?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'bw/no-python-autoconf'
[git/jrn.git] / git-svn.perl
blob4f7dc2b73b75dfed0aac5860c77117559cdd3bcd
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 $ENV{SVN_SSH} =~ s/(.*)/"$1"/;
30 }
31 }
32 }
33
34 $Git::SVN::Log::TZ = $ENV{TZ};
35 $ENV{TZ} = 'UTC';
36 $| = 1; # unbuffer STDOUT
37
38 sub fatal (@) { print STDERR "@_\n"; exit 1 }
39 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
40 require SVN::Ra;
41 require SVN::Delta;
42 if ($SVN::Core::VERSION lt '1.1.0') {
43 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
44 }
45 my $can_compress = eval { require Compress::Zlib; 1};
46 push @Git::SVN::Ra::ISA, 'SVN::Ra';
47 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
48 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
49 use Carp qw/croak/;
50 use Digest::MD5;
51 use IO::File qw//;
52 use File::Basename qw/dirname basename/;
53 use File::Path qw/mkpath/;
54 use File::Spec;
55 use File::Find;
56 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
57 use IPC::Open3;
58 use Git;
59
60 BEGIN {
61 # import functions from Git into our packages, en masse
62 no strict 'refs';
63 foreach (qw/command command_oneline command_noisy command_output_pipe
64 command_input_pipe command_close_pipe
65 command_bidi_pipe command_close_bidi_pipe/) {
66 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
67 Git::SVN::Migration Git::SVN::Log Git::SVN),
68 __PACKAGE__) {
69 *{"${package}::$_"} = \&{"Git::$_"};
70 }
71 }
72 }
73
74 my ($SVN);
75
76 $sha1 = qr/[a-f\d]{40}/;
77 $sha1_short = qr/[a-f\d]{4,40}/;
78 my ($_stdin, $_help, $_edit,
79 $_message, $_file, $_branch_dest,
80 $_template, $_shared,
81 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
82 $_merge, $_strategy, $_dry_run, $_local,
83 $_prefix, $_no_checkout, $_url, $_verbose,
84 $_git_format, $_commit_url, $_tag);
85 $Git::SVN::_follow_parent = 1;
86 $_q ||= 0;
87 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
88 'config-dir=s' => \$Git::SVN::Ra::config_dir,
89 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
90 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
91 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
92 'authors-file|A=s' => \$_authors,
93 'authors-prog=s' => \$_authors_prog,
94 'repack:i' => \$Git::SVN::_repack,
95 'noMetadata' => \$Git::SVN::_no_metadata,
96 'useSvmProps' => \$Git::SVN::_use_svm_props,
97 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
98 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
99 'no-checkout' => \$_no_checkout,
100 'quiet|q+' => \$_q,
101 'repack-flags|repack-args|repack-opts=s' =>
102 \$Git::SVN::_repack_flags,
103 'use-log-author' => \$Git::SVN::_use_log_author,
104 'add-author-from' => \$Git::SVN::_add_author_from,
105 'localtime' => \$Git::SVN::_localtime,
106 %remote_opts );
107
108 my ($_trunk, @_tags, @_branches, $_stdlayout);
109 my %icv;
110 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
111 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
112 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
113 'stdlayout|s' => \$_stdlayout,
114 'minimize-url|m!' => \$Git::SVN::_minimize_url,
115 'no-metadata' => sub { $icv{noMetadata} = 1 },
116 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
117 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
118 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
119 'rewrite-uuid=s' => sub { $icv{rewriteUUID} = $_[1] },
120 %remote_opts );
121 my %cmt_opts = ( 'edit|e' => \$_edit,
122 'rmdir' => \$SVN::Git::Editor::_rmdir,
123 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
124 'l=i' => \$SVN::Git::Editor::_rename_limit,
125 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
126 );
127
128 my %cmd = (
129 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
130 { 'revision|r=s' => \$_revision,
131 'fetch-all|all' => \$_fetch_all,
132 'parent|p' => \$_fetch_parent,
133 %fc_opts } ],
134 clone => [ \&cmd_clone, "Initialize and fetch revisions",
135 { 'revision|r=s' => \$_revision,
136 %fc_opts, %init_opts } ],
137 init => [ \&cmd_init, "Initialize a repo for tracking" .
138 " (requires URL argument)",
139 \%init_opts ],
140 'multi-init' => [ \&cmd_multi_init,
141 "Deprecated alias for ".
142 "'$0 init -T<trunk> -b<branches> -t<tags>'",
143 \%init_opts ],
144 dcommit => [ \&cmd_dcommit,
145 'Commit several diffs to merge with upstream',
146 { 'merge|m|M' => \$_merge,
147 'strategy|s=s' => \$_strategy,
148 'verbose|v' => \$_verbose,
149 'dry-run|n' => \$_dry_run,
150 'fetch-all|all' => \$_fetch_all,
151 'commit-url=s' => \$_commit_url,
152 'revision|r=i' => \$_revision,
153 'no-rebase' => \$_no_rebase,
154 %cmt_opts, %fc_opts } ],
155 branch => [ \&cmd_branch,
156 'Create a branch in the SVN repository',
157 { 'message|m=s' => \$_message,
158 'destination|d=s' => \$_branch_dest,
159 'dry-run|n' => \$_dry_run,
160 'tag|t' => \$_tag,
161 'username=s' => \$Git::SVN::Prompt::_username,
162 'commit-url=s' => \$_commit_url } ],
163 tag => [ sub { $_tag = 1; cmd_branch(@_) },
164 'Create a tag in the SVN repository',
165 { 'message|m=s' => \$_message,
166 'destination|d=s' => \$_branch_dest,
167 'dry-run|n' => \$_dry_run,
168 'username=s' => \$Git::SVN::Prompt::_username,
169 'commit-url=s' => \$_commit_url } ],
170 'set-tree' => [ \&cmd_set_tree,
171 "Set an SVN repository to a git tree-ish",
172 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
173 'create-ignore' => [ \&cmd_create_ignore,
174 'Create a .gitignore per svn:ignore',
175 { 'revision|r=i' => \$_revision
176 } ],
177 'mkdirs' => [ \&cmd_mkdirs ,
178 "recreate empty directories after a checkout",
179 { 'revision|r=i' => \$_revision } ],
180 'propget' => [ \&cmd_propget,
181 'Print the value of a property on a file or directory',
182 { 'revision|r=i' => \$_revision } ],
183 'proplist' => [ \&cmd_proplist,
184 'List all properties of a file or directory',
185 { 'revision|r=i' => \$_revision } ],
186 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
187 { 'revision|r=i' => \$_revision
188 } ],
189 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
190 { 'revision|r=i' => \$_revision
191 } ],
192 'multi-fetch' => [ \&cmd_multi_fetch,
193 "Deprecated alias for $0 fetch --all",
194 { 'revision|r=s' => \$_revision, %fc_opts } ],
195 'migrate' => [ sub { },
196 # no-op, we automatically run this anyways,
197 'Migrate configuration/metadata/layout from
198 previous versions of git-svn',
199 { 'minimize' => \$Git::SVN::Migration::_minimize,
200 %remote_opts } ],
201 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
202 { 'limit=i' => \$Git::SVN::Log::limit,
203 'revision|r=s' => \$_revision,
204 'verbose|v' => \$Git::SVN::Log::verbose,
205 'incremental' => \$Git::SVN::Log::incremental,
206 'oneline' => \$Git::SVN::Log::oneline,
207 'show-commit' => \$Git::SVN::Log::show_commit,
208 'non-recursive' => \$Git::SVN::Log::non_recursive,
209 'authors-file|A=s' => \$_authors,
210 'color' => \$Git::SVN::Log::color,
211 'pager=s' => \$Git::SVN::Log::pager
212 } ],
213 'find-rev' => [ \&cmd_find_rev,
214 "Translate between SVN revision numbers and tree-ish",
215 {} ],
216 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
217 { 'merge|m|M' => \$_merge,
218 'verbose|v' => \$_verbose,
219 'strategy|s=s' => \$_strategy,
220 'local|l' => \$_local,
221 'fetch-all|all' => \$_fetch_all,
222 'dry-run|n' => \$_dry_run,
223 %fc_opts } ],
224 'commit-diff' => [ \&cmd_commit_diff,
225 'Commit a diff between two trees',
226 { 'message|m=s' => \$_message,
227 'file|F=s' => \$_file,
228 'revision|r=s' => \$_revision,
229 %cmt_opts } ],
230 'info' => [ \&cmd_info,
231 "Show info about the latest SVN revision
232 on the current branch",
233 { 'url' => \$_url, } ],
234 'blame' => [ \&Git::SVN::Log::cmd_blame,
235 "Show what revision and author last modified each line of a file",
236 { 'git-format' => \$_git_format } ],
237 'reset' => [ \&cmd_reset,
238 "Undo fetches back to the specified SVN revision",
239 { 'revision|r=s' => \$_revision,
240 'parent|p' => \$_fetch_parent } ],
241 'gc' => [ \&cmd_gc,
242 "Compress unhandled.log files in .git/svn and remove " .
243 "index files in .git/svn",
244 {} ],
245 );
246
247 my $cmd;
248 for (my $i = 0; $i < @ARGV; $i++) {
249 if (defined $cmd{$ARGV[$i]}) {
250 $cmd = $ARGV[$i];
251 splice @ARGV, $i, 1;
252 last;
253 } elsif ($ARGV[$i] eq 'help') {
254 $cmd = $ARGV[$i+1];
255 usage(0);
256 }
257 };
258
259 # make sure we're always running at the top-level working directory
260 unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
261 unless (-d $ENV{GIT_DIR}) {
262 if ($git_dir_user_set) {
263 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
264 "but it is not a directory\n";
265 }
266 my $git_dir = delete $ENV{GIT_DIR};
267 my $cdup = undef;
268 git_cmd_try {
269 $cdup = command_oneline(qw/rev-parse --show-cdup/);
270 $git_dir = '.' unless ($cdup);
271 chomp $cdup if ($cdup);
272 $cdup = "." unless ($cdup && length $cdup);
273 } "Already at toplevel, but $git_dir not found\n";
274 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
275 unless (-d $git_dir) {
276 die "$git_dir still not found after going to ",
277 "'$cdup'\n";
278 }
279 $ENV{GIT_DIR} = $git_dir;
280 }
281 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
282 }
283
284 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
285
286 read_git_config(\%opts);
287 if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
288 Getopt::Long::Configure('pass_through');
289 }
290 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
291 'minimize-connections' => \$Git::SVN::Migration::_minimize,
292 'id|i=s' => \$Git::SVN::default_ref_id,
293 'svn-remote|remote|R=s' => sub {
294 $Git::SVN::no_reuse_existing = 1;
295 $Git::SVN::default_repo_id = $_[1] });
296 exit 1 if (!$rv && $cmd && $cmd ne 'log');
297
298 usage(0) if $_help;
299 version() if $_version;
300 usage(1) unless defined $cmd;
301 load_authors() if $_authors;
302 if (defined $_authors_prog) {
303 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
304 }
305
306 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
307 Git::SVN::Migration::migration_check();
308 }
309 Git::SVN::init_vars();
310 eval {
311 Git::SVN::verify_remotes_sanity();
312 $cmd{$cmd}->[0]->(@ARGV);
313 };
314 fatal $@ if $@;
315 post_fetch_checkout();
316 exit 0;
317
318 ####################### primary functions ######################
319 sub usage {
320 my $exit = shift || 0;
321 my $fd = $exit ? \*STDERR : \*STDOUT;
322 print $fd <<"";
323 git-svn - bidirectional operations between a single Subversion tree and git
324 Usage: git svn <command> [options] [arguments]\n
325
326 print $fd "Available commands:\n" unless $cmd;
327
328 foreach (sort keys %cmd) {
329 next if $cmd && $cmd ne $_;
330 next if /^multi-/; # don't show deprecated commands
331 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
332 foreach (sort keys %{$cmd{$_}->[2]}) {
333 # mixed-case options are for .git/config only
334 next if /[A-Z]/ && /^[a-z]+$/i;
335 # prints out arguments as they should be passed:
336 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
337 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
338 "--$_" : "-$_" }
339 split /\|/,$_)," $x\n";
340 }
341 }
342 print $fd <<"";
343 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
344 arbitrary identifier if you're tracking multiple SVN branches/repositories in
345 one git repository and want to keep them separate. See git-svn(1) for more
346 information.
347
348 exit $exit;
349 }
350
351 sub version {
352 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
353 exit 0;
354 }
355
356 sub do_git_init_db {
357 unless (-d $ENV{GIT_DIR}) {
358 my @init_db = ('init');
359 push @init_db, "--template=$_template" if defined $_template;
360 if (defined $_shared) {
361 if ($_shared =~ /[a-z]/) {
362 push @init_db, "--shared=$_shared";
363 } else {
364 push @init_db, "--shared";
365 }
366 }
367 command_noisy(@init_db);
368 $_repository = Git->repository(Repository => ".git");
369 }
370 command_noisy('config', 'core.autocrlf', 'false');
371 my $set;
372 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
373 foreach my $i (keys %icv) {
374 die "'$set' and '$i' cannot both be set\n" if $set;
375 next unless defined $icv{$i};
376 command_noisy('config', "$pfx.$i", $icv{$i});
377 $set = $i;
378 }
379 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
380 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
381 if defined $$ignore_regex;
382 }
383
384 sub init_subdir {
385 my $repo_path = shift or return;
386 mkpath([$repo_path]) unless -d $repo_path;
387 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
388 $ENV{GIT_DIR} = '.git';
389 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
390 }
391
392 sub cmd_clone {
393 my ($url, $path) = @_;
394 if (!defined $path &&
395 (defined $_trunk || @_branches || @_tags ||
396 defined $_stdlayout) &&
397 $url !~ m#^[a-z\+]+://#) {
398 $path = $url;
399 }
400 $path = basename($url) if !defined $path || !length $path;
401 my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
402 cmd_init($url, $path);
403 command_oneline('config', 'svn.authorsfile', $authors_absolute)
404 if $_authors;
405 Git::SVN::fetch_all($Git::SVN::default_repo_id);
406 }
407
408 sub cmd_init {
409 if (defined $_stdlayout) {
410 $_trunk = 'trunk' if (!defined $_trunk);
411 @_tags = 'tags' if (! @_tags);
412 @_branches = 'branches' if (! @_branches);
413 }
414 if (defined $_trunk || @_branches || @_tags) {
415 return cmd_multi_init(@_);
416 }
417 my $url = shift or die "SVN repository location required ",
418 "as a command-line argument\n";
419 $url = canonicalize_url($url);
420 init_subdir(@_);
421 do_git_init_db();
422
423 if ($Git::SVN::_minimize_url eq 'unset') {
424 $Git::SVN::_minimize_url = 0;
425 }
426
427 Git::SVN->init($url);
428 }
429
430 sub cmd_fetch {
431 if (grep /^\d+=./, @_) {
432 die "'<rev>=<commit>' fetch arguments are ",
433 "no longer supported.\n";
434 }
435 my ($remote) = @_;
436 if (@_ > 1) {
437 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
438 }
439 $Git::SVN::no_reuse_existing = undef;
440 if ($_fetch_parent) {
441 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
442 unless ($gs) {
443 die "Unable to determine upstream SVN information from ",
444 "working tree history\n";
445 }
446 # just fetch, don't checkout.
447 $_no_checkout = 'true';
448 $_fetch_all ? $gs->fetch_all : $gs->fetch;
449 } elsif ($_fetch_all) {
450 cmd_multi_fetch();
451 } else {
452 $remote ||= $Git::SVN::default_repo_id;
453 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
454 }
455 }
456
457 sub cmd_set_tree {
458 my (@commits) = @_;
459 if ($_stdin || !@commits) {
460 print "Reading from stdin...\n";
461 @commits = ();
462 while (<STDIN>) {
463 if (/\b($sha1_short)\b/o) {
464 unshift @commits, $1;
465 }
466 }
467 }
468 my @revs;
469 foreach my $c (@commits) {
470 my @tmp = command('rev-parse',$c);
471 if (scalar @tmp == 1) {
472 push @revs, $tmp[0];
473 } elsif (scalar @tmp > 1) {
474 push @revs, reverse(command('rev-list',@tmp));
475 } else {
476 fatal "Failed to rev-parse $c";
477 }
478 }
479 my $gs = Git::SVN->new;
480 my ($r_last, $cmt_last) = $gs->last_rev_commit;
481 $gs->fetch;
482 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
483 fatal "There are new revisions that were fetched ",
484 "and need to be merged (or acknowledged) ",
485 "before committing.\nlast rev: $r_last\n",
486 " current: $gs->{last_rev}";
487 }
488 $gs->set_tree($_) foreach @revs;
489 print "Done committing ",scalar @revs," revisions to SVN\n";
490 unlink $gs->{index};
491 }
492
493 sub cmd_dcommit {
494 my $head = shift;
495 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
496 'Cannot dcommit with a dirty index. Commit your changes first, '
497 . "or stash them with `git stash'.\n";
498 $head ||= 'HEAD';
499
500 my $old_head;
501 if ($head ne 'HEAD') {
502 $old_head = eval {
503 command_oneline([qw/symbolic-ref -q HEAD/])
504 };
505 if ($old_head) {
506 $old_head =~ s{^refs/heads/}{};
507 } else {
508 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
509 }
510 command(['checkout', $head], STDERR => 0);
511 }
512
513 my @refs;
514 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
515 unless ($gs) {
516 die "Unable to determine upstream SVN information from ",
517 "$head history.\nPerhaps the repository is empty.";
518 }
519
520 if (defined $_commit_url) {
521 $url = $_commit_url;
522 } else {
523 $url = eval { command_oneline('config', '--get',
524 "svn-remote.$gs->{repo_id}.commiturl") };
525 if (!$url) {
526 $url = $gs->full_url
527 }
528 }
529
530 my $last_rev = $_revision if defined $_revision;
531 if ($url) {
532 print "Committing to $url ...\n";
533 }
534 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
535 if ($_no_rebase && scalar(@$linear_refs) > 1) {
536 warn "Attempting to commit more than one change while ",
537 "--no-rebase is enabled.\n",
538 "If these changes depend on each other, re-running ",
539 "without --no-rebase may be required."
540 }
541 my $expect_url = $url;
542 Git::SVN::remove_username($expect_url);
543 while (1) {
544 my $d = shift @$linear_refs or last;
545 unless (defined $last_rev) {
546 (undef, $last_rev, undef) = cmt_metadata("$d~1");
547 unless (defined $last_rev) {
548 fatal "Unable to extract revision information ",
549 "from commit $d~1";
550 }
551 }
552 if ($_dry_run) {
553 print "diff-tree $d~1 $d\n";
554 } else {
555 my $cmt_rev;
556 my %ed_opts = ( r => $last_rev,
557 log => get_commit_entry($d)->{log},
558 ra => Git::SVN::Ra->new($url),
559 config => SVN::Core::config_get_config(
560 $Git::SVN::Ra::config_dir
561 ),
562 tree_a => "$d~1",
563 tree_b => $d,
564 editor_cb => sub {
565 print "Committed r$_[0]\n";
566 $cmt_rev = $_[0];
567 },
568 svn_path => '');
569 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
570 print "No changes\n$d~1 == $d\n";
571 } elsif ($parents->{$d} && @{$parents->{$d}}) {
572 $gs->{inject_parents_dcommit}->{$cmt_rev} =
573 $parents->{$d};
574 }
575 $_fetch_all ? $gs->fetch_all : $gs->fetch;
576 $last_rev = $cmt_rev;
577 next if $_no_rebase;
578
579 # we always want to rebase against the current HEAD,
580 # not any head that was passed to us
581 my @diff = command('diff-tree', $d,
582 $gs->refname, '--');
583 my @finish;
584 if (@diff) {
585 @finish = rebase_cmd();
586 print STDERR "W: $d and ", $gs->refname,
587 " differ, using @finish:\n",
588 join("\n", @diff), "\n";
589 } else {
590 print "No changes between current HEAD and ",
591 $gs->refname,
592 "\nResetting to the latest ",
593 $gs->refname, "\n";
594 @finish = qw/reset --mixed/;
595 }
596 command_noisy(@finish, $gs->refname);
597 if (@diff) {
598 @refs = ();
599 my ($url_, $rev_, $uuid_, $gs_) =
600 working_head_info('HEAD', \@refs);
601 my ($linear_refs_, $parents_) =
602 linearize_history($gs_, \@refs);
603 if (scalar(@$linear_refs) !=
604 scalar(@$linear_refs_)) {
605 fatal "# of revisions changed ",
606 "\nbefore:\n",
607 join("\n", @$linear_refs),
608 "\n\nafter:\n",
609 join("\n", @$linear_refs_), "\n",
610 'If you are attempting to commit ',
611 "merges, try running:\n\t",
612 'git rebase --interactive',
613 '--preserve-merges ',
614 $gs->refname,
615 "\nBefore dcommitting";
616 }
617 if ($url_ ne $expect_url) {
618 if ($url_ eq $gs->metadata_url) {
619 print
620 "Accepting rewritten URL:",
621 " $url_\n";
622 } else {
623 fatal
624 "URL mismatch after rebase:",
625 " $url_ != $expect_url";
626 }
627 }
628 if ($uuid_ ne $uuid) {
629 fatal "uuid mismatch after rebase: ",
630 "$uuid_ != $uuid";
631 }
632 # remap parents
633 my (%p, @l, $i);
634 for ($i = 0; $i < scalar @$linear_refs; $i++) {
635 my $new = $linear_refs_->[$i] or next;
636 $p{$new} =
637 $parents->{$linear_refs->[$i]};
638 push @l, $new;
639 }
640 $parents = \%p;
641 $linear_refs = \@l;
642 }
643 }
644 }
645
646 if ($old_head) {
647 my $new_head = command_oneline(qw/rev-parse HEAD/);
648 my $new_is_symbolic = eval {
649 command_oneline(qw/symbolic-ref -q HEAD/);
650 };
651 if ($new_is_symbolic) {
652 print "dcommitted the branch ", $head, "\n";
653 } else {
654 print "dcommitted on a detached HEAD because you gave ",
655 "a revision argument.\n",
656 "The rewritten commit is: ", $new_head, "\n";
657 }
658 command(['checkout', $old_head], STDERR => 0);
659 }
660
661 unlink $gs->{index};
662 }
663
664 sub cmd_branch {
665 my ($branch_name, $head) = @_;
666
667 unless (defined $branch_name && length $branch_name) {
668 die(($_tag ? "tag" : "branch") . " name required\n");
669 }
670 $head ||= 'HEAD';
671
672 my (undef, $rev, undef, $gs) = working_head_info($head);
673 my $src = $gs->full_url;
674
675 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
676 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
677 my $glob;
678 if ($#{$allglobs} == 0) {
679 $glob = $allglobs->[0];
680 } else {
681 unless(defined $_branch_dest) {
682 die "Multiple ",
683 $_tag ? "tag" : "branch",
684 " paths defined for Subversion repository.\n",
685 "You must specify where you want to create the ",
686 $_tag ? "tag" : "branch",
687 " with the --destination argument.\n";
688 }
689 foreach my $g (@{$allglobs}) {
690 # SVN::Git::Editor could probably be moved to Git.pm..
