2186
2187 $svnsync = { url => $url, uuid => $uuid }
2188 };
2189 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2190 return $self->{svnsync} = $svnsync;
2191 }
2192
2193 my $err = "useSvnsyncProps set, but failed to read " .
2194 "svnsync property: svn:sync-from-";
2195 my $rp = $self->ra->rev_proplist(0);
2196
2197 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2198 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2199 die "doesn't look right - svn:sync-from-url is '$url'\n";
2200
2201 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2202 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2203 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2204
2205 my $section = "svn-remote.$self->{repo_id}";
2206 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2207 tmp_config('--add', "$section.svnsync-url", $url);
2208 return $self->{svnsync} = { url => $url, uuid => $uuid };
2209 }
2210
2211 # this allows us to memoize our SVN::Ra UUID locally and avoid a
2212 # remote lookup (useful for 'git svn log').
2213 sub ra_uuid {
2214 my ($self) = @_;
2215 unless ($self->{ra_uuid}) {
2216 my $key = "svn-remote.$self->{repo_id}.uuid";
2217 my $uuid = eval { tmp_config('--get', $key) };
2218 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2219 $self->{ra_uuid} = $uuid;
2220 } else {
2221 die "ra_uuid called without URL\n" unless $self->{url};
2222 $self->{ra_uuid} = $self->ra->get_uuid;
2223 tmp_config('--add', $key, $self->{ra_uuid});
2224 }
2225 }
2226 $self->{ra_uuid};
2227 }
2228
2229 sub _set_repos_root {
2230 my ($self, $repos_root) = @_;
2231 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2232 $repos_root ||= $self->ra->{repos_root};
2233 tmp_config($k, $repos_root);
2234 $repos_root;
2235 }
2236
2237 sub repos_root {
2238 my ($self) = @_;
2239 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2240 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2241 }
2242
2243 sub ra {
2244 my ($self) = shift;
2245 my $ra = Git::SVN::Ra->new($self->{url});
2246 $self->_set_repos_root($ra->{repos_root});
2247 if ($self->use_svm_props && !$self->{svm}) {
2248 if ($self->no_metadata) {
2249 die "Can't have both 'noMetadata' and ",
2250 "'useSvmProps' options set!\n";
2251 } elsif ($self->use_svnsync_props) {
2252 die "Can't have both 'useSvnsyncProps' and ",
2253 "'useSvmProps' options set!\n";
2254 }
2255 $ra = $self->_set_svm_vars($ra);
2256 $self->{-want_revprops} = 1;
2257 }
2258 $ra;
2259 }
2260
2261 # prop_walk(PATH, REV, SUB)
2262 # -------------------------
2263 # Recursively traverse PATH at revision REV and invoke SUB for each
2264 # directory that contains a SVN property. SUB will be invoked as
2265 # follows: &SUB(gs, path, props); where `gs' is this instance of
2266 # Git::SVN, `path' the path to the directory where the properties
2267 # `props' were found. The `path' will be relative to point of checkout,
2268 # that is, if url://repo/trunk is the current Git branch, and that
2269 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
2270 # as `path' (note the trailing `/').
2271 sub prop_walk {
2272 my ($self, $path, $rev, $sub) = @_;
2273
2274 $path =~ s#^/##;
2275 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2276 $path =~ s#^/*#/#g;
2277 my $p = $path;
2278 # Strip the irrelevant part of the path.
2279 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2280 # Ensure the path is terminated by a `/'.
2281 $p =~ s#/*$#/#;
2282
2283 # The properties contain all the internal SVN stuff nobody
2284 # (usually) cares about.
2285 my $interesting_props = 0;
2286 foreach (keys %{$props}) {
2287 # If it doesn't start with `svn:', it must be a
2288 # user-defined property.
2289 ++$interesting_props and next if $_ !~ /^svn:/;
2290 # FIXME: Fragile, if SVN adds new public properties,
2291 # this needs to be updated.
2292 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2293 |eol-style|mime-type
2294 |externals|needs-lock)$/x;
2295 }
2296 &$sub($self, $p, $props) if $interesting_props;
2297
2298 foreach (sort keys %$dirent) {
2299 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2300 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2301 }
2302 }
2303
2304 sub last_rev { ($_[0]->last_rev_commit)[0] }
2305 sub last_commit { ($_[0]->last_rev_commit)[1] }
2306
2307 # returns the newest SVN revision number and newest commit SHA1
2308 sub last_rev_commit {
2309 my ($self) = @_;
2310 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2311 return ($self->{last_rev}, $self->{last_commit});
2312 }
2313 my $c = ::verify_ref($self->refname.'^0');
2314 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2315 my $rev = (::cmt_metadata($c))[1];
2316 if (defined $rev) {
2317 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2318 return ($rev, $c);
2319 }
2320 }
2321 my $map_path = $self->map_path;
2322 unless (-e $map_path) {
2323 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2324 return (undef, undef);
2325 }
2326 my ($rev, $commit) = $self->rev_map_max(1);
2327 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2328 return ($rev, $commit);
2329 }
2330
2331 sub get_fetch_range {
2332 my ($self, $min, $max) = @_;
2333 $max ||= $self->ra->get_latest_revnum;
2334 $min ||= $self->rev_map_max;
2335 (++$min, $max);
2336 }
2337
2338 sub tmp_config {
2339 my (@args) = @_;
2340 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2341 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2342 if (! -f $config && -f $old_def_config) {
2343 rename $old_def_config, $config or
2344 die "Failed rename $old_def_config => $config: $!\n";
2345 }
2346 my $old_config = $ENV{GIT_CONFIG};
2347 $ENV{GIT_CONFIG} = $config;
2348 $@ = undef;
2349 my @ret = eval {
2350 unless (-f $config) {
2351 mkfile($config);
2352 open my $fh, '>', $config or
2353 die "Can't open $config: $!\n";
2354 print $fh "; This file is used internally by ",
2355 "git-svn\n" or die
2356 "Couldn't write to $config: $!\n";
2357 print $fh "; You should not have to edit it\n" or
2358 die "Couldn't write to $config: $!\n";
2359 close $fh or die "Couldn't close $config: $!\n";
2360 }
2361 command('config', @args);
2362 };
2363 my $err = $@;
2364 if (defined $old_config) {
2365 $ENV{GIT_CONFIG} = $old_config;
2366 } else {
2367 delete $ENV{GIT_CONFIG};
2368 }
2369 die $err if $err;
2370 wantarray ? @ret : $ret[0];
2371 }
2372
2373 sub tmp_index_do {
2374 my ($self, $sub) = @_;
2375 my $old_index = $ENV{GIT_INDEX_FILE};
2376 $ENV{GIT_INDEX_FILE} = $self->{index};
2377 $@ = undef;
2378 my @ret = eval {
2379 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2380 mkpath([$dir]) unless -d $dir;
2381 &$sub;
2382 };
2383 my $err = $@;
2384 if (defined $old_index) {
2385 $ENV{GIT_INDEX_FILE} = $old_index;
2386 } else {
2387 delete $ENV{GIT_INDEX_FILE};
2388 }
2389 die $err if $err;
2390 wantarray ? @ret : $ret[0];
2391 }
2392
2393 sub assert_index_clean {
2394 my ($self, $treeish) = @_;
2395
2396 $self->tmp_index_do(sub {
2397 command_noisy('read-tree', $treeish) unless -e $self->{index};
2398 my $x = command_oneline('write-tree');
2399 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2400 /^tree ($::sha1)/mo);
2401 return if $y eq $x;
2402
2403 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2404 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2405 command_noisy('read-tree', $treeish);
2406 $x = command_oneline('write-tree');
2407 if ($y ne $x) {
2408 ::fatal "trees ($treeish) $y != $x\n",
2409 "Something is seriously wrong...";
2410 }
2411 });
2412 }
2413
2414 sub get_commit_parents {
2415 my ($self, $log_entry) = @_;
2416 my (%seen, @ret, @tmp);
2417 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2418 if (my $ip = $self->{inject_parents}) {
2419 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2420 push @tmp, $commit;
2421 }
2422 }
2423 if (my $cur = ::verify_ref($self->refname.'^0')) {
2424 push @tmp, $cur;
2425 }
2426 if (my $ipd = $self->{inject_parents_dcommit}) {
2427 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2428 push @tmp, @$commit;
2429 }
2430 }
2431 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2432 while (my $p = shift @tmp) {
2433 next if $seen{$p};
2434 $seen{$p} = 1;
2435 push @ret, $p;
2436 # MAXPARENT is defined to 16 in commit-tree.c:
2437 last if @ret >= 16;
2438 }
2439 if (@tmp) {
2440 die "r$log_entry->{revision}: No room for parents:\n\t",
2441 join("\n\t", @tmp), "\n";
2442 }
2443 @ret;
2444 }
2445
2446 sub rewrite_root {
2447 my ($self) = @_;
2448 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2449 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2450 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2451 if ($rwr) {
2452 $rwr =~ s#/+$##;
2453 if ($rwr !~ m#^[a-z\+]+://#) {
2454 die "$rwr is not a valid URL (key: $k)\n";
2455 }
2456 }
2457 $self->{-rewrite_root} = $rwr;
2458 }
2459
2460 sub metadata_url {
2461 my ($self) = @_;
2462 ($self->rewrite_root || $self->{url}) .
2463 (length $self->{path} ? '/' . $self->{path} : '');
2464 }
2465
2466 sub full_url {
2467 my ($self) = @_;
2468 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2469 }
2470
2471
2472 sub set_commit_header_env {
2473 my ($log_entry) = @_;
2474 my %env;
2475 foreach my $ned (qw/NAME EMAIL DATE/) {
2476 foreach my $ac (qw/AUTHOR COMMITTER/) {
2477 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2478 }
2479 }
2480
2481 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2482 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2483 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2484
2485 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2486 ? $log_entry->{commit_name}
2487 : $log_entry->{name};
2488 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2489 ? $log_entry->{commit_email}
2490 : $log_entry->{email};
2491 \%env;
2492 }
2493
2494 sub restore_commit_header_env {
2495 my ($env) = @_;
2496 foreach my $ned (qw/NAME EMAIL DATE/) {
2497 foreach my $ac (qw/AUTHOR COMMITTER/) {
2498 my $k = "GIT_${ac}_${ned}";
2499 if (defined $env->{$k}) {
2500 $ENV{$k} = $env->{$k};
2501 } else {
2502 delete $ENV{$k};
2503 }
2504 }
2505 }
2506 }
2507
2508 sub gc {
2509 command_noisy('gc', '--auto');
2510 };
2511
2512 sub do_git_commit {
2513 my ($self, $log_entry) = @_;
2514 my $lr = $self->last_rev;
2515 if (defined $lr && $lr >= $log_entry->{revision}) {
2516 die "Last fetched revision of ", $self->refname,
2517 " was r$lr, but we are about to fetch: ",
2518 "r$log_entry->{revision}!\n";
2519 }
2520 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2521 croak "$log_entry->{revision} = $c already exists! ",
2522 "Why are we refetching it?\n";
2523 }
2524 my $old_env = set_commit_header_env($log_entry);
2525 my $tree = $log_entry->{tree};
2526 if (!defined $tree) {
2527 $tree = $self->tmp_index_do(sub {
2528 command_oneline('write-tree') });
2529 }
2530 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2531
2532 my @exec = ('git', 'commit-tree', $tree);
2533 foreach ($self->get_commit_parents($log_entry)) {
2534 push @exec, '-p', $_;
2535 }
2536 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2537 or croak $!;
2538 binmode $msg_fh;
2539
2540 # we always get UTF-8 from SVN, but we may want our commits in
2541 # a different encoding.
