?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Fix handling of git-p4 on deleted files
[git/kirr.git] / git-svn.perl
blob2c86ea2e384e4b3ecf9f6c2a856c52a0203ce489
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 $| = 1; # unbuffer STDOUT
37
38 sub fatal (@) { print STDERR "@_\n"; exit 1 }
39 sub _req_svn {
40 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
41 require SVN::Ra;
42 require SVN::Delta;
43 if ($SVN::Core::VERSION lt '1.1.0') {
44 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
45 }
46 }
47 my $can_compress = eval { require Compress::Zlib; 1};
48 push @Git::SVN::Ra::ISA, 'SVN::Ra';
49 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
50 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
51 use Carp qw/croak/;
52 use Digest::MD5;
53 use IO::File qw//;
54 use File::Basename qw/dirname basename/;
55 use File::Path qw/mkpath/;
56 use File::Spec;
57 use File::Find;
58 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
59 use IPC::Open3;
60 use Git;
61
62 BEGIN {
63 # import functions from Git into our packages, en masse
64 no strict 'refs';
65 foreach (qw/command command_oneline command_noisy command_output_pipe
66 command_input_pipe command_close_pipe
67 command_bidi_pipe command_close_bidi_pipe/) {
68 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
69 Git::SVN::Migration Git::SVN::Log Git::SVN),
70 __PACKAGE__) {
71 *{"${package}::$_"} = \&{"Git::$_"};
72 }
73 }
74 }
75
76 my ($SVN);
77
78 $sha1 = qr/[a-f\d]{40}/;
79 $sha1_short = qr/[a-f\d]{4,40}/;
80 my ($_stdin, $_help, $_edit,
81 $_message, $_file, $_branch_dest,
82 $_template, $_shared,
83 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
84 $_merge, $_strategy, $_dry_run, $_local,
85 $_prefix, $_no_checkout, $_url, $_verbose,
86 $_git_format, $_commit_url, $_tag);
87 $Git::SVN::_follow_parent = 1;
88 $_q ||= 0;
89 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
90 'config-dir=s' => \$Git::SVN::Ra::config_dir,
91 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
92 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
93 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
94 'authors-file|A=s' => \$_authors,
95 'authors-prog=s' => \$_authors_prog,
96 'repack:i' => \$Git::SVN::_repack,
97 'noMetadata' => \$Git::SVN::_no_metadata,
98 'useSvmProps' => \$Git::SVN::_use_svm_props,
99 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
100 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
101 'no-checkout' => \$_no_checkout,
102 'quiet|q+' => \$_q,
103 'repack-flags|repack-args|repack-opts=s' =>
104 \$Git::SVN::_repack_flags,
105 'use-log-author' => \$Git::SVN::_use_log_author,
106 'add-author-from' => \$Git::SVN::_add_author_from,
107 'localtime' => \$Git::SVN::_localtime,
108 %remote_opts );
109
110 my ($_trunk, @_tags, @_branches, $_stdlayout);
111 my %icv;
112 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
113 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
114 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
115 'stdlayout|s' => \$_stdlayout,
116 'minimize-url|m!' => \$Git::SVN::_minimize_url,
117 'no-metadata' => sub { $icv{noMetadata} = 1 },
118 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
119 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
120 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
121 'rewrite-uuid=s' => sub { $icv{rewriteUUID} = $_[1] },
122 %remote_opts );
123 my %cmt_opts = ( 'edit|e' => \$_edit,
124 'rmdir' => \$SVN::Git::Editor::_rmdir,
125 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
126 'l=i' => \$SVN::Git::Editor::_rename_limit,
127 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
128 );
129
130 my %cmd = (
131 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
132 { 'revision|r=s' => \$_revision,
133 'fetch-all|all' => \$_fetch_all,
134 'parent|p' => \$_fetch_parent,
135 %fc_opts } ],
136 clone => [ \&cmd_clone, "Initialize and fetch revisions",
137 { 'revision|r=s' => \$_revision,
138 %fc_opts, %init_opts } ],
139 init => [ \&cmd_init, "Initialize a repo for tracking" .
140 " (requires URL argument)",
141 \%init_opts ],
142 'multi-init' => [ \&cmd_multi_init,
143 "Deprecated alias for ".
144 "'$0 init -T<trunk> -b<branches> -t<tags>'",
145 \%init_opts ],
146 dcommit => [ \&cmd_dcommit,
147 'Commit several diffs to merge with upstream',
148 { 'merge|m|M' => \$_merge,
149 'strategy|s=s' => \$_strategy,
150 'verbose|v' => \$_verbose,
151 'dry-run|n' => \$_dry_run,
152 'fetch-all|all' => \$_fetch_all,
153 'commit-url=s' => \$_commit_url,
154 'revision|r=i' => \$_revision,
155 'no-rebase' => \$_no_rebase,
156 %cmt_opts, %fc_opts } ],
157 branch => [ \&cmd_branch,
158 'Create a branch in the SVN repository',
159 { 'message|m=s' => \$_message,
160 'destination|d=s' => \$_branch_dest,
161 'dry-run|n' => \$_dry_run,
162 'tag|t' => \$_tag,
163 'username=s' => \$Git::SVN::Prompt::_username,
164 'commit-url=s' => \$_commit_url } ],
165 tag => [ sub { $_tag = 1; cmd_branch(@_) },
166 'Create a tag in the SVN repository',
167 { 'message|m=s' => \$_message,
168 'destination|d=s' => \$_branch_dest,
169 'dry-run|n' => \$_dry_run,
170 'username=s' => \$Git::SVN::Prompt::_username,
171 'commit-url=s' => \$_commit_url } ],
172 'set-tree' => [ \&cmd_set_tree,
173 "Set an SVN repository to a git tree-ish",
174 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
175 'create-ignore' => [ \&cmd_create_ignore,
176 'Create a .gitignore per svn:ignore',
177 { 'revision|r=i' => \$_revision
178 } ],
179 'mkdirs' => [ \&cmd_mkdirs ,
180 "recreate empty directories after a checkout",
181 { 'revision|r=i' => \$_revision } ],
182 'propget' => [ \&cmd_propget,
183 'Print the value of a property on a file or directory',
184 { 'revision|r=i' => \$_revision } ],
185 'proplist' => [ \&cmd_proplist,
186 'List all properties of a file or directory',
187 { 'revision|r=i' => \$_revision } ],
188 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
189 { 'revision|r=i' => \$_revision
190 } ],
191 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
192 { 'revision|r=i' => \$_revision
193 } ],
194 'multi-fetch' => [ \&cmd_multi_fetch,
195 "Deprecated alias for $0 fetch --all",
196 { 'revision|r=s' => \$_revision, %fc_opts } ],
197 'migrate' => [ sub { },
198 # no-op, we automatically run this anyways,
199 'Migrate configuration/metadata/layout from
200 previous versions of git-svn',
201 { 'minimize' => \$Git::SVN::Migration::_minimize,
202 %remote_opts } ],
203 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
204 { 'limit=i' => \$Git::SVN::Log::limit,
205 'revision|r=s' => \$_revision,
206 'verbose|v' => \$Git::SVN::Log::verbose,
207 'incremental' => \$Git::SVN::Log::incremental,
208 'oneline' => \$Git::SVN::Log::oneline,
209 'show-commit' => \$Git::SVN::Log::show_commit,
210 'non-recursive' => \$Git::SVN::Log::non_recursive,
211 'authors-file|A=s' => \$_authors,
212 'color' => \$Git::SVN::Log::color,
213 'pager=s' => \$Git::SVN::Log::pager
214 } ],
215 'find-rev' => [ \&cmd_find_rev,
216 "Translate between SVN revision numbers and tree-ish",
217 {} ],
218 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
219 { 'merge|m|M' => \$_merge,
220 'verbose|v' => \$_verbose,
221 'strategy|s=s' => \$_strategy,
222 'local|l' => \$_local,
223 'fetch-all|all' => \$_fetch_all,
224 'dry-run|n' => \$_dry_run,
225 %fc_opts } ],
226 'commit-diff' => [ \&cmd_commit_diff,
227 'Commit a diff between two trees',
228 { 'message|m=s' => \$_message,
229 'file|F=s' => \$_file,
230 'revision|r=s' => \$_revision,
231 %cmt_opts } ],
232 'info' => [ \&cmd_info,
233 "Show info about the latest SVN revision
234 on the current branch",
235 { 'url' => \$_url, } ],
236 'blame' => [ \&Git::SVN::Log::cmd_blame,
237 "Show what revision and author last modified each line of a file",
238 { 'git-format' => \$_git_format } ],
239 'reset' => [ \&cmd_reset,
240 "Undo fetches back to the specified SVN revision",
241 { 'revision|r=s' => \$_revision,
242 'parent|p' => \$_fetch_parent } ],
243 'gc' => [ \&cmd_gc,
244 "Compress unhandled.log files in .git/svn and remove " .
245 "index files in .git/svn",
246 {} ],
247 );
248
249 my $cmd;
250 for (my $i = 0; $i < @ARGV; $i++) {
251 if (defined $cmd{$ARGV[$i]}) {
252 $cmd = $ARGV[$i];
253 splice @ARGV, $i, 1;
254 last;
255 } elsif ($ARGV[$i] eq 'help') {
256 $cmd = $ARGV[$i+1];
257 usage(0);
258 }
259 };
260
261 # make sure we're always running at the top-level working directory
262 unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
263 unless (-d $ENV{GIT_DIR}) {
264 if ($git_dir_user_set) {
265 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
266 "but it is not a directory\n";
267 }
268 my $git_dir = delete $ENV{GIT_DIR};
269 my $cdup = undef;
270 git_cmd_try {
271 $cdup = command_oneline(qw/rev-parse --show-cdup/);
272 $git_dir = '.' unless ($cdup);
273 chomp $cdup if ($cdup);
274 $cdup = "." unless ($cdup && length $cdup);
275 } "Already at toplevel, but $git_dir not found\n";
276 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
277 unless (-d $git_dir) {
278 die "$git_dir still not found after going to ",
279 "'$cdup'\n";
280 }
281 $ENV{GIT_DIR} = $git_dir;
282 }
283 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
284 }
285
286 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
287
288 read_git_config(\%opts);
289 if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
290 Getopt::Long::Configure('pass_through');
291 }
292 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
293 'minimize-connections' => \$Git::SVN::Migration::_minimize,
294 'id|i=s' => \$Git::SVN::default_ref_id,
295 'svn-remote|remote|R=s' => sub {
296 $Git::SVN::no_reuse_existing = 1;
297 $Git::SVN::default_repo_id = $_[1] });
298 exit 1 if (!$rv && $cmd && $cmd ne 'log');
299
300 usage(0) if $_help;
301 version() if $_version;
302 usage(1) unless defined $cmd;
303 load_authors() if $_authors;
304 if (defined $_authors_prog) {
305 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
306 }
307
308 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
309 Git::SVN::Migration::migration_check();
310 }
311 Git::SVN::init_vars();
312 eval {
313 Git::SVN::verify_remotes_sanity();
314 $cmd{$cmd}->[0]->(@ARGV);
315 };
316 fatal $@ if $@;
317 post_fetch_checkout();
318 exit 0;
319
320 ####################### primary functions ######################
321 sub usage {
322 my $exit = shift || 0;
323 my $fd = $exit ? \*STDERR : \*STDOUT;
324 print $fd <<"";
325 git-svn - bidirectional operations between a single Subversion tree and git
326 Usage: git svn <command> [options] [arguments]\n
327
328 print $fd "Available commands:\n" unless $cmd;
329
330 foreach (sort keys %cmd) {
331 next if $cmd && $cmd ne $_;
332 next if /^multi-/; # don't show deprecated commands
333 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
334 foreach (sort keys %{$cmd{$_}->[2]}) {
335 # mixed-case options are for .git/config only
336 next if /[A-Z]/ && /^[a-z]+$/i;
337 # prints out arguments as they should be passed:
338 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
339 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
340 "--$_" : "-$_" }
341 split /\|/,$_)," $x\n";
342 }
343 }
344 print $fd <<"";
345 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
346 arbitrary identifier if you're tracking multiple SVN branches/repositories in
347 one git repository and want to keep them separate. See git-svn(1) for more
348 information.
349
350 exit $exit;
351 }
352
353 sub version {
354 ::_req_svn();
355 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
356 exit 0;
357 }
358
359 sub do_git_init_db {
360 unless (-d $ENV{GIT_DIR}) {
361 my @init_db = ('init');
362 push @init_db, "--template=$_template" if defined $_template;
363 if (defined $_shared) {
364 if ($_shared =~ /[a-z]/) {
365 push @init_db, "--shared=$_shared";
366 } else {
367 push @init_db, "--shared";
368 }
369 }
370 command_noisy(@init_db);
371 $_repository = Git->repository(Repository => ".git");
372 }
373 my $set;
374 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
375 foreach my $i (keys %icv) {
376 die "'$set' and '$i' cannot both be set\n" if $set;
377 next unless defined $icv{$i};
378 command_noisy('config', "$pfx.$i", $icv{$i});
379 $set = $i;
380 }
381 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
382 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
383 if defined $$ignore_regex;
384 }
385
386 sub init_subdir {
387 my $repo_path = shift or return;
388 mkpath([$repo_path]) unless -d $repo_path;
389 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
390 $ENV{GIT_DIR} = '.git';
391 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
392 }
393
394 sub cmd_clone {
395 my ($url, $path) = @_;
396 if (!defined $path &&
397 (defined $_trunk || @_branches || @_tags ||
398 defined $_stdlayout) &&
399 $url !~ m#^[a-z\+]+://#) {
400 $path = $url;
401 }
402 $path = basename($url) if !defined $path || !length $path;
403 my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
404 cmd_init($url, $path);
405 command_oneline('config', 'svn.authorsfile', $authors_absolute)
406 if $_authors;
407 Git::SVN::fetch_all($Git::SVN::default_repo_id);
408 }
409
410 sub cmd_init {
411 if (defined $_stdlayout) {
412 $_trunk = 'trunk' if (!defined $_trunk);
413 @_tags = 'tags' if (! @_tags);
414 @_branches = 'branches' if (! @_branches);
415 }
416 if (defined $_trunk || @_branches || @_tags) {
417 return cmd_multi_init(@_);
418 }
419 my $url = shift or die "SVN repository location required ",
420 "as a command-line argument\n";
421 $url = canonicalize_url($url);
422 init_subdir(@_);
423 do_git_init_db();
424
425 if ($Git::SVN::_minimize_url eq 'unset') {
426 $Git::SVN::_minimize_url = 0;
427 }
428
429 Git::SVN->init($url);
430 }
431
432 sub cmd_fetch {
433 if (grep /^\d+=./, @_) {
434 die "'<rev>=<commit>' fetch arguments are ",
435 "no longer supported.\n";
436 }
437 my ($remote) = @_;
438 if (@_ > 1) {
439 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
440 }
441 $Git::SVN::no_reuse_existing = undef;
442 if ($_fetch_parent) {
443 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
444 unless ($gs) {
445 die "Unable to determine upstream SVN information from ",
446 "working tree history\n";
447 }
448 # just fetch, don't checkout.
449 $_no_checkout = 'true';
450 $_fetch_all ? $gs->fetch_all : $gs->fetch;
451 } elsif ($_fetch_all) {
452 cmd_multi_fetch();
453 } else {
454 $remote ||= $Git::SVN::default_repo_id;
455 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
456 }
457 }
458
459 sub cmd_set_tree {
460 my (@commits) = @_;
461 if ($_stdin || !@commits) {
462 print "Reading from stdin...\n";
463 @commits = ();
464 while (<STDIN>) {
465 if (/\b($sha1_short)\b/o) {
466 unshift @commits, $1;
467 }
468 }
469 }
470 my @revs;
471 foreach my $c (@commits) {
472 my @tmp = command('rev-parse',$c);
473 if (scalar @tmp == 1) {
474 push @revs, $tmp[0];
475 } elsif (scalar @tmp > 1) {
476 push @revs, reverse(command('rev-list',@tmp));
477 } else {
478 fatal "Failed to rev-parse $c";
479 }
480 }
481 my $gs = Git::SVN->new;
482 my ($r_last, $cmt_last) = $gs->last_rev_commit;
483 $gs->fetch;
484 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
485 fatal "There are new revisions that were fetched ",
486 "and need to be merged (or acknowledged) ",
487 "before committing.\nlast rev: $r_last\n",
488 " current: $gs->{last_rev}";
489 }
490 $gs->set_tree($_) foreach @revs;
491 print "Done committing ",scalar @revs," revisions to SVN\n";
492 unlink $gs->{index};
493 }
494
495 sub cmd_dcommit {
496 my $head = shift;
497 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
498 'Cannot dcommit with a dirty index. Commit your changes first, '
499 . "or stash them with `git stash'.\n";
500 $head ||= 'HEAD';
501
502 my $old_head;
503 if ($head ne 'HEAD') {
504 $old_head = eval {
505 command_oneline([qw/symbolic-ref -q HEAD/])
506 };
507 if ($old_head) {
508 $old_head =~ s{^refs/heads/}{};
509 } else {
510 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
511 }
512 command(['checkout', $head], STDERR => 0);
513 }
514
515 my @refs;
516 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
517 unless ($gs) {
518 die "Unable to determine upstream SVN information from ",
519 "$head history.\nPerhaps the repository is empty.";
520 }
521
522 if (defined $_commit_url) {
523 $url = $_commit_url;
524 } else {
525 $url = eval { command_oneline('config', '--get',
526 "svn-remote.$gs->{repo_id}.commiturl") };
527 if (!$url) {
528 $url = $gs->full_url
529 }
530 }
531
532 my $last_rev = $_revision if defined $_revision;
533 if ($url) {
534 print "Committing to $url ...\n";
535 }
536 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
537 if ($_no_rebase && scalar(@$linear_refs) > 1) {
538 warn "Attempting to commit more than one change while ",
539 "--no-rebase is enabled.\n",
540 "If these changes depend on each other, re-running ",
541 "without --no-rebase may be required."
542 }
543 my $expect_url = $url;
544 Git::SVN::remove_username($expect_url);
545 while (1) {
546 my $d = shift @$linear_refs or last;
547 unless (defined $last_rev) {
548 (undef, $last_rev, undef) = cmt_metadata("$d~1");
549 unless (defined $last_rev) {
550 fatal "Unable to extract revision information ",
551 "from commit $d~1";
552 }
553 }
554 if ($_dry_run) {
555 print "diff-tree $d~1 $d\n";
556 } else {
557 my $cmt_rev;
558 my %ed_opts = ( r => $last_rev,
559 log => get_commit_entry($d)->{log},
560 ra => Git::SVN::Ra->new($url),
561 config => SVN::Core::config_get_config(
562 $Git::SVN::Ra::config_dir
563 ),
564 tree_a => "$d~1",
565 tree_b => $d,
566 editor_cb => sub {
567 print "Committed r$_[0]\n";
568 $cmt_rev = $_[0];
569 },
570 svn_path => '');
571 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
572 print "No changes\n$d~1 == $d\n";
573 } elsif ($parents->{$d} && @{$parents->{$d}}) {
574 $gs->{inject_parents_dcommit}->{$cmt_rev} =
575 $parents->{$d};
576 }
577 $_fetch_all ? $gs->fetch_all : $gs->fetch;
578 $last_rev = $cmt_rev;
579 next if $_no_rebase;
580
581 # we always want to rebase against the current HEAD,
582 # not any head that was passed to us
583 my @diff = command('diff-tree', $d,
584 $gs->refname, '--');
585 my @finish;
586 if (@diff) {
587 @finish = rebase_cmd();
588 print STDERR "W: $d and ", $gs->refname,
589 " differ, using @finish:\n",
590 join("\n", @diff), "\n";
591 } else {
592 print "No changes between current HEAD and ",
593 $gs->refname,
594 "\nResetting to the latest ",
595 $gs->refname, "\n";
596 @finish = qw/reset --mixed/;
597 }
598 command_noisy(@finish, $gs->refname);
599 if (@diff) {
600 @refs = ();
601 my ($url_, $rev_, $uuid_, $gs_) =
602 working_head_info('HEAD', \@refs);
603 my ($linear_refs_, $parents_) =
604 linearize_history($gs_, \@refs);
605 if (scalar(@$linear_refs) !=
606 scalar(@$linear_refs_)) {
607 fatal "# of revisions changed ",
608 "\nbefore:\n",
609 join("\n", @$linear_refs),
610 "\n\nafter:\n",
611 join("\n", @$linear_refs_), "\n",
612 'If you are attempting to commit ',
613 "merges, try running:\n\t",
614 'git rebase --interactive',
615 '--preserve-merges ',
616 $gs->refname,
617 "\nBefore dcommitting";
618 }
619 if ($url_ ne $expect_url) {
620 if ($url_ eq $gs->metadata_url) {
621 print
622 "Accepting rewritten URL:",
623 " $url_\n";
624 } else {
625 fatal
626 "URL mismatch after rebase:",
627 " $url_ != $expect_url";
628 }
629 }
630 if ($uuid_ ne $uuid) {
631 fatal "uuid mismatch after rebase: ",
632 "$uuid_ != $uuid";
633 }
634 # remap parents
635 my (%p, @l, $i);
636 for ($i = 0; $i < scalar @$linear_refs; $i++) {
637 my $new = $linear_refs_->[$i] or next;
638 $p{$new} =
639 $parents->{$linear_refs->[$i]};
640 push @l, $new;
641 }
642 $parents = \%p;
643 $linear_refs = \@l;
644 }
645 }
646 }
647
648 if ($old_head) {
649 my $new_head = command_oneline(qw/rev-parse HEAD/);
650 my $new_is_symbolic = eval {
651 command_oneline(qw/symbolic-ref -q HEAD/);
652 };
653 if ($new_is_symbolic) {
654 print "dcommitted the branch ", $head, "\n";
655 } else {
656 print "dcommitted on a detached HEAD because you gave ",
657 "a revision argument.\n",
658 "The rewritten commit is: ", $new_head, "\n";
659 }
660 command(['checkout', $old_head], STDERR => 0);
661 }
662
663 unlink $gs->{index};
664 }
665
666 sub cmd_branch {
667 my ($branch_name, $head) = @_;
668
669 unless (defined $branch_name && length $branch_name) {
670 die(($_tag ? "tag" : "branch") . " name required\n");
671 }
672 $head ||= 'HEAD';
673
674 my (undef, $rev, undef, $gs) = working_head_info($head);
675 my $src = $gs->full_url;
676
677 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
678 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
679 my $glob;
680 if ($#{$allglobs} == 0) {
681 $glob = $allglobs->[0];
682 } else {
683 unless(defined $_branch_dest) {
684 die "Multiple ",
685 $_tag ? "tag" : "branch",
686 " paths defined for Subversion repository.\n",
687 "You must specify where you want to create the ",
688 $_tag ? "tag" : "branch",
689 " with the --destination argument.\n";
690 }
691 foreach my $g (@{$allglobs}) {
692 # SVN::Git::Editor could probably be moved to Git.pm..