691 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
692 if ($_branch_dest =~ /$re/) {
693 $glob = $g;
694 last;
695 }
696 }
697 unless (defined $glob) {
698 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
699 foreach my $g (@{$allglobs}) {
700 $g->{path}->{left} =~ /$dest_re/ or next;
701 if (defined $glob) {
702 die "Ambiguous destination: ",
703 $_branch_dest, "\nmatches both '",
704 $glob->{path}->{left}, "' and '",
705 $g->{path}->{left}, "'\n";
706 }
707 $glob = $g;
708 }
709 unless (defined $glob) {
710 die "Unknown ",
711 $_tag ? "tag" : "branch",
712 " destination $_branch_dest\n";
713 }
714 }
715 }
716 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
717 my $url;
718 if (defined $_commit_url) {
719 $url = $_commit_url;
720 } else {
721 $url = eval { command_oneline('config', '--get',
722 "svn-remote.$gs->{repo_id}.commiturl") };
723 if (!$url) {
724 $url = $remote->{url};
725 }
726 }
727 my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
728
729 if ($dst =~ /^https:/ && $src =~ /^http:/) {
730 $src=~s/^http:/https:/;
731 }
732
733 my $ctx = SVN::Client->new(
734 auth => Git::SVN::Ra::_auth_providers(),
735 log_msg => sub {
736 ${ $_[0] } = defined $_message
737 ? $_message
738 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
739 . $branch_name;
740 },
741 );
742
743 eval {
744 $ctx->ls($dst, 'HEAD', 0);
745 } and die "branch ${branch_name} already exists\n";
746
747 print "Copying ${src} at r${rev} to ${dst}...\n";
748 $ctx->copy($src, $rev, $dst)
749 unless $_dry_run;
750
751 $gs->fetch_all;
752 }
753
754 sub cmd_find_rev {
755 my $revision_or_hash = shift or die "SVN or git revision required ",
756 "as a command-line argument\n";
757 my $result;
758 if ($revision_or_hash =~ /^r\d+$/) {
759 my $head = shift;
760 $head ||= 'HEAD';
761 my @refs;
762 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
763 unless ($gs) {
764 die "Unable to determine upstream SVN information from ",
765 "$head history\n";
766 }
767 my $desired_revision = substr($revision_or_hash, 1);
768 $result = $gs->rev_map_get($desired_revision, $uuid);
769 } else {
770 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
771 $result = $rev;
772 }
773 print "$result\n" if $result;
774 }
775
776 sub cmd_rebase {
777 command_noisy(qw/update-index --refresh/);
778 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
779 unless ($gs) {
780 die "Unable to determine upstream SVN information from ",
781 "working tree history\n";
782 }
783 if ($_dry_run) {
784 print "Remote Branch: " . $gs->refname . "\n";
785 print "SVN URL: " . $url . "\n";
786 return;
787 }
788 if (command(qw/diff-index HEAD --/)) {
789 print STDERR "Cannot rebase with uncommited changes:\n";
790 command_noisy('status');
791 exit 1;
792 }
793 unless ($_local) {
794 # rebase will checkout for us, so no need to do it explicitly
795 $_no_checkout = 'true';
796 $_fetch_all ? $gs->fetch_all : $gs->fetch;
797 }
798 command_noisy(rebase_cmd(), $gs->refname);
799 $gs->mkemptydirs;
800 }
801
802 sub cmd_show_ignore {
803 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
804 $gs ||= Git::SVN->new;
805 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
806 $gs->prop_walk($gs->{path}, $r, sub {
807 my ($gs, $path, $props) = @_;
808 print STDOUT "\n# $path\n";
809 my $s = $props->{'svn:ignore'} or return;
810 $s =~ s/[\r\n]+/\n/g;
811 $s =~ s/^\n+//;
812 chomp $s;
813 $s =~ s#^#$path#gm;
814 print STDOUT "$s\n";
815 });
816 }
817
818 sub cmd_show_externals {
819 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
820 $gs ||= Git::SVN->new;
821 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
822 $gs->prop_walk($gs->{path}, $r, sub {
823 my ($gs, $path, $props) = @_;
824 print STDOUT "\n# $path\n";
825 my $s = $props->{'svn:externals'} or return;
826 $s =~ s/[\r\n]+/\n/g;
827 chomp $s;
828 $s =~ s#^#$path#gm;
829 print STDOUT "$s\n";
830 });
831 }
832
833 sub cmd_create_ignore {
834 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
835 $gs ||= Git::SVN->new;
836 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
837 $gs->prop_walk($gs->{path}, $r, sub {
838 my ($gs, $path, $props) = @_;
839 # $path is of the form /path/to/dir/
840 $path = '.' . $path;
841 # SVN can have attributes on empty directories,
842 # which git won't track
843 mkpath([$path]) unless -d $path;
844 my $ignore = $path . '.gitignore';
845 my $s = $props->{'svn:ignore'} or return;
846 open(GITIGNORE, '>', $ignore)
847 or fatal("Failed to open `$ignore' for writing: $!");
848 $s =~ s/[\r\n]+/\n/g;
849 $s =~ s/^\n+//;
850 chomp $s;
851 # Prefix all patterns so that the ignore doesn't apply
852 # to sub-directories.
853 $s =~ s#^#/#gm;
854 print GITIGNORE "$s\n";
855 close(GITIGNORE)
856 or fatal("Failed to close `$ignore': $!");
857 command_noisy('add', '-f', $ignore);
858 });
859 }
860
861 sub cmd_mkdirs {
862 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
863 $gs ||= Git::SVN->new;
864 $gs->mkemptydirs($_revision);
865 }
866
867 sub canonicalize_path {
868 my ($path) = @_;
869 my $dot_slash_added = 0;
870 if (substr($path, 0, 1) ne "/") {
871 $path = "./" . $path;
872 $dot_slash_added = 1;
873 }
874 # File::Spec->canonpath doesn't collapse x/../y into y (for a
875 # good reason), so let's do this manually.
876 $path =~ s#/+#/#g;
877 $path =~ s#/\.(?:/|$)#/#g;
878 $path =~ s#/[^/]+/\.\.##g;
879 $path =~ s#/$##g;
880 $path =~ s#^\./## if $dot_slash_added;
881 $path =~ s#^/##;
882 $path =~ s#^\.$##;
883 return $path;
884 }
885
886 sub canonicalize_url {
887 my ($url) = @_;
888 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
889 return $url;
890 }
891
892 # get_svnprops(PATH)
893 # ------------------
894 # Helper for cmd_propget and cmd_proplist below.
895 sub get_svnprops {
896 my $path = shift;
897 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
898 $gs ||= Git::SVN->new;
899
900 # prefix THE PATH by the sub-directory from which the user
901 # invoked us.
902 $path = $cmd_dir_prefix . $path;
903 fatal("No such file or directory: $path") unless -e $path;
904 my $is_dir = -d $path ? 1 : 0;
905 $path = $gs->{path} . '/' . $path;
906
907 # canonicalize the path (otherwise libsvn will abort or fail to
908 # find the file)
909 $path = canonicalize_path($path);
910
911 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
912 my $props;
913 if ($is_dir) {
914 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
915 }
916 else {
917 (undef, $props) = $gs->ra->get_file($path, $r, undef);
918 }
919 return $props;
920 }
921
922 # cmd_propget (PROP, PATH)
923 # ------------------------
924 # Print the SVN property PROP for PATH.
925 sub cmd_propget {
926 my ($prop, $path) = @_;
927 $path = '.' if not defined $path;
928 usage(1) if not defined $prop;
929 my $props = get_svnprops($path);
930 if (not defined $props->{$prop}) {
931 fatal("`$path' does not have a `$prop' SVN property.");
932 }
933 print $props->{$prop} . "\n";
934 }
935
936 # cmd_proplist (PATH)
937 # -------------------
938 # Print the list of SVN properties for PATH.
939 sub cmd_proplist {
940 my $path = shift;
941 $path = '.' if not defined $path;
942 my $props = get_svnprops($path);
943 print "Properties on '$path':\n";
944 foreach (sort keys %{$props}) {
945 print " $_\n";
946 }
947 }
948
949 sub cmd_multi_init {
950 my $url = shift;
951 unless (defined $_trunk || @_branches || @_tags) {
952 usage(1);
953 }
954
955 $_prefix = '' unless defined $_prefix;
956 if (defined $url) {
957 $url = canonicalize_url($url);
958 init_subdir(@_);
959 }
960 do_git_init_db();
961 if (defined $_trunk) {
962 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
963 # try both old-style and new-style lookups:
964 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
965 unless ($gs_trunk) {
966 my ($trunk_url, $trunk_path) =
967 complete_svn_url($url, $_trunk);
968 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
969 undef, $trunk_ref);
970 }
971 }
972 return unless @_branches || @_tags;
973 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
974 foreach my $path (@_branches) {
975 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
976 }
977 foreach my $path (@_tags) {
978 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
979 }
980 }
981
982 sub cmd_multi_fetch {
983 $Git::SVN::no_reuse_existing = undef;
984 my $remotes = Git::SVN::read_all_remotes();
985 foreach my $repo_id (sort keys %$remotes) {
986 if ($remotes->{$repo_id}->{url}) {
987 Git::SVN::fetch_all($repo_id, $remotes);
988 }
989 }
990 }
991
992 # this command is special because it requires no metadata
993 sub cmd_commit_diff {
994 my ($ta, $tb, $url) = @_;
995 my $usage = "Usage: $0 commit-diff -r<revision> ".
996 "<tree-ish> <tree-ish> [<URL>]";
997 fatal($usage) if (!defined $ta || !defined $tb);
998 my $svn_path = '';
999 if (!defined $url) {
1000 my $gs = eval { Git::SVN->new };
1001 if (!$gs) {
1002 fatal("Needed URL or usable git-svn --id in ",
1003 "the command-line\n", $usage);
1004 }
1005 $url = $gs->{url};
1006 $svn_path = $gs->{path};
1007 }
1008 unless (defined $_revision) {
1009 fatal("-r|--revision is a required argument\n", $usage);
1010 }
1011 if (defined $_message && defined $_file) {
1012 fatal("Both --message/-m and --file/-F specified ",
1013 "for the commit message.\n",
1014 "I have no idea what you mean");
1015 }
1016 if (defined $_file) {
1017 $_message = file_to_s($_file);
1018 } else {
1019 $_message ||= get_commit_entry($tb)->{log};
1020 }
1021 my $ra ||= Git::SVN::Ra->new($url);
1022 my $r = $_revision;
1023 if ($r eq 'HEAD') {
1024 $r = $ra->get_latest_revnum;
1025 } elsif ($r !~ /^\d+$/) {
1026 die "revision argument: $r not understood by git-svn\n";
1027 }
1028 my %ed_opts = ( r => $r,
1029 log => $_message,
1030 ra => $ra,
1031 tree_a => $ta,
1032 tree_b => $tb,
1033 editor_cb => sub { print "Committed r$_[0]\n" },
1034 svn_path => $svn_path );
1035 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1036 print "No changes\n$ta == $tb\n";
1037 }
1038 }
1039
1040 sub escape_uri_only {
1041 my ($uri) = @_;
1042 my @tmp;
1043 foreach (split m{/}, $uri) {
1044 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1045 push @tmp, $_;
1046 }
1047 join('/', @tmp);
1048 }
1049
1050 sub escape_url {
1051 my ($url) = @_;
1052 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1053 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1054 $url = "$scheme://$domain$uri";
1055 }
1056 $url;
1057 }
1058
1059 sub cmd_info {
1060 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1061 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1062 if (exists $_[1]) {
1063 die "Too many arguments specified\n";
1064 }
1065
1066 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1067
1068 if (!$file_type && !$diff_status) {
1069 print STDERR "svn: '$path' is not under version control\n";
1070 exit 1;
1071 }
1072
1073 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1074 unless ($gs) {
1075 die "Unable to determine upstream SVN information from ",
1076 "working tree history\n";
1077 }
1078
1079 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1080 $path = "." if $path eq "";
1081
1082 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1083
1084 if ($_url) {
1085 print escape_url($full_url), "\n";
1086 return;
1087 }
1088
1089 my $result = "Path: $path\n";
1090 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1091 $result .= "URL: " . escape_url($full_url) . "\n";
1092
1093 eval {
1094 my $repos_root = $gs->repos_root;
1095 Git::SVN::remove_username($repos_root);
1096 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1097 };
1098 if ($@) {
1099 $result .= "Repository Root: (offline)\n";
1100 }
1101 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1102 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1103 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1104
1105 $result .= "Node Kind: " .
1106 ($file_type eq "dir" ? "directory" : "file") . "\n";
1107
1108 my $schedule = $diff_status eq "A"
1109 ? "add"
1110 : ($diff_status eq "D" ? "delete" : "normal");
1111 $result .= "Schedule: $schedule\n";
1112
1113 if ($diff_status eq "A") {
1114 print $result, "\n";
1115 return;
1116 }
1117
1118 my ($lc_author, $lc_rev, $lc_date_utc);
1119 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1120 my $log = command_output_pipe(@args);
1121 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1122 while (<$log>) {
1123 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1124 $lc_author = $1;
1125 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1126 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1127 (undef, $lc_rev, undef) = ::extract_metadata($1);
1128 }
1129 }
1130 close $log;
1131
1132 Git::SVN::Log::set_local_timezone();
1133
1134 $result .= "Last Changed Author: $lc_author\n";
1135 $result .= "Last Changed Rev: $lc_rev\n";
1136 $result .= "Last Changed Date: " .
1137 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1138
1139 if ($file_type ne "dir") {
1140 my $text_last_updated_date =
1141 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1142 $result .=
1143 "Text Last Updated: " .
1144 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1145 "\n";
1146 my $checksum;
1147 if ($diff_status eq "D") {
1148 my ($fh, $ctx) =
1149 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1150 if ($file_type eq "link") {
1151 my $file_name = <$fh>;
1152 $checksum = md5sum("link $file_name");
1153 } else {
1154 $checksum = md5sum($fh);
1155 }
1156 command_close_pipe($fh, $ctx);
1157 } elsif ($file_type eq "link") {
1158 my $file_name =
1159 command(qw(cat-file blob), "HEAD:$path");
1160 $checksum =
1161 md5sum("link " . $file_name);
1162 } else {
1163 open FILE, "<", $path or die $!;
1164 $checksum = md5sum(\*FILE);
1165 close FILE or die $!;
1166 }
1167 $result .= "Checksum: " . $checksum . "\n";
1168 }
1169
1170 print $result, "\n";
1171 }
1172
1173 sub cmd_reset {
1174 my $target = shift || $_revision or die "SVN revision required\n";
1175 $target = $1 if $target =~ /^r(\d+)$/;
1176 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1177 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1178 unless ($gs) {
1179 die "Unable to determine upstream SVN information from ".
1180 "history\n";
1181 }
1182 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1183 $gs->rev_map_set($r, $c, 'reset', $uuid);
1184 print "r$r = $c ($gs->{ref_id})\n";
1185 }
1186
1187 sub cmd_gc {
1188 if (!$can_compress) {
1189 warn "Compress::Zlib could not be found; unhandled.log " .
1190 "files will not be compressed.\n";
1191 }
1192 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1193 }
1194
1195 ########################### utility functions #########################
1196
1197 sub rebase_cmd {
1198 my @cmd = qw/rebase/;
1199 push @cmd, '-v' if $_verbose;
1200 push @cmd, qw/--merge/ if $_merge;
1201 push @cmd, "--strategy=$_strategy" if $_strategy;
1202 @cmd;
1203 }
1204
1205 sub post_fetch_checkout {
1206 return if $_no_checkout;
1207 my $gs = $Git::SVN::_head or return;
1208 return if verify_ref('refs/heads/master^0');
1209
1210 # look for "trunk" ref if it exists
1211 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1212 my $fetch = $remote->{fetch};
1213 if ($fetch) {
1214 foreach my $p (keys %$fetch) {
1215 basename($fetch->{$p}) eq 'trunk' or next;
1216 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1217 last;
1218 }
1219 }
1220
1221 my $valid_head = verify_ref('HEAD^0');
1222 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1223 return if ($valid_head || !verify_ref('HEAD^0'));
1224
1225 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1226 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1227 return if -f $index;
1228
1229 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1230 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1231 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1232 print STDERR "Checked out HEAD:\n ",
1233 $gs->full_url, " r", $gs->last_rev, "\n";
1234 $gs->mkemptydirs($gs->last_rev);
1235 }
1236
1237 sub complete_svn_url {
1238 my ($url, $path) = @_;
1239 $path =~ s#/+$##;
1240 if ($path !~ m#^[a-z\+]+://#) {
1241 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1242 fatal("E: '$path' is not a complete URL ",
1243 "and a separate URL is not specified");
1244 }
1245 return ($url, $path);
1246 }
1247 return ($path, '');
1248 }
1249
1250 sub complete_url_ls_init {
1251 my ($ra, $repo_path, $switch, $pfx) = @_;
1252 unless ($repo_path) {
1253 print STDERR "W: $switch not specified\n";
1254 return;
1255 }
1256 $repo_path =~ s#/+$##;
1257 if ($repo_path =~ m#^[a-z\+]+://#) {
1258 $ra = Git::SVN::Ra->new($repo_path);
1259 $repo_path = '';
1260 } else {
1261 $repo_path =~ s#^/+##;
1262 unless ($ra) {
1263 fatal("E: '$repo_path' is not a complete URL ",
1264 "and a separate URL is not specified");
1265 }
1266 }
1267 my $url = $ra->{url};
1268 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1269 my $k = "svn-remote.$gs->{repo_id}.url";
1270 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1271 if ($orig_url && ($orig_url ne $gs->{url})) {
1272 die "$k already set: $orig_url\n",
1273 "wanted to set to: $gs->{url}\n";
1274 }
1275 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1276 my $remote_path = "$gs->{path}/$repo_path";
1277 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1278 $remote_path =~ s#/+#/#g;
1279 $remote_path =~ s#^/##g;
1280 $remote_path .= "/*" if $remote_path !~ /\*/;
1281 my ($n) = ($switch =~ /^--(\w+)/);
1282 if (length $pfx && $pfx !~ m#/$#) {
1283 die "--prefix='$pfx' must have a trailing slash '/'\n";
1284 }
1285 command_noisy('config',
1286 '--add',
1287 "svn-remote.$gs->{repo_id}.$n",
1288 "$remote_path:refs/remotes/$pfx*" .
1289 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1290 }
1291
1292 sub verify_ref {
1293 my ($ref) = @_;
1294 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1295 { STDERR => 0 }); };
1296 }
1297
1298 sub get_tree_from_treeish {
1299 my ($treeish) = @_;
1300 # $treeish can be a symbolic ref, too:
1301 my $type = command_oneline(qw/cat-file -t/, $treeish);
1302 my $expected;
1303 while ($type eq 'tag') {
1304 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1305 }
1306 if ($type eq 'commit') {
1307 $expected = (grep /^tree /, command(qw/cat-file commit/,
1308 $treeish))[0];
1309 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1310 die "Unable to get tree from $treeish\n" unless $expected;
1311 } elsif ($type eq 'tree') {
1312 $expected = $treeish;
1313 } else {
1314 die "$treeish is a $type, expected tree, tag or commit\n";
1315 }
1316 return $expected;
1317 }
1318
1319 sub get_commit_entry {
1320 my ($treeish) = shift;
1321 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1322 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1323 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1324 open my $log_fh, '>', $commit_editmsg or croak $!;
1325
1326 my $type = command_oneline(qw/cat-file -t/, $treeish);
1327 if ($type eq 'commit' || $type eq 'tag') {
1328 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1329 $type, $treeish);
1330 my $in_msg = 0;
1331 my $author;
1332 my $saw_from = 0;
1333 my $msgbuf = "";
1334 while (<$msg_fh>) {
1335 if (!$in_msg) {
1336 $in_msg = 1 if (/^\s*$/);
1337 $author = $1 if (/^author (.*>)/);
1338 } elsif (/^git-svn-id: /) {
1339 # skip this for now, we regenerate the
1340 # correct one on re-fetch anyways
1341 # TODO: set *:merge properties or like...
1342 } else {
1343 if (/^From:/ || /^Signed-off-by:/) {
1344 $saw_from = 1;
1345 }
1346 $msgbuf .= $_;
1347 }
1348 }
1349 $msgbuf =~ s/\s+$//s;
1350 if ($Git::SVN::_add_author_from && defined($author)
1351 && !$saw_from) {
1352 $msgbuf .= "\n\nFrom: $author";
1353 }
1354 print $log_fh $msgbuf or croak $!;
1355 command_close_pipe($msg_fh, $ctx);
1356 }
1357 close $log_fh or croak $!;
1358
1359 if ($_edit || ($type eq 'tree')) {
1360 chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1361 system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1362 }
1363 rename $commit_editmsg, $commit_msg or croak $!;
1364 {
1365 require Encode;
1366 # SVN requires messages to be UTF-8 when entering the repo
1367 local $/;
1368 open $log_fh, '<', $commit_msg or croak $!;
1369 binmode $log_fh;
1370 chomp($log_entry{log} = <$log_fh>);
1371
1372 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1373 my $msg = $log_entry{log};
1374
1375 eval { $msg = Encode::decode($enc, $msg, 1) };
1376 if ($@) {
1377 die "Could not decode as $enc:\n", $msg,
1378 "\nPerhaps you need to set i18n.commitencoding\n";
1379 }
1380
1381 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1382 die "Could not encode as UTF-8:\n$msg\n" if $@;
1383
1384 $log_entry{log} = $msg;
1385
1386 close $log_fh or croak $!;
1387 }
1388 unlink $commit_msg;
1389 \%log_entry;
1390 }
1391
1392 sub s_to_file {
1393 my ($str, $file, $mode) = @_;
1394 open my $fd,'>',$file or croak $!;
1395 print $fd $str,"\n" or croak $!;
1396 close $fd or croak $!;
1397 chmod ($mode &~ umask, $file) if (defined $mode);
1398 }
1399
1400 sub file_to_s {
1401 my $file = shift;
1402 open my $fd,'<',$file or croak "$!: file: $file\n";
1403 local $/;
1404 my $ret = <$fd>;
1405 close $fd or croak $!;
1406 $ret =~ s/\s*$//s;
1407 return $ret;
1408 }
1409
1410 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1411 sub load_authors {
1412 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1413 my $log = $cmd eq 'log';
1414 while (<$authors>) {
1415 chomp;
1416 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1417 my ($user, $name, $email) = ($1, $2, $3);
1418 if ($log) {
1419 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1420 } else {
1421 $users{$user} = [$name, $email];
1422 }
1423 }
1424 close $authors or croak $!;
1425 }
1426
1427 # convert GetOpt::Long specs for use by git-config
1428 sub read_git_config {
1429 my $opts = shift;
1430 my @config_only;
1431 foreach my $o (keys %$opts) {
1432 # if we have mixedCase and a long option-only, then
1433 # it's a config-only variable that we don't need for
1434 # the command-line.
