?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
utf8.c: remove strbuf_write()
[git/gitweb.git] / git-svn.perl
blob2746895ae620a37c4b84419caf221db8f476e197
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision $_repository
8 $_q $_authors $_authors_prog %users/;
9 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10 $VERSION = '@@GIT_VERSION@@';
11
12 # From which subdir have we been invoked?
13 my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15 } || '';
16
17 my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
18 $ENV{GIT_DIR} ||= '.git';
19 $Git::SVN::default_repo_id = 'svn';
20 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
21 $Git::SVN::Ra::_log_window_size = 100;
22 $Git::SVN::_minimize_url = 'unset';
23
24 if (! exists $ENV{SVN_SSH}) {
25 if (exists $ENV{GIT_SSH}) {
26 $ENV{SVN_SSH} = $ENV{GIT_SSH};
27 if ($^O eq 'msys') {
28 $ENV{SVN_SSH} =~ s/\\/\\\\/g;
29 }
30 }
31 }
32
33 $Git::SVN::Log::TZ = $ENV{TZ};
34 $ENV{TZ} = 'UTC';
35 $| = 1; # unbuffer STDOUT
36
37 sub fatal (@) { print STDERR "@_\n"; exit 1 }
38 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
39 require SVN::Ra;
40 require SVN::Delta;
41 if ($SVN::Core::VERSION lt '1.1.0') {
42 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
43 }
44 my $can_compress = eval { require Compress::Zlib; 1};
45 push @Git::SVN::Ra::ISA, 'SVN::Ra';
46 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
47 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
48 use Carp qw/croak/;
49 use Digest::MD5;
50 use IO::File qw//;
51 use File::Basename qw/dirname basename/;
52 use File::Path qw/mkpath/;
53 use File::Spec;
54 use File::Find;
55 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
56 use IPC::Open3;
57 use Git;
58
59 BEGIN {
60 # import functions from Git into our packages, en masse
61 no strict 'refs';
62 foreach (qw/command command_oneline command_noisy command_output_pipe
63 command_input_pipe command_close_pipe
64 command_bidi_pipe command_close_bidi_pipe/) {
65 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
66 Git::SVN::Migration Git::SVN::Log Git::SVN),
67 __PACKAGE__) {
68 *{"${package}::$_"} = \&{"Git::$_"};
69 }
70 }
71 }
72
73 my ($SVN);
74
75 $sha1 = qr/[a-f\d]{40}/;
76 $sha1_short = qr/[a-f\d]{4,40}/;
77 my ($_stdin, $_help, $_edit,
78 $_message, $_file, $_branch_dest,
79 $_template, $_shared,
80 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
81 $_merge, $_strategy, $_dry_run, $_local,
82 $_prefix, $_no_checkout, $_url, $_verbose,
83 $_git_format, $_commit_url, $_tag);
84 $Git::SVN::_follow_parent = 1;
85 $_q ||= 0;
86 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
87 'config-dir=s' => \$Git::SVN::Ra::config_dir,
88 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
89 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
90 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
91 'authors-file|A=s' => \$_authors,
92 'authors-prog=s' => \$_authors_prog,
93 'repack:i' => \$Git::SVN::_repack,
94 'noMetadata' => \$Git::SVN::_no_metadata,
95 'useSvmProps' => \$Git::SVN::_use_svm_props,
96 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
97 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
98 'no-checkout' => \$_no_checkout,
99 'quiet|q+' => \$_q,
100 'repack-flags|repack-args|repack-opts=s' =>
101 \$Git::SVN::_repack_flags,
102 'use-log-author' => \$Git::SVN::_use_log_author,
103 'add-author-from' => \$Git::SVN::_add_author_from,
104 'localtime' => \$Git::SVN::_localtime,
105 %remote_opts );
106
107 my ($_trunk, @_tags, @_branches, $_stdlayout);
108 my %icv;
109 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
110 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
111 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
112 'stdlayout|s' => \$_stdlayout,
113 'minimize-url|m!' => \$Git::SVN::_minimize_url,
114 'no-metadata' => sub { $icv{noMetadata} = 1 },
115 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
116 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
117 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
118 %remote_opts );
119 my %cmt_opts = ( 'edit|e' => \$_edit,
120 'rmdir' => \$SVN::Git::Editor::_rmdir,
121 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
122 'l=i' => \$SVN::Git::Editor::_rename_limit,
123 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
124 );
125
126 my %cmd = (
127 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
128 { 'revision|r=s' => \$_revision,
129 'fetch-all|all' => \$_fetch_all,
130 'parent|p' => \$_fetch_parent,
131 %fc_opts } ],
132 clone => [ \&cmd_clone, "Initialize and fetch revisions",
133 { 'revision|r=s' => \$_revision,
134 %fc_opts, %init_opts } ],
135 init => [ \&cmd_init, "Initialize a repo for tracking" .
136 " (requires URL argument)",
137 \%init_opts ],
138 'multi-init' => [ \&cmd_multi_init,
139 "Deprecated alias for ".
140 "'$0 init -T<trunk> -b<branches> -t<tags>'",
141 \%init_opts ],
142 dcommit => [ \&cmd_dcommit,
143 'Commit several diffs to merge with upstream',
144 { 'merge|m|M' => \$_merge,
145 'strategy|s=s' => \$_strategy,
146 'verbose|v' => \$_verbose,
147 'dry-run|n' => \$_dry_run,
148 'fetch-all|all' => \$_fetch_all,
149 'commit-url=s' => \$_commit_url,
150 'revision|r=i' => \$_revision,
151 'no-rebase' => \$_no_rebase,
152 %cmt_opts, %fc_opts } ],
153 branch => [ \&cmd_branch,
154 'Create a branch in the SVN repository',
155 { 'message|m=s' => \$_message,
156 'destination|d=s' => \$_branch_dest,
157 'dry-run|n' => \$_dry_run,
158 'tag|t' => \$_tag } ],
159 tag => [ sub { $_tag = 1; cmd_branch(@_) },
160 'Create a tag in the SVN repository',
161 { 'message|m=s' => \$_message,
162 'destination|d=s' => \$_branch_dest,
163 'dry-run|n' => \$_dry_run } ],
164 'set-tree' => [ \&cmd_set_tree,
165 "Set an SVN repository to a git tree-ish",
166 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
167 'create-ignore' => [ \&cmd_create_ignore,
168 'Create a .gitignore per svn:ignore',
169 { 'revision|r=i' => \$_revision
170 } ],
171 'mkdirs' => [ \&cmd_mkdirs ,
172 "recreate empty directories after a checkout",
173 { 'revision|r=i' => \$_revision } ],
174 'propget' => [ \&cmd_propget,
175 'Print the value of a property on a file or directory',
176 { 'revision|r=i' => \$_revision } ],
177 'proplist' => [ \&cmd_proplist,
178 'List all properties of a file or directory',
179 { 'revision|r=i' => \$_revision } ],
180 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
181 { 'revision|r=i' => \$_revision
182 } ],
183 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
184 { 'revision|r=i' => \$_revision
185 } ],
186 'multi-fetch' => [ \&cmd_multi_fetch,
187 "Deprecated alias for $0 fetch --all",
188 { 'revision|r=s' => \$_revision, %fc_opts } ],
189 'migrate' => [ sub { },
190 # no-op, we automatically run this anyways,
191 'Migrate configuration/metadata/layout from
192 previous versions of git-svn',
193 { 'minimize' => \$Git::SVN::Migration::_minimize,
194 %remote_opts } ],
195 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
196 { 'limit=i' => \$Git::SVN::Log::limit,
197 'revision|r=s' => \$_revision,
198 'verbose|v' => \$Git::SVN::Log::verbose,
199 'incremental' => \$Git::SVN::Log::incremental,
200 'oneline' => \$Git::SVN::Log::oneline,
201 'show-commit' => \$Git::SVN::Log::show_commit,
202 'non-recursive' => \$Git::SVN::Log::non_recursive,
203 'authors-file|A=s' => \$_authors,
204 'color' => \$Git::SVN::Log::color,
205 'pager=s' => \$Git::SVN::Log::pager
206 } ],
207 'find-rev' => [ \&cmd_find_rev,
208 "Translate between SVN revision numbers and tree-ish",
209 {} ],
210 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
211 { 'merge|m|M' => \$_merge,
212 'verbose|v' => \$_verbose,
213 'strategy|s=s' => \$_strategy,
214 'local|l' => \$_local,
215 'fetch-all|all' => \$_fetch_all,
216 'dry-run|n' => \$_dry_run,
217 %fc_opts } ],
218 'commit-diff' => [ \&cmd_commit_diff,
219 'Commit a diff between two trees',
220 { 'message|m=s' => \$_message,
221 'file|F=s' => \$_file,
222 'revision|r=s' => \$_revision,
223 %cmt_opts } ],
224 'info' => [ \&cmd_info,
225 "Show info about the latest SVN revision
226 on the current branch",
227 { 'url' => \$_url, } ],
228 'blame' => [ \&Git::SVN::Log::cmd_blame,
229 "Show what revision and author last modified each line of a file",
230 { 'git-format' => \$_git_format } ],
231 'reset' => [ \&cmd_reset,
232 "Undo fetches back to the specified SVN revision",
233 { 'revision|r=s' => \$_revision,
234 'parent|p' => \$_fetch_parent } ],
235 'gc' => [ \&cmd_gc,
236 "Compress unhandled.log files in .git/svn and remove " .
237 "index files in .git/svn",
238 {} ],
239 );
240
241 my $cmd;
242 for (my $i = 0; $i < @ARGV; $i++) {
243 if (defined $cmd{$ARGV[$i]}) {
244 $cmd = $ARGV[$i];
245 splice @ARGV, $i, 1;
246 last;
247 } elsif ($ARGV[$i] eq 'help') {
248 $cmd = $ARGV[$i+1];
249 usage(0);
250 }
251 };
252
253 # make sure we're always running at the top-level working directory
254 unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
255 unless (-d $ENV{GIT_DIR}) {
256 if ($git_dir_user_set) {
257 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
258 "but it is not a directory\n";
259 }
260 my $git_dir = delete $ENV{GIT_DIR};
261 my $cdup = undef;
262 git_cmd_try {
263 $cdup = command_oneline(qw/rev-parse --show-cdup/);
264 $git_dir = '.' unless ($cdup);
265 chomp $cdup if ($cdup);
266 $cdup = "." unless ($cdup && length $cdup);
267 } "Already at toplevel, but $git_dir not found\n";
268 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
269 unless (-d $git_dir) {
270 die "$git_dir still not found after going to ",
271 "'$cdup'\n";
272 }
273 $ENV{GIT_DIR} = $git_dir;
274 }
275 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
276 }
277
278 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
279
280 read_git_config(\%opts);
281 if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
282 Getopt::Long::Configure('pass_through');
283 }
284 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
285 'minimize-connections' => \$Git::SVN::Migration::_minimize,
286 'id|i=s' => \$Git::SVN::default_ref_id,
287 'svn-remote|remote|R=s' => sub {
288 $Git::SVN::no_reuse_existing = 1;
289 $Git::SVN::default_repo_id = $_[1] });
290 exit 1 if (!$rv && $cmd && $cmd ne 'log');
291
292 usage(0) if $_help;
293 version() if $_version;
294 usage(1) unless defined $cmd;
295 load_authors() if $_authors;
296 if (defined $_authors_prog) {
297 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
298 }
299
300 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
301 Git::SVN::Migration::migration_check();
302 }
303 Git::SVN::init_vars();
304 eval {
305 Git::SVN::verify_remotes_sanity();
306 $cmd{$cmd}->[0]->(@ARGV);
307 };
308 fatal $@ if $@;
309 post_fetch_checkout();
310 exit 0;
311
312 ####################### primary functions ######################
313 sub usage {
314 my $exit = shift || 0;
315 my $fd = $exit ? \*STDERR : \*STDOUT;
316 print $fd <<"";
317 git-svn - bidirectional operations between a single Subversion tree and git
318 Usage: git svn <command> [options] [arguments]\n
319
320 print $fd "Available commands:\n" unless $cmd;
321
322 foreach (sort keys %cmd) {
323 next if $cmd && $cmd ne $_;
324 next if /^multi-/; # don't show deprecated commands
325 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
326 foreach (sort keys %{$cmd{$_}->[2]}) {
327 # mixed-case options are for .git/config only
328 next if /[A-Z]/ && /^[a-z]+$/i;
329 # prints out arguments as they should be passed:
330 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
331 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
332 "--$_" : "-$_" }
333 split /\|/,$_)," $x\n";
334 }
335 }
336 print $fd <<"";
337 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
338 arbitrary identifier if you're tracking multiple SVN branches/repositories in
339 one git repository and want to keep them separate. See git-svn(1) for more
340 information.
341
342 exit $exit;
343 }
344
345 sub version {
346 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
347 exit 0;
348 }
349
350 sub do_git_init_db {
351 unless (-d $ENV{GIT_DIR}) {
352 my @init_db = ('init');
353 push @init_db, "--template=$_template" if defined $_template;
354 if (defined $_shared) {
355 if ($_shared =~ /[a-z]/) {
356 push @init_db, "--shared=$_shared";
357 } else {
358 push @init_db, "--shared";
359 }
360 }
361 command_noisy(@init_db);
362 $_repository = Git->repository(Repository => ".git");
363 }
364 command_noisy('config', 'core.autocrlf', 'false');
365 my $set;
366 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
367 foreach my $i (keys %icv) {
368 die "'$set' and '$i' cannot both be set\n" if $set;
369 next unless defined $icv{$i};
370 command_noisy('config', "$pfx.$i", $icv{$i});
371 $set = $i;
372 }
373 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
374 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
375 if defined $$ignore_regex;
376 }
377
378 sub init_subdir {
379 my $repo_path = shift or return;
380 mkpath([$repo_path]) unless -d $repo_path;
381 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
382 $ENV{GIT_DIR} = '.git';
383 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
384 }
385
386 sub cmd_clone {
387 my ($url, $path) = @_;
388 if (!defined $path &&
389 (defined $_trunk || @_branches || @_tags ||
390 defined $_stdlayout) &&
391 $url !~ m#^[a-z\+]+://#) {
392 $path = $url;
393 }
394 $path = basename($url) if !defined $path || !length $path;
395 cmd_init($url, $path);
396 Git::SVN::fetch_all($Git::SVN::default_repo_id);
397 command_oneline('config', 'svn.authorsfile', $_authors) if $_authors;
398 }
399
400 sub cmd_init {
401 if (defined $_stdlayout) {
402 $_trunk = 'trunk' if (!defined $_trunk);
403 @_tags = 'tags' if (! @_tags);
404 @_branches = 'branches' if (! @_branches);
405 }
406 if (defined $_trunk || @_branches || @_tags) {
407 return cmd_multi_init(@_);
408 }
409 my $url = shift or die "SVN repository location required ",
410 "as a command-line argument\n";
411 $url = canonicalize_url($url);
412 init_subdir(@_);
413 do_git_init_db();
414
415 if ($Git::SVN::_minimize_url eq 'unset') {
416 $Git::SVN::_minimize_url = 0;
417 }
418
419 Git::SVN->init($url);
420 }
421
422 sub cmd_fetch {
423 if (grep /^\d+=./, @_) {
424 die "'<rev>=<commit>' fetch arguments are ",
425 "no longer supported.\n";
426 }
427 my ($remote) = @_;
428 if (@_ > 1) {
429 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
430 }
431 if ($_fetch_parent) {
432 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
433 unless ($gs) {
434 die "Unable to determine upstream SVN information from ",
435 "working tree history\n";
436 }
437 # just fetch, don't checkout.
438 $_no_checkout = 'true';
439 $_fetch_all ? $gs->fetch_all : $gs->fetch;
440 } elsif ($_fetch_all) {
441 cmd_multi_fetch();
442 } else {
443 $remote ||= $Git::SVN::default_repo_id;
444 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
445 }
446 }
447
448 sub cmd_set_tree {
449 my (@commits) = @_;
450 if ($_stdin || !@commits) {
451 print "Reading from stdin...\n";
452 @commits = ();
453 while (<STDIN>) {
454 if (/\b($sha1_short)\b/o) {
455 unshift @commits, $1;
456 }
457 }
458 }
459 my @revs;
460 foreach my $c (@commits) {
461 my @tmp = command('rev-parse',$c);
462 if (scalar @tmp == 1) {
463 push @revs, $tmp[0];
464 } elsif (scalar @tmp > 1) {
465 push @revs, reverse(command('rev-list',@tmp));
466 } else {
467 fatal "Failed to rev-parse $c";
468 }
469 }
470 my $gs = Git::SVN->new;
471 my ($r_last, $cmt_last) = $gs->last_rev_commit;
472 $gs->fetch;
473 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
474 fatal "There are new revisions that were fetched ",
475 "and need to be merged (or acknowledged) ",
476 "before committing.\nlast rev: $r_last\n",
477 " current: $gs->{last_rev}";
478 }
479 $gs->set_tree($_) foreach @revs;
480 print "Done committing ",scalar @revs," revisions to SVN\n";
481 unlink $gs->{index};
482 }
483
484 sub cmd_dcommit {
485 my $head = shift;
486 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
487 'Cannot dcommit with a dirty index. Commit your changes first, '
488 . "or stash them with `git stash'.\n";
489 $head ||= 'HEAD';
490
491 my $old_head;
492 if ($head ne 'HEAD') {
493 $old_head = eval {
494 command_oneline([qw/symbolic-ref -q HEAD/])
495 };
496 if ($old_head) {
497 $old_head =~ s{^refs/heads/}{};
498 } else {
499 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
500 }
501 command(['checkout', $head], STDERR => 0);
502 }
503
504 my @refs;
505 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
506 unless ($gs) {
507 die "Unable to determine upstream SVN information from ",
508 "$head history.\nPerhaps the repository is empty.";
509 }
510
511 if (defined $_commit_url) {
512 $url = $_commit_url;
513 } else {
514 $url = eval { command_oneline('config', '--get',
515 "svn-remote.$gs->{repo_id}.commiturl") };
516 if (!$url) {
517 $url = $gs->full_url
518 }
519 }
520
521 my $last_rev = $_revision if defined $_revision;
522 if ($url) {
523 print "Committing to $url ...\n";
524 }
525 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
526 if ($_no_rebase && scalar(@$linear_refs) > 1) {
527 warn "Attempting to commit more than one change while ",
528 "--no-rebase is enabled.\n",
529 "If these changes depend on each other, re-running ",
530 "without --no-rebase may be required."
531 }
532 my $expect_url = $url;
533 Git::SVN::remove_username($expect_url);
534 while (1) {
535 my $d = shift @$linear_refs or last;
536 unless (defined $last_rev) {
537 (undef, $last_rev, undef) = cmt_metadata("$d~1");
538 unless (defined $last_rev) {
539 fatal "Unable to extract revision information ",
540 "from commit $d~1";
541 }
542 }
543 if ($_dry_run) {
544 print "diff-tree $d~1 $d\n";
545 } else {
546 my $cmt_rev;
547 my %ed_opts = ( r => $last_rev,
548 log => get_commit_entry($d)->{log},
549 ra => Git::SVN::Ra->new($url),
550 config => SVN::Core::config_get_config(
551 $Git::SVN::Ra::config_dir
552 ),
553 tree_a => "$d~1",
554 tree_b => $d,
555 editor_cb => sub {
556 print "Committed r$_[0]\n";
557 $cmt_rev = $_[0];
558 },
559 svn_path => '');
560 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
561 print "No changes\n$d~1 == $d\n";
562 } elsif ($parents->{$d} && @{$parents->{$d}}) {
563 $gs->{inject_parents_dcommit}->{$cmt_rev} =
564 $parents->{$d};
565 }
566 $_fetch_all ? $gs->fetch_all : $gs->fetch;
567 $last_rev = $cmt_rev;
568 next if $_no_rebase;
569
570 # we always want to rebase against the current HEAD,
571 # not any head that was passed to us
572 my @diff = command('diff-tree', $d,
573 $gs->refname, '--');
574 my @finish;
575 if (@diff) {
576 @finish = rebase_cmd();
577 print STDERR "W: $d and ", $gs->refname,
578 " differ, using @finish:\n",
579 join("\n", @diff), "\n";
580 } else {
581 print "No changes between current HEAD and ",
582 $gs->refname,
583 "\nResetting to the latest ",
584 $gs->refname, "\n";
585 @finish = qw/reset --mixed/;
586 }
587 command_noisy(@finish, $gs->refname);
588 if (@diff) {
589 @refs = ();
590 my ($url_, $rev_, $uuid_, $gs_) =
591 working_head_info('HEAD', \@refs);
592 my ($linear_refs_, $parents_) =
593 linearize_history($gs_, \@refs);
594 if (scalar(@$linear_refs) !=
595 scalar(@$linear_refs_)) {
596 fatal "# of revisions changed ",
597 "\nbefore:\n",
598 join("\n", @$linear_refs),
599 "\n\nafter:\n",
600 join("\n", @$linear_refs_), "\n",
601 'If you are attempting to commit ',
602 "merges, try running:\n\t",
603 'git rebase --interactive',
604 '--preserve-merges ',
605 $gs->refname,
606 "\nBefore dcommitting";
607 }
608 if ($url_ ne $expect_url) {
609 if ($url_ eq $gs->metadata_url) {
610 print
611 "Accepting rewritten URL:",
612 " $url_\n";
613 } else {
614 fatal
615 "URL mismatch after rebase:",
616 " $url_ != $expect_url";
617 }
618 }
619 if ($uuid_ ne $uuid) {
620 fatal "uuid mismatch after rebase: ",
621 "$uuid_ != $uuid";
622 }
623 # remap parents
624 my (%p, @l, $i);
625 for ($i = 0; $i < scalar @$linear_refs; $i++) {
626 my $new = $linear_refs_->[$i] or next;
627 $p{$new} =
628 $parents->{$linear_refs->[$i]};
629 push @l, $new;
630 }
631 $parents = \%p;
632 $linear_refs = \@l;
633 }
634 }
635 }
636
637 if ($old_head) {
638 my $new_head = command_oneline(qw/rev-parse HEAD/);
639 my $new_is_symbolic = eval {
640 command_oneline(qw/symbolic-ref -q HEAD/);
641 };
642 if ($new_is_symbolic) {
643 print "dcommitted the branch ", $head, "\n";
644 } else {
645 print "dcommitted on a detached HEAD because you gave ",
646 "a revision argument.\n",
647 "The rewritten commit is: ", $new_head, "\n";
648 }
649 command(['checkout', $old_head], STDERR => 0);
650 }
651
652 unlink $gs->{index};
653 }
654
655 sub cmd_branch {
656 my ($branch_name, $head) = @_;
657
658 unless (defined $branch_name && length $branch_name) {
659 die(($_tag ? "tag" : "branch") . " name required\n");
660 }
661 $head ||= 'HEAD';
662
663 my ($src, $rev, undef, $gs) = working_head_info($head);
664
665 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
666 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
667 my $glob;
668 if ($#{$allglobs} == 0) {
669 $glob = $allglobs->[0];
670 } else {
671 unless(defined $_branch_dest) {
672 die "Multiple ",
673 $_tag ? "tag" : "branch",
674 " paths defined for Subversion repository.\n",
675 "You must specify where you want to create the ",
676 $_tag ? "tag" : "branch",
677 " with the --destination argument.\n";
678 }
679 foreach my $g (@{$allglobs}) {
680 # SVN::Git::Editor could probably be moved to Git.pm..