693 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
694 if ($_branch_dest =~ /$re/) {
695 $glob = $g;
696 last;
697 }
698 }
699 unless (defined $glob) {
700 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
701 foreach my $g (@{$allglobs}) {
702 $g->{path}->{left} =~ /$dest_re/ or next;
703 if (defined $glob) {
704 die "Ambiguous destination: ",
705 $_branch_dest, "\nmatches both '",
706 $glob->{path}->{left}, "' and '",
707 $g->{path}->{left}, "'\n";
708 }
709 $glob = $g;
710 }
711 unless (defined $glob) {
712 die "Unknown ",
713 $_tag ? "tag" : "branch",
714 " destination $_branch_dest\n";
715 }
716 }
717 }
718 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
719 my $url;
720 if (defined $_commit_url) {
721 $url = $_commit_url;
722 } else {
723 $url = eval { command_oneline('config', '--get',
724 "svn-remote.$gs->{repo_id}.commiturl") };
725 if (!$url) {
726 $url = $remote->{url};
727 }
728 }
729 my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
730
731 if ($dst =~ /^https:/ && $src =~ /^http:/) {
732 $src=~s/^http:/https:/;
733 }
734
735 ::_req_svn();
736
737 my $ctx = SVN::Client->new(
738 auth => Git::SVN::Ra::_auth_providers(),
739 log_msg => sub {
740 ${ $_[0] } = defined $_message
741 ? $_message
742 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
743 . $branch_name;
744 },
745 );
746
747 eval {
748 $ctx->ls($dst, 'HEAD', 0);
749 } and die "branch ${branch_name} already exists\n";
750
751 print "Copying ${src} at r${rev} to ${dst}...\n";
752 $ctx->copy($src, $rev, $dst)
753 unless $_dry_run;
754
755 $gs->fetch_all;
756 }
757
758 sub cmd_find_rev {
759 my $revision_or_hash = shift or die "SVN or git revision required ",
760 "as a command-line argument\n";
761 my $result;
762 if ($revision_or_hash =~ /^r\d+$/) {
763 my $head = shift;
764 $head ||= 'HEAD';
765 my @refs;
766 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
767 unless ($gs) {
768 die "Unable to determine upstream SVN information from ",
769 "$head history\n";
770 }
771 my $desired_revision = substr($revision_or_hash, 1);
772 $result = $gs->rev_map_get($desired_revision, $uuid);
773 } else {
774 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
775 $result = $rev;
776 }
777 print "$result\n" if $result;
778 }
779
780 sub cmd_rebase {
781 command_noisy(qw/update-index --refresh/);
782 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
783 unless ($gs) {
784 die "Unable to determine upstream SVN information from ",
785 "working tree history\n";
786 }
787 if ($_dry_run) {
788 print "Remote Branch: " . $gs->refname . "\n";
789 print "SVN URL: " . $url . "\n";
790 return;
791 }
792 if (command(qw/diff-index HEAD --/)) {
793 print STDERR "Cannot rebase with uncommited changes:\n";
794 command_noisy('status');
795 exit 1;
796 }
797 unless ($_local) {
798 # rebase will checkout for us, so no need to do it explicitly
799 $_no_checkout = 'true';
800 $_fetch_all ? $gs->fetch_all : $gs->fetch;
801 }
802 command_noisy(rebase_cmd(), $gs->refname);
803 $gs->mkemptydirs;
804 }
805
806 sub cmd_show_ignore {
807 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
808 $gs ||= Git::SVN->new;
809 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
810 $gs->prop_walk($gs->{path}, $r, sub {
811 my ($gs, $path, $props) = @_;
812 print STDOUT "\n# $path\n";
813 my $s = $props->{'svn:ignore'} or return;
814 $s =~ s/[\r\n]+/\n/g;
815 $s =~ s/^\n+//;
816 chomp $s;
817 $s =~ s#^#$path#gm;
818 print STDOUT "$s\n";
819 });
820 }
821
822 sub cmd_show_externals {
823 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
824 $gs ||= Git::SVN->new;
825 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
826 $gs->prop_walk($gs->{path}, $r, sub {
827 my ($gs, $path, $props) = @_;
828 print STDOUT "\n# $path\n";
829 my $s = $props->{'svn:externals'} or return;
830 $s =~ s/[\r\n]+/\n/g;
831 chomp $s;
832 $s =~ s#^#$path#gm;
833 print STDOUT "$s\n";
834 });
835 }
836
837 sub cmd_create_ignore {
838 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
839 $gs ||= Git::SVN->new;
840 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
841 $gs->prop_walk($gs->{path}, $r, sub {
842 my ($gs, $path, $props) = @_;
843 # $path is of the form /path/to/dir/
844 $path = '.' . $path;
845 # SVN can have attributes on empty directories,
846 # which git won't track
847 mkpath([$path]) unless -d $path;
848 my $ignore = $path . '.gitignore';
849 my $s = $props->{'svn:ignore'} or return;
850 open(GITIGNORE, '>', $ignore)
851 or fatal("Failed to open `$ignore' for writing: $!");
852 $s =~ s/[\r\n]+/\n/g;
853 $s =~ s/^\n+//;
854 chomp $s;
855 # Prefix all patterns so that the ignore doesn't apply
856 # to sub-directories.
857 $s =~ s#^#/#gm;
858 print GITIGNORE "$s\n";
859 close(GITIGNORE)
860 or fatal("Failed to close `$ignore': $!");
861 command_noisy('add', '-f', $ignore);
862 });
863 }
864
865 sub cmd_mkdirs {
866 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
867 $gs ||= Git::SVN->new;
868 $gs->mkemptydirs($_revision);
869 }
870
871 sub canonicalize_path {
872 my ($path) = @_;
873 my $dot_slash_added = 0;
874 if (substr($path, 0, 1) ne "/") {
875 $path = "./" . $path;
876 $dot_slash_added = 1;
877 }
878 # File::Spec->canonpath doesn't collapse x/../y into y (for a
879 # good reason), so let's do this manually.
880 $path =~ s#/+#/#g;
881 $path =~ s#/\.(?:/|$)#/#g;
882 $path =~ s#/[^/]+/\.\.##g;
883 $path =~ s#/$##g;
884 $path =~ s#^\./## if $dot_slash_added;
885 $path =~ s#^/##;
886 $path =~ s#^\.$##;
887 return $path;
888 }
889
890 sub canonicalize_url {
891 my ($url) = @_;
892 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
893 return $url;
894 }
895
896 # get_svnprops(PATH)
897 # ------------------
898 # Helper for cmd_propget and cmd_proplist below.
899 sub get_svnprops {
900 my $path = shift;
901 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
902 $gs ||= Git::SVN->new;
903
904 # prefix THE PATH by the sub-directory from which the user
905 # invoked us.
906 $path = $cmd_dir_prefix . $path;
907 fatal("No such file or directory: $path") unless -e $path;
908 my $is_dir = -d $path ? 1 : 0;
909 $path = $gs->{path} . '/' . $path;
910
911 # canonicalize the path (otherwise libsvn will abort or fail to
912 # find the file)
913 $path = canonicalize_path($path);
914
915 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
916 my $props;
917 if ($is_dir) {
918 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
919 }
920 else {
921 (undef, $props) = $gs->ra->get_file($path, $r, undef);
922 }
923 return $props;
924 }
925
926 # cmd_propget (PROP, PATH)
927 # ------------------------
928 # Print the SVN property PROP for PATH.
929 sub cmd_propget {
930 my ($prop, $path) = @_;
931 $path = '.' if not defined $path;
932 usage(1) if not defined $prop;
933 my $props = get_svnprops($path);
934 if (not defined $props->{$prop}) {
935 fatal("`$path' does not have a `$prop' SVN property.");
936 }
937 print $props->{$prop} . "\n";
938 }
939
940 # cmd_proplist (PATH)
941 # -------------------
942 # Print the list of SVN properties for PATH.
943 sub cmd_proplist {
944 my $path = shift;
945 $path = '.' if not defined $path;
946 my $props = get_svnprops($path);
947 print "Properties on '$path':\n";
948 foreach (sort keys %{$props}) {
949 print " $_\n";
950 }
951 }
952
953 sub cmd_multi_init {
954 my $url = shift;
955 unless (defined $_trunk || @_branches || @_tags) {
956 usage(1);
957 }
958
959 $_prefix = '' unless defined $_prefix;
960 if (defined $url) {
961 $url = canonicalize_url($url);
962 init_subdir(@_);
963 }
964 do_git_init_db();
965 if (defined $_trunk) {
966 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
967 # try both old-style and new-style lookups:
968 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
969 unless ($gs_trunk) {
970 my ($trunk_url, $trunk_path) =
971 complete_svn_url($url, $_trunk);
972 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
973 undef, $trunk_ref);
974 }
975 }
976 return unless @_branches || @_tags;
977 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
978 foreach my $path (@_branches) {
979 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
980 }
981 foreach my $path (@_tags) {
982 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
983 }
984 }
985
986 sub cmd_multi_fetch {
987 $Git::SVN::no_reuse_existing = undef;
988 my $remotes = Git::SVN::read_all_remotes();
989 foreach my $repo_id (sort keys %$remotes) {
990 if ($remotes->{$repo_id}->{url}) {
991 Git::SVN::fetch_all($repo_id, $remotes);
992 }
993 }
994 }
995
996 # this command is special because it requires no metadata
997 sub cmd_commit_diff {
998 my ($ta, $tb, $url) = @_;
999 my $usage = "Usage: $0 commit-diff -r<revision> ".
1000 "<tree-ish> <tree-ish> [<URL>]";
1001 fatal($usage) if (!defined $ta || !defined $tb);
1002 my $svn_path = '';
1003 if (!defined $url) {
1004 my $gs = eval { Git::SVN->new };
1005 if (!$gs) {
1006 fatal("Needed URL or usable git-svn --id in ",
1007 "the command-line\n", $usage);
1008 }
1009 $url = $gs->{url};
1010 $svn_path = $gs->{path};
1011 }
1012 unless (defined $_revision) {
1013 fatal("-r|--revision is a required argument\n", $usage);
1014 }
1015 if (defined $_message && defined $_file) {
1016 fatal("Both --message/-m and --file/-F specified ",
1017 "for the commit message.\n",
1018 "I have no idea what you mean");
1019 }
1020 if (defined $_file) {
1021 $_message = file_to_s($_file);
1022 } else {
1023 $_message ||= get_commit_entry($tb)->{log};
1024 }
1025 my $ra ||= Git::SVN::Ra->new($url);
1026 my $r = $_revision;
1027 if ($r eq 'HEAD') {
1028 $r = $ra->get_latest_revnum;
1029 } elsif ($r !~ /^\d+$/) {
1030 die "revision argument: $r not understood by git-svn\n";
1031 }
1032 my %ed_opts = ( r => $r,
1033 log => $_message,
1034 ra => $ra,
1035 tree_a => $ta,
1036 tree_b => $tb,
1037 editor_cb => sub { print "Committed r$_[0]\n" },
1038 svn_path => $svn_path );
1039 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1040 print "No changes\n$ta == $tb\n";
1041 }
1042 }
1043
1044 sub escape_uri_only {
1045 my ($uri) = @_;
1046 my @tmp;
1047 foreach (split m{/}, $uri) {
1048 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1049 push @tmp, $_;
1050 }
1051 join('/', @tmp);
1052 }
1053
1054 sub escape_url {
1055 my ($url) = @_;
1056 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1057 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1058 $url = "$scheme://$domain$uri";
1059 }
1060 $url;
1061 }
1062
1063 sub cmd_info {
1064 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1065 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1066 if (exists $_[1]) {
1067 die "Too many arguments specified\n";
1068 }
1069
1070 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1071
1072 if (!$file_type && !$diff_status) {
1073 print STDERR "svn: '$path' is not under version control\n";
1074 exit 1;
1075 }
1076
1077 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1078 unless ($gs) {
1079 die "Unable to determine upstream SVN information from ",
1080 "working tree history\n";
1081 }
1082
1083 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1084 $path = "." if $path eq "";
1085
1086 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1087
1088 if ($_url) {
1089 print escape_url($full_url), "\n";
1090 return;
1091 }
1092
1093 my $result = "Path: $path\n";
1094 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1095 $result .= "URL: " . escape_url($full_url) . "\n";
1096
1097 eval {
1098 my $repos_root = $gs->repos_root;
1099 Git::SVN::remove_username($repos_root);
1100 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1101 };
1102 if ($@) {
1103 $result .= "Repository Root: (offline)\n";
1104 }
1105 ::_req_svn();
1106 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1107 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1108 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1109
1110 $result .= "Node Kind: " .
1111 ($file_type eq "dir" ? "directory" : "file") . "\n";
1112
1113 my $schedule = $diff_status eq "A"
1114 ? "add"
1115 : ($diff_status eq "D" ? "delete" : "normal");
1116 $result .= "Schedule: $schedule\n";
1117
1118 if ($diff_status eq "A") {
1119 print $result, "\n";
1120 return;
1121 }
1122
1123 my ($lc_author, $lc_rev, $lc_date_utc);
1124 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1125 my $log = command_output_pipe(@args);
1126 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1127 while (<$log>) {
1128 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1129 $lc_author = $1;
1130 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1131 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1132 (undef, $lc_rev, undef) = ::extract_metadata($1);
1133 }
1134 }
1135 close $log;
1136
1137 Git::SVN::Log::set_local_timezone();
1138
1139 $result .= "Last Changed Author: $lc_author\n";
1140 $result .= "Last Changed Rev: $lc_rev\n";
1141 $result .= "Last Changed Date: " .
1142 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1143
1144 if ($file_type ne "dir") {
1145 my $text_last_updated_date =
1146 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1147 $result .=
1148 "Text Last Updated: " .
1149 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1150 "\n";
1151 my $checksum;
1152 if ($diff_status eq "D") {
1153 my ($fh, $ctx) =
1154 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1155 if ($file_type eq "link") {
1156 my $file_name = <$fh>;
1157 $checksum = md5sum("link $file_name");
1158 } else {
1159 $checksum = md5sum($fh);
1160 }
1161 command_close_pipe($fh, $ctx);
1162 } elsif ($file_type eq "link") {
1163 my $file_name =
1164 command(qw(cat-file blob), "HEAD:$path");
1165 $checksum =
1166 md5sum("link " . $file_name);
1167 } else {
1168 open FILE, "<", $path or die $!;
1169 $checksum = md5sum(\*FILE);
1170 close FILE or die $!;
1171 }
1172 $result .= "Checksum: " . $checksum . "\n";
1173 }
1174
1175 print $result, "\n";
1176 }
1177
1178 sub cmd_reset {
1179 my $target = shift || $_revision or die "SVN revision required\n";
1180 $target = $1 if $target =~ /^r(\d+)$/;
1181 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1182 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1183 unless ($gs) {
1184 die "Unable to determine upstream SVN information from ".
1185 "history\n";
1186 }
1187 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1188 $gs->rev_map_set($r, $c, 'reset', $uuid);
1189 print "r$r = $c ($gs->{ref_id})\n";
1190 }
1191
1192 sub cmd_gc {
1193 if (!$can_compress) {
1194 warn "Compress::Zlib could not be found; unhandled.log " .
1195 "files will not be compressed.\n";
1196 }
1197 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1198 }
1199
1200 ########################### utility functions #########################
1201
1202 sub rebase_cmd {
1203 my @cmd = qw/rebase/;
1204 push @cmd, '-v' if $_verbose;
1205 push @cmd, qw/--merge/ if $_merge;
1206 push @cmd, "--strategy=$_strategy" if $_strategy;
1207 @cmd;
1208 }
1209
1210 sub post_fetch_checkout {
1211 return if $_no_checkout;
1212 my $gs = $Git::SVN::_head or return;
1213 return if verify_ref('refs/heads/master^0');
1214
1215 # look for "trunk" ref if it exists
1216 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1217 my $fetch = $remote->{fetch};
1218 if ($fetch) {
1219 foreach my $p (keys %$fetch) {
1220 basename($fetch->{$p}) eq 'trunk' or next;
1221 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1222 last;
1223 }
1224 }
1225
1226 my $valid_head = verify_ref('HEAD^0');
1227 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1228 return if ($valid_head || !verify_ref('HEAD^0'));
1229
1230 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1231 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1232 return if -f $index;
1233
1234 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1235 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1236 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1237 print STDERR "Checked out HEAD:\n ",
1238 $gs->full_url, " r", $gs->last_rev, "\n";
1239 $gs->mkemptydirs($gs->last_rev);
1240 }
1241
1242 sub complete_svn_url {
1243 my ($url, $path) = @_;
1244 $path =~ s#/+$##;
1245 if ($path !~ m#^[a-z\+]+://#) {
1246 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1247 fatal("E: '$path' is not a complete URL ",
1248 "and a separate URL is not specified");
1249 }
1250 return ($url, $path);
1251 }
1252 return ($path, '');
1253 }
1254
1255 sub complete_url_ls_init {
1256 my ($ra, $repo_path, $switch, $pfx) = @_;
1257 unless ($repo_path) {
1258 print STDERR "W: $switch not specified\n";
1259 return;
1260 }
1261 $repo_path =~ s#/+$##;
1262 if ($repo_path =~ m#^[a-z\+]+://#) {
1263 $ra = Git::SVN::Ra->new($repo_path);
1264 $repo_path = '';
1265 } else {
1266 $repo_path =~ s#^/+##;
1267 unless ($ra) {
1268 fatal("E: '$repo_path' is not a complete URL ",
1269 "and a separate URL is not specified");
1270 }
1271 }
1272 my $url = $ra->{url};
1273 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1274 my $k = "svn-remote.$gs->{repo_id}.url";
1275 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1276 if ($orig_url && ($orig_url ne $gs->{url})) {
1277 die "$k already set: $orig_url\n",
1278 "wanted to set to: $gs->{url}\n";
1279 }
1280 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1281 my $remote_path = "$gs->{path}/$repo_path";
1282 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1283 $remote_path =~ s#/+#/#g;
1284 $remote_path =~ s#^/##g;
1285 $remote_path .= "/*" if $remote_path !~ /\*/;
1286 my ($n) = ($switch =~ /^--(\w+)/);
1287 if (length $pfx && $pfx !~ m#/$#) {
1288 die "--prefix='$pfx' must have a trailing slash '/'\n";
1289 }
1290 command_noisy('config',
1291 '--add',
1292 "svn-remote.$gs->{repo_id}.$n",
1293 "$remote_path:refs/remotes/$pfx*" .
1294 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1295 }
1296
1297 sub verify_ref {
1298 my ($ref) = @_;
1299 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1300 { STDERR => 0 }); };
1301 }
1302
1303 sub get_tree_from_treeish {
1304 my ($treeish) = @_;
1305 # $treeish can be a symbolic ref, too:
1306 my $type = command_oneline(qw/cat-file -t/, $treeish);
1307 my $expected;
1308 while ($type eq 'tag') {
1309 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1310 }
1311 if ($type eq 'commit') {
1312 $expected = (grep /^tree /, command(qw/cat-file commit/,
1313 $treeish))[0];
1314 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1315 die "Unable to get tree from $treeish\n" unless $expected;
1316 } elsif ($type eq 'tree') {
1317 $expected = $treeish;
1318 } else {
1319 die "$treeish is a $type, expected tree, tag or commit\n";
1320 }
1321 return $expected;
1322 }
1323
1324 sub get_commit_entry {
1325 my ($treeish) = shift;
1326 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1327 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1328 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1329 open my $log_fh, '>', $commit_editmsg or croak $!;
1330
1331 my $type = command_oneline(qw/cat-file -t/, $treeish);
1332 if ($type eq 'commit' || $type eq 'tag') {
1333 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1334 $type, $treeish);
1335 my $in_msg = 0;
1336 my $author;
1337 my $saw_from = 0;
1338 my $msgbuf = "";
1339 while (<$msg_fh>) {
1340 if (!$in_msg) {
1341 $in_msg = 1 if (/^\s*$/);
1342 $author = $1 if (/^author (.*>)/);
1343 } elsif (/^git-svn-id: /) {
1344 # skip this for now, we regenerate the
1345 # correct one on re-fetch anyways
1346 # TODO: set *:merge properties or like...
1347 } else {
1348 if (/^From:/ || /^Signed-off-by:/) {
1349 $saw_from = 1;
1350 }
1351 $msgbuf .= $_;
1352 }
1353 }
1354 $msgbuf =~ s/\s+$//s;
1355 if ($Git::SVN::_add_author_from && defined($author)
1356 && !$saw_from) {
1357 $msgbuf .= "\n\nFrom: $author";
1358 }
1359 print $log_fh $msgbuf or croak $!;
1360 command_close_pipe($msg_fh, $ctx);
1361 }
1362 close $log_fh or croak $!;
1363
1364 if ($_edit || ($type eq 'tree')) {
1365 chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1366 system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1367 }
1368 rename $commit_editmsg, $commit_msg or croak $!;
1369 {
1370 require Encode;
1371 # SVN requires messages to be UTF-8 when entering the repo
1372 local $/;
1373 open $log_fh, '<', $commit_msg or croak $!;
1374 binmode $log_fh;
1375 chomp($log_entry{log} = <$log_fh>);
1376
1377 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1378 my $msg = $log_entry{log};
1379
1380 eval { $msg = Encode::decode($enc, $msg, 1) };
1381 if ($@) {
1382 die "Could not decode as $enc:\n", $msg,
1383 "\nPerhaps you need to set i18n.commitencoding\n";
1384 }
1385
1386 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1387 die "Could not encode as UTF-8:\n$msg\n" if $@;
1388
1389 $log_entry{log} = $msg;
1390
1391 close $log_fh or croak $!;
1392 }
1393 unlink $commit_msg;
1394 \%log_entry;
1395 }
1396
1397 sub s_to_file {
1398 my ($str, $file, $mode) = @_;
1399 open my $fd,'>',$file or croak $!;
1400 print $fd $str,"\n" or croak $!;
1401 close $fd or croak $!;
1402 chmod ($mode &~ umask, $file) if (defined $mode);
1403 }
1404
1405 sub file_to_s {
1406 my $file = shift;
1407 open my $fd,'<',$file or croak "$!: file: $file\n";
1408 local $/;
1409 my $ret = <$fd>;
1410 close $fd or croak $!;
1411 $ret =~ s/\s*$//s;
1412 return $ret;
1413 }
1414
1415 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1416 sub load_authors {
1417 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1418 my $log = $cmd eq 'log';
1419 while (<$authors>) {
1420 chomp;
1421 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1422 my ($user, $name, $email) = ($1, $2, $3);
1423 if ($log) {
1424 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1425 } else {
1426 $users{$user} = [$name, $email];
1427 }
1428 }
1429 close $authors or croak $!;
1430 }
1431
1432 # convert GetOpt::Long specs for use by git-config
1433 sub read_git_config {
1434 my $opts = shift;
1435 my @config_only;
1436 foreach my $o (keys %$opts) {
1437 # if we have mixedCase and a long option-only, then
1438 # it's a config-only variable that we don't need for
1439 # the command-line.