1435 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1436 my $v = $opts->{$o};
1437 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1438 $key =~ s/-//g;
1439 my $arg = 'git config';
1440 $arg .= ' --int' if ($o =~ /[:=]i$/);
1441 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1442 if (ref $v eq 'ARRAY') {
1443 chomp(my @tmp = `$arg --get-all svn.$key`);
1444 @$v = @tmp if @tmp;
1445 } else {
1446 chomp(my $tmp = `$arg --get svn.$key`);
1447 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1448 $$v = $tmp;
1449 }
1450 }
1451 }
1452 delete @$opts{@config_only} if @config_only;
1453 }
1454
1455 sub extract_metadata {
1456 my $id = shift or return (undef, undef, undef);
1457 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1458 \s([a-f\d\-]+)$/ix);
1459 if (!defined $rev || !$uuid || !$url) {
1460 # some of the original repositories I made had
1461 # identifiers like this:
1462 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1463 }
1464 return ($url, $rev, $uuid);
1465 }
1466
1467 sub cmt_metadata {
1468 return extract_metadata((grep(/^git-svn-id: /,
1469 command(qw/cat-file commit/, shift)))[-1]);
1470 }
1471
1472 sub cmt_sha2rev_batch {
1473 my %s2r;
1474 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1475 my $list = shift;
1476
1477 foreach my $sha (@{$list}) {
1478 my $first = 1;
1479 my $size = 0;
1480 print $out $sha, "\n";
1481
1482 while (my $line = <$in>) {
1483 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1484 last;
1485 } elsif ($first &&
1486 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1487 $first = 0;
1488 $size = $1;
1489 next;
1490 } elsif ($line =~ /^(git-svn-id: )/) {
1491 my (undef, $rev, undef) =
1492 extract_metadata($line);
1493 $s2r{$sha} = $rev;
1494 }
1495
1496 $size -= length($line);
1497 last if ($size == 0);
1498 }
1499 }
1500
1501 command_close_bidi_pipe($pid, $in, $out, $ctx);
1502
1503 return \%s2r;
1504 }
1505
1506 sub working_head_info {
1507 my ($head, $refs) = @_;
1508 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1509 my ($fh, $ctx) = command_output_pipe(@args, $head);
1510 my $hash;
1511 my %max;
1512 while (<$fh>) {
1513 if ( m{^commit ($::sha1)$} ) {
1514 unshift @$refs, $hash if $hash and $refs;
1515 $hash = $1;
1516 next;
1517 }
1518 next unless s{^\s*(git-svn-id:)}{$1};
1519 my ($url, $rev, $uuid) = extract_metadata($_);
1520 if (defined $url && defined $rev) {
1521 next if $max{$url} and $max{$url} < $rev;
1522 if (my $gs = Git::SVN->find_by_url($url)) {
1523 my $c = $gs->rev_map_get($rev, $uuid);
1524 if ($c && $c eq $hash) {
1525 close $fh; # break the pipe
1526 return ($url, $rev, $uuid, $gs);
1527 } else {
1528 $max{$url} ||= $gs->rev_map_max;
1529 }
1530 }
1531 }
1532 }
1533 command_close_pipe($fh, $ctx);
1534 (undef, undef, undef, undef);
1535 }
1536
1537 sub read_commit_parents {
1538 my ($parents, $c) = @_;
1539 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1540 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1541 @{$parents->{$c}} = split(/ /, $p);
1542 }
1543
1544 sub linearize_history {
1545 my ($gs, $refs) = @_;
1546 my %parents;
1547 foreach my $c (@$refs) {
1548 read_commit_parents(\%parents, $c);
1549 }
1550
1551 my @linear_refs;
1552 my %skip = ();
1553 my $last_svn_commit = $gs->last_commit;
1554 foreach my $c (reverse @$refs) {
1555 next if $c eq $last_svn_commit;
1556 last if $skip{$c};
1557
1558 unshift @linear_refs, $c;
1559 $skip{$c} = 1;
1560
1561 # we only want the first parent to diff against for linear
1562 # history, we save the rest to inject when we finalize the
1563 # svn commit
1564 my $fp_a = verify_ref("$c~1");
1565 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1566 if (!$fp_a || !$fp_b) {
1567 die "Commit $c\n",
1568 "has no parent commit, and therefore ",
1569 "nothing to diff against.\n",
1570 "You should be working from a repository ",
1571 "originally created by git-svn\n";
1572 }
1573 if ($fp_a ne $fp_b) {
1574 die "$c~1 = $fp_a, however parsing commit $c ",
1575 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1576 }
1577
1578 foreach my $p (@{$parents{$c}}) {
1579 $skip{$p} = 1;
1580 }
1581 }
1582 (\@linear_refs, \%parents);
1583 }
1584
1585 sub find_file_type_and_diff_status {
1586 my ($path) = @_;
1587 return ('dir', '') if $path eq '';
1588
1589 my $diff_output =
1590 command_oneline(qw(diff --cached --name-status --), $path) || "";
1591 my $diff_status = (split(' ', $diff_output))[0] || "";
1592
1593 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1594
1595 return (undef, undef) if !$diff_status && !$ls_tree;
1596
1597 if ($diff_status eq "A") {
1598 return ("link", $diff_status) if -l $path;
1599 return ("dir", $diff_status) if -d $path;
1600 return ("file", $diff_status);
1601 }
1602
1603 my $mode = (split(' ', $ls_tree))[0] || "";
1604
1605 return ("link", $diff_status) if $mode eq "120000";
1606 return ("dir", $diff_status) if $mode eq "040000";
1607 return ("file", $diff_status);
1608 }
1609
1610 sub md5sum {
1611 my $arg = shift;
1612 my $ref = ref $arg;
1613 my $md5 = Digest::MD5->new();
1614 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1615 $md5->addfile($arg) or croak $!;
1616 } elsif ($ref eq 'SCALAR') {
1617 $md5->add($$arg) or croak $!;
1618 } elsif (!$ref) {
1619 $md5->add($arg) or croak $!;
1620 } else {
1621 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1622 }
1623 return $md5->hexdigest();
1624 }
1625
1626 sub gc_directory {
1627 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1628 my $out_filename = $_ . ".gz";
1629 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1630 binmode $in_fh;
1631 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1632 die "Unable to open $out_filename: $!\n";
1633
1634 my $res;
1635 while ($res = sysread($in_fh, my $str, 1024)) {
1636 $gz->gzwrite($str) or
1637 die "Unable to write: ".$gz->gzerror()."!\n";
1638 }
1639 unlink $_ or die "unlink $File::Find::name: $!\n";
1640 } elsif (-f $_ && basename($_) eq "index") {
1641 unlink $_ or die "unlink $_: $!\n";
1642 }
1643 }
1644
1645 package Git::SVN;
1646 use strict;
1647 use warnings;
1648 use Fcntl qw/:DEFAULT :seek/;
1649 use constant rev_map_fmt => 'NH40';
1650 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1651 $_repack $_repack_flags $_use_svm_props $_head
1652 $_use_svnsync_props $no_reuse_existing $_minimize_url
1653 $_use_log_author $_add_author_from $_localtime/;
1654 use Carp qw/croak/;
1655 use File::Path qw/mkpath/;
1656 use File::Copy qw/copy/;
1657 use IPC::Open3;
1658 use Memoize; # core since 5.8.0, Jul 2002
1659
1660 my ($_gc_nr, $_gc_period);
1661
1662 # properties that we do not log:
1663 my %SKIP_PROP;
1664 BEGIN {
1665 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1666 svn:special svn:executable
1667 svn:entry:committed-rev
1668 svn:entry:last-author
1669 svn:entry:uuid
1670 svn:entry:committed-date/;
1671
1672 # some options are read globally, but can be overridden locally
1673 # per [svn-remote "..."] section. Command-line options will *NOT*
1674 # override options set in an [svn-remote "..."] section
1675 no strict 'refs';
1676 for my $option (qw/follow_parent no_metadata use_svm_props
1677 use_svnsync_props/) {
1678 my $key = $option;
1679 $key =~ tr/_//d;
1680 my $prop = "-$option";
1681 *$option = sub {
1682 my ($self) = @_;
1683 return $self->{$prop} if exists $self->{$prop};
1684 my $k = "svn-remote.$self->{repo_id}.$key";
1685 eval { command_oneline(qw/config --get/, $k) };
1686 if ($@) {
1687 $self->{$prop} = ${"Git::SVN::_$option"};
1688 } else {
1689 my $v = command_oneline(qw/config --bool/,$k);
1690 $self->{$prop} = $v eq 'false' ? 0 : 1;
1691 }
1692 return $self->{$prop};
1693 }
1694 }
1695 }
1696
1697
1698 my (%LOCKFILES, %INDEX_FILES);
1699 END {
1700 unlink keys %LOCKFILES if %LOCKFILES;
1701 unlink keys %INDEX_FILES if %INDEX_FILES;
1702 }
1703
1704 sub resolve_local_globs {
1705 my ($url, $fetch, $glob_spec) = @_;
1706 return unless defined $glob_spec;
1707 my $ref = $glob_spec->{ref};
1708 my $path = $glob_spec->{path};
1709 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1710 next unless m#^$ref->{regex}$#;
1711 my $p = $1;
1712 my $pathname = desanitize_refname($path->full_path($p));
1713 my $refname = desanitize_refname($ref->full_path($p));
1714 if (my $existing = $fetch->{$pathname}) {
1715 if ($existing ne $refname) {
1716 die "Refspec conflict:\n",
1717 "existing: $existing\n",
1718 " globbed: $refname\n";
1719 }
1720 my $u = (::cmt_metadata("$refname"))[0];
1721 $u =~ s!^\Q$url\E(/|$)!! or die
1722 "$refname: '$url' not found in '$u'\n";
1723 if ($pathname ne $u) {
1724 warn "W: Refspec glob conflict ",
1725 "(ref: $refname):\n",
1726 "expected path: $pathname\n",
1727 " real path: $u\n",
1728 "Continuing ahead with $u\n";
1729 next;
1730 }
1731 } else {
1732 $fetch->{$pathname} = $refname;
1733 }
1734 }
1735 }
1736
1737 sub parse_revision_argument {
1738 my ($base, $head) = @_;
1739 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1740 return ($base, $head);
1741 }
1742 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1743 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1744 return ($head, $head) if ($::_revision eq 'HEAD');
1745 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1746 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1747 die "revision argument: $::_revision not understood by git-svn\n";
1748 }
1749
1750 sub fetch_all {
1751 my ($repo_id, $remotes) = @_;
1752 if (ref $repo_id) {
1753 my $gs = $repo_id;
1754 $repo_id = undef;
1755 $repo_id = $gs->{repo_id};
1756 }
1757 $remotes ||= read_all_remotes();
1758 my $remote = $remotes->{$repo_id} or
1759 die "[svn-remote \"$repo_id\"] unknown\n";
1760 my $fetch = $remote->{fetch};
1761 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1762 my (@gs, @globs);
1763 my $ra = Git::SVN::Ra->new($url);
1764 my $uuid = $ra->get_uuid;
1765 my $head = $ra->get_latest_revnum;
1766
1767 # ignore errors, $head revision may not even exist anymore
1768 eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
1769 warn "W: $@\n" if $@;
1770
1771 my $base = defined $fetch ? $head : 0;
1772
1773 # read the max revs for wildcard expansion (branches/*, tags/*)
1774 foreach my $t (qw/branches tags/) {
1775 defined $remote->{$t} or next;
1776 push @globs, @{$remote->{$t}};
1777
1778 my $max_rev = eval { tmp_config(qw/--int --get/,
1779 "svn-remote.$repo_id.${t}-maxRev") };
1780 if (defined $max_rev && ($max_rev < $base)) {
1781 $base = $max_rev;
1782 } elsif (!defined $max_rev) {
1783 $base = 0;
1784 }
1785 }
1786
1787 if ($fetch) {
1788 foreach my $p (sort keys %$fetch) {
1789 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1790 my $lr = $gs->rev_map_max;
1791 if (defined $lr) {
1792 $base = $lr if ($lr < $base);
1793 }
1794 push @gs, $gs;
1795 }
1796 }
1797
1798 ($base, $head) = parse_revision_argument($base, $head);
1799 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1800 }
1801
1802 sub read_all_remotes {
1803 my $r = {};
1804 my $use_svm_props = eval { command_oneline(qw/config --bool
1805 svn.useSvmProps/) };
1806 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1807 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1808 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1809 if (m!^(.+)\.fetch=$svn_refspec$!) {
1810 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1811 die("svn-remote.$remote: remote ref '$remote_ref' "
1812 . "must start with 'refs/'\n")
1813 unless $remote_ref =~ m{^refs/};
1814 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1815 $r->{$remote}->{svm} = {} if $use_svm_props;
1816 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1817 $r->{$1}->{svm} = {};
1818 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1819 $r->{$1}->{url} = $2;
1820 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1821 my ($remote, $t, $local_ref, $remote_ref) =
1822 ($1, $2, $3, $4);
1823 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1824 . "must start with 'refs/'\n")
1825 unless $remote_ref =~ m{^refs/};
1826 my $rs = {
1827 t => $t,
1828 remote => $remote,
1829 path => Git::SVN::GlobSpec->new($local_ref, 1),
1830 ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
1831 if (length($rs->{ref}->{right}) != 0) {
1832 die "The '*' glob character must be the last ",
1833 "character of '$remote_ref'\n";
1834 }
1835 push @{ $r->{$remote}->{$t} }, $rs;
1836 }
1837 }
1838
1839 map {
1840 if (defined $r->{$_}->{svm}) {
1841 my $svm;
1842 eval {
1843 my $section = "svn-remote.$_";
1844 $svm = {
1845 source => tmp_config('--get',
1846 "$section.svm-source"),
1847 replace => tmp_config('--get',
1848 "$section.svm-replace"),
1849 }
1850 };
1851 $r->{$_}->{svm} = $svm;
1852 }
1853 } keys %$r;
1854
1855 $r;
1856 }
1857
1858 sub init_vars {
1859 $_gc_nr = $_gc_period = 1000;
1860 if (defined $_repack || defined $_repack_flags) {
1861 warn "Repack options are obsolete; they have no effect.\n";
1862 }
1863 }
1864
1865 sub verify_remotes_sanity {
1866 return unless -d $ENV{GIT_DIR};
1867 my %seen;
1868 foreach (command(qw/config -l/)) {
1869 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1870 if ($seen{$1}) {
1871 die "Remote ref refs/remote/$1 is tracked by",
1872 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1873 "Please resolve this ambiguity in ",
1874 "your git configuration file before ",
1875 "continuing\n";
1876 }
1877 $seen{$1} = $_;
1878 }
1879 }
1880 }
1881
1882 sub find_existing_remote {
1883 my ($url, $remotes) = @_;
1884 return undef if $no_reuse_existing;
1885 my $existing;
1886 foreach my $repo_id (keys %$remotes) {
1887 my $u = $remotes->{$repo_id}->{url} or next;
1888 next if $u ne $url;
1889 $existing = $repo_id;
1890 last;
1891 }
1892 $existing;
1893 }
1894
1895 sub init_remote_config {
1896 my ($self, $url, $no_write) = @_;
1897 $url =~ s!/+$!!; # strip trailing slash
1898 my $r = read_all_remotes();
1899 my $existing = find_existing_remote($url, $r);
1900 if ($existing) {
1901 unless ($no_write) {
1902 print STDERR "Using existing ",
1903 "[svn-remote \"$existing\"]\n";
1904 }
1905 $self->{repo_id} = $existing;
1906 } elsif ($_minimize_url) {
1907 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1908 $existing = find_existing_remote($min_url, $r);
1909 if ($existing) {
1910 unless ($no_write) {
1911 print STDERR "Using existing ",
1912 "[svn-remote \"$existing\"]\n";
1913 }
1914 $self->{repo_id} = $existing;
1915 }
1916 if ($min_url ne $url) {
1917 unless ($no_write) {
1918 print STDERR "Using higher level of URL: ",
1919 "$url => $min_url\n";
1920 }
1921 my $old_path = $self->{path};
1922 $self->{path} = $url;
1923 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1924 if (length $old_path) {
1925 $self->{path} .= "/$old_path";
1926 }
1927 $url = $min_url;
1928 }
1929 }
1930 my $orig_url;
1931 if (!$existing) {
1932 # verify that we aren't overwriting anything:
1933 $orig_url = eval {
1934 command_oneline('config', '--get',
1935 "svn-remote.$self->{repo_id}.url")
1936 };
1937 if ($orig_url && ($orig_url ne $url)) {
1938 die "svn-remote.$self->{repo_id}.url already set: ",
1939 "$orig_url\nwanted to set to: $url\n";
1940 }
1941 }
1942 my ($xrepo_id, $xpath) = find_ref($self->refname);
1943 if (!$no_write && defined $xpath) {
1944 die "svn-remote.$xrepo_id.fetch already set to track ",
1945 "$xpath:", $self->refname, "\n";
1946 }
1947 unless ($no_write) {
1948 command_noisy('config',
1949 "svn-remote.$self->{repo_id}.url", $url);
1950 $self->{path} =~ s{^/}{};
1951 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1952 command_noisy('config', '--add',
1953 "svn-remote.$self->{repo_id}.fetch",
1954 "$self->{path}:".$self->refname);
1955 }
1956 $self->{url} = $url;
1957 }
1958
1959 sub find_by_url { # repos_root and, path are optional
1960 my ($class, $full_url, $repos_root, $path) = @_;
1961
1962 return undef unless defined $full_url;
1963 remove_username($full_url);
1964 remove_username($repos_root) if defined $repos_root;
1965 my $remotes = read_all_remotes();
1966 if (defined $full_url && defined $repos_root && !defined $path) {
1967 $path = $full_url;
1968 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1969 }
1970 foreach my $repo_id (keys %$remotes) {
1971 my $u = $remotes->{$repo_id}->{url} or next;
1972 remove_username($u);
1973 next if defined $repos_root && $repos_root ne $u;
1974
1975 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1976 foreach my $t (qw/branches tags/) {
1977 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1978 resolve_local_globs($u, $fetch, $globspec);
1979 }
1980 }
1981 my $p = $path;
1982 my $rwr = rewrite_root({repo_id => $repo_id});
1983 my $svm = $remotes->{$repo_id}->{svm}
1984 if defined $remotes->{$repo_id}->{svm};
1985 unless (defined $p) {
1986 $p = $full_url;
1987 my $z = $u;
1988 my $prefix = '';
1989 if ($rwr) {
1990 $z = $rwr;
1991 remove_username($z);
1992 } elsif (defined $svm) {
1993 $z = $svm->{source};
1994 $prefix = $svm->{replace};
1995 $prefix =~ s#^\Q$u\E(?:/|$)##;
1996 $prefix =~ s#/$##;
1997 }
1998 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1999 }
2000 foreach my $f (keys %$fetch) {
2001 next if $f ne $p;
2002 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2003 }
2004 }
2005 undef;
2006 }
2007
2008 sub init {
2009 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2010 my $self = _new($class, $repo_id, $ref_id, $path);
2011 if (defined $url) {
2012 $self->init_remote_config($url, $no_write);
2013 }
2014 $self;
2015 }
2016
2017 sub find_ref {
2018 my ($ref_id) = @_;
2019 foreach (command(qw/config -l/)) {
2020 next unless m!^svn-remote\.(.+)\.fetch=
2021 \s*(.*?)\s*:\s*(.+?)\s*$!x;
2022 my ($repo_id, $path, $ref) = ($1, $2, $3);
2023 if ($ref eq $ref_id) {
2024 $path = '' if ($path =~ m#^\./?#);
2025 return ($repo_id, $path);
2026 }
2027 }
2028 (undef, undef, undef);
2029 }
2030
2031 sub new {
2032 my ($class, $ref_id, $repo_id, $path) = @_;
2033 if (defined $ref_id && !defined $repo_id && !defined $path) {
2034 ($repo_id, $path) = find_ref($ref_id);
2035 if (!defined $repo_id) {
2036 die "Could not find a \"svn-remote.*.fetch\" key ",
2037 "in the repository configuration matching: ",
2038 "$ref_id\n";
2039 }
2040 }
2041 my $self = _new($class, $repo_id, $ref_id, $path);
2042 if (!defined $self->{path} || !length $self->{path}) {
2043 my $fetch = command_oneline('config', '--get',
2044 "svn-remote.$repo_id.fetch",
2045 ":$ref_id\$") or
2046 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2047 "\":$ref_id\$\" in config\n";
2048 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2049 }
2050 $self->{url} = command_oneline('config', '--get',
2051 "svn-remote.$repo_id.url") or
2052 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2053 $self->rebuild;
2054 $self;
2055 }
2056
2057 sub refname {
2058 my ($refname) = $_[0]->{ref_id} ;
2059
2060 # It cannot end with a slash /, we'll throw up on this because
2061 # SVN can't have directories with a slash in their name, either:
2062 if ($refname =~ m{/$}) {
2063 die "ref: '$refname' ends with a trailing slash, this is ",
2064 "not permitted by git nor Subversion\n";
2065 }
2066
2067 # It cannot have ASCII control character space, tilde ~, caret ^,
2068 # colon :, question-mark ?, asterisk *, space, or open bracket [
2069 # anywhere.
2070 #
2071 # Additionally, % must be escaped because it is used for escaping
2072 # and we want our escaped refname to be reversible
2073 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2074
2075 # no slash-separated component can begin with a dot .
2076 # /.* becomes /%2E*
2077 $refname =~ s{/\.}{/%2E}g;
2078
2079 # It cannot have two consecutive dots .. anywhere
2080 # .. becomes %2E%2E
2081 $refname =~ s{\.\.}{%2E%2E}g;
2082
2083 return $refname;
2084 }
2085
2086 sub desanitize_refname {
2087 my ($refname) = @_;
2088 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2089 return $refname;
2090 }
2091
2092 sub svm_uuid {
2093 my ($self) = @_;
2094 return $self->{svm}->{uuid} if $self->svm;
2095 $self->ra;
2096 unless ($self->{svm}) {
2097 die "SVM UUID not cached, and reading remotely failed\n";
2098 }
2099 $self->{svm}->{uuid};
2100 }
2101
2102 sub svm {
2103 my ($self) = @_;
2104 return $self->{svm} if $self->{svm};
2105 my $svm;
2106 # see if we have it in our config, first:
2107 eval {
2108 my $section = "svn-remote.$self->{repo_id}";
2109 $svm = {
2110 source => tmp_config('--get', "$section.svm-source"),
2111 uuid => tmp_config('--get', "$section.svm-uuid"),
2112 replace => tmp_config('--get', "$section.svm-replace"),
2113 }
2114 };
2115 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2116 $self->{svm} = $svm;
2117 }
2118 $self->{svm};
2119 }
2120
2121 sub _set_svm_vars {
2122 my ($self, $ra) = @_;
2123 return $ra if $self->svm;
2124
2125 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2126 "(svm:source, svm:uuid) ",
2127 "from the following URLs:\n" );
2128 sub read_svm_props {
2129 my ($self, $ra, $path, $r) = @_;
2130 my $props = ($ra->get_dir($path, $r))[2];
2131 my $src = $props->{'svm:source'};
2132 my $uuid = $props->{'svm:uuid'};
2133 return undef if (!$src || !$uuid);
2134
2135 chomp($src, $uuid);
2136
2137 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2138 or die "doesn't look right - svm:uuid is '$uuid'\n";
2139
2140 # the '!' is used to mark the repos_root!/relative/path
2141 $src =~ s{/?!/?}{/};
2142 $src =~ s{/+$}{}; # no trailing slashes please
2143 # username is of no interest
2144 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2145
2146 my $replace = $ra->{url};
2147 $replace .= "/$path" if length $path;
2148
2149 my $section = "svn-remote.$self->{repo_id}";
2150 tmp_config("$section.svm-source", $src);
2151 tmp_config("$section.svm-replace", $replace);
2152 tmp_config("$section.svm-uuid", $uuid);
2153 $self->{svm} = {
2154 source => $src,
2155 uuid => $uuid,
2156 replace => $replace
2157 };
2158 }
2159
2160 my $r = $ra->get_latest_revnum;
2161 my $path = $self->{path};
2162 my %tried;
2163 while (length $path) {
2164 unless ($tried{"$self->{url}/$path"}) {
2165 return $ra if $self->read_svm_props($ra, $path, $r);
2166 $tried{"$self->{url}/$path"} = 1;
2167 }
2168 $path =~ s#/?[^/]+$##;
2169 }
2170 die "Path: '$path' should be ''\n" if $path ne '';
2171 return $ra if $self->read_svm_props($ra, $path, $r);
2172 $tried{"$self->{url}/$path"} = 1;
2173
2174 if ($ra->{repos_root} eq $self->{url}) {
2175 die @err, (map { " $_\n" } keys %tried), "\n";
2176 }
2177
2178 # nope, make sure we're connected to the repository root:
2179 my $ok;
2180 my @tried_b;
2181 $path = $ra->{svn_path};
2182 $ra = Git::SVN::Ra->new($ra->{repos_root});
2183 while (length $path) {
2184 unless ($tried{"$ra->{url}/$path"}) {
2185 $ok = $self->read_svm_props($ra, $path, $r);
2186 last if $ok;
2187 $tried{"$ra->{url}/$path"} = 1;
2188 }
2189 $path =~ s#/?[^/]+$##;
2190 }
2191 die "Path: '$path' should be ''\n" if $path ne '';
2192 $ok ||= $self->read_svm_props($ra, $path, $r);
2193 $tried{"$ra->{url}/$path"} = 1;
2194 if (!$ok) {
2195 die @err, (map { " $_\n" } keys %tried), "\n";
2196 }
2197 Git::SVN::Ra->new($self->{url});
2198 }
2199
2200 sub svnsync {
2201 my ($self) = @_;
2202 return $self->{svnsync} if $self->{svnsync};
2203
2204 if ($self->no_metadata) {
2205 die "Can't have both 'noMetadata' and ",
2206 "'useSvnsyncProps' options set!\n";
2207 }
2208 if ($self->rewrite_root) {
2209 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2210 "options set!\n";
2211 }
2212 if ($self->rewrite_uuid) {
2213 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2214 "options set!\n";
2215 }
2216
2217 my $svnsync;
2218 # see if we have it in our config, first:
2219 eval {
2220 my $section = "svn-remote.$self->{repo_id}";
2221
2222 my $url = tmp_config('--get', "$section.svnsync-url");
2223 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2224 die "doesn't look right - svn:sync-from-url is '$url'\n";
2225
2226 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2227 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2228 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2229
2230 $svnsync = { url => $url, uuid => $uuid }
2231 };
2232 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2233 return $self->{svnsync} = $svnsync;
2234 }
2235
2236 my $err = "useSvnsyncProps set, but failed to read " .
2237 "svnsync property: svn:sync-from-";
2238 my $rp = $self->ra->rev_proplist(0);
2239
2240 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2241 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2242 die "doesn't look right - svn:sync-from-url is '$url'\n";
2243
2244 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2245 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2246 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2247
2248 my $section = "svn-remote.$self->{repo_id}";
2249 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2250 tmp_config('--add', "$section.svnsync-url", $url);
2251 return $self->{svnsync} = { url => $url, uuid => $uuid };
2252 }
2253
2254 # this allows us to memoize our SVN::Ra UUID locally and avoid a
2255 # remote lookup (useful for 'git svn log').
2256 sub ra_uuid {
2257 my ($self) = @_;
2258 unless ($self->{ra_uuid}) {
2259 my $key = "svn-remote.$self->{repo_id}.uuid";
2260 my $uuid = eval { tmp_config('--get', $key) };
2261 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2262 $self->{ra_uuid} = $uuid;
2263 } else {
2264 die "ra_uuid called without URL\n" unless $self->{url};
2265 $self->{ra_uuid} = $self->ra->get_uuid;
2266 tmp_config('--add', $key, $self->{ra_uuid});
2267 }
2268 }
2269 $self->{ra_uuid};
2270 }
2271
2272 sub _set_repos_root {
2273 my ($self, $repos_root) = @_;
2274 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2275 $repos_root ||= $self->ra->{repos_root};
2276 tmp_config($k, $repos_root);
2277 $repos_root;
2278 }
2279
2280 sub repos_root {
2281 my ($self) = @_;
2282 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2283 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2284 }
2285
2286 sub ra {
2287 my ($self) = shift;
2288 my $ra = Git::SVN::Ra->new($self->{url});
2289 $self->_set_repos_root($ra->{repos_root});
2290 if ($self->use_svm_props && !$self->{svm}) {
2291 if ($self->no_metadata) {
2292 die "Can't have both 'noMetadata' and ",
2293 "'useSvmProps' options set!\n";
2294 } elsif ($self->use_svnsync_props) {
2295 die "Can't have both 'useSvnsyncProps' and ",
2296 "'useSvmProps' options set!\n";
2297 }
2298 $ra = $self->_set_svm_vars($ra);
2299 $self->{-want_revprops} = 1;
2300 }
2301 $ra;
2302 }
2303
2304 # prop_walk(PATH, REV, SUB)
2305 # -------------------------
2306 # Recursively traverse PATH at revision REV and invoke SUB for each
2307 # directory that contains a SVN property. SUB will be invoked as
2308 # follows: &SUB(gs, path, props); where `gs' is this instance of
2309 # Git::SVN, `path' the path to the directory where the properties
2310 # `props' were found. The `path' will be relative to point of checkout,
2311 # that is, if url://repo/trunk is the current Git branch, and that
2312 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
2313 # as `path' (note the trailing `/').