681 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
682 if ($_branch_dest =~ /$re/) {
683 $glob = $g;
684 last;
685 }
686 }
687 unless (defined $glob) {
688 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
689 foreach my $g (@{$allglobs}) {
690 $g->{path}->{left} =~ /$dest_re/ or next;
691 if (defined $glob) {
692 die "Ambiguous destination: ",
693 $_branch_dest, "\nmatches both '",
694 $glob->{path}->{left}, "' and '",
695 $g->{path}->{left}, "'\n";
696 }
697 $glob = $g;
698 }
699 unless (defined $glob) {
700 die "Unknown ",
701 $_tag ? "tag" : "branch",
702 " destination $_branch_dest\n";
703 }
704 }
705 }
706 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
707 my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
708
709 my $ctx = SVN::Client->new(
710 auth => Git::SVN::Ra::_auth_providers(),
711 log_msg => sub {
712 ${ $_[0] } = defined $_message
713 ? $_message
714 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
715 . $branch_name;
716 },
717 );
718
719 eval {
720 $ctx->ls($dst, 'HEAD', 0);
721 } and die "branch ${branch_name} already exists\n";
722
723 print "Copying ${src} at r${rev} to ${dst}...\n";
724 $ctx->copy($src, $rev, $dst)
725 unless $_dry_run;
726
727 $gs->fetch_all;
728 }
729
730 sub cmd_find_rev {
731 my $revision_or_hash = shift or die "SVN or git revision required ",
732 "as a command-line argument\n";
733 my $result;
734 if ($revision_or_hash =~ /^r\d+$/) {
735 my $head = shift;
736 $head ||= 'HEAD';
737 my @refs;
738 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
739 unless ($gs) {
740 die "Unable to determine upstream SVN information from ",
741 "$head history\n";
742 }
743 my $desired_revision = substr($revision_or_hash, 1);
744 $result = $gs->rev_map_get($desired_revision, $uuid);
745 } else {
746 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
747 $result = $rev;
748 }
749 print "$result\n" if $result;
750 }
751
752 sub cmd_rebase {
753 command_noisy(qw/update-index --refresh/);
754 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
755 unless ($gs) {
756 die "Unable to determine upstream SVN information from ",
757 "working tree history\n";
758 }
759 if ($_dry_run) {
760 print "Remote Branch: " . $gs->refname . "\n";
761 print "SVN URL: " . $url . "\n";
762 return;
763 }
764 if (command(qw/diff-index HEAD --/)) {
765 print STDERR "Cannot rebase with uncommited changes:\n";
766 command_noisy('status');
767 exit 1;
768 }
769 unless ($_local) {
770 # rebase will checkout for us, so no need to do it explicitly
771 $_no_checkout = 'true';
772 $_fetch_all ? $gs->fetch_all : $gs->fetch;
773 }
774 command_noisy(rebase_cmd(), $gs->refname);
775 $gs->mkemptydirs;
776 }
777
778 sub cmd_show_ignore {
779 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
780 $gs ||= Git::SVN->new;
781 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
782 $gs->prop_walk($gs->{path}, $r, sub {
783 my ($gs, $path, $props) = @_;
784 print STDOUT "\n# $path\n";
785 my $s = $props->{'svn:ignore'} or return;
786 $s =~ s/[\r\n]+/\n/g;
787 $s =~ s/^\n+//;
788 chomp $s;
789 $s =~ s#^#$path#gm;
790 print STDOUT "$s\n";
791 });
792 }
793
794 sub cmd_show_externals {
795 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
796 $gs ||= Git::SVN->new;
797 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
798 $gs->prop_walk($gs->{path}, $r, sub {
799 my ($gs, $path, $props) = @_;
800 print STDOUT "\n# $path\n";
801 my $s = $props->{'svn:externals'} or return;
802 $s =~ s/[\r\n]+/\n/g;
803 chomp $s;
804 $s =~ s#^#$path#gm;
805 print STDOUT "$s\n";
806 });
807 }
808
809 sub cmd_create_ignore {
810 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
811 $gs ||= Git::SVN->new;
812 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
813 $gs->prop_walk($gs->{path}, $r, sub {
814 my ($gs, $path, $props) = @_;
815 # $path is of the form /path/to/dir/
816 $path = '.' . $path;
817 # SVN can have attributes on empty directories,
818 # which git won't track
819 mkpath([$path]) unless -d $path;
820 my $ignore = $path . '.gitignore';
821 my $s = $props->{'svn:ignore'} or return;
822 open(GITIGNORE, '>', $ignore)
823 or fatal("Failed to open `$ignore' for writing: $!");
824 $s =~ s/[\r\n]+/\n/g;
825 $s =~ s/^\n+//;
826 chomp $s;
827 # Prefix all patterns so that the ignore doesn't apply
828 # to sub-directories.
829 $s =~ s#^#/#gm;
830 print GITIGNORE "$s\n";
831 close(GITIGNORE)
832 or fatal("Failed to close `$ignore': $!");
833 command_noisy('add', '-f', $ignore);
834 });
835 }
836
837 sub cmd_mkdirs {
838 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
839 $gs ||= Git::SVN->new;
840 $gs->mkemptydirs($_revision);
841 }
842
843 sub canonicalize_path {
844 my ($path) = @_;
845 my $dot_slash_added = 0;
846 if (substr($path, 0, 1) ne "/") {
847 $path = "./" . $path;
848 $dot_slash_added = 1;
849 }
850 # File::Spec->canonpath doesn't collapse x/../y into y (for a
851 # good reason), so let's do this manually.
852 $path =~ s#/+#/#g;
853 $path =~ s#/\.(?:/|$)#/#g;
854 $path =~ s#/[^/]+/\.\.##g;
855 $path =~ s#/$##g;
856 $path =~ s#^\./## if $dot_slash_added;
857 $path =~ s#^/##;
858 $path =~ s#^\.$##;
859 return $path;
860 }
861
862 sub canonicalize_url {
863 my ($url) = @_;
864 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
865 return $url;
866 }
867
868 # get_svnprops(PATH)
869 # ------------------
870 # Helper for cmd_propget and cmd_proplist below.
871 sub get_svnprops {
872 my $path = shift;
873 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
874 $gs ||= Git::SVN->new;
875
876 # prefix THE PATH by the sub-directory from which the user
877 # invoked us.
878 $path = $cmd_dir_prefix . $path;
879 fatal("No such file or directory: $path") unless -e $path;
880 my $is_dir = -d $path ? 1 : 0;
881 $path = $gs->{path} . '/' . $path;
882
883 # canonicalize the path (otherwise libsvn will abort or fail to
884 # find the file)
885 $path = canonicalize_path($path);
886
887 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
888 my $props;
889 if ($is_dir) {
890 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
891 }
892 else {
893 (undef, $props) = $gs->ra->get_file($path, $r, undef);
894 }
895 return $props;
896 }
897
898 # cmd_propget (PROP, PATH)
899 # ------------------------
900 # Print the SVN property PROP for PATH.
901 sub cmd_propget {
902 my ($prop, $path) = @_;
903 $path = '.' if not defined $path;
904 usage(1) if not defined $prop;
905 my $props = get_svnprops($path);
906 if (not defined $props->{$prop}) {
907 fatal("`$path' does not have a `$prop' SVN property.");
908 }
909 print $props->{$prop} . "\n";
910 }
911
912 # cmd_proplist (PATH)
913 # -------------------
914 # Print the list of SVN properties for PATH.
915 sub cmd_proplist {
916 my $path = shift;
917 $path = '.' if not defined $path;
918 my $props = get_svnprops($path);
919 print "Properties on '$path':\n";
920 foreach (sort keys %{$props}) {
921 print " $_\n";
922 }
923 }
924
925 sub cmd_multi_init {
926 my $url = shift;
927 unless (defined $_trunk || @_branches || @_tags) {
928 usage(1);
929 }
930
931 $_prefix = '' unless defined $_prefix;
932 if (defined $url) {
933 $url = canonicalize_url($url);
934 init_subdir(@_);
935 }
936 do_git_init_db();
937 if (defined $_trunk) {
938 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
939 # try both old-style and new-style lookups:
940 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
941 unless ($gs_trunk) {
942 my ($trunk_url, $trunk_path) =
943 complete_svn_url($url, $_trunk);
944 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
945 undef, $trunk_ref);
946 }
947 }
948 return unless @_branches || @_tags;
949 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
950 foreach my $path (@_branches) {
951 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
952 }
953 foreach my $path (@_tags) {
954 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
955 }
956 }
957
958 sub cmd_multi_fetch {
959 my $remotes = Git::SVN::read_all_remotes();
960 foreach my $repo_id (sort keys %$remotes) {
961 if ($remotes->{$repo_id}->{url}) {
962 Git::SVN::fetch_all($repo_id, $remotes);
963 }
964 }
965 }
966
967 # this command is special because it requires no metadata
968 sub cmd_commit_diff {
969 my ($ta, $tb, $url) = @_;
970 my $usage = "Usage: $0 commit-diff -r<revision> ".
971 "<tree-ish> <tree-ish> [<URL>]";
972 fatal($usage) if (!defined $ta || !defined $tb);
973 my $svn_path = '';
974 if (!defined $url) {
975 my $gs = eval { Git::SVN->new };
976 if (!$gs) {
977 fatal("Needed URL or usable git-svn --id in ",
978 "the command-line\n", $usage);
979 }
980 $url = $gs->{url};
981 $svn_path = $gs->{path};
982 }
983 unless (defined $_revision) {
984 fatal("-r|--revision is a required argument\n", $usage);
985 }
986 if (defined $_message && defined $_file) {
987 fatal("Both --message/-m and --file/-F specified ",
988 "for the commit message.\n",
989 "I have no idea what you mean");
990 }
991 if (defined $_file) {
992 $_message = file_to_s($_file);
993 } else {
994 $_message ||= get_commit_entry($tb)->{log};
995 }
996 my $ra ||= Git::SVN::Ra->new($url);
997 my $r = $_revision;
998 if ($r eq 'HEAD') {
999 $r = $ra->get_latest_revnum;
1000 } elsif ($r !~ /^\d+$/) {
1001 die "revision argument: $r not understood by git-svn\n";
1002 }
1003 my %ed_opts = ( r => $r,
1004 log => $_message,
1005 ra => $ra,
1006 tree_a => $ta,
1007 tree_b => $tb,
1008 editor_cb => sub { print "Committed r$_[0]\n" },
1009 svn_path => $svn_path );
1010 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1011 print "No changes\n$ta == $tb\n";
1012 }
1013 }
1014
1015 sub escape_uri_only {
1016 my ($uri) = @_;
1017 my @tmp;
1018 foreach (split m{/}, $uri) {
1019 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1020 push @tmp, $_;
1021 }
1022 join('/', @tmp);
1023 }
1024
1025 sub escape_url {
1026 my ($url) = @_;
1027 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1028 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1029 $url = "$scheme://$domain$uri";
1030 }
1031 $url;
1032 }
1033
1034 sub cmd_info {
1035 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1036 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1037 if (exists $_[1]) {
1038 die "Too many arguments specified\n";
1039 }
1040
1041 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1042
1043 if (!$file_type && !$diff_status) {
1044 print STDERR "svn: '$path' is not under version control\n";
1045 exit 1;
1046 }
1047
1048 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1049 unless ($gs) {
1050 die "Unable to determine upstream SVN information from ",
1051 "working tree history\n";
1052 }
1053
1054 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1055 $path = "." if $path eq "";
1056
1057 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1058
1059 if ($_url) {
1060 print escape_url($full_url), "\n";
1061 return;
1062 }
1063
1064 my $result = "Path: $path\n";
1065 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1066 $result .= "URL: " . escape_url($full_url) . "\n";
1067
1068 eval {
1069 my $repos_root = $gs->repos_root;
1070 Git::SVN::remove_username($repos_root);
1071 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1072 };
1073 if ($@) {
1074 $result .= "Repository Root: (offline)\n";
1075 }
1076 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1077 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1078 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1079
1080 $result .= "Node Kind: " .
1081 ($file_type eq "dir" ? "directory" : "file") . "\n";
1082
1083 my $schedule = $diff_status eq "A"
1084 ? "add"
1085 : ($diff_status eq "D" ? "delete" : "normal");
1086 $result .= "Schedule: $schedule\n";
1087
1088 if ($diff_status eq "A") {
1089 print $result, "\n";
1090 return;
1091 }
1092
1093 my ($lc_author, $lc_rev, $lc_date_utc);
1094 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1095 my $log = command_output_pipe(@args);
1096 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1097 while (<$log>) {
1098 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1099 $lc_author = $1;
1100 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1101 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1102 (undef, $lc_rev, undef) = ::extract_metadata($1);
1103 }
1104 }
1105 close $log;
1106
1107 Git::SVN::Log::set_local_timezone();
1108
1109 $result .= "Last Changed Author: $lc_author\n";
1110 $result .= "Last Changed Rev: $lc_rev\n";
1111 $result .= "Last Changed Date: " .
1112 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1113
1114 if ($file_type ne "dir") {
1115 my $text_last_updated_date =
1116 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1117 $result .=
1118 "Text Last Updated: " .
1119 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1120 "\n";
1121 my $checksum;
1122 if ($diff_status eq "D") {
1123 my ($fh, $ctx) =
1124 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1125 if ($file_type eq "link") {
1126 my $file_name = <$fh>;
1127 $checksum = md5sum("link $file_name");
1128 } else {
1129 $checksum = md5sum($fh);
1130 }
1131 command_close_pipe($fh, $ctx);
1132 } elsif ($file_type eq "link") {
1133 my $file_name =
1134 command(qw(cat-file blob), "HEAD:$path");
1135 $checksum =
1136 md5sum("link " . $file_name);
1137 } else {
1138 open FILE, "<", $path or die $!;
1139 $checksum = md5sum(\*FILE);
1140 close FILE or die $!;
1141 }
1142 $result .= "Checksum: " . $checksum . "\n";
1143 }
1144
1145 print $result, "\n";
1146 }
1147
1148 sub cmd_reset {
1149 my $target = shift || $_revision or die "SVN revision required\n";
1150 $target = $1 if $target =~ /^r(\d+)$/;
1151 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1152 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1153 unless ($gs) {
1154 die "Unable to determine upstream SVN information from ".
1155 "history\n";
1156 }
1157 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1158 $gs->rev_map_set($r, $c, 'reset', $uuid);
1159 print "r$r = $c ($gs->{ref_id})\n";
1160 }
1161
1162 sub cmd_gc {
1163 if (!$can_compress) {
1164 warn "Compress::Zlib could not be found; unhandled.log " .
1165 "files will not be compressed.\n";
1166 }
1167 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1168 }
1169
1170 ########################### utility functions #########################
1171
1172 sub rebase_cmd {
1173 my @cmd = qw/rebase/;
1174 push @cmd, '-v' if $_verbose;
1175 push @cmd, qw/--merge/ if $_merge;
1176 push @cmd, "--strategy=$_strategy" if $_strategy;
1177 @cmd;
1178 }
1179
1180 sub post_fetch_checkout {
1181 return if $_no_checkout;
1182 my $gs = $Git::SVN::_head or return;
1183 return if verify_ref('refs/heads/master^0');
1184
1185 # look for "trunk" ref if it exists
1186 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1187 my $fetch = $remote->{fetch};
1188 if ($fetch) {
1189 foreach my $p (keys %$fetch) {
1190 basename($fetch->{$p}) eq 'trunk' or next;
1191 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1192 last;
1193 }
1194 }
1195
1196 my $valid_head = verify_ref('HEAD^0');
1197 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1198 return if ($valid_head || !verify_ref('HEAD^0'));
1199
1200 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1201 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1202 return if -f $index;
1203
1204 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1205 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1206 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1207 print STDERR "Checked out HEAD:\n ",
1208 $gs->full_url, " r", $gs->last_rev, "\n";
1209 $gs->mkemptydirs($gs->last_rev);
1210 }
1211
1212 sub complete_svn_url {
1213 my ($url, $path) = @_;
1214 $path =~ s#/+$##;
1215 if ($path !~ m#^[a-z\+]+://#) {
1216 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1217 fatal("E: '$path' is not a complete URL ",
1218 "and a separate URL is not specified");
1219 }
1220 return ($url, $path);
1221 }
1222 return ($path, '');
1223 }
1224
1225 sub complete_url_ls_init {
1226 my ($ra, $repo_path, $switch, $pfx) = @_;
1227 unless ($repo_path) {
1228 print STDERR "W: $switch not specified\n";
1229 return;
1230 }
1231 $repo_path =~ s#/+$##;
1232 if ($repo_path =~ m#^[a-z\+]+://#) {
1233 $ra = Git::SVN::Ra->new($repo_path);
1234 $repo_path = '';
1235 } else {
1236 $repo_path =~ s#^/+##;
1237 unless ($ra) {
1238 fatal("E: '$repo_path' is not a complete URL ",
1239 "and a separate URL is not specified");
1240 }
1241 }
1242 my $url = $ra->{url};
1243 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1244 my $k = "svn-remote.$gs->{repo_id}.url";
1245 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1246 if ($orig_url && ($orig_url ne $gs->{url})) {
1247 die "$k already set: $orig_url\n",
1248 "wanted to set to: $gs->{url}\n";
1249 }
1250 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1251 my $remote_path = "$gs->{path}/$repo_path";
1252 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1253 $remote_path =~ s#/+#/#g;
1254 $remote_path =~ s#^/##g;
1255 $remote_path .= "/*" if $remote_path !~ /\*/;
1256 my ($n) = ($switch =~ /^--(\w+)/);
1257 if (length $pfx && $pfx !~ m#/$#) {
1258 die "--prefix='$pfx' must have a trailing slash '/'\n";
1259 }
1260 command_noisy('config',
1261 '--add',
1262 "svn-remote.$gs->{repo_id}.$n",
1263 "$remote_path:refs/remotes/$pfx*" .
1264 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1265 }
1266
1267 sub verify_ref {
1268 my ($ref) = @_;
1269 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1270 { STDERR => 0 }); };
1271 }
1272
1273 sub get_tree_from_treeish {
1274 my ($treeish) = @_;
1275 # $treeish can be a symbolic ref, too:
1276 my $type = command_oneline(qw/cat-file -t/, $treeish);
1277 my $expected;
1278 while ($type eq 'tag') {
1279 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1280 }
1281 if ($type eq 'commit') {
1282 $expected = (grep /^tree /, command(qw/cat-file commit/,
1283 $treeish))[0];
1284 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1285 die "Unable to get tree from $treeish\n" unless $expected;
1286 } elsif ($type eq 'tree') {
1287 $expected = $treeish;
1288 } else {
1289 die "$treeish is a $type, expected tree, tag or commit\n";
1290 }
1291 return $expected;
1292 }
1293
1294 sub get_commit_entry {
1295 my ($treeish) = shift;
1296 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1297 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1298 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1299 open my $log_fh, '>', $commit_editmsg or croak $!;
1300
1301 my $type = command_oneline(qw/cat-file -t/, $treeish);
1302 if ($type eq 'commit' || $type eq 'tag') {
1303 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1304 $type, $treeish);
1305 my $in_msg = 0;
1306 my $author;
1307 my $saw_from = 0;
1308 my $msgbuf = "";
1309 while (<$msg_fh>) {
1310 if (!$in_msg) {
1311 $in_msg = 1 if (/^\s*$/);
1312 $author = $1 if (/^author (.*>)/);
1313 } elsif (/^git-svn-id: /) {
1314 # skip this for now, we regenerate the
1315 # correct one on re-fetch anyways
1316 # TODO: set *:merge properties or like...
1317 } else {
1318 if (/^From:/ || /^Signed-off-by:/) {
1319 $saw_from = 1;
1320 }
1321 $msgbuf .= $_;
1322 }
1323 }
1324 $msgbuf =~ s/\s+$//s;
1325 if ($Git::SVN::_add_author_from && defined($author)
1326 && !$saw_from) {
1327 $msgbuf .= "\n\nFrom: $author";
1328 }
1329 print $log_fh $msgbuf or croak $!;
1330 command_close_pipe($msg_fh, $ctx);
1331 }
1332 close $log_fh or croak $!;
1333
1334 if ($_edit || ($type eq 'tree')) {
1335 chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1336 system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1337 }
1338 rename $commit_editmsg, $commit_msg or croak $!;
1339 {
1340 require Encode;
1341 # SVN requires messages to be UTF-8 when entering the repo
1342 local $/;
1343 open $log_fh, '<', $commit_msg or croak $!;
1344 binmode $log_fh;
1345 chomp($log_entry{log} = <$log_fh>);
1346
1347 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1348 my $msg = $log_entry{log};
1349
1350 eval { $msg = Encode::decode($enc, $msg, 1) };
1351 if ($@) {
1352 die "Could not decode as $enc:\n", $msg,
1353 "\nPerhaps you need to set i18n.commitencoding\n";
1354 }
1355
1356 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1357 die "Could not encode as UTF-8:\n$msg\n" if $@;
1358
1359 $log_entry{log} = $msg;
1360
1361 close $log_fh or croak $!;
1362 }
1363 unlink $commit_msg;
1364 \%log_entry;
1365 }
1366
1367 sub s_to_file {
1368 my ($str, $file, $mode) = @_;
1369 open my $fd,'>',$file or croak $!;
1370 print $fd $str,"\n" or croak $!;
1371 close $fd or croak $!;
1372 chmod ($mode &~ umask, $file) if (defined $mode);
1373 }
1374
1375 sub file_to_s {
1376 my $file = shift;
1377 open my $fd,'<',$file or croak "$!: file: $file\n";
1378 local $/;
1379 my $ret = <$fd>;
1380 close $fd or croak $!;
1381 $ret =~ s/\s*$//s;
1382 return $ret;
1383 }
1384
1385 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1386 sub load_authors {
1387 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1388 my $log = $cmd eq 'log';
1389 while (<$authors>) {
1390 chomp;
1391 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1392 my ($user, $name, $email) = ($1, $2, $3);
1393 if ($log) {
1394 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1395 } else {
1396 $users{$user} = [$name, $email];
1397 }
1398 }
1399 close $authors or croak $!;
1400 }
1401
1402 # convert GetOpt::Long specs for use by git-config
1403 sub read_git_config {
1404 my $opts = shift;
1405 my @config_only;
1406 foreach my $o (keys %$opts) {
1407 # if we have mixedCase and a long option-only, then
1408 # it's a config-only variable that we don't need for
1409 # the command-line.