2542 if (my $enc = Git::config('i18n.commitencoding')) {
2543 require Encode;
2544 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2545 }
2546 print $msg_fh $log_entry->{log} or croak $!;
2547 restore_commit_header_env($old_env);
2548 unless ($self->no_metadata) {
2549 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2550 or croak $!;
2551 }
2552 $msg_fh->flush == 0 or croak $!;
2553 close $msg_fh or croak $!;
2554 chomp(my $commit = do { local $/; <$out_fh> });
2555 close $out_fh or croak $!;
2556 waitpid $pid, 0;
2557 croak $? if $?;
2558 if ($commit !~ /^$::sha1$/o) {
2559 die "Failed to commit, invalid sha1: $commit\n";
2560 }
2561
2562 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2563
2564 $self->{last_rev} = $log_entry->{revision};
2565 $self->{last_commit} = $commit;
2566 print "r$log_entry->{revision}" unless $::_q > 1;
2567 if (defined $log_entry->{svm_revision}) {
2568 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2569 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2570 0, $self->svm_uuid);
2571 }
2572 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2573 if (--$_gc_nr == 0) {
2574 $_gc_nr = $_gc_period;
2575 gc();
2576 }
2577 return $commit;
2578 }
2579
2580 sub match_paths {
2581 my ($self, $paths, $r) = @_;
2582 return 1 if $self->{path} eq '';
2583 if (my $path = $paths->{"/$self->{path}"}) {
2584 return ($path->{action} eq 'D') ? 0 : 1;
2585 }
2586 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2587 if (grep /$self->{path_regex}/, keys %$paths) {
2588 return 1;
2589 }
2590 my $c = '';
2591 foreach (split m#/#, $self->{path}) {
2592 $c .= "/$_";
2593 next unless ($paths->{$c} &&
2594 ($paths->{$c}->{action} =~ /^[AR]$/));
2595 if ($self->ra->check_path($self->{path}, $r) ==
2596 $SVN::Node::dir) {
2597 return 1;
2598 }
2599 }
2600 return 0;
2601 }
2602
2603 sub find_parent_branch {
2604 my ($self, $paths, $rev) = @_;
2605 return undef unless $self->follow_parent;
2606 unless (defined $paths) {
2607 my $err_handler = $SVN::Error::handler;
2608 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2609 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2610 sub { $paths = $_[0] });
2611 $SVN::Error::handler = $err_handler;
2612 }
2613 return undef unless defined $paths;
2614
2615 # look for a parent from another branch:
2616 my @b_path_components = split m#/#, $self->{path};
2617 my @a_path_components;
2618 my $i;
2619 while (@b_path_components) {
2620 $i = $paths->{'/'.join('/', @b_path_components)};
2621 last if $i && defined $i->{copyfrom_path};
2622 unshift(@a_path_components, pop(@b_path_components));
2623 }
2624 return undef unless defined $i && defined $i->{copyfrom_path};
2625 my $branch_from = $i->{copyfrom_path};
2626 if (@a_path_components) {
2627 print STDERR "branch_from: $branch_from => ";
2628 $branch_from .= '/'.join('/', @a_path_components);
2629 print STDERR $branch_from, "\n";
2630 }
2631 my $r = $i->{copyfrom_rev};
2632 my $repos_root = $self->ra->{repos_root};
2633 my $url = $self->ra->{url};
2634 my $new_url = $url . $branch_from;
2635 print STDERR "Found possible branch point: ",
2636 "$new_url => ", $self->full_url, ", $r\n"
2637 unless $::_q > 1;
2638 $branch_from =~ s#^/##;
2639 my $gs = $self->other_gs($new_url, $url,
2640 $branch_from, $r, $self->{ref_id});
2641 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2642 {
2643 my ($base, $head);
2644 if (!defined $r0 || !defined $parent) {
2645 ($base, $head) = parse_revision_argument(0, $r);
2646 } else {
2647 if ($r0 < $r) {
2648 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2649 0, 1, sub { $base = $_[1] - 1 });
2650 }
2651 }
2652 if (defined $base && $base <= $r) {
2653 $gs->fetch($base, $r);
2654 }
2655 ($r0, $parent) = $gs->find_rev_before($r, 1);
2656 }
2657 if (defined $r0 && defined $parent) {
2658 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2659 unless $::_q > 1;
2660 my $ed;
2661 if ($self->ra->can_do_switch) {
2662 $self->assert_index_clean($parent);
2663 print STDERR "Following parent with do_switch\n"
2664 unless $::_q > 1;
2665 # do_switch works with svn/trunk >= r22312, but that
2666 # is not included with SVN 1.4.3 (the latest version
2667 # at the moment), so we can't rely on it
2668 $self->{last_rev} = $r0;
2669 $self->{last_commit} = $parent;
2670 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2671 $gs->ra->gs_do_switch($r0, $rev, $gs,
2672 $self->full_url, $ed)
2673 or die "SVN connection failed somewhere...\n";
2674 } elsif ($self->ra->trees_match($new_url, $r0,
2675 $self->full_url, $rev)) {
2676 print STDERR "Trees match:\n",
2677 " $new_url\@$r0\n",
2678 " ${\$self->full_url}\@$rev\n",
2679 "Following parent with no changes\n"
2680 unless $::_q > 1;
2681 $self->tmp_index_do(sub {
2682 command_noisy('read-tree', $parent);
2683 });
2684 $self->{last_commit} = $parent;
2685 } else {
2686 print STDERR "Following parent with do_update\n"
2687 unless $::_q > 1;
2688 $ed = SVN::Git::Fetcher->new($self);
2689 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2690 or die "SVN connection failed somewhere...\n";
2691 }
2692 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2693 return $self->make_log_entry($rev, [$parent], $ed);
2694 }
2695 return undef;
2696 }
2697
2698 sub do_fetch {
2699 my ($self, $paths, $rev) = @_;
2700 my $ed;
2701 my ($last_rev, @parents);
2702 if (my $lc = $self->last_commit) {
2703 # we can have a branch that was deleted, then re-added
2704 # under the same name but copied from another path, in
2705 # which case we'll have multiple parents (we don't
2706 # want to break the original ref, nor lose copypath info):
2707 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2708 push @{$log_entry->{parents}}, $lc;
2709 return $log_entry;
2710 }
2711 $ed = SVN::Git::Fetcher->new($self);
2712 $last_rev = $self->{last_rev};
2713 $ed->{c} = $lc;
2714 @parents = ($lc);
2715 } else {
2716 $last_rev = $rev;
2717 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2718 return $log_entry;
2719 }
2720 $ed = SVN::Git::Fetcher->new($self);
2721 }
2722 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2723 die "SVN connection failed somewhere...\n";
2724 }
2725 $self->make_log_entry($rev, \@parents, $ed);
2726 }
2727
2728 sub get_untracked {
2729 my ($self, $ed) = @_;
2730 my @out;
2731 my $h = $ed->{empty};
2732 foreach (sort keys %$h) {
2733 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2734 push @out, " $act: " . uri_encode($_);
2735 warn "W: $act: $_\n";
2736 }
2737 foreach my $t (qw/dir_prop file_prop/) {
2738 $h = $ed->{$t} or next;
2739 foreach my $path (sort keys %$h) {
2740 my $ppath = $path eq '' ? '.' : $path;
2741 foreach my $prop (sort keys %{$h->{$path}}) {
2742 next if $SKIP_PROP{$prop};
2743 my $v = $h->{$path}->{$prop};
2744 my $t_ppath_prop = "$t: " .
2745 uri_encode($ppath) . ' ' .
2746 uri_encode($prop);
2747 if (defined $v) {
2748 push @out, " +$t_ppath_prop " .
2749 uri_encode($v);
2750 } else {
2751 push @out, " -$t_ppath_prop";
2752 }
2753 }
2754 }
2755 }
2756 foreach my $t (qw/absent_file absent_directory/) {
2757 $h = $ed->{$t} or next;
2758 foreach my $parent (sort keys %$h) {
2759 foreach my $path (sort @{$h->{$parent}}) {
2760 push @out, " $t: " .
2761 uri_encode("$parent/$path");
2762 warn "W: $t: $parent/$path ",
2763 "Insufficient permissions?\n";
2764 }
2765 }
2766 }
2767 \@out;
2768 }
2769
2770 # parse_svn_date(DATE)
2771 # --------------------
2772 # Given a date (in UTC) from Subversion, return a string in the format
2773 # "<TZ Offset> <local date/time>" that Git will use.
2774 #
2775 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
2776 # is true we'll convert it to the local timezone instead.
2777 sub parse_svn_date {
2778 my $date = shift || return '+0000 1970-01-01 00:00:00';
2779 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2780 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2781 croak "Unable to parse date: $date\n";
2782 my $parsed_date; # Set next.
2783
2784 if ($Git::SVN::_localtime) {
2785 # Translate the Subversion datetime to an epoch time.
2786 # Begin by switching ourselves to $date's timezone, UTC.
2787 my $old_env_TZ = $ENV{TZ};
2788 $ENV{TZ} = 'UTC';
2789
2790 my $epoch_in_UTC =
2791 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2792
2793 # Determine our local timezone (including DST) at the
2794 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2795 # value of TZ, if any, at the time we were run.
2796 if (defined $Git::SVN::Log::TZ) {
2797 $ENV{TZ} = $Git::SVN::Log::TZ;
2798 } else {
2799 delete $ENV{TZ};
2800 }
2801
2802 my $our_TZ =
2803 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2804
2805 # This converts $epoch_in_UTC into our local timezone.
2806 my ($sec, $min, $hour, $mday, $mon, $year,
2807 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2808
2809 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2810 $our_TZ, $year + 1900, $mon + 1,
2811 $mday, $hour, $min, $sec);
2812
2813 # Reset us to the timezone in effect when we entered
2814 # this routine.
2815 if (defined $old_env_TZ) {
2816 $ENV{TZ} = $old_env_TZ;
2817 } else {
2818 delete $ENV{TZ};
2819 }
2820 } else {
2821 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2822 }
2823
2824 return $parsed_date;
2825 }
2826
2827 sub other_gs {
2828 my ($self, $new_url, $url,
2829 $branch_from, $r, $old_ref_id) = @_;
2830 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2831 unless ($gs) {
2832 my $ref_id = $old_ref_id;
2833 $ref_id =~ s/\@\d+$//;
2834 $ref_id .= "\@$r";
2835 # just grow a tail if we're not unique enough :x
2836 $ref_id .= '-' while find_ref($ref_id);
2837 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2838 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2839 if ($u =~ s#^\Q$url\E(/|$)##) {
2840 $p = $u;
2841 $u = $url;
2842 $repo_id = $self->{repo_id};
2843 }
2844 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2845 }
2846 $gs
2847 }
2848
2849 sub call_authors_prog {
2850 my ($orig_author) = @_;
2851 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2852 my $author = `$::_authors_prog $orig_author`;
2853 if ($? != 0) {
2854 die "$::_authors_prog failed with exit code $?\n"
2855 }
2856 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2857 my ($name, $email) = ($1, $2);
2858 $email = undef if length $2 == 0;
2859 return [$name, $email];
2860 } else {
2861 die "Author: $orig_author: $::_authors_prog returned "
2862 . "invalid author format: $author\n";
2863 }
2864 }
2865
2866 sub check_author {
2867 my ($author) = @_;
2868 if (!defined $author || length $author == 0) {
2869 $author = '(no author)';
2870 }
2871 if (!defined $::users{$author}) {
2872 if (defined $::_authors_prog) {
2873 $::users{$author} = call_authors_prog($author);
2874 } elsif (defined $::_authors) {
2875 die "Author: $author not defined in $::_authors file\n";
2876 }
2877 }
2878 $author;
2879 }
2880
2881 sub make_log_entry {
2882 my ($self, $rev, $parents, $ed) = @_;
2883 my $untracked = $self->get_untracked($ed);
2884
2885 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2886 print $un "r$rev\n" or croak $!;
2887 print $un $_, "\n" foreach @$untracked;
2888 my %log_entry = ( parents => $parents || [], revision => $rev,
2889 log => '');
2890
2891 my $headrev;
2892 my $logged = delete $self->{logged_rev_props};
2893 if (!$logged || $self->{-want_revprops}) {
2894 my $rp = $self->ra->rev_proplist($rev);
2895 foreach (sort keys %$rp) {
2896 my $v = $rp->{$_};
2897 if (/^svn:(author|date|log)$/) {
2898 $log_entry{$1} = $v;
2899 } elsif ($_ eq 'svm:headrev') {
2900 $headrev = $v;
2901 } else {
2902 print $un " rev_prop: ", uri_encode($_), ' ',
2903 uri_encode($v), "\n";
2904 }
2905 }
2906 } else {
2907 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2908 }
2909 close $un or croak $!;
2910
2911 $log_entry{date} = parse_svn_date($log_entry{date});
2912 $log_entry{log} .= "\n";
2913 my $author = $log_entry{author} = check_author($log_entry{author});
2914 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2915 : ($author, undef);
2916
2917 my ($commit_name, $commit_email) = ($name, $email);
2918 if ($_use_log_author) {
2919 my $name_field;
2920 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2921 $name_field = $1;
2922 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2923 $name_field = $1;
2924 }
2925 if (!defined $name_field) {
2926 if (!defined $email) {
2927 $email = $name;
2928 }
2929 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2930 ($name, $email) = ($1, $2);
2931 } elsif ($name_field =~ /(.*)@/) {
2932 ($name, $email) = ($1, $name_field);
2933 } else {
2934 ($name, $email) = ($name_field, $name_field);
2935 }
2936 }
2937 if (defined $headrev && $self->use_svm_props) {
2938 if ($self->rewrite_root) {
2939 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2940 "options set!\n";
2941 }
2942 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
2943 # we don't want "SVM: initializing mirror for junk" ...
2944 return undef if $r == 0;
2945 my $svm = $self->svm;
2946 if ($uuid ne $svm->{uuid}) {
2947 die "UUID mismatch on SVM path:\n",
2948 "expected: $svm->{uuid}\n",
2949 " got: $uuid\n";
2950 }
2951 my $full_url = $self->full_url;
2952 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2953 die "Failed to replace '$svm->{replace}' with ",
2954 "'$svm->{source}' in $full_url\n";
2955 # throw away username for storing in records
2956 remove_username($full_url);
2957 $log_entry{metadata} = "$full_url\@$r $uuid";
2958 $log_entry{svm_revision} = $r;
2959 $email ||= "$author\@$uuid";
2960 $commit_email ||= "$author\@$uuid";
2961 } elsif ($self->use_svnsync_props) {
2962 my $full_url = $self->svnsync->{url};
2963 $full_url .= "/$self->{path}" if length $self->{path};
2964 remove_username($full_url);
2965 my $uuid = $self->svnsync->{uuid};
2966 $log_entry{metadata} = "$full_url\@$rev $uuid";
2967 $email ||= "$author\@$uuid";
2968 $commit_email ||= "$author\@$uuid";
2969 } else {
2970 my $url = $self->metadata_url;
2971 remove_username($url);
2972 $log_entry{metadata} = "$url\@$rev " .
2973 $self->ra->get_uuid;
2974 $email ||= "$author\@" . $self->ra->get_uuid;
2975 $commit_email ||= "$author\@" . $self->ra->get_uuid;
2976 }
2977 $log_entry{name} = $name;
2978 $log_entry{email} = $email;
2979 $log_entry{commit_name} = $commit_name;
2980 $log_entry{commit_email} = $commit_email;
2981 \%log_entry;
2982 }
2983
2984 sub fetch {
2985 my ($self, $min_rev, $max_rev, @parents) = @_;
2986 my ($last_rev, $last_commit) = $self->last_rev_commit;
2987 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2988 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2989 }
2990
2991 sub set_tree_cb {
2992 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2993 $self->{inject_parents} = { $rev => $tree };
2994 $self->fetch(undef, undef);
2995 }
2996
2997 sub set_tree {
2998 my ($self, $tree) = (shift, shift);
2999 my $log_entry = ::get_commit_entry($tree);
3000 unless ($self->{last_rev}) {
3001 ::fatal("Must have an existing revision to commit");
3002 }
3003 my %ed_opts = ( r => $self->{last_rev},
3004 log => $log_entry->{log},
3005 ra => $self->ra,
3006 tree_a => $self->{last_commit},
3007 tree_b => $tree,
3008 editor_cb => sub {
3009 $self->set_tree_cb($log_entry, $tree, @_) },
3010 svn_path => $self->{path} );
3011 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3012 print "No changes\nr$self->{last_rev} = $tree\n";
3013 }
3014 }
3015
3016 sub rebuild_from_rev_db {
3017 my ($self, $path) = @_;
3018 my $r = -1;
3019 open my $fh, '<', $path or croak "open: $!";
3020 binmode $fh or croak "binmode: $!";
3021 while (<$fh>) {
3022 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3023 chomp($_);
3024 ++$r;
3025 next if $_ eq ('0' x 40);
3026 $self->rev_map_set($r, $_);
3027 print "r$r = $_\n";
3028 }
3029 close $fh or croak "close: $!";
3030 unlink $path or croak "unlink: $!";
3031 }
3032
3033 sub rebuild {
3034 my ($self) = @_;
3035 my $map_path = $self->map_path;
3036 my $partial = (-e $map_path && ! -z $map_path);
3037 return unless ::verify_ref($self->refname.'^0');
3038 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3039 my $rev_db = $self->rev_db_path;
3040 $self->rebuild_from_rev_db($rev_db);
3041 if ($self->use_svm_props) {
3042 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3043 $self->rebuild_from_rev_db($svm_rev_db);
3044 }
3045 $self->unlink_rev_db_symlink;
3046 return;
3047 }
3048 print "Rebuilding $map_path ...\n" if (!$partial);
3049 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3050 (undef, undef));
3051 my ($log, $ctx) =
3052 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3053 ($head ? "$head.." : "") . $self->refname,
3054 '--');
3055 my $metadata_url = $self->metadata_url;
3056 remove_username($metadata_url);
3057 my $svn_uuid = $self->ra_uuid;
3058 my $c;
3059 while (<$log>) {
3060 if ( m{^commit ($::sha1)$} ) {
3061 $c = $1;
3062 next;
3063 }
3064 next unless s{^\s*(git-svn-id:)}{$1};
3065 my ($url, $rev, $uuid) = ::extract_metadata($_);
3066 remove_username($url);
3067
3068 # ignore merges (from set-tree)
3069 next if (!defined $rev || !$uuid);
3070
3071 # if we merged or otherwise started elsewhere, this is
3072 # how we break out of it
3073 if (($uuid ne $svn_uuid) ||
3074 ($metadata_url && $url && ($url ne $metadata_url))) {
3075 next;
3076 }
3077 if ($partial && $head) {
3078 print "Partial-rebuilding $map_path ...\n";
3079 print "Currently at $base_rev = $head\n";
3080 $head = undef;
3081 }
3082
3083 $self->rev_map_set($rev, $c);
3084 print "r$rev = $c\n";
3085 }
3086 command_close_pipe($log, $ctx);
3087 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3088 my $rev_db_path = $self->rev_db_path;
3089 if (-f $self->rev_db_path) {
3090 unlink $self->rev_db_path or croak "unlink: $!";
3091 }
3092 $self->unlink_rev_db_symlink;
3093 }
3094
3095 # rev_map:
3096 # Tie::File seems to be prone to offset errors if revisions get sparse,
3097 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
3098 # one of my favorite modules is out :< Next up would be one of the DBM
3099 # modules, but I'm not sure which is most portable...
