?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'jc/cherry-pick' (early part)
[git/jnareb-git/bp-gitweb.git] / git-svn.perl
blob1195569529401fd1c62414c8683a6b577721c1c7
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
8 $_q $_authors %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
23 $Git::SVN::Log::TZ = $ENV{TZ};
24 $ENV{TZ} = 'UTC';
25 $| = 1; # unbuffer STDOUT
26
27 sub fatal (@) { print STDERR "@_\n"; exit 1 }
28 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
29 require SVN::Ra;
30 require SVN::Delta;
31 if ($SVN::Core::VERSION lt '1.1.0') {
32 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
33 }
34 push @Git::SVN::Ra::ISA, 'SVN::Ra';
35 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
36 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
37 use Carp qw/croak/;
38 use Digest::MD5;
39 use IO::File qw//;
40 use File::Basename qw/dirname basename/;
41 use File::Path qw/mkpath/;
42 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
43 use IPC::Open3;
44 use Git;
45
46 BEGIN {
47 # import functions from Git into our packages, en masse
48 no strict 'refs';
49 foreach (qw/command command_oneline command_noisy command_output_pipe
50 command_input_pipe command_close_pipe/) {
51 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
52 Git::SVN::Migration Git::SVN::Log Git::SVN),
53 __PACKAGE__) {
54 *{"${package}::$_"} = \&{"Git::$_"};
55 }
56 }
57 }
58
59 my ($SVN);
60
61 $sha1 = qr/[a-f\d]{40}/;
62 $sha1_short = qr/[a-f\d]{4,40}/;
63 my ($_stdin, $_help, $_edit,
64 $_message, $_file,
65 $_template, $_shared,
66 $_version, $_fetch_all, $_no_rebase,
67 $_merge, $_strategy, $_dry_run, $_local,
68 $_prefix, $_no_checkout, $_url, $_verbose);
69 $Git::SVN::_follow_parent = 1;
70 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
71 'config-dir=s' => \$Git::SVN::Ra::config_dir,
72 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
73 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
74 'authors-file|A=s' => \$_authors,
75 'repack:i' => \$Git::SVN::_repack,
76 'noMetadata' => \$Git::SVN::_no_metadata,
77 'useSvmProps' => \$Git::SVN::_use_svm_props,
78 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
79 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
80 'no-checkout' => \$_no_checkout,
81 'quiet|q' => \$_q,
82 'repack-flags|repack-args|repack-opts=s' =>
83 \$Git::SVN::_repack_flags,
84 'use-log-author' => \$Git::SVN::_use_log_author,
85 %remote_opts );
86
87 my ($_trunk, $_tags, $_branches, $_stdlayout);
88 my %icv;
89 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
90 'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
91 'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
92 'stdlayout|s' => \$_stdlayout,
93 'minimize-url|m' => \$Git::SVN::_minimize_url,
94 'no-metadata' => sub { $icv{noMetadata} = 1 },
95 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
96 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
97 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
98 %remote_opts );
99 my %cmt_opts = ( 'edit|e' => \$_edit,
100 'rmdir' => \$SVN::Git::Editor::_rmdir,
101 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
102 'l=i' => \$SVN::Git::Editor::_rename_limit,
103 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
104 );
105
106 my %cmd = (
107 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
108 { 'revision|r=s' => \$_revision,
109 'fetch-all|all' => \$_fetch_all,
110 %fc_opts } ],
111 clone => [ \&cmd_clone, "Initialize and fetch revisions",
112 { 'revision|r=s' => \$_revision,
113 %fc_opts, %init_opts } ],
114 init => [ \&cmd_init, "Initialize a repo for tracking" .
115 " (requires URL argument)",
116 \%init_opts ],
117 'multi-init' => [ \&cmd_multi_init,
118 "Deprecated alias for ".
119 "'$0 init -T<trunk> -b<branches> -t<tags>'",
120 \%init_opts ],
121 dcommit => [ \&cmd_dcommit,
122 'Commit several diffs to merge with upstream',
123 { 'merge|m|M' => \$_merge,
124 'strategy|s=s' => \$_strategy,
125 'verbose|v' => \$_verbose,
126 'dry-run|n' => \$_dry_run,
127 'fetch-all|all' => \$_fetch_all,
128 'no-rebase' => \$_no_rebase,
129 %cmt_opts, %fc_opts } ],
130 'set-tree' => [ \&cmd_set_tree,
131 "Set an SVN repository to a git tree-ish",
132 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
133 'create-ignore' => [ \&cmd_create_ignore,
134 'Create a .gitignore per svn:ignore',
135 { 'revision|r=i' => \$_revision
136 } ],
137 'propget' => [ \&cmd_propget,
138 'Print the value of a property on a file or directory',
139 { 'revision|r=i' => \$_revision } ],
140 'proplist' => [ \&cmd_proplist,
141 'List all properties of a file or directory',
142 { 'revision|r=i' => \$_revision } ],
143 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
144 { 'revision|r=i' => \$_revision
145 } ],
146 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
147 { 'revision|r=i' => \$_revision
148 } ],
149 'multi-fetch' => [ \&cmd_multi_fetch,
150 "Deprecated alias for $0 fetch --all",
151 { 'revision|r=s' => \$_revision, %fc_opts } ],
152 'migrate' => [ sub { },
153 # no-op, we automatically run this anyways,
154 'Migrate configuration/metadata/layout from
155 previous versions of git-svn',
156 { 'minimize' => \$Git::SVN::Migration::_minimize,
157 %remote_opts } ],
158 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
159 { 'limit=i' => \$Git::SVN::Log::limit,
160 'revision|r=s' => \$_revision,
161 'verbose|v' => \$Git::SVN::Log::verbose,
162 'incremental' => \$Git::SVN::Log::incremental,
163 'oneline' => \$Git::SVN::Log::oneline,
164 'show-commit' => \$Git::SVN::Log::show_commit,
165 'non-recursive' => \$Git::SVN::Log::non_recursive,
166 'authors-file|A=s' => \$_authors,
167 'color' => \$Git::SVN::Log::color,
168 'pager=s' => \$Git::SVN::Log::pager
169 } ],
170 'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
171 {} ],
172 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
173 { 'merge|m|M' => \$_merge,
174 'verbose|v' => \$_verbose,
175 'strategy|s=s' => \$_strategy,
176 'local|l' => \$_local,
177 'fetch-all|all' => \$_fetch_all,
178 %fc_opts } ],
179 'commit-diff' => [ \&cmd_commit_diff,
180 'Commit a diff between two trees',
181 { 'message|m=s' => \$_message,
182 'file|F=s' => \$_file,
183 'revision|r=s' => \$_revision,
184 %cmt_opts } ],
185 'info' => [ \&cmd_info,
186 "Show info about the latest SVN revision
187 on the current branch",
188 { 'url' => \$_url, } ],
189 'blame' => [ \&Git::SVN::Log::cmd_blame,
190 "Show what revision and author last modified each line of a file",
191 {} ],
192 );
193
194 my $cmd;
195 for (my $i = 0; $i < @ARGV; $i++) {
196 if (defined $cmd{$ARGV[$i]}) {
197 $cmd = $ARGV[$i];
198 splice @ARGV, $i, 1;
199 last;
200 }
201 };
202
203 # make sure we're always running at the top-level working directory
204 unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
205 unless (-d $ENV{GIT_DIR}) {
206 if ($git_dir_user_set) {
207 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
208 "but it is not a directory\n";
209 }
210 my $git_dir = delete $ENV{GIT_DIR};
211 chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
212 unless (length $cdup) {
213 die "Already at toplevel, but $git_dir ",
214 "not found '$cdup'\n";
215 }
216 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
217 unless (-d $git_dir) {
218 die "$git_dir still not found after going to ",
219 "'$cdup'\n";
220 }
221 $ENV{GIT_DIR} = $git_dir;
222 }
223 }
224
225 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
226
227 read_repo_config(\%opts);
228 Getopt::Long::Configure('pass_through') if ($cmd && $cmd eq 'log');
229 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
230 'minimize-connections' => \$Git::SVN::Migration::_minimize,
231 'id|i=s' => \$Git::SVN::default_ref_id,
232 'svn-remote|remote|R=s' => sub {
233 $Git::SVN::no_reuse_existing = 1;
234 $Git::SVN::default_repo_id = $_[1] });
235 exit 1 if (!$rv && $cmd && $cmd ne 'log');
236
237 usage(0) if $_help;
238 version() if $_version;
239 usage(1) unless defined $cmd;
240 load_authors() if $_authors;
241
242 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
243 Git::SVN::Migration::migration_check();
244 }
245 Git::SVN::init_vars();
246 eval {
247 Git::SVN::verify_remotes_sanity();
248 $cmd{$cmd}->[0]->(@ARGV);
249 };
250 fatal $@ if $@;
251 post_fetch_checkout();
252 exit 0;
253
254 ####################### primary functions ######################
255 sub usage {
256 my $exit = shift || 0;
257 my $fd = $exit ? \*STDERR : \*STDOUT;
258 print $fd <<"";
259 git-svn - bidirectional operations between a single Subversion tree and git
260 Usage: $0 <command> [options] [arguments]\n
261
262 print $fd "Available commands:\n" unless $cmd;
263
264 foreach (sort keys %cmd) {
265 next if $cmd && $cmd ne $_;
266 next if /^multi-/; # don't show deprecated commands
267 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
268 foreach (sort keys %{$cmd{$_}->[2]}) {
269 # mixed-case options are for .git/config only
270 next if /[A-Z]/ && /^[a-z]+$/i;
271 # prints out arguments as they should be passed:
272 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
273 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
274 "--$_" : "-$_" }
275 split /\|/,$_)," $x\n";
276 }
277 }
278 print $fd <<"";
279 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
280 arbitrary identifier if you're tracking multiple SVN branches/repositories in
281 one git repository and want to keep them separate. See git-svn(1) for more
282 information.
283
284 exit $exit;
285 }
286
287 sub version {
288 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
289 exit 0;
290 }
291
292 sub do_git_init_db {
293 unless (-d $ENV{GIT_DIR}) {
294 my @init_db = ('init');
295 push @init_db, "--template=$_template" if defined $_template;
296 if (defined $_shared) {
297 if ($_shared =~ /[a-z]/) {
298 push @init_db, "--shared=$_shared";
299 } else {
300 push @init_db, "--shared";
301 }
302 }
303 command_noisy(@init_db);
304 }
305 my $set;
306 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
307 foreach my $i (keys %icv) {
308 die "'$set' and '$i' cannot both be set\n" if $set;
309 next unless defined $icv{$i};
310 command_noisy('config', "$pfx.$i", $icv{$i});
311 $set = $i;
312 }
313 }
314
315 sub init_subdir {
316 my $repo_path = shift or return;
317 mkpath([$repo_path]) unless -d $repo_path;
318 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
319 $ENV{GIT_DIR} = '.git';
320 }
321
322 sub cmd_clone {
323 my ($url, $path) = @_;
324 if (!defined $path &&
325 (defined $_trunk || defined $_branches || defined $_tags ||
326 defined $_stdlayout) &&
327 $url !~ m#^[a-z\+]+://#) {
328 $path = $url;
329 }
330 $path = basename($url) if !defined $path || !length $path;
331 cmd_init($url, $path);
332 Git::SVN::fetch_all($Git::SVN::default_repo_id);
333 }
334
335 sub cmd_init {
336 if (defined $_stdlayout) {
337 $_trunk = 'trunk' if (!defined $_trunk);
338 $_tags = 'tags' if (!defined $_tags);
339 $_branches = 'branches' if (!defined $_branches);
340 }
341 if (defined $_trunk || defined $_branches || defined $_tags) {
342 return cmd_multi_init(@_);
343 }
344 my $url = shift or die "SVN repository location required ",
345 "as a command-line argument\n";
346 init_subdir(@_);
347 do_git_init_db();
348
349 Git::SVN->init($url);
350 }
351
352 sub cmd_fetch {
353 if (grep /^\d+=./, @_) {
354 die "'<rev>=<commit>' fetch arguments are ",
355 "no longer supported.\n";
356 }
357 my ($remote) = @_;
358 if (@_ > 1) {
359 die "Usage: $0 fetch [--all] [svn-remote]\n";
360 }
361 $remote ||= $Git::SVN::default_repo_id;
362 if ($_fetch_all) {
363 cmd_multi_fetch();
364 } else {
365 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
366 }
367 }
368
369 sub cmd_set_tree {
370 my (@commits) = @_;
371 if ($_stdin || !@commits) {
372 print "Reading from stdin...\n";
373 @commits = ();
374 while (<STDIN>) {
375 if (/\b($sha1_short)\b/o) {
376 unshift @commits, $1;
377 }
378 }
379 }
380 my @revs;
381 foreach my $c (@commits) {
382 my @tmp = command('rev-parse',$c);
383 if (scalar @tmp == 1) {
384 push @revs, $tmp[0];
385 } elsif (scalar @tmp > 1) {
386 push @revs, reverse(command('rev-list',@tmp));
387 } else {
388 fatal "Failed to rev-parse $c";
389 }
390 }
391 my $gs = Git::SVN->new;
392 my ($r_last, $cmt_last) = $gs->last_rev_commit;
393 $gs->fetch;
394 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
395 fatal "There are new revisions that were fetched ",
396 "and need to be merged (or acknowledged) ",
397 "before committing.\nlast rev: $r_last\n",
398 " current: $gs->{last_rev}";
399 }
400 $gs->set_tree($_) foreach @revs;
401 print "Done committing ",scalar @revs," revisions to SVN\n";
402 unlink $gs->{index};
403 }
404
405 sub cmd_dcommit {
406 my $head = shift;
407 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
408 'Cannot dcommit with a dirty index. Commit your changes first, '
409 . "or stash them with `git stash'.\n";
410 $head ||= 'HEAD';
411 my @refs;
412 my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
413 print "Committing to $url ...\n";
414 unless ($gs) {
415 die "Unable to determine upstream SVN information from ",
416 "$head history\n";
417 }
418 my $last_rev;
419 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
420 if ($_no_rebase && scalar(@$linear_refs) > 1) {
421 warn "Attempting to commit more than one change while ",
422 "--no-rebase is enabled.\n",
423 "If these changes depend on each other, re-running ",
424 "without --no-rebase may be required."