1410 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1411 my $v = $opts->{$o};
1412 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1413 $key =~ s/-//g;
1414 my $arg = 'git config';
1415 $arg .= ' --int' if ($o =~ /[:=]i$/);
1416 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1417 if (ref $v eq 'ARRAY') {
1418 chomp(my @tmp = `$arg --get-all svn.$key`);
1419 @$v = @tmp if @tmp;
1420 } else {
1421 chomp(my $tmp = `$arg --get svn.$key`);
1422 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1423 $$v = $tmp;
1424 }
1425 }
1426 }
1427 delete @$opts{@config_only} if @config_only;
1428 }
1429
1430 sub extract_metadata {
1431 my $id = shift or return (undef, undef, undef);
1432 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1433 \s([a-f\d\-]+)$/ix);
1434 if (!defined $rev || !$uuid || !$url) {
1435 # some of the original repositories I made had
1436 # identifiers like this:
1437 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1438 }
1439 return ($url, $rev, $uuid);
1440 }
1441
1442 sub cmt_metadata {
1443 return extract_metadata((grep(/^git-svn-id: /,
1444 command(qw/cat-file commit/, shift)))[-1]);
1445 }
1446
1447 sub cmt_sha2rev_batch {
1448 my %s2r;
1449 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1450 my $list = shift;
1451
1452 foreach my $sha (@{$list}) {
1453 my $first = 1;
1454 my $size = 0;
1455 print $out $sha, "\n";
1456
1457 while (my $line = <$in>) {
1458 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1459 last;
1460 } elsif ($first &&
1461 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1462 $first = 0;
1463 $size = $1;
1464 next;
1465 } elsif ($line =~ /^(git-svn-id: )/) {
1466 my (undef, $rev, undef) =
1467 extract_metadata($line);
1468 $s2r{$sha} = $rev;
1469 }
1470
1471 $size -= length($line);
1472 last if ($size == 0);
1473 }
1474 }
1475
1476 command_close_bidi_pipe($pid, $in, $out, $ctx);
1477
1478 return \%s2r;
1479 }
1480
1481 sub working_head_info {
1482 my ($head, $refs) = @_;
1483 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1484 my ($fh, $ctx) = command_output_pipe(@args, $head);
1485 my $hash;
1486 my %max;
1487 while (<$fh>) {
1488 if ( m{^commit ($::sha1)$} ) {
1489 unshift @$refs, $hash if $hash and $refs;
1490 $hash = $1;
1491 next;
1492 }
1493 next unless s{^\s*(git-svn-id:)}{$1};
1494 my ($url, $rev, $uuid) = extract_metadata($_);
1495 if (defined $url && defined $rev) {
1496 next if $max{$url} and $max{$url} < $rev;
1497 if (my $gs = Git::SVN->find_by_url($url)) {
1498 my $c = $gs->rev_map_get($rev, $uuid);
1499 if ($c && $c eq $hash) {
1500 close $fh; # break the pipe
1501 return ($url, $rev, $uuid, $gs);
1502 } else {
1503 $max{$url} ||= $gs->rev_map_max;
1504 }
1505 }
1506 }
1507 }
1508 command_close_pipe($fh, $ctx);
1509 (undef, undef, undef, undef);
1510 }
1511
1512 sub read_commit_parents {
1513 my ($parents, $c) = @_;
1514 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1515 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1516 @{$parents->{$c}} = split(/ /, $p);
1517 }
1518
1519 sub linearize_history {
1520 my ($gs, $refs) = @_;
1521 my %parents;
1522 foreach my $c (@$refs) {
1523 read_commit_parents(\%parents, $c);
1524 }
1525
1526 my @linear_refs;
1527 my %skip = ();
1528 my $last_svn_commit = $gs->last_commit;
1529 foreach my $c (reverse @$refs) {
1530 next if $c eq $last_svn_commit;
1531 last if $skip{$c};
1532
1533 unshift @linear_refs, $c;
1534 $skip{$c} = 1;
1535
1536 # we only want the first parent to diff against for linear
1537 # history, we save the rest to inject when we finalize the
1538 # svn commit
1539 my $fp_a = verify_ref("$c~1");
1540 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1541 if (!$fp_a || !$fp_b) {
1542 die "Commit $c\n",
1543 "has no parent commit, and therefore ",
1544 "nothing to diff against.\n",
1545 "You should be working from a repository ",
1546 "originally created by git-svn\n";
1547 }
1548 if ($fp_a ne $fp_b) {
1549 die "$c~1 = $fp_a, however parsing commit $c ",
1550 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1551 }
1552
1553 foreach my $p (@{$parents{$c}}) {
1554 $skip{$p} = 1;
1555 }
1556 }
1557 (\@linear_refs, \%parents);
1558 }
1559
1560 sub find_file_type_and_diff_status {
1561 my ($path) = @_;
1562 return ('dir', '') if $path eq '';
1563
1564 my $diff_output =
1565 command_oneline(qw(diff --cached --name-status --), $path) || "";
1566 my $diff_status = (split(' ', $diff_output))[0] || "";
1567
1568 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1569
1570 return (undef, undef) if !$diff_status && !$ls_tree;
1571
1572 if ($diff_status eq "A") {
1573 return ("link", $diff_status) if -l $path;
1574 return ("dir", $diff_status) if -d $path;
1575 return ("file", $diff_status);
1576 }
1577
1578 my $mode = (split(' ', $ls_tree))[0] || "";
1579
1580 return ("link", $diff_status) if $mode eq "120000";
1581 return ("dir", $diff_status) if $mode eq "040000";
1582 return ("file", $diff_status);
1583 }
1584
1585 sub md5sum {
1586 my $arg = shift;
1587 my $ref = ref $arg;
1588 my $md5 = Digest::MD5->new();
1589 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1590 $md5->addfile($arg) or croak $!;
1591 } elsif ($ref eq 'SCALAR') {
1592 $md5->add($$arg) or croak $!;
1593 } elsif (!$ref) {
1594 $md5->add($arg) or croak $!;
1595 } else {
1596 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1597 }
1598 return $md5->hexdigest();
1599 }
1600
1601 sub gc_directory {
1602 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1603 my $out_filename = $_ . ".gz";
1604 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1605 binmode $in_fh;
1606 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1607 die "Unable to open $out_filename: $!\n";
1608
1609 my $res;
1610 while ($res = sysread($in_fh, my $str, 1024)) {
1611 $gz->gzwrite($str) or
1612 die "Unable to write: ".$gz->gzerror()."!\n";
1613 }
1614 unlink $_ or die "unlink $File::Find::name: $!\n";
1615 } elsif (-f $_ && basename($_) eq "index") {
1616 unlink $_ or die "unlink $_: $!\n";
1617 }
1618 }
1619
1620 package Git::SVN;
1621 use strict;
1622 use warnings;
1623 use Fcntl qw/:DEFAULT :seek/;
1624 use constant rev_map_fmt => 'NH40';
1625 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1626 $_repack $_repack_flags $_use_svm_props $_head
1627 $_use_svnsync_props $no_reuse_existing $_minimize_url
1628 $_use_log_author $_add_author_from $_localtime/;
1629 use Carp qw/croak/;
1630 use File::Path qw/mkpath/;
1631 use File::Copy qw/copy/;
1632 use IPC::Open3;
1633
1634 my ($_gc_nr, $_gc_period);
1635
1636 # properties that we do not log:
1637 my %SKIP_PROP;
1638 BEGIN {
1639 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1640 svn:special svn:executable
1641 svn:entry:committed-rev
1642 svn:entry:last-author
1643 svn:entry:uuid
1644 svn:entry:committed-date/;
1645
1646 # some options are read globally, but can be overridden locally
1647 # per [svn-remote "..."] section. Command-line options will *NOT*
1648 # override options set in an [svn-remote "..."] section
1649 no strict 'refs';
1650 for my $option (qw/follow_parent no_metadata use_svm_props
1651 use_svnsync_props/) {
1652 my $key = $option;
1653 $key =~ tr/_//d;
1654 my $prop = "-$option";
1655 *$option = sub {
1656 my ($self) = @_;
1657 return $self->{$prop} if exists $self->{$prop};
1658 my $k = "svn-remote.$self->{repo_id}.$key";
1659 eval { command_oneline(qw/config --get/, $k) };
1660 if ($@) {
1661 $self->{$prop} = ${"Git::SVN::_$option"};
1662 } else {
1663 my $v = command_oneline(qw/config --bool/,$k);
1664 $self->{$prop} = $v eq 'false' ? 0 : 1;
1665 }
1666 return $self->{$prop};
1667 }
1668 }
1669 }
1670
1671
1672 my (%LOCKFILES, %INDEX_FILES);
1673 END {
1674 unlink keys %LOCKFILES if %LOCKFILES;
1675 unlink keys %INDEX_FILES if %INDEX_FILES;
1676 }
1677
1678 sub resolve_local_globs {
1679 my ($url, $fetch, $glob_spec) = @_;
1680 return unless defined $glob_spec;
1681 my $ref = $glob_spec->{ref};
1682 my $path = $glob_spec->{path};
1683 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1684 next unless m#^$ref->{regex}$#;
1685 my $p = $1;
1686 my $pathname = desanitize_refname($path->full_path($p));
1687 my $refname = desanitize_refname($ref->full_path($p));
1688 if (my $existing = $fetch->{$pathname}) {
1689 if ($existing ne $refname) {
1690 die "Refspec conflict:\n",
1691 "existing: $existing\n",
1692 " globbed: $refname\n";
1693 }
1694 my $u = (::cmt_metadata("$refname"))[0];
1695 $u =~ s!^\Q$url\E(/|$)!! or die
1696 "$refname: '$url' not found in '$u'\n";
1697 if ($pathname ne $u) {
1698 warn "W: Refspec glob conflict ",
1699 "(ref: $refname):\n",
1700 "expected path: $pathname\n",
1701 " real path: $u\n",
1702 "Continuing ahead with $u\n";
1703 next;
1704 }
1705 } else {
1706 $fetch->{$pathname} = $refname;
1707 }
1708 }
1709 }
1710
1711 sub parse_revision_argument {
1712 my ($base, $head) = @_;
1713 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1714 return ($base, $head);
1715 }
1716 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1717 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1718 return ($head, $head) if ($::_revision eq 'HEAD');
1719 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1720 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1721 die "revision argument: $::_revision not understood by git-svn\n";
1722 }
1723
1724 sub fetch_all {
1725 my ($repo_id, $remotes) = @_;
1726 if (ref $repo_id) {
1727 my $gs = $repo_id;
1728 $repo_id = undef;
1729 $repo_id = $gs->{repo_id};
1730 }
1731 $remotes ||= read_all_remotes();
1732 my $remote = $remotes->{$repo_id} or
1733 die "[svn-remote \"$repo_id\"] unknown\n";
1734 my $fetch = $remote->{fetch};
1735 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1736 my (@gs, @globs);
1737 my $ra = Git::SVN::Ra->new($url);
1738 my $uuid = $ra->get_uuid;
1739 my $head = $ra->get_latest_revnum;
1740 $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1741 my $base = defined $fetch ? $head : 0;
1742
1743 # read the max revs for wildcard expansion (branches/*, tags/*)
1744 foreach my $t (qw/branches tags/) {
1745 defined $remote->{$t} or next;
1746 push @globs, @{$remote->{$t}};
1747
1748 my $max_rev = eval { tmp_config(qw/--int --get/,
1749 "svn-remote.$repo_id.${t}-maxRev") };
1750 if (defined $max_rev && ($max_rev < $base)) {
1751 $base = $max_rev;
1752 } elsif (!defined $max_rev) {
1753 $base = 0;
1754 }
1755 }
1756
1757 if ($fetch) {
1758 foreach my $p (sort keys %$fetch) {
1759 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1760 my $lr = $gs->rev_map_max;
1761 if (defined $lr) {
1762 $base = $lr if ($lr < $base);
1763 }
1764 push @gs, $gs;
1765 }
1766 }
1767
1768 ($base, $head) = parse_revision_argument($base, $head);
1769 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1770 }
1771
1772 sub read_all_remotes {
1773 my $r = {};
1774 my $use_svm_props = eval { command_oneline(qw/config --bool
1775 svn.useSvmProps/) };
1776 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1777 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1778 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1779 if (m!^(.+)\.fetch=$svn_refspec$!) {
1780 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1781 die("svn-remote.$remote: remote ref '$remote_ref' "
1782 . "must start with 'refs/'\n")
1783 unless $remote_ref =~ m{^refs/};
1784 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1785 $r->{$remote}->{svm} = {} if $use_svm_props;
1786 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1787 $r->{$1}->{svm} = {};
1788 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1789 $r->{$1}->{url} = $2;
1790 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1791 my ($remote, $t, $local_ref, $remote_ref) =
1792 ($1, $2, $3, $4);
1793 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1794 . "must start with 'refs/'\n")
1795 unless $remote_ref =~ m{^refs/};
1796 my $rs = {
1797 t => $t,
1798 remote => $remote,
1799 path => Git::SVN::GlobSpec->new($local_ref),
1800 ref => Git::SVN::GlobSpec->new($remote_ref) };
1801 if (length($rs->{ref}->{right}) != 0) {
1802 die "The '*' glob character must be the last ",
1803 "character of '$remote_ref'\n";
1804 }
1805 push @{ $r->{$remote}->{$t} }, $rs;
1806 }
1807 }
1808
1809 map {
1810 if (defined $r->{$_}->{svm}) {
1811 my $svm;
1812 eval {
1813 my $section = "svn-remote.$_";
1814 $svm = {
1815 source => tmp_config('--get',
1816 "$section.svm-source"),
1817 replace => tmp_config('--get',
1818 "$section.svm-replace"),
1819 }
1820 };
1821 $r->{$_}->{svm} = $svm;
1822 }
1823 } keys %$r;
1824
1825 $r;
1826 }
1827
1828 sub init_vars {
1829 $_gc_nr = $_gc_period = 1000;
1830 if (defined $_repack || defined $_repack_flags) {
1831 warn "Repack options are obsolete; they have no effect.\n";
1832 }
1833 }
1834
1835 sub verify_remotes_sanity {
1836 return unless -d $ENV{GIT_DIR};
1837 my %seen;
1838 foreach (command(qw/config -l/)) {
1839 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1840 if ($seen{$1}) {
1841 die "Remote ref refs/remote/$1 is tracked by",
1842 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1843 "Please resolve this ambiguity in ",
1844 "your git configuration file before ",
1845 "continuing\n";
1846 }
1847 $seen{$1} = $_;
1848 }
1849 }
1850 }
1851
1852 sub find_existing_remote {
1853 my ($url, $remotes) = @_;
1854 return undef if $no_reuse_existing;
1855 my $existing;
1856 foreach my $repo_id (keys %$remotes) {
1857 my $u = $remotes->{$repo_id}->{url} or next;
1858 next if $u ne $url;
1859 $existing = $repo_id;
1860 last;
1861 }
1862 $existing;
1863 }
1864
1865 sub init_remote_config {
1866 my ($self, $url, $no_write) = @_;
1867 $url =~ s!/+$!!; # strip trailing slash
1868 my $r = read_all_remotes();
1869 my $existing = find_existing_remote($url, $r);
1870 if ($existing) {
1871 unless ($no_write) {
1872 print STDERR "Using existing ",
1873 "[svn-remote \"$existing\"]\n";
1874 }
1875 $self->{repo_id} = $existing;
1876 } elsif ($_minimize_url) {
1877 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1878 $existing = find_existing_remote($min_url, $r);
1879 if ($existing) {
1880 unless ($no_write) {
1881 print STDERR "Using existing ",
1882 "[svn-remote \"$existing\"]\n";
1883 }
1884 $self->{repo_id} = $existing;
1885 }
1886 if ($min_url ne $url) {
1887 unless ($no_write) {
1888 print STDERR "Using higher level of URL: ",
1889 "$url => $min_url\n";
1890 }
1891 my $old_path = $self->{path};
1892 $self->{path} = $url;
1893 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1894 if (length $old_path) {
1895 $self->{path} .= "/$old_path";
1896 }
1897 $url = $min_url;
1898 }
1899 }
1900 my $orig_url;
1901 if (!$existing) {
1902 # verify that we aren't overwriting anything:
1903 $orig_url = eval {
1904 command_oneline('config', '--get',
1905 "svn-remote.$self->{repo_id}.url")
1906 };
1907 if ($orig_url && ($orig_url ne $url)) {
1908 die "svn-remote.$self->{repo_id}.url already set: ",
1909 "$orig_url\nwanted to set to: $url\n";
1910 }
1911 }
1912 my ($xrepo_id, $xpath) = find_ref($self->refname);
1913 if (!$no_write && defined $xpath) {
1914 die "svn-remote.$xrepo_id.fetch already set to track ",
1915 "$xpath:", $self->refname, "\n";
1916 }
1917 unless ($no_write) {
1918 command_noisy('config',
1919 "svn-remote.$self->{repo_id}.url", $url);
1920 $self->{path} =~ s{^/}{};
1921 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1922 command_noisy('config', '--add',
1923 "svn-remote.$self->{repo_id}.fetch",
1924 "$self->{path}:".$self->refname);
1925 }
1926 $self->{url} = $url;
1927 }
1928
1929 sub find_by_url { # repos_root and, path are optional
1930 my ($class, $full_url, $repos_root, $path) = @_;
1931
1932 return undef unless defined $full_url;
1933 remove_username($full_url);
1934 remove_username($repos_root) if defined $repos_root;
1935 my $remotes = read_all_remotes();
1936 if (defined $full_url && defined $repos_root && !defined $path) {
1937 $path = $full_url;
1938 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1939 }
1940 foreach my $repo_id (keys %$remotes) {
1941 my $u = $remotes->{$repo_id}->{url} or next;
1942 remove_username($u);
1943 next if defined $repos_root && $repos_root ne $u;
1944
1945 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1946 foreach my $t (qw/branches tags/) {
1947 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1948 resolve_local_globs($u, $fetch, $globspec);
1949 }
1950 }
1951 my $p = $path;
1952 my $rwr = rewrite_root({repo_id => $repo_id});
1953 my $svm = $remotes->{$repo_id}->{svm}
1954 if defined $remotes->{$repo_id}->{svm};
1955 unless (defined $p) {
1956 $p = $full_url;
1957 my $z = $u;
1958 my $prefix = '';
1959 if ($rwr) {
1960 $z = $rwr;
1961 remove_username($z);
1962 } elsif (defined $svm) {
1963 $z = $svm->{source};
1964 $prefix = $svm->{replace};
1965 $prefix =~ s#^\Q$u\E(?:/|$)##;
1966 $prefix =~ s#/$##;
1967 }
1968 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1969 }
1970 foreach my $f (keys %$fetch) {
1971 next if $f ne $p;
1972 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1973 }
1974 }
1975 undef;
1976 }
1977
1978 sub init {
1979 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1980 my $self = _new($class, $repo_id, $ref_id, $path);
1981 if (defined $url) {
1982 $self->init_remote_config($url, $no_write);
1983 }
1984 $self;
1985 }
1986
1987 sub find_ref {
1988 my ($ref_id) = @_;
1989 foreach (command(qw/config -l/)) {
1990 next unless m!^svn-remote\.(.+)\.fetch=
1991 \s*(.*?)\s*:\s*(.+?)\s*$!x;
1992 my ($repo_id, $path, $ref) = ($1, $2, $3);
1993 if ($ref eq $ref_id) {
1994 $path = '' if ($path =~ m#^\./?#);
1995 return ($repo_id, $path);
1996 }
1997 }
1998 (undef, undef, undef);
1999 }
2000
2001 sub new {
2002 my ($class, $ref_id, $repo_id, $path) = @_;
2003 if (defined $ref_id && !defined $repo_id && !defined $path) {
2004 ($repo_id, $path) = find_ref($ref_id);
2005 if (!defined $repo_id) {
2006 die "Could not find a \"svn-remote.*.fetch\" key ",
2007 "in the repository configuration matching: ",
2008 "$ref_id\n";
2009 }
2010 }
2011 my $self = _new($class, $repo_id, $ref_id, $path);
2012 if (!defined $self->{path} || !length $self->{path}) {
2013 my $fetch = command_oneline('config', '--get',
2014 "svn-remote.$repo_id.fetch",
2015 ":$ref_id\$") or
2016 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2017 "\":$ref_id\$\" in config\n";
2018 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2019 }
2020 $self->{url} = command_oneline('config', '--get',
2021 "svn-remote.$repo_id.url") or
2022 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2023 $self->rebuild;
2024 $self;
2025 }
2026
2027 sub refname {
2028 my ($refname) = $_[0]->{ref_id} ;
2029
2030 # It cannot end with a slash /, we'll throw up on this because
2031 # SVN can't have directories with a slash in their name, either:
2032 if ($refname =~ m{/$}) {
2033 die "ref: '$refname' ends with a trailing slash, this is ",
2034 "not permitted by git nor Subversion\n";
2035 }
2036
2037 # It cannot have ASCII control character space, tilde ~, caret ^,
2038 # colon :, question-mark ?, asterisk *, space, or open bracket [
2039 # anywhere.
2040 #
2041 # Additionally, % must be escaped because it is used for escaping
2042 # and we want our escaped refname to be reversible
2043 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2044
2045 # no slash-separated component can begin with a dot .
2046 # /.* becomes /%2E*
2047 $refname =~ s{/\.}{/%2E}g;
2048
2049 # It cannot have two consecutive dots .. anywhere
2050 # .. becomes %2E%2E
2051 $refname =~ s{\.\.}{%2E%2E}g;
2052
2053 return $refname;
2054 }
2055
2056 sub desanitize_refname {
2057 my ($refname) = @_;
2058 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2059 return $refname;
2060 }
2061
2062 sub svm_uuid {
2063 my ($self) = @_;
2064 return $self->{svm}->{uuid} if $self->svm;
2065 $self->ra;
2066 unless ($self->{svm}) {
2067 die "SVM UUID not cached, and reading remotely failed\n";
2068 }
2069 $self->{svm}->{uuid};
2070 }
2071
2072 sub svm {
2073 my ($self) = @_;
2074 return $self->{svm} if $self->{svm};
2075 my $svm;
2076 # see if we have it in our config, first:
2077 eval {
2078 my $section = "svn-remote.$self->{repo_id}";
2079 $svm = {
2080 source => tmp_config('--get', "$section.svm-source"),
2081 uuid => tmp_config('--get', "$section.svm-uuid"),
2082 replace => tmp_config('--get', "$section.svm-replace"),
2083 }
2084 };
2085 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2086 $self->{svm} = $svm;
2087 }
2088 $self->{svm};
2089 }
2090
2091 sub _set_svm_vars {
2092 my ($self, $ra) = @_;
2093 return $ra if $self->svm;
2094
2095 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2096 "(svm:source, svm:uuid) ",
2097 "from the following URLs:\n" );
2098 sub read_svm_props {
2099 my ($self, $ra, $path, $r) = @_;
2100 my $props = ($ra->get_dir($path, $r))[2];
2101 my $src = $props->{'svm:source'};
2102 my $uuid = $props->{'svm:uuid'};
2103 return undef if (!$src || !$uuid);
2104
2105 chomp($src, $uuid);
2106
2107 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2108 or die "doesn't look right - svm:uuid is '$uuid'\n";
2109
2110 # the '!' is used to mark the repos_root!/relative/path
2111 $src =~ s{/?!/?}{/};
2112 $src =~ s{/+$}{}; # no trailing slashes please
2113 # username is of no interest
2114 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2115
2116 my $replace = $ra->{url};
2117 $replace .= "/$path" if length $path;
2118
2119 my $section = "svn-remote.$self->{repo_id}";
2120 tmp_config("$section.svm-source", $src);
2121 tmp_config("$section.svm-replace", $replace);
2122 tmp_config("$section.svm-uuid", $uuid);
2123 $self->{svm} = {
2124 source => $src,
2125 uuid => $uuid,
2126 replace => $replace
2127 };
2128 }
2129
2130 my $r = $ra->get_latest_revnum;
2131 my $path = $self->{path};
2132 my %tried;
2133 while (length $path) {
2134 unless ($tried{"$self->{url}/$path"}) {
2135 return $ra if $self->read_svm_props($ra, $path, $r);
2136 $tried{"$self->{url}/$path"} = 1;
2137 }
2138 $path =~ s#/?[^/]+$##;
2139 }
2140 die "Path: '$path' should be ''\n" if $path ne '';
2141 return $ra if $self->read_svm_props($ra, $path, $r);
2142 $tried{"$self->{url}/$path"} = 1;
2143
2144 if ($ra->{repos_root} eq $self->{url}) {
2145 die @err, (map { " $_\n" } keys %tried), "\n";
2146 }
2147
2148 # nope, make sure we're connected to the repository root:
2149 my $ok;
2150 my @tried_b;
2151 $path = $ra->{svn_path};
2152 $ra = Git::SVN::Ra->new($ra->{repos_root});
2153 while (length $path) {
2154 unless ($tried{"$ra->{url}/$path"}) {
2155 $ok = $self->read_svm_props($ra, $path, $r);
2156 last if $ok;
2157 $tried{"$ra->{url}/$path"} = 1;
2158 }
2159 $path =~ s#/?[^/]+$##;
2160 }
2161 die "Path: '$path' should be ''\n" if $path ne '';
2162 $ok ||= $self->read_svm_props($ra, $path, $r);
2163 $tried{"$ra->{url}/$path"} = 1;
2164 if (!$ok) {
2165 die @err, (map { " $_\n" } keys %tried), "\n";
2166 }
2167 Git::SVN::Ra->new($self->{url});
2168 }
2169
2170 sub svnsync {
2171 my ($self) = @_;
2172 return $self->{svnsync} if $self->{svnsync};
2173
2174 if ($self->no_metadata) {
2175 die "Can't have both 'noMetadata' and ",
2176 "'useSvnsyncProps' options set!\n";
2177 }
2178 if ($self->rewrite_root) {
2179 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2180 "options set!\n";
2181 }
2182
2183 my $svnsync;
2184 # see if we have it in our config, first:
2185 eval {
2186 my $section = "svn-remote.$self->{repo_id}";
2187
2188 my $url = tmp_config('--get', "$section.svnsync-url");
2189 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2190 die "doesn't look right - svn:sync-from-url is '$url'\n";
2191
2192 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2193 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2194 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2195
2196 $svnsync = { url => $url, uuid => $uuid }
2197 };
2198 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2199 return $self->{svnsync} = $svnsync;
2200 }
2201
2202 my $err = "useSvnsyncProps set, but failed to read " .
2203 "svnsync property: svn:sync-from-";
2204 my $rp = $self->ra->rev_proplist(0);
2205
2206 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2207 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2208 die "doesn't look right - svn:sync-from-url is '$url'\n";
2209
2210 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2211 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2212 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2213
2214 my $section = "svn-remote.$self->{repo_id}";
2215 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2216 tmp_config('--add', "$section.svnsync-url", $url);
2217 return $self->{svnsync} = { url => $url, uuid => $uuid };
2218 }
2219
2220 # this allows us to memoize our SVN::Ra UUID locally and avoid a
2221 # remote lookup (useful for 'git svn log').