3100 #
3101 # This is the replacement for the rev_db format, which was too big
3102 # and inefficient for large repositories with a lot of sparse history
3103 # (mainly tags)
3104 #
3105 # The format is this:
3106 # - 24 bytes for every record,
3107 # * 4 bytes for the integer representing an SVN revision number
3108 # * 20 bytes representing the sha1 of a git commit
3109 # - No empty padding records like the old format
3110 # (except the last record, which can be overwritten)
3111 # - new records are written append-only since SVN revision numbers
3112 # increase monotonically
3113 # - lookups on SVN revision number are done via a binary search
3114 # - Piping the file to xxd -c24 is a good way of dumping it for
3115 # viewing or editing (piped back through xxd -r), should the need
3116 # ever arise.
3117 # - The last record can be padding revision with an all-zero sha1
3118 # This is used to optimize fetch performance when using multiple
3119 # "fetch" directives in .git/config
3120 #
3121 # These files are disposable unless noMetadata or useSvmProps is set
3122
3123 sub _rev_map_set {
3124 my ($fh, $rev, $commit) = @_;
3125
3126 binmode $fh or croak "binmode: $!";
3127 my $size = (stat($fh))[7];
3128 ($size % 24) == 0 or croak "inconsistent size: $size";
3129
3130 my $wr_offset = 0;
3131 if ($size > 0) {
3132 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3133 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3134 $read == 24 or croak "read only $read bytes (!= 24)";
3135 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3136 if ($last_commit eq ('0' x40)) {
3137 if ($size >= 48) {
3138 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3139 $read = sysread($fh, $buf, 24) or
3140 croak "read: $!";
3141 $read == 24 or
3142 croak "read only $read bytes (!= 24)";
3143 ($last_rev, $last_commit) =
3144 unpack(rev_map_fmt, $buf);
3145 if ($last_commit eq ('0' x40)) {
3146 croak "inconsistent .rev_map\n";
3147 }
3148 }
3149 if ($last_rev >= $rev) {
3150 croak "last_rev is higher!: $last_rev >= $rev";
3151 }
3152 $wr_offset = -24;
3153 }
3154 }
3155 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3156 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3157 croak "write: $!";
3158 }
3159
3160 sub _rev_map_reset {
3161 my ($fh, $rev, $commit) = @_;
3162 my $c = _rev_map_get($fh, $rev);
3163 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3164 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3165 truncate $fh, $offset or croak "truncate: $!";
3166 }
3167
3168 sub mkfile {
3169 my ($path) = @_;
3170 unless (-e $path) {
3171 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3172 mkpath([$dir]) unless -d $dir;
3173 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3174 close $fh or die "Couldn't close (create) $path: $!\n";
3175 }
3176 }
3177
3178 sub rev_map_set {
3179 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3180 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3181 my $db = $self->map_path($uuid);
3182 my $db_lock = "$db.lock";
3183 my $sig;
3184 $update_ref ||= 0;
3185 if ($update_ref) {
3186 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3187 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3188 }
3189 mkfile($db);
3190
3191 $LOCKFILES{$db_lock} = 1;
3192 my $sync;
3193 # both of these options make our .rev_db file very, very important
3194 # and we can't afford to lose it because rebuild() won't work
3195 if ($self->use_svm_props || $self->no_metadata) {
3196 $sync = 1;
3197 copy($db, $db_lock) or die "rev_map_set(@_): ",
3198 "Failed to copy: ",
3199 "$db => $db_lock ($!)\n";
3200 } else {
3201 rename $db, $db_lock or die "rev_map_set(@_): ",
3202 "Failed to rename: ",
3203 "$db => $db_lock ($!)\n";
3204 }
3205
3206 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3207 or croak "Couldn't open $db_lock: $!\n";
3208 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3209 _rev_map_set($fh, $rev, $commit);
3210 if ($sync) {
3211 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3212 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3213 }
3214 close $fh or croak $!;
3215 if ($update_ref) {
3216 $_head = $self;
3217 my $note = "";
3218 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3219 command_noisy('update-ref', '-m', "r$rev$note",
3220 $self->refname, $commit);
3221 }
3222 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3223 "$db_lock => $db ($!)\n";
3224 delete $LOCKFILES{$db_lock};
3225 if ($update_ref) {
3226 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3227 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3228 kill $sig, $$ if defined $sig;
3229 }
3230 }
3231
3232 # If want_commit, this will return an array of (rev, commit) where
3233 # commit _must_ be a valid commit in the archive.
3234 # Otherwise, it'll return the max revision (whether or not the
3235 # commit is valid or just a 0x40 placeholder).
3236 sub rev_map_max {
3237 my ($self, $want_commit) = @_;
3238 $self->rebuild;
3239 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3240 $want_commit ? ($r, $c) : $r;
3241 }
3242
3243 sub rev_map_max_norebuild {
3244 my ($self, $want_commit) = @_;
3245 my $map_path = $self->map_path;
3246 stat $map_path or return $want_commit ? (0, undef) : 0;
3247 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3248 binmode $fh or croak "binmode: $!";
3249 my $size = (stat($fh))[7];
3250 ($size % 24) == 0 or croak "inconsistent size: $size";
3251
3252 if ($size == 0) {
3253 close $fh or croak "close: $!";
3254 return $want_commit ? (0, undef) : 0;
3255 }
3256
3257 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3258 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3259 my ($r, $c) = unpack(rev_map_fmt, $buf);
3260 if ($want_commit && $c eq ('0' x40)) {
3261 if ($size < 48) {
3262 return $want_commit ? (0, undef) : 0;
3263 }
3264 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3265 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3266 ($r, $c) = unpack(rev_map_fmt, $buf);
3267 if ($c eq ('0'x40)) {
3268 croak "Penultimate record is all-zeroes in $map_path";
3269 }
3270 }
3271 close $fh or croak "close: $!";
3272 $want_commit ? ($r, $c) : $r;
3273 }
3274
3275 sub rev_map_get {
3276 my ($self, $rev, $uuid) = @_;
3277 my $map_path = $self->map_path($uuid);
3278 return undef unless -e $map_path;
3279
3280 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3281 my $c = _rev_map_get($fh, $rev);
3282 close($fh) or croak "close: $!";
3283 $c
3284 }
3285
3286 sub _rev_map_get {
3287 my ($fh, $rev) = @_;
3288
3289 binmode $fh or croak "binmode: $!";
3290 my $size = (stat($fh))[7];
3291 ($size % 24) == 0 or croak "inconsistent size: $size";
3292
3293 if ($size == 0) {
3294 return undef;
3295 }
3296
3297 my ($l, $u) = (0, $size - 24);
3298 my ($r, $c, $buf);
3299
3300 while ($l <= $u) {
3301 my $i = int(($l/24 + $u/24) / 2) * 24;
3302 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3303 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3304 my ($r, $c) = unpack(rev_map_fmt, $buf);
3305
3306 if ($r < $rev) {
3307 $l = $i + 24;
3308 } elsif ($r > $rev) {
3309 $u = $i - 24;
3310 } else { # $r == $rev
3311 return $c eq ('0' x 40) ? undef : $c;
3312 }
3313 }
3314 undef;
3315 }
3316
3317 # Finds the first svn revision that exists on (if $eq_ok is true) or
3318 # before $rev for the current branch. It will not search any lower
3319 # than $min_rev. Returns the git commit hash and svn revision number
3320 # if found, else (undef, undef).
3321 sub find_rev_before {
3322 my ($self, $rev, $eq_ok, $min_rev) = @_;
3323 --$rev unless $eq_ok;
3324 $min_rev ||= 1;
3325 my $max_rev = $self->rev_map_max;
3326 $rev = $max_rev if ($rev > $max_rev);
3327 while ($rev >= $min_rev) {
3328 if (my $c = $self->rev_map_get($rev)) {
3329 return ($rev, $c);
3330 }
3331 --$rev;
3332 }
3333 return (undef, undef);
3334 }
3335
3336 # Finds the first svn revision that exists on (if $eq_ok is true) or
3337 # after $rev for the current branch. It will not search any higher
3338 # than $max_rev. Returns the git commit hash and svn revision number
3339 # if found, else (undef, undef).
3340 sub find_rev_after {
3341 my ($self, $rev, $eq_ok, $max_rev) = @_;
3342 ++$rev unless $eq_ok;
3343 $max_rev ||= $self->rev_map_max;
3344 while ($rev <= $max_rev) {
3345 if (my $c = $self->rev_map_get($rev)) {
3346 return ($rev, $c);
3347 }
3348 ++$rev;
3349 }
3350 return (undef, undef);
3351 }
3352
3353 sub _new {
3354 my ($class, $repo_id, $ref_id, $path) = @_;
3355 unless (defined $repo_id && length $repo_id) {
3356 $repo_id = $Git::SVN::default_repo_id;
3357 }
3358 unless (defined $ref_id && length $ref_id) {
3359 $_prefix = '' unless defined($_prefix);
3360 $_[2] = $ref_id =
3361 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3362 }
3363 $_[1] = $repo_id;
3364 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3365
3366 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3367 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3368 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3369 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3370 if (-d $old_dir && ! -d $dir) {
3371 $dir = $old_dir;
3372 }
3373 }
3374
3375 $_[3] = $path = '' unless (defined $path);
3376 mkpath([$dir]);
3377 bless {
3378 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3379 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3380 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3381 }
3382
3383 # for read-only access of old .rev_db formats
3384 sub unlink_rev_db_symlink {
3385 my ($self) = @_;
3386 my $link = $self->rev_db_path;
3387 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3388 if (-l $link) {
3389 unlink $link or croak "unlink: $link failed!";
3390 }
3391 }
3392
3393 sub rev_db_path {
3394 my ($self, $uuid) = @_;
3395 my $db_path = $self->map_path($uuid);
3396 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3397 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3398 $db_path;
3399 }
3400
3401 # the new replacement for .rev_db
3402 sub map_path {
3403 my ($self, $uuid) = @_;
3404 $uuid ||= $self->ra_uuid;
3405 "$self->{map_root}.$uuid";
3406 }
3407
3408 sub uri_encode {
3409 my ($f) = @_;
3410 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3411 $f
3412 }
3413
3414 sub remove_username {
3415 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3416 }
3417
3418 package Git::SVN::Prompt;
3419 use strict;
3420 use warnings;
3421 require SVN::Core;
3422 use vars qw/$_no_auth_cache $_username/;
3423
3424 sub simple {
3425 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3426 $may_save = undef if $_no_auth_cache;
3427 $default_username = $_username if defined $_username;
3428 if (defined $default_username && length $default_username) {
3429 if (defined $realm && length $realm) {
3430 print STDERR "Authentication realm: $realm\n";
3431 STDERR->flush;
3432 }
3433 $cred->username($default_username);
3434 } else {
3435 username($cred, $realm, $may_save, $pool);
3436 }
3437 $cred->password(_read_password("Password for '" .
3438 $cred->username . "': ", $realm));
3439 $cred->may_save($may_save);
3440 $SVN::_Core::SVN_NO_ERROR;
3441 }
3442
3443 sub ssl_server_trust {
3444 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3445 $may_save = undef if $_no_auth_cache;
3446 print STDERR "Error validating server certificate for '$realm':\n";
3447 {
3448 no warnings 'once';
3449 # All variables SVN::Auth::SSL::* are used only once,
3450 # so we're shutting up Perl warnings about this.
3451 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3452 print STDERR " - The certificate is not issued ",
3453 "by a trusted authority. Use the\n",
3454 " fingerprint to validate ",
3455 "the certificate manually!\n";
3456 }
3457 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3458 print STDERR " - The certificate hostname ",
3459 "does not match.\n";
3460 }
3461 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3462 print STDERR " - The certificate is not yet valid.\n";
3463 }
3464 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3465 print STDERR " - The certificate has expired.\n";
3466 }
3467 if ($failures & $SVN::Auth::SSL::OTHER) {
3468 print STDERR " - The certificate has ",
3469 "an unknown error.\n";
3470 }
3471 } # no warnings 'once'
3472 printf STDERR
3473 "Certificate information:\n".
3474 " - Hostname: %s\n".
3475 " - Valid: from %s until %s\n".
3476 " - Issuer: %s\n".
3477 " - Fingerprint: %s\n",
3478 map $cert_info->$_, qw(hostname valid_from valid_until
3479 issuer_dname fingerprint);
3480 my $choice;
3481 prompt:
3482 print STDERR $may_save ?