425 }
426 while (1) {
427 my $d = shift @$linear_refs or last;
428 unless (defined $last_rev) {
429 (undef, $last_rev, undef) = cmt_metadata("$d~1");
430 unless (defined $last_rev) {
431 fatal "Unable to extract revision information ",
432 "from commit $d~1";
433 }
434 }
435 if ($_dry_run) {
436 print "diff-tree $d~1 $d\n";
437 } else {
438 my $cmt_rev;
439 my %ed_opts = ( r => $last_rev,
440 log => get_commit_entry($d)->{log},
441 ra => Git::SVN::Ra->new($gs->full_url),
442 config => SVN::Core::config_get_config(
443 $Git::SVN::Ra::config_dir
444 ),
445 tree_a => "$d~1",
446 tree_b => $d,
447 editor_cb => sub {
448 print "Committed r$_[0]\n";
449 $cmt_rev = $_[0];
450 },
451 svn_path => '');
452 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
453 print "No changes\n$d~1 == $d\n";
454 } elsif ($parents->{$d} && @{$parents->{$d}}) {
455 $gs->{inject_parents_dcommit}->{$cmt_rev} =
456 $parents->{$d};
457 }
458 $_fetch_all ? $gs->fetch_all : $gs->fetch;
459 $last_rev = $cmt_rev;
460 next if $_no_rebase;
461
462 # we always want to rebase against the current HEAD,
463 # not any head that was passed to us
464 my @diff = command('diff-tree', $d,
465 $gs->refname, '--');
466 my @finish;
467 if (@diff) {
468 @finish = rebase_cmd();
469 print STDERR "W: $d and ", $gs->refname,
470 " differ, using @finish:\n",
471 join("\n", @diff), "\n";
472 } else {
473 print "No changes between current HEAD and ",
474 $gs->refname,
475 "\nResetting to the latest ",
476 $gs->refname, "\n";
477 @finish = qw/reset --mixed/;
478 }
479 command_noisy(@finish, $gs->refname);
480 if (@diff) {
481 @refs = ();
482 my ($url_, $rev_, $uuid_, $gs_) =
483 working_head_info($head, \@refs);
484 my ($linear_refs_, $parents_) =
485 linearize_history($gs_, \@refs);
486 if (scalar(@$linear_refs) !=
487 scalar(@$linear_refs_)) {
488 fatal "# of revisions changed ",
489 "\nbefore:\n",
490 join("\n", @$linear_refs),
491 "\n\nafter:\n",
492 join("\n", @$linear_refs_), "\n",
493 'If you are attempting to commit ',
494 "merges, try running:\n\t",
495 'git rebase --interactive',
496 '--preserve-merges ',
497 $gs->refname,
498 "\nBefore dcommitting";
499 }
500 if ($url_ ne $url) {
501 fatal "URL mismatch after rebase: ",
502 "$url_ != $url";
503 }
504 if ($uuid_ ne $uuid) {
505 fatal "uuid mismatch after rebase: ",
506 "$uuid_ != $uuid";
507 }
508 # remap parents
509 my (%p, @l, $i);
510 for ($i = 0; $i < scalar @$linear_refs; $i++) {
511 my $new = $linear_refs_->[$i] or next;
512 $p{$new} =
513 $parents->{$linear_refs->[$i]};
514 push @l, $new;
515 }
516 $parents = \%p;
517 $linear_refs = \@l;
518 }
519 }
520 }
521 unlink $gs->{index};
522 }
523
524 sub cmd_find_rev {
525 my $revision_or_hash = shift;
526 my $result;
527 if ($revision_or_hash =~ /^r\d+$/) {
528 my $head = shift;
529 $head ||= 'HEAD';
530 my @refs;
531 my (undef, undef, undef, $gs) = working_head_info($head, \@refs);
532 unless ($gs) {
533 die "Unable to determine upstream SVN information from ",
534 "$head history\n";
535 }
536 my $desired_revision = substr($revision_or_hash, 1);
537 $result = $gs->rev_map_get($desired_revision);
538 } else {
539 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
540 $result = $rev;
541 }
542 print "$result\n" if $result;
543 }
544
545 sub cmd_rebase {
546 command_noisy(qw/update-index --refresh/);
547 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
548 unless ($gs) {
549 die "Unable to determine upstream SVN information from ",
550 "working tree history\n";
551 }
552 if (command(qw/diff-index HEAD --/)) {
553 print STDERR "Cannot rebase with uncommited changes:\n";
554 command_noisy('status');
555 exit 1;
556 }
557 unless ($_local) {
558 # rebase will checkout for us, so no need to do it explicitly
559 $_no_checkout = 'true';
560 $_fetch_all ? $gs->fetch_all : $gs->fetch;
561 }
562 command_noisy(rebase_cmd(), $gs->refname);
563 }
564
565 sub cmd_show_ignore {
566 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
567 $gs ||= Git::SVN->new;
568 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
569 $gs->prop_walk($gs->{path}, $r, sub {
570 my ($gs, $path, $props) = @_;
571 print STDOUT "\n# $path\n";
572 my $s = $props->{'svn:ignore'} or return;
573 $s =~ s/[\r\n]+/\n/g;
574 chomp $s;
575 $s =~ s#^#$path#gm;
576 print STDOUT "$s\n";
577 });
578 }
579
580 sub cmd_show_externals {
581 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
582 $gs ||= Git::SVN->new;
583 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
584 $gs->prop_walk($gs->{path}, $r, sub {
585 my ($gs, $path, $props) = @_;
586 print STDOUT "\n# $path\n";
587 my $s = $props->{'svn:externals'} or return;
588 $s =~ s/[\r\n]+/\n/g;
589 chomp $s;
590 $s =~ s#^#$path#gm;
591 print STDOUT "$s\n";
592 });
593 }
594
595 sub cmd_create_ignore {
596 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
597 $gs ||= Git::SVN->new;
598 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
599 $gs->prop_walk($gs->{path}, $r, sub {
600 my ($gs, $path, $props) = @_;
601 # $path is of the form /path/to/dir/
602 my $ignore = '.' . $path . '.gitignore';
603 my $s = $props->{'svn:ignore'} or return;
604 open(GITIGNORE, '>', $ignore)
605 or fatal("Failed to open `$ignore' for writing: $!");
606 $s =~ s/[\r\n]+/\n/g;
607 chomp $s;
608 # Prefix all patterns so that the ignore doesn't apply
609 # to sub-directories.
610 $s =~ s#^#/#gm;
611 print GITIGNORE "$s\n";
612 close(GITIGNORE)
613 or fatal("Failed to close `$ignore': $!");
614 command_noisy('add', $ignore);
615 });
616 }
617
618 sub canonicalize_path {
619 my ($path) = @_;
620 my $dot_slash_added = 0;
621 if (substr($path, 0, 1) ne "/") {
622 $path = "./" . $path;
623 $dot_slash_added = 1;
624 }
625 # File::Spec->canonpath doesn't collapse x/../y into y (for a
626 # good reason), so let's do this manually.
627 $path =~ s#/+#/#g;
628 $path =~ s#/\.(?:/|$)#/#g;
629 $path =~ s#/[^/]+/\.\.##g;
630 $path =~ s#/$##g;
631 $path =~ s#^\./## if $dot_slash_added;
632 return $path;
633 }
634
635 # get_svnprops(PATH)
636 # ------------------
637 # Helper for cmd_propget and cmd_proplist below.
638 sub get_svnprops {
639 my $path = shift;
640 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
641 $gs ||= Git::SVN->new;
642
643 # prefix THE PATH by the sub-directory from which the user
644 # invoked us.
645 $path = $cmd_dir_prefix . $path;
646 fatal("No such file or directory: $path") unless -e $path;
647 my $is_dir = -d $path ? 1 : 0;
648 $path = $gs->{path} . '/' . $path;
649
650 # canonicalize the path (otherwise libsvn will abort or fail to
651 # find the file)
652 $path = canonicalize_path($path);
653
654 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
655 my $props;
656 if ($is_dir) {
657 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
658 }
659 else {
660 (undef, $props) = $gs->ra->get_file($path, $r, undef);
661 }
662 return $props;
663 }
664
665 # cmd_propget (PROP, PATH)
666 # ------------------------
667 # Print the SVN property PROP for PATH.
668 sub cmd_propget {
669 my ($prop, $path) = @_;
670 $path = '.' if not defined $path;
671 usage(1) if not defined $prop;
672 my $props = get_svnprops($path);
673 if (not defined $props->{$prop}) {
674 fatal("`$path' does not have a `$prop' SVN property.");
675 }
676 print $props->{$prop} . "\n";
677 }
678
679 # cmd_proplist (PATH)
680 # -------------------
681 # Print the list of SVN properties for PATH.
682 sub cmd_proplist {
683 my $path = shift;
684 $path = '.' if not defined $path;
685 my $props = get_svnprops($path);
686 print "Properties on '$path':\n";
687 foreach (sort keys %{$props}) {
688 print " $_\n";
689 }
690 }
691
692 sub cmd_multi_init {
693 my $url = shift;
694 unless (defined $_trunk || defined $_branches || defined $_tags) {
695 usage(1);
696 }
697
698 # there are currently some bugs that prevent multi-init/multi-fetch
699 # setups from working well without this.
700 $Git::SVN::_minimize_url = 1;
701
702 $_prefix = '' unless defined $_prefix;
703 if (defined $url) {
704 $url =~ s#/+$##;
705 init_subdir(@_);
706 }
707 do_git_init_db();
708 if (defined $_trunk) {
709 my $trunk_ref = $_prefix . 'trunk';
710 # try both old-style and new-style lookups:
711 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
712 unless ($gs_trunk) {
713 my ($trunk_url, $trunk_path) =
714 complete_svn_url($url, $_trunk);
715 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
716 undef, $trunk_ref);
717 }
718 }
719 return unless defined $_branches || defined $_tags;
720 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
721 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
722 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
723 }
724
725 sub cmd_multi_fetch {
726 my $remotes = Git::SVN::read_all_remotes();
727 foreach my $repo_id (sort keys %$remotes) {
728 if ($remotes->{$repo_id}->{url}) {
729 Git::SVN::fetch_all($repo_id, $remotes);
730 }
731 }
732 }
733
734 # this command is special because it requires no metadata
735 sub cmd_commit_diff {
736 my ($ta, $tb, $url) = @_;
737 my $usage = "Usage: $0 commit-diff -r<revision> ".
738 "<tree-ish> <tree-ish> [<URL>]";
739 fatal($usage) if (!defined $ta || !defined $tb);
740 my $svn_path;
741 if (!defined $url) {
742 my $gs = eval { Git::SVN->new };
743 if (!$gs) {
744 fatal("Needed URL or usable git-svn --id in ",
745 "the command-line\n", $usage);
746 }
747 $url = $gs->{url};
748 $svn_path = $gs->{path};
749 }
750 unless (defined $_revision) {
751 fatal("-r|--revision is a required argument\n", $usage);
752 }
753 if (defined $_message && defined $_file) {
754 fatal("Both --message/-m and --file/-F specified ",
755 "for the commit message.\n",
756 "I have no idea what you mean");
757 }
758 if (defined $_file) {
759 $_message = file_to_s($_file);
760 } else {
761 $_message ||= get_commit_entry($tb)->{log};
762 }
763 my $ra ||= Git::SVN::Ra->new($url);
764 $svn_path ||= $ra->{svn_path};
765 my $r = $_revision;
766 if ($r eq 'HEAD') {
767 $r = $ra->get_latest_revnum;
768 } elsif ($r !~ /^\d+$/) {
769 die "revision argument: $r not understood by git-svn\n";
770 }
771 my %ed_opts = ( r => $r,
772 log => $_message,
773 ra => $ra,
774 tree_a => $ta,
775 tree_b => $tb,
776 editor_cb => sub { print "Committed r$_[0]\n" },
777 svn_path => $svn_path );
778 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
779 print "No changes\n$ta == $tb\n";
780 }
781 }
782
783 sub cmd_info {
784 my $path = canonicalize_path(shift or ".");
785 unless (scalar(@_) == 0) {
786 die "Too many arguments specified\n";
787 }
788
789 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
790
791 if (!$file_type && !$diff_status) {
792 print STDERR "$path: (Not a versioned resource)\n\n";
793 return;
794 }
795
796 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
797 unless ($gs) {
798 die "Unable to determine upstream SVN information from ",
799 "working tree history\n";
800 }
801 my $full_url = $url . ($path eq "." ? "" : "/$path");
802
803 if ($_url) {
804 print $full_url, "\n";
805 return;
806 }
807
808 my $result = "Path: $path\n";
809 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
810 $result .= "URL: " . $full_url . "\n";
811
812 eval {
813 my $repos_root = $gs->repos_root;
814 Git::SVN::remove_username($repos_root);
815 $result .= "Repository Root: $repos_root\n";
816 };
817 if ($@) {
818 $result .= "Repository Root: (offline)\n";
819 }
820 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
821 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
822
823 $result .= "Node Kind: " .
824 ($file_type eq "dir" ? "directory" : "file") . "\n";
825
826 my $schedule = $diff_status eq "A"
827 ? "add"
828 : ($diff_status eq "D" ? "delete" : "normal");
829 $result .= "Schedule: $schedule\n";
830
831 if ($diff_status eq "A") {
832 print $result, "\n";
833 return;
834 }
835
836 my ($lc_author, $lc_rev, $lc_date_utc);
837 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $path);
838 my $log = command_output_pipe(@args);
839 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
840 while (<$log>) {
841 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
842 $lc_author = $1;
843 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
844 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
845 (undef, $lc_rev, undef) = ::extract_metadata($1);
846 }
847 }
848 close $log;
849
850 Git::SVN::Log::set_local_timezone();
851
852 $result .= "Last Changed Author: $lc_author\n";
853 $result .= "Last Changed Rev: $lc_rev\n";
854 $result .= "Last Changed Date: " .
855 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
856
857 if ($file_type ne "dir") {
858 my $text_last_updated_date =
859 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
860 $result .=
861 "Text Last Updated: " .