2314 sub prop_walk {
2315 my ($self, $path, $rev, $sub) = @_;
2316
2317 $path =~ s#^/##;
2318 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2319 $path =~ s#^/*#/#g;
2320 my $p = $path;
2321 # Strip the irrelevant part of the path.
2322 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2323 # Ensure the path is terminated by a `/'.
2324 $p =~ s#/*$#/#;
2325
2326 # The properties contain all the internal SVN stuff nobody
2327 # (usually) cares about.
2328 my $interesting_props = 0;
2329 foreach (keys %{$props}) {
2330 # If it doesn't start with `svn:', it must be a
2331 # user-defined property.
2332 ++$interesting_props and next if $_ !~ /^svn:/;
2333 # FIXME: Fragile, if SVN adds new public properties,
2334 # this needs to be updated.
2335 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2336 |eol-style|mime-type
2337 |externals|needs-lock)$/x;
2338 }
2339 &$sub($self, $p, $props) if $interesting_props;
2340
2341 foreach (sort keys %$dirent) {
2342 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2343 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2344 }
2345 }
2346
2347 sub last_rev { ($_[0]->last_rev_commit)[0] }
2348 sub last_commit { ($_[0]->last_rev_commit)[1] }
2349
2350 # returns the newest SVN revision number and newest commit SHA1
2351 sub last_rev_commit {
2352 my ($self) = @_;
2353 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2354 return ($self->{last_rev}, $self->{last_commit});
2355 }
2356 my $c = ::verify_ref($self->refname.'^0');
2357 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2358 my $rev = (::cmt_metadata($c))[1];
2359 if (defined $rev) {
2360 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2361 return ($rev, $c);
2362 }
2363 }
2364 my $map_path = $self->map_path;
2365 unless (-e $map_path) {
2366 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2367 return (undef, undef);
2368 }
2369 my ($rev, $commit) = $self->rev_map_max(1);
2370 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2371 return ($rev, $commit);
2372 }
2373
2374 sub get_fetch_range {
2375 my ($self, $min, $max) = @_;
2376 $max ||= $self->ra->get_latest_revnum;
2377 $min ||= $self->rev_map_max;
2378 (++$min, $max);
2379 }
2380
2381 sub tmp_config {
2382 my (@args) = @_;
2383 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2384 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2385 if (! -f $config && -f $old_def_config) {
2386 rename $old_def_config, $config or
2387 die "Failed rename $old_def_config => $config: $!\n";
2388 }
2389 my $old_config = $ENV{GIT_CONFIG};
2390 $ENV{GIT_CONFIG} = $config;
2391 $@ = undef;
2392 my @ret = eval {
2393 unless (-f $config) {
2394 mkfile($config);
2395 open my $fh, '>', $config or
2396 die "Can't open $config: $!\n";
2397 print $fh "; This file is used internally by ",
2398 "git-svn\n" or die
2399 "Couldn't write to $config: $!\n";
2400 print $fh "; You should not have to edit it\n" or
2401 die "Couldn't write to $config: $!\n";
2402 close $fh or die "Couldn't close $config: $!\n";
2403 }
2404 command('config', @args);
2405 };
2406 my $err = $@;
2407 if (defined $old_config) {
2408 $ENV{GIT_CONFIG} = $old_config;
2409 } else {
2410 delete $ENV{GIT_CONFIG};
2411 }
2412 die $err if $err;
2413 wantarray ? @ret : $ret[0];
2414 }
2415
2416 sub tmp_index_do {
2417 my ($self, $sub) = @_;
2418 my $old_index = $ENV{GIT_INDEX_FILE};
2419 $ENV{GIT_INDEX_FILE} = $self->{index};
2420 $@ = undef;
2421 my @ret = eval {
2422 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2423 mkpath([$dir]) unless -d $dir;
2424 &$sub;
2425 };
2426 my $err = $@;
2427 if (defined $old_index) {
2428 $ENV{GIT_INDEX_FILE} = $old_index;
2429 } else {
2430 delete $ENV{GIT_INDEX_FILE};
2431 }
2432 die $err if $err;
2433 wantarray ? @ret : $ret[0];
2434 }
2435
2436 sub assert_index_clean {
2437 my ($self, $treeish) = @_;
2438
2439 $self->tmp_index_do(sub {
2440 command_noisy('read-tree', $treeish) unless -e $self->{index};
2441 my $x = command_oneline('write-tree');
2442 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2443 /^tree ($::sha1)/mo);
2444 return if $y eq $x;
2445
2446 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2447 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2448 command_noisy('read-tree', $treeish);
2449 $x = command_oneline('write-tree');
2450 if ($y ne $x) {
2451 ::fatal "trees ($treeish) $y != $x\n",
2452 "Something is seriously wrong...";
2453 }
2454 });
2455 }
2456
2457 sub get_commit_parents {
2458 my ($self, $log_entry) = @_;
2459 my (%seen, @ret, @tmp);
2460 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2461 if (my $ip = $self->{inject_parents}) {
2462 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2463 push @tmp, $commit;
2464 }
2465 }
2466 if (my $cur = ::verify_ref($self->refname.'^0')) {
2467 push @tmp, $cur;
2468 }
2469 if (my $ipd = $self->{inject_parents_dcommit}) {
2470 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2471 push @tmp, @$commit;
2472 }
2473 }
2474 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2475 while (my $p = shift @tmp) {
2476 next if $seen{$p};
2477 $seen{$p} = 1;
2478 push @ret, $p;
2479 }
2480 @ret;
2481 }
2482
2483 sub rewrite_root {
2484 my ($self) = @_;
2485 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2486 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2487 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2488 if ($rwr) {
2489 $rwr =~ s#/+$##;
2490 if ($rwr !~ m#^[a-z\+]+://#) {
2491 die "$rwr is not a valid URL (key: $k)\n";
2492 }
2493 }
2494 $self->{-rewrite_root} = $rwr;
2495 }
2496
2497 sub rewrite_uuid {
2498 my ($self) = @_;
2499 return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2500 my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2501 my $rwid = eval { command_oneline(qw/config --get/, $k) };
2502 if ($rwid) {
2503 $rwid =~ s#/+$##;
2504 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2505 die "$rwid is not a valid UUID (key: $k)\n";
2506 }
2507 }
2508 $self->{-rewrite_uuid} = $rwid;
2509 }
2510
2511 sub metadata_url {
2512 my ($self) = @_;
2513 ($self->rewrite_root || $self->{url}) .
2514 (length $self->{path} ? '/' . $self->{path} : '');
2515 }
2516
2517 sub full_url {
2518 my ($self) = @_;
2519 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2520 }
2521
2522
2523 sub set_commit_header_env {
2524 my ($log_entry) = @_;
2525 my %env;
2526 foreach my $ned (qw/NAME EMAIL DATE/) {
2527 foreach my $ac (qw/AUTHOR COMMITTER/) {
2528 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2529 }
2530 }
2531
2532 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2533 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2534 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2535
2536 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2537 ? $log_entry->{commit_name}
2538 : $log_entry->{name};
2539 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2540 ? $log_entry->{commit_email}
2541 : $log_entry->{email};
2542 \%env;
2543 }
2544
2545 sub restore_commit_header_env {
2546 my ($env) = @_;
2547 foreach my $ned (qw/NAME EMAIL DATE/) {
2548 foreach my $ac (qw/AUTHOR COMMITTER/) {
2549 my $k = "GIT_${ac}_${ned}";
2550 if (defined $env->{$k}) {
2551 $ENV{$k} = $env->{$k};
2552 } else {
2553 delete $ENV{$k};
2554 }
2555 }
2556 }
2557 }
2558
2559 sub gc {
2560 command_noisy('gc', '--auto');
2561 };
2562
2563 sub do_git_commit {
2564 my ($self, $log_entry) = @_;
2565 my $lr = $self->last_rev;
2566 if (defined $lr && $lr >= $log_entry->{revision}) {
2567 die "Last fetched revision of ", $self->refname,
2568 " was r$lr, but we are about to fetch: ",
2569 "r$log_entry->{revision}!\n";
2570 }
2571 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2572 croak "$log_entry->{revision} = $c already exists! ",
2573 "Why are we refetching it?\n";
2574 }
2575 my $old_env = set_commit_header_env($log_entry);
2576 my $tree = $log_entry->{tree};
2577 if (!defined $tree) {
2578 $tree = $self->tmp_index_do(sub {
2579 command_oneline('write-tree') });
2580 }
2581 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2582
2583 my @exec = ('git', 'commit-tree', $tree);
2584 foreach ($self->get_commit_parents($log_entry)) {
2585 push @exec, '-p', $_;
2586 }
2587 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2588 or croak $!;
2589 binmode $msg_fh;
2590
2591 # we always get UTF-8 from SVN, but we may want our commits in
2592 # a different encoding.
2593 if (my $enc = Git::config('i18n.commitencoding')) {
2594 require Encode;
2595 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2596 }
2597 print $msg_fh $log_entry->{log} or croak $!;
2598 restore_commit_header_env($old_env);
2599 unless ($self->no_metadata) {
2600 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2601 or croak $!;
2602 }
2603 $msg_fh->flush == 0 or croak $!;
2604 close $msg_fh or croak $!;
2605 chomp(my $commit = do { local $/; <$out_fh> });
2606 close $out_fh or croak $!;
2607 waitpid $pid, 0;
2608 croak $? if $?;
2609 if ($commit !~ /^$::sha1$/o) {
2610 die "Failed to commit, invalid sha1: $commit\n";
2611 }
2612
2613 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2614
2615 $self->{last_rev} = $log_entry->{revision};
2616 $self->{last_commit} = $commit;
2617 print "r$log_entry->{revision}" unless $::_q > 1;
2618 if (defined $log_entry->{svm_revision}) {
2619 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2620 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2621 0, $self->svm_uuid);
2622 }
2623 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2624 if (--$_gc_nr == 0) {
2625 $_gc_nr = $_gc_period;
2626 gc();
2627 }
2628 return $commit;
2629 }
2630
2631 sub match_paths {
2632 my ($self, $paths, $r) = @_;
2633 return 1 if $self->{path} eq '';
2634 if (my $path = $paths->{"/$self->{path}"}) {
2635 return ($path->{action} eq 'D') ? 0 : 1;
2636 }
2637 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2638 if (grep /$self->{path_regex}/, keys %$paths) {
2639 return 1;
2640 }
2641 my $c = '';
2642 foreach (split m#/#, $self->{path}) {
2643 $c .= "/$_";
2644 next unless ($paths->{$c} &&
2645 ($paths->{$c}->{action} =~ /^[AR]$/));
2646 if ($self->ra->check_path($self->{path}, $r) ==
2647 $SVN::Node::dir) {
2648 return 1;
2649 }
2650 }
2651 return 0;
2652 }
2653
2654 sub find_parent_branch {
2655 my ($self, $paths, $rev) = @_;
2656 return undef unless $self->follow_parent;
2657 unless (defined $paths) {
2658 my $err_handler = $SVN::Error::handler;
2659 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2660 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2661 sub { $paths = $_[0] });
2662 $SVN::Error::handler = $err_handler;
2663 }
2664 return undef unless defined $paths;
2665
2666 # look for a parent from another branch:
2667 my @b_path_components = split m#/#, $self->{path};
2668 my @a_path_components;
2669 my $i;
2670 while (@b_path_components) {
2671 $i = $paths->{'/'.join('/', @b_path_components)};
2672 last if $i && defined $i->{copyfrom_path};
2673 unshift(@a_path_components, pop(@b_path_components));
2674 }
2675 return undef unless defined $i && defined $i->{copyfrom_path};
2676 my $branch_from = $i->{copyfrom_path};
2677 if (@a_path_components) {
2678 print STDERR "branch_from: $branch_from => ";
2679 $branch_from .= '/'.join('/', @a_path_components);
2680 print STDERR $branch_from, "\n";
2681 }
2682 my $r = $i->{copyfrom_rev};
2683 my $repos_root = $self->ra->{repos_root};
2684 my $url = $self->ra->{url};
2685 my $new_url = $url . $branch_from;
2686 print STDERR "Found possible branch point: ",
2687 "$new_url => ", $self->full_url, ", $r\n"
2688 unless $::_q > 1;
2689 $branch_from =~ s#^/##;
2690 my $gs = $self->other_gs($new_url, $url,
2691 $branch_from, $r, $self->{ref_id});
2692 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2693 {
2694 my ($base, $head);
2695 if (!defined $r0 || !defined $parent) {
2696 ($base, $head) = parse_revision_argument(0, $r);
2697 } else {
2698 if ($r0 < $r) {
2699 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2700 0, 1, sub { $base = $_[1] - 1 });
2701 }
2702 }
2703 if (defined $base && $base <= $r) {
2704 $gs->fetch($base, $r);
2705 }
2706 ($r0, $parent) = $gs->find_rev_before($r, 1);
2707 }
2708 if (defined $r0 && defined $parent) {
2709 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2710 unless $::_q > 1;
2711 my $ed;
2712 if ($self->ra->can_do_switch) {
2713 $self->assert_index_clean($parent);
2714 print STDERR "Following parent with do_switch\n"
2715 unless $::_q > 1;
2716 # do_switch works with svn/trunk >= r22312, but that
2717 # is not included with SVN 1.4.3 (the latest version
2718 # at the moment), so we can't rely on it
2719 $self->{last_rev} = $r0;
2720 $self->{last_commit} = $parent;
2721 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2722 $gs->ra->gs_do_switch($r0, $rev, $gs,
2723 $self->full_url, $ed)
2724 or die "SVN connection failed somewhere...\n";
2725 } elsif ($self->ra->trees_match($new_url, $r0,
2726 $self->full_url, $rev)) {
2727 print STDERR "Trees match:\n",
2728 " $new_url\@$r0\n",
2729 " ${\$self->full_url}\@$rev\n",
2730 "Following parent with no changes\n"
2731 unless $::_q > 1;
2732 $self->tmp_index_do(sub {
2733 command_noisy('read-tree', $parent);
2734 });
2735 $self->{last_commit} = $parent;
2736 } else {
2737 print STDERR "Following parent with do_update\n"
2738 unless $::_q > 1;
2739 $ed = SVN::Git::Fetcher->new($self);
2740 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2741 or die "SVN connection failed somewhere...\n";
2742 }
2743 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2744 return $self->make_log_entry($rev, [$parent], $ed);
2745 }
2746 return undef;
2747 }
2748
2749 sub do_fetch {
2750 my ($self, $paths, $rev) = @_;
2751 my $ed;
2752 my ($last_rev, @parents);
2753 if (my $lc = $self->last_commit) {
2754 # we can have a branch that was deleted, then re-added
2755 # under the same name but copied from another path, in
2756 # which case we'll have multiple parents (we don't
2757 # want to break the original ref, nor lose copypath info):
2758 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2759 push @{$log_entry->{parents}}, $lc;
2760 return $log_entry;
2761 }
2762 $ed = SVN::Git::Fetcher->new($self);
2763 $last_rev = $self->{last_rev};
2764 $ed->{c} = $lc;
2765 @parents = ($lc);
2766 } else {
2767 $last_rev = $rev;
2768 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2769 return $log_entry;
2770 }
2771 $ed = SVN::Git::Fetcher->new($self);
2772 }
2773 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2774 die "SVN connection failed somewhere...\n";
2775 }
2776 $self->make_log_entry($rev, \@parents, $ed);
2777 }
2778
2779 sub mkemptydirs {
2780 my ($self, $r) = @_;
2781
2782 sub scan {
2783 my ($r, $empty_dirs, $line) = @_;
2784 if (defined $r && $line =~ /^r(\d+)$/) {
2785 return 0 if $1 > $r;
2786 } elsif ($line =~ /^ \+empty_dir: (.+)$/) {
2787 $empty_dirs->{$1} = 1;
2788 } elsif ($line =~ /^ \-empty_dir: (.+)$/) {
2789 my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
2790 delete @$empty_dirs{@d};
2791 }
2792 1; # continue
2793 };
2794
2795 my %empty_dirs = ();
2796 my $gz_file = "$self->{dir}/unhandled.log.gz";
2797 if (-f $gz_file) {
2798 if (!$can_compress) {
2799 warn "Compress::Zlib could not be found; ",
2800 "empty directories in $gz_file will not be read\n";
2801 } else {
2802 my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
2803 die "Unable to open $gz_file: $!\n";
2804 my $line;
2805 while ($gz->gzreadline($line) > 0) {
2806 scan($r, \%empty_dirs, $line) or last;
2807 }
2808 $gz->gzclose;
2809 }
2810 }
2811
2812 if (open my $fh, '<', "$self->{dir}/unhandled.log") {
2813 binmode $fh or croak "binmode: $!";
2814 while (<$fh>) {
2815 scan($r, \%empty_dirs, $_) or last;
2816 }
2817 close $fh;
2818 }
2819
2820 my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2821 foreach my $d (sort keys %empty_dirs) {
2822 $d = uri_decode($d);
2823 $d =~ s/$strip//;
2824 next if -d $d;
2825 if (-e _) {
2826 warn "$d exists but is not a directory\n";
2827 } else {
2828 print "creating empty directory: $d\n";
2829 mkpath([$d]);
2830 }
2831 }
2832 }
2833
2834 sub get_untracked {
2835 my ($self, $ed) = @_;
2836 my @out;
2837 my $h = $ed->{empty};
2838 foreach (sort keys %$h) {
2839 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2840 push @out, " $act: " . uri_encode($_);
2841 warn "W: $act: $_\n";
2842 }
2843 foreach my $t (qw/dir_prop file_prop/) {
2844 $h = $ed->{$t} or next;
2845 foreach my $path (sort keys %$h) {
2846 my $ppath = $path eq '' ? '.' : $path;
2847 foreach my $prop (sort keys %{$h->{$path}}) {
2848 next if $SKIP_PROP{$prop};
2849 my $v = $h->{$path}->{$prop};
2850 my $t_ppath_prop = "$t: " .
2851 uri_encode($ppath) . ' ' .
2852 uri_encode($prop);
2853 if (defined $v) {
2854 push @out, " +$t_ppath_prop " .
2855 uri_encode($v);
2856 } else {
2857 push @out, " -$t_ppath_prop";
2858 }
2859 }
2860 }
2861 }
2862 foreach my $t (qw/absent_file absent_directory/) {
2863 $h = $ed->{$t} or next;
2864 foreach my $parent (sort keys %$h) {
2865 foreach my $path (sort @{$h->{$parent}}) {
2866 push @out, " $t: " .
2867 uri_encode("$parent/$path");
2868 warn "W: $t: $parent/$path ",
2869 "Insufficient permissions?\n";
2870 }
2871 }
2872 }
2873 \@out;
2874 }
2875
2876 # parse_svn_date(DATE)
2877 # --------------------
2878 # Given a date (in UTC) from Subversion, return a string in the format
2879 # "<TZ Offset> <local date/time>" that Git will use.
2880 #
2881 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
2882 # is true we'll convert it to the local timezone instead.
2883 sub parse_svn_date {
2884 my $date = shift || return '+0000 1970-01-01 00:00:00';
2885 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2886 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2887 croak "Unable to parse date: $date\n";
2888 my $parsed_date; # Set next.
2889
2890 if ($Git::SVN::_localtime) {
2891 # Translate the Subversion datetime to an epoch time.
2892 # Begin by switching ourselves to $date's timezone, UTC.
2893 my $old_env_TZ = $ENV{TZ};
2894 $ENV{TZ} = 'UTC';
2895
2896 my $epoch_in_UTC =
2897 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2898
2899 # Determine our local timezone (including DST) at the
2900 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2901 # value of TZ, if any, at the time we were run.
2902 if (defined $Git::SVN::Log::TZ) {
2903 $ENV{TZ} = $Git::SVN::Log::TZ;
2904 } else {
2905 delete $ENV{TZ};
2906 }
2907
2908 my $our_TZ =
2909 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2910
2911 # This converts $epoch_in_UTC into our local timezone.
2912 my ($sec, $min, $hour, $mday, $mon, $year,
2913 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2914
2915 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2916 $our_TZ, $year + 1900, $mon + 1,
2917 $mday, $hour, $min, $sec);
2918
2919 # Reset us to the timezone in effect when we entered
2920 # this routine.
2921 if (defined $old_env_TZ) {
2922 $ENV{TZ} = $old_env_TZ;
2923 } else {
2924 delete $ENV{TZ};
2925 }
2926 } else {
2927 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2928 }
2929
2930 return $parsed_date;
2931 }
2932
2933 sub other_gs {
2934 my ($self, $new_url, $url,
2935 $branch_from, $r, $old_ref_id) = @_;
2936 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2937 unless ($gs) {
2938 my $ref_id = $old_ref_id;
2939 $ref_id =~ s/\@\d+$//;
2940 $ref_id .= "\@$r";
2941 # just grow a tail if we're not unique enough :x
2942 $ref_id .= '-' while find_ref($ref_id);
2943 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2944 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2945 if ($u =~ s#^\Q$url\E(/|$)##) {
2946 $p = $u;
2947 $u = $url;
2948 $repo_id = $self->{repo_id};
2949 }
2950 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2951 }
2952 $gs
2953 }
2954
2955 sub call_authors_prog {
2956 my ($orig_author) = @_;
2957 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2958 my $author = `$::_authors_prog $orig_author`;
2959 if ($? != 0) {
2960 die "$::_authors_prog failed with exit code $?\n"
2961 }
2962 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2963 my ($name, $email) = ($1, $2);
2964 $email = undef if length $2 == 0;
2965 return [$name, $email];
2966 } else {
2967 die "Author: $orig_author: $::_authors_prog returned "
2968 . "invalid author format: $author\n";
2969 }
2970 }
2971
2972 sub check_author {
2973 my ($author) = @_;
2974 if (!defined $author || length $author == 0) {
2975 $author = '(no author)';
2976 }
2977 if (!defined $::users{$author}) {
2978 if (defined $::_authors_prog) {
2979 $::users{$author} = call_authors_prog($author);
2980 } elsif (defined $::_authors) {
2981 die "Author: $author not defined in $::_authors file\n";
2982 }
2983 }
2984 $author;
2985 }
2986
2987 sub find_extra_svk_parents {
2988 my ($self, $ed, $tickets, $parents) = @_;
2989 # aha! svk:merge property changed...
2990 my @tickets = split "\n", $tickets;
2991 my @known_parents;
2992 for my $ticket ( @tickets ) {
2993 my ($uuid, $path, $rev) = split /:/, $ticket;
2994 if ( $uuid eq $self->ra_uuid ) {
2995 my $url = $self->rewrite_root || $self->{url};
2996 my $repos_root = $url;
2997 my $branch_from = $path;
2998 $branch_from =~ s{^/}{};
2999 my $gs = $self->other_gs($repos_root."/".$branch_from,
3000 $url,
3001 $branch_from,
3002 $rev,
3003 $self->{ref_id});
3004 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3005 # wahey! we found it, but it might be
3006 # an old one (!)
3007 push @known_parents, [ $rev, $commit ];
3008 }
3009 }
3010 }
3011 # Ordering matters; highest-numbered commit merge tickets
3012 # first, as they may account for later merge ticket additions
3013 # or changes.
3014 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3015 for my $parent ( @known_parents ) {
3016 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3017 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3018 my $new;
3019 while ( <$msg_fh> ) {
3020 $new=1;last;
3021 }
3022 command_close_pipe($msg_fh, $ctx);
3023 if ( $new ) {
3024 print STDERR
3025 "Found merge parent (svk:merge ticket): $parent\n";
3026 push @$parents, $parent;
3027 }
3028 }
3029 }
3030
3031 sub lookup_svn_merge {
3032 my $uuid = shift;
3033 my $url = shift;
3034 my $merge = shift;
3035
3036 my ($source, $revs) = split ":", $merge;
3037 my $path = $source;
3038 $path =~ s{^/}{};
3039 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3040 if ( !$gs ) {
3041 warn "Couldn't find revmap for $url$source\n";
3042 return;
3043 }
3044 my @ranges = split ",", $revs;
3045 my ($tip, $tip_commit);
3046 my @merged_commit_ranges;
3047 # find the tip
3048 for my $range ( @ranges ) {
3049 my ($bottom, $top) = split "-", $range;
3050 $top ||= $bottom;
3051 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3052 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3053
3054 unless ($top_commit and $bottom_commit) {
3055 warn "W:unknown path/rev in svn:mergeinfo "
3056 ."dirprop: $source:$range\n";
3057 next;
3058 }
3059
3060 push @merged_commit_ranges,
3061 "$bottom_commit^..$top_commit";
3062
3063 if ( !defined $tip or $top > $tip ) {
3064 $tip = $top;
3065 $tip_commit = $top_commit;
3066 }
3067 }
3068 return ($tip_commit, @merged_commit_ranges);
3069 }
3070
3071 sub _rev_list {
3072 my ($msg_fh, $ctx) = command_output_pipe(
3073 "rev-list", @_,
3074 );
3075 my @rv;
3076 while ( <$msg_fh> ) {
3077 chomp;
3078 push @rv, $_;
3079 }
3080 command_close_pipe($msg_fh, $ctx);
3081 @rv;
3082 }
3083
3084 sub check_cherry_pick {
3085 my $base = shift;
3086 my $tip = shift;
3087 my @ranges = @_;
3088 my %commits = map { $_ => 1 }
3089 _rev_list("--no-merges", $tip, "--not", $base);
3090 for my $range ( @ranges ) {
3091 delete @commits{_rev_list($range)};
3092 }
3093 for my $commit (keys %commits) {
3094 if (has_no_changes($commit)) {
3095 delete $commits{$commit};
3096 }
3097 }
3098 return (keys %commits);
3099 }
3100
3101 sub has_no_changes {
3102 my $commit = shift;
3103
3104 my @revs = split / /, command_oneline(
3105 qw(rev-list --parents -1 -m), $commit);
3106
3107 # Commits with no parents, e.g. the start of a partial branch,
3108 # have changes by definition.