3483 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3484 "(R)eject or accept (t)emporarily? ";
3485 STDERR->flush;
3486 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3487 if ($choice =~ /^t$/i) {
3488 $cred->may_save(undef);
3489 } elsif ($choice =~ /^r$/i) {
3490 return -1;
3491 } elsif ($may_save && $choice =~ /^p$/i) {
3492 $cred->may_save($may_save);
3493 } else {
3494 goto prompt;
3495 }
3496 $cred->accepted_failures($failures);
3497 $SVN::_Core::SVN_NO_ERROR;
3498 }
3499
3500 sub ssl_client_cert {
3501 my ($cred, $realm, $may_save, $pool) = @_;
3502 $may_save = undef if $_no_auth_cache;
3503 print STDERR "Client certificate filename: ";
3504 STDERR->flush;
3505 chomp(my $filename = <STDIN>);
3506 $cred->cert_file($filename);
3507 $cred->may_save($may_save);
3508 $SVN::_Core::SVN_NO_ERROR;
3509 }
3510
3511 sub ssl_client_cert_pw {
3512 my ($cred, $realm, $may_save, $pool) = @_;
3513 $may_save = undef if $_no_auth_cache;
3514 $cred->password(_read_password("Password: ", $realm));
3515 $cred->may_save($may_save);
3516 $SVN::_Core::SVN_NO_ERROR;
3517 }
3518
3519 sub username {
3520 my ($cred, $realm, $may_save, $pool) = @_;
3521 $may_save = undef if $_no_auth_cache;
3522 if (defined $realm && length $realm) {
3523 print STDERR "Authentication realm: $realm\n";
3524 }
3525 my $username;
3526 if (defined $_username) {
3527 $username = $_username;
3528 } else {
3529 print STDERR "Username: ";
3530 STDERR->flush;
3531 chomp($username = <STDIN>);
3532 }
3533 $cred->username($username);
3534 $cred->may_save($may_save);
3535 $SVN::_Core::SVN_NO_ERROR;
3536 }
3537
3538 sub _read_password {
3539 my ($prompt, $realm) = @_;
3540 print STDERR $prompt;
3541 STDERR->flush;
3542 require Term::ReadKey;
3543 Term::ReadKey::ReadMode('noecho');
3544 my $password = '';
3545 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3546 last if $key =~ /[\012\015]/; # \n\r
3547 $password .= $key;
3548 }
3549 Term::ReadKey::ReadMode('restore');
3550 print STDERR "\n";
3551 STDERR->flush;
3552 $password;
3553 }
3554
3555 package SVN::Git::Fetcher;
3556 use vars qw/@ISA/;
3557 use strict;
3558 use warnings;
3559 use Carp qw/croak/;
3560 use File::Temp qw/tempfile/;
3561 use IO::File qw//;
3562 use vars qw/$_ignore_regex/;
3563
3564 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
3565 sub new {
3566 my ($class, $git_svn, $switch_path) = @_;
3567 my $self = SVN::Delta::Editor->new;
3568 bless $self, $class;
3569 if (exists $git_svn->{last_commit}) {
3570 $self->{c} = $git_svn->{last_commit};
3571 $self->{empty_symlinks} =
3572 _mark_empty_symlinks($git_svn, $switch_path);
3573 }
3574 $self->{ignore_regex} = eval { command_oneline('config', '--get',
3575 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3576 $self->{empty} = {};
3577 $self->{dir_prop} = {};
3578 $self->{file_prop} = {};
3579 $self->{absent_dir} = {};
3580 $self->{absent_file} = {};
3581 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3582 $self;
3583 }
3584
3585 # this uses the Ra object, so it must be called before do_{switch,update},
3586 # not inside them (when the Git::SVN::Fetcher object is passed) to
3587 # do_{switch,update}
3588 sub _mark_empty_symlinks {
3589 my ($git_svn, $switch_path) = @_;
3590 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3591 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3592
3593 my %ret;
3594 my ($rev, $cmt) = $git_svn->last_rev_commit;
3595 return {} unless ($rev && $cmt);
3596
3597 # allow the warning to be printed for each revision we fetch to
3598 # ensure the user sees it. The user can also disable the workaround
3599 # on the repository even while git svn is running and the next
3600 # revision fetched will skip this expensive function.
3601 my $printed_warning;
3602 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3603 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3604 local $/ = "\0";
3605 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3606 $pfx .= '/' if length($pfx);
3607 while (<$ls>) {
3608 chomp;
3609 s/\A100644 blob $empty_blob\t//o or next;
3610 unless ($printed_warning) {
3611 print STDERR "Scanning for empty symlinks, ",
3612 "this may take a while if you have ",
3613 "many empty files\n",
3614 "You may disable this with `",
3615 "git config svn.brokenSymlinkWorkaround ",
3616 "false'.\n",
3617 "This may be done in a different ",
3618 "terminal without restarting ",
3619 "git svn\n";
3620 $printed_warning = 1;
3621 }
3622 my $path = $_;
3623 my (undef, $props) =
3624 $git_svn->ra->get_file($pfx.$path, $rev, undef);
3625 if ($props->{'svn:special'}) {
3626 $ret{$path} = 1;
3627 }
3628 }
3629 command_close_pipe($ls, $ctx);
3630 \%ret;
3631 }
3632
3633 # returns true if a given path is inside a ".git" directory
3634 sub in_dot_git {
3635 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3636 }
3637
3638 # return value: 0 -- don't ignore, 1 -- ignore
3639 sub is_path_ignored {
3640 my ($self, $path) = @_;
3641 return 1 if in_dot_git($path);
3642 return 1 if defined($self->{ignore_regex}) &&
3643 $path =~ m!$self->{ignore_regex}!;
3644 return 0 unless defined($_ignore_regex);
3645 return 1 if $path =~ m!$_ignore_regex!o;
3646 return 0;
3647 }
3648
3649 sub set_path_strip {
3650 my ($self, $path) = @_;
3651 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3652 }
3653
3654 sub open_root {
3655 { path => '' };
3656 }
3657
3658 sub open_directory {
3659 my ($self, $path, $pb, $rev) = @_;
3660 { path => $path };
3661 }
3662
3663 sub git_path {
3664 my ($self, $path) = @_;
3665 if ($self->{path_strip}) {
3666 $path =~ s!$self->{path_strip}!! or
3667 die "Failed to strip path '$path' ($self->{path_strip})\n";
3668 }
3669 $path;
3670 }
3671
3672 sub delete_entry {
3673 my ($self, $path, $rev, $pb) = @_;
3674 return undef if $self->is_path_ignored($path);
3675
3676 my $gpath = $self->git_path($path);
3677 return undef if ($gpath eq '');
3678
3679 # remove entire directories.
3680 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3681 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3682 if ($tree) {
3683 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3684 -r --name-only -z/,
3685 $tree);
3686 local $/ = "\0";
3687 while (<$ls>) {
3688 chomp;
3689 my $rmpath = "$gpath/$_";
3690 $self->{gii}->remove($rmpath);
3691 print "\tD\t$rmpath\n" unless $::_q;
3692 }
3693 print "\tD\t$gpath/\n" unless $::_q;
3694 command_close_pipe($ls, $ctx);
3695 $self->{empty}->{$path} = 0
3696 } else {
3697 $self->{gii}->remove($gpath);
3698 print "\tD\t$gpath\n" unless $::_q;
3699 }
3700 undef;
3701 }
3702
3703 sub open_file {
3704 my ($self, $path, $pb, $rev) = @_;
3705 my ($mode, $blob);
3706
3707 goto out if $self->is_path_ignored($path);
3708
3709 my $gpath = $self->git_path($path);
3710 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3711 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3712 unless (defined $mode && defined $blob) {
3713 die "$path was not found in commit $self->{c} (r$rev)\n";
3714 }
3715 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3716 $mode = '120000';
3717 }
3718 out:
3719 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3720 pool => SVN::Pool->new, action => 'M' };
3721 }
3722
3723 sub add_file {
3724 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3725 my $mode;
3726
3727 if (!$self->is_path_ignored($path)) {
3728 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3729 delete $self->{empty}->{$dir};
3730 $mode = '100644';
3731 }
3732 { path => $path, mode_a => $mode, mode_b => $mode,
3733 pool => SVN::Pool->new, action => 'A' };
3734 }
3735
3736 sub add_directory {
3737 my ($self, $path, $cp_path, $cp_rev) = @_;
3738 goto out if $self->is_path_ignored($path);
3739 my $gpath = $self->git_path($path);
3740 if ($gpath eq '') {
3741 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3742 -r --name-only -z/,
3743 $self->{c});
3744 local $/ = "\0";
3745 while (<$ls>) {
3746 chomp;
3747 $self->{gii}->remove($_);
3748 print "\tD\t$_\n" unless $::_q;
3749 }
3750 command_close_pipe($ls, $ctx);
3751 $self->{empty}->{$path} = 0;
3752 }
3753 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3754 delete $self->{empty}->{$dir};
3755 $self->{empty}->{$path} = 1;
3756 out:
3757 { path => $path };
3758 }
3759
3760 sub change_dir_prop {
3761 my ($self, $db, $prop, $value) = @_;
3762 return undef if $self->is_path_ignored($db->{path});
3763 $self->{dir_prop}->{$db->{path}} ||= {};
3764 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3765 undef;
3766 }
3767
3768 sub absent_directory {
3769 my ($self, $path, $pb) = @_;
3770 return undef if $self->is_path_ignored($path);
3771 $self->{absent_dir}->{$pb->{path}} ||= [];
3772 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3773 undef;
3774 }
3775
3776 sub absent_file {
3777 my ($self, $path, $pb) = @_;
3778 return undef if $self->is_path_ignored($path);
3779 $self->{absent_file}->{$pb->{path}} ||= [];
3780 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3781 undef;
3782 }
3783
3784 sub change_file_prop {
3785 my ($self, $fb, $prop, $value) = @_;
3786 return undef if $self->is_path_ignored($fb->{path});
3787 if ($prop eq 'svn:executable') {
3788 if ($fb->{mode_b} != 120000) {
3789 $fb->{mode_b} = defined $value ? 100755 : 100644;
3790 }
3791 } elsif ($prop eq 'svn:special') {
3792 $fb->{mode_b} = defined $value ? 120000 : 100644;
3793 } else {
3794 $self->{file_prop}->{$fb->{path}} ||= {};
3795 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3796 }
3797 undef;
3798 }
3799
3800 sub apply_textdelta {
3801 my ($self, $fb, $exp) = @_;
3802 return undef if $self->is_path_ignored($fb->{path});
3803 my $fh = $::_repository->temp_acquire('svn_delta');
3804 # $fh gets auto-closed() by SVN::TxDelta::apply(),
3805 # (but $base does not,) so dup() it for reading in close_file
3806 open my $dup, '<&', $fh or croak $!;
3807 my $base = $::_repository->temp_acquire('git_blob');
3808
3809 if ($fb->{blob}) {
3810 my ($base_is_link, $size);
3811
3812 if ($fb->{mode_a} eq '120000' &&
3813 ! $self->{empty_symlinks}->{$fb->{path}}) {
3814 print $base 'link ' or die "print $!\n";
3815 $base_is_link = 1;
3816 }
3817 retry:
3818 $size = $::_repository->cat_blob($fb->{blob}, $base);
3819 die "Failed to read object $fb->{blob}" if ($size < 0);
3820
3821 if (defined $exp) {
3822 seek $base, 0, 0 or croak $!;
3823 my $got = ::md5sum($base);
3824 if ($got ne $exp) {
3825 my $err = "Checksum mismatch: ".
3826 "$fb->{path} $fb->{blob}\n" .
3827 "expected: $exp\n" .
3828 " got: $got\n";
3829 if ($base_is_link) {
3830 warn $err,
3831 "Retrying... (possibly ",
3832 "a bad symlink from SVN)\n";
3833 $::_repository->temp_reset($base);
3834 $base_is_link = 0;
3835 goto retry;
3836 }
3837 die $err;
3838 }
3839 }
3840 }
3841 seek $base, 0, 0 or croak $!;
3842 $fb->{fh} = $fh;
3843 $fb->{base} = $base;
3844 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3845 }
3846
3847 sub close_file {
3848 my ($self, $fb, $exp) = @_;
3849 return undef if $self->is_path_ignored($fb->{path});
3850
3851 my $hash;
3852 my $path = $self->git_path($fb->{path});
3853 if (my $fh = $fb->{fh}) {
3854 if (defined $exp) {
3855 seek($fh, 0, 0) or croak $!;
3856 my $got = ::md5sum($fh);
3857 if ($got ne $exp) {
3858 die "Checksum mismatch: $path\n",
3859 "expected: $exp\n got: $got\n";
3860 }
3861 }
3862 if ($fb->{mode_b} == 120000) {
3863 sysseek($fh, 0, 0) or croak $!;
3864 my $rd = sysread($fh, my $buf, 5);
3865
3866 if (!defined $rd) {
3867 croak "sysread: $!\n";
3868 } elsif ($rd == 0) {
3869 warn "$path has mode 120000",
3870 " but it points to nothing\n",
3871 "converting to an empty file with mode",
3872 " 100644\n";
3873 $fb->{mode_b} = '100644';
3874 } elsif ($buf ne 'link ') {
3875 warn "$path has mode 120000",
3876 " but is not a link\n";
3877 } else {
3878 my $tmp_fh = $::_repository->temp_acquire(
3879 'svn_hash');
3880 my $res;
3881 while ($res = sysread($fh, my $str, 1024)) {
3882 my $out = syswrite($tmp_fh, $str, $res);
3883 defined($out) && $out == $res
3884 or croak("write ",
3885 Git::temp_path($tmp_fh),
3886 ": $!\n");
3887 }
3888 defined $res or croak $!;
3889
3890 ($fh, $tmp_fh) = ($tmp_fh, $fh);
3891 Git::temp_release($tmp_fh, 1);
3892 }
3893 }
3894
3895 $hash = $::_repository->hash_and_insert_object(
3896 Git::temp_path($fh));
3897 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3898
3899 Git::temp_release($fb->{base}, 1);
3900 Git::temp_release($fh, 1);
3901 } else {
3902 $hash = $fb->{blob} or die "no blob information\n";
3903 }
3904 $fb->{pool}->clear;
3905 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3906 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3907 undef;
3908 }
3909
3910 sub abort_edit {
3911 my $self = shift;
3912 $self->{nr} = $self->{gii}->{nr};
3913 delete $self->{gii};
3914 $self->SUPER::abort_edit(@_);
3915 }
3916
3917 sub close_edit {
3918 my $self = shift;
3919 $self->{git_commit_ok} = 1;
3920 $self->{nr} = $self->{gii}->{nr};
3921 delete $self->{gii};
3922 $self->SUPER::close_edit(@_);
3923 }
3924
3925 package SVN::Git::Editor;
3926 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3927 use strict;
3928 use warnings;
3929 use Carp qw/croak/;
3930 use IO::File;
3931
3932 sub new {
3933 my ($class, $opts) = @_;
3934 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3935 die "$_ required!\n" unless (defined $opts->{$_});
3936 }
3937
3938 my $pool = SVN::Pool->new;
3939 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3940 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3941 $opts->{r}, $mods);
3942
3943 # $opts->{ra} functions should not be used after this:
3944 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3945 $opts->{editor_cb}, $pool);
3946 my $self = SVN::Delta::Editor->new(@ce, $pool);
3947 bless $self, $class;
3948 foreach (qw/svn_path r tree_a tree_b/) {
3949 $self->{$_} = $opts->{$_};
3950 }
3951 $self->{url} = $opts->{ra}->{url};
3952 $self->{mods} = $mods;
3953 $self->{types} = $types;
3954 $self->{pool} = $pool;
3955 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3956 $self->{rm} = { };
3957 $self->{path_prefix} = length $self->{svn_path} ?