862 Git::SVN::Log::format_svn_date($text_last_updated_date) .
863 "\n";
864 my $checksum;
865 if ($diff_status eq "D") {
866 my ($fh, $ctx) =
867 command_output_pipe(qw(cat-file blob), "HEAD:$path");
868 if ($file_type eq "link") {
869 my $file_name = <$fh>;
870 $checksum = md5sum("link $file_name");
871 } else {
872 $checksum = md5sum($fh);
873 }
874 command_close_pipe($fh, $ctx);
875 } elsif ($file_type eq "link") {
876 my $file_name =
877 command(qw(cat-file blob), "HEAD:$path");
878 $checksum =
879 md5sum("link " . $file_name);
880 } else {
881 open FILE, "<", $path or die $!;
882 $checksum = md5sum(\*FILE);
883 close FILE or die $!;
884 }
885 $result .= "Checksum: " . $checksum . "\n";
886 }
887
888 print $result, "\n";
889 }
890
891 ########################### utility functions #########################
892
893 sub rebase_cmd {
894 my @cmd = qw/rebase/;
895 push @cmd, '-v' if $_verbose;
896 push @cmd, qw/--merge/ if $_merge;
897 push @cmd, "--strategy=$_strategy" if $_strategy;
898 @cmd;
899 }
900
901 sub post_fetch_checkout {
902 return if $_no_checkout;
903 my $gs = $Git::SVN::_head or return;
904 return if verify_ref('refs/heads/master^0');
905
906 my $valid_head = verify_ref('HEAD^0');
907 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
908 return if ($valid_head || !verify_ref('HEAD^0'));
909
910 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
911 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
912 return if -f $index;
913
914 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
915 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
916 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
917 print STDERR "Checked out HEAD:\n ",
918 $gs->full_url, " r", $gs->last_rev, "\n";
919 }
920
921 sub complete_svn_url {
922 my ($url, $path) = @_;
923 $path =~ s#/+$##;
924 if ($path !~ m#^[a-z\+]+://#) {
925 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
926 fatal("E: '$path' is not a complete URL ",
927 "and a separate URL is not specified");
928 }
929 return ($url, $path);
930 }
931 return ($path, '');
932 }
933
934 sub complete_url_ls_init {
935 my ($ra, $repo_path, $switch, $pfx) = @_;
936 unless ($repo_path) {
937 print STDERR "W: $switch not specified\n";
938 return;
939 }
940 $repo_path =~ s#/+$##;
941 if ($repo_path =~ m#^[a-z\+]+://#) {
942 $ra = Git::SVN::Ra->new($repo_path);
943 $repo_path = '';
944 } else {
945 $repo_path =~ s#^/+##;
946 unless ($ra) {
947 fatal("E: '$repo_path' is not a complete URL ",
948 "and a separate URL is not specified");
949 }
950 }
951 my $url = $ra->{url};
952 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
953 my $k = "svn-remote.$gs->{repo_id}.url";
954 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
955 if ($orig_url && ($orig_url ne $gs->{url})) {
956 die "$k already set: $orig_url\n",
957 "wanted to set to: $gs->{url}\n";
958 }
959 command_oneline('config', $k, $gs->{url}) unless $orig_url;
960 my $remote_path = "$ra->{svn_path}/$repo_path/*";
961 $remote_path =~ s#/+#/#g;
962 $remote_path =~ s#^/##g;
963 my ($n) = ($switch =~ /^--(\w+)/);
964 if (length $pfx && $pfx !~ m#/$#) {
965 die "--prefix='$pfx' must have a trailing slash '/'\n";
966 }
967 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
968 "$remote_path:refs/remotes/$pfx*");
969 }
970
971 sub verify_ref {
972 my ($ref) = @_;
973 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
974 { STDERR => 0 }); };
975 }
976
977 sub get_tree_from_treeish {
978 my ($treeish) = @_;
979 # $treeish can be a symbolic ref, too:
980 my $type = command_oneline(qw/cat-file -t/, $treeish);
981 my $expected;
982 while ($type eq 'tag') {
983 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
984 }
985 if ($type eq 'commit') {
986 $expected = (grep /^tree /, command(qw/cat-file commit/,
987 $treeish))[0];
988 ($expected) = ($expected =~ /^tree ($sha1)$/o);
989 die "Unable to get tree from $treeish\n" unless $expected;
990 } elsif ($type eq 'tree') {
991 $expected = $treeish;
992 } else {
993 die "$treeish is a $type, expected tree, tag or commit\n";
994 }
995 return $expected;
996 }
997
998 sub get_commit_entry {
999 my ($treeish) = shift;
1000 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1001 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1002 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1003 open my $log_fh, '>', $commit_editmsg or croak $!;
1004
1005 my $type = command_oneline(qw/cat-file -t/, $treeish);
1006 if ($type eq 'commit' || $type eq 'tag') {
1007 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1008 $type, $treeish);
1009 my $in_msg = 0;
1010 while (<$msg_fh>) {
1011 if (!$in_msg) {
1012 $in_msg = 1 if (/^\s*$/);
1013 } elsif (/^git-svn-id: /) {
1014 # skip this for now, we regenerate the
1015 # correct one on re-fetch anyways
1016 # TODO: set *:merge properties or like...
1017 } else {
1018 print $log_fh $_ or croak $!;
1019 }
1020 }
1021 command_close_pipe($msg_fh, $ctx);
1022 }
1023 close $log_fh or croak $!;
1024
1025 if ($_edit || ($type eq 'tree')) {
1026 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1027 # TODO: strip out spaces, comments, like git-commit.sh
1028 system($editor, $commit_editmsg);
1029 }
1030 rename $commit_editmsg, $commit_msg or croak $!;
1031 open $log_fh, '<', $commit_msg or croak $!;
1032 { local $/; chomp($log_entry{log} = <$log_fh>); }
1033 close $log_fh or croak $!;
1034 unlink $commit_msg;
1035 \%log_entry;
1036 }
1037
1038 sub s_to_file {
1039 my ($str, $file, $mode) = @_;
1040 open my $fd,'>',$file or croak $!;
1041 print $fd $str,"\n" or croak $!;
1042 close $fd or croak $!;
1043 chmod ($mode &~ umask, $file) if (defined $mode);
1044 }
1045
1046 sub file_to_s {
1047 my $file = shift;
1048 open my $fd,'<',$file or croak "$!: file: $file\n";
1049 local $/;
1050 my $ret = <$fd>;
1051 close $fd or croak $!;
1052 $ret =~ s/\s*$//s;
1053 return $ret;
1054 }
1055
1056 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1057 sub load_authors {
1058 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1059 my $log = $cmd eq 'log';
1060 while (<$authors>) {
1061 chomp;
1062 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1063 my ($user, $name, $email) = ($1, $2, $3);
1064 if ($log) {
1065 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1066 } else {
1067 $users{$user} = [$name, $email];
1068 }
1069 }
1070 close $authors or croak $!;
1071 }
1072
1073 # convert GetOpt::Long specs for use by git-config
1074 sub read_repo_config {
1075 return unless -d $ENV{GIT_DIR};
1076 my $opts = shift;
1077 my @config_only;
1078 foreach my $o (keys %$opts) {
1079 # if we have mixedCase and a long option-only, then
1080 # it's a config-only variable that we don't need for
1081 # the command-line.
1082 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1083 my $v = $opts->{$o};
1084 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1085 $key =~ s/-//g;
1086 my $arg = 'git-config';
1087 $arg .= ' --int' if ($o =~ /[:=]i$/);
1088 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1089 if (ref $v eq 'ARRAY') {
1090 chomp(my @tmp = `$arg --get-all svn.$key`);
1091 @$v = @tmp if @tmp;
1092 } else {
1093 chomp(my $tmp = `$arg --get svn.$key`);
1094 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1095 $$v = $tmp;
1096 }
1097 }
1098 }
1099 delete @$opts{@config_only} if @config_only;
1100 }
1101
1102 sub extract_metadata {
1103 my $id = shift or return (undef, undef, undef);
1104 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1105 \s([a-f\d\-]+)$/x);
1106 if (!defined $rev || !$uuid || !$url) {
1107 # some of the original repositories I made had
1108 # identifiers like this:
1109 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1110 }
1111 return ($url, $rev, $uuid);
1112 }
1113
1114 sub cmt_metadata {
1115 return extract_metadata((grep(/^git-svn-id: /,
1116 command(qw/cat-file commit/, shift)))[-1]);
1117 }
1118
1119 sub working_head_info {
1120 my ($head, $refs) = @_;
1121 my @args = ('log', '--no-color', '--first-parent');
1122 my ($fh, $ctx) = command_output_pipe(@args, $head);
1123 my $hash;
1124 my %max;
1125 while (<$fh>) {
1126 if ( m{^commit ($::sha1)$} ) {
1127 unshift @$refs, $hash if $hash and $refs;
1128 $hash = $1;
1129 next;
1130 }
1131 next unless s{^\s*(git-svn-id:)}{$1};
1132 my ($url, $rev, $uuid) = extract_metadata($_);
1133 if (defined $url && defined $rev) {
1134 next if $max{$url} and $max{$url} < $rev;
1135 if (my $gs = Git::SVN->find_by_url($url)) {
1136 my $c = $gs->rev_map_get($rev);
1137 if ($c && $c eq $hash) {
1138 close $fh; # break the pipe
1139 return ($url, $rev, $uuid, $gs);
1140 } else {
1141 $max{$url} ||= $gs->rev_map_max;
1142 }
1143 }
1144 }
1145 }
1146 command_close_pipe($fh, $ctx);
1147 (undef, undef, undef, undef);
1148 }
1149
1150 sub read_commit_parents {
1151 my ($parents, $c) = @_;
1152 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1153 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1154 @{$parents->{$c}} = split(/ /, $p);
1155 }
1156
1157 sub linearize_history {
1158 my ($gs, $refs) = @_;
1159 my %parents;
1160 foreach my $c (@$refs) {
1161 read_commit_parents(\%parents, $c);
1162 }
1163
1164 my @linear_refs;
1165 my %skip = ();
1166 my $last_svn_commit = $gs->last_commit;
1167 foreach my $c (reverse @$refs) {
1168 next if $c eq $last_svn_commit;
1169 last if $skip{$c};
1170
1171 unshift @linear_refs, $c;
1172 $skip{$c} = 1;
1173
1174 # we only want the first parent to diff against for linear
1175 # history, we save the rest to inject when we finalize the
1176 # svn commit
1177 my $fp_a = verify_ref("$c~1");
1178 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1179 if (!$fp_a || !$fp_b) {
1180 die "Commit $c\n",
1181 "has no parent commit, and therefore ",
1182 "nothing to diff against.\n",
1183 "You should be working from a repository ",
1184 "originally created by git-svn\n";
1185 }
1186 if ($fp_a ne $fp_b) {
1187 die "$c~1 = $fp_a, however parsing commit $c ",
1188 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1189 }
1190
1191 foreach my $p (@{$parents{$c}}) {
1192 $skip{$p} = 1;
1193 }
1194 }
1195 (\@linear_refs, \%parents);
1196 }
1197
1198 sub find_file_type_and_diff_status {
1199 my ($path) = @_;
1200 return ('dir', '') if $path eq '.';
1201
1202 my $diff_output =
1203 command_oneline(qw(diff --cached --name-status --), $path) || "";
1204 my $diff_status = (split(' ', $diff_output))[0] || "";
1205
1206 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1207
1208 return (undef, undef) if !$diff_status && !$ls_tree;
1209
1210 if ($diff_status eq "A") {
1211 return ("link", $diff_status) if -l $path;
1212 return ("dir", $diff_status) if -d $path;
1213 return ("file", $diff_status);
1214 }
1215
1216 my $mode = (split(' ', $ls_tree))[0] || "";
1217
1218 return ("link", $diff_status) if $mode eq "120000";
1219 return ("dir", $diff_status) if $mode eq "040000";
1220 return ("file", $diff_status);
1221 }
1222
1223 sub md5sum {
1224 my $arg = shift;
1225 my $ref = ref $arg;
1226 my $md5 = Digest::MD5->new();
1227 if ($ref eq 'GLOB' || $ref eq 'IO::File') {
1228 $md5->addfile($arg) or croak $!;
1229 } elsif ($ref eq 'SCALAR') {
1230 $md5->add($$arg) or croak $!;
1231 } elsif (!$ref) {
1232 $md5->add($arg) or croak $!;
1233 } else {
1234 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1235 }
1236 return $md5->hexdigest();
1237 }
1238
1239 package Git::SVN;
1240 use strict;
1241 use warnings;
1242 use Fcntl qw/:DEFAULT :seek/;
1243 use constant rev_map_fmt => 'NH40';
1244 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1245 $_repack $_repack_flags $_use_svm_props $_head
1246 $_use_svnsync_props $no_reuse_existing $_minimize_url
1247 $_use_log_author/;
1248 use Carp qw/croak/;
1249 use File::Path qw/mkpath/;
1250 use File::Copy qw/copy/;