1440 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1441 my $v = $opts->{$o};
1442 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1443 $key =~ s/-//g;
1444 my $arg = 'git config';
1445 $arg .= ' --int' if ($o =~ /[:=]i$/);
1446 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1447 if (ref $v eq 'ARRAY') {
1448 chomp(my @tmp = `$arg --get-all svn.$key`);
1449 @$v = @tmp if @tmp;
1450 } else {
1451 chomp(my $tmp = `$arg --get svn.$key`);
1452 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1453 $$v = $tmp;
1454 }
1455 }
1456 }
1457 delete @$opts{@config_only} if @config_only;
1458 }
1459
1460 sub extract_metadata {
1461 my $id = shift or return (undef, undef, undef);
1462 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1463 \s([a-f\d\-]+)$/ix);
1464 if (!defined $rev || !$uuid || !$url) {
1465 # some of the original repositories I made had
1466 # identifiers like this:
1467 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1468 }
1469 return ($url, $rev, $uuid);
1470 }
1471
1472 sub cmt_metadata {
1473 return extract_metadata((grep(/^git-svn-id: /,
1474 command(qw/cat-file commit/, shift)))[-1]);
1475 }
1476
1477 sub cmt_sha2rev_batch {
1478 my %s2r;
1479 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1480 my $list = shift;
1481
1482 foreach my $sha (@{$list}) {
1483 my $first = 1;
1484 my $size = 0;
1485 print $out $sha, "\n";
1486
1487 while (my $line = <$in>) {
1488 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1489 last;
1490 } elsif ($first &&
1491 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1492 $first = 0;
1493 $size = $1;
1494 next;
1495 } elsif ($line =~ /^(git-svn-id: )/) {
1496 my (undef, $rev, undef) =
1497 extract_metadata($line);
1498 $s2r{$sha} = $rev;
1499 }
1500
1501 $size -= length($line);
1502 last if ($size == 0);
1503 }
1504 }
1505
1506 command_close_bidi_pipe($pid, $in, $out, $ctx);
1507
1508 return \%s2r;
1509 }
1510
1511 sub working_head_info {
1512 my ($head, $refs) = @_;
1513 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1514 my ($fh, $ctx) = command_output_pipe(@args, $head);
1515 my $hash;
1516 my %max;
1517 while (<$fh>) {
1518 if ( m{^commit ($::sha1)$} ) {
1519 unshift @$refs, $hash if $hash and $refs;
1520 $hash = $1;
1521 next;
1522 }
1523 next unless s{^\s*(git-svn-id:)}{$1};
1524 my ($url, $rev, $uuid) = extract_metadata($_);
1525 if (defined $url && defined $rev) {
1526 next if $max{$url} and $max{$url} < $rev;
1527 if (my $gs = Git::SVN->find_by_url($url)) {
1528 my $c = $gs->rev_map_get($rev, $uuid);
1529 if ($c && $c eq $hash) {
1530 close $fh; # break the pipe
1531 return ($url, $rev, $uuid, $gs);
1532 } else {
1533 $max{$url} ||= $gs->rev_map_max;
1534 }
1535 }
1536 }
1537 }
1538 command_close_pipe($fh, $ctx);
1539 (undef, undef, undef, undef);
1540 }
1541
1542 sub read_commit_parents {
1543 my ($parents, $c) = @_;
1544 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1545 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1546 @{$parents->{$c}} = split(/ /, $p);
1547 }
1548
1549 sub linearize_history {
1550 my ($gs, $refs) = @_;
1551 my %parents;
1552 foreach my $c (@$refs) {
1553 read_commit_parents(\%parents, $c);
1554 }
1555
1556 my @linear_refs;
1557 my %skip = ();
1558 my $last_svn_commit = $gs->last_commit;
1559 foreach my $c (reverse @$refs) {
1560 next if $c eq $last_svn_commit;
1561 last if $skip{$c};
1562
1563 unshift @linear_refs, $c;
1564 $skip{$c} = 1;
1565
1566 # we only want the first parent to diff against for linear
1567 # history, we save the rest to inject when we finalize the
1568 # svn commit
1569 my $fp_a = verify_ref("$c~1");
1570 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1571 if (!$fp_a || !$fp_b) {
1572 die "Commit $c\n",
1573 "has no parent commit, and therefore ",
1574 "nothing to diff against.\n",
1575 "You should be working from a repository ",
1576 "originally created by git-svn\n";
1577 }
1578 if ($fp_a ne $fp_b) {
1579 die "$c~1 = $fp_a, however parsing commit $c ",
1580 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1581 }
1582
1583 foreach my $p (@{$parents{$c}}) {
1584 $skip{$p} = 1;
1585 }
1586 }
1587 (\@linear_refs, \%parents);
1588 }
1589
1590 sub find_file_type_and_diff_status {
1591 my ($path) = @_;
1592 return ('dir', '') if $path eq '';
1593
1594 my $diff_output =
1595 command_oneline(qw(diff --cached --name-status --), $path) || "";
1596 my $diff_status = (split(' ', $diff_output))[0] || "";
1597
1598 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1599
1600 return (undef, undef) if !$diff_status && !$ls_tree;
1601
1602 if ($diff_status eq "A") {
1603 return ("link", $diff_status) if -l $path;
1604 return ("dir", $diff_status) if -d $path;
1605 return ("file", $diff_status);
1606 }
1607
1608 my $mode = (split(' ', $ls_tree))[0] || "";
1609
1610 return ("link", $diff_status) if $mode eq "120000";
1611 return ("dir", $diff_status) if $mode eq "040000";
1612 return ("file", $diff_status);
1613 }
1614
1615 sub md5sum {
1616 my $arg = shift;
1617 my $ref = ref $arg;
1618 my $md5 = Digest::MD5->new();
1619 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1620 $md5->addfile($arg) or croak $!;
1621 } elsif ($ref eq 'SCALAR') {
1622 $md5->add($$arg) or croak $!;
1623 } elsif (!$ref) {
1624 $md5->add($arg) or croak $!;
1625 } else {
1626 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1627 }
1628 return $md5->hexdigest();
1629 }
1630
1631 sub gc_directory {
1632 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1633 my $out_filename = $_ . ".gz";
1634 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1635 binmode $in_fh;
1636 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1637 die "Unable to open $out_filename: $!\n";
1638
1639 my $res;
1640 while ($res = sysread($in_fh, my $str, 1024)) {
1641 $gz->gzwrite($str) or
1642 die "Unable to write: ".$gz->gzerror()."!\n";
1643 }
1644 unlink $_ or die "unlink $File::Find::name: $!\n";
1645 } elsif (-f $_ && basename($_) eq "index") {
1646 unlink $_ or die "unlink $_: $!\n";
1647 }
1648 }
1649
1650 package Git::SVN;
1651 use strict;
1652 use warnings;
1653 use Fcntl qw/:DEFAULT :seek/;
1654 use constant rev_map_fmt => 'NH40';
1655 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1656 $_repack $_repack_flags $_use_svm_props $_head
1657 $_use_svnsync_props $no_reuse_existing $_minimize_url
1658 $_use_log_author $_add_author_from $_localtime/;
1659 use Carp qw/croak/;
1660 use File::Path qw/mkpath/;
1661 use File::Copy qw/copy/;
1662 use IPC::Open3;
1663 use Memoize; # core since 5.8.0, Jul 2002
1664 use Memoize::Storable;
1665
1666 my ($_gc_nr, $_gc_period);
1667
1668 # properties that we do not log:
1669 my %SKIP_PROP;
1670 BEGIN {
1671 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1672 svn:special svn:executable
1673 svn:entry:committed-rev
1674 svn:entry:last-author
1675 svn:entry:uuid
1676 svn:entry:committed-date/;
1677
1678 # some options are read globally, but can be overridden locally
1679 # per [svn-remote "..."] section. Command-line options will *NOT*
1680 # override options set in an [svn-remote "..."] section
1681 no strict 'refs';
1682 for my $option (qw/follow_parent no_metadata use_svm_props
1683 use_svnsync_props/) {
1684 my $key = $option;
1685 $key =~ tr/_//d;
1686 my $prop = "-$option";
1687 *$option = sub {
1688 my ($self) = @_;
1689 return $self->{$prop} if exists $self->{$prop};
1690 my $k = "svn-remote.$self->{repo_id}.$key";
1691 eval { command_oneline(qw/config --get/, $k) };
1692 if ($@) {
1693 $self->{$prop} = ${"Git::SVN::_$option"};
1694 } else {
1695 my $v = command_oneline(qw/config --bool/,$k);
1696 $self->{$prop} = $v eq 'false' ? 0 : 1;
1697 }
1698 return $self->{$prop};
1699 }
1700 }
1701 }
1702
1703
1704 my (%LOCKFILES, %INDEX_FILES);
1705 END {
1706 unlink keys %LOCKFILES if %LOCKFILES;
1707 unlink keys %INDEX_FILES if %INDEX_FILES;
1708 }
1709
1710 sub resolve_local_globs {
1711 my ($url, $fetch, $glob_spec) = @_;
1712 return unless defined $glob_spec;
1713 my $ref = $glob_spec->{ref};
1714 my $path = $glob_spec->{path};
1715 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1716 next unless m#^$ref->{regex}$#;
1717 my $p = $1;
1718 my $pathname = desanitize_refname($path->full_path($p));
1719 my $refname = desanitize_refname($ref->full_path($p));
1720 if (my $existing = $fetch->{$pathname}) {
1721 if ($existing ne $refname) {
1722 die "Refspec conflict:\n",
1723 "existing: $existing\n",
1724 " globbed: $refname\n";
1725 }
1726 my $u = (::cmt_metadata("$refname"))[0];
1727 $u =~ s!^\Q$url\E(/|$)!! or die
1728 "$refname: '$url' not found in '$u'\n";
1729 if ($pathname ne $u) {
1730 warn "W: Refspec glob conflict ",
1731 "(ref: $refname):\n",
1732 "expected path: $pathname\n",
1733 " real path: $u\n",
1734 "Continuing ahead with $u\n";
1735 next;
1736 }
1737 } else {
1738 $fetch->{$pathname} = $refname;
1739 }
1740 }
1741 }
1742
1743 sub parse_revision_argument {
1744 my ($base, $head) = @_;
1745 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1746 return ($base, $head);
1747 }
1748 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1749 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1750 return ($head, $head) if ($::_revision eq 'HEAD');
1751 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1752 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1753 die "revision argument: $::_revision not understood by git-svn\n";
1754 }
1755
1756 sub fetch_all {
1757 my ($repo_id, $remotes) = @_;
1758 if (ref $repo_id) {
1759 my $gs = $repo_id;
1760 $repo_id = undef;
1761 $repo_id = $gs->{repo_id};
1762 }
1763 $remotes ||= read_all_remotes();
1764 my $remote = $remotes->{$repo_id} or
1765 die "[svn-remote \"$repo_id\"] unknown\n";
1766 my $fetch = $remote->{fetch};
1767 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1768 my (@gs, @globs);
1769 my $ra = Git::SVN::Ra->new($url);
1770 my $uuid = $ra->get_uuid;
1771 my $head = $ra->get_latest_revnum;
1772
1773 # ignore errors, $head revision may not even exist anymore
1774 eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
1775 warn "W: $@\n" if $@;
1776
1777 my $base = defined $fetch ? $head : 0;
1778
1779 # read the max revs for wildcard expansion (branches/*, tags/*)
1780 foreach my $t (qw/branches tags/) {
1781 defined $remote->{$t} or next;
1782 push @globs, @{$remote->{$t}};
1783
1784 my $max_rev = eval { tmp_config(qw/--int --get/,
1785 "svn-remote.$repo_id.${t}-maxRev") };
1786 if (defined $max_rev && ($max_rev < $base)) {
1787 $base = $max_rev;
1788 } elsif (!defined $max_rev) {
1789 $base = 0;
1790 }
1791 }
1792
1793 if ($fetch) {
1794 foreach my $p (sort keys %$fetch) {
1795 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1796 my $lr = $gs->rev_map_max;
1797 if (defined $lr) {
1798 $base = $lr if ($lr < $base);
1799 }
1800 push @gs, $gs;
1801 }
1802 }
1803
1804 ($base, $head) = parse_revision_argument($base, $head);
1805 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1806 }
1807
1808 sub read_all_remotes {
1809 my $r = {};
1810 my $use_svm_props = eval { command_oneline(qw/config --bool
1811 svn.useSvmProps/) };
1812 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1813 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1814 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1815 if (m!^(.+)\.fetch=$svn_refspec$!) {
1816 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1817 die("svn-remote.$remote: remote ref '$remote_ref' "
1818 . "must start with 'refs/'\n")
1819 unless $remote_ref =~ m{^refs/};
1820 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1821 $r->{$remote}->{svm} = {} if $use_svm_props;
1822 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1823 $r->{$1}->{svm} = {};
1824 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1825 $r->{$1}->{url} = $2;
1826 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1827 my ($remote, $t, $local_ref, $remote_ref) =
1828 ($1, $2, $3, $4);
1829 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1830 . "must start with 'refs/'\n")
1831 unless $remote_ref =~ m{^refs/};
1832 my $rs = {
1833 t => $t,
1834 remote => $remote,
1835 path => Git::SVN::GlobSpec->new($local_ref, 1),
1836 ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
1837 if (length($rs->{ref}->{right}) != 0) {
1838 die "The '*' glob character must be the last ",
1839 "character of '$remote_ref'\n";
1840 }
1841 push @{ $r->{$remote}->{$t} }, $rs;
1842 }
1843 }
1844
1845 map {
1846 if (defined $r->{$_}->{svm}) {
1847 my $svm;
1848 eval {
1849 my $section = "svn-remote.$_";
1850 $svm = {
1851 source => tmp_config('--get',
1852 "$section.svm-source"),
1853 replace => tmp_config('--get',
1854 "$section.svm-replace"),
1855 }
1856 };
1857 $r->{$_}->{svm} = $svm;
1858 }
1859 } keys %$r;
1860
1861 $r;
1862 }
1863
1864 sub init_vars {
1865 $_gc_nr = $_gc_period = 1000;
1866 if (defined $_repack || defined $_repack_flags) {
1867 warn "Repack options are obsolete; they have no effect.\n";
1868 }
1869 }
1870
1871 sub verify_remotes_sanity {
1872 return unless -d $ENV{GIT_DIR};
1873 my %seen;
1874 foreach (command(qw/config -l/)) {
1875 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1876 if ($seen{$1}) {
1877 die "Remote ref refs/remote/$1 is tracked by",
1878 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1879 "Please resolve this ambiguity in ",
1880 "your git configuration file before ",
1881 "continuing\n";
1882 }
1883 $seen{$1} = $_;
1884 }
1885 }
1886 }
1887
1888 sub find_existing_remote {
1889 my ($url, $remotes) = @_;
1890 return undef if $no_reuse_existing;
1891 my $existing;
1892 foreach my $repo_id (keys %$remotes) {
1893 my $u = $remotes->{$repo_id}->{url} or next;
1894 next if $u ne $url;
1895 $existing = $repo_id;
1896 last;
1897 }
1898 $existing;
1899 }
1900
1901 sub init_remote_config {
1902 my ($self, $url, $no_write) = @_;
1903 $url =~ s!/+$!!; # strip trailing slash
1904 my $r = read_all_remotes();
1905 my $existing = find_existing_remote($url, $r);
1906 if ($existing) {
1907 unless ($no_write) {
1908 print STDERR "Using existing ",
1909 "[svn-remote \"$existing\"]\n";
1910 }
1911 $self->{repo_id} = $existing;
1912 } elsif ($_minimize_url) {
1913 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1914 $existing = find_existing_remote($min_url, $r);
1915 if ($existing) {
1916 unless ($no_write) {
1917 print STDERR "Using existing ",
1918 "[svn-remote \"$existing\"]\n";
1919 }
1920 $self->{repo_id} = $existing;
1921 }
1922 if ($min_url ne $url) {
1923 unless ($no_write) {
1924 print STDERR "Using higher level of URL: ",
1925 "$url => $min_url\n";
1926 }
1927 my $old_path = $self->{path};
1928 $self->{path} = $url;
1929 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1930 if (length $old_path) {
1931 $self->{path} .= "/$old_path";
1932 }
1933 $url = $min_url;
1934 }
1935 }
1936 my $orig_url;
1937 if (!$existing) {
1938 # verify that we aren't overwriting anything:
1939 $orig_url = eval {
1940 command_oneline('config', '--get',
1941 "svn-remote.$self->{repo_id}.url")
1942 };
1943 if ($orig_url && ($orig_url ne $url)) {
1944 die "svn-remote.$self->{repo_id}.url already set: ",
1945 "$orig_url\nwanted to set to: $url\n";
1946 }
1947 }
1948 my ($xrepo_id, $xpath) = find_ref($self->refname);
1949 if (!$no_write && defined $xpath) {
1950 die "svn-remote.$xrepo_id.fetch already set to track ",
1951 "$xpath:", $self->refname, "\n";
1952 }
1953 unless ($no_write) {
1954 command_noisy('config',
1955 "svn-remote.$self->{repo_id}.url", $url);
1956 $self->{path} =~ s{^/}{};
1957 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1958 command_noisy('config', '--add',
1959 "svn-remote.$self->{repo_id}.fetch",
1960 "$self->{path}:".$self->refname);
1961 }
1962 $self->{url} = $url;
1963 }
1964
1965 sub find_by_url { # repos_root and, path are optional
1966 my ($class, $full_url, $repos_root, $path) = @_;
1967
1968 return undef unless defined $full_url;
1969 remove_username($full_url);
1970 remove_username($repos_root) if defined $repos_root;
1971 my $remotes = read_all_remotes();
1972 if (defined $full_url && defined $repos_root && !defined $path) {
1973 $path = $full_url;
1974 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1975 }
1976 foreach my $repo_id (keys %$remotes) {
1977 my $u = $remotes->{$repo_id}->{url} or next;
1978 remove_username($u);
1979 next if defined $repos_root && $repos_root ne $u;
1980
1981 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1982 foreach my $t (qw/branches tags/) {
1983 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1984 resolve_local_globs($u, $fetch, $globspec);
1985 }
1986 }
1987 my $p = $path;
1988 my $rwr = rewrite_root({repo_id => $repo_id});
1989 my $svm = $remotes->{$repo_id}->{svm}
1990 if defined $remotes->{$repo_id}->{svm};
1991 unless (defined $p) {
1992 $p = $full_url;
1993 my $z = $u;
1994 my $prefix = '';
1995 if ($rwr) {
1996 $z = $rwr;
1997 remove_username($z);
1998 } elsif (defined $svm) {
1999 $z = $svm->{source};
2000 $prefix = $svm->{replace};
2001 $prefix =~ s#^\Q$u\E(?:/|$)##;
2002 $prefix =~ s#/$##;
2003 }
2004 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2005 }
2006 foreach my $f (keys %$fetch) {
2007 next if $f ne $p;
2008 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2009 }
2010 }
2011 undef;
2012 }
2013
2014 sub init {
2015 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2016 my $self = _new($class, $repo_id, $ref_id, $path);
2017 if (defined $url) {
2018 $self->init_remote_config($url, $no_write);
2019 }
2020 $self;
2021 }
2022
2023 sub find_ref {
2024 my ($ref_id) = @_;
2025 foreach (command(qw/config -l/)) {
2026 next unless m!^svn-remote\.(.+)\.fetch=
2027 \s*(.*?)\s*:\s*(.+?)\s*$!x;
2028 my ($repo_id, $path, $ref) = ($1, $2, $3);
2029 if ($ref eq $ref_id) {
2030 $path = '' if ($path =~ m#^\./?#);
2031 return ($repo_id, $path);
2032 }
2033 }
2034 (undef, undef, undef);
2035 }
2036
2037 sub new {
2038 my ($class, $ref_id, $repo_id, $path) = @_;
2039 if (defined $ref_id && !defined $repo_id && !defined $path) {
2040 ($repo_id, $path) = find_ref($ref_id);
2041 if (!defined $repo_id) {
2042 die "Could not find a \"svn-remote.*.fetch\" key ",
2043 "in the repository configuration matching: ",
2044 "$ref_id\n";
2045 }
2046 }
2047 my $self = _new($class, $repo_id, $ref_id, $path);
2048 if (!defined $self->{path} || !length $self->{path}) {
2049 my $fetch = command_oneline('config', '--get',
2050 "svn-remote.$repo_id.fetch",
2051 ":$ref_id\$") or
2052 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2053 "\":$ref_id\$\" in config\n";
2054 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2055 }
2056 $self->{url} = command_oneline('config', '--get',
2057 "svn-remote.$repo_id.url") or
2058 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2059 $self->rebuild;
2060 $self;
2061 }
2062
2063 sub refname {
2064 my ($refname) = $_[0]->{ref_id} ;
2065
2066 # It cannot end with a slash /, we'll throw up on this because
2067 # SVN can't have directories with a slash in their name, either:
2068 if ($refname =~ m{/$}) {
2069 die "ref: '$refname' ends with a trailing slash, this is ",
2070 "not permitted by git nor Subversion\n";
2071 }
2072
2073 # It cannot have ASCII control character space, tilde ~, caret ^,
2074 # colon :, question-mark ?, asterisk *, space, or open bracket [
2075 # anywhere.
2076 #
2077 # Additionally, % must be escaped because it is used for escaping
2078 # and we want our escaped refname to be reversible
2079 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2080
2081 # no slash-separated component can begin with a dot .
2082 # /.* becomes /%2E*
2083 $refname =~ s{/\.}{/%2E}g;
2084
2085 # It cannot have two consecutive dots .. anywhere
2086 # .. becomes %2E%2E
2087 $refname =~ s{\.\.}{%2E%2E}g;
2088
2089 return $refname;
2090 }
2091
2092 sub desanitize_refname {
2093 my ($refname) = @_;
2094 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2095 return $refname;
2096 }
2097
2098 sub svm_uuid {
2099 my ($self) = @_;
2100 return $self->{svm}->{uuid} if $self->svm;
2101 $self->ra;
2102 unless ($self->{svm}) {
2103 die "SVM UUID not cached, and reading remotely failed\n";
2104 }
2105 $self->{svm}->{uuid};
2106 }
2107
2108 sub svm {
2109 my ($self) = @_;
2110 return $self->{svm} if $self->{svm};
2111 my $svm;
2112 # see if we have it in our config, first:
2113 eval {
2114 my $section = "svn-remote.$self->{repo_id}";
2115 $svm = {
2116 source => tmp_config('--get', "$section.svm-source"),
2117 uuid => tmp_config('--get', "$section.svm-uuid"),
2118 replace => tmp_config('--get', "$section.svm-replace"),
2119 }
2120 };
2121 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2122 $self->{svm} = $svm;
2123 }
2124 $self->{svm};
2125 }
2126
2127 sub _set_svm_vars {
2128 my ($self, $ra) = @_;
2129 return $ra if $self->svm;
2130
2131 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2132 "(svm:source, svm:uuid) ",
2133 "from the following URLs:\n" );
2134 sub read_svm_props {
2135 my ($self, $ra, $path, $r) = @_;
2136 my $props = ($ra->get_dir($path, $r))[2];
2137 my $src = $props->{'svm:source'};
2138 my $uuid = $props->{'svm:uuid'};
2139 return undef if (!$src || !$uuid);
2140
2141 chomp($src, $uuid);
2142
2143 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2144 or die "doesn't look right - svm:uuid is '$uuid'\n";
2145
2146 # the '!' is used to mark the repos_root!/relative/path
2147 $src =~ s{/?!/?}{/};
2148 $src =~ s{/+$}{}; # no trailing slashes please
2149 # username is of no interest
2150 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2151
2152 my $replace = $ra->{url};
2153 $replace .= "/$path" if length $path;
2154
2155 my $section = "svn-remote.$self->{repo_id}";
2156 tmp_config("$section.svm-source", $src);
2157 tmp_config("$section.svm-replace", $replace);
2158 tmp_config("$section.svm-uuid", $uuid);
2159 $self->{svm} = {
2160 source => $src,
2161 uuid => $uuid,
2162 replace => $replace
2163 };
2164 }
2165
2166 my $r = $ra->get_latest_revnum;
2167 my $path = $self->{path};
2168 my %tried;
2169 while (length $path) {
2170 unless ($tried{"$self->{url}/$path"}) {
2171 return $ra if $self->read_svm_props($ra, $path, $r);
2172 $tried{"$self->{url}/$path"} = 1;
2173 }
2174 $path =~ s#/?[^/]+$##;
2175 }
2176 die "Path: '$path' should be ''\n" if $path ne '';
2177 return $ra if $self->read_svm_props($ra, $path, $r);
2178 $tried{"$self->{url}/$path"} = 1;
2179
2180 if ($ra->{repos_root} eq $self->{url}) {
2181 die @err, (map { " $_\n" } keys %tried), "\n";
2182 }
2183
2184 # nope, make sure we're connected to the repository root:
2185 my $ok;
2186 my @tried_b;
2187 $path = $ra->{svn_path};
2188 $ra = Git::SVN::Ra->new($ra->{repos_root});
2189 while (length $path) {
2190 unless ($tried{"$ra->{url}/$path"}) {
2191 $ok = $self->read_svm_props($ra, $path, $r);
2192 last if $ok;
2193 $tried{"$ra->{url}/$path"} = 1;
2194 }
2195 $path =~ s#/?[^/]+$##;
2196 }
2197 die "Path: '$path' should be ''\n" if $path ne '';
2198 $ok ||= $self->read_svm_props($ra, $path, $r);
2199 $tried{"$ra->{url}/$path"} = 1;
2200 if (!$ok) {
2201 die @err, (map { " $_\n" } keys %tried), "\n";
2202 }
2203 Git::SVN::Ra->new($self->{url});
2204 }
2205
2206 sub svnsync {
2207 my ($self) = @_;
2208 return $self->{svnsync} if $self->{svnsync};
2209
2210 if ($self->no_metadata) {
2211 die "Can't have both 'noMetadata' and ",
2212 "'useSvnsyncProps' options set!\n";
2213 }
2214 if ($self->rewrite_root) {
2215 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2216 "options set!\n";
2217 }
2218 if ($self->rewrite_uuid) {
2219 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2220 "options set!\n";
2221 }
2222
2223 my $svnsync;
2224 # see if we have it in our config, first:
2225 eval {
2226 my $section = "svn-remote.$self->{repo_id}";
2227
2228 my $url = tmp_config('--get', "$section.svnsync-url");
2229 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2230 die "doesn't look right - svn:sync-from-url is '$url'\n";
2231
2232 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2233 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2234 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2235
2236 $svnsync = { url => $url, uuid => $uuid }
2237 };
2238 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2239 return $self->{svnsync} = $svnsync;
2240 }
2241
2242 my $err = "useSvnsyncProps set, but failed to read " .
2243 "svnsync property: svn:sync-from-";
2244 my $rp = $self->ra->rev_proplist(0);
2245
2246 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2247 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2248 die "doesn't look right - svn:sync-from-url is '$url'\n";
2249
2250 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2251 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2252 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2253
2254 my $section = "svn-remote.$self->{repo_id}";
2255 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2256 tmp_config('--add', "$section.svnsync-url", $url);
2257 return $self->{svnsync} = { url => $url, uuid => $uuid };
2258 }
2259
2260 # this allows us to memoize our SVN::Ra UUID locally and avoid a
2261 # remote lookup (useful for 'git svn log').
2262 sub ra_uuid {
2263 my ($self) = @_;
2264 unless ($self->{ra_uuid}) {
2265 my $key = "svn-remote.$self->{repo_id}.uuid";
2266 my $uuid = eval { tmp_config('--get', $key) };
2267 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2268 $self->{ra_uuid} = $uuid;
2269 } else {
2270 die "ra_uuid called without URL\n" unless $self->{url};
2271 $self->{ra_uuid} = $self->ra->get_uuid;
2272 tmp_config('--add', $key, $self->{ra_uuid});
2273 }
2274 }
2275 $self->{ra_uuid};
2276 }
2277
2278 sub _set_repos_root {
2279 my ($self, $repos_root) = @_;
2280 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2281 $repos_root ||= $self->ra->{repos_root};
2282 tmp_config($k, $repos_root);
2283 $repos_root;
2284 }
2285
2286 sub repos_root {
2287 my ($self) = @_;
2288 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2289 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2290 }
2291
2292 sub ra {
2293 my ($self) = shift;
2294 my $ra = Git::SVN::Ra->new($self->{url});
2295 $self->_set_repos_root($ra->{repos_root});
2296 if ($self->use_svm_props && !$self->{svm}) {
2297 if ($self->no_metadata) {
2298 die "Can't have both 'noMetadata' and ",
2299 "'useSvmProps' options set!\n";
2300 } elsif ($self->use_svnsync_props) {
2301 die "Can't have both 'useSvnsyncProps' and ",
2302 "'useSvmProps' options set!\n";
2303 }
2304 $ra = $self->_set_svm_vars($ra);
2305 $self->{-want_revprops} = 1;
2306 }
2307 $ra;
2308 }
2309
2310 # prop_walk(PATH, REV, SUB)
2311 # -------------------------
2312 # Recursively traverse PATH at revision REV and invoke SUB for each
2313 # directory that contains a SVN property. SUB will be invoked as
2314 # follows: &SUB(gs, path, props); where `gs' is this instance of
2315 # Git::SVN, `path' the path to the directory where the properties
2316 # `props' were found. The `path' will be relative to point of checkout,
2317 # that is, if url://repo/trunk is the current Git branch, and that
2318 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
2319 # as `path' (note the trailing `/').