2222 sub ra_uuid {
2223 my ($self) = @_;
2224 unless ($self->{ra_uuid}) {
2225 my $key = "svn-remote.$self->{repo_id}.uuid";
2226 my $uuid = eval { tmp_config('--get', $key) };
2227 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2228 $self->{ra_uuid} = $uuid;
2229 } else {
2230 die "ra_uuid called without URL\n" unless $self->{url};
2231 $self->{ra_uuid} = $self->ra->get_uuid;
2232 tmp_config('--add', $key, $self->{ra_uuid});
2233 }
2234 }
2235 $self->{ra_uuid};
2236 }
2237
2238 sub _set_repos_root {
2239 my ($self, $repos_root) = @_;
2240 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2241 $repos_root ||= $self->ra->{repos_root};
2242 tmp_config($k, $repos_root);
2243 $repos_root;
2244 }
2245
2246 sub repos_root {
2247 my ($self) = @_;
2248 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2249 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2250 }
2251
2252 sub ra {
2253 my ($self) = shift;
2254 my $ra = Git::SVN::Ra->new($self->{url});
2255 $self->_set_repos_root($ra->{repos_root});
2256 if ($self->use_svm_props && !$self->{svm}) {
2257 if ($self->no_metadata) {
2258 die "Can't have both 'noMetadata' and ",
2259 "'useSvmProps' options set!\n";
2260 } elsif ($self->use_svnsync_props) {
2261 die "Can't have both 'useSvnsyncProps' and ",
2262 "'useSvmProps' options set!\n";
2263 }
2264 $ra = $self->_set_svm_vars($ra);
2265 $self->{-want_revprops} = 1;
2266 }
2267 $ra;
2268 }
2269
2270 # prop_walk(PATH, REV, SUB)
2271 # -------------------------
2272 # Recursively traverse PATH at revision REV and invoke SUB for each
2273 # directory that contains a SVN property. SUB will be invoked as
2274 # follows: &SUB(gs, path, props); where `gs' is this instance of
2275 # Git::SVN, `path' the path to the directory where the properties
2276 # `props' were found. The `path' will be relative to point of checkout,
2277 # that is, if url://repo/trunk is the current Git branch, and that
2278 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
2279 # as `path' (note the trailing `/').
2280 sub prop_walk {
2281 my ($self, $path, $rev, $sub) = @_;
2282
2283 $path =~ s#^/##;
2284 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2285 $path =~ s#^/*#/#g;
2286 my $p = $path;
2287 # Strip the irrelevant part of the path.
2288 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2289 # Ensure the path is terminated by a `/'.
2290 $p =~ s#/*$#/#;
2291
2292 # The properties contain all the internal SVN stuff nobody
2293 # (usually) cares about.
2294 my $interesting_props = 0;
2295 foreach (keys %{$props}) {
2296 # If it doesn't start with `svn:', it must be a
2297 # user-defined property.
2298 ++$interesting_props and next if $_ !~ /^svn:/;
2299 # FIXME: Fragile, if SVN adds new public properties,
2300 # this needs to be updated.
2301 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2302 |eol-style|mime-type
2303 |externals|needs-lock)$/x;
2304 }
2305 &$sub($self, $p, $props) if $interesting_props;
2306
2307 foreach (sort keys %$dirent) {
2308 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2309 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2310 }
2311 }
2312
2313 sub last_rev { ($_[0]->last_rev_commit)[0] }
2314 sub last_commit { ($_[0]->last_rev_commit)[1] }
2315
2316 # returns the newest SVN revision number and newest commit SHA1
2317 sub last_rev_commit {
2318 my ($self) = @_;
2319 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2320 return ($self->{last_rev}, $self->{last_commit});
2321 }
2322 my $c = ::verify_ref($self->refname.'^0');
2323 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2324 my $rev = (::cmt_metadata($c))[1];
2325 if (defined $rev) {
2326 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2327 return ($rev, $c);
2328 }
2329 }
2330 my $map_path = $self->map_path;
2331 unless (-e $map_path) {
2332 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2333 return (undef, undef);
2334 }
2335 my ($rev, $commit) = $self->rev_map_max(1);
2336 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2337 return ($rev, $commit);
2338 }
2339
2340 sub get_fetch_range {
2341 my ($self, $min, $max) = @_;
2342 $max ||= $self->ra->get_latest_revnum;
2343 $min ||= $self->rev_map_max;
2344 (++$min, $max);
2345 }
2346
2347 sub tmp_config {
2348 my (@args) = @_;
2349 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2350 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2351 if (! -f $config && -f $old_def_config) {
2352 rename $old_def_config, $config or
2353 die "Failed rename $old_def_config => $config: $!\n";
2354 }
2355 my $old_config = $ENV{GIT_CONFIG};
2356 $ENV{GIT_CONFIG} = $config;
2357 $@ = undef;
2358 my @ret = eval {
2359 unless (-f $config) {
2360 mkfile($config);
2361 open my $fh, '>', $config or
2362 die "Can't open $config: $!\n";
2363 print $fh "; This file is used internally by ",
2364 "git-svn\n" or die
2365 "Couldn't write to $config: $!\n";
2366 print $fh "; You should not have to edit it\n" or
2367 die "Couldn't write to $config: $!\n";
2368 close $fh or die "Couldn't close $config: $!\n";
2369 }
2370 command('config', @args);
2371 };
2372 my $err = $@;
2373 if (defined $old_config) {
2374 $ENV{GIT_CONFIG} = $old_config;
2375 } else {
2376 delete $ENV{GIT_CONFIG};
2377 }
2378 die $err if $err;
2379 wantarray ? @ret : $ret[0];
2380 }
2381
2382 sub tmp_index_do {
2383 my ($self, $sub) = @_;
2384 my $old_index = $ENV{GIT_INDEX_FILE};
2385 $ENV{GIT_INDEX_FILE} = $self->{index};
2386 $@ = undef;
2387 my @ret = eval {
2388 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2389 mkpath([$dir]) unless -d $dir;
2390 &$sub;
2391 };
2392 my $err = $@;
2393 if (defined $old_index) {
2394 $ENV{GIT_INDEX_FILE} = $old_index;
2395 } else {
2396 delete $ENV{GIT_INDEX_FILE};
2397 }
2398 die $err if $err;
2399 wantarray ? @ret : $ret[0];
2400 }
2401
2402 sub assert_index_clean {
2403 my ($self, $treeish) = @_;
2404
2405 $self->tmp_index_do(sub {
2406 command_noisy('read-tree', $treeish) unless -e $self->{index};
2407 my $x = command_oneline('write-tree');
2408 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2409 /^tree ($::sha1)/mo);
2410 return if $y eq $x;
2411
2412 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2413 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2414 command_noisy('read-tree', $treeish);
2415 $x = command_oneline('write-tree');
2416 if ($y ne $x) {
2417 ::fatal "trees ($treeish) $y != $x\n",
2418 "Something is seriously wrong...";
2419 }
2420 });
2421 }
2422
2423 sub get_commit_parents {
2424 my ($self, $log_entry) = @_;
2425 my (%seen, @ret, @tmp);
2426 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2427 if (my $ip = $self->{inject_parents}) {
2428 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2429 push @tmp, $commit;
2430 }
2431 }
2432 if (my $cur = ::verify_ref($self->refname.'^0')) {
2433 push @tmp, $cur;
2434 }
2435 if (my $ipd = $self->{inject_parents_dcommit}) {
2436 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2437 push @tmp, @$commit;
2438 }
2439 }
2440 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2441 while (my $p = shift @tmp) {
2442 next if $seen{$p};
2443 $seen{$p} = 1;
2444 push @ret, $p;
2445 # MAXPARENT is defined to 16 in commit-tree.c:
2446 last if @ret >= 16;
2447 }
2448 if (@tmp) {
2449 die "r$log_entry->{revision}: No room for parents:\n\t",
2450 join("\n\t", @tmp), "\n";
2451 }
2452 @ret;
2453 }
2454
2455 sub rewrite_root {
2456 my ($self) = @_;
2457 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2458 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2459 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2460 if ($rwr) {
2461 $rwr =~ s#/+$##;
2462 if ($rwr !~ m#^[a-z\+]+://#) {
2463 die "$rwr is not a valid URL (key: $k)\n";
2464 }
2465 }
2466 $self->{-rewrite_root} = $rwr;
2467 }
2468
2469 sub metadata_url {
2470 my ($self) = @_;
2471 ($self->rewrite_root || $self->{url}) .
2472 (length $self->{path} ? '/' . $self->{path} : '');
2473 }
2474
2475 sub full_url {
2476 my ($self) = @_;
2477 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2478 }
2479
2480
2481 sub set_commit_header_env {
2482 my ($log_entry) = @_;
2483 my %env;
2484 foreach my $ned (qw/NAME EMAIL DATE/) {
2485 foreach my $ac (qw/AUTHOR COMMITTER/) {
2486 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2487 }
2488 }
2489
2490 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2491 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2492 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2493
2494 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2495 ? $log_entry->{commit_name}
2496 : $log_entry->{name};
2497 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2498 ? $log_entry->{commit_email}
2499 : $log_entry->{email};
2500 \%env;
2501 }
2502
2503 sub restore_commit_header_env {
2504 my ($env) = @_;
2505 foreach my $ned (qw/NAME EMAIL DATE/) {
2506 foreach my $ac (qw/AUTHOR COMMITTER/) {
2507 my $k = "GIT_${ac}_${ned}";
2508 if (defined $env->{$k}) {
2509 $ENV{$k} = $env->{$k};
2510 } else {
2511 delete $ENV{$k};
2512 }
2513 }
2514 }
2515 }
2516
2517 sub gc {
2518 command_noisy('gc', '--auto');
2519 };
2520
2521 sub do_git_commit {
2522 my ($self, $log_entry) = @_;
2523 my $lr = $self->last_rev;
2524 if (defined $lr && $lr >= $log_entry->{revision}) {
2525 die "Last fetched revision of ", $self->refname,
2526 " was r$lr, but we are about to fetch: ",
2527 "r$log_entry->{revision}!\n";
2528 }
2529 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2530 croak "$log_entry->{revision} = $c already exists! ",
2531 "Why are we refetching it?\n";
2532 }
2533 my $old_env = set_commit_header_env($log_entry);
2534 my $tree = $log_entry->{tree};
2535 if (!defined $tree) {
2536 $tree = $self->tmp_index_do(sub {
2537 command_oneline('write-tree') });
2538 }
2539 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2540
2541 my @exec = ('git', 'commit-tree', $tree);
2542 foreach ($self->get_commit_parents($log_entry)) {
2543 push @exec, '-p', $_;
2544 }
2545 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2546 or croak $!;
2547 binmode $msg_fh;
2548
2549 # we always get UTF-8 from SVN, but we may want our commits in
2550 # a different encoding.
2551 if (my $enc = Git::config('i18n.commitencoding')) {
2552 require Encode;
2553 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2554 }
2555 print $msg_fh $log_entry->{log} or croak $!;
2556 restore_commit_header_env($old_env);
2557 unless ($self->no_metadata) {
2558 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2559 or croak $!;
2560 }
2561 $msg_fh->flush == 0 or croak $!;
2562 close $msg_fh or croak $!;
2563 chomp(my $commit = do { local $/; <$out_fh> });
2564 close $out_fh or croak $!;
2565 waitpid $pid, 0;
2566 croak $? if $?;
2567 if ($commit !~ /^$::sha1$/o) {
2568 die "Failed to commit, invalid sha1: $commit\n";
2569 }
2570
2571 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2572
2573 $self->{last_rev} = $log_entry->{revision};
2574 $self->{last_commit} = $commit;
2575 print "r$log_entry->{revision}" unless $::_q > 1;
2576 if (defined $log_entry->{svm_revision}) {
2577 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2578 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2579 0, $self->svm_uuid);
2580 }
2581 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2582 if (--$_gc_nr == 0) {
2583 $_gc_nr = $_gc_period;
2584 gc();
2585 }
2586 return $commit;
2587 }
2588
2589 sub match_paths {
2590 my ($self, $paths, $r) = @_;
2591 return 1 if $self->{path} eq '';
2592 if (my $path = $paths->{"/$self->{path}"}) {
2593 return ($path->{action} eq 'D') ? 0 : 1;
2594 }
2595 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2596 if (grep /$self->{path_regex}/, keys %$paths) {
2597 return 1;
2598 }
2599 my $c = '';
2600 foreach (split m#/#, $self->{path}) {
2601 $c .= "/$_";
2602 next unless ($paths->{$c} &&
2603 ($paths->{$c}->{action} =~ /^[AR]$/));
2604 if ($self->ra->check_path($self->{path}, $r) ==
2605 $SVN::Node::dir) {
2606 return 1;
2607 }
2608 }
2609 return 0;
2610 }
2611
2612 sub find_parent_branch {
2613 my ($self, $paths, $rev) = @_;
2614 return undef unless $self->follow_parent;
2615 unless (defined $paths) {
2616 my $err_handler = $SVN::Error::handler;
2617 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2618 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2619 sub { $paths = $_[0] });
2620 $SVN::Error::handler = $err_handler;
2621 }
2622 return undef unless defined $paths;
2623
2624 # look for a parent from another branch:
2625 my @b_path_components = split m#/#, $self->{path};
2626 my @a_path_components;
2627 my $i;
2628 while (@b_path_components) {
2629 $i = $paths->{'/'.join('/', @b_path_components)};
2630 last if $i && defined $i->{copyfrom_path};
2631 unshift(@a_path_components, pop(@b_path_components));
2632 }
2633 return undef unless defined $i && defined $i->{copyfrom_path};
2634 my $branch_from = $i->{copyfrom_path};
2635 if (@a_path_components) {
2636 print STDERR "branch_from: $branch_from => ";
2637 $branch_from .= '/'.join('/', @a_path_components);
2638 print STDERR $branch_from, "\n";
2639 }
2640 my $r = $i->{copyfrom_rev};
2641 my $repos_root = $self->ra->{repos_root};
2642 my $url = $self->ra->{url};
2643 my $new_url = $url . $branch_from;
2644 print STDERR "Found possible branch point: ",
2645 "$new_url => ", $self->full_url, ", $r\n"
2646 unless $::_q > 1;
2647 $branch_from =~ s#^/##;
2648 my $gs = $self->other_gs($new_url, $url,
2649 $branch_from, $r, $self->{ref_id});
2650 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2651 {
2652 my ($base, $head);
2653 if (!defined $r0 || !defined $parent) {
2654 ($base, $head) = parse_revision_argument(0, $r);
2655 } else {
2656 if ($r0 < $r) {
2657 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2658 0, 1, sub { $base = $_[1] - 1 });
2659 }
2660 }
2661 if (defined $base && $base <= $r) {
2662 $gs->fetch($base, $r);
2663 }
2664 ($r0, $parent) = $gs->find_rev_before($r, 1);
2665 }
2666 if (defined $r0 && defined $parent) {
2667 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2668 unless $::_q > 1;
2669 my $ed;
2670 if ($self->ra->can_do_switch) {
2671 $self->assert_index_clean($parent);
2672 print STDERR "Following parent with do_switch\n"
2673 unless $::_q > 1;
2674 # do_switch works with svn/trunk >= r22312, but that
2675 # is not included with SVN 1.4.3 (the latest version
2676 # at the moment), so we can't rely on it
2677 $self->{last_rev} = $r0;
2678 $self->{last_commit} = $parent;
2679 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2680 $gs->ra->gs_do_switch($r0, $rev, $gs,
2681 $self->full_url, $ed)
2682 or die "SVN connection failed somewhere...\n";
2683 } elsif ($self->ra->trees_match($new_url, $r0,
2684 $self->full_url, $rev)) {
2685 print STDERR "Trees match:\n",
2686 " $new_url\@$r0\n",
2687 " ${\$self->full_url}\@$rev\n",
2688 "Following parent with no changes\n"
2689 unless $::_q > 1;
2690 $self->tmp_index_do(sub {
2691 command_noisy('read-tree', $parent);
2692 });
2693 $self->{last_commit} = $parent;
2694 } else {
2695 print STDERR "Following parent with do_update\n"
2696 unless $::_q > 1;
2697 $ed = SVN::Git::Fetcher->new($self);
2698 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2699 or die "SVN connection failed somewhere...\n";
2700 }
2701 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2702 return $self->make_log_entry($rev, [$parent], $ed);
2703 }
2704 return undef;
2705 }
2706
2707 sub do_fetch {
2708 my ($self, $paths, $rev) = @_;
2709 my $ed;
2710 my ($last_rev, @parents);
2711 if (my $lc = $self->last_commit) {
2712 # we can have a branch that was deleted, then re-added
2713 # under the same name but copied from another path, in
2714 # which case we'll have multiple parents (we don't
2715 # want to break the original ref, nor lose copypath info):
2716 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2717 push @{$log_entry->{parents}}, $lc;
2718 return $log_entry;
2719 }
2720 $ed = SVN::Git::Fetcher->new($self);
2721 $last_rev = $self->{last_rev};
2722 $ed->{c} = $lc;
2723 @parents = ($lc);
2724 } else {
2725 $last_rev = $rev;
2726 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2727 return $log_entry;
2728 }
2729 $ed = SVN::Git::Fetcher->new($self);
2730 }
2731 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2732 die "SVN connection failed somewhere...\n";
2733 }
2734 $self->make_log_entry($rev, \@parents, $ed);
2735 }
2736
2737 sub mkemptydirs {
2738 my ($self, $r) = @_;
2739 my %empty_dirs = ();
2740
2741 open my $fh, '<', "$self->{dir}/unhandled.log" or return;
2742 binmode $fh or croak "binmode: $!";
2743 while (<$fh>) {
2744 if (defined $r && /^r(\d+)$/) {
2745 last if $1 > $r;
2746 } elsif (/^ \+empty_dir: (.+)$/) {
2747 $empty_dirs{$1} = 1;
2748 } elsif (/^ \-empty_dir: (.+)$/) {
2749 delete $empty_dirs{$1};
2750 }
2751 }
2752 close $fh;
2753 foreach my $d (sort keys %empty_dirs) {
2754 $d = uri_decode($d);
2755 next if -d $d;
2756 if (-e _) {
2757 warn "$d exists but is not a directory\n";
2758 } else {
2759 print "creating empty directory: $d\n";
2760 mkpath([$d]);
2761 }
2762 }
2763 }
2764
2765 sub get_untracked {
2766 my ($self, $ed) = @_;
2767 my @out;
2768 my $h = $ed->{empty};
2769 foreach (sort keys %$h) {
2770 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2771 push @out, " $act: " . uri_encode($_);
2772 warn "W: $act: $_\n";
2773 }
2774 foreach my $t (qw/dir_prop file_prop/) {
2775 $h = $ed->{$t} or next;
2776 foreach my $path (sort keys %$h) {
2777 my $ppath = $path eq '' ? '.' : $path;
2778 foreach my $prop (sort keys %{$h->{$path}}) {
2779 next if $SKIP_PROP{$prop};
2780 my $v = $h->{$path}->{$prop};
2781 my $t_ppath_prop = "$t: " .
2782 uri_encode($ppath) . ' ' .
2783 uri_encode($prop);
2784 if (defined $v) {
2785 push @out, " +$t_ppath_prop " .
2786 uri_encode($v);
2787 } else {
2788 push @out, " -$t_ppath_prop";
2789 }
2790 }
2791 }
2792 }
2793 foreach my $t (qw/absent_file absent_directory/) {
2794 $h = $ed->{$t} or next;
2795 foreach my $parent (sort keys %$h) {
2796 foreach my $path (sort @{$h->{$parent}}) {
2797 push @out, " $t: " .
2798 uri_encode("$parent/$path");
2799 warn "W: $t: $parent/$path ",
2800 "Insufficient permissions?\n";
2801 }
2802 }
2803 }
2804 \@out;
2805 }
2806
2807 # parse_svn_date(DATE)
2808 # --------------------
2809 # Given a date (in UTC) from Subversion, return a string in the format
2810 # "<TZ Offset> <local date/time>" that Git will use.
2811 #
2812 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
2813 # is true we'll convert it to the local timezone instead.
2814 sub parse_svn_date {
2815 my $date = shift || return '+0000 1970-01-01 00:00:00';
2816 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2817 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2818 croak "Unable to parse date: $date\n";
2819 my $parsed_date; # Set next.
2820
2821 if ($Git::SVN::_localtime) {
2822 # Translate the Subversion datetime to an epoch time.
2823 # Begin by switching ourselves to $date's timezone, UTC.
2824 my $old_env_TZ = $ENV{TZ};
2825 $ENV{TZ} = 'UTC';
2826
2827 my $epoch_in_UTC =
2828 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2829
2830 # Determine our local timezone (including DST) at the
2831 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2832 # value of TZ, if any, at the time we were run.
2833 if (defined $Git::SVN::Log::TZ) {
2834 $ENV{TZ} = $Git::SVN::Log::TZ;
2835 } else {
2836 delete $ENV{TZ};
2837 }
2838
2839 my $our_TZ =
2840 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2841
2842 # This converts $epoch_in_UTC into our local timezone.
2843 my ($sec, $min, $hour, $mday, $mon, $year,
2844 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2845
2846 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2847 $our_TZ, $year + 1900, $mon + 1,
2848 $mday, $hour, $min, $sec);
2849
2850 # Reset us to the timezone in effect when we entered
2851 # this routine.
2852 if (defined $old_env_TZ) {
2853 $ENV{TZ} = $old_env_TZ;
2854 } else {
2855 delete $ENV{TZ};
2856 }
2857 } else {
2858 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2859 }
2860
2861 return $parsed_date;
2862 }
2863
2864 sub other_gs {
2865 my ($self, $new_url, $url,
2866 $branch_from, $r, $old_ref_id) = @_;
2867 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2868 unless ($gs) {
2869 my $ref_id = $old_ref_id;
2870 $ref_id =~ s/\@\d+$//;
2871 $ref_id .= "\@$r";
2872 # just grow a tail if we're not unique enough :x
2873 $ref_id .= '-' while find_ref($ref_id);
2874 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2875 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2876 if ($u =~ s#^\Q$url\E(/|$)##) {
2877 $p = $u;
2878 $u = $url;
2879 $repo_id = $self->{repo_id};
2880 }
2881 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2882 }
2883 $gs
2884 }
2885
2886 sub call_authors_prog {
2887 my ($orig_author) = @_;
2888 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2889 my $author = `$::_authors_prog $orig_author`;
2890 if ($? != 0) {
2891 die "$::_authors_prog failed with exit code $?\n"
2892 }
2893 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2894 my ($name, $email) = ($1, $2);
2895 $email = undef if length $2 == 0;
2896 return [$name, $email];
2897 } else {
2898 die "Author: $orig_author: $::_authors_prog returned "
2899 . "invalid author format: $author\n";
2900 }
2901 }
2902
2903 sub check_author {
2904 my ($author) = @_;
2905 if (!defined $author || length $author == 0) {
2906 $author = '(no author)';
2907 }
2908 if (!defined $::users{$author}) {
2909 if (defined $::_authors_prog) {
2910 $::users{$author} = call_authors_prog($author);
2911 } elsif (defined $::_authors) {
2912 die "Author: $author not defined in $::_authors file\n";
2913 }
2914 }
2915 $author;
2916 }
2917
2918 sub find_extra_svk_parents {
2919 my ($self, $ed, $tickets, $parents) = @_;
2920 # aha! svk:merge property changed...
2921 my @tickets = split "\n", $tickets;
2922 my @known_parents;
2923 for my $ticket ( @tickets ) {
2924 my ($uuid, $path, $rev) = split /:/, $ticket;
2925 if ( $uuid eq $self->ra_uuid ) {
2926 my $url = $self->rewrite_root || $self->{url};
2927 my $repos_root = $url;
2928 my $branch_from = $path;
2929 $branch_from =~ s{^/}{};
2930 my $gs = $self->other_gs($repos_root."/".$branch_from,
2931 $url,
2932 $branch_from,
2933 $rev,
2934 $self->{ref_id});
2935 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
2936 # wahey! we found it, but it might be
2937 # an old one (!)