1251 use IPC::Open3;
1252
1253 my ($_gc_nr, $_gc_period);
1254
1255 # properties that we do not log:
1256 my %SKIP_PROP;
1257 BEGIN {
1258 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1259 svn:special svn:executable
1260 svn:entry:committed-rev
1261 svn:entry:last-author
1262 svn:entry:uuid
1263 svn:entry:committed-date/;
1264
1265 # some options are read globally, but can be overridden locally
1266 # per [svn-remote "..."] section. Command-line options will *NOT*
1267 # override options set in an [svn-remote "..."] section
1268 no strict 'refs';
1269 for my $option (qw/follow_parent no_metadata use_svm_props
1270 use_svnsync_props/) {
1271 my $key = $option;
1272 $key =~ tr/_//d;
1273 my $prop = "-$option";
1274 *$option = sub {
1275 my ($self) = @_;
1276 return $self->{$prop} if exists $self->{$prop};
1277 my $k = "svn-remote.$self->{repo_id}.$key";
1278 eval { command_oneline(qw/config --get/, $k) };
1279 if ($@) {
1280 $self->{$prop} = ${"Git::SVN::_$option"};
1281 } else {
1282 my $v = command_oneline(qw/config --bool/,$k);
1283 $self->{$prop} = $v eq 'false' ? 0 : 1;
1284 }
1285 return $self->{$prop};
1286 }
1287 }
1288 }
1289
1290 my (%LOCKFILES, %INDEX_FILES);
1291 END {
1292 unlink keys %LOCKFILES if %LOCKFILES;
1293 unlink keys %INDEX_FILES if %INDEX_FILES;
1294 }
1295
1296 sub resolve_local_globs {
1297 my ($url, $fetch, $glob_spec) = @_;
1298 return unless defined $glob_spec;
1299 my $ref = $glob_spec->{ref};
1300 my $path = $glob_spec->{path};
1301 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1302 next unless m#^refs/remotes/$ref->{regex}$#;
1303 my $p = $1;
1304 my $pathname = desanitize_refname($path->full_path($p));
1305 my $refname = desanitize_refname($ref->full_path($p));
1306 if (my $existing = $fetch->{$pathname}) {
1307 if ($existing ne $refname) {
1308 die "Refspec conflict:\n",
1309 "existing: refs/remotes/$existing\n",
1310 " globbed: refs/remotes/$refname\n";
1311 }
1312 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1313 $u =~ s!^\Q$url\E(/|$)!! or die
1314 "refs/remotes/$refname: '$url' not found in '$u'\n";
1315 if ($pathname ne $u) {
1316 warn "W: Refspec glob conflict ",
1317 "(ref: refs/remotes/$refname):\n",
1318 "expected path: $pathname\n",
1319 " real path: $u\n",
1320 "Continuing ahead with $u\n";
1321 next;
1322 }
1323 } else {
1324 $fetch->{$pathname} = $refname;
1325 }
1326 }
1327 }
1328
1329 sub parse_revision_argument {
1330 my ($base, $head) = @_;
1331 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1332 return ($base, $head);
1333 }
1334 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1335 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1336 return ($head, $head) if ($::_revision eq 'HEAD');
1337 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1338 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1339 die "revision argument: $::_revision not understood by git-svn\n";
1340 }
1341
1342 sub fetch_all {
1343 my ($repo_id, $remotes) = @_;
1344 if (ref $repo_id) {
1345 my $gs = $repo_id;
1346 $repo_id = undef;
1347 $repo_id = $gs->{repo_id};
1348 }
1349 $remotes ||= read_all_remotes();
1350 my $remote = $remotes->{$repo_id} or
1351 die "[svn-remote \"$repo_id\"] unknown\n";
1352 my $fetch = $remote->{fetch};
1353 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1354 my (@gs, @globs);
1355 my $ra = Git::SVN::Ra->new($url);
1356 my $uuid = $ra->get_uuid;
1357 my $head = $ra->get_latest_revnum;
1358 my $base = defined $fetch ? $head : 0;
1359
1360 # read the max revs for wildcard expansion (branches/*, tags/*)
1361 foreach my $t (qw/branches tags/) {
1362 defined $remote->{$t} or next;
1363 push @globs, $remote->{$t};
1364 my $max_rev = eval { tmp_config(qw/--int --get/,
1365 "svn-remote.$repo_id.${t}-maxRev") };
1366 if (defined $max_rev && ($max_rev < $base)) {
1367 $base = $max_rev;
1368 } elsif (!defined $max_rev) {
1369 $base = 0;
1370 }
1371 }
1372
1373 if ($fetch) {
1374 foreach my $p (sort keys %$fetch) {
1375 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1376 my $lr = $gs->rev_map_max;
1377 if (defined $lr) {
1378 $base = $lr if ($lr < $base);
1379 }
1380 push @gs, $gs;
1381 }
1382 }
1383
1384 ($base, $head) = parse_revision_argument($base, $head);
1385 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1386 }
1387
1388 sub read_all_remotes {
1389 my $r = {};
1390 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1391 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*refs/remotes/(.+)\s*$!) {
1392 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1393 $local_ref =~ s{^/}{};
1394 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1395 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1396 $r->{$1}->{url} = $2;
1397 } elsif (m!^(.+)\.(branches|tags)=
1398 (.*):refs/remotes/(.+)\s*$/!x) {
1399 my ($p, $g) = ($3, $4);
1400 my $rs = $r->{$1}->{$2} = {
1401 t => $2,
1402 remote => $1,
1403 path => Git::SVN::GlobSpec->new($p),
1404 ref => Git::SVN::GlobSpec->new($g) };
1405 if (length($rs->{ref}->{right}) != 0) {
1406 die "The '*' glob character must be the last ",
1407 "character of '$g'\n";
1408 }
1409 }
1410 }
1411 $r;
1412 }
1413
1414 sub init_vars {
1415 $_gc_nr = $_gc_period = 1000;
1416 if (defined $_repack || defined $_repack_flags) {
1417 warn "Repack options are obsolete; they have no effect.\n";
1418 }
1419 }
1420
1421 sub verify_remotes_sanity {
1422 return unless -d $ENV{GIT_DIR};
1423 my %seen;
1424 foreach (command(qw/config -l/)) {
1425 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1426 if ($seen{$1}) {
1427 die "Remote ref refs/remote/$1 is tracked by",
1428 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1429 "Please resolve this ambiguity in ",
1430 "your git configuration file before ",
1431 "continuing\n";
1432 }
1433 $seen{$1} = $_;
1434 }
1435 }
1436 }
1437
1438 # we allow more chars than remotes2config.sh...
1439 sub sanitize_remote_name {
1440 my ($name) = @_;
1441 $name =~ tr{A-Za-z0-9:,/+-}{.}c;
1442 $name;
1443 }
1444
1445 sub find_existing_remote {
1446 my ($url, $remotes) = @_;
1447 return undef if $no_reuse_existing;
1448 my $existing;
1449 foreach my $repo_id (keys %$remotes) {
1450 my $u = $remotes->{$repo_id}->{url} or next;
1451 next if $u ne $url;
1452 $existing = $repo_id;
1453 last;
1454 }
1455 $existing;
1456 }
1457
1458 sub init_remote_config {
1459 my ($self, $url, $no_write) = @_;
1460 $url =~ s!/+$!!; # strip trailing slash
1461 my $r = read_all_remotes();
1462 my $existing = find_existing_remote($url, $r);
1463 if ($existing) {
1464 unless ($no_write) {
1465 print STDERR "Using existing ",
1466 "[svn-remote \"$existing\"]\n";
1467 }
1468 $self->{repo_id} = $existing;
1469 } elsif ($_minimize_url) {
1470 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1471 $existing = find_existing_remote($min_url, $r);
1472 if ($existing) {
1473 unless ($no_write) {
1474 print STDERR "Using existing ",
1475 "[svn-remote \"$existing\"]\n";
1476 }
1477 $self->{repo_id} = $existing;
1478 }
1479 if ($min_url ne $url) {
1480 unless ($no_write) {
1481 print STDERR "Using higher level of URL: ",
1482 "$url => $min_url\n";
1483 }
1484 my $old_path = $self->{path};
1485 $self->{path} = $url;
1486 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1487 if (length $old_path) {
1488 $self->{path} .= "/$old_path";
1489 }
1490 $url = $min_url;
1491 }
1492 }
1493 my $orig_url;
1494 if (!$existing) {
1495 # verify that we aren't overwriting anything:
1496 $orig_url = eval {
1497 command_oneline('config', '--get',
1498 "svn-remote.$self->{repo_id}.url")
1499 };
1500 if ($orig_url && ($orig_url ne $url)) {
1501 die "svn-remote.$self->{repo_id}.url already set: ",
1502 "$orig_url\nwanted to set to: $url\n";
1503 }
1504 }
1505 my ($xrepo_id, $xpath) = find_ref($self->refname);
1506 if (defined $xpath) {
1507 die "svn-remote.$xrepo_id.fetch already set to track ",
1508 "$xpath:refs/remotes/", $self->refname, "\n";
1509 }
1510 unless ($no_write) {
1511 command_noisy('config',
1512 "svn-remote.$self->{repo_id}.url", $url);
1513 $self->{path} =~ s{^/}{};
1514 command_noisy('config', '--add',
1515 "svn-remote.$self->{repo_id}.fetch",
1516 "$self->{path}:".$self->refname);
1517 }
1518 $self->{url} = $url;
1519 }
1520
1521 sub find_by_url { # repos_root and, path are optional
1522 my ($class, $full_url, $repos_root, $path) = @_;
1523
1524 return undef unless defined $full_url;
1525 remove_username($full_url);
1526 remove_username($repos_root) if defined $repos_root;
1527 my $remotes = read_all_remotes();
1528 if (defined $full_url && defined $repos_root && !defined $path) {
1529 $path = $full_url;
1530 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1531 }
1532 foreach my $repo_id (keys %$remotes) {
1533 my $u = $remotes->{$repo_id}->{url} or next;
1534 remove_username($u);
1535 next if defined $repos_root && $repos_root ne $u;
1536
1537 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1538 foreach (qw/branches tags/) {
1539 resolve_local_globs($u, $fetch,
1540 $remotes->{$repo_id}->{$_});
1541 }
1542 my $p = $path;
1543 my $rwr = rewrite_root({repo_id => $repo_id});
1544 unless (defined $p) {
1545 $p = $full_url;
1546 my $z = $u;
1547 if ($rwr) {
1548 $z = $rwr;
1549 }
1550 $p =~ s#^\Q$z\E(?:/|$)## or next;
1551 }
1552 foreach my $f (keys %$fetch) {
1553 next if $f ne $p;
1554 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1555 }
1556 }
1557 undef;
1558 }
1559
1560 sub init {
1561 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1562 my $self = _new($class, $repo_id, $ref_id, $path);
1563 if (defined $url) {
1564 $self->init_remote_config($url, $no_write);
1565 }
1566 $self;
1567 }
1568
1569 sub find_ref {
1570 my ($ref_id) = @_;
1571 foreach (command(qw/config -l/)) {
1572 next unless m!^svn-remote\.(.+)\.fetch=
1573 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1574 my ($repo_id, $path, $ref) = ($1, $2, $3);
1575 if ($ref eq $ref_id) {
1576 $path = '' if ($path =~ m#^\./?#);
1577 return ($repo_id, $path);
1578 }
1579 }
1580 (undef, undef, undef);
1581 }
1582
1583 sub new {
1584 my ($class, $ref_id, $repo_id, $path) = @_;
1585 if (defined $ref_id && !defined $repo_id && !defined $path) {
1586 ($repo_id, $path) = find_ref($ref_id);
1587 if (!defined $repo_id) {
1588 die "Could not find a \"svn-remote.*.fetch\" key ",
1589 "in the repository configuration matching: ",
1590 "refs/remotes/$ref_id\n";
1591 }
1592 }
1593 my $self = _new($class, $repo_id, $ref_id, $path);
1594 if (!defined $self->{path} || !length $self->{path}) {
1595 my $fetch = command_oneline('config', '--get',
1596 "svn-remote.$repo_id.fetch",
1597 ":refs/remotes/$ref_id\$") or
1598 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1599 "\":refs/remotes/$ref_id\$\" in config\n";
1600 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1601 }
1602 $self->{url} = command_oneline('config', '--get',
1603 "svn-remote.$repo_id.url") or
1604 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1605 $self->rebuild;
1606 $self;
1607 }
1608
1609 sub refname {
1610 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1611
1612 # It cannot end with a slash /, we'll throw up on this because
1613 # SVN can't have directories with a slash in their name, either:
1614 if ($refname =~ m{/$}) {
1615 die "ref: '$refname' ends with a trailing slash, this is ",
1616 "not permitted by git nor Subversion\n";
1617 }
1618
1619 # It cannot have ASCII control character space, tilde ~, caret ^,
1620 # colon :, question-mark ?, asterisk *, space, or open bracket [
1621 # anywhere.
1622 #
1623 # Additionally, % must be escaped because it is used for escaping
1624 # and we want our escaped refname to be reversible
1625 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1626
1627 # no slash-separated component can begin with a dot .