2320 sub prop_walk {
2321 my ($self, $path, $rev, $sub) = @_;
2322
2323 $path =~ s#^/##;
2324 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2325 $path =~ s#^/*#/#g;
2326 my $p = $path;
2327 # Strip the irrelevant part of the path.
2328 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2329 # Ensure the path is terminated by a `/'.
2330 $p =~ s#/*$#/#;
2331
2332 # The properties contain all the internal SVN stuff nobody
2333 # (usually) cares about.
2334 my $interesting_props = 0;
2335 foreach (keys %{$props}) {
2336 # If it doesn't start with `svn:', it must be a
2337 # user-defined property.
2338 ++$interesting_props and next if $_ !~ /^svn:/;
2339 # FIXME: Fragile, if SVN adds new public properties,
2340 # this needs to be updated.
2341 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2342 |eol-style|mime-type
2343 |externals|needs-lock)$/x;
2344 }
2345 &$sub($self, $p, $props) if $interesting_props;
2346
2347 foreach (sort keys %$dirent) {
2348 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2349 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2350 }
2351 }
2352
2353 sub last_rev { ($_[0]->last_rev_commit)[0] }
2354 sub last_commit { ($_[0]->last_rev_commit)[1] }
2355
2356 # returns the newest SVN revision number and newest commit SHA1
2357 sub last_rev_commit {
2358 my ($self) = @_;
2359 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2360 return ($self->{last_rev}, $self->{last_commit});
2361 }
2362 my $c = ::verify_ref($self->refname.'^0');
2363 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2364 my $rev = (::cmt_metadata($c))[1];
2365 if (defined $rev) {
2366 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2367 return ($rev, $c);
2368 }
2369 }
2370 my $map_path = $self->map_path;
2371 unless (-e $map_path) {
2372 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2373 return (undef, undef);
2374 }
2375 my ($rev, $commit) = $self->rev_map_max(1);
2376 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2377 return ($rev, $commit);
2378 }
2379
2380 sub get_fetch_range {
2381 my ($self, $min, $max) = @_;
2382 $max ||= $self->ra->get_latest_revnum;
2383 $min ||= $self->rev_map_max;
2384 (++$min, $max);
2385 }
2386
2387 sub tmp_config {
2388 my (@args) = @_;
2389 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2390 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2391 if (! -f $config && -f $old_def_config) {
2392 rename $old_def_config, $config or
2393 die "Failed rename $old_def_config => $config: $!\n";
2394 }
2395 my $old_config = $ENV{GIT_CONFIG};
2396 $ENV{GIT_CONFIG} = $config;
2397 $@ = undef;
2398 my @ret = eval {
2399 unless (-f $config) {
2400 mkfile($config);
2401 open my $fh, '>', $config or
2402 die "Can't open $config: $!\n";
2403 print $fh "; This file is used internally by ",
2404 "git-svn\n" or die
2405 "Couldn't write to $config: $!\n";
2406 print $fh "; You should not have to edit it\n" or
2407 die "Couldn't write to $config: $!\n";
2408 close $fh or die "Couldn't close $config: $!\n";
2409 }
2410 command('config', @args);
2411 };
2412 my $err = $@;
2413 if (defined $old_config) {
2414 $ENV{GIT_CONFIG} = $old_config;
2415 } else {
2416 delete $ENV{GIT_CONFIG};
2417 }
2418 die $err if $err;
2419 wantarray ? @ret : $ret[0];
2420 }
2421
2422 sub tmp_index_do {
2423 my ($self, $sub) = @_;
2424 my $old_index = $ENV{GIT_INDEX_FILE};
2425 $ENV{GIT_INDEX_FILE} = $self->{index};
2426 $@ = undef;
2427 my @ret = eval {
2428 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2429 mkpath([$dir]) unless -d $dir;
2430 &$sub;
2431 };
2432 my $err = $@;
2433 if (defined $old_index) {
2434 $ENV{GIT_INDEX_FILE} = $old_index;
2435 } else {
2436 delete $ENV{GIT_INDEX_FILE};
2437 }
2438 die $err if $err;
2439 wantarray ? @ret : $ret[0];
2440 }
2441
2442 sub assert_index_clean {
2443 my ($self, $treeish) = @_;
2444
2445 $self->tmp_index_do(sub {
2446 command_noisy('read-tree', $treeish) unless -e $self->{index};
2447 my $x = command_oneline('write-tree');
2448 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2449 /^tree ($::sha1)/mo);
2450 return if $y eq $x;
2451
2452 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2453 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2454 command_noisy('read-tree', $treeish);
2455 $x = command_oneline('write-tree');
2456 if ($y ne $x) {
2457 ::fatal "trees ($treeish) $y != $x\n",
2458 "Something is seriously wrong...";
2459 }
2460 });
2461 }
2462
2463 sub get_commit_parents {
2464 my ($self, $log_entry) = @_;
2465 my (%seen, @ret, @tmp);
2466 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2467 if (my $ip = $self->{inject_parents}) {
2468 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2469 push @tmp, $commit;
2470 }
2471 }
2472 if (my $cur = ::verify_ref($self->refname.'^0')) {
2473 push @tmp, $cur;
2474 }
2475 if (my $ipd = $self->{inject_parents_dcommit}) {
2476 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2477 push @tmp, @$commit;
2478 }
2479 }
2480 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2481 while (my $p = shift @tmp) {
2482 next if $seen{$p};
2483 $seen{$p} = 1;
2484 push @ret, $p;
2485 }
2486 @ret;
2487 }
2488
2489 sub rewrite_root {
2490 my ($self) = @_;
2491 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2492 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2493 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2494 if ($rwr) {
2495 $rwr =~ s#/+$##;
2496 if ($rwr !~ m#^[a-z\+]+://#) {
2497 die "$rwr is not a valid URL (key: $k)\n";
2498 }
2499 }
2500 $self->{-rewrite_root} = $rwr;
2501 }
2502
2503 sub rewrite_uuid {
2504 my ($self) = @_;
2505 return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2506 my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2507 my $rwid = eval { command_oneline(qw/config --get/, $k) };
2508 if ($rwid) {
2509 $rwid =~ s#/+$##;
2510 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2511 die "$rwid is not a valid UUID (key: $k)\n";
2512 }
2513 }
2514 $self->{-rewrite_uuid} = $rwid;
2515 }
2516
2517 sub metadata_url {
2518 my ($self) = @_;
2519 ($self->rewrite_root || $self->{url}) .
2520 (length $self->{path} ? '/' . $self->{path} : '');
2521 }
2522
2523 sub full_url {
2524 my ($self) = @_;
2525 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2526 }
2527
2528
2529 sub set_commit_header_env {
2530 my ($log_entry) = @_;
2531 my %env;
2532 foreach my $ned (qw/NAME EMAIL DATE/) {
2533 foreach my $ac (qw/AUTHOR COMMITTER/) {
2534 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2535 }
2536 }
2537
2538 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2539 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2540 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2541
2542 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2543 ? $log_entry->{commit_name}
2544 : $log_entry->{name};
2545 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2546 ? $log_entry->{commit_email}
2547 : $log_entry->{email};
2548 \%env;
2549 }
2550
2551 sub restore_commit_header_env {
2552 my ($env) = @_;
2553 foreach my $ned (qw/NAME EMAIL DATE/) {
2554 foreach my $ac (qw/AUTHOR COMMITTER/) {
2555 my $k = "GIT_${ac}_${ned}";
2556 if (defined $env->{$k}) {
2557 $ENV{$k} = $env->{$k};
2558 } else {
2559 delete $ENV{$k};
2560 }
2561 }
2562 }
2563 }
2564
2565 sub gc {
2566 command_noisy('gc', '--auto');
2567 };
2568
2569 sub do_git_commit {
2570 my ($self, $log_entry) = @_;
2571 my $lr = $self->last_rev;
2572 if (defined $lr && $lr >= $log_entry->{revision}) {
2573 die "Last fetched revision of ", $self->refname,
2574 " was r$lr, but we are about to fetch: ",
2575 "r$log_entry->{revision}!\n";
2576 }
2577 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2578 croak "$log_entry->{revision} = $c already exists! ",
2579 "Why are we refetching it?\n";
2580 }
2581 my $old_env = set_commit_header_env($log_entry);
2582 my $tree = $log_entry->{tree};
2583 if (!defined $tree) {
2584 $tree = $self->tmp_index_do(sub {
2585 command_oneline('write-tree') });
2586 }
2587 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2588
2589 my @exec = ('git', 'commit-tree', $tree);
2590 foreach ($self->get_commit_parents($log_entry)) {
2591 push @exec, '-p', $_;
2592 }
2593 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2594 or croak $!;
2595 binmode $msg_fh;
2596
2597 # we always get UTF-8 from SVN, but we may want our commits in
2598 # a different encoding.
2599 if (my $enc = Git::config('i18n.commitencoding')) {
2600 require Encode;
2601 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2602 }
2603 print $msg_fh $log_entry->{log} or croak $!;
2604 restore_commit_header_env($old_env);
2605 unless ($self->no_metadata) {
2606 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2607 or croak $!;
2608 }
2609 $msg_fh->flush == 0 or croak $!;
2610 close $msg_fh or croak $!;
2611 chomp(my $commit = do { local $/; <$out_fh> });
2612 close $out_fh or croak $!;
2613 waitpid $pid, 0;
2614 croak $? if $?;
2615 if ($commit !~ /^$::sha1$/o) {
2616 die "Failed to commit, invalid sha1: $commit\n";
2617 }
2618
2619 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2620
2621 $self->{last_rev} = $log_entry->{revision};
2622 $self->{last_commit} = $commit;
2623 print "r$log_entry->{revision}" unless $::_q > 1;
2624 if (defined $log_entry->{svm_revision}) {
2625 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2626 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2627 0, $self->svm_uuid);
2628 }
2629 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2630 if (--$_gc_nr == 0) {
2631 $_gc_nr = $_gc_period;
2632 gc();
2633 }
2634 return $commit;
2635 }
2636
2637 sub match_paths {
2638 my ($self, $paths, $r) = @_;
2639 return 1 if $self->{path} eq '';
2640 if (my $path = $paths->{"/$self->{path}"}) {
2641 return ($path->{action} eq 'D') ? 0 : 1;
2642 }
2643 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2644 if (grep /$self->{path_regex}/, keys %$paths) {
2645 return 1;
2646 }
2647 my $c = '';
2648 foreach (split m#/#, $self->{path}) {
2649 $c .= "/$_";
2650 next unless ($paths->{$c} &&
2651 ($paths->{$c}->{action} =~ /^[AR]$/));
2652 if ($self->ra->check_path($self->{path}, $r) ==
2653 $SVN::Node::dir) {
2654 return 1;
2655 }
2656 }
2657 return 0;
2658 }
2659
2660 sub find_parent_branch {
2661 my ($self, $paths, $rev) = @_;
2662 return undef unless $self->follow_parent;
2663 unless (defined $paths) {
2664 my $err_handler = $SVN::Error::handler;
2665 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2666 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2667 sub { $paths = $_[0] });
2668 $SVN::Error::handler = $err_handler;
2669 }
2670 return undef unless defined $paths;
2671
2672 # look for a parent from another branch:
2673 my @b_path_components = split m#/#, $self->{path};
2674 my @a_path_components;
2675 my $i;
2676 while (@b_path_components) {
2677 $i = $paths->{'/'.join('/', @b_path_components)};
2678 last if $i && defined $i->{copyfrom_path};
2679 unshift(@a_path_components, pop(@b_path_components));
2680 }
2681 return undef unless defined $i && defined $i->{copyfrom_path};
2682 my $branch_from = $i->{copyfrom_path};
2683 if (@a_path_components) {
2684 print STDERR "branch_from: $branch_from => ";
2685 $branch_from .= '/'.join('/', @a_path_components);
2686 print STDERR $branch_from, "\n";
2687 }
2688 my $r = $i->{copyfrom_rev};
2689 my $repos_root = $self->ra->{repos_root};
2690 my $url = $self->ra->{url};
2691 my $new_url = $url . $branch_from;
2692 print STDERR "Found possible branch point: ",
2693 "$new_url => ", $self->full_url, ", $r\n"
2694 unless $::_q > 1;
2695 $branch_from =~ s#^/##;
2696 my $gs = $self->other_gs($new_url, $url,
2697 $branch_from, $r, $self->{ref_id});
2698 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2699 {
2700 my ($base, $head);
2701 if (!defined $r0 || !defined $parent) {
2702 ($base, $head) = parse_revision_argument(0, $r);
2703 } else {
2704 if ($r0 < $r) {
2705 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2706 0, 1, sub { $base = $_[1] - 1 });
2707 }
2708 }
2709 if (defined $base && $base <= $r) {
2710 $gs->fetch($base, $r);
2711 }
2712 ($r0, $parent) = $gs->find_rev_before($r, 1);
2713 }
2714 if (defined $r0 && defined $parent) {
2715 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2716 unless $::_q > 1;
2717 my $ed;
2718 if ($self->ra->can_do_switch) {
2719 $self->assert_index_clean($parent);
2720 print STDERR "Following parent with do_switch\n"
2721 unless $::_q > 1;
2722 # do_switch works with svn/trunk >= r22312, but that
2723 # is not included with SVN 1.4.3 (the latest version
2724 # at the moment), so we can't rely on it
2725 $self->{last_rev} = $r0;
2726 $self->{last_commit} = $parent;
2727 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2728 $gs->ra->gs_do_switch($r0, $rev, $gs,
2729 $self->full_url, $ed)
2730 or die "SVN connection failed somewhere...\n";
2731 } elsif ($self->ra->trees_match($new_url, $r0,
2732 $self->full_url, $rev)) {
2733 print STDERR "Trees match:\n",
2734 " $new_url\@$r0\n",
2735 " ${\$self->full_url}\@$rev\n",
2736 "Following parent with no changes\n"
2737 unless $::_q > 1;
2738 $self->tmp_index_do(sub {
2739 command_noisy('read-tree', $parent);
2740 });
2741 $self->{last_commit} = $parent;
2742 } else {
2743 print STDERR "Following parent with do_update\n"
2744 unless $::_q > 1;
2745 $ed = SVN::Git::Fetcher->new($self);
2746 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2747 or die "SVN connection failed somewhere...\n";
2748 }
2749 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2750 return $self->make_log_entry($rev, [$parent], $ed);
2751 }
2752 return undef;
2753 }
2754
2755 sub do_fetch {
2756 my ($self, $paths, $rev) = @_;
2757 my $ed;
2758 my ($last_rev, @parents);
2759 if (my $lc = $self->last_commit) {
2760 # we can have a branch that was deleted, then re-added
2761 # under the same name but copied from another path, in
2762 # which case we'll have multiple parents (we don't
2763 # want to break the original ref, nor lose copypath info):
2764 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2765 push @{$log_entry->{parents}}, $lc;
2766 return $log_entry;
2767 }
2768 $ed = SVN::Git::Fetcher->new($self);
2769 $last_rev = $self->{last_rev};
2770 $ed->{c} = $lc;
2771 @parents = ($lc);
2772 } else {
2773 $last_rev = $rev;
2774 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2775 return $log_entry;
2776 }
2777 $ed = SVN::Git::Fetcher->new($self);
2778 }
2779 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2780 die "SVN connection failed somewhere...\n";
2781 }
2782 $self->make_log_entry($rev, \@parents, $ed);
2783 }
2784
2785 sub mkemptydirs {
2786 my ($self, $r) = @_;
2787
2788 sub scan {
2789 my ($r, $empty_dirs, $line) = @_;
2790 if (defined $r && $line =~ /^r(\d+)$/) {
2791 return 0 if $1 > $r;
2792 } elsif ($line =~ /^ \+empty_dir: (.+)$/) {
2793 $empty_dirs->{$1} = 1;
2794 } elsif ($line =~ /^ \-empty_dir: (.+)$/) {
2795 my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
2796 delete @$empty_dirs{@d};
2797 }
2798 1; # continue
2799 };
2800
2801 my %empty_dirs = ();
2802 my $gz_file = "$self->{dir}/unhandled.log.gz";
2803 if (-f $gz_file) {
2804 if (!$can_compress) {
2805 warn "Compress::Zlib could not be found; ",
2806 "empty directories in $gz_file will not be read\n";
2807 } else {
2808 my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
2809 die "Unable to open $gz_file: $!\n";
2810 my $line;
2811 while ($gz->gzreadline($line) > 0) {
2812 scan($r, \%empty_dirs, $line) or last;
2813 }
2814 $gz->gzclose;
2815 }
2816 }
2817
2818 if (open my $fh, '<', "$self->{dir}/unhandled.log") {
2819 binmode $fh or croak "binmode: $!";
2820 while (<$fh>) {
2821 scan($r, \%empty_dirs, $_) or last;
2822 }
2823 close $fh;
2824 }
2825
2826 my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2827 foreach my $d (sort keys %empty_dirs) {
2828 $d = uri_decode($d);
2829 $d =~ s/$strip//;
2830 next if -d $d;
2831 if (-e _) {
2832 warn "$d exists but is not a directory\n";
2833 } else {
2834 print "creating empty directory: $d\n";
2835 mkpath([$d]);
2836 }
2837 }
2838 }
2839
2840 sub get_untracked {
2841 my ($self, $ed) = @_;
2842 my @out;
2843 my $h = $ed->{empty};
2844 foreach (sort keys %$h) {
2845 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2846 push @out, " $act: " . uri_encode($_);
2847 warn "W: $act: $_\n";
2848 }
2849 foreach my $t (qw/dir_prop file_prop/) {
2850 $h = $ed->{$t} or next;
2851 foreach my $path (sort keys %$h) {
2852 my $ppath = $path eq '' ? '.' : $path;
2853 foreach my $prop (sort keys %{$h->{$path}}) {
2854 next if $SKIP_PROP{$prop};
2855 my $v = $h->{$path}->{$prop};
2856 my $t_ppath_prop = "$t: " .
2857 uri_encode($ppath) . ' ' .
2858 uri_encode($prop);
2859 if (defined $v) {
2860 push @out, " +$t_ppath_prop " .
2861 uri_encode($v);
2862 } else {
2863 push @out, " -$t_ppath_prop";
2864 }
2865 }
2866 }
2867 }
2868 foreach my $t (qw/absent_file absent_directory/) {
2869 $h = $ed->{$t} or next;
2870 foreach my $parent (sort keys %$h) {
2871 foreach my $path (sort @{$h->{$parent}}) {
2872 push @out, " $t: " .
2873 uri_encode("$parent/$path");
2874 warn "W: $t: $parent/$path ",
2875 "Insufficient permissions?\n";
2876 }
2877 }
2878 }
2879 \@out;
2880 }
2881
2882 # parse_svn_date(DATE)
2883 # --------------------
2884 # Given a date (in UTC) from Subversion, return a string in the format
2885 # "<TZ Offset> <local date/time>" that Git will use.
2886 #
2887 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
2888 # is true we'll convert it to the local timezone instead.
2889 sub parse_svn_date {
2890 my $date = shift || return '+0000 1970-01-01 00:00:00';
2891 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2892 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2893 croak "Unable to parse date: $date\n";
2894 my $parsed_date; # Set next.
2895
2896 if ($Git::SVN::_localtime) {
2897 # Translate the Subversion datetime to an epoch time.
2898 # Begin by switching ourselves to $date's timezone, UTC.
2899 my $old_env_TZ = $ENV{TZ};
2900 $ENV{TZ} = 'UTC';
2901
2902 my $epoch_in_UTC =
2903 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2904
2905 # Determine our local timezone (including DST) at the
2906 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2907 # value of TZ, if any, at the time we were run.
2908 if (defined $Git::SVN::Log::TZ) {
2909 $ENV{TZ} = $Git::SVN::Log::TZ;
2910 } else {
2911 delete $ENV{TZ};
2912 }
2913
2914 my $our_TZ =
2915 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2916
2917 # This converts $epoch_in_UTC into our local timezone.
2918 my ($sec, $min, $hour, $mday, $mon, $year,
2919 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2920
2921 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2922 $our_TZ, $year + 1900, $mon + 1,
2923 $mday, $hour, $min, $sec);
2924
2925 # Reset us to the timezone in effect when we entered
2926 # this routine.
2927 if (defined $old_env_TZ) {
2928 $ENV{TZ} = $old_env_TZ;
2929 } else {
2930 delete $ENV{TZ};
2931 }
2932 } else {
2933 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2934 }
2935
2936 return $parsed_date;
2937 }
2938
2939 sub other_gs {
2940 my ($self, $new_url, $url,
2941 $branch_from, $r, $old_ref_id) = @_;
2942 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2943 unless ($gs) {
2944 my $ref_id = $old_ref_id;
2945 $ref_id =~ s/\@\d+$//;
2946 $ref_id .= "\@$r";
2947 # just grow a tail if we're not unique enough :x
2948 $ref_id .= '-' while find_ref($ref_id);
2949 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2950 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2951 if ($u =~ s#^\Q$url\E(/|$)##) {
2952 $p = $u;
2953 $u = $url;
2954 $repo_id = $self->{repo_id};
2955 }
2956 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2957 }
2958 $gs
2959 }
2960
2961 sub call_authors_prog {
2962 my ($orig_author) = @_;
2963 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2964 my $author = `$::_authors_prog $orig_author`;
2965 if ($? != 0) {
2966 die "$::_authors_prog failed with exit code $?\n"
2967 }
2968 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2969 my ($name, $email) = ($1, $2);
2970 $email = undef if length $2 == 0;
2971 return [$name, $email];
2972 } else {
2973 die "Author: $orig_author: $::_authors_prog returned "
2974 . "invalid author format: $author\n";
2975 }
2976 }
2977
2978 sub check_author {
2979 my ($author) = @_;
2980 if (!defined $author || length $author == 0) {
2981 $author = '(no author)';
2982 }
2983 if (!defined $::users{$author}) {
2984 if (defined $::_authors_prog) {
2985 $::users{$author} = call_authors_prog($author);
2986 } elsif (defined $::_authors) {
2987 die "Author: $author not defined in $::_authors file\n";
2988 }
2989 }
2990 $author;
2991 }
2992
2993 sub find_extra_svk_parents {
2994 my ($self, $ed, $tickets, $parents) = @_;
2995 # aha! svk:merge property changed...
2996 my @tickets = split "\n", $tickets;
2997 my @known_parents;
2998 for my $ticket ( @tickets ) {
2999 my ($uuid, $path, $rev) = split /:/, $ticket;
3000 if ( $uuid eq $self->ra_uuid ) {
3001 my $url = $self->{url};
3002 my $repos_root = $url;
3003 my $branch_from = $path;
3004 $branch_from =~ s{^/}{};
3005 my $gs = $self->other_gs($repos_root."/".$branch_from,
3006 $url,
3007 $branch_from,
3008 $rev,
3009 $self->{ref_id});
3010 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3011 # wahey! we found it, but it might be
3012 # an old one (!)
3013 push @known_parents, [ $rev, $commit ];
3014 }
3015 }
3016 }
3017 # Ordering matters; highest-numbered commit merge tickets
3018 # first, as they may account for later merge ticket additions
3019 # or changes.
3020 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3021 for my $parent ( @known_parents ) {
3022 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3023 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3024 my $new;
3025 while ( <$msg_fh> ) {
3026 $new=1;last;
3027 }
3028 command_close_pipe($msg_fh, $ctx);
3029 if ( $new ) {
3030 print STDERR
3031 "Found merge parent (svk:merge ticket): $parent\n";
3032 push @$parents, $parent;
3033 }
3034 }
3035 }
3036
3037 sub lookup_svn_merge {
3038 my $uuid = shift;
3039 my $url = shift;
3040 my $merge = shift;
3041
3042 my ($source, $revs) = split ":", $merge;
3043 my $path = $source;
3044 $path =~ s{^/}{};
3045 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3046 if ( !$gs ) {
3047 warn "Couldn't find revmap for $url$source\n";
3048 return;
3049 }
3050 my @ranges = split ",", $revs;
3051 my ($tip, $tip_commit);
3052 my @merged_commit_ranges;
3053 # find the tip
3054 for my $range ( @ranges ) {
3055 my ($bottom, $top) = split "-", $range;
3056 $top ||= $bottom;
3057 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3058 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3059
3060 unless ($top_commit and $bottom_commit) {
3061 warn "W:unknown path/rev in svn:mergeinfo "
3062 ."dirprop: $source:$range\n";
3063 next;
3064 }
3065
3066 push @merged_commit_ranges,
3067 "$bottom_commit^..$top_commit";
3068
3069 if ( !defined $tip or $top > $tip ) {
3070 $tip = $top;
3071 $tip_commit = $top_commit;
3072 }
3073 }
3074 return ($tip_commit, @merged_commit_ranges);
3075 }
3076
3077 sub _rev_list {
3078 my ($msg_fh, $ctx) = command_output_pipe(
3079 "rev-list", @_,
3080 );
3081 my @rv;
3082 while ( <$msg_fh> ) {
3083 chomp;
3084 push @rv, $_;
3085 }
3086 command_close_pipe($msg_fh, $ctx);
3087 @rv;
3088 }
3089
3090 sub check_cherry_pick {
3091 my $base = shift;
3092 my $tip = shift;
3093 my @ranges = @_;
3094 my %commits = map { $_ => 1 }
3095 _rev_list("--no-merges", $tip, "--not", $base);
3096 for my $range ( @ranges ) {
3097 delete @commits{_rev_list($range)};
3098 }
3099 for my $commit (keys %commits) {
3100 if (has_no_changes($commit)) {
3101 delete $commits{$commit};
3102 }
3103 }
3104 return (keys %commits);
3105 }
3106
3107 sub has_no_changes {
3108 my $commit = shift;
3109
3110 my @revs = split / /, command_oneline(
3111 qw(rev-list --parents -1 -m), $commit);
3112
3113 # Commits with no parents, e.g. the start of a partial branch,
3114 # have changes by definition.