2938 push @known_parents, $commit;
2939 }
2940 }
2941 }
2942 for my $parent ( @known_parents ) {
2943 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
2944 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
2945 my $new;
2946 while ( <$msg_fh> ) {
2947 $new=1;last;
2948 }
2949 command_close_pipe($msg_fh, $ctx);
2950 if ( $new ) {
2951 print STDERR
2952 "Found merge parent (svk:merge ticket): $parent\n";
2953 push @$parents, $parent;
2954 }
2955 }
2956 }
2957
2958 # note: this function should only be called if the various dirprops
2959 # have actually changed
2960 sub find_extra_svn_parents {
2961 my ($self, $ed, $mergeinfo, $parents) = @_;
2962 # aha! svk:merge property changed...
2963
2964 # We first search for merged tips which are not in our
2965 # history. Then, we figure out which git revisions are in
2966 # that tip, but not this revision. If all of those revisions
2967 # are now marked as merge, we can add the tip as a parent.
2968 my @merges = split "\n", $mergeinfo;
2969 my @merge_tips;
2970 my @merged_commit_ranges;
2971 my $url = $self->rewrite_root || $self->{url};
2972 for my $merge ( @merges ) {
2973 my ($source, $revs) = split ":", $merge;
2974 my $path = $source;
2975 $path =~ s{^/}{};
2976 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
2977 if ( !$gs ) {
2978 warn "Couldn't find revmap for $url$source\n";
2979 next;
2980 }
2981 my @ranges = split ",", $revs;
2982 my ($tip, $tip_commit);
2983 # find the tip
2984 for my $range ( @ranges ) {
2985 my ($bottom, $top) = split "-", $range;
2986 $top ||= $bottom;
2987 my $bottom_commit =
2988 $gs->rev_map_get($bottom, $self->ra_uuid) ||
2989 $gs->rev_map_get($bottom+1, $self->ra_uuid);
2990 my $top_commit;
2991 for (; !$top_commit && $top >= $bottom; --$top) {
2992 $top_commit =
2993 $gs->rev_map_get($top, $self->ra_uuid);
2994 }
2995
2996 unless ($top_commit and $bottom_commit) {
2997 warn "W:unknown path/rev in svn:mergeinfo "
2998 ."dirprop: $source:$range\n";
2999 next;
3000 }
3001
3002 push @merged_commit_ranges,
3003 "$bottom_commit..$top_commit";
3004
3005 if ( !defined $tip or $top > $tip ) {
3006 $tip = $top;
3007 $tip_commit = $top_commit;
3008 }
3009 }
3010 unless (!$tip_commit or
3011 grep { $_ eq $tip_commit } @$parents ) {
3012 push @merge_tips, $tip_commit;
3013 } else {
3014 push @merge_tips, undef;
3015 }
3016 }
3017 for my $merge_tip ( @merge_tips ) {
3018 my $spec = shift @merges;
3019 next unless $merge_tip;
3020 my @cmd = ('rev-list', "-1", $merge_tip,
3021 "--not", @$parents );
3022 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3023 my $new;
3024 while ( <$msg_fh> ) {
3025 $new=1;last;
3026 }
3027 command_close_pipe($msg_fh, $ctx);
3028 if ( $new ) {
3029 push @cmd, @merged_commit_ranges;
3030 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3031 my $unmerged;
3032 while ( <$msg_fh> ) {
3033 $unmerged=1;last;
3034 }
3035 command_close_pipe($msg_fh, $ctx);
3036 if ( $unmerged ) {
3037 warn "W:svn cherry-pick ignored ($spec)\n";
3038 } else {
3039 warn
3040 "Found merge parent (svn:mergeinfo prop): ",
3041 $merge_tip, "\n";
3042 push @$parents, $merge_tip;
3043 }
3044 }
3045 }
3046 }
3047
3048 sub make_log_entry {
3049 my ($self, $rev, $parents, $ed) = @_;
3050 my $untracked = $self->get_untracked($ed);
3051
3052 my @parents = @$parents;
3053 my $ps = $ed->{path_strip} || "";
3054 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3055 my $props = $ed->{dir_prop}{$path};
3056 if ( $props->{"svk:merge"} ) {
3057 $self->find_extra_svk_parents
3058 ($ed, $props->{"svk:merge"}, \@parents);
3059 }
3060 if ( $props->{"svn:mergeinfo"} ) {
3061 $self->find_extra_svn_parents
3062 ($ed,
3063 $props->{"svn:mergeinfo"},
3064 \@parents);
3065 }
3066 }
3067
3068 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3069 print $un "r$rev\n" or croak $!;
3070 print $un $_, "\n" foreach @$untracked;
3071 my %log_entry = ( parents => \@parents, revision => $rev,
3072 log => '');
3073
3074 my $headrev;
3075 my $logged = delete $self->{logged_rev_props};
3076 if (!$logged || $self->{-want_revprops}) {
3077 my $rp = $self->ra->rev_proplist($rev);
3078 foreach (sort keys %$rp) {
3079 my $v = $rp->{$_};
3080 if (/^svn:(author|date|log)$/) {
3081 $log_entry{$1} = $v;
3082 } elsif ($_ eq 'svm:headrev') {
3083 $headrev = $v;
3084 } else {
3085 print $un " rev_prop: ", uri_encode($_), ' ',
3086 uri_encode($v), "\n";
3087 }
3088 }
3089 } else {
3090 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3091 }
3092 close $un or croak $!;
3093
3094 $log_entry{date} = parse_svn_date($log_entry{date});
3095 $log_entry{log} .= "\n";
3096 my $author = $log_entry{author} = check_author($log_entry{author});
3097 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3098 : ($author, undef);
3099
3100 my ($commit_name, $commit_email) = ($name, $email);
3101 if ($_use_log_author) {
3102 my $name_field;
3103 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3104 $name_field = $1;
3105 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3106 $name_field = $1;
3107 }
3108 if (!defined $name_field) {
3109 if (!defined $email) {
3110 $email = $name;
3111 }
3112 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3113 ($name, $email) = ($1, $2);
3114 } elsif ($name_field =~ /(.*)@/) {
3115 ($name, $email) = ($1, $name_field);
3116 } else {
3117 ($name, $email) = ($name_field, $name_field);
3118 }
3119 }
3120 if (defined $headrev && $self->use_svm_props) {
3121 if ($self->rewrite_root) {
3122 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3123 "options set!\n";
3124 }
3125 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3126 # we don't want "SVM: initializing mirror for junk" ...
3127 return undef if $r == 0;
3128 my $svm = $self->svm;
3129 if ($uuid ne $svm->{uuid}) {
3130 die "UUID mismatch on SVM path:\n",
3131 "expected: $svm->{uuid}\n",
3132 " got: $uuid\n";
3133 }
3134 my $full_url = $self->full_url;
3135 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3136 die "Failed to replace '$svm->{replace}' with ",
3137 "'$svm->{source}' in $full_url\n";
3138 # throw away username for storing in records
3139 remove_username($full_url);
3140 $log_entry{metadata} = "$full_url\@$r $uuid";
3141 $log_entry{svm_revision} = $r;
3142 $email ||= "$author\@$uuid";
3143 $commit_email ||= "$author\@$uuid";
3144 } elsif ($self->use_svnsync_props) {
3145 my $full_url = $self->svnsync->{url};
3146 $full_url .= "/$self->{path}" if length $self->{path};
3147 remove_username($full_url);
3148 my $uuid = $self->svnsync->{uuid};
3149 $log_entry{metadata} = "$full_url\@$rev $uuid";
3150 $email ||= "$author\@$uuid";
3151 $commit_email ||= "$author\@$uuid";
3152 } else {
3153 my $url = $self->metadata_url;
3154 remove_username($url);
3155 $log_entry{metadata} = "$url\@$rev " .
3156 $self->ra->get_uuid;
3157 $email ||= "$author\@" . $self->ra->get_uuid;
3158 $commit_email ||= "$author\@" . $self->ra->get_uuid;
3159 }
3160 $log_entry{name} = $name;
3161 $log_entry{email} = $email;
3162 $log_entry{commit_name} = $commit_name;
3163 $log_entry{commit_email} = $commit_email;
3164 \%log_entry;
3165 }
3166
3167 sub fetch {
3168 my ($self, $min_rev, $max_rev, @parents) = @_;
3169 my ($last_rev, $last_commit) = $self->last_rev_commit;
3170 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3171 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3172 }
3173
3174 sub set_tree_cb {
3175 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3176 $self->{inject_parents} = { $rev => $tree };
3177 $self->fetch(undef, undef);
3178 }
3179
3180 sub set_tree {
3181 my ($self, $tree) = (shift, shift);
3182 my $log_entry = ::get_commit_entry($tree);
3183 unless ($self->{last_rev}) {
3184 ::fatal("Must have an existing revision to commit");
3185 }
3186 my %ed_opts = ( r => $self->{last_rev},
3187 log => $log_entry->{log},
3188 ra => $self->ra,
3189 tree_a => $self->{last_commit},
3190 tree_b => $tree,
3191 editor_cb => sub {
3192 $self->set_tree_cb($log_entry, $tree, @_) },
3193 svn_path => $self->{path} );
3194 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3195 print "No changes\nr$self->{last_rev} = $tree\n";
3196 }
3197 }
3198
3199 sub rebuild_from_rev_db {
3200 my ($self, $path) = @_;
3201 my $r = -1;
3202 open my $fh, '<', $path or croak "open: $!";
3203 binmode $fh or croak "binmode: $!";
3204 while (<$fh>) {
3205 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3206 chomp($_);
3207 ++$r;
3208 next if $_ eq ('0' x 40);
3209 $self->rev_map_set($r, $_);
3210 print "r$r = $_\n";
3211 }
3212 close $fh or croak "close: $!";
3213 unlink $path or croak "unlink: $!";
3214 }
3215
3216 sub rebuild {
3217 my ($self) = @_;
3218 my $map_path = $self->map_path;
3219 my $partial = (-e $map_path && ! -z $map_path);
3220 return unless ::verify_ref($self->refname.'^0');
3221 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3222 my $rev_db = $self->rev_db_path;
3223 $self->rebuild_from_rev_db($rev_db);
3224 if ($self->use_svm_props) {
3225 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3226 $self->rebuild_from_rev_db($svm_rev_db);
3227 }
3228 $self->unlink_rev_db_symlink;
3229 return;
3230 }
3231 print "Rebuilding $map_path ...\n" if (!$partial);
3232 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3233 (undef, undef));
3234 my ($log, $ctx) =
3235 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3236 ($head ? "$head.." : "") . $self->refname,
3237 '--');
3238 my $metadata_url = $self->metadata_url;
3239 remove_username($metadata_url);
3240 my $svn_uuid = $self->ra_uuid;
3241 my $c;
3242 while (<$log>) {
3243 if ( m{^commit ($::sha1)$} ) {
3244 $c = $1;
3245 next;
3246 }
3247 next unless s{^\s*(git-svn-id:)}{$1};
3248 my ($url, $rev, $uuid) = ::extract_metadata($_);
3249 remove_username($url);
3250
3251 # ignore merges (from set-tree)
3252 next if (!defined $rev || !$uuid);
3253
3254 # if we merged or otherwise started elsewhere, this is
3255 # how we break out of it
3256 if (($uuid ne $svn_uuid) ||
3257 ($metadata_url && $url && ($url ne $metadata_url))) {
3258 next;
3259 }
3260 if ($partial && $head) {
3261 print "Partial-rebuilding $map_path ...\n";
3262 print "Currently at $base_rev = $head\n";
3263 $head = undef;
3264 }
3265
3266 $self->rev_map_set($rev, $c);
3267 print "r$rev = $c\n";
3268 }
3269 command_close_pipe($log, $ctx);
3270 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3271 my $rev_db_path = $self->rev_db_path;
3272 if (-f $self->rev_db_path) {
3273 unlink $self->rev_db_path or croak "unlink: $!";
3274 }
3275 $self->unlink_rev_db_symlink;
3276 }
3277
3278 # rev_map:
3279 # Tie::File seems to be prone to offset errors if revisions get sparse,
3280 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
3281 # one of my favorite modules is out :< Next up would be one of the DBM
3282 # modules, but I'm not sure which is most portable...
3283 #
3284 # This is the replacement for the rev_db format, which was too big
3285 # and inefficient for large repositories with a lot of sparse history
3286 # (mainly tags)
3287 #
3288 # The format is this:
3289 # - 24 bytes for every record,
3290 # * 4 bytes for the integer representing an SVN revision number
3291 # * 20 bytes representing the sha1 of a git commit
3292 # - No empty padding records like the old format
3293 # (except the last record, which can be overwritten)
3294 # - new records are written append-only since SVN revision numbers
3295 # increase monotonically
3296 # - lookups on SVN revision number are done via a binary search
3297 # - Piping the file to xxd -c24 is a good way of dumping it for
3298 # viewing or editing (piped back through xxd -r), should the need
3299 # ever arise.
3300 # - The last record can be padding revision with an all-zero sha1
3301 # This is used to optimize fetch performance when using multiple
3302 # "fetch" directives in .git/config
3303 #
3304 # These files are disposable unless noMetadata or useSvmProps is set
3305
3306 sub _rev_map_set {
3307 my ($fh, $rev, $commit) = @_;
3308
3309 binmode $fh or croak "binmode: $!";
3310 my $size = (stat($fh))[7];
3311 ($size % 24) == 0 or croak "inconsistent size: $size";
3312
3313 my $wr_offset = 0;
3314 if ($size > 0) {
3315 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3316 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3317 $read == 24 or croak "read only $read bytes (!= 24)";
3318 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3319 if ($last_commit eq ('0' x40)) {
3320 if ($size >= 48) {
3321 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3322 $read = sysread($fh, $buf, 24) or
3323 croak "read: $!";
3324 $read == 24 or
3325 croak "read only $read bytes (!= 24)";
3326 ($last_rev, $last_commit) =
3327 unpack(rev_map_fmt, $buf);
3328 if ($last_commit eq ('0' x40)) {
3329 croak "inconsistent .rev_map\n";
3330 }
3331 }
3332 if ($last_rev >= $rev) {
3333 croak "last_rev is higher!: $last_rev >= $rev";
3334 }
3335 $wr_offset = -24;
3336 }
3337 }
3338 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3339 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3340 croak "write: $!";
3341 }
3342
3343 sub _rev_map_reset {
3344 my ($fh, $rev, $commit) = @_;
3345 my $c = _rev_map_get($fh, $rev);
3346 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3347 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3348 truncate $fh, $offset or croak "truncate: $!";
3349 }
3350
3351 sub mkfile {
3352 my ($path) = @_;
3353 unless (-e $path) {
3354 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3355 mkpath([$dir]) unless -d $dir;
3356 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3357 close $fh or die "Couldn't close (create) $path: $!\n";
3358 }
3359 }
3360
3361 sub rev_map_set {
3362 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3363 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3364 my $db = $self->map_path($uuid);
3365 my $db_lock = "$db.lock";
3366 my $sig;
3367 $update_ref ||= 0;
3368 if ($update_ref) {
3369 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3370 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3371 }
3372 mkfile($db);
3373
3374 $LOCKFILES{$db_lock} = 1;
3375 my $sync;
3376 # both of these options make our .rev_db file very, very important
3377 # and we can't afford to lose it because rebuild() won't work
3378 if ($self->use_svm_props || $self->no_metadata) {
3379 $sync = 1;
3380 copy($db, $db_lock) or die "rev_map_set(@_): ",
3381 "Failed to copy: ",
3382 "$db => $db_lock ($!)\n";
3383 } else {
3384 rename $db, $db_lock or die "rev_map_set(@_): ",
3385 "Failed to rename: ",
3386 "$db => $db_lock ($!)\n";
3387 }
3388
3389 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3390 or croak "Couldn't open $db_lock: $!\n";
3391 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3392 _rev_map_set($fh, $rev, $commit);
3393 if ($sync) {
3394 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3395 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3396 }
3397 close $fh or croak $!;
3398 if ($update_ref) {
3399 $_head = $self;
3400 my $note = "";
3401 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3402 command_noisy('update-ref', '-m', "r$rev$note",
3403 $self->refname, $commit);
3404 }
3405 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3406 "$db_lock => $db ($!)\n";
3407 delete $LOCKFILES{$db_lock};
3408 if ($update_ref) {
3409 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3410 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3411 kill $sig, $$ if defined $sig;
3412 }
3413 }
3414
3415 # If want_commit, this will return an array of (rev, commit) where
3416 # commit _must_ be a valid commit in the archive.
3417 # Otherwise, it'll return the max revision (whether or not the
3418 # commit is valid or just a 0x40 placeholder).
3419 sub rev_map_max {
3420 my ($self, $want_commit) = @_;
3421 $self->rebuild;
3422 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3423 $want_commit ? ($r, $c) : $r;
3424 }
3425
3426 sub rev_map_max_norebuild {
3427 my ($self, $want_commit) = @_;
3428 my $map_path = $self->map_path;
3429 stat $map_path or return $want_commit ? (0, undef) : 0;
3430 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3431 binmode $fh or croak "binmode: $!";
3432 my $size = (stat($fh))[7];
3433 ($size % 24) == 0 or croak "inconsistent size: $size";
3434
3435 if ($size == 0) {
3436 close $fh or croak "close: $!";
3437 return $want_commit ? (0, undef) : 0;
3438 }
3439
3440 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3441 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3442 my ($r, $c) = unpack(rev_map_fmt, $buf);
3443 if ($want_commit && $c eq ('0' x40)) {
3444 if ($size < 48) {
3445 return $want_commit ? (0, undef) : 0;
3446 }
3447 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3448 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3449 ($r, $c) = unpack(rev_map_fmt, $buf);
3450 if ($c eq ('0'x40)) {
3451 croak "Penultimate record is all-zeroes in $map_path";
3452 }
3453 }
3454 close $fh or croak "close: $!";
3455 $want_commit ? ($r, $c) : $r;
3456 }
3457
3458 sub rev_map_get {
3459 my ($self, $rev, $uuid) = @_;
3460 my $map_path = $self->map_path($uuid);
3461 return undef unless -e $map_path;
3462
3463 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3464 my $c = _rev_map_get($fh, $rev);
3465 close($fh) or croak "close: $!";
3466 $c
3467 }
3468
3469 sub _rev_map_get {
3470 my ($fh, $rev) = @_;
3471
3472 binmode $fh or croak "binmode: $!";
3473 my $size = (stat($fh))[7];
3474 ($size % 24) == 0 or croak "inconsistent size: $size";
3475
3476 if ($size == 0) {
3477 return undef;
3478 }
3479
3480 my ($l, $u) = (0, $size - 24);
3481 my ($r, $c, $buf);
3482
3483 while ($l <= $u) {
3484 my $i = int(($l/24 + $u/24) / 2) * 24;
3485 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3486 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3487 my ($r, $c) = unpack(rev_map_fmt, $buf);
3488
3489 if ($r < $rev) {
3490 $l = $i + 24;
3491 } elsif ($r > $rev) {
3492 $u = $i - 24;
3493 } else { # $r == $rev
3494 return $c eq ('0' x 40) ? undef : $c;
3495 }
3496 }
3497 undef;
3498 }
3499
3500 # Finds the first svn revision that exists on (if $eq_ok is true) or
3501 # before $rev for the current branch. It will not search any lower
3502 # than $min_rev. Returns the git commit hash and svn revision number
3503 # if found, else (undef, undef).
3504 sub find_rev_before {
3505 my ($self, $rev, $eq_ok, $min_rev) = @_;
3506 --$rev unless $eq_ok;
3507 $min_rev ||= 1;
3508 my $max_rev = $self->rev_map_max;
3509 $rev = $max_rev if ($rev > $max_rev);
3510 while ($rev >= $min_rev) {
3511 if (my $c = $self->rev_map_get($rev)) {
3512 return ($rev, $c);
3513 }
3514 --$rev;
3515 }
3516 return (undef, undef);
3517 }
3518
3519 # Finds the first svn revision that exists on (if $eq_ok is true) or
3520 # after $rev for the current branch. It will not search any higher
3521 # than $max_rev. Returns the git commit hash and svn revision number
3522 # if found, else (undef, undef).
3523 sub find_rev_after {
3524 my ($self, $rev, $eq_ok, $max_rev) = @_;
3525 ++$rev unless $eq_ok;
3526 $max_rev ||= $self->rev_map_max;
3527 while ($rev <= $max_rev) {
3528 if (my $c = $self->rev_map_get($rev)) {
3529 return ($rev, $c);
3530 }
3531 ++$rev;
3532 }
3533 return (undef, undef);
3534 }
3535
3536 sub _new {
3537 my ($class, $repo_id, $ref_id, $path) = @_;
3538 unless (defined $repo_id && length $repo_id) {
3539 $repo_id = $Git::SVN::default_repo_id;
3540 }
3541 unless (defined $ref_id && length $ref_id) {
3542 $_prefix = '' unless defined($_prefix);
3543 $_[2] = $ref_id =
3544 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3545 }
3546 $_[1] = $repo_id;
3547 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3548
3549 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3550 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3551 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3552 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3553 if (-d $old_dir && ! -d $dir) {
3554 $dir = $old_dir;
3555 }
3556 }
3557
3558 $_[3] = $path = '' unless (defined $path);
3559 mkpath([$dir]);
3560 bless {
3561 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3562 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3563 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3564 }
3565
3566 # for read-only access of old .rev_db formats
3567 sub unlink_rev_db_symlink {
3568 my ($self) = @_;
3569 my $link = $self->rev_db_path;
3570 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3571 if (-l $link) {
3572 unlink $link or croak "unlink: $link failed!";
3573 }
3574 }
3575
3576 sub rev_db_path {
3577 my ($self, $uuid) = @_;
3578 my $db_path = $self->map_path($uuid);
3579 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3580 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3581 $db_path;
3582 }
3583
3584 # the new replacement for .rev_db
3585 sub map_path {
3586 my ($self, $uuid) = @_;
3587 $uuid ||= $self->ra_uuid;
3588 "$self->{map_root}.$uuid";
3589 }
3590
3591 sub uri_encode {
3592 my ($f) = @_;
3593 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3594 $f
3595 }
3596
3597 sub uri_decode {
3598 my ($f) = @_;
3599 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3600 $f
3601 }
3602
3603 sub remove_username {
3604 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3605 }
3606
3607 package Git::SVN::Prompt;
3608 use strict;
3609 use warnings;
3610 require SVN::Core;
3611 use vars qw/$_no_auth_cache $_username/;
3612
3613 sub simple {
3614 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3615 $may_save = undef if $_no_auth_cache;
3616 $default_username = $_username if defined $_username;
3617 if (defined $default_username && length $default_username) {
3618 if (defined $realm && length $realm) {
3619 print STDERR "Authentication realm: $realm\n";
3620 STDERR->flush;
3621 }
3622 $cred->username($default_username);
3623 } else {
3624 username($cred, $realm, $may_save, $pool);
3625 }
3626 $cred->password(_read_password("Password for '" .
3627 $cred->username . "': ", $realm));
3628 $cred->may_save($may_save);
3629 $SVN::_Core::SVN_NO_ERROR;
3630 }
3631
3632 sub ssl_server_trust {
3633 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3634 $may_save = undef if $_no_auth_cache;
3635 print STDERR "Error validating server certificate for '$realm':\n";
3636 {
3637 no warnings 'once';
3638 # All variables SVN::Auth::SSL::* are used only once,
3639 # so we're shutting up Perl warnings about this.
3640 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3641 print STDERR " - The certificate is not issued ",
3642 "by a trusted authority. Use the\n",
3643 " fingerprint to validate ",
3644 "the certificate manually!\n";
3645 }
3646 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3647 print STDERR " - The certificate hostname ",
3648 "does not match.\n";
3649 }
3650 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3651 print STDERR " - The certificate is not yet valid.\n";
3652 }
3653 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3654 print STDERR " - The certificate has expired.\n";
3655 }
3656 if ($failures & $SVN::Auth::SSL::OTHER) {
3657 print STDERR " - The certificate has ",
3658 "an unknown error.\n";
3659 }
3660 } # no warnings 'once'
3661 printf STDERR
3662 "Certificate information:\n".
3663 " - Hostname: %s\n".
3664 " - Valid: from %s until %s\n".
3665 " - Issuer: %s\n".
3666 " - Fingerprint: %s\n",
3667 map $cert_info->$_, qw(hostname valid_from valid_until
3668 issuer_dname fingerprint);
3669 my $choice;
3670 prompt:
3671 print STDERR $may_save ?