3109 return 1 if (@revs < 2);
3110
3111 # Commits with multiple parents, e.g a merge, have no changes
3112 # by definition.
3113 return 0 if (@revs > 2);
3114
3115 return (command_oneline("rev-parse", "$commit^{tree}") eq
3116 command_oneline("rev-parse", "$commit~1^{tree}"));
3117 }
3118
3119 BEGIN {
3120 memoize 'lookup_svn_merge';
3121 memoize 'check_cherry_pick';
3122 memoize 'has_no_changes';
3123 }
3124
3125 sub parents_exclude {
3126 my $parents = shift;
3127 my @commits = @_;
3128 return unless @commits;
3129
3130 my @excluded;
3131 my $excluded;
3132 do {
3133 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3134 $excluded = command_oneline(@cmd);
3135 if ( $excluded ) {
3136 my @new;
3137 my $found;
3138 for my $commit ( @commits ) {
3139 if ( $commit eq $excluded ) {
3140 push @excluded, $commit;
3141 $found++;
3142 last;
3143 }
3144 else {
3145 push @new, $commit;
3146 }
3147 }
3148 die "saw commit '$excluded' in rev-list output, "
3149 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3150 unless $found;
3151 @commits = @new;
3152 }
3153 }
3154 while ($excluded and @commits);
3155
3156 return @excluded;
3157 }
3158
3159
3160 # note: this function should only be called if the various dirprops
3161 # have actually changed
3162 sub find_extra_svn_parents {
3163 my ($self, $ed, $mergeinfo, $parents) = @_;
3164 # aha! svk:merge property changed...
3165
3166 # We first search for merged tips which are not in our
3167 # history. Then, we figure out which git revisions are in
3168 # that tip, but not this revision. If all of those revisions
3169 # are now marked as merge, we can add the tip as a parent.
3170 my @merges = split "\n", $mergeinfo;
3171 my @merge_tips;
3172 my $url = $self->rewrite_root || $self->{url};
3173 my $uuid = $self->ra_uuid;
3174 my %ranges;
3175 for my $merge ( @merges ) {
3176 my ($tip_commit, @ranges) =
3177 lookup_svn_merge( $uuid, $url, $merge );
3178 unless (!$tip_commit or
3179 grep { $_ eq $tip_commit } @$parents ) {
3180 push @merge_tips, $tip_commit;
3181 $ranges{$tip_commit} = \@ranges;
3182 } else {
3183 push @merge_tips, undef;
3184 }
3185 }
3186
3187 my %excluded = map { $_ => 1 }
3188 parents_exclude($parents, grep { defined } @merge_tips);
3189
3190 # check merge tips for new parents
3191 my @new_parents;
3192 for my $merge_tip ( @merge_tips ) {
3193 my $spec = shift @merges;
3194 next unless $merge_tip and $excluded{$merge_tip};
3195
3196 my $ranges = $ranges{$merge_tip};
3197
3198 # check out 'new' tips
3199 my $merge_base;
3200 eval {
3201 $merge_base = command_oneline(
3202 "merge-base",
3203 @$parents, $merge_tip,
3204 );
3205 };
3206 if ($@) {
3207 die "An error occurred during merge-base"
3208 unless $@->isa("Git::Error::Command");
3209
3210 warn "W: Cannot find common ancestor between ".
3211 "@$parents and $merge_tip. Ignoring merge info.\n";
3212 next;
3213 }
3214
3215 # double check that there are no missing non-merge commits
3216 my (@incomplete) = check_cherry_pick(
3217 $merge_base, $merge_tip,
3218 @$ranges,
3219 );
3220
3221 if ( @incomplete ) {
3222 warn "W:svn cherry-pick ignored ($spec) - missing "
3223 .@incomplete." commit(s) (eg $incomplete[0])\n";
3224 } else {
3225 warn
3226 "Found merge parent (svn:mergeinfo prop): ",
3227 $merge_tip, "\n";
3228 push @new_parents, $merge_tip;
3229 }
3230 }
3231
3232 # cater for merges which merge commits from multiple branches
3233 if ( @new_parents > 1 ) {
3234 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3235 for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3236 next if $i == $j;
3237 next unless $new_parents[$i];
3238 next unless $new_parents[$j];
3239 my $revs = command_oneline(
3240 "rev-list", "-1",
3241 "$new_parents[$i]..$new_parents[$j]",
3242 );
3243 if ( !$revs ) {
3244 undef($new_parents[$i]);
3245 }
3246 }
3247 }
3248 }
3249 push @$parents, grep { defined } @new_parents;
3250 }
3251
3252 sub make_log_entry {
3253 my ($self, $rev, $parents, $ed) = @_;
3254 my $untracked = $self->get_untracked($ed);
3255
3256 my @parents = @$parents;
3257 my $ps = $ed->{path_strip} || "";
3258 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3259 my $props = $ed->{dir_prop}{$path};
3260 if ( $props->{"svk:merge"} ) {
3261 $self->find_extra_svk_parents
3262 ($ed, $props->{"svk:merge"}, \@parents);
3263 }
3264 if ( $props->{"svn:mergeinfo"} ) {
3265 $self->find_extra_svn_parents
3266 ($ed,
3267 $props->{"svn:mergeinfo"},
3268 \@parents);
3269 }
3270 }
3271
3272 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3273 print $un "r$rev\n" or croak $!;
3274 print $un $_, "\n" foreach @$untracked;
3275 my %log_entry = ( parents => \@parents, revision => $rev,
3276 log => '');
3277
3278 my $headrev;
3279 my $logged = delete $self->{logged_rev_props};
3280 if (!$logged || $self->{-want_revprops}) {
3281 my $rp = $self->ra->rev_proplist($rev);
3282 foreach (sort keys %$rp) {
3283 my $v = $rp->{$_};
3284 if (/^svn:(author|date|log)$/) {
3285 $log_entry{$1} = $v;
3286 } elsif ($_ eq 'svm:headrev') {
3287 $headrev = $v;
3288 } else {
3289 print $un " rev_prop: ", uri_encode($_), ' ',
3290 uri_encode($v), "\n";
3291 }
3292 }
3293 } else {
3294 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3295 }
3296 close $un or croak $!;
3297
3298 $log_entry{date} = parse_svn_date($log_entry{date});
3299 $log_entry{log} .= "\n";
3300 my $author = $log_entry{author} = check_author($log_entry{author});
3301 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3302 : ($author, undef);
3303
3304 my ($commit_name, $commit_email) = ($name, $email);
3305 if ($_use_log_author) {
3306 my $name_field;
3307 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3308 $name_field = $1;
3309 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3310 $name_field = $1;
3311 }
3312 if (!defined $name_field) {
3313 if (!defined $email) {
3314 $email = $name;
3315 }
3316 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3317 ($name, $email) = ($1, $2);
3318 } elsif ($name_field =~ /(.*)@/) {
3319 ($name, $email) = ($1, $name_field);
3320 } else {
3321 ($name, $email) = ($name_field, $name_field);
3322 }
3323 }
3324 if (defined $headrev && $self->use_svm_props) {
3325 if ($self->rewrite_root) {
3326 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3327 "options set!\n";
3328 }
3329 if ($self->rewrite_uuid) {
3330 die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3331 "options set!\n";
3332 }
3333 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3334 # we don't want "SVM: initializing mirror for junk" ...
3335 return undef if $r == 0;
3336 my $svm = $self->svm;
3337 if ($uuid ne $svm->{uuid}) {
3338 die "UUID mismatch on SVM path:\n",
3339 "expected: $svm->{uuid}\n",
3340 " got: $uuid\n";
3341 }
3342 my $full_url = $self->full_url;
3343 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3344 die "Failed to replace '$svm->{replace}' with ",
3345 "'$svm->{source}' in $full_url\n";
3346 # throw away username for storing in records
3347 remove_username($full_url);
3348 $log_entry{metadata} = "$full_url\@$r $uuid";
3349 $log_entry{svm_revision} = $r;
3350 $email ||= "$author\@$uuid";
3351 $commit_email ||= "$author\@$uuid";
3352 } elsif ($self->use_svnsync_props) {
3353 my $full_url = $self->svnsync->{url};
3354 $full_url .= "/$self->{path}" if length $self->{path};
3355 remove_username($full_url);
3356 my $uuid = $self->svnsync->{uuid};
3357 $log_entry{metadata} = "$full_url\@$rev $uuid";
3358 $email ||= "$author\@$uuid";
3359 $commit_email ||= "$author\@$uuid";
3360 } else {
3361 my $url = $self->metadata_url;
3362 remove_username($url);
3363 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3364 $log_entry{metadata} = "$url\@$rev " . $uuid;
3365 $email ||= "$author\@" . $uuid;
3366 $commit_email ||= "$author\@" . $uuid;
3367 }
3368 $log_entry{name} = $name;
3369 $log_entry{email} = $email;
3370 $log_entry{commit_name} = $commit_name;
3371 $log_entry{commit_email} = $commit_email;
3372 \%log_entry;
3373 }
3374
3375 sub fetch {
3376 my ($self, $min_rev, $max_rev, @parents) = @_;
3377 my ($last_rev, $last_commit) = $self->last_rev_commit;
3378 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3379 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3380 }
3381
3382 sub set_tree_cb {
3383 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3384 $self->{inject_parents} = { $rev => $tree };
3385 $self->fetch(undef, undef);
3386 }
3387
3388 sub set_tree {
3389 my ($self, $tree) = (shift, shift);
3390 my $log_entry = ::get_commit_entry($tree);
3391 unless ($self->{last_rev}) {
3392 ::fatal("Must have an existing revision to commit");
3393 }
3394 my %ed_opts = ( r => $self->{last_rev},
3395 log => $log_entry->{log},
3396 ra => $self->ra,
3397 tree_a => $self->{last_commit},
3398 tree_b => $tree,
3399 editor_cb => sub {
3400 $self->set_tree_cb($log_entry, $tree, @_) },
3401 svn_path => $self->{path} );
3402 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3403 print "No changes\nr$self->{last_rev} = $tree\n";
3404 }
3405 }
3406
3407 sub rebuild_from_rev_db {
3408 my ($self, $path) = @_;
3409 my $r = -1;
3410 open my $fh, '<', $path or croak "open: $!";
3411 binmode $fh or croak "binmode: $!";
3412 while (<$fh>) {
3413 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3414 chomp($_);
3415 ++$r;
3416 next if $_ eq ('0' x 40);
3417 $self->rev_map_set($r, $_);
3418 print "r$r = $_\n";
3419 }
3420 close $fh or croak "close: $!";
3421 unlink $path or croak "unlink: $!";
3422 }
3423
3424 sub rebuild {
3425 my ($self) = @_;
3426 my $map_path = $self->map_path;
3427 my $partial = (-e $map_path && ! -z $map_path);
3428 return unless ::verify_ref($self->refname.'^0');
3429 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3430 my $rev_db = $self->rev_db_path;
3431 $self->rebuild_from_rev_db($rev_db);
3432 if ($self->use_svm_props) {
3433 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3434 $self->rebuild_from_rev_db($svm_rev_db);
3435 }
3436 $self->unlink_rev_db_symlink;
3437 return;
3438 }
3439 print "Rebuilding $map_path ...\n" if (!$partial);
3440 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3441 (undef, undef));
3442 my ($log, $ctx) =
3443 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3444 ($head ? "$head.." : "") . $self->refname,
3445 '--');
3446 my $metadata_url = $self->metadata_url;
3447 remove_username($metadata_url);
3448 my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3449 my $c;
3450 while (<$log>) {
3451 if ( m{^commit ($::sha1)$} ) {
3452 $c = $1;
3453 next;
3454 }
3455 next unless s{^\s*(git-svn-id:)}{$1};
3456 my ($url, $rev, $uuid) = ::extract_metadata($_);
3457 remove_username($url);
3458
3459 # ignore merges (from set-tree)
3460 next if (!defined $rev || !$uuid);
3461
3462 # if we merged or otherwise started elsewhere, this is
3463 # how we break out of it
3464 if (($uuid ne $svn_uuid) ||
3465 ($metadata_url && $url && ($url ne $metadata_url))) {
3466 next;
3467 }
3468 if ($partial && $head) {
3469 print "Partial-rebuilding $map_path ...\n";
3470 print "Currently at $base_rev = $head\n";
3471 $head = undef;
3472 }
3473
3474 $self->rev_map_set($rev, $c);
3475 print "r$rev = $c\n";
3476 }
3477 command_close_pipe($log, $ctx);
3478 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3479 my $rev_db_path = $self->rev_db_path;
3480 if (-f $self->rev_db_path) {
3481 unlink $self->rev_db_path or croak "unlink: $!";
3482 }
3483 $self->unlink_rev_db_symlink;
3484 }
3485
3486 # rev_map:
3487 # Tie::File seems to be prone to offset errors if revisions get sparse,
3488 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
3489 # one of my favorite modules is out :< Next up would be one of the DBM
3490 # modules, but I'm not sure which is most portable...
3491 #
3492 # This is the replacement for the rev_db format, which was too big
3493 # and inefficient for large repositories with a lot of sparse history
3494 # (mainly tags)
3495 #
3496 # The format is this:
3497 # - 24 bytes for every record,
3498 # * 4 bytes for the integer representing an SVN revision number
3499 # * 20 bytes representing the sha1 of a git commit
3500 # - No empty padding records like the old format
3501 # (except the last record, which can be overwritten)
3502 # - new records are written append-only since SVN revision numbers
3503 # increase monotonically
3504 # - lookups on SVN revision number are done via a binary search
3505 # - Piping the file to xxd -c24 is a good way of dumping it for
3506 # viewing or editing (piped back through xxd -r), should the need
3507 # ever arise.
3508 # - The last record can be padding revision with an all-zero sha1
3509 # This is used to optimize fetch performance when using multiple
3510 # "fetch" directives in .git/config
3511 #
3512 # These files are disposable unless noMetadata or useSvmProps is set
3513
3514 sub _rev_map_set {
3515 my ($fh, $rev, $commit) = @_;
3516
3517 binmode $fh or croak "binmode: $!";
3518 my $size = (stat($fh))[7];
3519 ($size % 24) == 0 or croak "inconsistent size: $size";
3520
3521 my $wr_offset = 0;
3522 if ($size > 0) {
3523 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3524 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3525 $read == 24 or croak "read only $read bytes (!= 24)";
3526 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3527 if ($last_commit eq ('0' x40)) {
3528 if ($size >= 48) {
3529 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3530 $read = sysread($fh, $buf, 24) or
3531 croak "read: $!";
3532 $read == 24 or
3533 croak "read only $read bytes (!= 24)";
3534 ($last_rev, $last_commit) =
3535 unpack(rev_map_fmt, $buf);
3536 if ($last_commit eq ('0' x40)) {
3537 croak "inconsistent .rev_map\n";
3538 }
3539 }
3540 if ($last_rev >= $rev) {
3541 croak "last_rev is higher!: $last_rev >= $rev";
3542 }
3543 $wr_offset = -24;
3544 }
3545 }
3546 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3547 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3548 croak "write: $!";
3549 }
3550
3551 sub _rev_map_reset {
3552 my ($fh, $rev, $commit) = @_;
3553 my $c = _rev_map_get($fh, $rev);
3554 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3555 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3556 truncate $fh, $offset or croak "truncate: $!";
3557 }
3558
3559 sub mkfile {
3560 my ($path) = @_;
3561 unless (-e $path) {
3562 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3563 mkpath([$dir]) unless -d $dir;
3564 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3565 close $fh or die "Couldn't close (create) $path: $!\n";
3566 }
3567 }
3568
3569 sub rev_map_set {
3570 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3571 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3572 my $db = $self->map_path($uuid);
3573 my $db_lock = "$db.lock";
3574 my $sig;
3575 $update_ref ||= 0;
3576 if ($update_ref) {
3577 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3578 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3579 }
3580 mkfile($db);
3581
3582 $LOCKFILES{$db_lock} = 1;
3583 my $sync;
3584 # both of these options make our .rev_db file very, very important
3585 # and we can't afford to lose it because rebuild() won't work
3586 if ($self->use_svm_props || $self->no_metadata) {
3587 $sync = 1;
3588 copy($db, $db_lock) or die "rev_map_set(@_): ",
3589 "Failed to copy: ",
3590 "$db => $db_lock ($!)\n";
3591 } else {
3592 rename $db, $db_lock or die "rev_map_set(@_): ",
3593 "Failed to rename: ",
3594 "$db => $db_lock ($!)\n";
3595 }
3596
3597 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3598 or croak "Couldn't open $db_lock: $!\n";
3599 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3600 _rev_map_set($fh, $rev, $commit);
3601 if ($sync) {
3602 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3603 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3604 }
3605 close $fh or croak $!;
3606 if ($update_ref) {
3607 $_head = $self;
3608 my $note = "";
3609 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3610 command_noisy('update-ref', '-m', "r$rev$note",
3611 $self->refname, $commit);
3612 }
3613 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3614 "$db_lock => $db ($!)\n";
3615 delete $LOCKFILES{$db_lock};
3616 if ($update_ref) {
3617 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3618 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3619 kill $sig, $$ if defined $sig;
3620 }
3621 }
3622
3623 # If want_commit, this will return an array of (rev, commit) where
3624 # commit _must_ be a valid commit in the archive.
3625 # Otherwise, it'll return the max revision (whether or not the
3626 # commit is valid or just a 0x40 placeholder).
3627 sub rev_map_max {
3628 my ($self, $want_commit) = @_;
3629 $self->rebuild;
3630 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3631 $want_commit ? ($r, $c) : $r;
3632 }
3633
3634 sub rev_map_max_norebuild {
3635 my ($self, $want_commit) = @_;
3636 my $map_path = $self->map_path;
3637 stat $map_path or return $want_commit ? (0, undef) : 0;
3638 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3639 binmode $fh or croak "binmode: $!";
3640 my $size = (stat($fh))[7];
3641 ($size % 24) == 0 or croak "inconsistent size: $size";
3642
3643 if ($size == 0) {
3644 close $fh or croak "close: $!";
3645 return $want_commit ? (0, undef) : 0;
3646 }
3647
3648 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3649 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3650 my ($r, $c) = unpack(rev_map_fmt, $buf);
3651 if ($want_commit && $c eq ('0' x40)) {
3652 if ($size < 48) {
3653 return $want_commit ? (0, undef) : 0;
3654 }
3655 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3656 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3657 ($r, $c) = unpack(rev_map_fmt, $buf);
3658 if ($c eq ('0'x40)) {
3659 croak "Penultimate record is all-zeroes in $map_path";
3660 }
3661 }
3662 close $fh or croak "close: $!";
3663 $want_commit ? ($r, $c) : $r;
3664 }
3665
3666 sub rev_map_get {
3667 my ($self, $rev, $uuid) = @_;
3668 my $map_path = $self->map_path($uuid);
3669 return undef unless -e $map_path;
3670
3671 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3672 my $c = _rev_map_get($fh, $rev);
3673 close($fh) or croak "close: $!";
3674 $c
3675 }
3676
3677 sub _rev_map_get {
3678 my ($fh, $rev) = @_;
3679
3680 binmode $fh or croak "binmode: $!";
3681 my $size = (stat($fh))[7];
3682 ($size % 24) == 0 or croak "inconsistent size: $size";
3683
3684 if ($size == 0) {
3685 return undef;
3686 }
3687
3688 my ($l, $u) = (0, $size - 24);
3689 my ($r, $c, $buf);
3690
3691 while ($l <= $u) {
3692 my $i = int(($l/24 + $u/24) / 2) * 24;
3693 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3694 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3695 my ($r, $c) = unpack(rev_map_fmt, $buf);
3696
3697 if ($r < $rev) {
3698 $l = $i + 24;
3699 } elsif ($r > $rev) {
3700 $u = $i - 24;
3701 } else { # $r == $rev
3702 return $c eq ('0' x 40) ? undef : $c;
3703 }
3704 }
3705 undef;
3706 }
3707
3708 # Finds the first svn revision that exists on (if $eq_ok is true) or
3709 # before $rev for the current branch. It will not search any lower
3710 # than $min_rev. Returns the git commit hash and svn revision number
3711 # if found, else (undef, undef).
3712 sub find_rev_before {
3713 my ($self, $rev, $eq_ok, $min_rev) = @_;
3714 --$rev unless $eq_ok;
3715 $min_rev ||= 1;
3716 my $max_rev = $self->rev_map_max;
3717 $rev = $max_rev if ($rev > $max_rev);
3718 while ($rev >= $min_rev) {
3719 if (my $c = $self->rev_map_get($rev)) {
3720 return ($rev, $c);
3721 }
3722 --$rev;
3723 }
3724 return (undef, undef);
3725 }
3726
3727 # Finds the first svn revision that exists on (if $eq_ok is true) or
3728 # after $rev for the current branch. It will not search any higher
3729 # than $max_rev. Returns the git commit hash and svn revision number
3730 # if found, else (undef, undef).
3731 sub find_rev_after {
3732 my ($self, $rev, $eq_ok, $max_rev) = @_;
3733 ++$rev unless $eq_ok;
3734 $max_rev ||= $self->rev_map_max;
3735 while ($rev <= $max_rev) {
3736 if (my $c = $self->rev_map_get($rev)) {
3737 return ($rev, $c);
3738 }
3739 ++$rev;
3740 }
3741 return (undef, undef);
3742 }
3743
3744 sub _new {
3745 my ($class, $repo_id, $ref_id, $path) = @_;
3746 unless (defined $repo_id && length $repo_id) {
3747 $repo_id = $Git::SVN::default_repo_id;
3748 }
3749 unless (defined $ref_id && length $ref_id) {
3750 $_prefix = '' unless defined($_prefix);
3751 $_[2] = $ref_id =
3752 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3753 }
3754 $_[1] = $repo_id;
3755 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3756
3757 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3758 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3759 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3760 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3761 if (-d $old_dir && ! -d $dir) {
3762 $dir = $old_dir;
3763 }
3764 }
3765
3766 $_[3] = $path = '' unless (defined $path);
3767 mkpath([$dir]);
3768 bless {
3769 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3770 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3771 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3772 }
3773
3774 # for read-only access of old .rev_db formats
3775 sub unlink_rev_db_symlink {
3776 my ($self) = @_;
3777 my $link = $self->rev_db_path;
3778 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3779 if (-l $link) {
3780 unlink $link or croak "unlink: $link failed!";
3781 }
3782 }
3783
3784 sub rev_db_path {
3785 my ($self, $uuid) = @_;
3786 my $db_path = $self->map_path($uuid);
3787 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3788 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3789 $db_path;
3790 }
3791
3792 # the new replacement for .rev_db
3793 sub map_path {
3794 my ($self, $uuid) = @_;
3795 $uuid ||= $self->ra_uuid;
3796 "$self->{map_root}.$uuid";
3797 }
3798
3799 sub uri_encode {
3800 my ($f) = @_;
3801 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3802 $f
3803 }
3804
3805 sub uri_decode {
3806 my ($f) = @_;
3807 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3808 $f
3809 }
3810
3811 sub remove_username {
3812 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3813 }
3814
3815 package Git::SVN::Prompt;
3816 use strict;
3817 use warnings;
3818 require SVN::Core;
3819 use vars qw/$_no_auth_cache $_username/;
3820
3821 sub simple {
3822 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3823 $may_save = undef if $_no_auth_cache;
3824 $default_username = $_username if defined $_username;
3825 if (defined $default_username && length $default_username) {
3826 if (defined $realm && length $realm) {
3827 print STDERR "Authentication realm: $realm\n";
3828 STDERR->flush;
3829 }
3830 $cred->username($default_username);
3831 } else {
3832 username($cred, $realm, $may_save, $pool);
3833 }
3834 $cred->password(_read_password("Password for '" .