3958 "$self->{svn_path}/" : '';
3959 $self->{config} = $opts->{config};
3960 return $self;
3961 }
3962
3963 sub generate_diff {
3964 my ($tree_a, $tree_b) = @_;
3965 my @diff_tree = qw(diff-tree -z -r);
3966 if ($_cp_similarity) {
3967 push @diff_tree, "-C$_cp_similarity";
3968 } else {
3969 push @diff_tree, '-C';
3970 }
3971 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3972 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3973 push @diff_tree, $tree_a, $tree_b;
3974 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3975 local $/ = "\0";
3976 my $state = 'meta';
3977 my @mods;
3978 while (<$diff_fh>) {
3979 chomp $_; # this gets rid of the trailing "\0"
3980 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3981 ($::sha1)\s($::sha1)\s
3982 ([MTCRAD])\d*$/xo) {
3983 push @mods, { mode_a => $1, mode_b => $2,
3984 sha1_a => $3, sha1_b => $4,
3985 chg => $5 };
3986 if ($5 =~ /^(?:C|R)$/) {
3987 $state = 'file_a';
3988 } else {
3989 $state = 'file_b';
3990 }
3991 } elsif ($state eq 'file_a') {
3992 my $x = $mods[$#mods] or croak "Empty array\n";
3993 if ($x->{chg} !~ /^(?:C|R)$/) {
3994 croak "Error parsing $_, $x->{chg}\n";
3995 }
3996 $x->{file_a} = $_;
3997 $state = 'file_b';
3998 } elsif ($state eq 'file_b') {
3999 my $x = $mods[$#mods] or croak "Empty array\n";
4000 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4001 croak "Error parsing $_, $x->{chg}\n";
4002 }
4003 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4004 croak "Error parsing $_, $x->{chg}\n";
4005 }
4006 $x->{file_b} = $_;
4007 $state = 'meta';
4008 } else {
4009 croak "Error parsing $_\n";
4010 }
4011 }
4012 command_close_pipe($diff_fh, $ctx);
4013 \@mods;
4014 }
4015
4016 sub check_diff_paths {
4017 my ($ra, $pfx, $rev, $mods) = @_;
4018 my %types;
4019 $pfx .= '/' if length $pfx;
4020
4021 sub type_diff_paths {
4022 my ($ra, $types, $path, $rev) = @_;
4023 my @p = split m#/+#, $path;
4024 my $c = shift @p;
4025 unless (defined $types->{$c}) {
4026 $types->{$c} = $ra->check_path($c, $rev);
4027 }
4028 while (@p) {
4029 $c .= '/' . shift @p;
4030 next if defined $types->{$c};
4031 $types->{$c} = $ra->check_path($c, $rev);
4032 }
4033 }
4034
4035 foreach my $m (@$mods) {
4036 foreach my $f (qw/file_a file_b/) {
4037 next unless defined $m->{$f};
4038 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4039 if (length $pfx.$dir && ! defined $types{$dir}) {
4040 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4041 }
4042 }
4043 }
4044 \%types;
4045 }
4046
4047 sub split_path {
4048 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4049 }
4050
4051 sub repo_path {
4052 my ($self, $path) = @_;
4053 $self->{path_prefix}.(defined $path ? $path : '');
4054 }
4055
4056 sub url_path {
4057 my ($self, $path) = @_;
4058 if ($self->{url} =~ m#^https?://#) {
4059 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4060 }
4061 $self->{url} . '/' . $self->repo_path($path);
4062 }
4063
4064 sub rmdirs {
4065 my ($self) = @_;
4066 my $rm = $self->{rm};
4067 delete $rm->{''}; # we never delete the url we're tracking
4068 return unless %$rm;
4069
4070 foreach (keys %$rm) {
4071 my @d = split m#/#, $_;
4072 my $c = shift @d;
4073 $rm->{$c} = 1;
4074 while (@d) {
4075 $c .= '/' . shift @d;
4076 $rm->{$c} = 1;
4077 }
4078 }
4079 delete $rm->{$self->{svn_path}};
4080 delete $rm->{''}; # we never delete the url we're tracking
4081 return unless %$rm;
4082
4083 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4084 $self->{tree_b});
4085 local $/ = "\0";
4086 while (<$fh>) {
4087 chomp;
4088 my @dn = split m#/#, $_;
4089 while (pop @dn) {
4090 delete $rm->{join '/', @dn};
4091 }
4092 unless (%$rm) {
4093 close $fh;
4094 return;
4095 }
4096 }
4097 command_close_pipe($fh, $ctx);
4098
4099 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4100 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4101 $self->close_directory($bat->{$d}, $p);
4102 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4103 print "\tD+\t$d/\n" unless $::_q;
4104 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4105 delete $bat->{$d};
4106 }
4107 }
4108
4109 sub open_or_add_dir {
4110 my ($self, $full_path, $baton) = @_;
4111 my $t = $self->{types}->{$full_path};
4112 if (!defined $t) {
4113 die "$full_path not known in r$self->{r} or we have a bug!\n";
4114 }
4115 {
4116 no warnings 'once';
4117 # SVN::Node::none and SVN::Node::file are used only once,
4118 # so we're shutting up Perl's warnings about them.
4119 if ($t == $SVN::Node::none) {
4120 return $self->add_directory($full_path, $baton,
4121 undef, -1, $self->{pool});
4122 } elsif ($t == $SVN::Node::dir) {
4123 return $self->open_directory($full_path, $baton,
4124 $self->{r}, $self->{pool});
4125 } # no warnings 'once'
4126 print STDERR "$full_path already exists in repository at ",
4127 "r$self->{r} and it is not a directory (",
4128 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4129 } # no warnings 'once'
4130 exit 1;
4131 }
4132
4133 sub ensure_path {
4134 my ($self, $path) = @_;
4135 my $bat = $self->{bat};
4136 my $repo_path = $self->repo_path($path);
4137 return $bat->{''} unless (length $repo_path);
4138 my @p = split m#/+#, $repo_path;
4139 my $c = shift @p;
4140 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4141 while (@p) {
4142 my $c0 = $c;
4143 $c .= '/' . shift @p;
4144 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4145 }
4146 return $bat->{$c};
4147 }
4148
4149 # Subroutine to convert a globbing pattern to a regular expression.
4150 # From perl cookbook.
4151 sub glob2pat {
4152 my $globstr = shift;
4153 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4154 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4155 return '^' . $globstr . '$';
4156 }
4157
4158 sub check_autoprop {
4159 my ($self, $pattern, $properties, $file, $fbat) = @_;
4160 # Convert the globbing pattern to a regular expression.
4161 my $regex = glob2pat($pattern);
4162 # Check if the pattern matches the file name.
4163 if($file =~ m/($regex)/) {
4164 # Parse the list of properties to set.
4165 my @props = split(/;/, $properties);
4166 foreach my $prop (@props) {
4167 # Parse 'name=value' syntax and set the property.
4168 if ($prop =~ /([^=]+)=(.*)/) {
4169 my ($n,$v) = ($1,$2);
4170 for ($n, $v) {
4171 s/^\s+//; s/\s+$//;
4172 }
4173 $self->change_file_prop($fbat, $n, $v);
4174 }
4175 }
4176 }
4177 }
4178
4179 sub apply_autoprops {
4180 my ($self, $file, $fbat) = @_;
4181 my $conf_t = ${$self->{config}}{'config'};
4182 no warnings 'once';
4183 # Check [miscellany]/enable-auto-props in svn configuration.
4184 if (SVN::_Core::svn_config_get_bool(
4185 $conf_t,
4186 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4187 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4188 0)) {
4189 # Auto-props are enabled. Enumerate them to look for matches.
4190 my $callback = sub {
4191 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4192 };
4193 SVN::_Core::svn_config_enumerate(
4194 $conf_t,
4195 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4196 $callback);
4197 }
4198 }
4199
4200 sub A {
4201 my ($self, $m) = @_;
4202 my ($dir, $file) = split_path($m->{file_b});
4203 my $pbat = $self->ensure_path($dir);
4204 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4205 undef, -1);
4206 print "\tA\t$m->{file_b}\n" unless $::_q;
4207 $self->apply_autoprops($file, $fbat);
4208 $self->chg_file($fbat, $m);
4209 $self->close_file($fbat,undef,$self->{pool});
4210 }
4211
4212 sub C {
4213 my ($self, $m) = @_;
4214 my ($dir, $file) = split_path($m->{file_b});
4215 my $pbat = $self->ensure_path($dir);
4216 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4217 $self->url_path($m->{file_a}), $self->{r});
4218 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4219 $self->chg_file($fbat, $m);
4220 $self->close_file($fbat,undef,$self->{pool});
4221 }
4222
4223 sub delete_entry {
4224 my ($self, $path, $pbat) = @_;
4225 my $rpath = $self->repo_path($path);
4226 my ($dir, $file) = split_path($rpath);
4227 $self->{rm}->{$dir} = 1;
4228 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4229 }
4230
4231 sub R {
4232 my ($self, $m) = @_;
4233 my ($dir, $file) = split_path($m->{file_b});
4234 my $pbat = $self->ensure_path($dir);
4235 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4236 $self->url_path($m->{file_a}), $self->{r});
4237 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4238 $self->apply_autoprops($file, $fbat);
4239 $self->chg_file($fbat, $m);
4240 $self->close_file($fbat,undef,$self->{pool});
4241
4242 ($dir, $file) = split_path($m->{file_a});
4243 $pbat = $self->ensure_path($dir);
4244 $self->delete_entry($m->{file_a}, $pbat);
4245 }
4246
4247 sub M {
4248 my ($self, $m) = @_;
4249 my ($dir, $file) = split_path($m->{file_b});
4250 my $pbat = $self->ensure_path($dir);
4251 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4252 $pbat,$self->{r},$self->{pool});
4253 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4254 $self->chg_file($fbat, $m);
4255 $self->close_file($fbat,undef,$self->{pool});
4256 }
4257
4258 sub T { shift->M(@_) }
4259
4260 sub change_file_prop {
4261 my ($self, $fbat, $pname, $pval) = @_;
4262 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4263 }
4264
4265 sub _chg_file_get_blob ($$$$) {
4266 my ($self, $fbat, $m, $which) = @_;
4267 my $fh = $::_repository->temp_acquire("git_blob_$which");
4268 if ($m->{"mode_$which"} =~ /^120/) {
4269 print $fh 'link ' or croak $!;
4270 $self->change_file_prop($fbat,'svn:special','*');
4271 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4272 $self->change_file_prop($fbat,'svn:special',undef);
4273 }
4274 my $blob = $m->{"sha1_$which"};
4275 return ($fh,) if ($blob =~ /^0{40}$/);
4276 my $size = $::_repository->cat_blob($blob, $fh);
4277 croak "Failed to read object $blob" if ($size < 0);
4278 $fh->flush == 0 or croak $!;
4279 seek $fh, 0, 0 or croak $!;
4280
4281 my $exp = ::md5sum($fh);
4282 seek $fh, 0, 0 or croak $!;
4283 return ($fh, $exp);
4284 }
4285
4286 sub chg_file {
4287 my ($self, $fbat, $m) = @_;
4288 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4289 $self->change_file_prop($fbat,'svn:executable','*');
4290 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4291 $self->change_file_prop($fbat,'svn:executable',undef);
4292 }
4293 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4294 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4295 my $pool = SVN::Pool->new;
4296 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4297 if (-s $fh_a) {
4298 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4299 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4300 if (defined $res) {
4301 die "Unexpected result from send_txstream: $res\n",
4302 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4303 }
4304 } else {
4305 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4306 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4307 if ($got ne $exp_b);
4308 }
4309 Git::temp_release($fh_b, 1);
4310 Git::temp_release($fh_a, 1);
4311 $pool->clear;
4312 }
4313
4314 sub D {
4315 my ($self, $m) = @_;
4316 my ($dir, $file) = split_path($m->{file_b});
4317 my $pbat = $self->ensure_path($dir);
4318 print "\tD\t$m->{file_b}\n" unless $::_q;
4319 $self->delete_entry($m->{file_b}, $pbat);
4320 }
4321
4322 sub close_edit {
4323 my ($self) = @_;
4324 my ($p,$bat) = ($self->{pool}, $self->{bat});
4325 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4326 next if $_ eq '';
4327 $self->close_directory($bat->{$_}, $p);
4328 }
4329 $self->close_directory($bat->{''}, $p);
4330 $self->SUPER::close_edit($p);
4331 $p->clear;
4332 }
4333
4334 sub abort_edit {
4335 my ($self) = @_;
4336 $self->SUPER::abort_edit($self->{pool});
4337 }
4338
4339 sub DESTROY {
4340 my $self = shift;
4341 $self->SUPER::DESTROY(@_);
4342 $self->{pool}->clear;
4343 }
4344
4345 # this drives the editor
4346 sub apply_diff {
4347 my ($self) = @_;
4348 my $mods = $self->{mods};
4349 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4350 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4351 my $f = $m->{chg};
4352 if (defined $o{$f}) {
4353 $self->$f($m);
4354 } else {
4355 fatal("Invalid change type: $f");
4356 }
4357 }
4358 $self->rmdirs if $_rmdir;
4359 if (@$mods == 0) {
4360 $self->abort_edit;
4361 } else {
4362 $self->close_edit;
4363 }
4364 return scalar @$mods;
4365 }
4366
4367 package Git::SVN::Ra;
4368 use vars qw/@ISA $config_dir $_log_window_size/;
4369 use strict;
4370 use warnings;
4371 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4372
4373 BEGIN {
4374 # enforce temporary pool usage for some simple functions
4375 no strict 'refs';
4376 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4377 get_file/) {
4378 my $SUPER = "SUPER::$f";
4379 *$f = sub {
4380 my $self = shift;
4381 my $pool = SVN::Pool->new;
4382 my @ret = $self->$SUPER(@_,$pool);
4383 $pool->clear;
4384 wantarray ? @ret : $ret[0];
4385 };
4386 }
4387 }
4388
4389 sub _auth_providers () {
4390 [
4391 SVN::Client::get_simple_provider(),
4392 SVN::Client::get_ssl_server_trust_file_provider(),
4393 SVN::Client::get_simple_prompt_provider(
4394 \&Git::SVN::Prompt::simple, 2),
4395 SVN::Client::get_ssl_client_cert_file_provider(),
4396 SVN::Client::get_ssl_client_cert_prompt_provider(
4397 \&Git::SVN::Prompt::ssl_client_cert, 2),
4398 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4399 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4400 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4401 SVN::Client::get_username_provider(),
4402 SVN::Client::get_ssl_server_trust_prompt_provider(
4403 \&Git::SVN::Prompt::ssl_server_trust),
4404 SVN::Client::get_username_prompt_provider(
4405 \&Git::SVN::Prompt::username, 2)
4406 ]
4407 }
4408
4409 sub escape_uri_only {
4410 my ($uri) = @_;
4411 my @tmp;
4412 foreach (split m{/}, $uri) {
4413 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4414 push @tmp, $_;
4415 }
4416 join('/', @tmp);
4417 }
4418
4419 sub escape_url {
4420 my ($url) = @_;
4421 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4422 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4423 $url = "$scheme://$domain$uri";
4424 }
4425 $url;
4426 }
4427
4428 sub new {
4429 my ($class, $url) = @_;
4430 $url =~ s!/+$!!;
4431 return $RA if ($RA && $RA->{url} eq $url);
4432
4433 SVN::_Core::svn_config_ensure($config_dir, undef);
4434 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4435 my $config = SVN::Core::config_get_config($config_dir);
4436 $RA = undef;
4437 my $dont_store_passwords = 1;
4438 my $conf_t = ${$config}{'config'};
4439 {
4440 no warnings 'once';
4441 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4442 # produces warnings that variables are used only once.