3672 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3673 "(R)eject or accept (t)emporarily? ";
3674 STDERR->flush;
3675 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3676 if ($choice =~ /^t$/i) {
3677 $cred->may_save(undef);
3678 } elsif ($choice =~ /^r$/i) {
3679 return -1;
3680 } elsif ($may_save && $choice =~ /^p$/i) {
3681 $cred->may_save($may_save);
3682 } else {
3683 goto prompt;
3684 }
3685 $cred->accepted_failures($failures);
3686 $SVN::_Core::SVN_NO_ERROR;
3687 }
3688
3689 sub ssl_client_cert {
3690 my ($cred, $realm, $may_save, $pool) = @_;
3691 $may_save = undef if $_no_auth_cache;
3692 print STDERR "Client certificate filename: ";
3693 STDERR->flush;
3694 chomp(my $filename = <STDIN>);
3695 $cred->cert_file($filename);
3696 $cred->may_save($may_save);
3697 $SVN::_Core::SVN_NO_ERROR;
3698 }
3699
3700 sub ssl_client_cert_pw {
3701 my ($cred, $realm, $may_save, $pool) = @_;
3702 $may_save = undef if $_no_auth_cache;
3703 $cred->password(_read_password("Password: ", $realm));
3704 $cred->may_save($may_save);
3705 $SVN::_Core::SVN_NO_ERROR;
3706 }
3707
3708 sub username {
3709 my ($cred, $realm, $may_save, $pool) = @_;
3710 $may_save = undef if $_no_auth_cache;
3711 if (defined $realm && length $realm) {
3712 print STDERR "Authentication realm: $realm\n";
3713 }
3714 my $username;
3715 if (defined $_username) {
3716 $username = $_username;
3717 } else {
3718 print STDERR "Username: ";
3719 STDERR->flush;
3720 chomp($username = <STDIN>);
3721 }
3722 $cred->username($username);
3723 $cred->may_save($may_save);
3724 $SVN::_Core::SVN_NO_ERROR;
3725 }
3726
3727 sub _read_password {
3728 my ($prompt, $realm) = @_;
3729 print STDERR $prompt;
3730 STDERR->flush;
3731 require Term::ReadKey;
3732 Term::ReadKey::ReadMode('noecho');
3733 my $password = '';
3734 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3735 last if $key =~ /[\012\015]/; # \n\r
3736 $password .= $key;
3737 }
3738 Term::ReadKey::ReadMode('restore');
3739 print STDERR "\n";
3740 STDERR->flush;
3741 $password;
3742 }
3743
3744 package SVN::Git::Fetcher;
3745 use vars qw/@ISA/;
3746 use strict;
3747 use warnings;
3748 use Carp qw/croak/;
3749 use File::Temp qw/tempfile/;
3750 use IO::File qw//;
3751 use vars qw/$_ignore_regex/;
3752
3753 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
3754 sub new {
3755 my ($class, $git_svn, $switch_path) = @_;
3756 my $self = SVN::Delta::Editor->new;
3757 bless $self, $class;
3758 if (exists $git_svn->{last_commit}) {
3759 $self->{c} = $git_svn->{last_commit};
3760 $self->{empty_symlinks} =
3761 _mark_empty_symlinks($git_svn, $switch_path);
3762 }
3763 $self->{ignore_regex} = eval { command_oneline('config', '--get',
3764 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3765 $self->{empty} = {};
3766 $self->{dir_prop} = {};
3767 $self->{file_prop} = {};
3768 $self->{absent_dir} = {};
3769 $self->{absent_file} = {};
3770 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3771 $self;
3772 }
3773
3774 # this uses the Ra object, so it must be called before do_{switch,update},
3775 # not inside them (when the Git::SVN::Fetcher object is passed) to
3776 # do_{switch,update}
3777 sub _mark_empty_symlinks {
3778 my ($git_svn, $switch_path) = @_;
3779 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3780 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3781
3782 my %ret;
3783 my ($rev, $cmt) = $git_svn->last_rev_commit;
3784 return {} unless ($rev && $cmt);
3785
3786 # allow the warning to be printed for each revision we fetch to
3787 # ensure the user sees it. The user can also disable the workaround
3788 # on the repository even while git svn is running and the next
3789 # revision fetched will skip this expensive function.
3790 my $printed_warning;
3791 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3792 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3793 local $/ = "\0";
3794 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3795 $pfx .= '/' if length($pfx);
3796 while (<$ls>) {
3797 chomp;
3798 s/\A100644 blob $empty_blob\t//o or next;
3799 unless ($printed_warning) {
3800 print STDERR "Scanning for empty symlinks, ",
3801 "this may take a while if you have ",
3802 "many empty files\n",
3803 "You may disable this with `",
3804 "git config svn.brokenSymlinkWorkaround ",
3805 "false'.\n",
3806 "This may be done in a different ",
3807 "terminal without restarting ",
3808 "git svn\n";
3809 $printed_warning = 1;
3810 }
3811 my $path = $_;
3812 my (undef, $props) =
3813 $git_svn->ra->get_file($pfx.$path, $rev, undef);
3814 if ($props->{'svn:special'}) {
3815 $ret{$path} = 1;
3816 }
3817 }
3818 command_close_pipe($ls, $ctx);
3819 \%ret;
3820 }
3821
3822 # returns true if a given path is inside a ".git" directory
3823 sub in_dot_git {
3824 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3825 }
3826
3827 # return value: 0 -- don't ignore, 1 -- ignore
3828 sub is_path_ignored {
3829 my ($self, $path) = @_;
3830 return 1 if in_dot_git($path);
3831 return 1 if defined($self->{ignore_regex}) &&
3832 $path =~ m!$self->{ignore_regex}!;
3833 return 0 unless defined($_ignore_regex);
3834 return 1 if $path =~ m!$_ignore_regex!o;
3835 return 0;
3836 }
3837
3838 sub set_path_strip {
3839 my ($self, $path) = @_;
3840 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3841 }
3842
3843 sub open_root {
3844 { path => '' };
3845 }
3846
3847 sub open_directory {
3848 my ($self, $path, $pb, $rev) = @_;
3849 { path => $path };
3850 }
3851
3852 sub git_path {
3853 my ($self, $path) = @_;
3854 if ($self->{path_strip}) {
3855 $path =~ s!$self->{path_strip}!! or
3856 die "Failed to strip path '$path' ($self->{path_strip})\n";
3857 }
3858 $path;
3859 }
3860
3861 sub delete_entry {
3862 my ($self, $path, $rev, $pb) = @_;
3863 return undef if $self->is_path_ignored($path);
3864
3865 my $gpath = $self->git_path($path);
3866 return undef if ($gpath eq '');
3867
3868 # remove entire directories.
3869 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3870 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3871 if ($tree) {
3872 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3873 -r --name-only -z/,
3874 $tree);
3875 local $/ = "\0";
3876 while (<$ls>) {
3877 chomp;
3878 my $rmpath = "$gpath/$_";
3879 $self->{gii}->remove($rmpath);
3880 print "\tD\t$rmpath\n" unless $::_q;
3881 }
3882 print "\tD\t$gpath/\n" unless $::_q;
3883 command_close_pipe($ls, $ctx);
3884 $self->{empty}->{$path} = 0
3885 } else {
3886 $self->{gii}->remove($gpath);
3887 print "\tD\t$gpath\n" unless $::_q;
3888 }
3889 undef;
3890 }
3891
3892 sub open_file {
3893 my ($self, $path, $pb, $rev) = @_;
3894 my ($mode, $blob);
3895
3896 goto out if $self->is_path_ignored($path);
3897
3898 my $gpath = $self->git_path($path);
3899 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3900 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3901 unless (defined $mode && defined $blob) {
3902 die "$path was not found in commit $self->{c} (r$rev)\n";
3903 }
3904 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3905 $mode = '120000';
3906 }
3907 out:
3908 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3909 pool => SVN::Pool->new, action => 'M' };
3910 }
3911
3912 sub add_file {
3913 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3914 my $mode;
3915
3916 if (!$self->is_path_ignored($path)) {
3917 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3918 delete $self->{empty}->{$dir};
3919 $mode = '100644';
3920 }
3921 { path => $path, mode_a => $mode, mode_b => $mode,
3922 pool => SVN::Pool->new, action => 'A' };
3923 }
3924
3925 sub add_directory {
3926 my ($self, $path, $cp_path, $cp_rev) = @_;
3927 goto out if $self->is_path_ignored($path);
3928 my $gpath = $self->git_path($path);
3929 if ($gpath eq '') {
3930 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3931 -r --name-only -z/,
3932 $self->{c});
3933 local $/ = "\0";
3934 while (<$ls>) {
3935 chomp;
3936 $self->{gii}->remove($_);
3937 print "\tD\t$_\n" unless $::_q;
3938 }
3939 command_close_pipe($ls, $ctx);
3940 $self->{empty}->{$path} = 0;
3941 }
3942 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3943 delete $self->{empty}->{$dir};
3944 $self->{empty}->{$path} = 1;
3945 out:
3946 { path => $path };
3947 }
3948
3949 sub change_dir_prop {
3950 my ($self, $db, $prop, $value) = @_;
3951 return undef if $self->is_path_ignored($db->{path});
3952 $self->{dir_prop}->{$db->{path}} ||= {};
3953 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3954 undef;
3955 }
3956
3957 sub absent_directory {
3958 my ($self, $path, $pb) = @_;
3959 return undef if $self->is_path_ignored($path);
3960 $self->{absent_dir}->{$pb->{path}} ||= [];
3961 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3962 undef;
3963 }
3964
3965 sub absent_file {
3966 my ($self, $path, $pb) = @_;
3967 return undef if $self->is_path_ignored($path);
3968 $self->{absent_file}->{$pb->{path}} ||= [];
3969 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3970 undef;
3971 }
3972
3973 sub change_file_prop {
3974 my ($self, $fb, $prop, $value) = @_;
3975 return undef if $self->is_path_ignored($fb->{path});
3976 if ($prop eq 'svn:executable') {
3977 if ($fb->{mode_b} != 120000) {
3978 $fb->{mode_b} = defined $value ? 100755 : 100644;
3979 }
3980 } elsif ($prop eq 'svn:special') {
3981 $fb->{mode_b} = defined $value ? 120000 : 100644;
3982 } else {
3983 $self->{file_prop}->{$fb->{path}} ||= {};
3984 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3985 }
3986 undef;
3987 }
3988
3989 sub apply_textdelta {
3990 my ($self, $fb, $exp) = @_;
3991 return undef if $self->is_path_ignored($fb->{path});
3992 my $fh = $::_repository->temp_acquire('svn_delta');
3993 # $fh gets auto-closed() by SVN::TxDelta::apply(),
3994 # (but $base does not,) so dup() it for reading in close_file
3995 open my $dup, '<&', $fh or croak $!;
3996 my $base = $::_repository->temp_acquire('git_blob');
3997
3998 if ($fb->{blob}) {
3999 my ($base_is_link, $size);
4000
4001 if ($fb->{mode_a} eq '120000' &&
4002 ! $self->{empty_symlinks}->{$fb->{path}}) {
4003 print $base 'link ' or die "print $!\n";
4004 $base_is_link = 1;
4005 }
4006 retry:
4007 $size = $::_repository->cat_blob($fb->{blob}, $base);
4008 die "Failed to read object $fb->{blob}" if ($size < 0);
4009
4010 if (defined $exp) {
4011 seek $base, 0, 0 or croak $!;
4012 my $got = ::md5sum($base);
4013 if ($got ne $exp) {
4014 my $err = "Checksum mismatch: ".
4015 "$fb->{path} $fb->{blob}\n" .
4016 "expected: $exp\n" .
4017 " got: $got\n";
4018 if ($base_is_link) {
4019 warn $err,
4020 "Retrying... (possibly ",
4021 "a bad symlink from SVN)\n";
4022 $::_repository->temp_reset($base);
4023 $base_is_link = 0;
4024 goto retry;
4025 }
4026 die $err;
4027 }
4028 }
4029 }
4030 seek $base, 0, 0 or croak $!;
4031 $fb->{fh} = $fh;
4032 $fb->{base} = $base;
4033 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4034 }
4035
4036 sub close_file {
4037 my ($self, $fb, $exp) = @_;
4038 return undef if $self->is_path_ignored($fb->{path});
4039
4040 my $hash;
4041 my $path = $self->git_path($fb->{path});
4042 if (my $fh = $fb->{fh}) {
4043 if (defined $exp) {
4044 seek($fh, 0, 0) or croak $!;
4045 my $got = ::md5sum($fh);
4046 if ($got ne $exp) {
4047 die "Checksum mismatch: $path\n",
4048 "expected: $exp\n got: $got\n";
4049 }
4050 }
4051 if ($fb->{mode_b} == 120000) {
4052 sysseek($fh, 0, 0) or croak $!;
4053 my $rd = sysread($fh, my $buf, 5);
4054
4055 if (!defined $rd) {
4056 croak "sysread: $!\n";
4057 } elsif ($rd == 0) {
4058 warn "$path has mode 120000",
4059 " but it points to nothing\n",
4060 "converting to an empty file with mode",
4061 " 100644\n";
4062 $fb->{mode_b} = '100644';
4063 } elsif ($buf ne 'link ') {
4064 warn "$path has mode 120000",
4065 " but is not a link\n";
4066 } else {
4067 my $tmp_fh = $::_repository->temp_acquire(
4068 'svn_hash');
4069 my $res;
4070 while ($res = sysread($fh, my $str, 1024)) {
4071 my $out = syswrite($tmp_fh, $str, $res);
4072 defined($out) && $out == $res
4073 or croak("write ",
4074 Git::temp_path($tmp_fh),
4075 ": $!\n");
4076 }
4077 defined $res or croak $!;
4078
4079 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4080 Git::temp_release($tmp_fh, 1);
4081 }
4082 }
4083
4084 $hash = $::_repository->hash_and_insert_object(
4085 Git::temp_path($fh));
4086 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4087
4088 Git::temp_release($fb->{base}, 1);
4089 Git::temp_release($fh, 1);
4090 } else {
4091 $hash = $fb->{blob} or die "no blob information\n";
4092 }
4093 $fb->{pool}->clear;
4094 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4095 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4096 undef;
4097 }
4098
4099 sub abort_edit {
4100 my $self = shift;
4101 $self->{nr} = $self->{gii}->{nr};
4102 delete $self->{gii};
4103 $self->SUPER::abort_edit(@_);
4104 }
4105
4106 sub close_edit {
4107 my $self = shift;
4108 $self->{git_commit_ok} = 1;
4109 $self->{nr} = $self->{gii}->{nr};
4110 delete $self->{gii};
4111 $self->SUPER::close_edit(@_);
4112 }
4113
4114 package SVN::Git::Editor;
4115 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4116 use strict;
4117 use warnings;
4118 use Carp qw/croak/;
4119 use IO::File;
4120
4121 sub new {
4122 my ($class, $opts) = @_;
4123 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4124 die "$_ required!\n" unless (defined $opts->{$_});
4125 }
4126
4127 my $pool = SVN::Pool->new;
4128 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4129 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4130 $opts->{r}, $mods);
4131
4132 # $opts->{ra} functions should not be used after this:
4133 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4134 $opts->{editor_cb}, $pool);
4135 my $self = SVN::Delta::Editor->new(@ce, $pool);
4136 bless $self, $class;
4137 foreach (qw/svn_path r tree_a tree_b/) {
4138 $self->{$_} = $opts->{$_};
4139 }
4140 $self->{url} = $opts->{ra}->{url};
4141 $self->{mods} = $mods;
4142 $self->{types} = $types;
4143 $self->{pool} = $pool;
4144 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4145 $self->{rm} = { };
4146 $self->{path_prefix} = length $self->{svn_path} ?
4147 "$self->{svn_path}/" : '';
4148 $self->{config} = $opts->{config};
4149 return $self;
4150 }
4151
4152 sub generate_diff {
4153 my ($tree_a, $tree_b) = @_;
4154 my @diff_tree = qw(diff-tree -z -r);
4155 if ($_cp_similarity) {
4156 push @diff_tree, "-C$_cp_similarity";
4157 } else {
4158 push @diff_tree, '-C';
4159 }
4160 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4161 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4162 push @diff_tree, $tree_a, $tree_b;
4163 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4164 local $/ = "\0";
4165 my $state = 'meta';
4166 my @mods;
4167 while (<$diff_fh>) {
4168 chomp $_; # this gets rid of the trailing "\0"
4169 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4170 ($::sha1)\s($::sha1)\s
4171 ([MTCRAD])\d*$/xo) {
4172 push @mods, { mode_a => $1, mode_b => $2,
4173 sha1_a => $3, sha1_b => $4,
4174 chg => $5 };
4175 if ($5 =~ /^(?:C|R)$/) {
4176 $state = 'file_a';
4177 } else {
4178 $state = 'file_b';
4179 }
4180 } elsif ($state eq 'file_a') {
4181 my $x = $mods[$#mods] or croak "Empty array\n";
4182 if ($x->{chg} !~ /^(?:C|R)$/) {
4183 croak "Error parsing $_, $x->{chg}\n";
4184 }
4185 $x->{file_a} = $_;
4186 $state = 'file_b';
4187 } elsif ($state eq 'file_b') {
4188 my $x = $mods[$#mods] or croak "Empty array\n";
4189 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4190 croak "Error parsing $_, $x->{chg}\n";
4191 }
4192 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4193 croak "Error parsing $_, $x->{chg}\n";
4194 }
4195 $x->{file_b} = $_;
4196 $state = 'meta';
4197 } else {
4198 croak "Error parsing $_\n";
4199 }
4200 }
4201 command_close_pipe($diff_fh, $ctx);
4202 \@mods;
4203 }
4204
4205 sub check_diff_paths {
4206 my ($ra, $pfx, $rev, $mods) = @_;
4207 my %types;
4208 $pfx .= '/' if length $pfx;
4209
4210 sub type_diff_paths {
4211 my ($ra, $types, $path, $rev) = @_;
4212 my @p = split m#/+#, $path;
4213 my $c = shift @p;
4214 unless (defined $types->{$c}) {
4215 $types->{$c} = $ra->check_path($c, $rev);
4216 }
4217 while (@p) {
4218 $c .= '/' . shift @p;
4219 next if defined $types->{$c};
4220 $types->{$c} = $ra->check_path($c, $rev);
4221 }
4222 }
4223
4224 foreach my $m (@$mods) {
4225 foreach my $f (qw/file_a file_b/) {
4226 next unless defined $m->{$f};
4227 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4228 if (length $pfx.$dir && ! defined $types{$dir}) {
4229 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4230 }
4231 }
4232 }
4233 \%types;
4234 }
4235
4236 sub split_path {
4237 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4238 }
4239
4240 sub repo_path {
4241 my ($self, $path) = @_;
4242 $self->{path_prefix}.(defined $path ? $path : '');
4243 }
4244
4245 sub url_path {
4246 my ($self, $path) = @_;
4247 if ($self->{url} =~ m#^https?://#) {
4248 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4249 }
4250 $self->{url} . '/' . $self->repo_path($path);
4251 }
4252
4253 sub rmdirs {
4254 my ($self) = @_;
4255 my $rm = $self->{rm};
4256 delete $rm->{''}; # we never delete the url we're tracking
4257 return unless %$rm;
4258
4259 foreach (keys %$rm) {
4260 my @d = split m#/#, $_;
4261 my $c = shift @d;
4262 $rm->{$c} = 1;
4263 while (@d) {
4264 $c .= '/' . shift @d;
4265 $rm->{$c} = 1;
4266 }
4267 }
4268 delete $rm->{$self->{svn_path}};
4269 delete $rm->{''}; # we never delete the url we're tracking
4270 return unless %$rm;
4271
4272 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4273 $self->{tree_b});
4274 local $/ = "\0";
4275 while (<$fh>) {
4276 chomp;
4277 my @dn = split m#/#, $_;
4278 while (pop @dn) {
4279 delete $rm->{join '/', @dn};
4280 }
4281 unless (%$rm) {
4282 close $fh;
4283 return;
4284 }
4285 }
4286 command_close_pipe($fh, $ctx);
4287
4288 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4289 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4290 $self->close_directory($bat->{$d}, $p);
4291 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4292 print "\tD+\t$d/\n" unless $::_q;
4293 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4294 delete $bat->{$d};
4295 }
4296 }
4297
4298 sub open_or_add_dir {
4299 my ($self, $full_path, $baton) = @_;
4300 my $t = $self->{types}->{$full_path};
4301 if (!defined $t) {
4302 die "$full_path not known in r$self->{r} or we have a bug!\n";
4303 }
4304 {
4305 no warnings 'once';
4306 # SVN::Node::none and SVN::Node::file are used only once,
4307 # so we're shutting up Perl's warnings about them.
4308 if ($t == $SVN::Node::none) {
4309 return $self->add_directory($full_path, $baton,
4310 undef, -1, $self->{pool});
4311 } elsif ($t == $SVN::Node::dir) {
4312 return $self->open_directory($full_path, $baton,
4313 $self->{r}, $self->{pool});
4314 } # no warnings 'once'
4315 print STDERR "$full_path already exists in repository at ",
4316 "r$self->{r} and it is not a directory (",
4317 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4318 } # no warnings 'once'
4319 exit 1;
4320 }
4321
4322 sub ensure_path {
4323 my ($self, $path) = @_;
4324 my $bat = $self->{bat};
4325 my $repo_path = $self->repo_path($path);
4326 return $bat->{''} unless (length $repo_path);
4327 my @p = split m#/+#, $repo_path;
4328 my $c = shift @p;
4329 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4330 while (@p) {
4331 my $c0 = $c;
4332 $c .= '/' . shift @p;
4333 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4334 }
4335 return $bat->{$c};
4336 }
4337
4338 # Subroutine to convert a globbing pattern to a regular expression.
4339 # From perl cookbook.
4340 sub glob2pat {
4341 my $globstr = shift;
4342 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4343 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4344 return '^' . $globstr . '$';
4345 }
4346
4347 sub check_autoprop {
4348 my ($self, $pattern, $properties, $file, $fbat) = @_;
4349 # Convert the globbing pattern to a regular expression.
4350 my $regex = glob2pat($pattern);
4351 # Check if the pattern matches the file name.
4352 if($file =~ m/($regex)/) {
4353 # Parse the list of properties to set.
4354 my @props = split(/;/, $properties);
4355 foreach my $prop (@props) {
4356 # Parse 'name=value' syntax and set the property.
4357 if ($prop =~ /([^=]+)=(.*)/) {
4358 my ($n,$v) = ($1,$2);
4359 for ($n, $v) {
4360 s/^\s+//; s/\s+$//;
4361 }
4362 $self->change_file_prop($fbat, $n, $v);
4363 }
4364 }
4365 }
4366 }
4367
4368 sub apply_autoprops {
4369 my ($self, $file, $fbat) = @_;
4370 my $conf_t = ${$self->{config}}{'config'};
4371 no warnings 'once';
4372 # Check [miscellany]/enable-auto-props in svn configuration.
4373 if (SVN::_Core::svn_config_get_bool(
4374 $conf_t,
4375 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4376 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4377 0)) {
4378 # Auto-props are enabled. Enumerate them to look for matches.