3835 $cred->username . "': ", $realm));
3836 $cred->may_save($may_save);
3837 $SVN::_Core::SVN_NO_ERROR;
3838 }
3839
3840 sub ssl_server_trust {
3841 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3842 $may_save = undef if $_no_auth_cache;
3843 print STDERR "Error validating server certificate for '$realm':\n";
3844 {
3845 no warnings 'once';
3846 # All variables SVN::Auth::SSL::* are used only once,
3847 # so we're shutting up Perl warnings about this.
3848 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3849 print STDERR " - The certificate is not issued ",
3850 "by a trusted authority. Use the\n",
3851 " fingerprint to validate ",
3852 "the certificate manually!\n";
3853 }
3854 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3855 print STDERR " - The certificate hostname ",
3856 "does not match.\n";
3857 }
3858 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3859 print STDERR " - The certificate is not yet valid.\n";
3860 }
3861 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3862 print STDERR " - The certificate has expired.\n";
3863 }
3864 if ($failures & $SVN::Auth::SSL::OTHER) {
3865 print STDERR " - The certificate has ",
3866 "an unknown error.\n";
3867 }
3868 } # no warnings 'once'
3869 printf STDERR
3870 "Certificate information:\n".
3871 " - Hostname: %s\n".
3872 " - Valid: from %s until %s\n".
3873 " - Issuer: %s\n".
3874 " - Fingerprint: %s\n",
3875 map $cert_info->$_, qw(hostname valid_from valid_until
3876 issuer_dname fingerprint);
3877 my $choice;
3878 prompt:
3879 print STDERR $may_save ?
3880 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3881 "(R)eject or accept (t)emporarily? ";
3882 STDERR->flush;
3883 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3884 if ($choice =~ /^t$/i) {
3885 $cred->may_save(undef);
3886 } elsif ($choice =~ /^r$/i) {
3887 return -1;
3888 } elsif ($may_save && $choice =~ /^p$/i) {
3889 $cred->may_save($may_save);
3890 } else {
3891 goto prompt;
3892 }
3893 $cred->accepted_failures($failures);
3894 $SVN::_Core::SVN_NO_ERROR;
3895 }
3896
3897 sub ssl_client_cert {
3898 my ($cred, $realm, $may_save, $pool) = @_;
3899 $may_save = undef if $_no_auth_cache;
3900 print STDERR "Client certificate filename: ";
3901 STDERR->flush;
3902 chomp(my $filename = <STDIN>);
3903 $cred->cert_file($filename);
3904 $cred->may_save($may_save);
3905 $SVN::_Core::SVN_NO_ERROR;
3906 }
3907
3908 sub ssl_client_cert_pw {
3909 my ($cred, $realm, $may_save, $pool) = @_;
3910 $may_save = undef if $_no_auth_cache;
3911 $cred->password(_read_password("Password: ", $realm));
3912 $cred->may_save($may_save);
3913 $SVN::_Core::SVN_NO_ERROR;
3914 }
3915
3916 sub username {
3917 my ($cred, $realm, $may_save, $pool) = @_;
3918 $may_save = undef if $_no_auth_cache;
3919 if (defined $realm && length $realm) {
3920 print STDERR "Authentication realm: $realm\n";
3921 }
3922 my $username;
3923 if (defined $_username) {
3924 $username = $_username;
3925 } else {
3926 print STDERR "Username: ";
3927 STDERR->flush;
3928 chomp($username = <STDIN>);
3929 }
3930 $cred->username($username);
3931 $cred->may_save($may_save);
3932 $SVN::_Core::SVN_NO_ERROR;
3933 }
3934
3935 sub _read_password {
3936 my ($prompt, $realm) = @_;
3937 print STDERR $prompt;
3938 STDERR->flush;
3939 require Term::ReadKey;
3940 Term::ReadKey::ReadMode('noecho');
3941 my $password = '';
3942 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3943 last if $key =~ /[\012\015]/; # \n\r
3944 $password .= $key;
3945 }
3946 Term::ReadKey::ReadMode('restore');
3947 print STDERR "\n";
3948 STDERR->flush;
3949 $password;
3950 }
3951
3952 package SVN::Git::Fetcher;
3953 use vars qw/@ISA/;
3954 use strict;
3955 use warnings;
3956 use Carp qw/croak/;
3957 use File::Temp qw/tempfile/;
3958 use IO::File qw//;
3959 use vars qw/$_ignore_regex/;
3960
3961 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
3962 sub new {
3963 my ($class, $git_svn, $switch_path) = @_;
3964 my $self = SVN::Delta::Editor->new;
3965 bless $self, $class;
3966 if (exists $git_svn->{last_commit}) {
3967 $self->{c} = $git_svn->{last_commit};
3968 $self->{empty_symlinks} =
3969 _mark_empty_symlinks($git_svn, $switch_path);
3970 }
3971 $self->{ignore_regex} = eval { command_oneline('config', '--get',
3972 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3973 $self->{empty} = {};
3974 $self->{dir_prop} = {};
3975 $self->{file_prop} = {};
3976 $self->{absent_dir} = {};
3977 $self->{absent_file} = {};
3978 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3979 $self;
3980 }
3981
3982 # this uses the Ra object, so it must be called before do_{switch,update},
3983 # not inside them (when the Git::SVN::Fetcher object is passed) to
3984 # do_{switch,update}
3985 sub _mark_empty_symlinks {
3986 my ($git_svn, $switch_path) = @_;
3987 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3988 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3989
3990 my %ret;
3991 my ($rev, $cmt) = $git_svn->last_rev_commit;
3992 return {} unless ($rev && $cmt);
3993
3994 # allow the warning to be printed for each revision we fetch to
3995 # ensure the user sees it. The user can also disable the workaround
3996 # on the repository even while git svn is running and the next
3997 # revision fetched will skip this expensive function.
3998 my $printed_warning;
3999 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4000 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4001 local $/ = "\0";
4002 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4003 $pfx .= '/' if length($pfx);
4004 while (<$ls>) {
4005 chomp;
4006 s/\A100644 blob $empty_blob\t//o or next;
4007 unless ($printed_warning) {
4008 print STDERR "Scanning for empty symlinks, ",
4009 "this may take a while if you have ",
4010 "many empty files\n",
4011 "You may disable this with `",
4012 "git config svn.brokenSymlinkWorkaround ",
4013 "false'.\n",
4014 "This may be done in a different ",
4015 "terminal without restarting ",
4016 "git svn\n";
4017 $printed_warning = 1;
4018 }
4019 my $path = $_;
4020 my (undef, $props) =
4021 $git_svn->ra->get_file($pfx.$path, $rev, undef);
4022 if ($props->{'svn:special'}) {
4023 $ret{$path} = 1;
4024 }
4025 }
4026 command_close_pipe($ls, $ctx);
4027 \%ret;
4028 }
4029
4030 # returns true if a given path is inside a ".git" directory
4031 sub in_dot_git {
4032 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4033 }
4034
4035 # return value: 0 -- don't ignore, 1 -- ignore
4036 sub is_path_ignored {
4037 my ($self, $path) = @_;
4038 return 1 if in_dot_git($path);
4039 return 1 if defined($self->{ignore_regex}) &&
4040 $path =~ m!$self->{ignore_regex}!;
4041 return 0 unless defined($_ignore_regex);
4042 return 1 if $path =~ m!$_ignore_regex!o;
4043 return 0;
4044 }
4045
4046 sub set_path_strip {
4047 my ($self, $path) = @_;
4048 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4049 }
4050
4051 sub open_root {
4052 { path => '' };
4053 }
4054
4055 sub open_directory {
4056 my ($self, $path, $pb, $rev) = @_;
4057 { path => $path };
4058 }
4059
4060 sub git_path {
4061 my ($self, $path) = @_;
4062 if ($self->{path_strip}) {
4063 $path =~ s!$self->{path_strip}!! or
4064 die "Failed to strip path '$path' ($self->{path_strip})\n";
4065 }
4066 $path;
4067 }
4068
4069 sub delete_entry {
4070 my ($self, $path, $rev, $pb) = @_;
4071 return undef if $self->is_path_ignored($path);
4072
4073 my $gpath = $self->git_path($path);
4074 return undef if ($gpath eq '');
4075
4076 # remove entire directories.
4077 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4078 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4079 if ($tree) {
4080 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4081 -r --name-only -z/,
4082 $tree);
4083 local $/ = "\0";
4084 while (<$ls>) {
4085 chomp;
4086 my $rmpath = "$gpath/$_";
4087 $self->{gii}->remove($rmpath);
4088 print "\tD\t$rmpath\n" unless $::_q;
4089 }
4090 print "\tD\t$gpath/\n" unless $::_q;
4091 command_close_pipe($ls, $ctx);
4092 } else {
4093 $self->{gii}->remove($gpath);
4094 print "\tD\t$gpath\n" unless $::_q;
4095 }
4096 $self->{empty}->{$path} = 0;
4097 undef;
4098 }
4099
4100 sub open_file {
4101 my ($self, $path, $pb, $rev) = @_;
4102 my ($mode, $blob);
4103
4104 goto out if $self->is_path_ignored($path);
4105
4106 my $gpath = $self->git_path($path);
4107 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4108 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4109 unless (defined $mode && defined $blob) {
4110 die "$path was not found in commit $self->{c} (r$rev)\n";
4111 }
4112 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4113 $mode = '120000';
4114 }
4115 out:
4116 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4117 pool => SVN::Pool->new, action => 'M' };
4118 }
4119
4120 sub add_file {
4121 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4122 my $mode;
4123
4124 if (!$self->is_path_ignored($path)) {
4125 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4126 delete $self->{empty}->{$dir};
4127 $mode = '100644';
4128 }
4129 { path => $path, mode_a => $mode, mode_b => $mode,
4130 pool => SVN::Pool->new, action => 'A' };
4131 }
4132
4133 sub add_directory {
4134 my ($self, $path, $cp_path, $cp_rev) = @_;
4135 goto out if $self->is_path_ignored($path);
4136 my $gpath = $self->git_path($path);
4137 if ($gpath eq '') {
4138 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4139 -r --name-only -z/,
4140 $self->{c});
4141 local $/ = "\0";
4142 while (<$ls>) {
4143 chomp;
4144 $self->{gii}->remove($_);
4145 print "\tD\t$_\n" unless $::_q;
4146 }
4147 command_close_pipe($ls, $ctx);
4148 $self->{empty}->{$path} = 0;
4149 }
4150 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4151 delete $self->{empty}->{$dir};
4152 $self->{empty}->{$path} = 1;
4153 out:
4154 { path => $path };
4155 }
4156
4157 sub change_dir_prop {
4158 my ($self, $db, $prop, $value) = @_;
4159 return undef if $self->is_path_ignored($db->{path});
4160 $self->{dir_prop}->{$db->{path}} ||= {};
4161 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4162 undef;
4163 }
4164
4165 sub absent_directory {
4166 my ($self, $path, $pb) = @_;
4167 return undef if $self->is_path_ignored($path);
4168 $self->{absent_dir}->{$pb->{path}} ||= [];
4169 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4170 undef;
4171 }
4172
4173 sub absent_file {
4174 my ($self, $path, $pb) = @_;
4175 return undef if $self->is_path_ignored($path);
4176 $self->{absent_file}->{$pb->{path}} ||= [];
4177 push @{$self->{absent_file}->{$pb->{path}}}, $path;
4178 undef;
4179 }
4180
4181 sub change_file_prop {
4182 my ($self, $fb, $prop, $value) = @_;
4183 return undef if $self->is_path_ignored($fb->{path});
4184 if ($prop eq 'svn:executable') {
4185 if ($fb->{mode_b} != 120000) {
4186 $fb->{mode_b} = defined $value ? 100755 : 100644;
4187 }
4188 } elsif ($prop eq 'svn:special') {
4189 $fb->{mode_b} = defined $value ? 120000 : 100644;
4190 } else {
4191 $self->{file_prop}->{$fb->{path}} ||= {};
4192 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4193 }
4194 undef;
4195 }
4196
4197 sub apply_textdelta {
4198 my ($self, $fb, $exp) = @_;
4199 return undef if $self->is_path_ignored($fb->{path});
4200 my $fh = $::_repository->temp_acquire('svn_delta');
4201 # $fh gets auto-closed() by SVN::TxDelta::apply(),
4202 # (but $base does not,) so dup() it for reading in close_file
4203 open my $dup, '<&', $fh or croak $!;
4204 my $base = $::_repository->temp_acquire('git_blob');
4205
4206 if ($fb->{blob}) {
4207 my ($base_is_link, $size);
4208
4209 if ($fb->{mode_a} eq '120000' &&
4210 ! $self->{empty_symlinks}->{$fb->{path}}) {
4211 print $base 'link ' or die "print $!\n";
4212 $base_is_link = 1;
4213 }
4214 retry:
4215 $size = $::_repository->cat_blob($fb->{blob}, $base);
4216 die "Failed to read object $fb->{blob}" if ($size < 0);
4217
4218 if (defined $exp) {
4219 seek $base, 0, 0 or croak $!;
4220 my $got = ::md5sum($base);
4221 if ($got ne $exp) {
4222 my $err = "Checksum mismatch: ".
4223 "$fb->{path} $fb->{blob}\n" .
4224 "expected: $exp\n" .
4225 " got: $got\n";
4226 if ($base_is_link) {
4227 warn $err,
4228 "Retrying... (possibly ",
4229 "a bad symlink from SVN)\n";
4230 $::_repository->temp_reset($base);
4231 $base_is_link = 0;
4232 goto retry;
4233 }
4234 die $err;
4235 }
4236 }
4237 }
4238 seek $base, 0, 0 or croak $!;
4239 $fb->{fh} = $fh;
4240 $fb->{base} = $base;
4241 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4242 }
4243
4244 sub close_file {
4245 my ($self, $fb, $exp) = @_;
4246 return undef if $self->is_path_ignored($fb->{path});
4247
4248 my $hash;
4249 my $path = $self->git_path($fb->{path});
4250 if (my $fh = $fb->{fh}) {
4251 if (defined $exp) {
4252 seek($fh, 0, 0) or croak $!;
4253 my $got = ::md5sum($fh);
4254 if ($got ne $exp) {
4255 die "Checksum mismatch: $path\n",
4256 "expected: $exp\n got: $got\n";
4257 }
4258 }
4259 if ($fb->{mode_b} == 120000) {
4260 sysseek($fh, 0, 0) or croak $!;
4261 my $rd = sysread($fh, my $buf, 5);
4262
4263 if (!defined $rd) {
4264 croak "sysread: $!\n";
4265 } elsif ($rd == 0) {
4266 warn "$path has mode 120000",
4267 " but it points to nothing\n",
4268 "converting to an empty file with mode",
4269 " 100644\n";
4270 $fb->{mode_b} = '100644';
4271 } elsif ($buf ne 'link ') {
4272 warn "$path has mode 120000",
4273 " but is not a link\n";
4274 } else {
4275 my $tmp_fh = $::_repository->temp_acquire(
4276 'svn_hash');
4277 my $res;
4278 while ($res = sysread($fh, my $str, 1024)) {
4279 my $out = syswrite($tmp_fh, $str, $res);
4280 defined($out) && $out == $res
4281 or croak("write ",
4282 Git::temp_path($tmp_fh),
4283 ": $!\n");
4284 }
4285 defined $res or croak $!;
4286
4287 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4288 Git::temp_release($tmp_fh, 1);
4289 }
4290 }
4291
4292 $hash = $::_repository->hash_and_insert_object(
4293 Git::temp_path($fh));
4294 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4295
4296 Git::temp_release($fb->{base}, 1);
4297 Git::temp_release($fh, 1);
4298 } else {
4299 $hash = $fb->{blob} or die "no blob information\n";
4300 }
4301 $fb->{pool}->clear;
4302 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4303 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4304 undef;
4305 }
4306
4307 sub abort_edit {
4308 my $self = shift;
4309 $self->{nr} = $self->{gii}->{nr};
4310 delete $self->{gii};
4311 $self->SUPER::abort_edit(@_);
4312 }
4313
4314 sub close_edit {
4315 my $self = shift;
4316 $self->{git_commit_ok} = 1;
4317 $self->{nr} = $self->{gii}->{nr};
4318 delete $self->{gii};
4319 $self->SUPER::close_edit(@_);
4320 }
4321
4322 package SVN::Git::Editor;
4323 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4324 use strict;
4325 use warnings;
4326 use Carp qw/croak/;
4327 use IO::File;
4328
4329 sub new {
4330 my ($class, $opts) = @_;
4331 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4332 die "$_ required!\n" unless (defined $opts->{$_});
4333 }
4334
4335 my $pool = SVN::Pool->new;
4336 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4337 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4338 $opts->{r}, $mods);
4339
4340 # $opts->{ra} functions should not be used after this:
4341 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4342 $opts->{editor_cb}, $pool);
4343 my $self = SVN::Delta::Editor->new(@ce, $pool);
4344 bless $self, $class;
4345 foreach (qw/svn_path r tree_a tree_b/) {
4346 $self->{$_} = $opts->{$_};
4347 }
4348 $self->{url} = $opts->{ra}->{url};
4349 $self->{mods} = $mods;
4350 $self->{types} = $types;
4351 $self->{pool} = $pool;
4352 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4353 $self->{rm} = { };
4354 $self->{path_prefix} = length $self->{svn_path} ?
4355 "$self->{svn_path}/" : '';
4356 $self->{config} = $opts->{config};
4357 return $self;
4358 }
4359
4360 sub generate_diff {
4361 my ($tree_a, $tree_b) = @_;
4362 my @diff_tree = qw(diff-tree -z -r);
4363 if ($_cp_similarity) {
4364 push @diff_tree, "-C$_cp_similarity";
4365 } else {
4366 push @diff_tree, '-C';
4367 }
4368 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4369 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4370 push @diff_tree, $tree_a, $tree_b;
4371 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4372 local $/ = "\0";
4373 my $state = 'meta';
4374 my @mods;
4375 while (<$diff_fh>) {
4376 chomp $_; # this gets rid of the trailing "\0"
4377 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4378 ($::sha1)\s($::sha1)\s
4379 ([MTCRAD])\d*$/xo) {
4380 push @mods, { mode_a => $1, mode_b => $2,
4381 sha1_a => $3, sha1_b => $4,
4382 chg => $5 };
4383 if ($5 =~ /^(?:C|R)$/) {
4384 $state = 'file_a';
4385 } else {
4386 $state = 'file_b';
4387 }
4388 } elsif ($state eq 'file_a') {
4389 my $x = $mods[$#mods] or croak "Empty array\n";
4390 if ($x->{chg} !~ /^(?:C|R)$/) {
4391 croak "Error parsing $_, $x->{chg}\n";
4392 }
4393 $x->{file_a} = $_;
4394 $state = 'file_b';
4395 } elsif ($state eq 'file_b') {
4396 my $x = $mods[$#mods] or croak "Empty array\n";
4397 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4398 croak "Error parsing $_, $x->{chg}\n";
4399 }
4400 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4401 croak "Error parsing $_, $x->{chg}\n";
4402 }
4403 $x->{file_b} = $_;
4404 $state = 'meta';
4405 } else {
4406 croak "Error parsing $_\n";
4407 }
4408 }
4409 command_close_pipe($diff_fh, $ctx);
4410 \@mods;
4411 }
4412
4413 sub check_diff_paths {
4414 my ($ra, $pfx, $rev, $mods) = @_;
4415 my %types;
4416 $pfx .= '/' if length $pfx;
4417
4418 sub type_diff_paths {
4419 my ($ra, $types, $path, $rev) = @_;
4420 my @p = split m#/+#, $path;
4421 my $c = shift @p;
4422 unless (defined $types->{$c}) {
4423 $types->{$c} = $ra->check_path($c, $rev);
4424 }
4425 while (@p) {
4426 $c .= '/' . shift @p;
4427 next if defined $types->{$c};
4428 $types->{$c} = $ra->check_path($c, $rev);
4429 }
4430 }
4431
4432 foreach my $m (@$mods) {
4433 foreach my $f (qw/file_a file_b/) {
4434 next unless defined $m->{$f};
4435 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4436 if (length $pfx.$dir && ! defined $types{$dir}) {
4437 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4438 }
4439 }
4440 }
4441 \%types;
4442 }
4443
4444 sub split_path {
4445 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4446 }
4447
4448 sub repo_path {
4449 my ($self, $path) = @_;
4450 $self->{path_prefix}.(defined $path ? $path : '');
4451 }
4452
4453 sub url_path {
4454 my ($self, $path) = @_;
4455 if ($self->{url} =~ m#^https?://#) {
4456 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4457 }
4458 $self->{url} . '/' . $self->repo_path($path);
4459 }
4460
4461 sub rmdirs {
4462 my ($self) = @_;
4463 my $rm = $self->{rm};
4464 delete $rm->{''}; # we never delete the url we're tracking
4465 return unless %$rm;
4466
4467 foreach (keys %$rm) {
4468 my @d = split m#/#, $_;
4469 my $c = shift @d;
4470 $rm->{$c} = 1;
4471 while (@d) {
4472 $c .= '/' . shift @d;
4473 $rm->{$c} = 1;
4474 }
4475 }
4476 delete $rm->{$self->{svn_path}};
4477 delete $rm->{''}; # we never delete the url we're tracking
4478 return unless %$rm;
4479
4480 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4481 $self->{tree_b});
4482 local $/ = "\0";
4483 while (<$fh>) {
4484 chomp;
4485 my @dn = split m#/#, $_;
4486 while (pop @dn) {
4487 delete $rm->{join '/', @dn};
4488 }
4489 unless (%$rm) {
4490 close $fh;
4491 return;
4492 }
4493 }
4494 command_close_pipe($fh, $ctx);
4495
4496 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4497 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4498 $self->close_directory($bat->{$d}, $p);
4499 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4500 print "\tD+\t$d/\n" unless $::_q;
4501 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4502 delete $bat->{$d};
4503 }
4504 }
4505
4506 sub open_or_add_dir {
4507 my ($self, $full_path, $baton) = @_;
4508 my $t = $self->{types}->{$full_path};
4509 if (!defined $t) {
4510 die "$full_path not known in r$self->{r} or we have a bug!\n";
4511 }
4512 {
4513 no warnings 'once';
4514 # SVN::Node::none and SVN::Node::file are used only once,
4515 # so we're shutting up Perl's warnings about them.
4516 if ($t == $SVN::Node::none) {
4517 return $self->add_directory($full_path, $baton,
4518 undef, -1, $self->{pool});
4519 } elsif ($t == $SVN::Node::dir) {
4520 return $self->open_directory($full_path, $baton,
4521 $self->{r}, $self->{pool});
4522 } # no warnings 'once'
4523 print STDERR "$full_path already exists in repository at ",
4524 "r$self->{r} and it is not a directory (",
4525 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4526 } # no warnings 'once'
4527 exit 1;
4528 }
4529
4530 sub ensure_path {
4531 my ($self, $path) = @_;
4532 my $bat = $self->{bat};
4533 my $repo_path = $self->repo_path($path);
4534 return $bat->{''} unless (length $repo_path);
4535 my @p = split m#/+#, $repo_path;
4536 my $c = shift @p;
4537 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4538 while (@p) {
4539 my $c0 = $c;
4540 $c .= '/' . shift @p;
4541 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4542 }
4543 return $bat->{$c};
4544 }
4545
4546 # Subroutine to convert a globbing pattern to a regular expression.
4547 # From perl cookbook.
4548 sub glob2pat {
4549 my $globstr = shift;
4550 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4551 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4552 return '^' . $globstr . '$';
4553 }
4554
4555 sub check_autoprop {
4556 my ($self, $pattern, $properties, $file, $fbat) = @_;
4557 # Convert the globbing pattern to a regular expression.
4558 my $regex = glob2pat($pattern);
4559 # Check if the pattern matches the file name.
4560 if($file =~ m/($regex)/) {
4561 # Parse the list of properties to set.
4562 my @props = split(/;/, $properties);
4563 foreach my $prop (@props) {
4564 # Parse 'name=value' syntax and set the property.
4565 if ($prop =~ /([^=]+)=(.*)/) {
4566 my ($n,$v) = ($1,$2);
4567 for ($n, $v) {
4568 s/^\s+//; s/\s+$//;
4569 }
4570 $self->change_file_prop($fbat, $n, $v);
4571 }
4572 }
4573 }
4574 }
4575
4576 sub apply_autoprops {
4577 my ($self, $file, $fbat) = @_;
4578 my $conf_t = ${$self->{config}}{'config'};
4579 no warnings 'once';
4580 # Check [miscellany]/enable-auto-props in svn configuration.
4581 if (SVN::_Core::svn_config_get_bool(
4582 $conf_t,
4583 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4584 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4585 0)) {
4586 # Auto-props are enabled. Enumerate them to look for matches.