1628 # /.* becomes /%2E*
1629 $refname =~ s{/\.}{/%2E}g;
1630
1631 # It cannot have two consecutive dots .. anywhere
1632 # .. becomes %2E%2E
1633 $refname =~ s{\.\.}{%2E%2E}g;
1634
1635 return $refname;
1636 }
1637
1638 sub desanitize_refname {
1639 my ($refname) = @_;
1640 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1641 return $refname;
1642 }
1643
1644 sub svm_uuid {
1645 my ($self) = @_;
1646 return $self->{svm}->{uuid} if $self->svm;
1647 $self->ra;
1648 unless ($self->{svm}) {
1649 die "SVM UUID not cached, and reading remotely failed\n";
1650 }
1651 $self->{svm}->{uuid};
1652 }
1653
1654 sub svm {
1655 my ($self) = @_;
1656 return $self->{svm} if $self->{svm};
1657 my $svm;
1658 # see if we have it in our config, first:
1659 eval {
1660 my $section = "svn-remote.$self->{repo_id}";
1661 $svm = {
1662 source => tmp_config('--get', "$section.svm-source"),
1663 uuid => tmp_config('--get', "$section.svm-uuid"),
1664 replace => tmp_config('--get', "$section.svm-replace"),
1665 }
1666 };
1667 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1668 $self->{svm} = $svm;
1669 }
1670 $self->{svm};
1671 }
1672
1673 sub _set_svm_vars {
1674 my ($self, $ra) = @_;
1675 return $ra if $self->svm;
1676
1677 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1678 "(svm:source, svm:uuid) ",
1679 "from the following URLs:\n" );
1680 sub read_svm_props {
1681 my ($self, $ra, $path, $r) = @_;
1682 my $props = ($ra->get_dir($path, $r))[2];
1683 my $src = $props->{'svm:source'};
1684 my $uuid = $props->{'svm:uuid'};
1685 return undef if (!$src || !$uuid);
1686
1687 chomp($src, $uuid);
1688
1689 $uuid =~ m{^[0-9a-f\-]{30,}$}
1690 or die "doesn't look right - svm:uuid is '$uuid'\n";
1691
1692 # the '!' is used to mark the repos_root!/relative/path
1693 $src =~ s{/?!/?}{/};
1694 $src =~ s{/+$}{}; # no trailing slashes please
1695 # username is of no interest
1696 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1697
1698 my $replace = $ra->{url};
1699 $replace .= "/$path" if length $path;
1700
1701 my $section = "svn-remote.$self->{repo_id}";
1702 tmp_config("$section.svm-source", $src);
1703 tmp_config("$section.svm-replace", $replace);
1704 tmp_config("$section.svm-uuid", $uuid);
1705 $self->{svm} = {
1706 source => $src,
1707 uuid => $uuid,
1708 replace => $replace
1709 };
1710 }
1711
1712 my $r = $ra->get_latest_revnum;
1713 my $path = $self->{path};
1714 my %tried;
1715 while (length $path) {
1716 unless ($tried{"$self->{url}/$path"}) {
1717 return $ra if $self->read_svm_props($ra, $path, $r);
1718 $tried{"$self->{url}/$path"} = 1;
1719 }
1720 $path =~ s#/?[^/]+$##;
1721 }
1722 die "Path: '$path' should be ''\n" if $path ne '';
1723 return $ra if $self->read_svm_props($ra, $path, $r);
1724 $tried{"$self->{url}/$path"} = 1;
1725
1726 if ($ra->{repos_root} eq $self->{url}) {
1727 die @err, (map { " $_\n" } keys %tried), "\n";
1728 }
1729
1730 # nope, make sure we're connected to the repository root:
1731 my $ok;
1732 my @tried_b;
1733 $path = $ra->{svn_path};
1734 $ra = Git::SVN::Ra->new($ra->{repos_root});
1735 while (length $path) {
1736 unless ($tried{"$ra->{url}/$path"}) {
1737 $ok = $self->read_svm_props($ra, $path, $r);
1738 last if $ok;
1739 $tried{"$ra->{url}/$path"} = 1;
1740 }
1741 $path =~ s#/?[^/]+$##;
1742 }
1743 die "Path: '$path' should be ''\n" if $path ne '';
1744 $ok ||= $self->read_svm_props($ra, $path, $r);
1745 $tried{"$ra->{url}/$path"} = 1;
1746 if (!$ok) {
1747 die @err, (map { " $_\n" } keys %tried), "\n";
1748 }
1749 Git::SVN::Ra->new($self->{url});
1750 }
1751
1752 sub svnsync {
1753 my ($self) = @_;
1754 return $self->{svnsync} if $self->{svnsync};
1755
1756 if ($self->no_metadata) {
1757 die "Can't have both 'noMetadata' and ",
1758 "'useSvnsyncProps' options set!\n";
1759 }
1760 if ($self->rewrite_root) {
1761 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1762 "options set!\n";
1763 }
1764
1765 my $svnsync;
1766 # see if we have it in our config, first:
1767 eval {
1768 my $section = "svn-remote.$self->{repo_id}";
1769
1770 my $url = tmp_config('--get', "$section.svnsync-url");
1771 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1772 die "doesn't look right - svn:sync-from-url is '$url'\n";
1773
1774 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
1775 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1776 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1777
1778 $svnsync = { url => $url, uuid => $uuid }
1779 };
1780 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1781 return $self->{svnsync} = $svnsync;
1782 }
1783
1784 my $err = "useSvnsyncProps set, but failed to read " .
1785 "svnsync property: svn:sync-from-";
1786 my $rp = $self->ra->rev_proplist(0);
1787
1788 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1789 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1790 die "doesn't look right - svn:sync-from-url is '$url'\n";
1791
1792 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1793 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1794 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1795
1796 my $section = "svn-remote.$self->{repo_id}";
1797 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1798 tmp_config('--add', "$section.svnsync-url", $url);
1799 return $self->{svnsync} = { url => $url, uuid => $uuid };
1800 }
1801
1802 # this allows us to memoize our SVN::Ra UUID locally and avoid a
1803 # remote lookup (useful for 'git svn log').
1804 sub ra_uuid {
1805 my ($self) = @_;
1806 unless ($self->{ra_uuid}) {
1807 my $key = "svn-remote.$self->{repo_id}.uuid";
1808 my $uuid = eval { tmp_config('--get', $key) };
1809 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1810 $self->{ra_uuid} = $uuid;
1811 } else {
1812 die "ra_uuid called without URL\n" unless $self->{url};
1813 $self->{ra_uuid} = $self->ra->get_uuid;
1814 tmp_config('--add', $key, $self->{ra_uuid});
1815 }
1816 }
1817 $self->{ra_uuid};
1818 }
1819
1820 sub _set_repos_root {
1821 my ($self, $repos_root) = @_;
1822 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1823 $repos_root ||= $self->ra->{repos_root};
1824 tmp_config($k, $repos_root);
1825 $repos_root;
1826 }
1827
1828 sub repos_root {
1829 my ($self) = @_;
1830 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1831 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1832 }
1833
1834 sub ra {
1835 my ($self) = shift;
1836 my $ra = Git::SVN::Ra->new($self->{url});
1837 $self->_set_repos_root($ra->{repos_root});
1838 if ($self->use_svm_props && !$self->{svm}) {
1839 if ($self->no_metadata) {
1840 die "Can't have both 'noMetadata' and ",
1841 "'useSvmProps' options set!\n";
1842 } elsif ($self->use_svnsync_props) {
1843 die "Can't have both 'useSvnsyncProps' and ",
1844 "'useSvmProps' options set!\n";
1845 }
1846 $ra = $self->_set_svm_vars($ra);
1847 $self->{-want_revprops} = 1;
1848 }
1849 $ra;
1850 }
1851
1852 sub rel_path {
1853 my ($self) = @_;
1854 my $repos_root = $self->ra->{repos_root};
1855 return $self->{path} if ($self->{url} eq $repos_root);
1856 my $url = $self->{url} .
1857 (length $self->{path} ? "/$self->{path}" : $self->{path});
1858 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1859 $url;
1860 }
1861
1862 # prop_walk(PATH, REV, SUB)
1863 # -------------------------
1864 # Recursively traverse PATH at revision REV and invoke SUB for each
1865 # directory that contains a SVN property. SUB will be invoked as
1866 # follows: &SUB(gs, path, props); where `gs' is this instance of
1867 # Git::SVN, `path' the path to the directory where the properties
1868 # `props' were found. The `path' will be relative to point of checkout,
1869 # that is, if url://repo/trunk is the current Git branch, and that
1870 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
1871 # as `path' (note the trailing `/').
1872 sub prop_walk {
1873 my ($self, $path, $rev, $sub) = @_;
1874
1875 $path =~ s#^/##;
1876 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1877 $path =~ s#^/*#/#g;
1878 my $p = $path;
1879 # Strip the irrelevant part of the path.
1880 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1881 # Ensure the path is terminated by a `/'.
1882 $p =~ s#/*$#/#;
1883
1884 # The properties contain all the internal SVN stuff nobody
1885 # (usually) cares about.
1886 my $interesting_props = 0;
1887 foreach (keys %{$props}) {
1888 # If it doesn't start with `svn:', it must be a
1889 # user-defined property.
1890 ++$interesting_props and next if $_ !~ /^svn:/;
1891 # FIXME: Fragile, if SVN adds new public properties,
1892 # this needs to be updated.
1893 ++$interesting_props if /^svn:(?:ignore|keywords|executable
1894 |eol-style|mime-type
1895 |externals|needs-lock)$/x;
1896 }
1897 &$sub($self, $p, $props) if $interesting_props;
1898
1899 foreach (sort keys %$dirent) {
1900 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1901 $self->prop_walk($path . '/' . $_, $rev, $sub);
1902 }
1903 }
1904
1905 sub last_rev { ($_[0]->last_rev_commit)[0] }
1906 sub last_commit { ($_[0]->last_rev_commit)[1] }
1907
1908 # returns the newest SVN revision number and newest commit SHA1
1909 sub last_rev_commit {
1910 my ($self) = @_;
1911 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1912 return ($self->{last_rev}, $self->{last_commit});
1913 }
1914 my $c = ::verify_ref($self->refname.'^0');
1915 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1916 my $rev = (::cmt_metadata($c))[1];
1917 if (defined $rev) {
1918 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1919 return ($rev, $c);
1920 }
1921 }
1922 my $map_path = $self->map_path;
1923 unless (-e $map_path) {
1924 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1925 return (undef, undef);
1926 }
1927 my ($rev, $commit) = $self->rev_map_max(1);
1928 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
1929 return ($rev, $commit);
1930 }
1931
1932 sub get_fetch_range {
1933 my ($self, $min, $max) = @_;
1934 $max ||= $self->ra->get_latest_revnum;
1935 $min ||= $self->rev_map_max;
1936 (++$min, $max);
1937 }
1938
1939 sub tmp_config {
1940 my (@args) = @_;
1941 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1942 my $config = "$ENV{GIT_DIR}/svn/.metadata";
1943 if (! -f $config && -f $old_def_config) {
1944 rename $old_def_config, $config or
1945 die "Failed rename $old_def_config => $config: $!\n";
1946 }
1947 my $old_config = $ENV{GIT_CONFIG};
1948 $ENV{GIT_CONFIG} = $config;
1949 $@ = undef;
1950 my @ret = eval {
1951 unless (-f $config) {
1952 mkfile($config);
1953 open my $fh, '>', $config or
1954 die "Can't open $config: $!\n";
1955 print $fh "; This file is used internally by ",
1956 "git-svn\n" or die
1957 "Couldn't write to $config: $!\n";
1958 print $fh "; You should not have to edit it\n" or
1959 die "Couldn't write to $config: $!\n";
1960 close $fh or die "Couldn't close $config: $!\n";
1961 }
1962 command('config', @args);
1963 };
1964 my $err = $@;
1965 if (defined $old_config) {
1966 $ENV{GIT_CONFIG} = $old_config;
1967 } else {
1968 delete $ENV{GIT_CONFIG};
1969 }
1970 die $err if $err;
1971 wantarray ? @ret : $ret[0];
1972 }
1973
1974 sub tmp_index_do {
1975 my ($self, $sub) = @_;
1976 my $old_index = $ENV{GIT_INDEX_FILE};
1977 $ENV{GIT_INDEX_FILE} = $self->{index};
1978 $@ = undef;
1979 my @ret = eval {
1980 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
1981 mkpath([$dir]) unless -d $dir;
1982 &$sub;
1983 };
1984 my $err = $@;
1985 if (defined $old_index) {
1986 $ENV{GIT_INDEX_FILE} = $old_index;
1987 } else {
1988 delete $ENV{GIT_INDEX_FILE};
1989 }
1990 die $err if $err;
1991 wantarray ? @ret : $ret[0];
1992 }
1993
1994 sub assert_index_clean {
1995 my ($self, $treeish) = @_;
1996
1997 $self->tmp_index_do(sub {
1998 command_noisy('read-tree', $treeish) unless -e $self->{index};
1999 my $x = command_oneline('write-tree');
2000 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2001 /^tree ($::sha1)/mo);
2002 return if $y eq $x;
2003
2004 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2005 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2006 command_noisy('read-tree', $treeish);
2007 $x = command_oneline('write-tree');
2008 if ($y ne $x) {
2009 ::fatal "trees ($treeish) $y != $x\n",
2010 "Something is seriously wrong...";
2011 }
2012 });
2013 }
2014
2015 sub get_commit_parents {
2016 my ($self, $log_entry) = @_;
2017 my (%seen, @ret, @tmp);
2018 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2019 if (my $ip = $self->{inject_parents}) {
2020 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2021 push @tmp, $commit;
2022 }
2023 }
2024 if (my $cur = ::verify_ref($self->refname.'^0')) {
2025 push @tmp, $cur;
2026 }
2027 if (my $ipd = $self->{inject_parents_dcommit}) {
2028 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2029 push @tmp, @$commit;
2030 }
2031 }
2032 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2033 while (my $p = shift @tmp) {
2034 next if $seen{$p};
2035 $seen{$p} = 1;
2036 push @ret, $p;
2037 # MAXPARENT is defined to 16 in commit-tree.c:
2038 last if @ret >= 16;
2039 }
2040 if (@tmp) {
2041 die "r$log_entry->{revision}: No room for parents:\n\t",
2042 join("\n\t", @tmp), "\n";
2043 }
2044 @ret;
2045 }
2046
2047 sub rewrite_root {
2048 my ($self) = @_;
2049 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2050 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2051 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2052 if ($rwr) {
2053 $rwr =~ s#/+$##;
2054 if ($rwr !~ m#^[a-z\+]+://#) {
2055 die "$rwr is not a valid URL (key: $k)\n";
2056 }
2057 }
2058 $self->{-rewrite_root} = $rwr;
2059 }
2060
2061 sub metadata_url {
2062 my ($self) = @_;
2063 ($self->rewrite_root || $self->{url}) .