4443 # I had not found the better way to shut them up, so
4444 # the warnings of type 'once' are disabled in this block.
4445 if (SVN::_Core::svn_config_get_bool($conf_t,
4446 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4447 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4448 1) == 0) {
4449 SVN::_Core::svn_auth_set_parameter($baton,
4450 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4451 bless (\$dont_store_passwords, "_p_void"));
4452 }
4453 if (SVN::_Core::svn_config_get_bool($conf_t,
4454 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4455 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4456 1) == 0) {
4457 $Git::SVN::Prompt::_no_auth_cache = 1;
4458 }
4459 } # no warnings 'once'
4460 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4461 config => $config,
4462 pool => SVN::Pool->new,
4463 auth_provider_callbacks => $callbacks);
4464 $self->{url} = $url;
4465 $self->{svn_path} = $url;
4466 $self->{repos_root} = $self->get_repos_root;
4467 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4468 $self->{cache} = { check_path => { r => 0, data => {} },
4469 get_dir => { r => 0, data => {} } };
4470 $RA = bless $self, $class;
4471 }
4472
4473 sub check_path {
4474 my ($self, $path, $r) = @_;
4475 my $cache = $self->{cache}->{check_path};
4476 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4477 return $cache->{data}->{$path};
4478 }
4479 my $pool = SVN::Pool->new;
4480 my $t = $self->SUPER::check_path($path, $r, $pool);
4481 $pool->clear;
4482 if ($r != $cache->{r}) {
4483 %{$cache->{data}} = ();
4484 $cache->{r} = $r;
4485 }
4486 $cache->{data}->{$path} = $t;
4487 }
4488
4489 sub get_dir {
4490 my ($self, $dir, $r) = @_;
4491 my $cache = $self->{cache}->{get_dir};
4492 if ($r == $cache->{r}) {
4493 if (my $x = $cache->{data}->{$dir}) {
4494 return wantarray ? @$x : $x->[0];
4495 }
4496 }
4497 my $pool = SVN::Pool->new;
4498 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4499 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4500 $pool->clear;
4501 if ($r != $cache->{r}) {
4502 %{$cache->{data}} = ();
4503 $cache->{r} = $r;
4504 }
4505 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4506 wantarray ? (\%dirents, $r, $props) : \%dirents;
4507 }
4508
4509 sub DESTROY {
4510 # do not call the real DESTROY since we store ourselves in $RA
4511 }
4512
4513 # get_log(paths, start, end, limit,
4514 # discover_changed_paths, strict_node_history, receiver)
4515 sub get_log {
4516 my ($self, @args) = @_;
4517 my $pool = SVN::Pool->new;
4518
4519 # svn_log_changed_path_t objects passed to get_log are likely to be
4520 # overwritten even if only the refs are copied to an external variable,
4521 # so we should dup the structures in their entirety. Using an
4522 # externally passed pool (instead of our temporary and quickly cleared
4523 # pool in Git::SVN::Ra) does not help matters at all...
4524 my $receiver = pop @args;
4525 my $prefix = "/".$self->{svn_path};
4526 $prefix =~ s#/+($)##;
4527 my $prefix_regex = qr#^\Q$prefix\E#;
4528 push(@args, sub {
4529 my ($paths) = $_[0];
4530 return &$receiver(@_) unless $paths;
4531 $_[0] = ();
4532 foreach my $p (keys %$paths) {
4533 my $i = $paths->{$p};
4534 # Make path relative to our url, not repos_root
4535 $p =~ s/$prefix_regex//;
4536 my %s = map { $_ => $i->$_; }
4537 qw/copyfrom_path copyfrom_rev action/;
4538 if ($s{'copyfrom_path'}) {
4539 $s{'copyfrom_path'} =~ s/$prefix_regex//;
4540 }
4541 $_[0]{$p} = \%s;
4542 }
4543 &$receiver(@_);
4544 });
4545
4546
4547 # the limit parameter was not supported in SVN 1.1.x, so we
4548 # drop it. Therefore, the receiver callback passed to it
4549 # is made aware of this limitation by being wrapped if
4550 # the limit passed to is being wrapped.
4551 if ($SVN::Core::VERSION le '1.2.0') {
4552 my $limit = splice(@args, 3, 1);
4553 if ($limit > 0) {
4554 my $receiver = pop @args;
4555 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4556 }
4557 }
4558 my $ret = $self->SUPER::get_log(@args, $pool);
4559 $pool->clear;
4560 $ret;
4561 }
4562
4563 sub trees_match {
4564 my ($self, $url1, $rev1, $url2, $rev2) = @_;
4565 my $ctx = SVN::Client->new(auth => _auth_providers);
4566 my $out = IO::File->new_tmpfile;
4567
4568 # older SVN (1.1.x) doesn't take $pool as the last parameter for
4569 # $ctx->diff(), so we'll create a default one
4570 my $pool = SVN::Pool->new_default_sub;
4571
4572 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4573 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4574 $out->flush;
4575 my $ret = (($out->stat)[7] == 0);
4576 close $out or croak $!;
4577
4578 $ret;
4579 }
4580
4581 sub get_commit_editor {
4582 my ($self, $log, $cb, $pool) = @_;
4583 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4584 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4585 }
4586
4587 sub gs_do_update {
4588 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4589 my $new = ($rev_a == $rev_b);
4590 my $path = $gs->{path};
4591
4592 if ($new && -e $gs->{index}) {
4593 unlink $gs->{index} or die
4594 "Couldn't unlink index: $gs->{index}: $!\n";
4595 }
4596 my $pool = SVN::Pool->new;
4597 $editor->set_path_strip($path);
4598 my (@pc) = split m#/#, $path;
4599 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4600 1, $editor, $pool);
4601 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4602
4603 # Since we can't rely on svn_ra_reparent being available, we'll
4604 # just have to do some magic with set_path to make it so
4605 # we only want a partial path.
4606 my $sp = '';
4607 my $final = join('/', @pc);
4608 while (@pc) {
4609 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4610 $sp .= '/' if length $sp;
4611 $sp .= shift @pc;
4612 }
4613 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4614
4615 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4616
4617 $reporter->finish_report($pool);
4618 $pool->clear;
4619 $editor->{git_commit_ok};
4620 }
4621
4622 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4623 # svn_ra_reparent didn't work before 1.4)
4624 sub gs_do_switch {
4625 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4626 my $path = $gs->{path};
4627 my $pool = SVN::Pool->new;
4628
4629 my $full_url = $self->{url};
4630 my $old_url = $full_url;
4631 $full_url .= '/' . $path if length $path;
4632 my ($ra, $reparented);
4633
4634 if ($old_url =~ m#^svn(\+ssh)?://# ||
4635 ($full_url =~ m#^https?://# &&
4636 escape_url($full_url) ne $full_url)) {
4637 $_[0] = undef;
4638 $self = undef;
4639 $RA = undef;
4640 $ra = Git::SVN::Ra->new($full_url);
4641 $ra_invalid = 1;
4642 } elsif ($old_url ne $full_url) {
4643 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4644 $self->{url} = $full_url;
4645 $reparented = 1;
4646 }
4647
4648 $ra ||= $self;
4649 $url_b = escape_url($url_b);
4650 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4651 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4652 $reporter->set_path('', $rev_a, 0, @lock, $pool);
4653 $reporter->finish_report($pool);
4654
4655 if ($reparented) {
4656 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4657 $self->{url} = $old_url;
4658 }
4659
4660 $pool->clear;
4661 $editor->{git_commit_ok};
4662 }
4663
4664 sub longest_common_path {
4665 my ($gsv, $globs) = @_;
4666 my %common;
4667 my $common_max = scalar @$gsv;
4668
4669 foreach my $gs (@$gsv) {
4670 my @tmp = split m#/#, $gs->{path};
4671 my $p = '';
4672 foreach (@tmp) {
4673 $p .= length($p) ? "/$_" : $_;
4674 $common{$p} ||= 0;
4675 $common{$p}++;
4676 }
4677 }
4678 $globs ||= [];
4679 $common_max += scalar @$globs;
4680 foreach my $glob (@$globs) {
4681 my @tmp = split m#/#, $glob->{path}->{left};
4682 my $p = '';
4683 foreach (@tmp) {
4684 $p .= length($p) ? "/$_" : $_;
4685 $common{$p} ||= 0;
4686 $common{$p}++;
4687 }
4688 }
4689
4690 my $longest_path = '';
4691 foreach (sort {length $b <=> length $a} keys %common) {
4692 if ($common{$_} == $common_max) {
4693 $longest_path = $_;
4694 last;
4695 }
4696 }
4697 $longest_path;
4698 }
4699
4700 sub gs_fetch_loop_common {
4701 my ($self, $base, $head, $gsv, $globs) = @_;
4702 return if ($base > $head);
4703 my $inc = $_log_window_size;
4704 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4705 my $longest_path = longest_common_path($gsv, $globs);
4706 my $ra_url = $self->{url};
4707 my $find_trailing_edge;
4708 while (1) {
4709 my %revs;
4710 my $err;
4711 my $err_handler = $SVN::Error::handler;
4712 $SVN::Error::handler = sub {
4713 ($err) = @_;
4714 skip_unknown_revs($err);
4715 };
4716 sub _cb {
4717 my ($paths, $r, $author, $date, $log) = @_;
4718 [ $paths,
4719 { author => $author, date => $date, log => $log } ];
4720 }
4721 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4722 sub { $revs{$_[1]} = _cb(@_) });
4723 if ($err) {
4724 print "Checked through r$max\r";
4725 } else {
4726 $find_trailing_edge = 1;
4727 }
4728 if ($err and $find_trailing_edge) {
4729 print STDERR "Path '$longest_path' ",
4730 "was probably deleted:\n",
4731 $err->expanded_message,
4732 "\nWill attempt to follow ",
4733 "revisions r$min .. r$max ",
4734 "committed before the deletion\n";
4735 my $hi = $max;
4736 while (--$hi >= $min) {
4737 my $ok;
4738 $self->get_log([$longest_path], $min, $hi,
4739 0, 1, 1, sub {
4740 $ok = $_[1];
4741 $revs{$_[1]} = _cb(@_) });
4742 if ($ok) {
4743 print STDERR "r$min .. r$ok OK\n";
4744 last;
4745 }
4746 }
4747 $find_trailing_edge = 0;
4748 }
4749 $SVN::Error::handler = $err_handler;
4750
4751 my %exists = map { $_->{path} => $_ } @$gsv;
4752 foreach my $r (sort {$a <=> $b} keys %revs) {
4753 my ($paths, $logged) = @{$revs{$r}};
4754
4755 foreach my $gs ($self->match_globs(\%exists, $paths,
4756 $globs, $r)) {
4757 if ($gs->rev_map_max >= $r) {
4758 next;
4759 }
4760 next unless $gs->match_paths($paths, $r);
4761 $gs->{logged_rev_props} = $logged;
4762 if (my $last_commit = $gs->last_commit) {
4763 $gs->assert_index_clean($last_commit);
4764 }
4765 my $log_entry = $gs->do_fetch($paths, $r);
4766 if ($log_entry) {
4767 $gs->do_git_commit($log_entry);
4768 }
4769 $INDEX_FILES{$gs->{index}} = 1;
4770 }
4771 foreach my $g (@$globs) {
4772 my $k = "svn-remote.$g->{remote}." .