4379 my $callback = sub {
4380 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4381 };
4382 SVN::_Core::svn_config_enumerate(
4383 $conf_t,
4384 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4385 $callback);
4386 }
4387 }
4388
4389 sub A {
4390 my ($self, $m) = @_;
4391 my ($dir, $file) = split_path($m->{file_b});
4392 my $pbat = $self->ensure_path($dir);
4393 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4394 undef, -1);
4395 print "\tA\t$m->{file_b}\n" unless $::_q;
4396 $self->apply_autoprops($file, $fbat);
4397 $self->chg_file($fbat, $m);
4398 $self->close_file($fbat,undef,$self->{pool});
4399 }
4400
4401 sub C {
4402 my ($self, $m) = @_;
4403 my ($dir, $file) = split_path($m->{file_b});
4404 my $pbat = $self->ensure_path($dir);
4405 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4406 $self->url_path($m->{file_a}), $self->{r});
4407 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4408 $self->chg_file($fbat, $m);
4409 $self->close_file($fbat,undef,$self->{pool});
4410 }
4411
4412 sub delete_entry {
4413 my ($self, $path, $pbat) = @_;
4414 my $rpath = $self->repo_path($path);
4415 my ($dir, $file) = split_path($rpath);
4416 $self->{rm}->{$dir} = 1;
4417 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4418 }
4419
4420 sub R {
4421 my ($self, $m) = @_;
4422 my ($dir, $file) = split_path($m->{file_b});
4423 my $pbat = $self->ensure_path($dir);
4424 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4425 $self->url_path($m->{file_a}), $self->{r});
4426 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4427 $self->apply_autoprops($file, $fbat);
4428 $self->chg_file($fbat, $m);
4429 $self->close_file($fbat,undef,$self->{pool});
4430
4431 ($dir, $file) = split_path($m->{file_a});
4432 $pbat = $self->ensure_path($dir);
4433 $self->delete_entry($m->{file_a}, $pbat);
4434 }
4435
4436 sub M {
4437 my ($self, $m) = @_;
4438 my ($dir, $file) = split_path($m->{file_b});
4439 my $pbat = $self->ensure_path($dir);
4440 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4441 $pbat,$self->{r},$self->{pool});
4442 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4443 $self->chg_file($fbat, $m);
4444 $self->close_file($fbat,undef,$self->{pool});
4445 }
4446
4447 sub T { shift->M(@_) }
4448
4449 sub change_file_prop {
4450 my ($self, $fbat, $pname, $pval) = @_;
4451 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4452 }
4453
4454 sub _chg_file_get_blob ($$$$) {
4455 my ($self, $fbat, $m, $which) = @_;
4456 my $fh = $::_repository->temp_acquire("git_blob_$which");
4457 if ($m->{"mode_$which"} =~ /^120/) {
4458 print $fh 'link ' or croak $!;
4459 $self->change_file_prop($fbat,'svn:special','*');
4460 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4461 $self->change_file_prop($fbat,'svn:special',undef);
4462 }
4463 my $blob = $m->{"sha1_$which"};
4464 return ($fh,) if ($blob =~ /^0{40}$/);
4465 my $size = $::_repository->cat_blob($blob, $fh);
4466 croak "Failed to read object $blob" if ($size < 0);
4467 $fh->flush == 0 or croak $!;
4468 seek $fh, 0, 0 or croak $!;
4469
4470 my $exp = ::md5sum($fh);
4471 seek $fh, 0, 0 or croak $!;
4472 return ($fh, $exp);
4473 }
4474
4475 sub chg_file {
4476 my ($self, $fbat, $m) = @_;
4477 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4478 $self->change_file_prop($fbat,'svn:executable','*');
4479 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4480 $self->change_file_prop($fbat,'svn:executable',undef);
4481 }
4482 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4483 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4484 my $pool = SVN::Pool->new;
4485 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4486 if (-s $fh_a) {
4487 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4488 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4489 if (defined $res) {
4490 die "Unexpected result from send_txstream: $res\n",
4491 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4492 }
4493 } else {
4494 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4495 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4496 if ($got ne $exp_b);
4497 }
4498 Git::temp_release($fh_b, 1);
4499 Git::temp_release($fh_a, 1);
4500 $pool->clear;
4501 }
4502
4503 sub D {
4504 my ($self, $m) = @_;
4505 my ($dir, $file) = split_path($m->{file_b});
4506 my $pbat = $self->ensure_path($dir);
4507 print "\tD\t$m->{file_b}\n" unless $::_q;
4508 $self->delete_entry($m->{file_b}, $pbat);
4509 }
4510
4511 sub close_edit {
4512 my ($self) = @_;
4513 my ($p,$bat) = ($self->{pool}, $self->{bat});
4514 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4515 next if $_ eq '';
4516 $self->close_directory($bat->{$_}, $p);
4517 }
4518 $self->close_directory($bat->{''}, $p);
4519 $self->SUPER::close_edit($p);
4520 $p->clear;
4521 }
4522
4523 sub abort_edit {
4524 my ($self) = @_;
4525 $self->SUPER::abort_edit($self->{pool});
4526 }
4527
4528 sub DESTROY {
4529 my $self = shift;
4530 $self->SUPER::DESTROY(@_);
4531 $self->{pool}->clear;
4532 }
4533
4534 # this drives the editor
4535 sub apply_diff {
4536 my ($self) = @_;
4537 my $mods = $self->{mods};
4538 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4539 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4540 my $f = $m->{chg};
4541 if (defined $o{$f}) {
4542 $self->$f($m);
4543 } else {
4544 fatal("Invalid change type: $f");
4545 }
4546 }
4547 $self->rmdirs if $_rmdir;
4548 if (@$mods == 0) {
4549 $self->abort_edit;
4550 } else {
4551 $self->close_edit;
4552 }
4553 return scalar @$mods;
4554 }
4555
4556 package Git::SVN::Ra;
4557 use vars qw/@ISA $config_dir $_log_window_size/;
4558 use strict;
4559 use warnings;
4560 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4561
4562 BEGIN {
4563 # enforce temporary pool usage for some simple functions
4564 no strict 'refs';
4565 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4566 get_file/) {
4567 my $SUPER = "SUPER::$f";
4568 *$f = sub {
4569 my $self = shift;
4570 my $pool = SVN::Pool->new;
4571 my @ret = $self->$SUPER(@_,$pool);
4572 $pool->clear;
4573 wantarray ? @ret : $ret[0];
4574 };
4575 }
4576 }
4577
4578 sub _auth_providers () {
4579 [
4580 SVN::Client::get_simple_provider(),
4581 SVN::Client::get_ssl_server_trust_file_provider(),
4582 SVN::Client::get_simple_prompt_provider(
4583 \&Git::SVN::Prompt::simple, 2),
4584 SVN::Client::get_ssl_client_cert_file_provider(),
4585 SVN::Client::get_ssl_client_cert_prompt_provider(
4586 \&Git::SVN::Prompt::ssl_client_cert, 2),
4587 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4588 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4589 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4590 SVN::Client::get_username_provider(),
4591 SVN::Client::get_ssl_server_trust_prompt_provider(
4592 \&Git::SVN::Prompt::ssl_server_trust),
4593 SVN::Client::get_username_prompt_provider(
4594 \&Git::SVN::Prompt::username, 2)
4595 ]
4596 }
4597
4598 sub escape_uri_only {
4599 my ($uri) = @_;
4600 my @tmp;
4601 foreach (split m{/}, $uri) {
4602 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4603 push @tmp, $_;
4604 }
4605 join('/', @tmp);
4606 }
4607
4608 sub escape_url {
4609 my ($url) = @_;
4610 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4611 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4612 $url = "$scheme://$domain$uri";
4613 }
4614 $url;
4615 }
4616
4617 sub new {
4618 my ($class, $url) = @_;
4619 $url =~ s!/+$!!;
4620 return $RA if ($RA && $RA->{url} eq $url);
4621
4622 SVN::_Core::svn_config_ensure($config_dir, undef);
4623 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4624 my $config = SVN::Core::config_get_config($config_dir);
4625 $RA = undef;
4626 my $dont_store_passwords = 1;
4627 my $conf_t = ${$config}{'config'};
4628 {
4629 no warnings 'once';
4630 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4631 # produces warnings that variables are used only once.
4632 # I had not found the better way to shut them up, so
4633 # the warnings of type 'once' are disabled in this block.
4634 if (SVN::_Core::svn_config_get_bool($conf_t,
4635 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4636 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4637 1) == 0) {
4638 SVN::_Core::svn_auth_set_parameter($baton,
4639 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4640 bless (\$dont_store_passwords, "_p_void"));
4641 }
4642 if (SVN::_Core::svn_config_get_bool($conf_t,
4643 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4644 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4645 1) == 0) {
4646 $Git::SVN::Prompt::_no_auth_cache = 1;
4647 }
4648 } # no warnings 'once'
4649 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4650 config => $config,
4651 pool => SVN::Pool->new,
4652 auth_provider_callbacks => $callbacks);
4653 $self->{url} = $url;
4654 $self->{svn_path} = $url;
4655 $self->{repos_root} = $self->get_repos_root;
4656 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4657 $self->{cache} = { check_path => { r => 0, data => {} },
4658 get_dir => { r => 0, data => {} } };
4659 $RA = bless $self, $class;
4660 }
4661
4662 sub check_path {
4663 my ($self, $path, $r) = @_;
4664 my $cache = $self->{cache}->{check_path};
4665 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4666 return $cache->{data}->{$path};
4667 }
4668 my $pool = SVN::Pool->new;
4669 my $t = $self->SUPER::check_path($path, $r, $pool);
4670 $pool->clear;
4671 if ($r != $cache->{r}) {
4672 %{$cache->{data}} = ();
4673 $cache->{r} = $r;
4674 }
4675 $cache->{data}->{$path} = $t;
4676 }
4677
4678 sub get_dir {
4679 my ($self, $dir, $r) = @_;
4680 my $cache = $self->{cache}->{get_dir};
4681 if ($r == $cache->{r}) {
4682 if (my $x = $cache->{data}->{$dir}) {
4683 return wantarray ? @$x : $x->[0];
4684 }
4685 }
4686 my $pool = SVN::Pool->new;
4687 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4688 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4689 $pool->clear;
4690 if ($r != $cache->{r}) {
4691 %{$cache->{data}} = ();
4692 $cache->{r} = $r;
4693 }
4694 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4695 wantarray ? (\%dirents, $r, $props) : \%dirents;
4696 }
4697
4698 sub DESTROY {
4699 # do not call the real DESTROY since we store ourselves in $RA
4700 }
4701
4702 # get_log(paths, start, end, limit,
4703 # discover_changed_paths, strict_node_history, receiver)
4704 sub get_log {
4705 my ($self, @args) = @_;
4706 my $pool = SVN::Pool->new;
4707
4708 # svn_log_changed_path_t objects passed to get_log are likely to be
4709 # overwritten even if only the refs are copied to an external variable,
4710 # so we should dup the structures in their entirety. Using an
4711 # externally passed pool (instead of our temporary and quickly cleared
4712 # pool in Git::SVN::Ra) does not help matters at all...
4713 my $receiver = pop @args;
4714 my $prefix = "/".$self->{svn_path};
4715 $prefix =~ s#/+($)##;
4716 my $prefix_regex = qr#^\Q$prefix\E#;
4717 push(@args, sub {
4718 my ($paths) = $_[0];
4719 return &$receiver(@_) unless $paths;
4720 $_[0] = ();
4721 foreach my $p (keys %$paths) {
4722 my $i = $paths->{$p};
4723 # Make path relative to our url, not repos_root
4724 $p =~ s/$prefix_regex//;
4725 my %s = map { $_ => $i->$_; }
4726 qw/copyfrom_path copyfrom_rev action/;
4727 if ($s{'copyfrom_path'}) {
4728 $s{'copyfrom_path'} =~ s/$prefix_regex//;
4729 }
4730 $_[0]{$p} = \%s;
4731 }
4732 &$receiver(@_);
4733 });
4734
4735
4736 # the limit parameter was not supported in SVN 1.1.x, so we
4737 # drop it. Therefore, the receiver callback passed to it
4738 # is made aware of this limitation by being wrapped if
4739 # the limit passed to is being wrapped.
4740 if ($SVN::Core::VERSION le '1.2.0') {
4741 my $limit = splice(@args, 3, 1);
4742 if ($limit > 0) {
4743 my $receiver = pop @args;
4744 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4745 }
4746 }
4747 my $ret = $self->SUPER::get_log(@args, $pool);
4748 $pool->clear;
4749 $ret;
4750 }
4751
4752 sub trees_match {
4753 my ($self, $url1, $rev1, $url2, $rev2) = @_;
4754 my $ctx = SVN::Client->new(auth => _auth_providers);
4755 my $out = IO::File->new_tmpfile;
4756
4757 # older SVN (1.1.x) doesn't take $pool as the last parameter for
4758 # $ctx->diff(), so we'll create a default one
4759 my $pool = SVN::Pool->new_default_sub;
4760
4761 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4762 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4763 $out->flush;
4764 my $ret = (($out->stat)[7] == 0);
4765 close $out or croak $!;
4766
4767 $ret;
4768 }
4769
4770 sub get_commit_editor {
4771 my ($self, $log, $cb, $pool) = @_;
4772 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4773 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4774 }
4775
4776 sub gs_do_update {
4777 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4778 my $new = ($rev_a == $rev_b);
4779 my $path = $gs->{path};
4780
4781 if ($new && -e $gs->{index}) {
4782 unlink $gs->{index} or die
4783 "Couldn't unlink index: $gs->{index}: $!\n";
4784 }
4785 my $pool = SVN::Pool->new;
4786 $editor->set_path_strip($path);
4787 my (@pc) = split m#/#, $path;
4788 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4789 1, $editor, $pool);
4790 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4791
4792 # Since we can't rely on svn_ra_reparent being available, we'll
4793 # just have to do some magic with set_path to make it so
4794 # we only want a partial path.
4795 my $sp = '';
4796 my $final = join('/', @pc);
4797 while (@pc) {
4798 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4799 $sp .= '/' if length $sp;
4800 $sp .= shift @pc;
4801 }
4802 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4803
4804 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4805
4806 $reporter->finish_report($pool);
4807 $pool->clear;
4808 $editor->{git_commit_ok};
4809 }
4810
4811 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4812 # svn_ra_reparent didn't work before 1.4)
4813 sub gs_do_switch {
4814 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4815 my $path = $gs->{path};
4816 my $pool = SVN::Pool->new;
4817
4818 my $full_url = $self->{url};
4819 my $old_url = $full_url;
4820 $full_url .= '/' . $path if length $path;
4821 my ($ra, $reparented);
4822
4823 if ($old_url =~ m#^svn(\+ssh)?://# ||
4824 ($full_url =~ m#^https?://# &&
4825 escape_url($full_url) ne $full_url)) {
4826 $_[0] = undef;
4827 $self = undef;
4828 $RA = undef;
4829 $ra = Git::SVN::Ra->new($full_url);
4830 $ra_invalid = 1;
4831 } elsif ($old_url ne $full_url) {
4832 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4833 $self->{url} = $full_url;
4834 $reparented = 1;
4835 }
4836
4837 $ra ||= $self;
4838 $url_b = escape_url($url_b);
4839 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4840 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4841 $reporter->set_path('', $rev_a, 0, @lock, $pool);
4842 $reporter->finish_report($pool);
4843
4844 if ($reparented) {
4845 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4846 $self->{url} = $old_url;
4847 }
4848
4849 $pool->clear;
4850 $editor->{git_commit_ok};
4851 }
4852
4853 sub longest_common_path {
4854 my ($gsv, $globs) = @_;
4855 my %common;
4856 my $common_max = scalar @$gsv;
4857
4858 foreach my $gs (@$gsv) {
4859 my @tmp = split m#/#, $gs->{path};
4860 my $p = '';
4861 foreach (@tmp) {
4862 $p .= length($p) ? "/$_" : $_;
4863 $common{$p} ||= 0;
4864 $common{$p}++;
4865 }
4866 }
4867 $globs ||= [];
4868 $common_max += scalar @$globs;
4869 foreach my $glob (@$globs) {
4870 my @tmp = split m#/#, $glob->{path}->{left};
4871 my $p = '';
4872 foreach (@tmp) {
4873 $p .= length($p) ? "/$_" : $_;
4874 $common{$p} ||= 0;
4875 $common{$p}++;
4876 }
4877 }
4878
4879 my $longest_path = '';
4880 foreach (sort {length $b <=> length $a} keys %common) {
4881 if ($common{$_} == $common_max) {
4882 $longest_path = $_;
4883 last;
4884 }
4885 }
4886 $longest_path;
4887 }
4888
4889 sub gs_fetch_loop_common {
4890 my ($self, $base, $head, $gsv, $globs) = @_;
4891 return if ($base > $head);
4892 my $inc = $_log_window_size;
4893 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4894 my $longest_path = longest_common_path($gsv, $globs);
4895 my $ra_url = $self->{url};
4896 my $find_trailing_edge;
4897 while (1) {
4898 my %revs;
4899 my $err;
4900 my $err_handler = $SVN::Error::handler;
4901 $SVN::Error::handler = sub {
4902 ($err) = @_;
4903 skip_unknown_revs($err);
4904 };
4905 sub _cb {
4906 my ($paths, $r, $author, $date, $log) = @_;
4907 [ $paths,
4908 { author => $author, date => $date, log => $log } ];
4909 }
4910 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4911 sub { $revs{$_[1]} = _cb(@_) });
4912 if ($err) {
4913 print "Checked through r$max\r";
4914 } else {
4915 $find_trailing_edge = 1;
4916 }
4917 if ($err and $find_trailing_edge) {
4918 print STDERR "Path '$longest_path' ",
4919 "was probably deleted:\n",
4920 $err->expanded_message,
4921 "\nWill attempt to follow ",
4922 "revisions r$min .. r$max ",
4923 "committed before the deletion\n";
4924 my $hi = $max;
4925 while (--$hi >= $min) {
4926 my $ok;
4927 $self->get_log([$longest_path], $min, $hi,
4928 0, 1, 1, sub {
4929 $ok = $_[1];
4930 $revs{$_[1]} = _cb(@_) });
4931 if ($ok) {
4932 print STDERR "r$min .. r$ok OK\n";
4933 last;
4934 }
4935 }
4936 $find_trailing_edge = 0;
4937 }
4938 $SVN::Error::handler = $err_handler;
4939
4940 my %exists = map { $_->{path} => $_ } @$gsv;
4941 foreach my $r (sort {$a <=> $b} keys %revs) {
4942 my ($paths, $logged) = @{$revs{$r}};
4943
4944 foreach my $gs ($self->match_globs(\%exists, $paths,
4945 $globs, $r)) {
4946 if ($gs->rev_map_max >= $r) {
4947 next;
4948 }
4949 next unless $gs->match_paths($paths, $r);
4950 $gs->{logged_rev_props} = $logged;
4951 if (my $last_commit = $gs->last_commit) {
4952 $gs->assert_index_clean($last_commit);
4953 }
4954 my $log_entry = $gs->do_fetch($paths, $r);
4955 if ($log_entry) {
4956 $gs->do_git_commit($log_entry);
4957 }
4958 $INDEX_FILES{$gs->{index}} = 1;
4959 }
4960 foreach my $g (@$globs) {
4961 my $k = "svn-remote.$g->{remote}." .
4962 "$g->{t}-maxRev";
4963 Git::SVN::tmp_config($k, $r);
4964 }
4965 if ($ra_invalid) {
4966 $_[0] = undef;
4967 $self = undef;
4968 $RA = undef;
4969 $self = Git::SVN::Ra->new($ra_url);
4970 $ra_invalid = undef;
4971 }
4972 }
4973 # pre-fill the .rev_db since it'll eventually get filled in
4974 # with '0' x40 if something new gets committed
4975 foreach my $gs (@$gsv) {
4976 next if $gs->rev_map_max >= $max;
4977 next if defined $gs->rev_map_get($max);
4978 $gs->rev_map_set($max, 0 x40);
4979 }
4980 foreach my $g (@$globs) {
4981 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4982 Git::SVN::tmp_config($k, $max);
4983 }
4984 last if $max >= $head;
4985 $min = $max + 1;
4986 $max += $inc;
4987 $max = $head if ($max > $head);
4988 }
4989 Git::SVN::gc();
4990 }
4991
4992 sub get_dir_globbed {
4993 my ($self, $left, $depth, $r) = @_;
4994
4995 my @x = eval { $self->get_dir($left, $r) };
4996 return unless scalar @x == 3;
4997 my $dirents = $x[0];
4998 my @finalents;
4999 foreach my $de (keys %$dirents) {
5000 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5001 if ($depth > 1) {
5002 my @args = ("$left/$de", $depth - 1, $r);
5003 foreach my $dir ($self->get_dir_globbed(@args)) {
5004 push @finalents, "$de/$dir";
5005 }
5006 } else {
5007 push @finalents, $de;
5008 }
5009 }
5010 @finalents;
5011 }
5012
5013 sub match_globs {
5014 my ($self, $exists, $paths, $globs, $r) = @_;
5015
5016 sub get_dir_check {
5017 my ($self, $exists, $g, $r) = @_;
5018
5019 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5020 $g->{path}->{depth},
5021 $r);
5022
5023 foreach my $de (@dirs) {
5024 my $p = $g->{path}->full_path($de);
5025 next if $exists->{$p};
5026 next if (length $g->{path}->{right} &&
5027 ($self->check_path($p, $r) !=
5028 $SVN::Node::dir));
5029 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5030 $g->{ref}->full_path($de), 1);
5031 }
5032 }
5033 foreach my $g (@$globs) {
5034 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5035 if ($path->{action} =~ /^[AR]$/) {
5036 get_dir_check($self, $exists, $g, $r);
5037 }
5038 }
5039 foreach (keys %$paths) {
5040 if (/$g->{path}->{left_regex}/ &&
5041 !/$g->{path}->{regex}/) {
5042 next if $paths->{$_}->{action} !~ /^[AR]$/;
5043 get_dir_check($self, $exists, $g, $r);
5044 }
5045 next unless /$g->{path}->{regex}/;
5046 my $p = $1;
5047 my $pathname = $g->{path}->full_path($p);
5048 next if $exists->{$pathname};
5049 next if ($self->check_path($pathname, $r) !=
5050 $SVN::Node::dir);
5051 $exists->{$pathname} = Git::SVN->init(
5052 $self->{url}, $pathname, undef,
5053 $g->{ref}->full_path($p), 1);
5054 }
5055 my $c = '';
5056 foreach (split m#/#, $g->{path}->{left}) {
5057 $c .= "/$_";
5058 next unless ($paths->{$c} &&
5059 ($paths->{$c}->{action} =~ /^[AR]$/));
5060 get_dir_check($self, $exists, $g, $r);
5061 }
5062 }
5063 values %$exists;
5064 }
5065
5066 sub minimize_url {
5067 my ($self) = @_;
5068 return $self->{url} if ($self->{url} eq $self->{repos_root});
5069 my $url = $self->{repos_root};
5070 my @components = split(m!/!, $self->{svn_path});
5071 my $c = '';
5072 do {
5073 $url .= "/$c" if length $c;
5074 eval {
5075 my $ra = (ref $self)->new($url);
5076 my $latest = $ra->get_latest_revnum;
5077 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5078 };
5079 } while ($@ && ($c = shift @components));
5080 $url;
5081 }
5082
5083 sub can_do_switch {
5084 my $self = shift;
5085 unless (defined $can_do_switch) {
5086 my $pool = SVN::Pool->new;
5087 my $rep = eval {
5088 $self->do_switch(1, '', 0, $self->{url},
5089 SVN::Delta::Editor->new, $pool);
5090 };
5091 if ($@) {
5092 $can_do_switch = 0;
5093 } else {
5094 $rep->abort_report($pool);
5095 $can_do_switch = 1;
5096 }
5097 $pool->clear;
5098 }
5099 $can_do_switch;
5100 }
5101
5102 sub skip_unknown_revs {
5103 my ($err) = @_;
5104 my $errno = $err->apr_err();
5105 # Maybe the branch we're tracking didn't
5106 # exist when the repo started, so it's
5107 # not an error if it doesn't, just continue
5108 #
5109 # Wonderfully consistent library, eh?
5110 # 160013 - svn:// and file://
5111 # 175002 - http(s)://
5112 # 175007 - http(s):// (this repo required authorization, too...)
5113 # More codes may be discovered later...