2064 (length $self->{path} ? '/' . $self->{path} : '');
2065 }
2066
2067 sub full_url {
2068 my ($self) = @_;
2069 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2070 }
2071
2072
2073 sub set_commit_header_env {
2074 my ($log_entry) = @_;
2075 my %env;
2076 foreach my $ned (qw/NAME EMAIL DATE/) {
2077 foreach my $ac (qw/AUTHOR COMMITTER/) {
2078 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2079 }
2080 }
2081
2082 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2083 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2084 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2085
2086 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2087 ? $log_entry->{commit_name}
2088 : $log_entry->{name};
2089 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2090 ? $log_entry->{commit_email}
2091 : $log_entry->{email};
2092 \%env;
2093 }
2094
2095 sub restore_commit_header_env {
2096 my ($env) = @_;
2097 foreach my $ned (qw/NAME EMAIL DATE/) {
2098 foreach my $ac (qw/AUTHOR COMMITTER/) {
2099 my $k = "GIT_${ac}_${ned}";
2100 if (defined $env->{$k}) {
2101 $ENV{$k} = $env->{$k};
2102 } else {
2103 delete $ENV{$k};
2104 }
2105 }
2106 }
2107 }
2108
2109 sub gc {
2110 command_noisy('gc', '--auto');
2111 };
2112
2113 sub do_git_commit {
2114 my ($self, $log_entry) = @_;
2115 my $lr = $self->last_rev;
2116 if (defined $lr && $lr >= $log_entry->{revision}) {
2117 die "Last fetched revision of ", $self->refname,
2118 " was r$lr, but we are about to fetch: ",
2119 "r$log_entry->{revision}!\n";
2120 }
2121 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2122 croak "$log_entry->{revision} = $c already exists! ",
2123 "Why are we refetching it?\n";
2124 }
2125 my $old_env = set_commit_header_env($log_entry);
2126 my $tree = $log_entry->{tree};
2127 if (!defined $tree) {
2128 $tree = $self->tmp_index_do(sub {
2129 command_oneline('write-tree') });
2130 }
2131 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2132
2133 my @exec = ('git-commit-tree', $tree);
2134 foreach ($self->get_commit_parents($log_entry)) {
2135 push @exec, '-p', $_;
2136 }
2137 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2138 or croak $!;
2139 print $msg_fh $log_entry->{log} or croak $!;
2140 restore_commit_header_env($old_env);
2141 unless ($self->no_metadata) {
2142 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2143 or croak $!;
2144 }
2145 $msg_fh->flush == 0 or croak $!;
2146 close $msg_fh or croak $!;
2147 chomp(my $commit = do { local $/; <$out_fh> });
2148 close $out_fh or croak $!;
2149 waitpid $pid, 0;
2150 croak $? if $?;
2151 if ($commit !~ /^$::sha1$/o) {
2152 die "Failed to commit, invalid sha1: $commit\n";
2153 }
2154
2155 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2156
2157 $self->{last_rev} = $log_entry->{revision};
2158 $self->{last_commit} = $commit;
2159 print "r$log_entry->{revision}";
2160 if (defined $log_entry->{svm_revision}) {
2161 print " (\@$log_entry->{svm_revision})";
2162 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2163 0, $self->svm_uuid);
2164 }
2165 print " = $commit ($self->{ref_id})\n";
2166 if (--$_gc_nr == 0) {
2167 $_gc_nr = $_gc_period;
2168 gc();
2169 }
2170 return $commit;
2171 }
2172
2173 sub match_paths {
2174 my ($self, $paths, $r) = @_;
2175 return 1 if $self->{path} eq '';
2176 if (my $path = $paths->{"/$self->{path}"}) {
2177 return ($path->{action} eq 'D') ? 0 : 1;
2178 }
2179 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2180 if (grep /$self->{path_regex}/, keys %$paths) {
2181 return 1;
2182 }
2183 my $c = '';
2184 foreach (split m#/#, $self->{path}) {
2185 $c .= "/$_";
2186 next unless ($paths->{$c} &&
2187 ($paths->{$c}->{action} =~ /^[AR]$/));
2188 if ($self->ra->check_path($self->{path}, $r) ==
2189 $SVN::Node::dir) {
2190 return 1;
2191 }
2192 }
2193 return 0;
2194 }
2195
2196 sub find_parent_branch {
2197 my ($self, $paths, $rev) = @_;
2198 return undef unless $self->follow_parent;
2199 unless (defined $paths) {
2200 my $err_handler = $SVN::Error::handler;
2201 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2202 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2203 $paths =
2204 Git::SVN::Ra::dup_changed_paths($_[0]) });
2205 $SVN::Error::handler = $err_handler;
2206 }
2207 return undef unless defined $paths;
2208
2209 # look for a parent from another branch:
2210 my @b_path_components = split m#/#, $self->rel_path;
2211 my @a_path_components;
2212 my $i;
2213 while (@b_path_components) {
2214 $i = $paths->{'/'.join('/', @b_path_components)};
2215 last if $i && defined $i->{copyfrom_path};
2216 unshift(@a_path_components, pop(@b_path_components));
2217 }
2218 return undef unless defined $i && defined $i->{copyfrom_path};
2219 my $branch_from = $i->{copyfrom_path};
2220 if (@a_path_components) {
2221 print STDERR "branch_from: $branch_from => ";
2222 $branch_from .= '/'.join('/', @a_path_components);
2223 print STDERR $branch_from, "\n";
2224 }
2225 my $r = $i->{copyfrom_rev};
2226 my $repos_root = $self->ra->{repos_root};
2227 my $url = $self->ra->{url};
2228 my $new_url = $repos_root . $branch_from;
2229 print STDERR "Found possible branch point: ",
2230 "$new_url => ", $self->full_url, ", $r\n";
2231 $branch_from =~ s#^/##;
2232 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2233 unless ($gs) {
2234 my $ref_id = $self->{ref_id};
2235 $ref_id =~ s/\@\d+$//;
2236 $ref_id .= "\@$r";
2237 # just grow a tail if we're not unique enough :x
2238 $ref_id .= '-' while find_ref($ref_id);
2239 print STDERR "Initializing parent: $ref_id\n";
2240 my ($u, $p) = ($new_url, '');
2241 if ($u =~ s#^\Q$url\E(/|$)##) {
2242 $p = $u;
2243 $u = $url;
2244 }
2245 $gs = Git::SVN->init($u, $p, $self->{repo_id}, $ref_id, 1);
2246 }
2247 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2248 if (!defined $r0 || !defined $parent) {
2249 my ($base, $head) = parse_revision_argument(0, $r);
2250 if ($base <= $r) {
2251 $gs->fetch($base, $r);
2252 }
2253 ($r0, $parent) = $gs->last_rev_commit;
2254 }
2255 if (defined $r0 && defined $parent) {
2256 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2257 my $ed;
2258 if ($self->ra->can_do_switch) {
2259 $self->assert_index_clean($parent);
2260 print STDERR "Following parent with do_switch\n";
2261 # do_switch works with svn/trunk >= r22312, but that
2262 # is not included with SVN 1.4.3 (the latest version
2263 # at the moment), so we can't rely on it
2264 $self->{last_commit} = $parent;
2265 $ed = SVN::Git::Fetcher->new($self);
2266 $gs->ra->gs_do_switch($r0, $rev, $gs,
2267 $self->full_url, $ed)
2268 or die "SVN connection failed somewhere...\n";
2269 } elsif ($self->ra->trees_match($new_url, $r0,
2270 $self->full_url, $rev)) {
2271 print STDERR "Trees match:\n",
2272 " $new_url\@$r0\n",
2273 " ${\$self->full_url}\@$rev\n",
2274 "Following parent with no changes\n";
2275 $self->tmp_index_do(sub {
2276 command_noisy('read-tree', $parent);
2277 });
2278 $self->{last_commit} = $parent;
2279 } else {
2280 print STDERR "Following parent with do_update\n";
2281 $ed = SVN::Git::Fetcher->new($self);
2282 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2283 or die "SVN connection failed somewhere...\n";
2284 }
2285 print STDERR "Successfully followed parent\n";
2286 return $self->make_log_entry($rev, [$parent], $ed);
2287 }
2288 return undef;
2289 }
2290
2291 sub do_fetch {
2292 my ($self, $paths, $rev) = @_;
2293 my $ed;
2294 my ($last_rev, @parents);
2295 if (my $lc = $self->last_commit) {
2296 # we can have a branch that was deleted, then re-added
2297 # under the same name but copied from another path, in
2298 # which case we'll have multiple parents (we don't
2299 # want to break the original ref, nor lose copypath info):
2300 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2301 push @{$log_entry->{parents}}, $lc;
2302 return $log_entry;
2303 }
2304 $ed = SVN::Git::Fetcher->new($self);
2305 $last_rev = $self->{last_rev};
2306 $ed->{c} = $lc;
2307 @parents = ($lc);
2308 } else {
2309 $last_rev = $rev;
2310 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2311 return $log_entry;
2312 }
2313 $ed = SVN::Git::Fetcher->new($self);
2314 }
2315 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2316 die "SVN connection failed somewhere...\n";
2317 }
2318 $self->make_log_entry($rev, \@parents, $ed);
2319 }
2320
2321 sub get_untracked {
2322 my ($self, $ed) = @_;
2323 my @out;
2324 my $h = $ed->{empty};
2325 foreach (sort keys %$h) {
2326 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2327 push @out, " $act: " . uri_encode($_);
2328 warn "W: $act: $_\n";
2329 }
2330 foreach my $t (qw/dir_prop file_prop/) {
2331 $h = $ed->{$t} or next;
2332 foreach my $path (sort keys %$h) {
2333 my $ppath = $path eq '' ? '.' : $path;
2334 foreach my $prop (sort keys %{$h->{$path}}) {
2335 next if $SKIP_PROP{$prop};
2336 my $v = $h->{$path}->{$prop};
2337 my $t_ppath_prop = "$t: " .
2338 uri_encode($ppath) . ' ' .
2339 uri_encode($prop);
2340 if (defined $v) {
2341 push @out, " +$t_ppath_prop " .
2342 uri_encode($v);
2343 } else {
2344 push @out, " -$t_ppath_prop";
2345 }
2346 }
2347 }
2348 }
2349 foreach my $t (qw/absent_file absent_directory/) {
2350 $h = $ed->{$t} or next;
2351 foreach my $parent (sort keys %$h) {
2352 foreach my $path (sort @{$h->{$parent}}) {
2353 push @out, " $t: " .
2354 uri_encode("$parent/$path");
2355 warn "W: $t: $parent/$path ",
2356 "Insufficient permissions?\n";
2357 }
2358 }
2359 }
2360 \@out;
2361 }
2362
2363 sub parse_svn_date {
2364 my $date = shift || return '+0000 1970-01-01 00:00:00';
2365 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2366 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2367 croak "Unable to parse date: $date\n";
2368 "+0000 $Y-$m-$d $H:$M:$S";
2369 }
2370
2371 sub check_author {
2372 my ($author) = @_;
2373 if (!defined $author || length $author == 0) {
2374 $author = '(no author)';
2375 }
2376 if (defined $::_authors && ! defined $::users{$author}) {
2377 die "Author: $author not defined in $::_authors file\n";
2378 }
2379 $author;
2380 }
2381
2382 sub make_log_entry {
2383 my ($self, $rev, $parents, $ed) = @_;
2384 my $untracked = $self->get_untracked($ed);
2385
2386 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2387 print $un "r$rev\n" or croak $!;
2388 print $un $_, "\n" foreach @$untracked;
2389 my %log_entry = ( parents => $parents || [], revision => $rev,
2390 log => '');
2391
2392 my $headrev;
2393 my $logged = delete $self->{logged_rev_props};
2394 if (!$logged || $self->{-want_revprops}) {
2395 my $rp = $self->ra->rev_proplist($rev);
2396 foreach (sort keys %$rp) {
2397 my $v = $rp->{$_};
2398 if (/^svn:(author|date|log)$/) {
2399 $log_entry{$1} = $v;
2400 } elsif ($_ eq 'svm:headrev') {
2401 $headrev = $v;
2402 } else {
2403 print $un " rev_prop: ", uri_encode($_), ' ',
2404 uri_encode($v), "\n";
2405 }
2406 }
2407 } else {
2408 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2409 }
2410 close $un or croak $!;
2411
2412 $log_entry{date} = parse_svn_date($log_entry{date});
2413 $log_entry{log} .= "\n";
2414 my $author = $log_entry{author} = check_author($log_entry{author});
2415 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2416 : ($author, undef);
2417
2418 my ($commit_name, $commit_email) = ($name, $email);
2419 if ($_use_log_author) {
2420 my $name_field;
2421 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2422 $name_field = $1;
2423 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2424 $name_field = $1;
2425 }
2426 if (!defined $name_field) {
2427 #
2428 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2429 ($name, $email) = ($1, $2);
2430 } elsif ($name_field =~ /(.*)@/) {
2431 ($name, $email) = ($1, $name_field);
2432 } else {
2433 ($name, $email) = ($name_field, 'unknown');
2434 }
2435 }
2436 if (defined $headrev && $self->use_svm_props) {
2437 if ($self->rewrite_root) {
2438 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2439 "options set!\n";
2440 }
2441 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2442 # we don't want "SVM: initializing mirror for junk" ...
2443 return undef if $r == 0;
2444 my $svm = $self->svm;
2445 if ($uuid ne $svm->{uuid}) {
2446 die "UUID mismatch on SVM path:\n",
2447 "expected: $svm->{uuid}\n",
2448 " got: $uuid\n";
2449 }
2450 my $full_url = $self->full_url;
2451 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2452 die "Failed to replace '$svm->{replace}' with ",
2453 "'$svm->{source}' in $full_url\n";
2454 # throw away username for storing in records
2455 remove_username($full_url);
2456 $log_entry{metadata} = "$full_url\@$r $uuid";
2457 $log_entry{svm_revision} = $r;
2458 $email ||= "$author\@$uuid";
2459 $commit_email ||= "$author\@$uuid";
2460 } elsif ($self->use_svnsync_props) {
2461 my $full_url = $self->svnsync->{url};
2462 $full_url .= "/$self->{path}" if length $self->{path};
2463 remove_username($full_url);
2464 my $uuid = $self->svnsync->{uuid};
2465 $log_entry{metadata} = "$full_url\@$rev $uuid";
2466 $email ||= "$author\@$uuid";
2467 $commit_email ||= "$author\@$uuid";
2468 } else {
2469 my $url = $self->metadata_url;
2470 remove_username($url);
2471 $log_entry{metadata} = "$url\@$rev " .