3115 return 1 if (@revs < 2);
3116
3117 # Commits with multiple parents, e.g a merge, have no changes
3118 # by definition.
3119 return 0 if (@revs > 2);
3120
3121 return (command_oneline("rev-parse", "$commit^{tree}") eq
3122 command_oneline("rev-parse", "$commit~1^{tree}"));
3123 }
3124
3125 # The GIT_DIR environment variable is not always set until after the command
3126 # line arguments are processed, so we can't memoize in a BEGIN block.
3127 {
3128 my $memoized = 0;
3129
3130 sub memoize_svn_mergeinfo_functions {
3131 return if $memoized;
3132 $memoized = 1;
3133
3134 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3135 mkpath([$cache_path]) unless -d $cache_path;
3136
3137 tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3138 "$cache_path/lookup_svn_merge.db", 'nstore';
3139 memoize 'lookup_svn_merge',
3140 SCALAR_CACHE => 'FAULT',
3141 LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3142 ;
3143
3144 tie my %check_cherry_pick_cache => 'Memoize::Storable',
3145 "$cache_path/check_cherry_pick.db", 'nstore';
3146 memoize 'check_cherry_pick',
3147 SCALAR_CACHE => 'FAULT',
3148 LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3149 ;
3150
3151 tie my %has_no_changes_cache => 'Memoize::Storable',
3152 "$cache_path/has_no_changes.db", 'nstore';
3153 memoize 'has_no_changes',
3154 SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3155 LIST_CACHE => 'FAULT',
3156 ;
3157 }
3158 }
3159
3160 sub parents_exclude {
3161 my $parents = shift;
3162 my @commits = @_;
3163 return unless @commits;
3164
3165 my @excluded;
3166 my $excluded;
3167 do {
3168 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3169 $excluded = command_oneline(@cmd);
3170 if ( $excluded ) {
3171 my @new;
3172 my $found;
3173 for my $commit ( @commits ) {
3174 if ( $commit eq $excluded ) {
3175 push @excluded, $commit;
3176 $found++;
3177 last;
3178 }
3179 else {
3180 push @new, $commit;
3181 }
3182 }
3183 die "saw commit '$excluded' in rev-list output, "
3184 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3185 unless $found;
3186 @commits = @new;
3187 }
3188 }
3189 while ($excluded and @commits);
3190
3191 return @excluded;
3192 }
3193
3194
3195 # note: this function should only be called if the various dirprops
3196 # have actually changed
3197 sub find_extra_svn_parents {
3198 my ($self, $ed, $mergeinfo, $parents) = @_;
3199 # aha! svk:merge property changed...
3200
3201 memoize_svn_mergeinfo_functions();
3202
3203 # We first search for merged tips which are not in our
3204 # history. Then, we figure out which git revisions are in
3205 # that tip, but not this revision. If all of those revisions
3206 # are now marked as merge, we can add the tip as a parent.
3207 my @merges = split "\n", $mergeinfo;
3208 my @merge_tips;
3209 my $url = $self->{url};
3210 my $uuid = $self->ra_uuid;
3211 my %ranges;
3212 for my $merge ( @merges ) {
3213 my ($tip_commit, @ranges) =
3214 lookup_svn_merge( $uuid, $url, $merge );
3215 unless (!$tip_commit or
3216 grep { $_ eq $tip_commit } @$parents ) {
3217 push @merge_tips, $tip_commit;
3218 $ranges{$tip_commit} = \@ranges;
3219 } else {
3220 push @merge_tips, undef;
3221 }
3222 }
3223
3224 my %excluded = map { $_ => 1 }
3225 parents_exclude($parents, grep { defined } @merge_tips);
3226
3227 # check merge tips for new parents
3228 my @new_parents;
3229 for my $merge_tip ( @merge_tips ) {
3230 my $spec = shift @merges;
3231 next unless $merge_tip and $excluded{$merge_tip};
3232
3233 my $ranges = $ranges{$merge_tip};
3234
3235 # check out 'new' tips
3236 my $merge_base;
3237 eval {
3238 $merge_base = command_oneline(
3239 "merge-base",
3240 @$parents, $merge_tip,
3241 );
3242 };
3243 if ($@) {
3244 die "An error occurred during merge-base"
3245 unless $@->isa("Git::Error::Command");
3246
3247 warn "W: Cannot find common ancestor between ".
3248 "@$parents and $merge_tip. Ignoring merge info.\n";
3249 next;
3250 }
3251
3252 # double check that there are no missing non-merge commits
3253 my (@incomplete) = check_cherry_pick(
3254 $merge_base, $merge_tip,
3255 @$ranges,
3256 );
3257
3258 if ( @incomplete ) {
3259 warn "W:svn cherry-pick ignored ($spec) - missing "
3260 .@incomplete." commit(s) (eg $incomplete[0])\n";
3261 } else {
3262 warn
3263 "Found merge parent (svn:mergeinfo prop): ",
3264 $merge_tip, "\n";
3265 push @new_parents, $merge_tip;
3266 }
3267 }
3268
3269 # cater for merges which merge commits from multiple branches
3270 if ( @new_parents > 1 ) {
3271 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3272 for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3273 next if $i == $j;
3274 next unless $new_parents[$i];
3275 next unless $new_parents[$j];
3276 my $revs = command_oneline(
3277 "rev-list", "-1",
3278 "$new_parents[$i]..$new_parents[$j]",
3279 );
3280 if ( !$revs ) {
3281 undef($new_parents[$j]);
3282 }
3283 }
3284 }
3285 }
3286 push @$parents, grep { defined } @new_parents;
3287 }
3288
3289 sub make_log_entry {
3290 my ($self, $rev, $parents, $ed) = @_;
3291 my $untracked = $self->get_untracked($ed);
3292
3293 my @parents = @$parents;
3294 my $ps = $ed->{path_strip} || "";
3295 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3296 my $props = $ed->{dir_prop}{$path};
3297 if ( $props->{"svk:merge"} ) {
3298 $self->find_extra_svk_parents
3299 ($ed, $props->{"svk:merge"}, \@parents);
3300 }
3301 if ( $props->{"svn:mergeinfo"} ) {
3302 $self->find_extra_svn_parents
3303 ($ed,
3304 $props->{"svn:mergeinfo"},
3305 \@parents);
3306 }
3307 }
3308
3309 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3310 print $un "r$rev\n" or croak $!;
3311 print $un $_, "\n" foreach @$untracked;
3312 my %log_entry = ( parents => \@parents, revision => $rev,
3313 log => '');
3314
3315 my $headrev;
3316 my $logged = delete $self->{logged_rev_props};
3317 if (!$logged || $self->{-want_revprops}) {
3318 my $rp = $self->ra->rev_proplist($rev);
3319 foreach (sort keys %$rp) {
3320 my $v = $rp->{$_};
3321 if (/^svn:(author|date|log)$/) {
3322 $log_entry{$1} = $v;
3323 } elsif ($_ eq 'svm:headrev') {
3324 $headrev = $v;
3325 } else {
3326 print $un " rev_prop: ", uri_encode($_), ' ',
3327 uri_encode($v), "\n";
3328 }
3329 }
3330 } else {
3331 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3332 }
3333 close $un or croak $!;
3334
3335 $log_entry{date} = parse_svn_date($log_entry{date});
3336 $log_entry{log} .= "\n";
3337 my $author = $log_entry{author} = check_author($log_entry{author});
3338 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3339 : ($author, undef);
3340
3341 my ($commit_name, $commit_email) = ($name, $email);
3342 if ($_use_log_author) {
3343 my $name_field;
3344 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3345 $name_field = $1;
3346 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3347 $name_field = $1;
3348 }
3349 if (!defined $name_field) {
3350 if (!defined $email) {
3351 $email = $name;
3352 }
3353 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3354 ($name, $email) = ($1, $2);
3355 } elsif ($name_field =~ /(.*)@/) {
3356 ($name, $email) = ($1, $name_field);
3357 } else {
3358 ($name, $email) = ($name_field, $name_field);
3359 }
3360 }
3361 if (defined $headrev && $self->use_svm_props) {
3362 if ($self->rewrite_root) {
3363 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3364 "options set!\n";
3365 }
3366 if ($self->rewrite_uuid) {
3367 die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3368 "options set!\n";
3369 }
3370 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3371 # we don't want "SVM: initializing mirror for junk" ...
3372 return undef if $r == 0;
3373 my $svm = $self->svm;
3374 if ($uuid ne $svm->{uuid}) {
3375 die "UUID mismatch on SVM path:\n",
3376 "expected: $svm->{uuid}\n",
3377 " got: $uuid\n";
3378 }
3379 my $full_url = $self->full_url;
3380 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3381 die "Failed to replace '$svm->{replace}' with ",
3382 "'$svm->{source}' in $full_url\n";
3383 # throw away username for storing in records
3384 remove_username($full_url);
3385 $log_entry{metadata} = "$full_url\@$r $uuid";
3386 $log_entry{svm_revision} = $r;
3387 $email ||= "$author\@$uuid";
3388 $commit_email ||= "$author\@$uuid";
3389 } elsif ($self->use_svnsync_props) {
3390 my $full_url = $self->svnsync->{url};
3391 $full_url .= "/$self->{path}" if length $self->{path};
3392 remove_username($full_url);
3393 my $uuid = $self->svnsync->{uuid};
3394 $log_entry{metadata} = "$full_url\@$rev $uuid";
3395 $email ||= "$author\@$uuid";
3396 $commit_email ||= "$author\@$uuid";
3397 } else {
3398 my $url = $self->metadata_url;
3399 remove_username($url);
3400 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3401 $log_entry{metadata} = "$url\@$rev " . $uuid;
3402 $email ||= "$author\@" . $uuid;
3403 $commit_email ||= "$author\@" . $uuid;
3404 }
3405 $log_entry{name} = $name;
3406 $log_entry{email} = $email;
3407 $log_entry{commit_name} = $commit_name;
3408 $log_entry{commit_email} = $commit_email;
3409 \%log_entry;
3410 }
3411
3412 sub fetch {
3413 my ($self, $min_rev, $max_rev, @parents) = @_;
3414 my ($last_rev, $last_commit) = $self->last_rev_commit;
3415 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3416 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3417 }
3418
3419 sub set_tree_cb {
3420 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3421 $self->{inject_parents} = { $rev => $tree };
3422 $self->fetch(undef, undef);
3423 }
3424
3425 sub set_tree {
3426 my ($self, $tree) = (shift, shift);
3427 my $log_entry = ::get_commit_entry($tree);
3428 unless ($self->{last_rev}) {
3429 ::fatal("Must have an existing revision to commit");
3430 }
3431 my %ed_opts = ( r => $self->{last_rev},
3432 log => $log_entry->{log},
3433 ra => $self->ra,
3434 tree_a => $self->{last_commit},
3435 tree_b => $tree,
3436 editor_cb => sub {
3437 $self->set_tree_cb($log_entry, $tree, @_) },
3438 svn_path => $self->{path} );
3439 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3440 print "No changes\nr$self->{last_rev} = $tree\n";
3441 }
3442 }
3443
3444 sub rebuild_from_rev_db {
3445 my ($self, $path) = @_;
3446 my $r = -1;
3447 open my $fh, '<', $path or croak "open: $!";
3448 binmode $fh or croak "binmode: $!";
3449 while (<$fh>) {
3450 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3451 chomp($_);
3452 ++$r;
3453 next if $_ eq ('0' x 40);
3454 $self->rev_map_set($r, $_);
3455 print "r$r = $_\n";
3456 }
3457 close $fh or croak "close: $!";
3458 unlink $path or croak "unlink: $!";
3459 }
3460
3461 sub rebuild {
3462 my ($self) = @_;
3463 my $map_path = $self->map_path;
3464 my $partial = (-e $map_path && ! -z $map_path);
3465 return unless ::verify_ref($self->refname.'^0');
3466 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3467 my $rev_db = $self->rev_db_path;
3468 $self->rebuild_from_rev_db($rev_db);
3469 if ($self->use_svm_props) {
3470 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3471 $self->rebuild_from_rev_db($svm_rev_db);
3472 }
3473 $self->unlink_rev_db_symlink;
3474 return;
3475 }
3476 print "Rebuilding $map_path ...\n" if (!$partial);
3477 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3478 (undef, undef));
3479 my ($log, $ctx) =
3480 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3481 ($head ? "$head.." : "") . $self->refname,
3482 '--');
3483 my $metadata_url = $self->metadata_url;
3484 remove_username($metadata_url);
3485 my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3486 my $c;
3487 while (<$log>) {
3488 if ( m{^commit ($::sha1)$} ) {
3489 $c = $1;
3490 next;
3491 }
3492 next unless s{^\s*(git-svn-id:)}{$1};
3493 my ($url, $rev, $uuid) = ::extract_metadata($_);
3494 remove_username($url);
3495
3496 # ignore merges (from set-tree)
3497 next if (!defined $rev || !$uuid);
3498
3499 # if we merged or otherwise started elsewhere, this is
3500 # how we break out of it
3501 if (($uuid ne $svn_uuid) ||
3502 ($metadata_url && $url && ($url ne $metadata_url))) {
3503 next;
3504 }
3505 if ($partial && $head) {
3506 print "Partial-rebuilding $map_path ...\n";
3507 print "Currently at $base_rev = $head\n";
3508 $head = undef;
3509 }
3510
3511 $self->rev_map_set($rev, $c);
3512 print "r$rev = $c\n";
3513 }
3514 command_close_pipe($log, $ctx);
3515 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3516 my $rev_db_path = $self->rev_db_path;
3517 if (-f $self->rev_db_path) {
3518 unlink $self->rev_db_path or croak "unlink: $!";
3519 }
3520 $self->unlink_rev_db_symlink;
3521 }
3522
3523 # rev_map:
3524 # Tie::File seems to be prone to offset errors if revisions get sparse,
3525 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
3526 # one of my favorite modules is out :< Next up would be one of the DBM
3527 # modules, but I'm not sure which is most portable...
3528 #
3529 # This is the replacement for the rev_db format, which was too big
3530 # and inefficient for large repositories with a lot of sparse history
3531 # (mainly tags)
3532 #
3533 # The format is this:
3534 # - 24 bytes for every record,
3535 # * 4 bytes for the integer representing an SVN revision number
3536 # * 20 bytes representing the sha1 of a git commit
3537 # - No empty padding records like the old format
3538 # (except the last record, which can be overwritten)
3539 # - new records are written append-only since SVN revision numbers
3540 # increase monotonically
3541 # - lookups on SVN revision number are done via a binary search
3542 # - Piping the file to xxd -c24 is a good way of dumping it for
3543 # viewing or editing (piped back through xxd -r), should the need
3544 # ever arise.
3545 # - The last record can be padding revision with an all-zero sha1
3546 # This is used to optimize fetch performance when using multiple
3547 # "fetch" directives in .git/config
3548 #
3549 # These files are disposable unless noMetadata or useSvmProps is set
3550
3551 sub _rev_map_set {
3552 my ($fh, $rev, $commit) = @_;
3553
3554 binmode $fh or croak "binmode: $!";
3555 my $size = (stat($fh))[7];
3556 ($size % 24) == 0 or croak "inconsistent size: $size";
3557
3558 my $wr_offset = 0;
3559 if ($size > 0) {
3560 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3561 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3562 $read == 24 or croak "read only $read bytes (!= 24)";
3563 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3564 if ($last_commit eq ('0' x40)) {
3565 if ($size >= 48) {
3566 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3567 $read = sysread($fh, $buf, 24) or
3568 croak "read: $!";
3569 $read == 24 or
3570 croak "read only $read bytes (!= 24)";
3571 ($last_rev, $last_commit) =
3572 unpack(rev_map_fmt, $buf);
3573 if ($last_commit eq ('0' x40)) {
3574 croak "inconsistent .rev_map\n";
3575 }
3576 }
3577 if ($last_rev >= $rev) {
3578 croak "last_rev is higher!: $last_rev >= $rev";
3579 }
3580 $wr_offset = -24;
3581 }
3582 }
3583 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3584 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3585 croak "write: $!";
3586 }
3587
3588 sub _rev_map_reset {
3589 my ($fh, $rev, $commit) = @_;
3590 my $c = _rev_map_get($fh, $rev);
3591 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3592 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3593 truncate $fh, $offset or croak "truncate: $!";
3594 }
3595
3596 sub mkfile {
3597 my ($path) = @_;
3598 unless (-e $path) {
3599 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3600 mkpath([$dir]) unless -d $dir;
3601 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3602 close $fh or die "Couldn't close (create) $path: $!\n";
3603 }
3604 }
3605
3606 sub rev_map_set {
3607 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3608 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3609 my $db = $self->map_path($uuid);
3610 my $db_lock = "$db.lock";
3611 my $sig;
3612 $update_ref ||= 0;
3613 if ($update_ref) {
3614 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3615 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3616 }
3617 mkfile($db);
3618
3619 $LOCKFILES{$db_lock} = 1;
3620 my $sync;
3621 # both of these options make our .rev_db file very, very important
3622 # and we can't afford to lose it because rebuild() won't work
3623 if ($self->use_svm_props || $self->no_metadata) {
3624 $sync = 1;
3625 copy($db, $db_lock) or die "rev_map_set(@_): ",
3626 "Failed to copy: ",
3627 "$db => $db_lock ($!)\n";
3628 } else {
3629 rename $db, $db_lock or die "rev_map_set(@_): ",
3630 "Failed to rename: ",
3631 "$db => $db_lock ($!)\n";
3632 }
3633
3634 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3635 or croak "Couldn't open $db_lock: $!\n";
3636 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3637 _rev_map_set($fh, $rev, $commit);
3638 if ($sync) {
3639 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3640 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3641 }
3642 close $fh or croak $!;
3643 if ($update_ref) {
3644 $_head = $self;
3645 my $note = "";
3646 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3647 command_noisy('update-ref', '-m', "r$rev$note",
3648 $self->refname, $commit);
3649 }
3650 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3651 "$db_lock => $db ($!)\n";
3652 delete $LOCKFILES{$db_lock};
3653 if ($update_ref) {
3654 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3655 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3656 kill $sig, $$ if defined $sig;
3657 }
3658 }
3659
3660 # If want_commit, this will return an array of (rev, commit) where
3661 # commit _must_ be a valid commit in the archive.
3662 # Otherwise, it'll return the max revision (whether or not the
3663 # commit is valid or just a 0x40 placeholder).
3664 sub rev_map_max {
3665 my ($self, $want_commit) = @_;
3666 $self->rebuild;
3667 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3668 $want_commit ? ($r, $c) : $r;
3669 }
3670
3671 sub rev_map_max_norebuild {
3672 my ($self, $want_commit) = @_;
3673 my $map_path = $self->map_path;
3674 stat $map_path or return $want_commit ? (0, undef) : 0;
3675 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3676 binmode $fh or croak "binmode: $!";
3677 my $size = (stat($fh))[7];
3678 ($size % 24) == 0 or croak "inconsistent size: $size";
3679
3680 if ($size == 0) {
3681 close $fh or croak "close: $!";
3682 return $want_commit ? (0, undef) : 0;
3683 }
3684
3685 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3686 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3687 my ($r, $c) = unpack(rev_map_fmt, $buf);
3688 if ($want_commit && $c eq ('0' x40)) {
3689 if ($size < 48) {
3690 return $want_commit ? (0, undef) : 0;
3691 }
3692 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3693 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3694 ($r, $c) = unpack(rev_map_fmt, $buf);
3695 if ($c eq ('0'x40)) {
3696 croak "Penultimate record is all-zeroes in $map_path";
3697 }
3698 }
3699 close $fh or croak "close: $!";
3700 $want_commit ? ($r, $c) : $r;
3701 }
3702
3703 sub rev_map_get {
3704 my ($self, $rev, $uuid) = @_;
3705 my $map_path = $self->map_path($uuid);
3706 return undef unless -e $map_path;
3707
3708 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3709 my $c = _rev_map_get($fh, $rev);
3710 close($fh) or croak "close: $!";
3711 $c
3712 }
3713
3714 sub _rev_map_get {
3715 my ($fh, $rev) = @_;
3716
3717 binmode $fh or croak "binmode: $!";
3718 my $size = (stat($fh))[7];
3719 ($size % 24) == 0 or croak "inconsistent size: $size";
3720
3721 if ($size == 0) {
3722 return undef;
3723 }
3724
3725 my ($l, $u) = (0, $size - 24);
3726 my ($r, $c, $buf);
3727
3728 while ($l <= $u) {
3729 my $i = int(($l/24 + $u/24) / 2) * 24;
3730 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3731 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3732 my ($r, $c) = unpack(rev_map_fmt, $buf);
3733
3734 if ($r < $rev) {
3735 $l = $i + 24;
3736 } elsif ($r > $rev) {
3737 $u = $i - 24;
3738 } else { # $r == $rev
3739 return $c eq ('0' x 40) ? undef : $c;
3740 }
3741 }
3742 undef;
3743 }
3744
3745 # Finds the first svn revision that exists on (if $eq_ok is true) or
3746 # before $rev for the current branch. It will not search any lower
3747 # than $min_rev. Returns the git commit hash and svn revision number
3748 # if found, else (undef, undef).
3749 sub find_rev_before {
3750 my ($self, $rev, $eq_ok, $min_rev) = @_;
3751 --$rev unless $eq_ok;
3752 $min_rev ||= 1;
3753 my $max_rev = $self->rev_map_max;
3754 $rev = $max_rev if ($rev > $max_rev);
3755 while ($rev >= $min_rev) {
3756 if (my $c = $self->rev_map_get($rev)) {
3757 return ($rev, $c);
3758 }
3759 --$rev;
3760 }
3761 return (undef, undef);
3762 }
3763
3764 # Finds the first svn revision that exists on (if $eq_ok is true) or
3765 # after $rev for the current branch. It will not search any higher
3766 # than $max_rev. Returns the git commit hash and svn revision number
3767 # if found, else (undef, undef).
3768 sub find_rev_after {
3769 my ($self, $rev, $eq_ok, $max_rev) = @_;
3770 ++$rev unless $eq_ok;
3771 $max_rev ||= $self->rev_map_max;
3772 while ($rev <= $max_rev) {
3773 if (my $c = $self->rev_map_get($rev)) {
3774 return ($rev, $c);
3775 }
3776 ++$rev;
3777 }
3778 return (undef, undef);
3779 }
3780
3781 sub _new {
3782 my ($class, $repo_id, $ref_id, $path) = @_;
3783 unless (defined $repo_id && length $repo_id) {
3784 $repo_id = $Git::SVN::default_repo_id;
3785 }
3786 unless (defined $ref_id && length $ref_id) {
3787 $_prefix = '' unless defined($_prefix);
3788 $_[2] = $ref_id =
3789 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3790 }
3791 $_[1] = $repo_id;
3792 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3793
3794 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3795 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3796 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3797 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3798 if (-d $old_dir && ! -d $dir) {
3799 $dir = $old_dir;
3800 }
3801 }
3802
3803 $_[3] = $path = '' unless (defined $path);
3804 mkpath([$dir]);
3805 bless {
3806 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3807 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3808 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3809 }
3810
3811 # for read-only access of old .rev_db formats
3812 sub unlink_rev_db_symlink {
3813 my ($self) = @_;
3814 my $link = $self->rev_db_path;
3815 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3816 if (-l $link) {
3817 unlink $link or croak "unlink: $link failed!";
3818 }
3819 }
3820
3821 sub rev_db_path {
3822 my ($self, $uuid) = @_;
3823 my $db_path = $self->map_path($uuid);
3824 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3825 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3826 $db_path;
3827 }
3828
3829 # the new replacement for .rev_db
3830 sub map_path {
3831 my ($self, $uuid) = @_;
3832 $uuid ||= $self->ra_uuid;
3833 "$self->{map_root}.$uuid";
3834 }
3835
3836 sub uri_encode {
3837 my ($f) = @_;
3838 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3839 $f
3840 }
3841
3842 sub uri_decode {
3843 my ($f) = @_;
3844 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3845 $f
3846 }
3847
3848 sub remove_username {
3849 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3850 }
3851
3852 package Git::SVN::Prompt;
3853 use strict;
3854 use warnings;
3855 require SVN::Core;
3856 use vars qw/$_no_auth_cache $_username/;
3857
3858 sub simple {
3859 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3860 $may_save = undef if $_no_auth_cache;
3861 $default_username = $_username if defined $_username;
3862 if (defined $default_username && length $default_username) {
3863 if (defined $realm && length $realm) {
3864 print STDERR "Authentication realm: $realm\n";
3865 STDERR->flush;
3866 }
3867 $cred->username($default_username);
3868 } else {
3869 username($cred, $realm, $may_save, $pool);
3870 }
3871 $cred->password(_read_password("Password for '" .