5114 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5115 my $err_key = $err->expanded_message;
5116 # revision numbers change every time, filter them out
5117 $err_key =~ s/\d+/\0/g;
5118 $err_key = "$errno\0$err_key";
5119 unless ($ignored_err{$err_key}) {
5120 warn "W: Ignoring error from SVN, path probably ",
5121 "does not exist: ($errno): ",
5122 $err->expanded_message,"\n";
5123 warn "W: Do not be alarmed at the above message ",
5124 "git-svn is just searching aggressively for ",
5125 "old history.\n",
5126 "This may take a while on large repositories\n";
5127 $ignored_err{$err_key} = 1;
5128 }
5129 return;
5130 }
5131 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5132 }
5133
5134 package Git::SVN::Log;
5135 use strict;
5136 use warnings;
5137 use POSIX qw/strftime/;
5138 use Time::Local;
5139 use constant commit_log_separator => ('-' x 72) . "\n";
5140 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5141 %rusers $show_commit $incremental/;
5142 my $l_fmt;
5143
5144 sub cmt_showable {
5145 my ($c) = @_;
5146 return 1 if defined $c->{r};
5147
5148 # big commit message got truncated by the 16k pretty buffer in rev-list
5149 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5150 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5151 @{$c->{l}} = ();
5152 my @log = command(qw/cat-file commit/, $c->{c});
5153
5154 # shift off the headers
5155 shift @log while ($log[0] ne '');
5156 shift @log;
5157
5158 # TODO: make $c->{l} not have a trailing newline in the future
5159 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5160
5161 (undef, $c->{r}, undef) = ::extract_metadata(
5162 (grep(/^git-svn-id: /, @log))[-1]);
5163 }
5164 return defined $c->{r};
5165 }
5166
5167 sub log_use_color {
5168 return $color || Git->repository->get_colorbool('color.diff');
5169 }
5170
5171 sub git_svn_log_cmd {
5172 my ($r_min, $r_max, @args) = @_;
5173 my $head = 'HEAD';
5174 my (@files, @log_opts);
5175 foreach my $x (@args) {
5176 if ($x eq '--' || @files) {
5177 push @files, $x;
5178 } else {
5179 if (::verify_ref("$x^0")) {
5180 $head = $x;
5181 } else {
5182 push @log_opts, $x;
5183 }
5184 }
5185 }
5186
5187 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5188 $gs ||= Git::SVN->_new;
5189 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5190 $gs->refname);
5191 push @cmd, '-r' unless $non_recursive;
5192 push @cmd, qw/--raw --name-status/ if $verbose;
5193 push @cmd, '--color' if log_use_color();
5194 push @cmd, @log_opts;
5195 if (defined $r_max && $r_max == $r_min) {
5196 push @cmd, '--max-count=1';
5197 if (my $c = $gs->rev_map_get($r_max)) {
5198 push @cmd, $c;
5199 }
5200 } elsif (defined $r_max) {
5201 if ($r_max < $r_min) {
5202 ($r_min, $r_max) = ($r_max, $r_min);
5203 }
5204 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5205 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5206 # If there are no commits in the range, both $c_max and $c_min
5207 # will be undefined. If there is at least 1 commit in the
5208 # range, both will be defined.
5209 return () if !defined $c_min || !defined $c_max;
5210 if ($c_min eq $c_max) {
5211 push @cmd, '--max-count=1', $c_min;
5212 } else {
5213 push @cmd, '--boundary', "$c_min..$c_max";
5214 }
5215 }
5216 return (@cmd, @files);
5217 }
5218
5219 # adapted from pager.c
5220 sub config_pager {
5221 chomp(my $pager = command_oneline(qw(var GIT_PAGER)));
5222 if ($pager eq 'cat') {
5223 $pager = undef;
5224 }
5225 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5226 }
5227
5228 sub run_pager {
5229 return unless -t *STDOUT && defined $pager;
5230 pipe my ($rfd, $wfd) or return;
5231 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5232 if (!$pid) {
5233 open STDOUT, '>&', $wfd or
5234 ::fatal "Can't redirect to stdout: $!";
5235 return;
5236 }
5237 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5238 $ENV{LESS} ||= 'FRSX';
5239 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5240 }
5241
5242 sub format_svn_date {
5243 # some systmes don't handle or mishandle %z, so be creative.
5244 my $t = shift || time;
5245 my $gm = timelocal(gmtime($t));
5246 my $sign = qw( + + - )[ $t <=> $gm ];
5247 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5248 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5249 }
5250
5251 sub parse_git_date {
5252 my ($t, $tz) = @_;
5253 # Date::Parse isn't in the standard Perl distro :(
5254 if ($tz =~ s/^\+//) {
5255 $t += tz_to_s_offset($tz);
5256 } elsif ($tz =~ s/^\-//) {
5257 $t -= tz_to_s_offset($tz);
5258 }
5259 return $t;
5260 }
5261
5262 sub set_local_timezone {
5263 if (defined $TZ) {
5264 $ENV{TZ} = $TZ;
5265 } else {
5266 delete $ENV{TZ};
5267 }
5268 }
5269
5270 sub tz_to_s_offset {
5271 my ($tz) = @_;
5272 $tz =~ s/(\d\d)$//;
5273 return ($1 * 60) + ($tz * 3600);
5274 }
5275
5276 sub get_author_info {
5277 my ($dest, $author, $t, $tz) = @_;
5278 $author =~ s/(?:^\s*|\s*$)//g;
5279 $dest->{a_raw} = $author;
5280 my $au;
5281 if ($::_authors) {
5282 $au = $rusers{$author} || undef;
5283 }
5284 if (!$au) {
5285 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5286 }
5287 $dest->{t} = $t;
5288 $dest->{tz} = $tz;
5289 $dest->{a} = $au;
5290 $dest->{t_utc} = parse_git_date($t, $tz);
5291 }
5292
5293 sub process_commit {
5294 my ($c, $r_min, $r_max, $defer) = @_;
5295 if (defined $r_min && defined $r_max) {
5296 if ($r_min == $c->{r} && $r_min == $r_max) {
5297 show_commit($c);
5298 return 0;
5299 }
5300 return 1 if $r_min == $r_max;
5301 if ($r_min < $r_max) {
5302 # we need to reverse the print order
5303 return 0 if (defined $limit && --$limit < 0);
5304 push @$defer, $c;
5305 return 1;
5306 }
5307 if ($r_min != $r_max) {
5308 return 1 if ($r_min < $c->{r});
5309 return 1 if ($r_max > $c->{r});
5310 }
5311 }
5312 return 0 if (defined $limit && --$limit < 0);
5313 show_commit($c);
5314 return 1;
5315 }
5316
5317 sub show_commit {
5318 my $c = shift;
5319 if ($oneline) {
5320 my $x = "\n";
5321 if (my $l = $c->{l}) {
5322 while ($l->[0] =~ /^\s*$/) { shift @$l }
5323 $x = $l->[0];
5324 }
5325 $l_fmt ||= 'A' . length($c->{r});
5326 print 'r',pack($l_fmt, $c->{r}),' | ';
5327 print "$c->{c} | " if $show_commit;
5328 print $x;
5329 } else {
5330 show_commit_normal($c);
5331 }
5332 }
5333
5334 sub show_commit_changed_paths {
5335 my ($c) = @_;
5336 return unless $c->{changed};
5337 print "Changed paths:\n", @{$c->{changed}};
5338 }
5339
5340 sub show_commit_normal {
5341 my ($c) = @_;
5342 print commit_log_separator, "r$c->{r} | ";
5343 print "$c->{c} | " if $show_commit;
5344 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5345 my $nr_line = 0;
5346
5347 if (my $l = $c->{l}) {
5348 while ($l->[$#$l] eq "\n" && $#$l > 0
5349 && $l->[($#$l - 1)] eq "\n") {
5350 pop @$l;
5351 }
5352 $nr_line = scalar @$l;
5353 if (!$nr_line) {
5354 print "1 line\n\n\n";
5355 } else {
5356 if ($nr_line == 1) {
5357 $nr_line = '1 line';
5358 } else {
5359 $nr_line .= ' lines';
5360 }
5361 print $nr_line, "\n";
5362 show_commit_changed_paths($c);
5363 print "\n";
5364 print $_ foreach @$l;
5365 }
5366 } else {
5367 print "1 line\n";
5368 show_commit_changed_paths($c);
5369 print "\n";
5370
5371 }
5372 foreach my $x (qw/raw stat diff/) {
5373 if ($c->{$x}) {
5374 print "\n";
5375 print $_ foreach @{$c->{$x}}
5376 }
5377 }
5378 }
5379
5380 sub cmd_show_log {
5381 my (@args) = @_;
5382 my ($r_min, $r_max);
5383 my $r_last = -1; # prevent dupes
5384 set_local_timezone();
5385 if (defined $::_revision) {
5386 if ($::_revision =~ /^(\d+):(\d+)$/) {
5387 ($r_min, $r_max) = ($1, $2);
5388 } elsif ($::_revision =~ /^\d+$/) {
5389 $r_min = $r_max = $::_revision;
5390 } else {
5391 ::fatal "-r$::_revision is not supported, use ",
5392 "standard 'git log' arguments instead";
5393 }
5394 }
5395
5396 config_pager();
5397 @args = git_svn_log_cmd($r_min, $r_max, @args);
5398 if (!@args) {
5399 print commit_log_separator unless $incremental || $oneline;
5400 return;
5401 }
5402 my $log = command_output_pipe(@args);
5403 run_pager();
5404 my (@k, $c, $d, $stat);
5405 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5406 while (<$log>) {
5407 if (/^${esc_color}commit -?($::sha1_short)/o) {
5408 my $cmt = $1;
5409 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5410 $r_last = $c->{r};
5411 process_commit($c, $r_min, $r_max, \@k) or
5412 goto out;
5413 }
5414 $d = undef;
5415 $c = { c => $cmt };
5416 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5417 get_author_info($c, $1, $2, $3);
5418 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5419 # ignore
5420 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5421 push @{$c->{raw}}, $_;
5422 } elsif (/^${esc_color}[ACRMDT]\t/) {
5423 # we could add $SVN->{svn_path} here, but that requires
5424 # remote access at the moment (repo_path_split)...
5425 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5426 push @{$c->{changed}}, $_;
5427 } elsif (/^${esc_color}diff /o) {
5428 $d = 1;
5429 push @{$c->{diff}}, $_;
5430 } elsif ($d) {
5431 push @{$c->{diff}}, $_;
5432 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5433 $esc_color*[\+\-]*$esc_color$/x) {
5434 $stat = 1;
5435 push @{$c->{stat}}, $_;
5436 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5437 push @{$c->{stat}}, $_;
5438 $stat = undef;
5439 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5440 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5441 } elsif (s/^${esc_color} //o) {
5442 push @{$c->{l}}, $_;
5443 }
5444 }
5445 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5446 $r_last = $c->{r};
5447 process_commit($c, $r_min, $r_max, \@k);
5448 }
5449 if (@k) {
5450 ($r_min, $r_max) = ($r_max, $r_min);
5451 process_commit($_, $r_min, $r_max) foreach reverse @k;
5452 }
5453 out:
5454 close $log;
5455 print commit_log_separator unless $incremental || $oneline;
5456 }
5457
5458 sub cmd_blame {
5459 my $path = pop;
5460
5461 config_pager();
5462 run_pager();
5463
5464 my ($fh, $ctx, $rev);
5465
5466 if ($_git_format) {
5467 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5468 while (my $line = <$fh>) {
5469 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5470 # Uncommitted edits show up as a rev ID of
5471 # all zeros, which we can't look up with
5472 # cmt_metadata
5473 if ($1 !~ /^0+$/) {
5474 (undef, $rev, undef) =
5475 ::cmt_metadata($1);
5476 $rev = '0' if (!$rev);
5477 } else {
5478 $rev = '0';
5479 }
5480 $rev = sprintf('%-10s', $rev);
5481 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5482 }
5483 print $line;
5484 }
5485 } else {
5486 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5487 '--', $path);
5488 my ($sha1);
5489 my %authors;
5490 my @buffer;
5491 my %dsha; #distinct sha keys
5492
5493 while (my $line = <$fh>) {
5494 push @buffer, $line;
5495 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5496 $dsha{$1} = 1;
5497 }
5498 }
5499
5500 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5501
5502 foreach my $line (@buffer) {
5503 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5504 $rev = $s2r->{$1};
5505 $rev = '0' if (!$rev)
5506 }
5507 elsif ($line =~ /^author (.*)/) {
5508 $authors{$rev} = $1;
5509 $authors{$rev} =~ s/\s/_/g;
5510 }
5511 elsif ($line =~ /^\t(.*)$/) {
5512 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5513 }
5514 }
5515 }
5516 command_close_pipe($fh, $ctx);
5517 }
5518
5519 package Git::SVN::Migration;
5520 # these version numbers do NOT correspond to actual version numbers
5521 # of git nor git-svn. They are just relative.
5522 #
5523 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5524 #
5525 # v1 layout: .git/$id/info/url, refs/remotes/$id
5526 #
5527 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5528 #
5529 # v3 layout: .git/svn/$id, refs/remotes/$id
5530 # - info/url may remain for backwards compatibility
5531 # - this is what we migrate up to this layout automatically,
5532 # - this will be used by git svn init on single branches
5533 # v3.1 layout (auto migrated):
5534 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5535 # for backwards compatibility
5536 #
5537 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5538 # - this is only created for newly multi-init-ed
5539 # repositories. Similar in spirit to the
5540 # --use-separate-remotes option in git-clone (now default)
5541 # - we do not automatically migrate to this (following
5542 # the example set by core git)
5543 #
5544 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
5545 # - newer, more-efficient format that uses 24-bytes per record
5546 # with no filler space.
5547 # - use xxd -c24 < .rev_map.$UUID to view and debug
5548 # - This is a one-way migration, repositories updated to the
5549 # new format will not be able to use old git-svn without
5550 # rebuilding the .rev_db. Rebuilding the rev_db is not
5551 # possible if noMetadata or useSvmProps are set; but should
5552 # be no problem for users that use the (sensible) defaults.
5553 use strict;
5554 use warnings;
5555 use Carp qw/croak/;
5556 use File::Path qw/mkpath/;
5557 use File::Basename qw/dirname basename/;
5558 use vars qw/$_minimize/;
5559
5560 sub migrate_from_v0 {
5561 my $git_dir = $ENV{GIT_DIR};
5562 return undef unless -d $git_dir;
5563 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5564 my $migrated = 0;
5565 while (<$fh>) {
5566 chomp;
5567 my ($id, $orig_ref) = ($_, $_);
5568 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5569 next unless -f "$git_dir/$id/info/url";
5570 my $new_ref = "refs/remotes/$id";
5571 if (::verify_ref("$new_ref^0")) {
5572 print STDERR "W: $orig_ref is probably an old ",
5573 "branch used by an ancient version of ",
5574 "git-svn.\n",
5575 "However, $new_ref also exists.\n",
5576 "We will not be able ",
5577 "to use this branch until this ",
5578 "ambiguity is resolved.\n";
5579 next;
5580 }
5581 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5582 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5583 command_noisy('update-ref', $new_ref, $orig_ref);
5584 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5585 $migrated++;
5586 }
5587 command_close_pipe($fh, $ctx);
5588 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5589 $migrated;
5590 }
5591
5592 sub migrate_from_v1 {
5593 my $git_dir = $ENV{GIT_DIR};
5594 my $migrated = 0;
5595 return $migrated unless -d $git_dir;
5596 my $svn_dir = "$git_dir/svn";
5597
5598 # just in case somebody used 'svn' as their $id at some point...
5599 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5600
5601 print STDERR "Migrating from a git-svn v1 layout...\n";
5602 mkpath([$svn_dir]);
5603 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5604 "$svn_dir\n\t(required for this version ",
5605 "($::VERSION) of git-svn) does not exist.\n";
5606 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5607 while (<$fh>) {
5608 my $x = $_;
5609 next unless $x =~ s#^refs/remotes/##;
5610 chomp $x;
5611 next unless -f "$git_dir/$x/info/url";
5612 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5613 next unless $u;
5614 my $dn = dirname("$git_dir/svn/$x");
5615 mkpath([$dn]) unless -d $dn;
5616 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5617 mkpath(["$git_dir/svn/svn"]);
5618 print STDERR " - $git_dir/$x/info => ",
5619 "$git_dir/svn/$x/info\n";
5620 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5621 croak "$!: $x";
5622 # don't worry too much about these, they probably
5623 # don't exist with repos this old (save for index,
5624 # and we can easily regenerate that)
5625 foreach my $f (qw/unhandled.log index .rev_db/) {
5626 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5627 }
5628 } else {
5629 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5630 rename "$git_dir/$x", "$git_dir/svn/$x" or
5631 croak "$!: $x";
5632 }
5633 $migrated++;
5634 }
5635 command_close_pipe($fh, $ctx);
5636 print STDERR "Done migrating from a git-svn v1 layout\n";
5637 $migrated;
5638 }
5639
5640 sub read_old_urls {
5641 my ($l_map, $pfx, $path) = @_;
5642 my @dir;
5643 foreach (<$path/*>) {
5644 if (-r "$_/info/url") {
5645 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5646 my $ref_id = $pfx . basename $_;
5647 my $url = ::file_to_s("$_/info/url");
5648 $l_map->{$ref_id} = $url;
5649 } elsif (-d $_) {
5650 push @dir, $_;
5651 }
5652 }
5653 foreach (@dir) {
5654 my $x = $_;
5655 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5656 read_old_urls($l_map, $x, $_);
5657 }
5658 }
5659
5660 sub migrate_from_v2 {
5661 my @cfg = command(qw/config -l/);
5662 return if grep /^svn-remote\..+\.url=/, @cfg;
5663 my %l_map;
5664 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5665 my $migrated = 0;
5666
5667 foreach my $ref_id (sort keys %l_map) {
5668 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5669 if ($@) {
5670 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5671 }
5672 $migrated++;
5673 }
5674 $migrated;
5675 }
5676
5677 sub minimize_connections {
5678 my $r = Git::SVN::read_all_remotes();
5679 my $new_urls = {};
5680 my $root_repos = {};
5681 foreach my $repo_id (keys %$r) {
5682 my $url = $r->{$repo_id}->{url} or next;
5683 my $fetch = $r->{$repo_id}->{fetch} or next;
5684 my $ra = Git::SVN::Ra->new($url);
5685
5686 # skip existing cases where we already connect to the root
5687 if (($ra->{url} eq $ra->{repos_root}) ||
5688 ($ra->{repos_root} eq $repo_id)) {
5689 $root_repos->{$ra->{url}} = $repo_id;
5690 next;
5691 }
5692
5693 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5694 my $root_path = $ra->{url};
5695 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5696 foreach my $path (keys %$fetch) {
5697 my $ref_id = $fetch->{$path};
5698 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5699
5700 # make sure we can read when connecting to
5701 # a higher level of a repository
5702 my ($last_rev, undef) = $gs->last_rev_commit;
5703 if (!defined $last_rev) {
5704 $last_rev = eval {
5705 $root_ra->get_latest_revnum;
5706 };
5707 next if $@;
5708 }
5709 my $new = $root_path;
5710 $new .= length $path ? "/$path" : '';
5711 eval {
5712 $root_ra->get_log([$new], $last_rev, $last_rev,
5713 0, 0, 1, sub { });
5714 };
5715 next if $@;
5716 $new_urls->{$ra->{repos_root}}->{$new} =
5717 { ref_id => $ref_id,
5718 old_repo_id => $repo_id,
5719 old_path => $path };
5720 }
5721 }
5722
5723 my @emptied;
5724 foreach my $url (keys %$new_urls) {
5725 # see if we can re-use an existing [svn-remote "repo_id"]
5726 # instead of creating a(n ugly) new section:
5727 my $repo_id = $root_repos->{$url} || $url;
5728
5729 my $fetch = $new_urls->{$url};
5730 foreach my $path (keys %$fetch) {
5731 my $x = $fetch->{$path};
5732 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5733 my $pfx = "svn-remote.$x->{old_repo_id}";
5734
5735 my $old_fetch = quotemeta("$x->{old_path}:".
5736 "$x->{ref_id}");
5737 command_noisy(qw/config --unset/,
5738 "$pfx.fetch", '^'. $old_fetch . '$');
5739 delete $r->{$x->{old_repo_id}}->
5740 {fetch}->{$x->{old_path}};
5741 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5742 command_noisy(qw/config --unset/,
5743 "$pfx.url");
5744 push @emptied, $x->{old_repo_id}
5745 }
5746 }
5747 }
5748 if (@emptied) {
5749 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5750 print STDERR <<EOF;
5751 The following [svn-remote] sections in your config file ($file) are empty
5752 and can be safely removed:
5753 EOF
5754 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5755 }
5756 }
5757
5758 sub migration_check {
5759 migrate_from_v0();
5760 migrate_from_v1();
5761 migrate_from_v2();
5762 minimize_connections() if $_minimize;
5763 }
5764
5765 package Git::IndexInfo;
5766 use strict;
5767 use warnings;
5768 use Git qw/command_input_pipe command_close_pipe/;
5769
5770 sub new {
5771 my ($class) = @_;
5772 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5773 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5774 }
5775
5776 sub remove {
5777 my ($self, $path) = @_;
5778 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5779 return ++$self->{nr};
5780 }
5781 undef;
5782 }
5783
5784 sub update {
5785 my ($self, $mode, $hash, $path) = @_;
5786 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5787 return ++$self->{nr};
5788 }
5789 undef;
5790 }
5791
5792 sub DESTROY {
5793 my ($self) = @_;
5794 command_close_pipe($self->{gui}, $self->{ctx});
5795 }
5796
5797 package Git::SVN::GlobSpec;
5798 use strict;
5799 use warnings;
5800
5801 sub new {
5802 my ($class, $glob) = @_;
5803 my $re = $glob;
5804 $re =~ s!/+$!!g; # no need for trailing slashes
5805 $re =~ m!^([^*]*)(\*(?:/\*)*)(.*)$!;
5806 my $temp = $re;
5807 my ($left, $right) = ($1, $3);
5808 $re = $2;
5809 my $depth = $re =~ tr/*/*/;
5810 if ($depth != $temp =~ tr/*/*/) {
5811 die "Only one set of wildcard directories " .
5812 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5813 }
5814 if ($depth == 0) {
5815 die "One '*' is needed for glob: '$glob'\n";
5816 }
5817 $re =~ s!\*!\[^/\]*!g;
5818 $re = quotemeta($left) . "($re)" . quotemeta($right);
5819 if (length $left && !($left =~ s!/+$!!g)) {
5820 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5821 }
5822 if (length $right && !($right =~ s!^/+!!g)) {
5823 die "Missing leading '/' on right side of: '$glob' ($right)\n";
5824 }
5825 my $left_re = qr/^\/\Q$left\E(\/|$)/;
5826 bless { left => $left, right => $right, left_regex => $left_re,
5827 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5828 }
5829
5830 sub full_path {
5831 my ($self, $path) = @_;
5832 return (length $self->{left} ? "$self->{left}/" : '') .
5833 $path . (length $self->{right} ? "/$self->{right}" : '');
5834 }
5835
5836 __END__
5837
5838 Data structures:
5839
5840
5841 $remotes = { # returned by read_all_remotes()
5842 'svn' => {
5843 # svn-remote.svn.url=https://svn.musicpd.org
5844 url => 'https://svn.musicpd.org',
5845 # svn-remote.svn.fetch=mpd/trunk:trunk
5846 fetch => {
5847 'mpd/trunk' => 'trunk',
5848 },
5849 # svn-remote.svn.tags=mpd/tags/*:tags/*
5850 tags => {
5851 path => {
5852 left => 'mpd/tags',
5853 right => '',
5854 regex => qr!mpd/tags/([^/]+)$!,
5855 glob => 'tags/*',
5856 },
5857 ref => {
5858 left => 'tags',
5859 right => '',
5860 regex => qr!tags/([^/]+)$!,
5861 glob => 'tags/*',
5862 },
5863 }
5864 }
5865 };
5866
5867 $log_entry hashref as returned by libsvn_log_entry()
5868 {
5869 log => 'whitespace-formatted log entry
5870 ', # trailing newline is preserved
5871 revision => '8', # integer
5872 date => '2004-02-24T17:01:44.108345Z', # commit date
5873 author => 'committer name'
5874 };
5875
5876
5877 # this is generated by generate_diff();
5878 @mods = array of diff-index line hashes, each element represents one line
5879 of diff-index output
5880
5881 diff-index line ($m hash)
5882 {
5883 mode_a => first column of diff-index output, no leading ':',
5884 mode_b => second column of diff-index output,
5885 sha1_b => sha1sum of the final blob,
5886 chg => change type [MCRADT],
5887 file_a => original file name of a file (iff chg is 'C' or 'R')
5888 file_b => new/current file name of a file (any chg)
5889 }
5890 ;
5891
5892 # retval of read_url_paths{,_all}();
5893 $l_map = {
5894 # repository root url
5895 'https://svn.musicpd.org' => {
5896 # repository path # GIT_SVN_ID
5897 'mpd/trunk' => 'trunk',
5898 'mpd/tags/0.11.5' => 'tags/0.11.5',
5899 },
5900 }
5901
5902 Notes:
5903 I don't trust the each() function on unless I created %hash myself
5904 because the internal iterator may not have started at base.
Gitweb development nest