2472 $self->ra->get_uuid;
2473 $email ||= "$author\@" . $self->ra->get_uuid;
2474 $commit_email ||= "$author\@" . $self->ra->get_uuid;
2475 }
2476 $log_entry{name} = $name;
2477 $log_entry{email} = $email;
2478 $log_entry{commit_name} = $commit_name;
2479 $log_entry{commit_email} = $commit_email;
2480 \%log_entry;
2481 }
2482
2483 sub fetch {
2484 my ($self, $min_rev, $max_rev, @parents) = @_;
2485 my ($last_rev, $last_commit) = $self->last_rev_commit;
2486 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2487 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2488 }
2489
2490 sub set_tree_cb {
2491 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2492 $self->{inject_parents} = { $rev => $tree };
2493 $self->fetch(undef, undef);
2494 }
2495
2496 sub set_tree {
2497 my ($self, $tree) = (shift, shift);
2498 my $log_entry = ::get_commit_entry($tree);
2499 unless ($self->{last_rev}) {
2500 fatal("Must have an existing revision to commit");
2501 }
2502 my %ed_opts = ( r => $self->{last_rev},
2503 log => $log_entry->{log},
2504 ra => $self->ra,
2505 tree_a => $self->{last_commit},
2506 tree_b => $tree,
2507 editor_cb => sub {
2508 $self->set_tree_cb($log_entry, $tree, @_) },
2509 svn_path => $self->{path} );
2510 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2511 print "No changes\nr$self->{last_rev} = $tree\n";
2512 }
2513 }
2514
2515 sub rebuild_from_rev_db {
2516 my ($self, $path) = @_;
2517 my $r = -1;
2518 open my $fh, '<', $path or croak "open: $!";
2519 while (<$fh>) {
2520 length($_) == 41 or croak "inconsistent size in ($_) != 41";
2521 chomp($_);
2522 ++$r;
2523 next if $_ eq ('0' x 40);
2524 $self->rev_map_set($r, $_);
2525 print "r$r = $_\n";
2526 }
2527 close $fh or croak "close: $!";
2528 unlink $path or croak "unlink: $!";
2529 }
2530
2531 sub rebuild {
2532 my ($self) = @_;
2533 my $map_path = $self->map_path;
2534 return if (-e $map_path && ! -z $map_path);
2535 return unless ::verify_ref($self->refname.'^0');
2536 if ($self->use_svm_props || $self->no_metadata) {
2537 my $rev_db = $self->rev_db_path;
2538 $self->rebuild_from_rev_db($rev_db);
2539 if ($self->use_svm_props) {
2540 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2541 $self->rebuild_from_rev_db($svm_rev_db);
2542 }
2543 $self->unlink_rev_db_symlink;
2544 return;
2545 }
2546 print "Rebuilding $map_path ...\n";
2547 my ($log, $ctx) =
2548 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2549 $self->refname, '--');
2550 my $full_url = $self->full_url;
2551 remove_username($full_url);
2552 my $svn_uuid = $self->ra_uuid;
2553 my $c;
2554 while (<$log>) {
2555 if ( m{^commit ($::sha1)$} ) {
2556 $c = $1;
2557 next;
2558 }
2559 next unless s{^\s*(git-svn-id:)}{$1};
2560 my ($url, $rev, $uuid) = ::extract_metadata($_);
2561 remove_username($url);
2562
2563 # ignore merges (from set-tree)
2564 next if (!defined $rev || !$uuid);
2565
2566 # if we merged or otherwise started elsewhere, this is
2567 # how we break out of it
2568 if (($uuid ne $svn_uuid) ||
2569 ($full_url && $url && ($url ne $full_url))) {
2570 next;
2571 }
2572
2573 $self->rev_map_set($rev, $c);
2574 print "r$rev = $c\n";
2575 }
2576 command_close_pipe($log, $ctx);
2577 print "Done rebuilding $map_path\n";
2578 my $rev_db_path = $self->rev_db_path;
2579 if (-f $self->rev_db_path) {
2580 unlink $self->rev_db_path or croak "unlink: $!";
2581 }
2582 $self->unlink_rev_db_symlink;
2583 }
2584
2585 # rev_map:
2586 # Tie::File seems to be prone to offset errors if revisions get sparse,
2587 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
2588 # one of my favorite modules is out :< Next up would be one of the DBM
2589 # modules, but I'm not sure which is most portable...
2590 #
2591 # This is the replacement for the rev_db format, which was too big
2592 # and inefficient for large repositories with a lot of sparse history
2593 # (mainly tags)
2594 #
2595 # The format is this:
2596 # - 24 bytes for every record,
2597 # * 4 bytes for the integer representing an SVN revision number
2598 # * 20 bytes representing the sha1 of a git commit
2599 # - No empty padding records like the old format
2600 # (except the last record, which can be overwritten)
2601 # - new records are written append-only since SVN revision numbers
2602 # increase monotonically
2603 # - lookups on SVN revision number are done via a binary search
2604 # - Piping the file to xxd -c24 is a good way of dumping it for
2605 # viewing or editing (piped back through xxd -r), should the need
2606 # ever arise.
2607 # - The last record can be padding revision with an all-zero sha1
2608 # This is used to optimize fetch performance when using multiple
2609 # "fetch" directives in .git/config
2610 #
2611 # These files are disposable unless noMetadata or useSvmProps is set
2612
2613 sub _rev_map_set {
2614 my ($fh, $rev, $commit) = @_;
2615
2616 my $size = (stat($fh))[7];
2617 ($size % 24) == 0 or croak "inconsistent size: $size";
2618
2619 my $wr_offset = 0;
2620 if ($size > 0) {
2621 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2622 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2623 $read == 24 or croak "read only $read bytes (!= 24)";
2624 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2625 if ($last_commit eq ('0' x40)) {
2626 if ($size >= 48) {
2627 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2628 $read = sysread($fh, $buf, 24) or
2629 croak "read: $!";
2630 $read == 24 or
2631 croak "read only $read bytes (!= 24)";
2632 ($last_rev, $last_commit) =
2633 unpack(rev_map_fmt, $buf);
2634 if ($last_commit eq ('0' x40)) {
2635 croak "inconsistent .rev_map\n";
2636 }
2637 }
2638 if ($last_rev >= $rev) {
2639 croak "last_rev is higher!: $last_rev >= $rev";
2640 }
2641 $wr_offset = -24;
2642 }
2643 }
2644 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2645 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2646 croak "write: $!";
2647 }
2648
2649 sub mkfile {
2650 my ($path) = @_;
2651 unless (-e $path) {
2652 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2653 mkpath([$dir]) unless -d $dir;
2654 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2655 close $fh or die "Couldn't close (create) $path: $!\n";
2656 }
2657 }
2658
2659 sub rev_map_set {
2660 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2661 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2662 my $db = $self->map_path($uuid);
2663 my $db_lock = "$db.lock";
2664 my $sig;
2665 if ($update_ref) {
2666 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2667 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2668 }
2669 mkfile($db);
2670
2671 $LOCKFILES{$db_lock} = 1;
2672 my $sync;
2673 # both of these options make our .rev_db file very, very important
2674 # and we can't afford to lose it because rebuild() won't work
2675 if ($self->use_svm_props || $self->no_metadata) {
2676 $sync = 1;
2677 copy($db, $db_lock) or die "rev_map_set(@_): ",
2678 "Failed to copy: ",
2679 "$db => $db_lock ($!)\n";
2680 } else {
2681 rename $db, $db_lock or die "rev_map_set(@_): ",
2682 "Failed to rename: ",
2683 "$db => $db_lock ($!)\n";
2684 }
2685
2686 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2687 or croak "Couldn't open $db_lock: $!\n";
2688 _rev_map_set($fh, $rev, $commit);
2689 if ($sync) {
2690 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2691 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2692 }
2693 close $fh or croak $!;
2694 if ($update_ref) {
2695 $_head = $self;
2696 command_noisy('update-ref', '-m', "r$rev",
2697 $self->refname, $commit);
2698 }
2699 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2700 "$db_lock => $db ($!)\n";
2701 delete $LOCKFILES{$db_lock};
2702 if ($update_ref) {
2703 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2704 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2705 kill $sig, $$ if defined $sig;
2706 }
2707 }
2708
2709 # If want_commit, this will return an array of (rev, commit) where
2710 # commit _must_ be a valid commit in the archive.
2711 # Otherwise, it'll return the max revision (whether or not the
2712 # commit is valid or just a 0x40 placeholder).
2713 sub rev_map_max {
2714 my ($self, $want_commit) = @_;
2715 $self->rebuild;
2716 my $map_path = $self->map_path;
2717 stat $map_path or return $want_commit ? (0, undef) : 0;
2718 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2719 my $size = (stat($fh))[7];
2720 ($size % 24) == 0 or croak "inconsistent size: $size";
2721
2722 if ($size == 0) {
2723 close $fh or croak "close: $!";
2724 return $want_commit ? (0, undef) : 0;
2725 }
2726
2727 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2728 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2729 my ($r, $c) = unpack(rev_map_fmt, $buf);
2730 if ($want_commit && $c eq ('0' x40)) {
2731 if ($size < 48) {
2732 return $want_commit ? (0, undef) : 0;
2733 }
2734 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2735 sysread($fh, $buf, 24) == 24 or croak "read: $!";
2736 ($r, $c) = unpack(rev_map_fmt, $buf);
2737 if ($c eq ('0'x40)) {
2738 croak "Penultimate record is all-zeroes in $map_path";
2739 }
2740 }
2741 close $fh or croak "close: $!";
2742 $want_commit ? ($r, $c) : $r;
2743 }
2744
2745 sub rev_map_get {
2746 my ($self, $rev, $uuid) = @_;
2747 my $map_path = $self->map_path($uuid);
2748 return undef unless -e $map_path;
2749
2750 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2751 my $size = (stat($fh))[7];
2752 ($size % 24) == 0 or croak "inconsistent size: $size";
2753
2754 if ($size == 0) {
2755 close $fh or croak "close: $fh";
2756 return undef;
2757 }
2758
2759 my ($l, $u) = (0, $size - 24);
2760 my ($r, $c, $buf);
2761
2762 while ($l <= $u) {
2763 my $i = int(($l/24 + $u/24) / 2) * 24;
2764 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2765 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2766 my ($r, $c) = unpack('NH40', $buf);
2767
2768 if ($r < $rev) {
2769 $l = $i + 24;
2770 } elsif ($r > $rev) {
2771 $u = $i - 24;
2772 } else { # $r == $rev
2773 close($fh) or croak "close: $!";
2774 return $c eq ('0' x 40) ? undef : $c;
2775 }
2776 }
2777 close($fh) or croak "close: $!";
2778 undef;
2779 }
2780
2781 # Finds the first svn revision that exists on (if $eq_ok is true) or
2782 # before $rev for the current branch. It will not search any lower
2783 # than $min_rev. Returns the git commit hash and svn revision number
2784 # if found, else (undef, undef).
2785 sub find_rev_before {
2786 my ($self, $rev, $eq_ok, $min_rev) = @_;
2787 --$rev unless $eq_ok;
2788 $min_rev ||= 1;
2789 while ($rev >= $min_rev) {
2790 if (my $c = $self->rev_map_get($rev)) {
2791 return ($rev, $c);
2792 }
2793 --$rev;
2794 }
2795 return (undef, undef);
2796 }
2797
2798 # Finds the first svn revision that exists on (if $eq_ok is true) or
2799 # after $rev for the current branch. It will not search any higher
2800 # than $max_rev. Returns the git commit hash and svn revision number
2801 # if found, else (undef, undef).
2802 sub find_rev_after {
2803 my ($self, $rev, $eq_ok, $max_rev) = @_;
2804 ++$rev unless $eq_ok;
2805 $max_rev ||= $self->rev_map_max;
2806 while ($rev <= $max_rev) {
2807 if (my $c = $self->rev_map_get($rev)) {
2808 return ($rev, $c);
2809 }
2810 ++$rev;
2811 }
2812 return (undef, undef);
2813 }
2814
2815 sub _new {
2816 my ($class, $repo_id, $ref_id, $path) = @_;
2817 unless (defined $repo_id && length $repo_id) {
2818 $repo_id = $Git::SVN::default_repo_id;
2819 }
2820 unless (defined $ref_id && length $ref_id) {
2821 $_[2] = $ref_id = $Git::SVN::default_ref_id;
2822 }
2823 $_[1] = $repo_id = sanitize_remote_name($repo_id);
2824 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2825 $_[3] = $path = '' unless (defined $path);
2826 mkpath(["$ENV{GIT_DIR}/svn"]);
2827 bless {
2828 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2829 path => $path, config => "$ENV{GIT_DIR}/svn/config",
2830 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
2831 }
2832
2833 # for read-only access of old .rev_db formats
2834 sub unlink_rev_db_symlink {
2835 my ($self) = @_;
2836 my $link = $self->rev_db_path;
2837 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2838 if (-l $link) {
2839 unlink $link or croak "unlink: $link failed!";
2840 }
2841 }
2842
2843 sub rev_db_path {
2844 my ($self, $uuid) = @_;
2845 my $db_path = $self->map_path($uuid);
2846 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2847 or croak "map_path: $db_path does not contain '/.rev_map.' !";
2848 $db_path;
2849 }
2850
2851 # the new replacement for .rev_db
2852 sub map_path {
2853 my ($self, $uuid) = @_;
2854 $uuid ||= $self->ra_uuid;
2855 "$self->{map_root}.$uuid";
2856 }
2857
2858 sub uri_encode {
2859 my ($f) = @_;
2860 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2861 $f
2862 }
2863
2864 sub remove_username {
2865 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2866 }
2867
2868 package Git::SVN::Prompt;
2869 use strict;
2870 use warnings;
2871 require SVN::Core;
2872 use vars qw/$_no_auth_cache $_username/;
2873
2874 sub simple {
2875 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2876 $may_save = undef if $_no_auth_cache;
2877 $default_username = $_username if defined $_username;
2878 if (defined $default_username && length $default_username) {
2879 if (defined $realm && length $realm) {
2880 print STDERR "Authentication realm: $realm\n";
2881 STDERR->flush;
2882 }
2883 $cred->username($default_username);
2884 } else {
2885 username($cred, $realm, $may_save, $pool);
2886 }
2887 $cred->password(_read_password("Password for '" .
2888 $cred->username . "': ", $realm));
2889 $cred->may_save($may_save);
2890 $SVN::_Core::SVN_NO_ERROR;
2891 }
2892
2893 sub ssl_server_trust {
2894 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2895 $may_save = undef if $_no_auth_cache;
2896 print STDERR "Error validating server certificate for '$realm':\n";
2897 {
2898 no warnings 'once';
2899 # All variables SVN::Auth::SSL::* are used only once,
2900 # so we're shutting up Perl warnings about this.
2901 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2902 print STDERR " - The certificate is not issued ",
2903 "by a trusted authority. Use the\n",
2904 " fingerprint to validate ",
2905 "the certificate manually!\n";
2906 }
2907 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2908 print STDERR " - The certificate hostname ",
2909 "does not match.\n";
2910 }
2911 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2912 print STDERR " - The certificate is not yet valid.\n";
2913 }
2914 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2915 print STDERR " - The certificate has expired.\n";
2916 }
2917 if ($failures & $SVN::Auth::SSL::OTHER) {
2918 print STDERR " - The certificate has ",
2919 "an unknown error.\n";
2920 }
2921 } # no warnings 'once'
2922 printf STDERR
2923 "Certificate information:\n".