4587 my $callback = sub {
4588 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4589 };
4590 SVN::_Core::svn_config_enumerate(
4591 $conf_t,
4592 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4593 $callback);
4594 }
4595 }
4596
4597 sub A {
4598 my ($self, $m) = @_;
4599 my ($dir, $file) = split_path($m->{file_b});
4600 my $pbat = $self->ensure_path($dir);
4601 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4602 undef, -1);
4603 print "\tA\t$m->{file_b}\n" unless $::_q;
4604 $self->apply_autoprops($file, $fbat);
4605 $self->chg_file($fbat, $m);
4606 $self->close_file($fbat,undef,$self->{pool});
4607 }
4608
4609 sub C {
4610 my ($self, $m) = @_;
4611 my ($dir, $file) = split_path($m->{file_b});
4612 my $pbat = $self->ensure_path($dir);
4613 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4614 $self->url_path($m->{file_a}), $self->{r});
4615 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4616 $self->chg_file($fbat, $m);
4617 $self->close_file($fbat,undef,$self->{pool});
4618 }
4619
4620 sub delete_entry {
4621 my ($self, $path, $pbat) = @_;
4622 my $rpath = $self->repo_path($path);
4623 my ($dir, $file) = split_path($rpath);
4624 $self->{rm}->{$dir} = 1;
4625 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4626 }
4627
4628 sub R {
4629 my ($self, $m) = @_;
4630 my ($dir, $file) = split_path($m->{file_b});
4631 my $pbat = $self->ensure_path($dir);
4632 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4633 $self->url_path($m->{file_a}), $self->{r});
4634 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4635 $self->apply_autoprops($file, $fbat);
4636 $self->chg_file($fbat, $m);
4637 $self->close_file($fbat,undef,$self->{pool});
4638
4639 ($dir, $file) = split_path($m->{file_a});
4640 $pbat = $self->ensure_path($dir);
4641 $self->delete_entry($m->{file_a}, $pbat);
4642 }
4643
4644 sub M {
4645 my ($self, $m) = @_;
4646 my ($dir, $file) = split_path($m->{file_b});
4647 my $pbat = $self->ensure_path($dir);
4648 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4649 $pbat,$self->{r},$self->{pool});
4650 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4651 $self->chg_file($fbat, $m);
4652 $self->close_file($fbat,undef,$self->{pool});
4653 }
4654
4655 sub T { shift->M(@_) }
4656
4657 sub change_file_prop {
4658 my ($self, $fbat, $pname, $pval) = @_;
4659 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4660 }
4661
4662 sub _chg_file_get_blob ($$$$) {
4663 my ($self, $fbat, $m, $which) = @_;
4664 my $fh = $::_repository->temp_acquire("git_blob_$which");
4665 if ($m->{"mode_$which"} =~ /^120/) {
4666 print $fh 'link ' or croak $!;
4667 $self->change_file_prop($fbat,'svn:special','*');
4668 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4669 $self->change_file_prop($fbat,'svn:special',undef);
4670 }
4671 my $blob = $m->{"sha1_$which"};
4672 return ($fh,) if ($blob =~ /^0{40}$/);
4673 my $size = $::_repository->cat_blob($blob, $fh);
4674 croak "Failed to read object $blob" if ($size < 0);
4675 $fh->flush == 0 or croak $!;
4676 seek $fh, 0, 0 or croak $!;
4677
4678 my $exp = ::md5sum($fh);
4679 seek $fh, 0, 0 or croak $!;
4680 return ($fh, $exp);
4681 }
4682
4683 sub chg_file {
4684 my ($self, $fbat, $m) = @_;
4685 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4686 $self->change_file_prop($fbat,'svn:executable','*');
4687 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4688 $self->change_file_prop($fbat,'svn:executable',undef);
4689 }
4690 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4691 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4692 my $pool = SVN::Pool->new;
4693 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4694 if (-s $fh_a) {
4695 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4696 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4697 if (defined $res) {
4698 die "Unexpected result from send_txstream: $res\n",
4699 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4700 }
4701 } else {
4702 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4703 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4704 if ($got ne $exp_b);
4705 }
4706 Git::temp_release($fh_b, 1);
4707 Git::temp_release($fh_a, 1);
4708 $pool->clear;
4709 }
4710
4711 sub D {
4712 my ($self, $m) = @_;
4713 my ($dir, $file) = split_path($m->{file_b});
4714 my $pbat = $self->ensure_path($dir);
4715 print "\tD\t$m->{file_b}\n" unless $::_q;
4716 $self->delete_entry($m->{file_b}, $pbat);
4717 }
4718
4719 sub close_edit {
4720 my ($self) = @_;
4721 my ($p,$bat) = ($self->{pool}, $self->{bat});
4722 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4723 next if $_ eq '';
4724 $self->close_directory($bat->{$_}, $p);
4725 }
4726 $self->close_directory($bat->{''}, $p);
4727 $self->SUPER::close_edit($p);
4728 $p->clear;
4729 }
4730
4731 sub abort_edit {
4732 my ($self) = @_;
4733 $self->SUPER::abort_edit($self->{pool});
4734 }
4735
4736 sub DESTROY {
4737 my $self = shift;
4738 $self->SUPER::DESTROY(@_);
4739 $self->{pool}->clear;
4740 }
4741
4742 # this drives the editor
4743 sub apply_diff {
4744 my ($self) = @_;
4745 my $mods = $self->{mods};
4746 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4747 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4748 my $f = $m->{chg};
4749 if (defined $o{$f}) {
4750 $self->$f($m);
4751 } else {
4752 fatal("Invalid change type: $f");
4753 }
4754 }
4755 $self->rmdirs if $_rmdir;
4756 if (@$mods == 0) {
4757 $self->abort_edit;
4758 } else {
4759 $self->close_edit;
4760 }
4761 return scalar @$mods;
4762 }
4763
4764 package Git::SVN::Ra;
4765 use vars qw/@ISA $config_dir $_log_window_size/;
4766 use strict;
4767 use warnings;
4768 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4769
4770 BEGIN {
4771 # enforce temporary pool usage for some simple functions
4772 no strict 'refs';
4773 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4774 get_file/) {
4775 my $SUPER = "SUPER::$f";
4776 *$f = sub {
4777 my $self = shift;
4778 my $pool = SVN::Pool->new;
4779 my @ret = $self->$SUPER(@_,$pool);
4780 $pool->clear;
4781 wantarray ? @ret : $ret[0];
4782 };
4783 }
4784 }
4785
4786 sub _auth_providers () {
4787 [
4788 SVN::Client::get_simple_provider(),
4789 SVN::Client::get_ssl_server_trust_file_provider(),
4790 SVN::Client::get_simple_prompt_provider(
4791 \&Git::SVN::Prompt::simple, 2),
4792 SVN::Client::get_ssl_client_cert_file_provider(),
4793 SVN::Client::get_ssl_client_cert_prompt_provider(
4794 \&Git::SVN::Prompt::ssl_client_cert, 2),
4795 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4796 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4797 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4798 SVN::Client::get_username_provider(),
4799 SVN::Client::get_ssl_server_trust_prompt_provider(
4800 \&Git::SVN::Prompt::ssl_server_trust),
4801 SVN::Client::get_username_prompt_provider(
4802 \&Git::SVN::Prompt::username, 2)
4803 ]
4804 }
4805
4806 sub escape_uri_only {
4807 my ($uri) = @_;
4808 my @tmp;
4809 foreach (split m{/}, $uri) {
4810 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4811 push @tmp, $_;
4812 }
4813 join('/', @tmp);
4814 }
4815
4816 sub escape_url {
4817 my ($url) = @_;
4818 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4819 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4820 $url = "$scheme://$domain$uri";
4821 }
4822 $url;
4823 }
4824
4825 sub new {
4826 my ($class, $url) = @_;
4827 $url =~ s!/+$!!;
4828 return $RA if ($RA && $RA->{url} eq $url);
4829
4830 SVN::_Core::svn_config_ensure($config_dir, undef);
4831 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4832 my $config = SVN::Core::config_get_config($config_dir);
4833 $RA = undef;
4834 my $dont_store_passwords = 1;
4835 my $conf_t = ${$config}{'config'};
4836 {
4837 no warnings 'once';
4838 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4839 # produces warnings that variables are used only once.
4840 # I had not found the better way to shut them up, so
4841 # the warnings of type 'once' are disabled in this block.
4842 if (SVN::_Core::svn_config_get_bool($conf_t,
4843 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4844 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4845 1) == 0) {
4846 SVN::_Core::svn_auth_set_parameter($baton,
4847 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4848 bless (\$dont_store_passwords, "_p_void"));
4849 }
4850 if (SVN::_Core::svn_config_get_bool($conf_t,
4851 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4852 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4853 1) == 0) {
4854 $Git::SVN::Prompt::_no_auth_cache = 1;
4855 }
4856 } # no warnings 'once'
4857 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4858 config => $config,
4859 pool => SVN::Pool->new,
4860 auth_provider_callbacks => $callbacks);
4861 $self->{url} = $url;
4862 $self->{svn_path} = $url;
4863 $self->{repos_root} = $self->get_repos_root;
4864 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4865 $self->{cache} = { check_path => { r => 0, data => {} },
4866 get_dir => { r => 0, data => {} } };
4867 $RA = bless $self, $class;
4868 }
4869
4870 sub check_path {
4871 my ($self, $path, $r) = @_;
4872 my $cache = $self->{cache}->{check_path};
4873 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4874 return $cache->{data}->{$path};
4875 }
4876 my $pool = SVN::Pool->new;
4877 my $t = $self->SUPER::check_path($path, $r, $pool);
4878 $pool->clear;
4879 if ($r != $cache->{r}) {
4880 %{$cache->{data}} = ();
4881 $cache->{r} = $r;
4882 }
4883 $cache->{data}->{$path} = $t;
4884 }
4885
4886 sub get_dir {
4887 my ($self, $dir, $r) = @_;
4888 my $cache = $self->{cache}->{get_dir};
4889 if ($r == $cache->{r}) {
4890 if (my $x = $cache->{data}->{$dir}) {
4891 return wantarray ? @$x : $x->[0];
4892 }
4893 }
4894 my $pool = SVN::Pool->new;
4895 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4896 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4897 $pool->clear;
4898 if ($r != $cache->{r}) {
4899 %{$cache->{data}} = ();
4900 $cache->{r} = $r;
4901 }
4902 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4903 wantarray ? (\%dirents, $r, $props) : \%dirents;
4904 }
4905
4906 sub DESTROY {
4907 # do not call the real DESTROY since we store ourselves in $RA
4908 }
4909
4910 # get_log(paths, start, end, limit,
4911 # discover_changed_paths, strict_node_history, receiver)
4912 sub get_log {
4913 my ($self, @args) = @_;
4914 my $pool = SVN::Pool->new;
4915
4916 # svn_log_changed_path_t objects passed to get_log are likely to be
4917 # overwritten even if only the refs are copied to an external variable,
4918 # so we should dup the structures in their entirety. Using an
4919 # externally passed pool (instead of our temporary and quickly cleared
4920 # pool in Git::SVN::Ra) does not help matters at all...
4921 my $receiver = pop @args;
4922 my $prefix = "/".$self->{svn_path};
4923 $prefix =~ s#/+($)##;
4924 my $prefix_regex = qr#^\Q$prefix\E#;
4925 push(@args, sub {
4926 my ($paths) = $_[0];
4927 return &$receiver(@_) unless $paths;
4928 $_[0] = ();
4929 foreach my $p (keys %$paths) {
4930 my $i = $paths->{$p};
4931 # Make path relative to our url, not repos_root
4932 $p =~ s/$prefix_regex//;
4933 my %s = map { $_ => $i->$_; }
4934 qw/copyfrom_path copyfrom_rev action/;
4935 if ($s{'copyfrom_path'}) {
4936 $s{'copyfrom_path'} =~ s/$prefix_regex//;
4937 }
4938 $_[0]{$p} = \%s;
4939 }
4940 &$receiver(@_);
4941 });
4942
4943
4944 # the limit parameter was not supported in SVN 1.1.x, so we
4945 # drop it. Therefore, the receiver callback passed to it
4946 # is made aware of this limitation by being wrapped if
4947 # the limit passed to is being wrapped.
4948 if ($SVN::Core::VERSION le '1.2.0') {
4949 my $limit = splice(@args, 3, 1);
4950 if ($limit > 0) {
4951 my $receiver = pop @args;
4952 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4953 }
4954 }
4955 my $ret = $self->SUPER::get_log(@args, $pool);
4956 $pool->clear;
4957 $ret;
4958 }
4959
4960 sub trees_match {
4961 my ($self, $url1, $rev1, $url2, $rev2) = @_;
4962 my $ctx = SVN::Client->new(auth => _auth_providers);
4963 my $out = IO::File->new_tmpfile;
4964
4965 # older SVN (1.1.x) doesn't take $pool as the last parameter for
4966 # $ctx->diff(), so we'll create a default one
4967 my $pool = SVN::Pool->new_default_sub;
4968
4969 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4970 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4971 $out->flush;
4972 my $ret = (($out->stat)[7] == 0);
4973 close $out or croak $!;
4974
4975 $ret;
4976 }
4977
4978 sub get_commit_editor {
4979 my ($self, $log, $cb, $pool) = @_;
4980 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4981 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4982 }
4983
4984 sub gs_do_update {
4985 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4986 my $new = ($rev_a == $rev_b);
4987 my $path = $gs->{path};
4988
4989 if ($new && -e $gs->{index}) {
4990 unlink $gs->{index} or die
4991 "Couldn't unlink index: $gs->{index}: $!\n";
4992 }
4993 my $pool = SVN::Pool->new;
4994 $editor->set_path_strip($path);
4995 my (@pc) = split m#/#, $path;
4996 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4997 1, $editor, $pool);
4998 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4999
5000 # Since we can't rely on svn_ra_reparent being available, we'll
5001 # just have to do some magic with set_path to make it so
5002 # we only want a partial path.
5003 my $sp = '';
5004 my $final = join('/', @pc);
5005 while (@pc) {
5006 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5007 $sp .= '/' if length $sp;
5008 $sp .= shift @pc;
5009 }
5010 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5011
5012 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5013
5014 $reporter->finish_report($pool);
5015 $pool->clear;
5016 $editor->{git_commit_ok};
5017 }
5018
5019 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5020 # svn_ra_reparent didn't work before 1.4)
5021 sub gs_do_switch {
5022 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5023 my $path = $gs->{path};
5024 my $pool = SVN::Pool->new;
5025
5026 my $full_url = $self->{url};
5027 my $old_url = $full_url;
5028 $full_url .= '/' . $path if length $path;
5029 my ($ra, $reparented);
5030
5031 if ($old_url =~ m#^svn(\+ssh)?://# ||
5032 ($full_url =~ m#^https?://# &&
5033 escape_url($full_url) ne $full_url)) {
5034 $_[0] = undef;
5035 $self = undef;
5036 $RA = undef;
5037 $ra = Git::SVN::Ra->new($full_url);
5038 $ra_invalid = 1;
5039 } elsif ($old_url ne $full_url) {
5040 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5041 $self->{url} = $full_url;
5042 $reparented = 1;
5043 }
5044
5045 $ra ||= $self;
5046 $url_b = escape_url($url_b);
5047 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5048 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5049 $reporter->set_path('', $rev_a, 0, @lock, $pool);
5050 $reporter->finish_report($pool);
5051
5052 if ($reparented) {
5053 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5054 $self->{url} = $old_url;
5055 }
5056
5057 $pool->clear;
5058 $editor->{git_commit_ok};
5059 }
5060
5061 sub longest_common_path {
5062 my ($gsv, $globs) = @_;
5063 my %common;
5064 my $common_max = scalar @$gsv;
5065
5066 foreach my $gs (@$gsv) {
5067 my @tmp = split m#/#, $gs->{path};
5068 my $p = '';
5069 foreach (@tmp) {
5070 $p .= length($p) ? "/$_" : $_;
5071 $common{$p} ||= 0;
5072 $common{$p}++;
5073 }
5074 }
5075 $globs ||= [];
5076 $common_max += scalar @$globs;
5077 foreach my $glob (@$globs) {
5078 my @tmp = split m#/#, $glob->{path}->{left};
5079 my $p = '';
5080 foreach (@tmp) {
5081 $p .= length($p) ? "/$_" : $_;
5082 $common{$p} ||= 0;
5083 $common{$p}++;
5084 }
5085 }
5086
5087 my $longest_path = '';
5088 foreach (sort {length $b <=> length $a} keys %common) {
5089 if ($common{$_} == $common_max) {
5090 $longest_path = $_;
5091 last;
5092 }
5093 }
5094 $longest_path;
5095 }
5096
5097 sub gs_fetch_loop_common {
5098 my ($self, $base, $head, $gsv, $globs) = @_;
5099 return if ($base > $head);
5100 my $inc = $_log_window_size;
5101 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5102 my $longest_path = longest_common_path($gsv, $globs);
5103 my $ra_url = $self->{url};
5104 my $find_trailing_edge;
5105 while (1) {
5106 my %revs;
5107 my $err;
5108 my $err_handler = $SVN::Error::handler;
5109 $SVN::Error::handler = sub {
5110 ($err) = @_;
5111 skip_unknown_revs($err);
5112 };
5113 sub _cb {
5114 my ($paths, $r, $author, $date, $log) = @_;
5115 [ $paths,
5116 { author => $author, date => $date, log => $log } ];
5117 }
5118 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5119 sub { $revs{$_[1]} = _cb(@_) });
5120 if ($err) {
5121 print "Checked through r$max\r";
5122 } else {
5123 $find_trailing_edge = 1;
5124 }
5125 if ($err and $find_trailing_edge) {
5126 print STDERR "Path '$longest_path' ",
5127 "was probably deleted:\n",
5128 $err->expanded_message,
5129 "\nWill attempt to follow ",
5130 "revisions r$min .. r$max ",
5131 "committed before the deletion\n";
5132 my $hi = $max;
5133 while (--$hi >= $min) {
5134 my $ok;
5135 $self->get_log([$longest_path], $min, $hi,
5136 0, 1, 1, sub {
5137 $ok = $_[1];
5138 $revs{$_[1]} = _cb(@_) });
5139 if ($ok) {
5140 print STDERR "r$min .. r$ok OK\n";
5141 last;
5142 }
5143 }
5144 $find_trailing_edge = 0;
5145 }
5146 $SVN::Error::handler = $err_handler;
5147
5148 my %exists = map { $_->{path} => $_ } @$gsv;
5149 foreach my $r (sort {$a <=> $b} keys %revs) {
5150 my ($paths, $logged) = @{$revs{$r}};
5151
5152 foreach my $gs ($self->match_globs(\%exists, $paths,
5153 $globs, $r)) {
5154 if ($gs->rev_map_max >= $r) {
5155 next;
5156 }
5157 next unless $gs->match_paths($paths, $r);
5158 $gs->{logged_rev_props} = $logged;
5159 if (my $last_commit = $gs->last_commit) {
5160 $gs->assert_index_clean($last_commit);
5161 }
5162 my $log_entry = $gs->do_fetch($paths, $r);
5163 if ($log_entry) {
5164 $gs->do_git_commit($log_entry);
5165 }
5166 $INDEX_FILES{$gs->{index}} = 1;
5167 }
5168 foreach my $g (@$globs) {
5169 my $k = "svn-remote.$g->{remote}." .
5170 "$g->{t}-maxRev";
5171 Git::SVN::tmp_config($k, $r);
5172 }
5173 if ($ra_invalid) {
5174 $_[0] = undef;
5175 $self = undef;
5176 $RA = undef;
5177 $self = Git::SVN::Ra->new($ra_url);
5178 $ra_invalid = undef;
5179 }
5180 }
5181 # pre-fill the .rev_db since it'll eventually get filled in
5182 # with '0' x40 if something new gets committed
5183 foreach my $gs (@$gsv) {
5184 next if $gs->rev_map_max >= $max;
5185 next if defined $gs->rev_map_get($max);
5186 $gs->rev_map_set($max, 0 x40);
5187 }
5188 foreach my $g (@$globs) {
5189 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5190 Git::SVN::tmp_config($k, $max);
5191 }
5192 last if $max >= $head;
5193 $min = $max + 1;
5194 $max += $inc;
5195 $max = $head if ($max > $head);
5196 }
5197 Git::SVN::gc();
5198 }
5199
5200 sub get_dir_globbed {
5201 my ($self, $left, $depth, $r) = @_;
5202
5203 my @x = eval { $self->get_dir($left, $r) };
5204 return unless scalar @x == 3;
5205 my $dirents = $x[0];
5206 my @finalents;
5207 foreach my $de (keys %$dirents) {
5208 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5209 if ($depth > 1) {
5210 my @args = ("$left/$de", $depth - 1, $r);
5211 foreach my $dir ($self->get_dir_globbed(@args)) {
5212 push @finalents, "$de/$dir";
5213 }
5214 } else {
5215 push @finalents, $de;
5216 }
5217 }
5218 @finalents;
5219 }
5220
5221 sub match_globs {
5222 my ($self, $exists, $paths, $globs, $r) = @_;
5223
5224 sub get_dir_check {
5225 my ($self, $exists, $g, $r) = @_;
5226
5227 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5228 $g->{path}->{depth},
5229 $r);
5230
5231 foreach my $de (@dirs) {
5232 my $p = $g->{path}->full_path($de);
5233 next if $exists->{$p};
5234 next if (length $g->{path}->{right} &&
5235 ($self->check_path($p, $r) !=
5236 $SVN::Node::dir));
5237 next unless $p =~ /$g->{path}->{regex}/;
5238 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5239 $g->{ref}->full_path($de), 1);
5240 }
5241 }
5242 foreach my $g (@$globs) {
5243 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5244 if ($path->{action} =~ /^[AR]$/) {
5245 get_dir_check($self, $exists, $g, $r);
5246 }
5247 }
5248 foreach (keys %$paths) {
5249 if (/$g->{path}->{left_regex}/ &&
5250 !/$g->{path}->{regex}/) {
5251 next if $paths->{$_}->{action} !~ /^[AR]$/;
5252 get_dir_check($self, $exists, $g, $r);
5253 }
5254 next unless /$g->{path}->{regex}/;
5255 my $p = $1;
5256 my $pathname = $g->{path}->full_path($p);
5257 next if $exists->{$pathname};
5258 next if ($self->check_path($pathname, $r) !=
5259 $SVN::Node::dir);
5260 $exists->{$pathname} = Git::SVN->init(
5261 $self->{url}, $pathname, undef,
5262 $g->{ref}->full_path($p), 1);
5263 }
5264 my $c = '';
5265 foreach (split m#/#, $g->{path}->{left}) {
5266 $c .= "/$_";
5267 next unless ($paths->{$c} &&
5268 ($paths->{$c}->{action} =~ /^[AR]$/));
5269 get_dir_check($self, $exists, $g, $r);
5270 }
5271 }
5272 values %$exists;
5273 }
5274
5275 sub minimize_url {
5276 my ($self) = @_;
5277 return $self->{url} if ($self->{url} eq $self->{repos_root});
5278 my $url = $self->{repos_root};
5279 my @components = split(m!/!, $self->{svn_path});
5280 my $c = '';
5281 do {
5282 $url .= "/$c" if length $c;
5283 eval {
5284 my $ra = (ref $self)->new($url);
5285 my $latest = $ra->get_latest_revnum;
5286 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5287 };
5288 } while ($@ && ($c = shift @components));
5289 $url;
5290 }
5291
5292 sub can_do_switch {
5293 my $self = shift;
5294 unless (defined $can_do_switch) {
5295 my $pool = SVN::Pool->new;
5296 my $rep = eval {
5297 $self->do_switch(1, '', 0, $self->{url},
5298 SVN::Delta::Editor->new, $pool);
5299 };
5300 if ($@) {
5301 $can_do_switch = 0;
5302 } else {
5303 $rep->abort_report($pool);
5304 $can_do_switch = 1;
5305 }
5306 $pool->clear;
5307 }
5308 $can_do_switch;
5309 }
5310
5311 sub skip_unknown_revs {
5312 my ($err) = @_;
5313 my $errno = $err->apr_err();
5314 # Maybe the branch we're tracking didn't
5315 # exist when the repo started, so it's
5316 # not an error if it doesn't, just continue
5317 #
5318 # Wonderfully consistent library, eh?
5319 # 160013 - svn:// and file://
5320 # 175002 - http(s)://
5321 # 175007 - http(s):// (this repo required authorization, too...)
5322 # More codes may be discovered later...