3872 $cred->username . "': ", $realm));
3873 $cred->may_save($may_save);
3874 $SVN::_Core::SVN_NO_ERROR;
3875 }
3876
3877 sub ssl_server_trust {
3878 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3879 $may_save = undef if $_no_auth_cache;
3880 print STDERR "Error validating server certificate for '$realm':\n";
3881 {
3882 no warnings 'once';
3883 # All variables SVN::Auth::SSL::* are used only once,
3884 # so we're shutting up Perl warnings about this.
3885 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3886 print STDERR " - The certificate is not issued ",
3887 "by a trusted authority. Use the\n",
3888 " fingerprint to validate ",
3889 "the certificate manually!\n";
3890 }
3891 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3892 print STDERR " - The certificate hostname ",
3893 "does not match.\n";
3894 }
3895 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3896 print STDERR " - The certificate is not yet valid.\n";
3897 }
3898 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3899 print STDERR " - The certificate has expired.\n";
3900 }
3901 if ($failures & $SVN::Auth::SSL::OTHER) {
3902 print STDERR " - The certificate has ",
3903 "an unknown error.\n";
3904 }
3905 } # no warnings 'once'
3906 printf STDERR
3907 "Certificate information:\n".
3908 " - Hostname: %s\n".
3909 " - Valid: from %s until %s\n".
3910 " - Issuer: %s\n".
3911 " - Fingerprint: %s\n",
3912 map $cert_info->$_, qw(hostname valid_from valid_until
3913 issuer_dname fingerprint);
3914 my $choice;
3915 prompt:
3916 print STDERR $may_save ?
3917 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3918 "(R)eject or accept (t)emporarily? ";
3919 STDERR->flush;
3920 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3921 if ($choice =~ /^t$/i) {
3922 $cred->may_save(undef);
3923 } elsif ($choice =~ /^r$/i) {
3924 return -1;
3925 } elsif ($may_save && $choice =~ /^p$/i) {
3926 $cred->may_save($may_save);
3927 } else {
3928 goto prompt;
3929 }
3930 $cred->accepted_failures($failures);
3931 $SVN::_Core::SVN_NO_ERROR;
3932 }
3933
3934 sub ssl_client_cert {
3935 my ($cred, $realm, $may_save, $pool) = @_;
3936 $may_save = undef if $_no_auth_cache;
3937 print STDERR "Client certificate filename: ";
3938 STDERR->flush;
3939 chomp(my $filename = <STDIN>);
3940 $cred->cert_file($filename);
3941 $cred->may_save($may_save);
3942 $SVN::_Core::SVN_NO_ERROR;
3943 }
3944
3945 sub ssl_client_cert_pw {
3946 my ($cred, $realm, $may_save, $pool) = @_;
3947 $may_save = undef if $_no_auth_cache;
3948 $cred->password(_read_password("Password: ", $realm));
3949 $cred->may_save($may_save);
3950 $SVN::_Core::SVN_NO_ERROR;
3951 }
3952
3953 sub username {
3954 my ($cred, $realm, $may_save, $pool) = @_;
3955 $may_save = undef if $_no_auth_cache;
3956 if (defined $realm && length $realm) {
3957 print STDERR "Authentication realm: $realm\n";
3958 }
3959 my $username;
3960 if (defined $_username) {
3961 $username = $_username;
3962 } else {
3963 print STDERR "Username: ";
3964 STDERR->flush;
3965 chomp($username = <STDIN>);
3966 }
3967 $cred->username($username);
3968 $cred->may_save($may_save);
3969 $SVN::_Core::SVN_NO_ERROR;
3970 }
3971
3972 sub _read_password {
3973 my ($prompt, $realm) = @_;
3974 my $password = '';
3975 if (exists $ENV{GIT_ASKPASS}) {
3976 open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
3977 $password = <PH>;
3978 $password =~ s/[\012\015]//; # \n\r
3979 close(PH);
3980 } else {
3981 print STDERR $prompt;
3982 STDERR->flush;
3983 require Term::ReadKey;
3984 Term::ReadKey::ReadMode('noecho');
3985 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3986 last if $key =~ /[\012\015]/; # \n\r
3987 $password .= $key;
3988 }
3989 Term::ReadKey::ReadMode('restore');
3990 print STDERR "\n";
3991 STDERR->flush;
3992 }
3993 $password;
3994 }
3995
3996 package SVN::Git::Fetcher;
3997 use vars qw/@ISA/;
3998 use strict;
3999 use warnings;
4000 use Carp qw/croak/;
4001 use File::Temp qw/tempfile/;
4002 use IO::File qw//;
4003 use vars qw/$_ignore_regex/;
4004
4005 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
4006 sub new {
4007 my ($class, $git_svn, $switch_path) = @_;
4008 my $self = SVN::Delta::Editor->new;
4009 bless $self, $class;
4010 if (exists $git_svn->{last_commit}) {
4011 $self->{c} = $git_svn->{last_commit};
4012 $self->{empty_symlinks} =
4013 _mark_empty_symlinks($git_svn, $switch_path);
4014 }
4015 $self->{ignore_regex} = eval { command_oneline('config', '--get',
4016 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
4017 $self->{empty} = {};
4018 $self->{dir_prop} = {};
4019 $self->{file_prop} = {};
4020 $self->{absent_dir} = {};
4021 $self->{absent_file} = {};
4022 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4023 $self;
4024 }
4025
4026 # this uses the Ra object, so it must be called before do_{switch,update},
4027 # not inside them (when the Git::SVN::Fetcher object is passed) to
4028 # do_{switch,update}
4029 sub _mark_empty_symlinks {
4030 my ($git_svn, $switch_path) = @_;
4031 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4032 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4033
4034 my %ret;
4035 my ($rev, $cmt) = $git_svn->last_rev_commit;
4036 return {} unless ($rev && $cmt);
4037
4038 # allow the warning to be printed for each revision we fetch to
4039 # ensure the user sees it. The user can also disable the workaround
4040 # on the repository even while git svn is running and the next
4041 # revision fetched will skip this expensive function.
4042 my $printed_warning;
4043 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4044 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4045 local $/ = "\0";
4046 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4047 $pfx .= '/' if length($pfx);
4048 while (<$ls>) {
4049 chomp;
4050 s/\A100644 blob $empty_blob\t//o or next;
4051 unless ($printed_warning) {
4052 print STDERR "Scanning for empty symlinks, ",
4053 "this may take a while if you have ",
4054 "many empty files\n",
4055 "You may disable this with `",
4056 "git config svn.brokenSymlinkWorkaround ",
4057 "false'.\n",
4058 "This may be done in a different ",
4059 "terminal without restarting ",
4060 "git svn\n";
4061 $printed_warning = 1;
4062 }
4063 my $path = $_;
4064 my (undef, $props) =
4065 $git_svn->ra->get_file($pfx.$path, $rev, undef);
4066 if ($props->{'svn:special'}) {
4067 $ret{$path} = 1;
4068 }
4069 }
4070 command_close_pipe($ls, $ctx);
4071 \%ret;
4072 }
4073
4074 # returns true if a given path is inside a ".git" directory
4075 sub in_dot_git {
4076 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4077 }
4078
4079 # return value: 0 -- don't ignore, 1 -- ignore
4080 sub is_path_ignored {
4081 my ($self, $path) = @_;
4082 return 1 if in_dot_git($path);
4083 return 1 if defined($self->{ignore_regex}) &&
4084 $path =~ m!$self->{ignore_regex}!;
4085 return 0 unless defined($_ignore_regex);
4086 return 1 if $path =~ m!$_ignore_regex!o;
4087 return 0;
4088 }
4089
4090 sub set_path_strip {
4091 my ($self, $path) = @_;
4092 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4093 }
4094
4095 sub open_root {
4096 { path => '' };
4097 }
4098
4099 sub open_directory {
4100 my ($self, $path, $pb, $rev) = @_;
4101 { path => $path };
4102 }
4103
4104 sub git_path {
4105 my ($self, $path) = @_;
4106 if ($self->{path_strip}) {
4107 $path =~ s!$self->{path_strip}!! or
4108 die "Failed to strip path '$path' ($self->{path_strip})\n";
4109 }
4110 $path;
4111 }
4112
4113 sub delete_entry {
4114 my ($self, $path, $rev, $pb) = @_;
4115 return undef if $self->is_path_ignored($path);
4116
4117 my $gpath = $self->git_path($path);
4118 return undef if ($gpath eq '');
4119
4120 # remove entire directories.
4121 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4122 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4123 if ($tree) {
4124 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4125 -r --name-only -z/,
4126 $tree);
4127 local $/ = "\0";
4128 while (<$ls>) {
4129 chomp;
4130 my $rmpath = "$gpath/$_";
4131 $self->{gii}->remove($rmpath);
4132 print "\tD\t$rmpath\n" unless $::_q;
4133 }
4134 print "\tD\t$gpath/\n" unless $::_q;
4135 command_close_pipe($ls, $ctx);
4136 } else {
4137 $self->{gii}->remove($gpath);
4138 print "\tD\t$gpath\n" unless $::_q;
4139 }
4140 $self->{empty}->{$path} = 0;
4141 undef;
4142 }
4143
4144 sub open_file {
4145 my ($self, $path, $pb, $rev) = @_;
4146 my ($mode, $blob);
4147
4148 goto out if $self->is_path_ignored($path);
4149
4150 my $gpath = $self->git_path($path);
4151 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4152 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4153 unless (defined $mode && defined $blob) {
4154 die "$path was not found in commit $self->{c} (r$rev)\n";
4155 }
4156 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4157 $mode = '120000';
4158 }
4159 out:
4160 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4161 pool => SVN::Pool->new, action => 'M' };
4162 }
4163
4164 sub add_file {
4165 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4166 my $mode;
4167
4168 if (!$self->is_path_ignored($path)) {
4169 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4170 delete $self->{empty}->{$dir};
4171 $mode = '100644';
4172 }
4173 { path => $path, mode_a => $mode, mode_b => $mode,
4174 pool => SVN::Pool->new, action => 'A' };
4175 }
4176
4177 sub add_directory {
4178 my ($self, $path, $cp_path, $cp_rev) = @_;
4179 goto out if $self->is_path_ignored($path);
4180 my $gpath = $self->git_path($path);
4181 if ($gpath eq '') {
4182 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4183 -r --name-only -z/,
4184 $self->{c});
4185 local $/ = "\0";
4186 while (<$ls>) {
4187 chomp;
4188 $self->{gii}->remove($_);
4189 print "\tD\t$_\n" unless $::_q;
4190 }
4191 command_close_pipe($ls, $ctx);
4192 $self->{empty}->{$path} = 0;
4193 }
4194 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4195 delete $self->{empty}->{$dir};
4196 $self->{empty}->{$path} = 1;
4197 out:
4198 { path => $path };
4199 }
4200
4201 sub change_dir_prop {
4202 my ($self, $db, $prop, $value) = @_;
4203 return undef if $self->is_path_ignored($db->{path});
4204 $self->{dir_prop}->{$db->{path}} ||= {};
4205 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4206 undef;
4207 }
4208
4209 sub absent_directory {
4210 my ($self, $path, $pb) = @_;
4211 return undef if $self->is_path_ignored($path);
4212 $self->{absent_dir}->{$pb->{path}} ||= [];
4213 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4214 undef;
4215 }
4216
4217 sub absent_file {
4218 my ($self, $path, $pb) = @_;
4219 return undef if $self->is_path_ignored($path);
4220 $self->{absent_file}->{$pb->{path}} ||= [];
4221 push @{$self->{absent_file}->{$pb->{path}}}, $path;
4222 undef;
4223 }
4224
4225 sub change_file_prop {
4226 my ($self, $fb, $prop, $value) = @_;
4227 return undef if $self->is_path_ignored($fb->{path});
4228 if ($prop eq 'svn:executable') {
4229 if ($fb->{mode_b} != 120000) {
4230 $fb->{mode_b} = defined $value ? 100755 : 100644;
4231 }
4232 } elsif ($prop eq 'svn:special') {
4233 $fb->{mode_b} = defined $value ? 120000 : 100644;
4234 } else {
4235 $self->{file_prop}->{$fb->{path}} ||= {};
4236 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4237 }
4238 undef;
4239 }
4240
4241 sub apply_textdelta {
4242 my ($self, $fb, $exp) = @_;
4243 return undef if $self->is_path_ignored($fb->{path});
4244 my $fh = $::_repository->temp_acquire('svn_delta');
4245 # $fh gets auto-closed() by SVN::TxDelta::apply(),
4246 # (but $base does not,) so dup() it for reading in close_file
4247 open my $dup, '<&', $fh or croak $!;
4248 my $base = $::_repository->temp_acquire('git_blob');
4249
4250 if ($fb->{blob}) {
4251 my ($base_is_link, $size);
4252
4253 if ($fb->{mode_a} eq '120000' &&
4254 ! $self->{empty_symlinks}->{$fb->{path}}) {
4255 print $base 'link ' or die "print $!\n";
4256 $base_is_link = 1;
4257 }
4258 retry:
4259 $size = $::_repository->cat_blob($fb->{blob}, $base);
4260 die "Failed to read object $fb->{blob}" if ($size < 0);
4261
4262 if (defined $exp) {
4263 seek $base, 0, 0 or croak $!;
4264 my $got = ::md5sum($base);
4265 if ($got ne $exp) {
4266 my $err = "Checksum mismatch: ".
4267 "$fb->{path} $fb->{blob}\n" .
4268 "expected: $exp\n" .
4269 " got: $got\n";
4270 if ($base_is_link) {
4271 warn $err,
4272 "Retrying... (possibly ",
4273 "a bad symlink from SVN)\n";
4274 $::_repository->temp_reset($base);
4275 $base_is_link = 0;
4276 goto retry;
4277 }
4278 die $err;
4279 }
4280 }
4281 }
4282 seek $base, 0, 0 or croak $!;
4283 $fb->{fh} = $fh;
4284 $fb->{base} = $base;
4285 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4286 }
4287
4288 sub close_file {
4289 my ($self, $fb, $exp) = @_;
4290 return undef if $self->is_path_ignored($fb->{path});
4291
4292 my $hash;
4293 my $path = $self->git_path($fb->{path});
4294 if (my $fh = $fb->{fh}) {
4295 if (defined $exp) {
4296 seek($fh, 0, 0) or croak $!;
4297 my $got = ::md5sum($fh);
4298 if ($got ne $exp) {
4299 die "Checksum mismatch: $path\n",
4300 "expected: $exp\n got: $got\n";
4301 }
4302 }
4303 if ($fb->{mode_b} == 120000) {
4304 sysseek($fh, 0, 0) or croak $!;
4305 my $rd = sysread($fh, my $buf, 5);
4306
4307 if (!defined $rd) {
4308 croak "sysread: $!\n";
4309 } elsif ($rd == 0) {
4310 warn "$path has mode 120000",
4311 " but it points to nothing\n",
4312 "converting to an empty file with mode",
4313 " 100644\n";
4314 $fb->{mode_b} = '100644';
4315 } elsif ($buf ne 'link ') {
4316 warn "$path has mode 120000",
4317 " but is not a link\n";
4318 } else {
4319 my $tmp_fh = $::_repository->temp_acquire(
4320 'svn_hash');
4321 my $res;
4322 while ($res = sysread($fh, my $str, 1024)) {
4323 my $out = syswrite($tmp_fh, $str, $res);
4324 defined($out) && $out == $res
4325 or croak("write ",
4326 Git::temp_path($tmp_fh),
4327 ": $!\n");
4328 }
4329 defined $res or croak $!;
4330
4331 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4332 Git::temp_release($tmp_fh, 1);
4333 }
4334 }
4335
4336 $hash = $::_repository->hash_and_insert_object(
4337 Git::temp_path($fh));
4338 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4339
4340 Git::temp_release($fb->{base}, 1);
4341 Git::temp_release($fh, 1);
4342 } else {
4343 $hash = $fb->{blob} or die "no blob information\n";
4344 }
4345 $fb->{pool}->clear;
4346 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4347 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4348 undef;
4349 }
4350
4351 sub abort_edit {
4352 my $self = shift;
4353 $self->{nr} = $self->{gii}->{nr};
4354 delete $self->{gii};
4355 $self->SUPER::abort_edit(@_);
4356 }
4357
4358 sub close_edit {
4359 my $self = shift;
4360 $self->{git_commit_ok} = 1;
4361 $self->{nr} = $self->{gii}->{nr};
4362 delete $self->{gii};
4363 $self->SUPER::close_edit(@_);
4364 }
4365
4366 package SVN::Git::Editor;
4367 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4368 use strict;
4369 use warnings;
4370 use Carp qw/croak/;
4371 use IO::File;
4372
4373 sub new {
4374 my ($class, $opts) = @_;
4375 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4376 die "$_ required!\n" unless (defined $opts->{$_});
4377 }
4378
4379 my $pool = SVN::Pool->new;
4380 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4381 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4382 $opts->{r}, $mods);
4383
4384 # $opts->{ra} functions should not be used after this:
4385 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4386 $opts->{editor_cb}, $pool);
4387 my $self = SVN::Delta::Editor->new(@ce, $pool);
4388 bless $self, $class;
4389 foreach (qw/svn_path r tree_a tree_b/) {
4390 $self->{$_} = $opts->{$_};
4391 }
4392 $self->{url} = $opts->{ra}->{url};
4393 $self->{mods} = $mods;
4394 $self->{types} = $types;
4395 $self->{pool} = $pool;
4396 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4397 $self->{rm} = { };
4398 $self->{path_prefix} = length $self->{svn_path} ?
4399 "$self->{svn_path}/" : '';
4400 $self->{config} = $opts->{config};
4401 return $self;
4402 }
4403
4404 sub generate_diff {
4405 my ($tree_a, $tree_b) = @_;
4406 my @diff_tree = qw(diff-tree -z -r);
4407 if ($_cp_similarity) {
4408 push @diff_tree, "-C$_cp_similarity";
4409 } else {
4410 push @diff_tree, '-C';
4411 }
4412 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4413 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4414 push @diff_tree, $tree_a, $tree_b;
4415 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4416 local $/ = "\0";
4417 my $state = 'meta';
4418 my @mods;
4419 while (<$diff_fh>) {
4420 chomp $_; # this gets rid of the trailing "\0"
4421 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4422 ($::sha1)\s($::sha1)\s
4423 ([MTCRAD])\d*$/xo) {
4424 push @mods, { mode_a => $1, mode_b => $2,
4425 sha1_a => $3, sha1_b => $4,
4426 chg => $5 };
4427 if ($5 =~ /^(?:C|R)$/) {
4428 $state = 'file_a';
4429 } else {
4430 $state = 'file_b';
4431 }
4432 } elsif ($state eq 'file_a') {
4433 my $x = $mods[$#mods] or croak "Empty array\n";
4434 if ($x->{chg} !~ /^(?:C|R)$/) {
4435 croak "Error parsing $_, $x->{chg}\n";
4436 }
4437 $x->{file_a} = $_;
4438 $state = 'file_b';
4439 } elsif ($state eq 'file_b') {
4440 my $x = $mods[$#mods] or croak "Empty array\n";
4441 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4442 croak "Error parsing $_, $x->{chg}\n";
4443 }
4444 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4445 croak "Error parsing $_, $x->{chg}\n";
4446 }
4447 $x->{file_b} = $_;
4448 $state = 'meta';
4449 } else {
4450 croak "Error parsing $_\n";
4451 }
4452 }
4453 command_close_pipe($diff_fh, $ctx);
4454 \@mods;
4455 }
4456
4457 sub check_diff_paths {
4458 my ($ra, $pfx, $rev, $mods) = @_;
4459 my %types;
4460 $pfx .= '/' if length $pfx;
4461
4462 sub type_diff_paths {
4463 my ($ra, $types, $path, $rev) = @_;
4464 my @p = split m#/+#, $path;
4465 my $c = shift @p;
4466 unless (defined $types->{$c}) {
4467 $types->{$c} = $ra->check_path($c, $rev);
4468 }
4469 while (@p) {
4470 $c .= '/' . shift @p;
4471 next if defined $types->{$c};
4472 $types->{$c} = $ra->check_path($c, $rev);
4473 }
4474 }
4475
4476 foreach my $m (@$mods) {
4477 foreach my $f (qw/file_a file_b/) {
4478 next unless defined $m->{$f};
4479 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4480 if (length $pfx.$dir && ! defined $types{$dir}) {
4481 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4482 }
4483 }
4484 }
4485 \%types;
4486 }
4487
4488 sub split_path {
4489 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4490 }
4491
4492 sub repo_path {
4493 my ($self, $path) = @_;
4494 $self->{path_prefix}.(defined $path ? $path : '');
4495 }
4496
4497 sub url_path {
4498 my ($self, $path) = @_;
4499 if ($self->{url} =~ m#^https?://#) {
4500 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4501 }
4502 $self->{url} . '/' . $self->repo_path($path);
4503 }
4504
4505 sub rmdirs {
4506 my ($self) = @_;
4507 my $rm = $self->{rm};
4508 delete $rm->{''}; # we never delete the url we're tracking
4509 return unless %$rm;
4510
4511 foreach (keys %$rm) {
4512 my @d = split m#/#, $_;
4513 my $c = shift @d;
4514 $rm->{$c} = 1;
4515 while (@d) {
4516 $c .= '/' . shift @d;
4517 $rm->{$c} = 1;
4518 }
4519 }
4520 delete $rm->{$self->{svn_path}};
4521 delete $rm->{''}; # we never delete the url we're tracking
4522 return unless %$rm;
4523
4524 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4525 $self->{tree_b});
4526 local $/ = "\0";
4527 while (<$fh>) {
4528 chomp;
4529 my @dn = split m#/#, $_;
4530 while (pop @dn) {
4531 delete $rm->{join '/', @dn};
4532 }
4533 unless (%$rm) {
4534 close $fh;
4535 return;
4536 }
4537 }
4538 command_close_pipe($fh, $ctx);
4539
4540 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4541 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4542 $self->close_directory($bat->{$d}, $p);
4543 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4544 print "\tD+\t$d/\n" unless $::_q;
4545 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4546 delete $bat->{$d};
4547 }
4548 }
4549
4550 sub open_or_add_dir {
4551 my ($self, $full_path, $baton) = @_;
4552 my $t = $self->{types}->{$full_path};
4553 if (!defined $t) {
4554 die "$full_path not known in r$self->{r} or we have a bug!\n";
4555 }
4556 {
4557 no warnings 'once';
4558 # SVN::Node::none and SVN::Node::file are used only once,
4559 # so we're shutting up Perl's warnings about them.
4560 if ($t == $SVN::Node::none) {
4561 return $self->add_directory($full_path, $baton,
4562 undef, -1, $self->{pool});
4563 } elsif ($t == $SVN::Node::dir) {
4564 return $self->open_directory($full_path, $baton,
4565 $self->{r}, $self->{pool});
4566 } # no warnings 'once'
4567 print STDERR "$full_path already exists in repository at ",
4568 "r$self->{r} and it is not a directory (",
4569 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4570 } # no warnings 'once'
4571 exit 1;
4572 }
4573
4574 sub ensure_path {
4575 my ($self, $path) = @_;
4576 my $bat = $self->{bat};
4577 my $repo_path = $self->repo_path($path);
4578 return $bat->{''} unless (length $repo_path);
4579 my @p = split m#/+#, $repo_path;
4580 my $c = shift @p;
4581 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4582 while (@p) {
4583 my $c0 = $c;
4584 $c .= '/' . shift @p;
4585 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4586 }
4587 return $bat->{$c};
4588 }
4589
4590 # Subroutine to convert a globbing pattern to a regular expression.
4591 # From perl cookbook.
4592 sub glob2pat {
4593 my $globstr = shift;
4594 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4595 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4596 return '^' . $globstr . '$';
4597 }
4598
4599 sub check_autoprop {
4600 my ($self, $pattern, $properties, $file, $fbat) = @_;
4601 # Convert the globbing pattern to a regular expression.
4602 my $regex = glob2pat($pattern);
4603 # Check if the pattern matches the file name.
4604 if($file =~ m/($regex)/) {
4605 # Parse the list of properties to set.
4606 my @props = split(/;/, $properties);
4607 foreach my $prop (@props) {
4608 # Parse 'name=value' syntax and set the property.
4609 if ($prop =~ /([^=]+)=(.*)/) {
4610 my ($n,$v) = ($1,$2);
4611 for ($n, $v) {
4612 s/^\s+//; s/\s+$//;
4613 }
4614 $self->change_file_prop($fbat, $n, $v);
4615 }
4616 }
4617 }
4618 }
4619
4620 sub apply_autoprops {
4621 my ($self, $file, $fbat) = @_;
4622 my $conf_t = ${$self->{config}}{'config'};
4623 no warnings 'once';
4624 # Check [miscellany]/enable-auto-props in svn configuration.
4625 if (SVN::_Core::svn_config_get_bool(
4626 $conf_t,
4627 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4628 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4629 0)) {
4630 # Auto-props are enabled. Enumerate them to look for matches.