4773 "$g->{t}-maxRev";
4774 Git::SVN::tmp_config($k, $r);
4775 }
4776 if ($ra_invalid) {
4777 $_[0] = undef;
4778 $self = undef;
4779 $RA = undef;
4780 $self = Git::SVN::Ra->new($ra_url);
4781 $ra_invalid = undef;
4782 }
4783 }
4784 # pre-fill the .rev_db since it'll eventually get filled in
4785 # with '0' x40 if something new gets committed
4786 foreach my $gs (@$gsv) {
4787 next if $gs->rev_map_max >= $max;
4788 next if defined $gs->rev_map_get($max);
4789 $gs->rev_map_set($max, 0 x40);
4790 }
4791 foreach my $g (@$globs) {
4792 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4793 Git::SVN::tmp_config($k, $max);
4794 }
4795 last if $max >= $head;
4796 $min = $max + 1;
4797 $max += $inc;
4798 $max = $head if ($max > $head);
4799 }
4800 Git::SVN::gc();
4801 }
4802
4803 sub get_dir_globbed {
4804 my ($self, $left, $depth, $r) = @_;
4805
4806 my @x = eval { $self->get_dir($left, $r) };
4807 return unless scalar @x == 3;
4808 my $dirents = $x[0];
4809 my @finalents;
4810 foreach my $de (keys %$dirents) {
4811 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4812 if ($depth > 1) {
4813 my @args = ("$left/$de", $depth - 1, $r);
4814 foreach my $dir ($self->get_dir_globbed(@args)) {
4815 push @finalents, "$de/$dir";
4816 }
4817 } else {
4818 push @finalents, $de;
4819 }
4820 }
4821 @finalents;
4822 }
4823
4824 sub match_globs {
4825 my ($self, $exists, $paths, $globs, $r) = @_;
4826
4827 sub get_dir_check {
4828 my ($self, $exists, $g, $r) = @_;
4829
4830 my @dirs = $self->get_dir_globbed($g->{path}->{left},
4831 $g->{path}->{depth},
4832 $r);
4833
4834 foreach my $de (@dirs) {
4835 my $p = $g->{path}->full_path($de);
4836 next if $exists->{$p};
4837 next if (length $g->{path}->{right} &&
4838 ($self->check_path($p, $r) !=
4839 $SVN::Node::dir));
4840 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4841 $g->{ref}->full_path($de), 1);
4842 }
4843 }
4844 foreach my $g (@$globs) {
4845 if (my $path = $paths->{"/$g->{path}->{left}"}) {
4846 if ($path->{action} =~ /^[AR]$/) {
4847 get_dir_check($self, $exists, $g, $r);
4848 }
4849 }
4850 foreach (keys %$paths) {
4851 if (/$g->{path}->{left_regex}/ &&
4852 !/$g->{path}->{regex}/) {
4853 next if $paths->{$_}->{action} !~ /^[AR]$/;
4854 get_dir_check($self, $exists, $g, $r);
4855 }
4856 next unless /$g->{path}->{regex}/;
4857 my $p = $1;
4858 my $pathname = $g->{path}->full_path($p);
4859 next if $exists->{$pathname};
4860 next if ($self->check_path($pathname, $r) !=
4861 $SVN::Node::dir);
4862 $exists->{$pathname} = Git::SVN->init(
4863 $self->{url}, $pathname, undef,
4864 $g->{ref}->full_path($p), 1);
4865 }
4866 my $c = '';
4867 foreach (split m#/#, $g->{path}->{left}) {
4868 $c .= "/$_";
4869 next unless ($paths->{$c} &&
4870 ($paths->{$c}->{action} =~ /^[AR]$/));
4871 get_dir_check($self, $exists, $g, $r);
4872 }
4873 }
4874 values %$exists;
4875 }
4876
4877 sub minimize_url {
4878 my ($self) = @_;
4879 return $self->{url} if ($self->{url} eq $self->{repos_root});
4880 my $url = $self->{repos_root};
4881 my @components = split(m!/!, $self->{svn_path});
4882 my $c = '';
4883 do {
4884 $url .= "/$c" if length $c;
4885 eval {
4886 my $ra = (ref $self)->new($url);
4887 my $latest = $ra->get_latest_revnum;
4888 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
4889 };
4890 } while ($@ && ($c = shift @components));
4891 $url;
4892 }
4893
4894 sub can_do_switch {
4895 my $self = shift;
4896 unless (defined $can_do_switch) {
4897 my $pool = SVN::Pool->new;
4898 my $rep = eval {
4899 $self->do_switch(1, '', 0, $self->{url},
4900 SVN::Delta::Editor->new, $pool);
4901 };
4902 if ($@) {
4903 $can_do_switch = 0;
4904 } else {
4905 $rep->abort_report($pool);
4906 $can_do_switch = 1;
4907 }
4908 $pool->clear;
4909 }
4910 $can_do_switch;
4911 }
4912
4913 sub skip_unknown_revs {
4914 my ($err) = @_;
4915 my $errno = $err->apr_err();
4916 # Maybe the branch we're tracking didn't
4917 # exist when the repo started, so it's
4918 # not an error if it doesn't, just continue
4919 #
4920 # Wonderfully consistent library, eh?
4921 # 160013 - svn:// and file://
4922 # 175002 - http(s)://
4923 # 175007 - http(s):// (this repo required authorization, too...)
4924 # More codes may be discovered later...
4925 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4926 my $err_key = $err->expanded_message;
4927 # revision numbers change every time, filter them out
4928 $err_key =~ s/\d+/\0/g;
4929 $err_key = "$errno\0$err_key";
4930 unless ($ignored_err{$err_key}) {
4931 warn "W: Ignoring error from SVN, path probably ",
4932 "does not exist: ($errno): ",
4933 $err->expanded_message,"\n";
4934 warn "W: Do not be alarmed at the above message ",
4935 "git-svn is just searching aggressively for ",
4936 "old history.\n",
4937 "This may take a while on large repositories\n";
4938 $ignored_err{$err_key} = 1;
4939 }
4940 return;
4941 }
4942 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4943 }
4944
4945 package Git::SVN::Log;
4946 use strict;
4947 use warnings;
4948 use POSIX qw/strftime/;
4949 use Time::Local;
4950 use constant commit_log_separator => ('-' x 72) . "\n";
4951 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4952 %rusers $show_commit $incremental/;
4953 my $l_fmt;
4954
4955 sub cmt_showable {
4956 my ($c) = @_;
4957 return 1 if defined $c->{r};
4958
4959 # big commit message got truncated by the 16k pretty buffer in rev-list
4960 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4961 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4962 @{$c->{l}} = ();
4963 my @log = command(qw/cat-file commit/, $c->{c});
4964
4965 # shift off the headers
4966 shift @log while ($log[0] ne '');
4967 shift @log;
4968
4969 # TODO: make $c->{l} not have a trailing newline in the future
4970 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4971
4972 (undef, $c->{r}, undef) = ::extract_metadata(
4973 (grep(/^git-svn-id: /, @log))[-1]);
4974 }
4975 return defined $c->{r};
4976 }
4977
4978 sub log_use_color {
4979 return $color || Git->repository->get_colorbool('color.diff');
4980 }
4981
4982 sub git_svn_log_cmd {
4983 my ($r_min, $r_max, @args) = @_;
4984 my $head = 'HEAD';
4985 my (@files, @log_opts);
4986 foreach my $x (@args) {
4987 if ($x eq '--' || @files) {
4988 push @files, $x;
4989 } else {
4990 if (::verify_ref("$x^0")) {
4991 $head = $x;
4992 } else {
4993 push @log_opts, $x;
4994 }
4995 }
4996 }
4997
4998 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4999 $gs ||= Git::SVN->_new;
5000 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5001 $gs->refname);
5002 push @cmd, '-r' unless $non_recursive;
5003 push @cmd, qw/--raw --name-status/ if $verbose;
5004 push @cmd, '--color' if log_use_color();
5005 push @cmd, @log_opts;
5006 if (defined $r_max && $r_max == $r_min) {
5007 push @cmd, '--max-count=1';
5008 if (my $c = $gs->rev_map_get($r_max)) {
5009 push @cmd, $c;
5010 }
5011 } elsif (defined $r_max) {
5012 if ($r_max < $r_min) {
5013 ($r_min, $r_max) = ($r_max, $r_min);
5014 }
5015 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5016 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5017 # If there are no commits in the range, both $c_max and $c_min
5018 # will be undefined. If there is at least 1 commit in the
5019 # range, both will be defined.
5020 return () if !defined $c_min || !defined $c_max;
5021 if ($c_min eq $c_max) {
5022 push @cmd, '--max-count=1', $c_min;
5023 } else {
5024 push @cmd, '--boundary', "$c_min..$c_max";
5025 }
5026 }
5027 return (@cmd, @files);
5028 }
5029
5030 # adapted from pager.c
5031 sub config_pager {
5032 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
5033 if (!defined $pager) {
5034 $pager = 'less';
5035 } elsif (length $pager == 0 || $pager eq 'cat') {
5036 $pager = undef;
5037 }
5038 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5039 }
5040
5041 sub run_pager {
5042 return unless -t *STDOUT && defined $pager;
5043 pipe my ($rfd, $wfd) or return;
5044 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5045 if (!$pid) {
5046 open STDOUT, '>&', $wfd or
5047 ::fatal "Can't redirect to stdout: $!";
5048 return;
5049 }
5050 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5051 $ENV{LESS} ||= 'FRSX';
5052 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5053 }
5054
5055 sub format_svn_date {
5056 # some systmes don't handle or mishandle %z, so be creative.
5057 my $t = shift || time;
5058 my $gm = timelocal(gmtime($t));
5059 my $sign = qw( + + - )[ $t <=> $gm ];
5060 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5061 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5062 }
5063
5064 sub parse_git_date {
5065 my ($t, $tz) = @_;
5066 # Date::Parse isn't in the standard Perl distro :(
5067 if ($tz =~ s/^\+//) {
5068 $t += tz_to_s_offset($tz);
5069 } elsif ($tz =~ s/^\-//) {
5070 $t -= tz_to_s_offset($tz);
5071 }
5072 return $t;
5073 }
5074
5075 sub set_local_timezone {
5076 if (defined $TZ) {
5077 $ENV{TZ} = $TZ;
5078 } else {
5079 delete $ENV{TZ};
5080 }
5081 }
5082
5083 sub tz_to_s_offset {
5084 my ($tz) = @_;
5085 $tz =~ s/(\d\d)$//;
5086 return ($1 * 60) + ($tz * 3600);
5087 }
5088
5089 sub get_author_info {
5090 my ($dest, $author, $t, $tz) = @_;
5091 $author =~ s/(?:^\s*|\s*$)//g;
5092 $dest->{a_raw} = $author;
5093 my $au;
5094 if ($::_authors) {
5095 $au = $rusers{$author} || undef;
5096 }
5097 if (!$au) {
5098 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5099 }
5100 $dest->{t} = $t;
5101 $dest->{tz} = $tz;
5102 $dest->{a} = $au;
5103 $dest->{t_utc} = parse_git_date($t, $tz);
5104 }
5105
5106 sub process_commit {
5107 my ($c, $r_min, $r_max, $defer) = @_;
5108 if (defined $r_min && defined $r_max) {
5109 if ($r_min == $c->{r} && $r_min == $r_max) {
5110 show_commit($c);
5111 return 0;
5112 }
5113 return 1 if $r_min == $r_max;
5114 if ($r_min < $r_max) {
5115 # we need to reverse the print order
5116 return 0 if (defined $limit && --$limit < 0);
5117 push @$defer, $c;
5118 return 1;
5119 }
5120 if ($r_min != $r_max) {
5121 return 1 if ($r_min < $c->{r});
5122 return 1 if ($r_max > $c->{r});
5123 }
5124 }
5125 return 0 if (defined $limit && --$limit < 0);
5126 show_commit($c);
5127 return 1;
5128 }
5129
5130 sub show_commit {
5131 my $c = shift;
5132 if ($oneline) {
5133 my $x = "\n";
5134 if (my $l = $c->{l}) {
5135 while ($l->[0] =~ /^\s*$/) { shift @$l }
5136 $x = $l->[0];
5137 }
5138 $l_fmt ||= 'A' . length($c->{r});
5139 print 'r',pack($l_fmt, $c->{r}),' | ';
5140 print "$c->{c} | " if $show_commit;
5141 print $x;
5142 } else {
5143 show_commit_normal($c);
5144 }
5145 }
5146
5147 sub show_commit_changed_paths {
5148 my ($c) = @_;
5149 return unless $c->{changed};
5150 print "Changed paths:\n", @{$c->{changed}};
5151 }
5152
5153 sub show_commit_normal {
5154 my ($c) = @_;
5155 print commit_log_separator, "r$c->{r} | ";
5156 print "$c->{c} | " if $show_commit;
5157 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5158 my $nr_line = 0;
5159
5160 if (my $l = $c->{l}) {
5161 while ($l->[$#$l] eq "\n" && $#$l > 0
5162 && $l->[($#$l - 1)] eq "\n") {
5163 pop @$l;
5164 }
5165 $nr_line = scalar @$l;
5166 if (!$nr_line) {
5167 print "1 line\n\n\n";
5168 } else {
5169 if ($nr_line == 1) {
5170 $nr_line = '1 line';
5171 } else {
5172 $nr_line .= ' lines';
5173 }
5174 print $nr_line, "\n";
5175 show_commit_changed_paths($c);
5176 print "\n";
5177 print $_ foreach @$l;
5178 }
5179 } else {
5180 print "1 line\n";
5181 show_commit_changed_paths($c);
5182 print "\n";
5183
5184 }
5185 foreach my $x (qw/raw stat diff/) {
5186 if ($c->{$x}) {
5187 print "\n";
5188 print $_ foreach @{$c->{$x}}
5189 }
5190 }
5191 }
5192
5193 sub cmd_show_log {
5194 my (@args) = @_;
5195 my ($r_min, $r_max);
5196 my $r_last = -1; # prevent dupes
5197 set_local_timezone();
5198 if (defined $::_revision) {
5199 if ($::_revision =~ /^(\d+):(\d+)$/) {
5200 ($r_min, $r_max) = ($1, $2);
5201 } elsif ($::_revision =~ /^\d+$/) {
5202 $r_min = $r_max = $::_revision;
5203 } else {
5204 ::fatal "-r$::_revision is not supported, use ",
5205 "standard 'git log' arguments instead";
5206 }
5207 }
5208
5209 config_pager();
5210 @args = git_svn_log_cmd($r_min, $r_max, @args);
5211 if (!@args) {
5212 print commit_log_separator unless $incremental || $oneline;
5213 return;
5214 }
5215 my $log = command_output_pipe(@args);
5216 run_pager();
5217 my (@k, $c, $d, $stat);
5218 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5219 while (<$log>) {
5220 if (/^${esc_color}commit -?($::sha1_short)/o) {
5221 my $cmt = $1;
5222 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5223 $r_last = $c->{r};
5224 process_commit($c, $r_min, $r_max, \@k) or
5225 goto out;
5226 }
5227 $d = undef;
5228 $c = { c => $cmt };
5229 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5230 get_author_info($c, $1, $2, $3);
5231 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5232 # ignore
5233 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5234 push @{$c->{raw}}, $_;
5235 } elsif (/^${esc_color}[ACRMDT]\t/) {
5236 # we could add $SVN->{svn_path} here, but that requires
5237 # remote access at the moment (repo_path_split)...