5323 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5324 my $err_key = $err->expanded_message;
5325 # revision numbers change every time, filter them out
5326 $err_key =~ s/\d+/\0/g;
5327 $err_key = "$errno\0$err_key";
5328 unless ($ignored_err{$err_key}) {
5329 warn "W: Ignoring error from SVN, path probably ",
5330 "does not exist: ($errno): ",
5331 $err->expanded_message,"\n";
5332 warn "W: Do not be alarmed at the above message ",
5333 "git-svn is just searching aggressively for ",
5334 "old history.\n",
5335 "This may take a while on large repositories\n";
5336 $ignored_err{$err_key} = 1;
5337 }
5338 return;
5339 }
5340 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5341 }
5342
5343 package Git::SVN::Log;
5344 use strict;
5345 use warnings;
5346 use POSIX qw/strftime/;
5347 use Time::Local;
5348 use constant commit_log_separator => ('-' x 72) . "\n";
5349 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5350 %rusers $show_commit $incremental/;
5351 my $l_fmt;
5352
5353 sub cmt_showable {
5354 my ($c) = @_;
5355 return 1 if defined $c->{r};
5356
5357 # big commit message got truncated by the 16k pretty buffer in rev-list
5358 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5359 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5360 @{$c->{l}} = ();
5361 my @log = command(qw/cat-file commit/, $c->{c});
5362
5363 # shift off the headers
5364 shift @log while ($log[0] ne '');
5365 shift @log;
5366
5367 # TODO: make $c->{l} not have a trailing newline in the future
5368 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5369
5370 (undef, $c->{r}, undef) = ::extract_metadata(
5371 (grep(/^git-svn-id: /, @log))[-1]);
5372 }
5373 return defined $c->{r};
5374 }
5375
5376 sub log_use_color {
5377 return $color || Git->repository->get_colorbool('color.diff');
5378 }
5379
5380 sub git_svn_log_cmd {
5381 my ($r_min, $r_max, @args) = @_;
5382 my $head = 'HEAD';
5383 my (@files, @log_opts);
5384 foreach my $x (@args) {
5385 if ($x eq '--' || @files) {
5386 push @files, $x;
5387 } else {
5388 if (::verify_ref("$x^0")) {
5389 $head = $x;
5390 } else {
5391 push @log_opts, $x;
5392 }
5393 }
5394 }
5395
5396 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5397 $gs ||= Git::SVN->_new;
5398 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5399 $gs->refname);
5400 push @cmd, '-r' unless $non_recursive;
5401 push @cmd, qw/--raw --name-status/ if $verbose;
5402 push @cmd, '--color' if log_use_color();
5403 push @cmd, @log_opts;
5404 if (defined $r_max && $r_max == $r_min) {
5405 push @cmd, '--max-count=1';
5406 if (my $c = $gs->rev_map_get($r_max)) {
5407 push @cmd, $c;
5408 }
5409 } elsif (defined $r_max) {
5410 if ($r_max < $r_min) {
5411 ($r_min, $r_max) = ($r_max, $r_min);
5412 }
5413 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5414 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5415 # If there are no commits in the range, both $c_max and $c_min
5416 # will be undefined. If there is at least 1 commit in the
5417 # range, both will be defined.
5418 return () if !defined $c_min || !defined $c_max;
5419 if ($c_min eq $c_max) {
5420 push @cmd, '--max-count=1', $c_min;
5421 } else {
5422 push @cmd, '--boundary', "$c_min..$c_max";
5423 }
5424 }
5425 return (@cmd, @files);
5426 }
5427
5428 # adapted from pager.c
5429 sub config_pager {
5430 chomp(my $pager = command_oneline(qw(var GIT_PAGER)));
5431 if ($pager eq 'cat') {
5432 $pager = undef;
5433 }
5434 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5435 }
5436
5437 sub run_pager {
5438 return unless -t *STDOUT && defined $pager;
5439 pipe my ($rfd, $wfd) or return;
5440 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5441 if (!$pid) {
5442 open STDOUT, '>&', $wfd or
5443 ::fatal "Can't redirect to stdout: $!";
5444 return;
5445 }
5446 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5447 $ENV{LESS} ||= 'FRSX';
5448 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5449 }
5450
5451 sub format_svn_date {
5452 # some systmes don't handle or mishandle %z, so be creative.
5453 my $t = shift || time;
5454 my $gm = timelocal(gmtime($t));
5455 my $sign = qw( + + - )[ $t <=> $gm ];
5456 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5457 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5458 }
5459
5460 sub parse_git_date {
5461 my ($t, $tz) = @_;
5462 # Date::Parse isn't in the standard Perl distro :(
5463 if ($tz =~ s/^\+//) {
5464 $t += tz_to_s_offset($tz);
5465 } elsif ($tz =~ s/^\-//) {
5466 $t -= tz_to_s_offset($tz);
5467 }
5468 return $t;
5469 }
5470
5471 sub set_local_timezone {
5472 if (defined $TZ) {
5473 $ENV{TZ} = $TZ;
5474 } else {
5475 delete $ENV{TZ};
5476 }
5477 }
5478
5479 sub tz_to_s_offset {
5480 my ($tz) = @_;
5481 $tz =~ s/(\d\d)$//;
5482 return ($1 * 60) + ($tz * 3600);
5483 }
5484
5485 sub get_author_info {
5486 my ($dest, $author, $t, $tz) = @_;
5487 $author =~ s/(?:^\s*|\s*$)//g;
5488 $dest->{a_raw} = $author;
5489 my $au;
5490 if ($::_authors) {
5491 $au = $rusers{$author} || undef;
5492 }
5493 if (!$au) {
5494 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5495 }
5496 $dest->{t} = $t;
5497 $dest->{tz} = $tz;
5498 $dest->{a} = $au;
5499 $dest->{t_utc} = parse_git_date($t, $tz);
5500 }
5501
5502 sub process_commit {
5503 my ($c, $r_min, $r_max, $defer) = @_;
5504 if (defined $r_min && defined $r_max) {
5505 if ($r_min == $c->{r} && $r_min == $r_max) {
5506 show_commit($c);
5507 return 0;
5508 }
5509 return 1 if $r_min == $r_max;
5510 if ($r_min < $r_max) {
5511 # we need to reverse the print order
5512 return 0 if (defined $limit && --$limit < 0);
5513 push @$defer, $c;
5514 return 1;
5515 }
5516 if ($r_min != $r_max) {
5517 return 1 if ($r_min < $c->{r});
5518 return 1 if ($r_max > $c->{r});
5519 }
5520 }
5521 return 0 if (defined $limit && --$limit < 0);
5522 show_commit($c);
5523 return 1;
5524 }
5525
5526 sub show_commit {
5527 my $c = shift;
5528 if ($oneline) {
5529 my $x = "\n";
5530 if (my $l = $c->{l}) {
5531 while ($l->[0] =~ /^\s*$/) { shift @$l }
5532 $x = $l->[0];
5533 }
5534 $l_fmt ||= 'A' . length($c->{r});
5535 print 'r',pack($l_fmt, $c->{r}),' | ';
5536 print "$c->{c} | " if $show_commit;
5537 print $x;
5538 } else {
5539 show_commit_normal($c);
5540 }
5541 }
5542
5543 sub show_commit_changed_paths {
5544 my ($c) = @_;
5545 return unless $c->{changed};
5546 print "Changed paths:\n", @{$c->{changed}};
5547 }
5548
5549 sub show_commit_normal {
5550 my ($c) = @_;
5551 print commit_log_separator, "r$c->{r} | ";
5552 print "$c->{c} | " if $show_commit;
5553 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5554 my $nr_line = 0;
5555
5556 if (my $l = $c->{l}) {
5557 while ($l->[$#$l] eq "\n" && $#$l > 0
5558 && $l->[($#$l - 1)] eq "\n") {
5559 pop @$l;
5560 }
5561 $nr_line = scalar @$l;
5562 if (!$nr_line) {
5563 print "1 line\n\n\n";
5564 } else {
5565 if ($nr_line == 1) {
5566 $nr_line = '1 line';
5567 } else {
5568 $nr_line .= ' lines';
5569 }
5570 print $nr_line, "\n";
5571 show_commit_changed_paths($c);
5572 print "\n";
5573 print $_ foreach @$l;
5574 }
5575 } else {
5576 print "1 line\n";
5577 show_commit_changed_paths($c);
5578 print "\n";
5579
5580 }
5581 foreach my $x (qw/raw stat diff/) {
5582 if ($c->{$x}) {
5583 print "\n";
5584 print $_ foreach @{$c->{$x}}
5585 }
5586 }
5587 }
5588
5589 sub cmd_show_log {
5590 my (@args) = @_;
5591 my ($r_min, $r_max);
5592 my $r_last = -1; # prevent dupes
5593 set_local_timezone();
5594 if (defined $::_revision) {
5595 if ($::_revision =~ /^(\d+):(\d+)$/) {
5596 ($r_min, $r_max) = ($1, $2);
5597 } elsif ($::_revision =~ /^\d+$/) {
5598 $r_min = $r_max = $::_revision;
5599 } else {
5600 ::fatal "-r$::_revision is not supported, use ",
5601 "standard 'git log' arguments instead";
5602 }
5603 }
5604
5605 config_pager();
5606 @args = git_svn_log_cmd($r_min, $r_max, @args);
5607 if (!@args) {
5608 print commit_log_separator unless $incremental || $oneline;
5609 return;
5610 }
5611 my $log = command_output_pipe(@args);
5612 run_pager();
5613 my (@k, $c, $d, $stat);
5614 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5615 while (<$log>) {
5616 if (/^${esc_color}commit -?($::sha1_short)/o) {
5617 my $cmt = $1;
5618 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5619 $r_last = $c->{r};
5620 process_commit($c, $r_min, $r_max, \@k) or
5621 goto out;
5622 }
5623 $d = undef;
5624 $c = { c => $cmt };
5625 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5626 get_author_info($c, $1, $2, $3);
5627 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5628 # ignore
5629 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5630 push @{$c->{raw}}, $_;
5631 } elsif (/^${esc_color}[ACRMDT]\t/) {
5632 # we could add $SVN->{svn_path} here, but that requires
5633 # remote access at the moment (repo_path_split)...
5634 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5635 push @{$c->{changed}}, $_;
5636 } elsif (/^${esc_color}diff /o) {
5637 $d = 1;
5638 push @{$c->{diff}}, $_;
5639 } elsif ($d) {
5640 push @{$c->{diff}}, $_;
5641 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5642 $esc_color*[\+\-]*$esc_color$/x) {
5643 $stat = 1;
5644 push @{$c->{stat}}, $_;
5645 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5646 push @{$c->{stat}}, $_;
5647 $stat = undef;
5648 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5649 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5650 } elsif (s/^${esc_color} //o) {
5651 push @{$c->{l}}, $_;
5652 }
5653 }
5654 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5655 $r_last = $c->{r};
5656 process_commit($c, $r_min, $r_max, \@k);
5657 }
5658 if (@k) {
5659 ($r_min, $r_max) = ($r_max, $r_min);
5660 process_commit($_, $r_min, $r_max) foreach reverse @k;
5661 }
5662 out:
5663 close $log;
5664 print commit_log_separator unless $incremental || $oneline;
5665 }
5666
5667 sub cmd_blame {
5668 my $path = pop;
5669
5670 config_pager();
5671 run_pager();
5672
5673 my ($fh, $ctx, $rev);
5674
5675 if ($_git_format) {
5676 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5677 while (my $line = <$fh>) {
5678 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5679 # Uncommitted edits show up as a rev ID of
5680 # all zeros, which we can't look up with
5681 # cmt_metadata
5682 if ($1 !~ /^0+$/) {
5683 (undef, $rev, undef) =
5684 ::cmt_metadata($1);
5685 $rev = '0' if (!$rev);
5686 } else {
5687 $rev = '0';
5688 }
5689 $rev = sprintf('%-10s', $rev);
5690 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5691 }
5692 print $line;
5693 }
5694 } else {
5695 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5696 '--', $path);
5697 my ($sha1);
5698 my %authors;
5699 my @buffer;
5700 my %dsha; #distinct sha keys
5701
5702 while (my $line = <$fh>) {
5703 push @buffer, $line;
5704 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5705 $dsha{$1} = 1;
5706 }
5707 }
5708
5709 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5710
5711 foreach my $line (@buffer) {
5712 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5713 $rev = $s2r->{$1};
5714 $rev = '0' if (!$rev)
5715 }
5716 elsif ($line =~ /^author (.*)/) {
5717 $authors{$rev} = $1;
5718 $authors{$rev} =~ s/\s/_/g;
5719 }
5720 elsif ($line =~ /^\t(.*)$/) {
5721 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5722 }
5723 }
5724 }
5725 command_close_pipe($fh, $ctx);
5726 }
5727
5728 package Git::SVN::Migration;
5729 # these version numbers do NOT correspond to actual version numbers
5730 # of git nor git-svn. They are just relative.
5731 #
5732 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5733 #
5734 # v1 layout: .git/$id/info/url, refs/remotes/$id
5735 #
5736 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5737 #
5738 # v3 layout: .git/svn/$id, refs/remotes/$id
5739 # - info/url may remain for backwards compatibility
5740 # - this is what we migrate up to this layout automatically,
5741 # - this will be used by git svn init on single branches
5742 # v3.1 layout (auto migrated):
5743 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5744 # for backwards compatibility
5745 #
5746 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5747 # - this is only created for newly multi-init-ed
5748 # repositories. Similar in spirit to the
5749 # --use-separate-remotes option in git-clone (now default)
5750 # - we do not automatically migrate to this (following
5751 # the example set by core git)
5752 #
5753 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
5754 # - newer, more-efficient format that uses 24-bytes per record
5755 # with no filler space.
5756 # - use xxd -c24 < .rev_map.$UUID to view and debug
5757 # - This is a one-way migration, repositories updated to the
5758 # new format will not be able to use old git-svn without
5759 # rebuilding the .rev_db. Rebuilding the rev_db is not
5760 # possible if noMetadata or useSvmProps are set; but should
5761 # be no problem for users that use the (sensible) defaults.
5762 use strict;
5763 use warnings;
5764 use Carp qw/croak/;
5765 use File::Path qw/mkpath/;
5766 use File::Basename qw/dirname basename/;
5767 use vars qw/$_minimize/;
5768
5769 sub migrate_from_v0 {
5770 my $git_dir = $ENV{GIT_DIR};
5771 return undef unless -d $git_dir;
5772 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5773 my $migrated = 0;
5774 while (<$fh>) {
5775 chomp;
5776 my ($id, $orig_ref) = ($_, $_);
5777 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5778 next unless -f "$git_dir/$id/info/url";
5779 my $new_ref = "refs/remotes/$id";
5780 if (::verify_ref("$new_ref^0")) {
5781 print STDERR "W: $orig_ref is probably an old ",
5782 "branch used by an ancient version of ",
5783 "git-svn.\n",
5784 "However, $new_ref also exists.\n",
5785 "We will not be able ",
5786 "to use this branch until this ",
5787 "ambiguity is resolved.\n";
5788 next;
5789 }
5790 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5791 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5792 command_noisy('update-ref', $new_ref, $orig_ref);
5793 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5794 $migrated++;
5795 }
5796 command_close_pipe($fh, $ctx);
5797 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5798 $migrated;
5799 }
5800
5801 sub migrate_from_v1 {
5802 my $git_dir = $ENV{GIT_DIR};
5803 my $migrated = 0;
5804 return $migrated unless -d $git_dir;
5805 my $svn_dir = "$git_dir/svn";
5806
5807 # just in case somebody used 'svn' as their $id at some point...
5808 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5809
5810 print STDERR "Migrating from a git-svn v1 layout...\n";
5811 mkpath([$svn_dir]);
5812 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5813 "$svn_dir\n\t(required for this version ",
5814 "($::VERSION) of git-svn) does not exist.\n";
5815 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5816 while (<$fh>) {
5817 my $x = $_;
5818 next unless $x =~ s#^refs/remotes/##;
5819 chomp $x;
5820 next unless -f "$git_dir/$x/info/url";
5821 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5822 next unless $u;
5823 my $dn = dirname("$git_dir/svn/$x");
5824 mkpath([$dn]) unless -d $dn;
5825 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5826 mkpath(["$git_dir/svn/svn"]);
5827 print STDERR " - $git_dir/$x/info => ",
5828 "$git_dir/svn/$x/info\n";
5829 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5830 croak "$!: $x";
5831 # don't worry too much about these, they probably
5832 # don't exist with repos this old (save for index,
5833 # and we can easily regenerate that)
5834 foreach my $f (qw/unhandled.log index .rev_db/) {
5835 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5836 }
5837 } else {
5838 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5839 rename "$git_dir/$x", "$git_dir/svn/$x" or
5840 croak "$!: $x";
5841 }
5842 $migrated++;
5843 }
5844 command_close_pipe($fh, $ctx);
5845 print STDERR "Done migrating from a git-svn v1 layout\n";
5846 $migrated;
5847 }
5848
5849 sub read_old_urls {
5850 my ($l_map, $pfx, $path) = @_;
5851 my @dir;
5852 foreach (<$path/*>) {
5853 if (-r "$_/info/url") {
5854 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5855 my $ref_id = $pfx . basename $_;
5856 my $url = ::file_to_s("$_/info/url");
5857 $l_map->{$ref_id} = $url;
5858 } elsif (-d $_) {
5859 push @dir, $_;
5860 }
5861 }
5862 foreach (@dir) {
5863 my $x = $_;
5864 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5865 read_old_urls($l_map, $x, $_);
5866 }
5867 }
5868
5869 sub migrate_from_v2 {
5870 my @cfg = command(qw/config -l/);
5871 return if grep /^svn-remote\..+\.url=/, @cfg;
5872 my %l_map;
5873 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5874 my $migrated = 0;
5875
5876 foreach my $ref_id (sort keys %l_map) {
5877 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5878 if ($@) {
5879 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5880 }
5881 $migrated++;
5882 }
5883 $migrated;
5884 }
5885
5886 sub minimize_connections {
5887 my $r = Git::SVN::read_all_remotes();
5888 my $new_urls = {};
5889 my $root_repos = {};
5890 foreach my $repo_id (keys %$r) {
5891 my $url = $r->{$repo_id}->{url} or next;
5892 my $fetch = $r->{$repo_id}->{fetch} or next;
5893 my $ra = Git::SVN::Ra->new($url);
5894
5895 # skip existing cases where we already connect to the root
5896 if (($ra->{url} eq $ra->{repos_root}) ||
5897 ($ra->{repos_root} eq $repo_id)) {
5898 $root_repos->{$ra->{url}} = $repo_id;
5899 next;
5900 }
5901
5902 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5903 my $root_path = $ra->{url};
5904 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5905 foreach my $path (keys %$fetch) {
5906 my $ref_id = $fetch->{$path};
5907 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5908
5909 # make sure we can read when connecting to
5910 # a higher level of a repository
5911 my ($last_rev, undef) = $gs->last_rev_commit;
5912 if (!defined $last_rev) {
5913 $last_rev = eval {
5914 $root_ra->get_latest_revnum;
5915 };
5916 next if $@;
5917 }
5918 my $new = $root_path;
5919 $new .= length $path ? "/$path" : '';
5920 eval {
5921 $root_ra->get_log([$new], $last_rev, $last_rev,
5922 0, 0, 1, sub { });
5923 };
5924 next if $@;
5925 $new_urls->{$ra->{repos_root}}->{$new} =
5926 { ref_id => $ref_id,
5927 old_repo_id => $repo_id,
5928 old_path => $path };
5929 }
5930 }
5931
5932 my @emptied;
5933 foreach my $url (keys %$new_urls) {
5934 # see if we can re-use an existing [svn-remote "repo_id"]
5935 # instead of creating a(n ugly) new section:
5936 my $repo_id = $root_repos->{$url} || $url;
5937
5938 my $fetch = $new_urls->{$url};
5939 foreach my $path (keys %$fetch) {
5940 my $x = $fetch->{$path};
5941 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5942 my $pfx = "svn-remote.$x->{old_repo_id}";
5943
5944 my $old_fetch = quotemeta("$x->{old_path}:".
5945 "$x->{ref_id}");
5946 command_noisy(qw/config --unset/,
5947 "$pfx.fetch", '^'. $old_fetch . '$');
5948 delete $r->{$x->{old_repo_id}}->
5949 {fetch}->{$x->{old_path}};
5950 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5951 command_noisy(qw/config --unset/,
5952 "$pfx.url");
5953 push @emptied, $x->{old_repo_id}
5954 }
5955 }
5956 }
5957 if (@emptied) {
5958 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5959 print STDERR <<EOF;
5960 The following [svn-remote] sections in your config file ($file) are empty
5961 and can be safely removed:
5962 EOF
5963 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5964 }
5965 }
5966
5967 sub migration_check {
5968 migrate_from_v0();
5969 migrate_from_v1();
5970 migrate_from_v2();
5971 minimize_connections() if $_minimize;
5972 }
5973
5974 package Git::IndexInfo;
5975 use strict;
5976 use warnings;
5977 use Git qw/command_input_pipe command_close_pipe/;
5978
5979 sub new {
5980 my ($class) = @_;
5981 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5982 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5983 }
5984
5985 sub remove {
5986 my ($self, $path) = @_;
5987 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5988 return ++$self->{nr};
5989 }
5990 undef;
5991 }
5992
5993 sub update {
5994 my ($self, $mode, $hash, $path) = @_;
5995 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5996 return ++$self->{nr};
5997 }
5998 undef;
5999 }
6000
6001 sub DESTROY {
6002 my ($self) = @_;
6003 command_close_pipe($self->{gui}, $self->{ctx});
6004 }
6005
6006 package Git::SVN::GlobSpec;
6007 use strict;
6008 use warnings;
6009
6010 sub new {
6011 my ($class, $glob, $pattern_ok) = @_;
6012 my $re = $glob;
6013 $re =~ s!/+$!!g; # no need for trailing slashes
6014 my (@left, @right, @patterns);
6015 my $state = "left";
6016 my $die_msg = "Only one set of wildcard directories " .
6017 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6018 for my $part (split(m|/|, $glob)) {
6019 if ($part =~ /\*/ && $part ne "*") {
6020 die "Invalid pattern in '$glob': $part\n";
6021 } elsif ($pattern_ok && $part =~ /[{}]/ &&
6022 $part !~ /^\{[^{}]+\}/) {
6023 die "Invalid pattern in '$glob': $part\n";
6024 }
6025 if ($part eq "*") {
6026 die $die_msg if $state eq "right";
6027 $state = "pattern";
6028 push(@patterns, "[^/]*");
6029 } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6030 die $die_msg if $state eq "right";
6031 $state = "pattern";
6032 my $p = quotemeta($1);
6033 $p =~ s/\\,/|/g;
6034 push(@patterns, "(?:$p)");
6035 } else {
6036 if ($state eq "left") {
6037 push(@left, $part);
6038 } else {
6039 push(@right, $part);
6040 $state = "right";
6041 }
6042 }
6043 }
6044 my $depth = @patterns;
6045 if ($depth == 0) {
6046 die "One '*' is needed in glob: '$glob'\n";
6047 }
6048 my $left = join('/', @left);
6049 my $right = join('/', @right);
6050 $re = join('/', @patterns);
6051 $re = join('\/',
6052 grep(length, quotemeta($left), "($re)", quotemeta($right)));
6053 my $left_re = qr/^\/\Q$left\E(\/|$)/;
6054 bless { left => $left, right => $right, left_regex => $left_re,
6055 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6056 }
6057
6058 sub full_path {
6059 my ($self, $path) = @_;
6060 return (length $self->{left} ? "$self->{left}/" : '') .
6061 $path . (length $self->{right} ? "/$self->{right}" : '');
6062 }
6063
6064 __END__
6065
6066 Data structures:
6067
6068
6069 $remotes = { # returned by read_all_remotes()
6070 'svn' => {
6071 # svn-remote.svn.url=https://svn.musicpd.org
6072 url => 'https://svn.musicpd.org',
6073 # svn-remote.svn.fetch=mpd/trunk:trunk
6074 fetch => {
6075 'mpd/trunk' => 'trunk',
6076 },
6077 # svn-remote.svn.tags=mpd/tags/*:tags/*
6078 tags => {
6079 path => {
6080 left => 'mpd/tags',
6081 right => '',
6082 regex => qr!mpd/tags/([^/]+)$!,
6083 glob => 'tags/*',
6084 },
6085 ref => {
6086 left => 'tags',
6087 right => '',
6088 regex => qr!tags/([^/]+)$!,
6089 glob => 'tags/*',
6090 },
6091 }
6092 }
6093 };
6094
6095 $log_entry hashref as returned by libsvn_log_entry()
6096 {
6097 log => 'whitespace-formatted log entry
6098 ', # trailing newline is preserved
6099 revision => '8', # integer
6100 date => '2004-02-24T17:01:44.108345Z', # commit date
6101 author => 'committer name'
6102 };
6103
6104
6105 # this is generated by generate_diff();
6106 @mods = array of diff-index line hashes, each element represents one line
6107 of diff-index output
6108
6109 diff-index line ($m hash)
6110 {
6111 mode_a => first column of diff-index output, no leading ':',
6112 mode_b => second column of diff-index output,
6113 sha1_b => sha1sum of the final blob,
6114 chg => change type [MCRADT],
6115 file_a => original file name of a file (iff chg is 'C' or 'R')
6116 file_b => new/current file name of a file (any chg)
6117 }
6118 ;
6119
6120 # retval of read_url_paths{,_all}();
6121 $l_map = {
6122 # repository root url
6123 'https://svn.musicpd.org' => {
6124 # repository path # GIT_SVN_ID
6125 'mpd/trunk' => 'trunk',
6126 'mpd/tags/0.11.5' => 'tags/0.11.5',
6127 },
6128 }
6129
6130 Notes:
6131 I don't trust the each() function on unless I created %hash myself
6132 because the internal iterator may not have started at base.
Jonathan Nieder’s miscellaneous git branches