4631 my $callback = sub {
4632 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4633 };
4634 SVN::_Core::svn_config_enumerate(
4635 $conf_t,
4636 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4637 $callback);
4638 }
4639 }
4640
4641 sub A {
4642 my ($self, $m) = @_;
4643 my ($dir, $file) = split_path($m->{file_b});
4644 my $pbat = $self->ensure_path($dir);
4645 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4646 undef, -1);
4647 print "\tA\t$m->{file_b}\n" unless $::_q;
4648 $self->apply_autoprops($file, $fbat);
4649 $self->chg_file($fbat, $m);
4650 $self->close_file($fbat,undef,$self->{pool});
4651 }
4652
4653 sub C {
4654 my ($self, $m) = @_;
4655 my ($dir, $file) = split_path($m->{file_b});
4656 my $pbat = $self->ensure_path($dir);
4657 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4658 $self->url_path($m->{file_a}), $self->{r});
4659 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4660 $self->chg_file($fbat, $m);
4661 $self->close_file($fbat,undef,$self->{pool});
4662 }
4663
4664 sub delete_entry {
4665 my ($self, $path, $pbat) = @_;
4666 my $rpath = $self->repo_path($path);
4667 my ($dir, $file) = split_path($rpath);
4668 $self->{rm}->{$dir} = 1;
4669 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4670 }
4671
4672 sub R {
4673 my ($self, $m) = @_;
4674 my ($dir, $file) = split_path($m->{file_b});
4675 my $pbat = $self->ensure_path($dir);
4676 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4677 $self->url_path($m->{file_a}), $self->{r});
4678 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4679 $self->apply_autoprops($file, $fbat);
4680 $self->chg_file($fbat, $m);
4681 $self->close_file($fbat,undef,$self->{pool});
4682
4683 ($dir, $file) = split_path($m->{file_a});
4684 $pbat = $self->ensure_path($dir);
4685 $self->delete_entry($m->{file_a}, $pbat);
4686 }
4687
4688 sub M {
4689 my ($self, $m) = @_;
4690 my ($dir, $file) = split_path($m->{file_b});
4691 my $pbat = $self->ensure_path($dir);
4692 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4693 $pbat,$self->{r},$self->{pool});
4694 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4695 $self->chg_file($fbat, $m);
4696 $self->close_file($fbat,undef,$self->{pool});
4697 }
4698
4699 sub T { shift->M(@_) }
4700
4701 sub change_file_prop {
4702 my ($self, $fbat, $pname, $pval) = @_;
4703 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4704 }
4705
4706 sub _chg_file_get_blob ($$$$) {
4707 my ($self, $fbat, $m, $which) = @_;
4708 my $fh = $::_repository->temp_acquire("git_blob_$which");
4709 if ($m->{"mode_$which"} =~ /^120/) {
4710 print $fh 'link ' or croak $!;
4711 $self->change_file_prop($fbat,'svn:special','*');
4712 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4713 $self->change_file_prop($fbat,'svn:special',undef);
4714 }
4715 my $blob = $m->{"sha1_$which"};
4716 return ($fh,) if ($blob =~ /^0{40}$/);
4717 my $size = $::_repository->cat_blob($blob, $fh);
4718 croak "Failed to read object $blob" if ($size < 0);
4719 $fh->flush == 0 or croak $!;
4720 seek $fh, 0, 0 or croak $!;
4721
4722 my $exp = ::md5sum($fh);
4723 seek $fh, 0, 0 or croak $!;
4724 return ($fh, $exp);
4725 }
4726
4727 sub chg_file {
4728 my ($self, $fbat, $m) = @_;
4729 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4730 $self->change_file_prop($fbat,'svn:executable','*');
4731 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4732 $self->change_file_prop($fbat,'svn:executable',undef);
4733 }
4734 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4735 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4736 my $pool = SVN::Pool->new;
4737 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4738 if (-s $fh_a) {
4739 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4740 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4741 if (defined $res) {
4742 die "Unexpected result from send_txstream: $res\n",
4743 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4744 }
4745 } else {
4746 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4747 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4748 if ($got ne $exp_b);
4749 }
4750 Git::temp_release($fh_b, 1);
4751 Git::temp_release($fh_a, 1);
4752 $pool->clear;
4753 }
4754
4755 sub D {
4756 my ($self, $m) = @_;
4757 my ($dir, $file) = split_path($m->{file_b});
4758 my $pbat = $self->ensure_path($dir);
4759 print "\tD\t$m->{file_b}\n" unless $::_q;
4760 $self->delete_entry($m->{file_b}, $pbat);
4761 }
4762
4763 sub close_edit {
4764 my ($self) = @_;
4765 my ($p,$bat) = ($self->{pool}, $self->{bat});
4766 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4767 next if $_ eq '';
4768 $self->close_directory($bat->{$_}, $p);
4769 }
4770 $self->close_directory($bat->{''}, $p);
4771 $self->SUPER::close_edit($p);
4772 $p->clear;
4773 }
4774
4775 sub abort_edit {
4776 my ($self) = @_;
4777 $self->SUPER::abort_edit($self->{pool});
4778 }
4779
4780 sub DESTROY {
4781 my $self = shift;
4782 $self->SUPER::DESTROY(@_);
4783 $self->{pool}->clear;
4784 }
4785
4786 # this drives the editor
4787 sub apply_diff {
4788 my ($self) = @_;
4789 my $mods = $self->{mods};
4790 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4791 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4792 my $f = $m->{chg};
4793 if (defined $o{$f}) {
4794 $self->$f($m);
4795 } else {
4796 fatal("Invalid change type: $f");
4797 }
4798 }
4799 $self->rmdirs if $_rmdir;
4800 if (@$mods == 0) {
4801 $self->abort_edit;
4802 } else {
4803 $self->close_edit;
4804 }
4805 return scalar @$mods;
4806 }
4807
4808 package Git::SVN::Ra;
4809 use vars qw/@ISA $config_dir $_log_window_size/;
4810 use strict;
4811 use warnings;
4812 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4813
4814 BEGIN {
4815 # enforce temporary pool usage for some simple functions
4816 no strict 'refs';
4817 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4818 get_file/) {
4819 my $SUPER = "SUPER::$f";
4820 *$f = sub {
4821 my $self = shift;
4822 my $pool = SVN::Pool->new;
4823 my @ret = $self->$SUPER(@_,$pool);
4824 $pool->clear;
4825 wantarray ? @ret : $ret[0];
4826 };
4827 }
4828 }
4829
4830 sub _auth_providers () {
4831 [
4832 SVN::Client::get_simple_provider(),
4833 SVN::Client::get_ssl_server_trust_file_provider(),
4834 SVN::Client::get_simple_prompt_provider(
4835 \&Git::SVN::Prompt::simple, 2),
4836 SVN::Client::get_ssl_client_cert_file_provider(),
4837 SVN::Client::get_ssl_client_cert_prompt_provider(
4838 \&Git::SVN::Prompt::ssl_client_cert, 2),
4839 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4840 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4841 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4842 SVN::Client::get_username_provider(),
4843 SVN::Client::get_ssl_server_trust_prompt_provider(
4844 \&Git::SVN::Prompt::ssl_server_trust),
4845 SVN::Client::get_username_prompt_provider(
4846 \&Git::SVN::Prompt::username, 2)
4847 ]
4848 }
4849
4850 sub escape_uri_only {
4851 my ($uri) = @_;
4852 my @tmp;
4853 foreach (split m{/}, $uri) {
4854 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4855 push @tmp, $_;
4856 }
4857 join('/', @tmp);
4858 }
4859
4860 sub escape_url {
4861 my ($url) = @_;
4862 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4863 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4864 $url = "$scheme://$domain$uri";
4865 }
4866 $url;
4867 }
4868
4869 sub new {
4870 my ($class, $url) = @_;
4871 $url =~ s!/+$!!;
4872 return $RA if ($RA && $RA->{url} eq $url);
4873
4874 ::_req_svn();
4875
4876 SVN::_Core::svn_config_ensure($config_dir, undef);
4877 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4878 my $config = SVN::Core::config_get_config($config_dir);
4879 $RA = undef;
4880 my $dont_store_passwords = 1;
4881 my $conf_t = ${$config}{'config'};
4882 {
4883 no warnings 'once';
4884 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4885 # produces warnings that variables are used only once.
4886 # I had not found the better way to shut them up, so
4887 # the warnings of type 'once' are disabled in this block.
4888 if (SVN::_Core::svn_config_get_bool($conf_t,
4889 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4890 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4891 1) == 0) {
4892 SVN::_Core::svn_auth_set_parameter($baton,
4893 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4894 bless (\$dont_store_passwords, "_p_void"));
4895 }
4896 if (SVN::_Core::svn_config_get_bool($conf_t,
4897 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4898 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4899 1) == 0) {
4900 $Git::SVN::Prompt::_no_auth_cache = 1;
4901 }
4902 } # no warnings 'once'
4903 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4904 config => $config,
4905 pool => SVN::Pool->new,
4906 auth_provider_callbacks => $callbacks);
4907 $self->{url} = $url;
4908 $self->{svn_path} = $url;
4909 $self->{repos_root} = $self->get_repos_root;
4910 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4911 $self->{cache} = { check_path => { r => 0, data => {} },
4912 get_dir => { r => 0, data => {} } };
4913 $RA = bless $self, $class;
4914 }
4915
4916 sub check_path {
4917 my ($self, $path, $r) = @_;
4918 my $cache = $self->{cache}->{check_path};
4919 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4920 return $cache->{data}->{$path};
4921 }
4922 my $pool = SVN::Pool->new;
4923 my $t = $self->SUPER::check_path($path, $r, $pool);
4924 $pool->clear;
4925 if ($r != $cache->{r}) {
4926 %{$cache->{data}} = ();
4927 $cache->{r} = $r;
4928 }
4929 $cache->{data}->{$path} = $t;
4930 }
4931
4932 sub get_dir {
4933 my ($self, $dir, $r) = @_;
4934 my $cache = $self->{cache}->{get_dir};
4935 if ($r == $cache->{r}) {
4936 if (my $x = $cache->{data}->{$dir}) {
4937 return wantarray ? @$x : $x->[0];
4938 }
4939 }
4940 my $pool = SVN::Pool->new;
4941 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4942 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4943 $pool->clear;
4944 if ($r != $cache->{r}) {
4945 %{$cache->{data}} = ();
4946 $cache->{r} = $r;
4947 }
4948 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4949 wantarray ? (\%dirents, $r, $props) : \%dirents;
4950 }
4951
4952 sub DESTROY {
4953 # do not call the real DESTROY since we store ourselves in $RA
4954 }
4955
4956 # get_log(paths, start, end, limit,
4957 # discover_changed_paths, strict_node_history, receiver)
4958 sub get_log {
4959 my ($self, @args) = @_;
4960 my $pool = SVN::Pool->new;
4961
4962 # svn_log_changed_path_t objects passed to get_log are likely to be
4963 # overwritten even if only the refs are copied to an external variable,
4964 # so we should dup the structures in their entirety. Using an
4965 # externally passed pool (instead of our temporary and quickly cleared
4966 # pool in Git::SVN::Ra) does not help matters at all...
4967 my $receiver = pop @args;
4968 my $prefix = "/".$self->{svn_path};
4969 $prefix =~ s#/+($)##;
4970 my $prefix_regex = qr#^\Q$prefix\E#;
4971 push(@args, sub {
4972 my ($paths) = $_[0];
4973 return &$receiver(@_) unless $paths;
4974 $_[0] = ();
4975 foreach my $p (keys %$paths) {
4976 my $i = $paths->{$p};
4977 # Make path relative to our url, not repos_root
4978 $p =~ s/$prefix_regex//;
4979 my %s = map { $_ => $i->$_; }
4980 qw/copyfrom_path copyfrom_rev action/;
4981 if ($s{'copyfrom_path'}) {
4982 $s{'copyfrom_path'} =~ s/$prefix_regex//;
4983 }
4984 $_[0]{$p} = \%s;
4985 }
4986 &$receiver(@_);
4987 });
4988
4989
4990 # the limit parameter was not supported in SVN 1.1.x, so we
4991 # drop it. Therefore, the receiver callback passed to it
4992 # is made aware of this limitation by being wrapped if
4993 # the limit passed to is being wrapped.
4994 if ($SVN::Core::VERSION le '1.2.0') {
4995 my $limit = splice(@args, 3, 1);
4996 if ($limit > 0) {
4997 my $receiver = pop @args;
4998 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4999 }
5000 }
5001 my $ret = $self->SUPER::get_log(@args, $pool);
5002 $pool->clear;
5003 $ret;
5004 }
5005
5006 sub trees_match {
5007 my ($self, $url1, $rev1, $url2, $rev2) = @_;
5008 my $ctx = SVN::Client->new(auth => _auth_providers);
5009 my $out = IO::File->new_tmpfile;
5010
5011 # older SVN (1.1.x) doesn't take $pool as the last parameter for
5012 # $ctx->diff(), so we'll create a default one
5013 my $pool = SVN::Pool->new_default_sub;
5014
5015 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5016 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5017 $out->flush;
5018 my $ret = (($out->stat)[7] == 0);
5019 close $out or croak $!;
5020
5021 $ret;
5022 }
5023
5024 sub get_commit_editor {
5025 my ($self, $log, $cb, $pool) = @_;
5026 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
5027 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5028 }
5029
5030 sub gs_do_update {
5031 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5032 my $new = ($rev_a == $rev_b);
5033 my $path = $gs->{path};
5034
5035 if ($new && -e $gs->{index}) {
5036 unlink $gs->{index} or die
5037 "Couldn't unlink index: $gs->{index}: $!\n";
5038 }
5039 my $pool = SVN::Pool->new;
5040 $editor->set_path_strip($path);
5041 my (@pc) = split m#/#, $path;
5042 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5043 1, $editor, $pool);
5044 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5045
5046 # Since we can't rely on svn_ra_reparent being available, we'll
5047 # just have to do some magic with set_path to make it so
5048 # we only want a partial path.
5049 my $sp = '';
5050 my $final = join('/', @pc);
5051 while (@pc) {
5052 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5053 $sp .= '/' if length $sp;
5054 $sp .= shift @pc;
5055 }
5056 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5057
5058 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5059
5060 $reporter->finish_report($pool);
5061 $pool->clear;
5062 $editor->{git_commit_ok};
5063 }
5064
5065 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5066 # svn_ra_reparent didn't work before 1.4)
5067 sub gs_do_switch {
5068 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5069 my $path = $gs->{path};
5070 my $pool = SVN::Pool->new;
5071
5072 my $full_url = $self->{url};
5073 my $old_url = $full_url;
5074 $full_url .= '/' . $path if length $path;
5075 my ($ra, $reparented);
5076
5077 if ($old_url =~ m#^svn(\+ssh)?://# ||
5078 ($full_url =~ m#^https?://# &&
5079 escape_url($full_url) ne $full_url)) {
5080 $_[0] = undef;
5081 $self = undef;
5082 $RA = undef;
5083 $ra = Git::SVN::Ra->new($full_url);
5084 $ra_invalid = 1;
5085 } elsif ($old_url ne $full_url) {
5086 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5087 $self->{url} = $full_url;
5088 $reparented = 1;
5089 }
5090
5091 $ra ||= $self;
5092 $url_b = escape_url($url_b);
5093 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5094 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5095 $reporter->set_path('', $rev_a, 0, @lock, $pool);
5096 $reporter->finish_report($pool);
5097
5098 if ($reparented) {
5099 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5100 $self->{url} = $old_url;
5101 }
5102
5103 $pool->clear;
5104 $editor->{git_commit_ok};
5105 }
5106
5107 sub longest_common_path {
5108 my ($gsv, $globs) = @_;
5109 my %common;
5110 my $common_max = scalar @$gsv;
5111
5112 foreach my $gs (@$gsv) {
5113 my @tmp = split m#/#, $gs->{path};
5114 my $p = '';
5115 foreach (@tmp) {
5116 $p .= length($p) ? "/$_" : $_;
5117 $common{$p} ||= 0;
5118 $common{$p}++;
5119 }
5120 }
5121 $globs ||= [];
5122 $common_max += scalar @$globs;
5123 foreach my $glob (@$globs) {
5124 my @tmp = split m#/#, $glob->{path}->{left};
5125 my $p = '';
5126 foreach (@tmp) {
5127 $p .= length($p) ? "/$_" : $_;
5128 $common{$p} ||= 0;
5129 $common{$p}++;
5130 }
5131 }
5132
5133 my $longest_path = '';
5134 foreach (sort {length $b <=> length $a} keys %common) {
5135 if ($common{$_} == $common_max) {
5136 $longest_path = $_;
5137 last;
5138 }
5139 }
5140 $longest_path;
5141 }
5142
5143 sub gs_fetch_loop_common {
5144 my ($self, $base, $head, $gsv, $globs) = @_;
5145 return if ($base > $head);
5146 my $inc = $_log_window_size;
5147 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5148 my $longest_path = longest_common_path($gsv, $globs);
5149 my $ra_url = $self->{url};
5150 my $find_trailing_edge;
5151 while (1) {
5152 my %revs;
5153 my $err;
5154 my $err_handler = $SVN::Error::handler;
5155 $SVN::Error::handler = sub {
5156 ($err) = @_;
5157 skip_unknown_revs($err);
5158 };
5159 sub _cb {
5160 my ($paths, $r, $author, $date, $log) = @_;
5161 [ $paths,
5162 { author => $author, date => $date, log => $log } ];
5163 }
5164 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5165 sub { $revs{$_[1]} = _cb(@_) });
5166 if ($err) {
5167 print "Checked through r$max\r";
5168 } else {
5169 $find_trailing_edge = 1;
5170 }
5171 if ($err and $find_trailing_edge) {
5172 print STDERR "Path '$longest_path' ",
5173 "was probably deleted:\n",
5174 $err->expanded_message,
5175 "\nWill attempt to follow ",
5176 "revisions r$min .. r$max ",
5177 "committed before the deletion\n";
5178 my $hi = $max;
5179 while (--$hi >= $min) {
5180 my $ok;
5181 $self->get_log([$longest_path], $min, $hi,
5182 0, 1, 1, sub {
5183 $ok = $_[1];
5184 $revs{$_[1]} = _cb(@_) });
5185 if ($ok) {
5186 print STDERR "r$min .. r$ok OK\n";
5187 last;
5188 }
5189 }
5190 $find_trailing_edge = 0;
5191 }
5192 $SVN::Error::handler = $err_handler;
5193
5194 my %exists = map { $_->{path} => $_ } @$gsv;
5195 foreach my $r (sort {$a <=> $b} keys %revs) {
5196 my ($paths, $logged) = @{$revs{$r}};
5197
5198 foreach my $gs ($self->match_globs(\%exists, $paths,
5199 $globs, $r)) {
5200 if ($gs->rev_map_max >= $r) {
5201 next;
5202 }
5203 next unless $gs->match_paths($paths, $r);
5204 $gs->{logged_rev_props} = $logged;
5205 if (my $last_commit = $gs->last_commit) {
5206 $gs->assert_index_clean($last_commit);
5207 }
5208 my $log_entry = $gs->do_fetch($paths, $r);
5209 if ($log_entry) {
5210 $gs->do_git_commit($log_entry);
5211 }
5212 $INDEX_FILES{$gs->{index}} = 1;
5213 }
5214 foreach my $g (@$globs) {
5215 my $k = "svn-remote.$g->{remote}." .
5216 "$g->{t}-maxRev";
5217 Git::SVN::tmp_config($k, $r);
5218 }
5219 if ($ra_invalid) {
5220 $_[0] = undef;
5221 $self = undef;
5222 $RA = undef;
5223 $self = Git::SVN::Ra->new($ra_url);
5224 $ra_invalid = undef;
5225 }
5226 }
5227 # pre-fill the .rev_db since it'll eventually get filled in
5228 # with '0' x40 if something new gets committed
5229 foreach my $gs (@$gsv) {
5230 next if $gs->rev_map_max >= $max;
5231 next if defined $gs->rev_map_get($max);
5232 $gs->rev_map_set($max, 0 x40);
5233 }
5234 foreach my $g (@$globs) {
5235 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5236 Git::SVN::tmp_config($k, $max);
5237 }
5238 last if $max >= $head;
5239 $min = $max + 1;
5240 $max += $inc;
5241 $max = $head if ($max > $head);
5242 }
5243 Git::SVN::gc();
5244 }
5245
5246 sub get_dir_globbed {
5247 my ($self, $left, $depth, $r) = @_;
5248
5249 my @x = eval { $self->get_dir($left, $r) };
5250 return unless scalar @x == 3;
5251 my $dirents = $x[0];
5252 my @finalents;
5253 foreach my $de (keys %$dirents) {
5254 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5255 if ($depth > 1) {
5256 my @args = ("$left/$de", $depth - 1, $r);
5257 foreach my $dir ($self->get_dir_globbed(@args)) {
5258 push @finalents, "$de/$dir";
5259 }
5260 } else {
5261 push @finalents, $de;
5262 }
5263 }
5264 @finalents;
5265 }
5266
5267 sub match_globs {
5268 my ($self, $exists, $paths, $globs, $r) = @_;
5269
5270 sub get_dir_check {
5271 my ($self, $exists, $g, $r) = @_;
5272
5273 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5274 $g->{path}->{depth},
5275 $r);
5276
5277 foreach my $de (@dirs) {
5278 my $p = $g->{path}->full_path($de);
5279 next if $exists->{$p};
5280 next if (length $g->{path}->{right} &&
5281 ($self->check_path($p, $r) !=
5282 $SVN::Node::dir));
5283 next unless $p =~ /$g->{path}->{regex}/;
5284 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5285 $g->{ref}->full_path($de), 1);
5286 }
5287 }
5288 foreach my $g (@$globs) {
5289 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5290 if ($path->{action} =~ /^[AR]$/) {
5291 get_dir_check($self, $exists, $g, $r);
5292 }
5293 }
5294 foreach (keys %$paths) {
5295 if (/$g->{path}->{left_regex}/ &&
5296 !/$g->{path}->{regex}/) {
5297 next if $paths->{$_}->{action} !~ /^[AR]$/;
5298 get_dir_check($self, $exists, $g, $r);
5299 }
5300 next unless /$g->{path}->{regex}/;
5301 my $p = $1;
5302 my $pathname = $g->{path}->full_path($p);
5303 next if $exists->{$pathname};
5304 next if ($self->check_path($pathname, $r) !=
5305 $SVN::Node::dir);
5306 $exists->{$pathname} = Git::SVN->init(
5307 $self->{url}, $pathname, undef,
5308 $g->{ref}->full_path($p), 1);
5309 }
5310 my $c = '';
5311 foreach (split m#/#, $g->{path}->{left}) {
5312 $c .= "/$_";
5313 next unless ($paths->{$c} &&
5314 ($paths->{$c}->{action} =~ /^[AR]$/));
5315 get_dir_check($self, $exists, $g, $r);
5316 }
5317 }
5318 values %$exists;
5319 }
5320
5321 sub minimize_url {
5322 my ($self) = @_;
5323 return $self->{url} if ($self->{url} eq $self->{repos_root});
5324 my $url = $self->{repos_root};
5325 my @components = split(m!/!, $self->{svn_path});
5326 my $c = '';
5327 do {
5328 $url .= "/$c" if length $c;
5329 eval {
5330 my $ra = (ref $self)->new($url);
5331 my $latest = $ra->get_latest_revnum;
5332 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5333 };
5334 } while ($@ && ($c = shift @components));
5335 $url;
5336 }
5337
5338 sub can_do_switch {
5339 my $self = shift;
5340 unless (defined $can_do_switch) {
5341 my $pool = SVN::Pool->new;
5342 my $rep = eval {
5343 $self->do_switch(1, '', 0, $self->{url},
5344 SVN::Delta::Editor->new, $pool);
5345 };
5346 if ($@) {
5347 $can_do_switch = 0;
5348 } else {
5349 $rep->abort_report($pool);
5350 $can_do_switch = 1;
5351 }
5352 $pool->clear;
5353 }
5354 $can_do_switch;
5355 }
5356
5357 sub skip_unknown_revs {
5358 my ($err) = @_;
5359 my $errno = $err->apr_err();
5360 # Maybe the branch we're tracking didn't
5361 # exist when the repo started, so it's
5362 # not an error if it doesn't, just continue
5363 #
5364 # Wonderfully consistent library, eh?
5365 # 160013 - svn:// and file://
5366 # 175002 - http(s)://
5367 # 175007 - http(s):// (this repo required authorization, too...)
5368 # More codes may be discovered later...