5238 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5239 push @{$c->{changed}}, $_;
5240 } elsif (/^${esc_color}diff /o) {
5241 $d = 1;
5242 push @{$c->{diff}}, $_;
5243 } elsif ($d) {
5244 push @{$c->{diff}}, $_;
5245 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5246 $esc_color*[\+\-]*$esc_color$/x) {
5247 $stat = 1;
5248 push @{$c->{stat}}, $_;
5249 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5250 push @{$c->{stat}}, $_;
5251 $stat = undef;
5252 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5253 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5254 } elsif (s/^${esc_color} //o) {
5255 push @{$c->{l}}, $_;
5256 }
5257 }
5258 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5259 $r_last = $c->{r};
5260 process_commit($c, $r_min, $r_max, \@k);
5261 }
5262 if (@k) {
5263 ($r_min, $r_max) = ($r_max, $r_min);
5264 process_commit($_, $r_min, $r_max) foreach reverse @k;
5265 }
5266 out:
5267 close $log;
5268 print commit_log_separator unless $incremental || $oneline;
5269 }
5270
5271 sub cmd_blame {
5272 my $path = pop;
5273
5274 config_pager();
5275 run_pager();
5276
5277 my ($fh, $ctx, $rev);
5278
5279 if ($_git_format) {
5280 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5281 while (my $line = <$fh>) {
5282 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5283 # Uncommitted edits show up as a rev ID of
5284 # all zeros, which we can't look up with
5285 # cmt_metadata
5286 if ($1 !~ /^0+$/) {
5287 (undef, $rev, undef) =
5288 ::cmt_metadata($1);
5289 $rev = '0' if (!$rev);
5290 } else {
5291 $rev = '0';
5292 }
5293 $rev = sprintf('%-10s', $rev);
5294 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5295 }
5296 print $line;
5297 }
5298 } else {
5299 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5300 '--', $path);
5301 my ($sha1);
5302 my %authors;
5303 my @buffer;
5304 my %dsha; #distinct sha keys
5305
5306 while (my $line = <$fh>) {
5307 push @buffer, $line;
5308 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5309 $dsha{$1} = 1;
5310 }
5311 }
5312
5313 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5314
5315 foreach my $line (@buffer) {
5316 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5317 $rev = $s2r->{$1};
5318 $rev = '0' if (!$rev)
5319 }
5320 elsif ($line =~ /^author (.*)/) {
5321 $authors{$rev} = $1;
5322 $authors{$rev} =~ s/\s/_/g;
5323 }
5324 elsif ($line =~ /^\t(.*)$/) {
5325 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5326 }
5327 }
5328 }
5329 command_close_pipe($fh, $ctx);
5330 }
5331
5332 package Git::SVN::Migration;
5333 # these version numbers do NOT correspond to actual version numbers
5334 # of git nor git-svn. They are just relative.
5335 #
5336 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5337 #
5338 # v1 layout: .git/$id/info/url, refs/remotes/$id
5339 #
5340 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5341 #
5342 # v3 layout: .git/svn/$id, refs/remotes/$id
5343 # - info/url may remain for backwards compatibility
5344 # - this is what we migrate up to this layout automatically,
5345 # - this will be used by git svn init on single branches
5346 # v3.1 layout (auto migrated):
5347 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5348 # for backwards compatibility
5349 #
5350 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5351 # - this is only created for newly multi-init-ed
5352 # repositories. Similar in spirit to the
5353 # --use-separate-remotes option in git-clone (now default)
5354 # - we do not automatically migrate to this (following
5355 # the example set by core git)
5356 #
5357 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
5358 # - newer, more-efficient format that uses 24-bytes per record
5359 # with no filler space.
5360 # - use xxd -c24 < .rev_map.$UUID to view and debug
5361 # - This is a one-way migration, repositories updated to the
5362 # new format will not be able to use old git-svn without
5363 # rebuilding the .rev_db. Rebuilding the rev_db is not
5364 # possible if noMetadata or useSvmProps are set; but should
5365 # be no problem for users that use the (sensible) defaults.
5366 use strict;
5367 use warnings;
5368 use Carp qw/croak/;
5369 use File::Path qw/mkpath/;
5370 use File::Basename qw/dirname basename/;
5371 use vars qw/$_minimize/;
5372
5373 sub migrate_from_v0 {
5374 my $git_dir = $ENV{GIT_DIR};
5375 return undef unless -d $git_dir;
5376 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5377 my $migrated = 0;
5378 while (<$fh>) {
5379 chomp;
5380 my ($id, $orig_ref) = ($_, $_);
5381 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5382 next unless -f "$git_dir/$id/info/url";
5383 my $new_ref = "refs/remotes/$id";
5384 if (::verify_ref("$new_ref^0")) {
5385 print STDERR "W: $orig_ref is probably an old ",
5386 "branch used by an ancient version of ",
5387 "git-svn.\n",
5388 "However, $new_ref also exists.\n",
5389 "We will not be able ",
5390 "to use this branch until this ",
5391 "ambiguity is resolved.\n";
5392 next;
5393 }
5394 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5395 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5396 command_noisy('update-ref', $new_ref, $orig_ref);
5397 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5398 $migrated++;
5399 }
5400 command_close_pipe($fh, $ctx);
5401 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5402 $migrated;
5403 }
5404
5405 sub migrate_from_v1 {
5406 my $git_dir = $ENV{GIT_DIR};
5407 my $migrated = 0;
5408 return $migrated unless -d $git_dir;
5409 my $svn_dir = "$git_dir/svn";
5410
5411 # just in case somebody used 'svn' as their $id at some point...
5412 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5413
5414 print STDERR "Migrating from a git-svn v1 layout...\n";
5415 mkpath([$svn_dir]);
5416 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5417 "$svn_dir\n\t(required for this version ",
5418 "($::VERSION) of git-svn) does not exist.\n";
5419 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5420 while (<$fh>) {
5421 my $x = $_;
5422 next unless $x =~ s#^refs/remotes/##;
5423 chomp $x;
5424 next unless -f "$git_dir/$x/info/url";
5425 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5426 next unless $u;
5427 my $dn = dirname("$git_dir/svn/$x");
5428 mkpath([$dn]) unless -d $dn;
5429 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5430 mkpath(["$git_dir/svn/svn"]);
5431 print STDERR " - $git_dir/$x/info => ",
5432 "$git_dir/svn/$x/info\n";
5433 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5434 croak "$!: $x";
5435 # don't worry too much about these, they probably
5436 # don't exist with repos this old (save for index,
5437 # and we can easily regenerate that)
5438 foreach my $f (qw/unhandled.log index .rev_db/) {
5439 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5440 }
5441 } else {
5442 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5443 rename "$git_dir/$x", "$git_dir/svn/$x" or
5444 croak "$!: $x";
5445 }
5446 $migrated++;
5447 }
5448 command_close_pipe($fh, $ctx);
5449 print STDERR "Done migrating from a git-svn v1 layout\n";
5450 $migrated;
5451 }
5452
5453 sub read_old_urls {
5454 my ($l_map, $pfx, $path) = @_;
5455 my @dir;
5456 foreach (<$path/*>) {
5457 if (-r "$_/info/url") {
5458 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5459 my $ref_id = $pfx . basename $_;
5460 my $url = ::file_to_s("$_/info/url");
5461 $l_map->{$ref_id} = $url;
5462 } elsif (-d $_) {
5463 push @dir, $_;
5464 }
5465 }
5466 foreach (@dir) {
5467 my $x = $_;
5468 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5469 read_old_urls($l_map, $x, $_);
5470 }
5471 }
5472
5473 sub migrate_from_v2 {
5474 my @cfg = command(qw/config -l/);
5475 return if grep /^svn-remote\..+\.url=/, @cfg;
5476 my %l_map;
5477 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5478 my $migrated = 0;
5479
5480 foreach my $ref_id (sort keys %l_map) {
5481 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5482 if ($@) {
5483 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5484 }
5485 $migrated++;
5486 }
5487 $migrated;
5488 }
5489
5490 sub minimize_connections {
5491 my $r = Git::SVN::read_all_remotes();
5492 my $new_urls = {};
5493 my $root_repos = {};
5494 foreach my $repo_id (keys %$r) {
5495 my $url = $r->{$repo_id}->{url} or next;
5496 my $fetch = $r->{$repo_id}->{fetch} or next;
5497 my $ra = Git::SVN::Ra->new($url);
5498
5499 # skip existing cases where we already connect to the root
5500 if (($ra->{url} eq $ra->{repos_root}) ||
5501 ($ra->{repos_root} eq $repo_id)) {
5502 $root_repos->{$ra->{url}} = $repo_id;
5503 next;
5504 }
5505
5506 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5507 my $root_path = $ra->{url};
5508 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5509 foreach my $path (keys %$fetch) {
5510 my $ref_id = $fetch->{$path};
5511 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5512
5513 # make sure we can read when connecting to
5514 # a higher level of a repository
5515 my ($last_rev, undef) = $gs->last_rev_commit;
5516 if (!defined $last_rev) {
5517 $last_rev = eval {
5518 $root_ra->get_latest_revnum;
5519 };
5520 next if $@;
5521 }
5522 my $new = $root_path;
5523 $new .= length $path ? "/$path" : '';
5524 eval {
5525 $root_ra->get_log([$new], $last_rev, $last_rev,
5526 0, 0, 1, sub { });
5527 };
5528 next if $@;
5529 $new_urls->{$ra->{repos_root}}->{$new} =
5530 { ref_id => $ref_id,
5531 old_repo_id => $repo_id,
5532 old_path => $path };
5533 }
5534 }
5535
5536 my @emptied;
5537 foreach my $url (keys %$new_urls) {
5538 # see if we can re-use an existing [svn-remote "repo_id"]
5539 # instead of creating a(n ugly) new section:
5540 my $repo_id = $root_repos->{$url} || $url;
5541
5542 my $fetch = $new_urls->{$url};
5543 foreach my $path (keys %$fetch) {
5544 my $x = $fetch->{$path};
5545 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5546 my $pfx = "svn-remote.$x->{old_repo_id}";
5547
5548 my $old_fetch = quotemeta("$x->{old_path}:".
5549 "$x->{ref_id}");
5550 command_noisy(qw/config --unset/,
5551 "$pfx.fetch", '^'. $old_fetch . '$');
5552 delete $r->{$x->{old_repo_id}}->
5553 {fetch}->{$x->{old_path}};
5554 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5555 command_noisy(qw/config --unset/,
5556 "$pfx.url");
5557 push @emptied, $x->{old_repo_id}
5558 }
5559 }
5560 }
5561 if (@emptied) {
5562 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5563 print STDERR <<EOF;
5564 The following [svn-remote] sections in your config file ($file) are empty
5565 and can be safely removed:
5566 EOF
5567 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5568 }
5569 }
5570
5571 sub migration_check {
5572 migrate_from_v0();
5573 migrate_from_v1();
5574 migrate_from_v2();
5575 minimize_connections() if $_minimize;
5576 }
5577
5578 package Git::IndexInfo;
5579 use strict;
5580 use warnings;
5581 use Git qw/command_input_pipe command_close_pipe/;
5582
5583 sub new {
5584 my ($class) = @_;
5585 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5586 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5587 }
5588
5589 sub remove {
5590 my ($self, $path) = @_;
5591 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5592 return ++$self->{nr};
5593 }
5594 undef;
5595 }
5596
5597 sub update {
5598 my ($self, $mode, $hash, $path) = @_;
5599 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5600 return ++$self->{nr};
5601 }
5602 undef;
5603 }
5604
5605 sub DESTROY {
5606 my ($self) = @_;
5607 command_close_pipe($self->{gui}, $self->{ctx});
5608 }
5609
5610 package Git::SVN::GlobSpec;
5611 use strict;
5612 use warnings;
5613
5614 sub new {
5615 my ($class, $glob) = @_;
5616 my $re = $glob;
5617 $re =~ s!/+$!!g; # no need for trailing slashes
5618 $re =~ m!^([^*]*)(\*(?:/\*)*)(.*)$!;
5619 my $temp = $re;
5620 my ($left, $right) = ($1, $3);
5621 $re = $2;
5622 my $depth = $re =~ tr/*/*/;
5623 if ($depth != $temp =~ tr/*/*/) {
5624 die "Only one set of wildcard directories " .
5625 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5626 }
5627 if ($depth == 0) {
5628 die "One '*' is needed for glob: '$glob'\n";
5629 }
5630 $re =~ s!\*!\[^/\]*!g;
5631 $re = quotemeta($left) . "($re)" . quotemeta($right);
5632 if (length $left && !($left =~ s!/+$!!g)) {
5633 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5634 }
5635 if (length $right && !($right =~ s!^/+!!g)) {
5636 die "Missing leading '/' on right side of: '$glob' ($right)\n";
5637 }
5638 my $left_re = qr/^\/\Q$left\E(\/|$)/;
5639 bless { left => $left, right => $right, left_regex => $left_re,
5640 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5641 }
5642
5643 sub full_path {
5644 my ($self, $path) = @_;
5645 return (length $self->{left} ? "$self->{left}/" : '') .
5646 $path . (length $self->{right} ? "/$self->{right}" : '');
5647 }
5648
5649 __END__
5650
5651 Data structures:
5652
5653
5654 $remotes = { # returned by read_all_remotes()
5655 'svn' => {
5656 # svn-remote.svn.url=https://svn.musicpd.org
5657 url => 'https://svn.musicpd.org',
5658 # svn-remote.svn.fetch=mpd/trunk:trunk
5659 fetch => {
5660 'mpd/trunk' => 'trunk',
5661 },
5662 # svn-remote.svn.tags=mpd/tags/*:tags/*
5663 tags => {
5664 path => {
5665 left => 'mpd/tags',
5666 right => '',
5667 regex => qr!mpd/tags/([^/]+)$!,
5668 glob => 'tags/*',
5669 },
5670 ref => {
5671 left => 'tags',
5672 right => '',
5673 regex => qr!tags/([^/]+)$!,
5674 glob => 'tags/*',
5675 },
5676 }
5677 }
5678 };
5679
5680 $log_entry hashref as returned by libsvn_log_entry()
5681 {
5682 log => 'whitespace-formatted log entry
5683 ', # trailing newline is preserved
5684 revision => '8', # integer
5685 date => '2004-02-24T17:01:44.108345Z', # commit date
5686 author => 'committer name'
5687 };
5688
5689
5690 # this is generated by generate_diff();
5691 @mods = array of diff-index line hashes, each element represents one line
5692 of diff-index output
5693
5694 diff-index line ($m hash)
5695 {
5696 mode_a => first column of diff-index output, no leading ':',
5697 mode_b => second column of diff-index output,
5698 sha1_b => sha1sum of the final blob,
5699 chg => change type [MCRADT],
5700 file_a => original file name of a file (iff chg is 'C' or 'R')
5701 file_b => new/current file name of a file (any chg)
5702 }
5703 ;
5704
5705 # retval of read_url_paths{,_all}();
5706 $l_map = {
5707 # repository root url
5708 'https://svn.musicpd.org' => {
5709 # repository path # GIT_SVN_ID
5710 'mpd/trunk' => 'trunk',
5711 'mpd/tags/0.11.5' => 'tags/0.11.5',
5712 },
5713 }
5714
5715 Notes:
5716 I don't trust the each() function on unless I created %hash myself
5717 because the internal iterator may not have started at base.