5369 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5370 my $err_key = $err->expanded_message;
5371 # revision numbers change every time, filter them out
5372 $err_key =~ s/\d+/\0/g;
5373 $err_key = "$errno\0$err_key";
5374 unless ($ignored_err{$err_key}) {
5375 warn "W: Ignoring error from SVN, path probably ",
5376 "does not exist: ($errno): ",
5377 $err->expanded_message,"\n";
5378 warn "W: Do not be alarmed at the above message ",
5379 "git-svn is just searching aggressively for ",
5380 "old history.\n",
5381 "This may take a while on large repositories\n";
5382 $ignored_err{$err_key} = 1;
5383 }
5384 return;
5385 }
5386 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5387 }
5388
5389 package Git::SVN::Log;
5390 use strict;
5391 use warnings;
5392 use POSIX qw/strftime/;
5393 use Time::Local;
5394 use constant commit_log_separator => ('-' x 72) . "\n";
5395 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5396 %rusers $show_commit $incremental/;
5397 my $l_fmt;
5398
5399 sub cmt_showable {
5400 my ($c) = @_;
5401 return 1 if defined $c->{r};
5402
5403 # big commit message got truncated by the 16k pretty buffer in rev-list
5404 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5405 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5406 @{$c->{l}} = ();
5407 my @log = command(qw/cat-file commit/, $c->{c});
5408
5409 # shift off the headers
5410 shift @log while ($log[0] ne '');
5411 shift @log;
5412
5413 # TODO: make $c->{l} not have a trailing newline in the future
5414 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5415
5416 (undef, $c->{r}, undef) = ::extract_metadata(
5417 (grep(/^git-svn-id: /, @log))[-1]);
5418 }
5419 return defined $c->{r};
5420 }
5421
5422 sub log_use_color {
5423 return $color || Git->repository->get_colorbool('color.diff');
5424 }
5425
5426 sub git_svn_log_cmd {
5427 my ($r_min, $r_max, @args) = @_;
5428 my $head = 'HEAD';
5429 my (@files, @log_opts);
5430 foreach my $x (@args) {
5431 if ($x eq '--' || @files) {
5432 push @files, $x;
5433 } else {
5434 if (::verify_ref("$x^0")) {
5435 $head = $x;
5436 } else {
5437 push @log_opts, $x;
5438 }
5439 }
5440 }
5441
5442 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5443 $gs ||= Git::SVN->_new;
5444 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5445 $gs->refname);
5446 push @cmd, '-r' unless $non_recursive;
5447 push @cmd, qw/--raw --name-status/ if $verbose;
5448 push @cmd, '--color' if log_use_color();
5449 push @cmd, @log_opts;
5450 if (defined $r_max && $r_max == $r_min) {
5451 push @cmd, '--max-count=1';
5452 if (my $c = $gs->rev_map_get($r_max)) {
5453 push @cmd, $c;
5454 }
5455 } elsif (defined $r_max) {
5456 if ($r_max < $r_min) {
5457 ($r_min, $r_max) = ($r_max, $r_min);
5458 }
5459 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5460 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5461 # If there are no commits in the range, both $c_max and $c_min
5462 # will be undefined. If there is at least 1 commit in the
5463 # range, both will be defined.
5464 return () if !defined $c_min || !defined $c_max;
5465 if ($c_min eq $c_max) {
5466 push @cmd, '--max-count=1', $c_min;
5467 } else {
5468 push @cmd, '--boundary', "$c_min..$c_max";
5469 }
5470 }
5471 return (@cmd, @files);
5472 }
5473
5474 # adapted from pager.c
5475 sub config_pager {
5476 if (! -t *STDOUT) {
5477 $ENV{GIT_PAGER_IN_USE} = 'false';
5478 $pager = undef;
5479 return;
5480 }
5481 chomp($pager = command_oneline(qw(var GIT_PAGER)));
5482 if ($pager eq 'cat') {
5483 $pager = undef;
5484 }
5485 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5486 }
5487
5488 sub run_pager {
5489 return unless defined $pager;
5490 pipe my ($rfd, $wfd) or return;
5491 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5492 if (!$pid) {
5493 open STDOUT, '>&', $wfd or
5494 ::fatal "Can't redirect to stdout: $!";
5495 return;
5496 }
5497 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5498 $ENV{LESS} ||= 'FRSX';
5499 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5500 }
5501
5502 sub format_svn_date {
5503 # some systmes don't handle or mishandle %z, so be creative.
5504 my $t = shift || time;
5505 my $gm = timelocal(gmtime($t));
5506 my $sign = qw( + + - )[ $t <=> $gm ];
5507 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5508 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5509 }
5510
5511 sub parse_git_date {
5512 my ($t, $tz) = @_;
5513 # Date::Parse isn't in the standard Perl distro :(
5514 if ($tz =~ s/^\+//) {
5515 $t += tz_to_s_offset($tz);
5516 } elsif ($tz =~ s/^\-//) {
5517 $t -= tz_to_s_offset($tz);
5518 }
5519 return $t;
5520 }
5521
5522 sub set_local_timezone {
5523 if (defined $TZ) {
5524 $ENV{TZ} = $TZ;
5525 } else {
5526 delete $ENV{TZ};
5527 }
5528 }
5529
5530 sub tz_to_s_offset {
5531 my ($tz) = @_;
5532 $tz =~ s/(\d\d)$//;
5533 return ($1 * 60) + ($tz * 3600);
5534 }
5535
5536 sub get_author_info {
5537 my ($dest, $author, $t, $tz) = @_;
5538 $author =~ s/(?:^\s*|\s*$)//g;
5539 $dest->{a_raw} = $author;
5540 my $au;
5541 if ($::_authors) {
5542 $au = $rusers{$author} || undef;
5543 }
5544 if (!$au) {
5545 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5546 }
5547 $dest->{t} = $t;
5548 $dest->{tz} = $tz;
5549 $dest->{a} = $au;
5550 $dest->{t_utc} = parse_git_date($t, $tz);
5551 }
5552
5553 sub process_commit {
5554 my ($c, $r_min, $r_max, $defer) = @_;
5555 if (defined $r_min && defined $r_max) {
5556 if ($r_min == $c->{r} && $r_min == $r_max) {
5557 show_commit($c);
5558 return 0;
5559 }
5560 return 1 if $r_min == $r_max;
5561 if ($r_min < $r_max) {
5562 # we need to reverse the print order
5563 return 0 if (defined $limit && --$limit < 0);
5564 push @$defer, $c;
5565 return 1;
5566 }
5567 if ($r_min != $r_max) {
5568 return 1 if ($r_min < $c->{r});
5569 return 1 if ($r_max > $c->{r});
5570 }
5571 }
5572 return 0 if (defined $limit && --$limit < 0);
5573 show_commit($c);
5574 return 1;
5575 }
5576
5577 sub show_commit {
5578 my $c = shift;
5579 if ($oneline) {
5580 my $x = "\n";
5581 if (my $l = $c->{l}) {
5582 while ($l->[0] =~ /^\s*$/) { shift @$l }
5583 $x = $l->[0];
5584 }
5585 $l_fmt ||= 'A' . length($c->{r});
5586 print 'r',pack($l_fmt, $c->{r}),' | ';
5587 print "$c->{c} | " if $show_commit;
5588 print $x;
5589 } else {
5590 show_commit_normal($c);
5591 }
5592 }
5593
5594 sub show_commit_changed_paths {
5595 my ($c) = @_;
5596 return unless $c->{changed};
5597 print "Changed paths:\n", @{$c->{changed}};
5598 }
5599
5600 sub show_commit_normal {
5601 my ($c) = @_;
5602 print commit_log_separator, "r$c->{r} | ";
5603 print "$c->{c} | " if $show_commit;
5604 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5605 my $nr_line = 0;
5606
5607 if (my $l = $c->{l}) {
5608 while ($l->[$#$l] eq "\n" && $#$l > 0
5609 && $l->[($#$l - 1)] eq "\n") {
5610 pop @$l;
5611 }
5612 $nr_line = scalar @$l;
5613 if (!$nr_line) {
5614 print "1 line\n\n\n";
5615 } else {
5616 if ($nr_line == 1) {
5617 $nr_line = '1 line';
5618 } else {
5619 $nr_line .= ' lines';
5620 }
5621 print $nr_line, "\n";
5622 show_commit_changed_paths($c);
5623 print "\n";
5624 print $_ foreach @$l;
5625 }
5626 } else {
5627 print "1 line\n";
5628 show_commit_changed_paths($c);
5629 print "\n";
5630
5631 }
5632 foreach my $x (qw/raw stat diff/) {
5633 if ($c->{$x}) {
5634 print "\n";
5635 print $_ foreach @{$c->{$x}}
5636 }
5637 }
5638 }
5639
5640 sub cmd_show_log {
5641 my (@args) = @_;
5642 my ($r_min, $r_max);
5643 my $r_last = -1; # prevent dupes
5644 set_local_timezone();
5645 if (defined $::_revision) {
5646 if ($::_revision =~ /^(\d+):(\d+)$/) {
5647 ($r_min, $r_max) = ($1, $2);
5648 } elsif ($::_revision =~ /^\d+$/) {
5649 $r_min = $r_max = $::_revision;
5650 } else {
5651 ::fatal "-r$::_revision is not supported, use ",
5652 "standard 'git log' arguments instead";
5653 }
5654 }
5655
5656 config_pager();
5657 @args = git_svn_log_cmd($r_min, $r_max, @args);
5658 if (!@args) {
5659 print commit_log_separator unless $incremental || $oneline;
5660 return;
5661 }
5662 my $log = command_output_pipe(@args);
5663 run_pager();
5664 my (@k, $c, $d, $stat);
5665 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5666 while (<$log>) {
5667 if (/^${esc_color}commit -?($::sha1_short)/o) {
5668 my $cmt = $1;
5669 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5670 $r_last = $c->{r};
5671 process_commit($c, $r_min, $r_max, \@k) or
5672 goto out;
5673 }
5674 $d = undef;
5675 $c = { c => $cmt };
5676 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5677 get_author_info($c, $1, $2, $3);
5678 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5679 # ignore
5680 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5681 push @{$c->{raw}}, $_;
5682 } elsif (/^${esc_color}[ACRMDT]\t/) {
5683 # we could add $SVN->{svn_path} here, but that requires
5684 # remote access at the moment (repo_path_split)...
5685 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5686 push @{$c->{changed}}, $_;
5687 } elsif (/^${esc_color}diff /o) {
5688 $d = 1;
5689 push @{$c->{diff}}, $_;
5690 } elsif ($d) {
5691 push @{$c->{diff}}, $_;
5692 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5693 $esc_color*[\+\-]*$esc_color$/x) {
5694 $stat = 1;
5695 push @{$c->{stat}}, $_;
5696 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5697 push @{$c->{stat}}, $_;
5698 $stat = undef;
5699 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5700 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5701 } elsif (s/^${esc_color} //o) {
5702 push @{$c->{l}}, $_;
5703 }
5704 }
5705 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5706 $r_last = $c->{r};
5707 process_commit($c, $r_min, $r_max, \@k);
5708 }
5709 if (@k) {
5710 ($r_min, $r_max) = ($r_max, $r_min);
5711 process_commit($_, $r_min, $r_max) foreach reverse @k;
5712 }
5713 out:
5714 close $log;
5715 print commit_log_separator unless $incremental || $oneline;
5716 }
5717
5718 sub cmd_blame {
5719 my $path = pop;
5720
5721 config_pager();
5722 run_pager();
5723
5724 my ($fh, $ctx, $rev);
5725
5726 if ($_git_format) {
5727 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5728 while (my $line = <$fh>) {
5729 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5730 # Uncommitted edits show up as a rev ID of
5731 # all zeros, which we can't look up with
5732 # cmt_metadata
5733 if ($1 !~ /^0+$/) {
5734 (undef, $rev, undef) =
5735 ::cmt_metadata($1);
5736 $rev = '0' if (!$rev);
5737 } else {
5738 $rev = '0';
5739 }
5740 $rev = sprintf('%-10s', $rev);
5741 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5742 }
5743 print $line;
5744 }
5745 } else {
5746 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5747 '--', $path);
5748 my ($sha1);
5749 my %authors;
5750 my @buffer;
5751 my %dsha; #distinct sha keys
5752
5753 while (my $line = <$fh>) {
5754 push @buffer, $line;
5755 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5756 $dsha{$1} = 1;
5757 }
5758 }
5759
5760 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5761
5762 foreach my $line (@buffer) {
5763 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5764 $rev = $s2r->{$1};
5765 $rev = '0' if (!$rev)
5766 }
5767 elsif ($line =~ /^author (.*)/) {
5768 $authors{$rev} = $1;
5769 $authors{$rev} =~ s/\s/_/g;
5770 }
5771 elsif ($line =~ /^\t(.*)$/) {
5772 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5773 }
5774 }
5775 }
5776 command_close_pipe($fh, $ctx);
5777 }
5778
5779 package Git::SVN::Migration;
5780 # these version numbers do NOT correspond to actual version numbers
5781 # of git nor git-svn. They are just relative.
5782 #
5783 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5784 #
5785 # v1 layout: .git/$id/info/url, refs/remotes/$id
5786 #
5787 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5788 #
5789 # v3 layout: .git/svn/$id, refs/remotes/$id
5790 # - info/url may remain for backwards compatibility
5791 # - this is what we migrate up to this layout automatically,
5792 # - this will be used by git svn init on single branches
5793 # v3.1 layout (auto migrated):
5794 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5795 # for backwards compatibility
5796 #
5797 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5798 # - this is only created for newly multi-init-ed
5799 # repositories. Similar in spirit to the
5800 # --use-separate-remotes option in git-clone (now default)
5801 # - we do not automatically migrate to this (following
5802 # the example set by core git)
5803 #
5804 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
5805 # - newer, more-efficient format that uses 24-bytes per record
5806 # with no filler space.
5807 # - use xxd -c24 < .rev_map.$UUID to view and debug
5808 # - This is a one-way migration, repositories updated to the
5809 # new format will not be able to use old git-svn without
5810 # rebuilding the .rev_db. Rebuilding the rev_db is not
5811 # possible if noMetadata or useSvmProps are set; but should
5812 # be no problem for users that use the (sensible) defaults.
5813 use strict;
5814 use warnings;
5815 use Carp qw/croak/;
5816 use File::Path qw/mkpath/;
5817 use File::Basename qw/dirname basename/;
5818 use vars qw/$_minimize/;
5819
5820 sub migrate_from_v0 {
5821 my $git_dir = $ENV{GIT_DIR};
5822 return undef unless -d $git_dir;
5823 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5824 my $migrated = 0;
5825 while (<$fh>) {
5826 chomp;
5827 my ($id, $orig_ref) = ($_, $_);
5828 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5829 next unless -f "$git_dir/$id/info/url";
5830 my $new_ref = "refs/remotes/$id";
5831 if (::verify_ref("$new_ref^0")) {
5832 print STDERR "W: $orig_ref is probably an old ",
5833 "branch used by an ancient version of ",
5834 "git-svn.\n",
5835 "However, $new_ref also exists.\n",
5836 "We will not be able ",
5837 "to use this branch until this ",
5838 "ambiguity is resolved.\n";
5839 next;
5840 }
5841 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5842 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5843 command_noisy('update-ref', $new_ref, $orig_ref);
5844 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5845 $migrated++;
5846 }
5847 command_close_pipe($fh, $ctx);
5848 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5849 $migrated;
5850 }
5851
5852 sub migrate_from_v1 {
5853 my $git_dir = $ENV{GIT_DIR};
5854 my $migrated = 0;
5855 return $migrated unless -d $git_dir;
5856 my $svn_dir = "$git_dir/svn";
5857
5858 # just in case somebody used 'svn' as their $id at some point...
5859 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5860
5861 print STDERR "Migrating from a git-svn v1 layout...\n";
5862 mkpath([$svn_dir]);
5863 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5864 "$svn_dir\n\t(required for this version ",
5865 "($::VERSION) of git-svn) does not exist.\n";
5866 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5867 while (<$fh>) {
5868 my $x = $_;
5869 next unless $x =~ s#^refs/remotes/##;
5870 chomp $x;
5871 next unless -f "$git_dir/$x/info/url";
5872 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5873 next unless $u;
5874 my $dn = dirname("$git_dir/svn/$x");
5875 mkpath([$dn]) unless -d $dn;
5876 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5877 mkpath(["$git_dir/svn/svn"]);
5878 print STDERR " - $git_dir/$x/info => ",
5879 "$git_dir/svn/$x/info\n";
5880 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5881 croak "$!: $x";
5882 # don't worry too much about these, they probably
5883 # don't exist with repos this old (save for index,
5884 # and we can easily regenerate that)
5885 foreach my $f (qw/unhandled.log index .rev_db/) {
5886 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5887 }
5888 } else {
5889 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5890 rename "$git_dir/$x", "$git_dir/svn/$x" or
5891 croak "$!: $x";
5892 }
5893 $migrated++;
5894 }
5895 command_close_pipe($fh, $ctx);
5896 print STDERR "Done migrating from a git-svn v1 layout\n";
5897 $migrated;
5898 }
5899
5900 sub read_old_urls {
5901 my ($l_map, $pfx, $path) = @_;
5902 my @dir;
5903 foreach (<$path/*>) {
5904 if (-r "$_/info/url") {
5905 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5906 my $ref_id = $pfx . basename $_;
5907 my $url = ::file_to_s("$_/info/url");
5908 $l_map->{$ref_id} = $url;
5909 } elsif (-d $_) {
5910 push @dir, $_;
5911 }
5912 }
5913 foreach (@dir) {
5914 my $x = $_;
5915 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5916 read_old_urls($l_map, $x, $_);
5917 }
5918 }
5919
5920 sub migrate_from_v2 {
5921 my @cfg = command(qw/config -l/);
5922 return if grep /^svn-remote\..+\.url=/, @cfg;
5923 my %l_map;
5924 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5925 my $migrated = 0;
5926
5927 foreach my $ref_id (sort keys %l_map) {
5928 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5929 if ($@) {
5930 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5931 }
5932 $migrated++;
5933 }
5934 $migrated;
5935 }
5936
5937 sub minimize_connections {
5938 my $r = Git::SVN::read_all_remotes();
5939 my $new_urls = {};
5940 my $root_repos = {};
5941 foreach my $repo_id (keys %$r) {
5942 my $url = $r->{$repo_id}->{url} or next;
5943 my $fetch = $r->{$repo_id}->{fetch} or next;
5944 my $ra = Git::SVN::Ra->new($url);
5945
5946 # skip existing cases where we already connect to the root
5947 if (($ra->{url} eq $ra->{repos_root}) ||
5948 ($ra->{repos_root} eq $repo_id)) {
5949 $root_repos->{$ra->{url}} = $repo_id;
5950 next;
5951 }
5952
5953 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5954 my $root_path = $ra->{url};
5955 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5956 foreach my $path (keys %$fetch) {
5957 my $ref_id = $fetch->{$path};
5958 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5959
5960 # make sure we can read when connecting to
5961 # a higher level of a repository
5962 my ($last_rev, undef) = $gs->last_rev_commit;
5963 if (!defined $last_rev) {
5964 $last_rev = eval {
5965 $root_ra->get_latest_revnum;
5966 };
5967 next if $@;
5968 }
5969 my $new = $root_path;
5970 $new .= length $path ? "/$path" : '';
5971 eval {
5972 $root_ra->get_log([$new], $last_rev, $last_rev,
5973 0, 0, 1, sub { });
5974 };
5975 next if $@;
5976 $new_urls->{$ra->{repos_root}}->{$new} =
5977 { ref_id => $ref_id,
5978 old_repo_id => $repo_id,
5979 old_path => $path };
5980 }
5981 }
5982
5983 my @emptied;
5984 foreach my $url (keys %$new_urls) {
5985 # see if we can re-use an existing [svn-remote "repo_id"]
5986 # instead of creating a(n ugly) new section:
5987 my $repo_id = $root_repos->{$url} || $url;
5988
5989 my $fetch = $new_urls->{$url};
5990 foreach my $path (keys %$fetch) {
5991 my $x = $fetch->{$path};
5992 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5993 my $pfx = "svn-remote.$x->{old_repo_id}";
5994
5995 my $old_fetch = quotemeta("$x->{old_path}:".
5996 "$x->{ref_id}");
5997 command_noisy(qw/config --unset/,
5998 "$pfx.fetch", '^'. $old_fetch . '$');
5999 delete $r->{$x->{old_repo_id}}->
6000 {fetch}->{$x->{old_path}};
6001 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6002 command_noisy(qw/config --unset/,
6003 "$pfx.url");
6004 push @emptied, $x->{old_repo_id}
6005 }
6006 }
6007 }
6008 if (@emptied) {
6009 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6010 print STDERR <<EOF;
6011 The following [svn-remote] sections in your config file ($file) are empty
6012 and can be safely removed:
6013 EOF
6014 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6015 }
6016 }
6017
6018 sub migration_check {
6019 migrate_from_v0();
6020 migrate_from_v1();
6021 migrate_from_v2();
6022 minimize_connections() if $_minimize;
6023 }
6024
6025 package Git::IndexInfo;
6026 use strict;
6027 use warnings;
6028 use Git qw/command_input_pipe command_close_pipe/;
6029
6030 sub new {
6031 my ($class) = @_;
6032 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6033 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6034 }
6035
6036 sub remove {
6037 my ($self, $path) = @_;
6038 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6039 return ++$self->{nr};
6040 }
6041 undef;
6042 }
6043
6044 sub update {
6045 my ($self, $mode, $hash, $path) = @_;
6046 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6047 return ++$self->{nr};
6048 }
6049 undef;
6050 }
6051
6052 sub DESTROY {
6053 my ($self) = @_;
6054 command_close_pipe($self->{gui}, $self->{ctx});
6055 }
6056
6057 package Git::SVN::GlobSpec;
6058 use strict;
6059 use warnings;
6060
6061 sub new {
6062 my ($class, $glob, $pattern_ok) = @_;
6063 my $re = $glob;
6064 $re =~ s!/+$!!g; # no need for trailing slashes
6065 my (@left, @right, @patterns);
6066 my $state = "left";
6067 my $die_msg = "Only one set of wildcard directories " .
6068 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6069 for my $part (split(m|/|, $glob)) {
6070 if ($part =~ /\*/ && $part ne "*") {
6071 die "Invalid pattern in '$glob': $part\n";
6072 } elsif ($pattern_ok && $part =~ /[{}]/ &&
6073 $part !~ /^\{[^{}]+\}/) {
6074 die "Invalid pattern in '$glob': $part\n";
6075 }
6076 if ($part eq "*") {
6077 die $die_msg if $state eq "right";
6078 $state = "pattern";
6079 push(@patterns, "[^/]*");
6080 } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6081 die $die_msg if $state eq "right";
6082 $state = "pattern";
6083 my $p = quotemeta($1);
6084 $p =~ s/\\,/|/g;
6085 push(@patterns, "(?:$p)");
6086 } else {
6087 if ($state eq "left") {
6088 push(@left, $part);
6089 } else {
6090 push(@right, $part);
6091 $state = "right";
6092 }
6093 }
6094 }
6095 my $depth = @patterns;
6096 if ($depth == 0) {
6097 die "One '*' is needed in glob: '$glob'\n";
6098 }
6099 my $left = join('/', @left);
6100 my $right = join('/', @right);
6101 $re = join('/', @patterns);
6102 $re = join('\/',
6103 grep(length, quotemeta($left), "($re)", quotemeta($right)));
6104 my $left_re = qr/^\/\Q$left\E(\/|$)/;
6105 bless { left => $left, right => $right, left_regex => $left_re,
6106 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6107 }
6108
6109 sub full_path {
6110 my ($self, $path) = @_;
6111 return (length $self->{left} ? "$self->{left}/" : '') .
6112 $path . (length $self->{right} ? "/$self->{right}" : '');
6113 }
6114
6115 __END__
6116
6117 Data structures:
6118
6119
6120 $remotes = { # returned by read_all_remotes()
6121 'svn' => {
6122 # svn-remote.svn.url=https://svn.musicpd.org
6123 url => 'https://svn.musicpd.org',
6124 # svn-remote.svn.fetch=mpd/trunk:trunk
6125 fetch => {
6126 'mpd/trunk' => 'trunk',
6127 },
6128 # svn-remote.svn.tags=mpd/tags/*:tags/*
6129 tags => {
6130 path => {
6131 left => 'mpd/tags',
6132 right => '',
6133 regex => qr!mpd/tags/([^/]+)$!,
6134 glob => 'tags/*',
6135 },
6136 ref => {
6137 left => 'tags',
6138 right => '',
6139 regex => qr!tags/([^/]+)$!,
6140 glob => 'tags/*',
6141 },
6142 }
6143 }
6144 };
6145
6146 $log_entry hashref as returned by libsvn_log_entry()
6147 {
6148 log => 'whitespace-formatted log entry
6149 ', # trailing newline is preserved
6150 revision => '8', # integer
6151 date => '2004-02-24T17:01:44.108345Z', # commit date
6152 author => 'committer name'
6153 };
6154
6155
6156 # this is generated by generate_diff();
6157 @mods = array of diff-index line hashes, each element represents one line
6158 of diff-index output
6159
6160 diff-index line ($m hash)
6161 {
6162 mode_a => first column of diff-index output, no leading ':',
6163 mode_b => second column of diff-index output,
6164 sha1_b => sha1sum of the final blob,
6165 chg => change type [MCRADT],
6166 file_a => original file name of a file (iff chg is 'C' or 'R')
6167 file_b => new/current file name of a file (any chg)
6168 }
6169 ;
6170
6171 # retval of read_url_paths{,_all}();
6172 $l_map = {
6173 # repository root url
6174 'https://svn.musicpd.org' => {
6175 # repository path # GIT_SVN_ID
6176 'mpd/trunk' => 'trunk',
6177 'mpd/tags/0.11.5' => 'tags/0.11.5',
6178 },
6179 }
6180
6181 Notes:
6182 I don't trust the each() function on unless I created %hash myself
6183 because the internal iterator may not have started at base.
kirr's git tree