2924 " - Hostname: %s\n".
2925 " - Valid: from %s until %s\n".
2926 " - Issuer: %s\n".
2927 " - Fingerprint: %s\n",
2928 map $cert_info->$_, qw(hostname valid_from valid_until
2929 issuer_dname fingerprint);
2930 my $choice;
2931 prompt:
2932 print STDERR $may_save ?
2933 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2934 "(R)eject or accept (t)emporarily? ";
2935 STDERR->flush;
2936 $choice = lc(substr(<STDIN> || 'R', 0, 1));
2937 if ($choice =~ /^t$/i) {
2938 $cred->may_save(undef);
2939 } elsif ($choice =~ /^r$/i) {
2940 return -1;
2941 } elsif ($may_save && $choice =~ /^p$/i) {
2942 $cred->may_save($may_save);
2943 } else {
2944 goto prompt;
2945 }
2946 $cred->accepted_failures($failures);
2947 $SVN::_Core::SVN_NO_ERROR;
2948 }
2949
2950 sub ssl_client_cert {
2951 my ($cred, $realm, $may_save, $pool) = @_;
2952 $may_save = undef if $_no_auth_cache;
2953 print STDERR "Client certificate filename: ";
2954 STDERR->flush;
2955 chomp(my $filename = <STDIN>);
2956 $cred->cert_file($filename);
2957 $cred->may_save($may_save);
2958 $SVN::_Core::SVN_NO_ERROR;
2959 }
2960
2961 sub ssl_client_cert_pw {
2962 my ($cred, $realm, $may_save, $pool) = @_;
2963 $may_save = undef if $_no_auth_cache;
2964 $cred->password(_read_password("Password: ", $realm));
2965 $cred->may_save($may_save);
2966 $SVN::_Core::SVN_NO_ERROR;
2967 }
2968
2969 sub username {
2970 my ($cred, $realm, $may_save, $pool) = @_;
2971 $may_save = undef if $_no_auth_cache;
2972 if (defined $realm && length $realm) {
2973 print STDERR "Authentication realm: $realm\n";
2974 }
2975 my $username;
2976 if (defined $_username) {
2977 $username = $_username;
2978 } else {
2979 print STDERR "Username: ";
2980 STDERR->flush;
2981 chomp($username = <STDIN>);
2982 }
2983 $cred->username($username);
2984 $cred->may_save($may_save);
2985 $SVN::_Core::SVN_NO_ERROR;
2986 }
2987
2988 sub _read_password {
2989 my ($prompt, $realm) = @_;
2990 print STDERR $prompt;
2991 STDERR->flush;
2992 require Term::ReadKey;
2993 Term::ReadKey::ReadMode('noecho');
2994 my $password = '';
2995 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
2996 last if $key =~ /[\012\015]/; # \n\r
2997 $password .= $key;
2998 }
2999 Term::ReadKey::ReadMode('restore');
3000 print STDERR "\n";
3001 STDERR->flush;
3002 $password;
3003 }
3004
3005 package SVN::Git::Fetcher;
3006 use vars qw/@ISA/;
3007 use strict;
3008 use warnings;
3009 use Carp qw/croak/;
3010 use IO::File qw//;
3011
3012 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
3013 sub new {
3014 my ($class, $git_svn) = @_;
3015 my $self = SVN::Delta::Editor->new;
3016 bless $self, $class;
3017 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
3018 $self->{empty} = {};
3019 $self->{dir_prop} = {};
3020 $self->{file_prop} = {};
3021 $self->{absent_dir} = {};
3022 $self->{absent_file} = {};
3023 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3024 $self;
3025 }
3026
3027 sub set_path_strip {
3028 my ($self, $path) = @_;
3029 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3030 }
3031
3032 sub open_root {
3033 { path => '' };
3034 }
3035
3036 sub open_directory {
3037 my ($self, $path, $pb, $rev) = @_;
3038 { path => $path };
3039 }
3040
3041 sub git_path {
3042 my ($self, $path) = @_;
3043 if ($self->{path_strip}) {
3044 $path =~ s!$self->{path_strip}!! or
3045 die "Failed to strip path '$path' ($self->{path_strip})\n";
3046 }
3047 $path;
3048 }
3049
3050 sub delete_entry {
3051 my ($self, $path, $rev, $pb) = @_;
3052
3053 my $gpath = $self->git_path($path);
3054 return undef if ($gpath eq '');
3055
3056 # remove entire directories.
3057 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
3058 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3059 -r --name-only -z/,
3060 $self->{c}, '--', $gpath);
3061 local $/ = "\0";
3062 while (<$ls>) {
3063 chomp;
3064 $self->{gii}->remove($_);
3065 print "\tD\t$_\n" unless $::_q;
3066 }
3067 print "\tD\t$gpath/\n" unless $::_q;
3068 command_close_pipe($ls, $ctx);
3069 $self->{empty}->{$path} = 0
3070 } else {
3071 $self->{gii}->remove($gpath);
3072 print "\tD\t$gpath\n" unless $::_q;
3073 }
3074 undef;
3075 }
3076
3077 sub open_file {
3078 my ($self, $path, $pb, $rev) = @_;
3079 my $gpath = $self->git_path($path);
3080 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
3081 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3082 unless (defined $mode && defined $blob) {
3083 die "$path was not found in commit $self->{c} (r$rev)\n";
3084 }
3085 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3086 pool => SVN::Pool->new, action => 'M' };
3087 }
3088
3089 sub add_file {
3090 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3091 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3092 delete $self->{empty}->{$dir};
3093 { path => $path, mode_a => 100644, mode_b => 100644,
3094 pool => SVN::Pool->new, action => 'A' };
3095 }
3096
3097 sub add_directory {
3098 my ($self, $path, $cp_path, $cp_rev) = @_;
3099 my $gpath = $self->git_path($path);
3100 if ($gpath eq '') {
3101 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3102 -r --name-only -z/,
3103 $self->{c});
3104 local $/ = "\0";
3105 while (<$ls>) {
3106 chomp;
3107 $self->{gii}->remove($_);
3108 print "\tD\t$_\n" unless $::_q;
3109 }
3110 command_close_pipe($ls, $ctx);
3111 $self->{empty}->{$path} = 0;
3112 }
3113 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3114 delete $self->{empty}->{$dir};
3115 $self->{empty}->{$path} = 1;
3116 { path => $path };
3117 }
3118
3119 sub change_dir_prop {
3120 my ($self, $db, $prop, $value) = @_;
3121 $self->{dir_prop}->{$db->{path}} ||= {};
3122 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3123 undef;
3124 }
3125
3126 sub absent_directory {
3127 my ($self, $path, $pb) = @_;
3128 $self->{absent_dir}->{$pb->{path}} ||= [];
3129 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3130 undef;
3131 }
3132
3133 sub absent_file {
3134 my ($self, $path, $pb) = @_;
3135 $self->{absent_file}->{$pb->{path}} ||= [];
3136 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3137 undef;
3138 }
3139
3140 sub change_file_prop {
3141 my ($self, $fb, $prop, $value) = @_;
3142 if ($prop eq 'svn:executable') {
3143 if ($fb->{mode_b} != 120000) {
3144 $fb->{mode_b} = defined $value ? 100755 : 100644;
3145 }
3146 } elsif ($prop eq 'svn:special') {
3147 $fb->{mode_b} = defined $value ? 120000 : 100644;
3148 } else {
3149 $self->{file_prop}->{$fb->{path}} ||= {};
3150 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3151 }
3152 undef;
3153 }
3154
3155 sub apply_textdelta {
3156 my ($self, $fb, $exp) = @_;
3157 my $fh = IO::File->new_tmpfile;
3158 $fh->autoflush(1);
3159 # $fh gets auto-closed() by SVN::TxDelta::apply(),
3160 # (but $base does not,) so dup() it for reading in close_file
3161 open my $dup, '<&', $fh or croak $!;
3162 my $base = IO::File->new_tmpfile;
3163 $base->autoflush(1);
3164 if ($fb->{blob}) {
3165 defined (my $pid = fork) or croak $!;
3166 if (!$pid) {
3167 open STDOUT, '>&', $base or croak $!;
3168 print STDOUT 'link ' if ($fb->{mode_a} == 120000);
3169 exec qw/git-cat-file blob/, $fb->{blob} or croak $!;
3170 }
3171 waitpid $pid, 0;
3172 croak $? if $?;
3173
3174 if (defined $exp) {
3175 seek $base, 0, 0 or croak $!;
3176 my $got = ::md5sum($base);
3177 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3178 "expected: $exp\n",
3179 " got: $got\n" if ($got ne $exp);
3180 }
3181 }
3182 seek $base, 0, 0 or croak $!;
3183 $fb->{fh} = $dup;
3184 $fb->{base} = $base;
3185 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
3186 }
3187
3188 sub close_file {
3189 my ($self, $fb, $exp) = @_;
3190 my $hash;
3191 my $path = $self->git_path($fb->{path});
3192 if (my $fh = $fb->{fh}) {
3193 if (defined $exp) {
3194 seek($fh, 0, 0) or croak $!;
3195 my $got = ::md5sum($fh);
3196 if ($got ne $exp) {
3197 die "Checksum mismatch: $path\n",
3198 "expected: $exp\n got: $got\n";
3199 }
3200 }
3201 sysseek($fh, 0, 0) or croak $!;
3202 if ($fb->{mode_b} == 120000) {
3203 eval {
3204 sysread($fh, my $buf, 5) == 5 or croak $!;
3205 $buf eq 'link ' or die "$path has mode 120000",
3206 " but is not a link";
3207 };
3208 if ($@) {
3209 warn "$@\n";
3210 sysseek($fh, 0, 0) or croak $!;
3211 }
3212 }
3213 defined(my $pid = open my $out,'-|') or die "Can't fork: $!\n";
3214 if (!$pid) {
3215 open STDIN, '<&', $fh or croak $!;
3216 exec qw/git-hash-object -w --stdin/ or croak $!;
3217 }
3218 chomp($hash = do { local $/; <$out> });
3219 close $out or croak $!;
3220 close $fh or croak $!;
3221 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3222 close $fb->{base} or croak $!;
3223 } else {
3224 $hash = $fb->{blob} or die "no blob information\n";
3225 }
3226 $fb->{pool}->clear;
3227 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3228 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3229 undef;
3230 }
3231
3232 sub abort_edit {
3233 my $self = shift;
3234 $self->{nr} = $self->{gii}->{nr};
3235 delete $self->{gii};
3236 $self->SUPER::abort_edit(@_);
3237 }
3238
3239 sub close_edit {
3240 my $self = shift;
3241 $self->{git_commit_ok} = 1;
3242 $self->{nr} = $self->{gii}->{nr};
3243 delete $self->{gii};
3244 $self->SUPER::close_edit(@_);
3245 }
3246
3247 package SVN::Git::Editor;
3248 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3249 use strict;
3250 use warnings;
3251 use Carp qw/croak/;
3252 use IO::File;
3253
3254 sub new {
3255 my ($class, $opts) = @_;
3256 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3257 die "$_ required!\n" unless (defined $opts->{$_});
3258 }
3259
3260 my $pool = SVN::Pool->new;
3261 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3262 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3263 $opts->{r}, $mods);
3264
3265 # $opts->{ra} functions should not be used after this:
3266 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3267 $opts->{editor_cb}, $pool);
3268 my $self = SVN::Delta::Editor->new(@ce, $pool);
3269 bless $self, $class;
3270 foreach (qw/svn_path r tree_a tree_b/) {
3271 $self->{$_} = $opts->{$_};
3272 }
3273 $self->{url} = $opts->{ra}->{url};
3274 $self->{mods} = $mods;
3275 $self->{types} = $types;
3276 $self->{pool} = $pool;
3277 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3278 $self->{rm} = { };
3279 $self->{path_prefix} = length $self->{svn_path} ?
3280 "$self->{svn_path}/" : '';
3281 return $self;
3282 }
3283
3284 sub generate_diff {
3285 my ($tree_a, $tree_b) = @_;
3286 my @diff_tree = qw(diff-tree -z -r);
3287 if ($_cp_similarity) {
3288 push @diff_tree, "-C$_cp_similarity";
3289 } else {
3290 push @diff_tree, '-C';
3291 }
3292 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3293 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3294 push @diff_tree, $tree_a, $tree_b;
3295 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3296 local $/ = "\0";
3297 my $state = 'meta';
3298 my @mods;
3299 while (<$diff_fh>) {
3300 chomp $_; # this gets rid of the trailing "\0"
3301 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3302 $::sha1\s($::sha1)\s
3303 ([MTCRAD])\d*$/xo) {
3304 push @mods, { mode_a => $1, mode_b => $2,
3305 sha1_b => $3, chg => $4 };
3306 if ($4 =~ /^(?:C|R)$/) {
3307 $state = 'file_a';
3308 } else {
3309 $state = 'file_b';
3310 }
3311 } elsif ($state eq 'file_a') {
3312 my $x = $mods[$#mods] or croak "Empty array\n";
3313 if ($x->{chg} !~ /^(?:C|R)$/) {
3314 croak "Error parsing $_, $x->{chg}\n";
3315 }
3316 $x->{file_a} = $_;
3317 $state = 'file_b';
3318 } elsif ($state eq 'file_b') {
3319 my $x = $mods[$#mods] or croak "Empty array\n";
3320 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3321 croak "Error parsing $_, $x->{chg}\n";
3322 }
3323 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3324 croak "Error parsing $_, $x->{chg}\n";
3325 }
3326 $x->{file_b} = $_;
3327 $state = 'meta';
3328 } else {
3329 croak "Error parsing $_\n";
3330 }
3331 }
3332 command_close_pipe($diff_fh, $ctx);
3333 \@mods;
3334 }
3335
3336 sub check_diff_paths {
3337 my ($ra, $pfx, $rev, $mods) = @_;
3338 my %types;
3339 $pfx .= '/' if length $pfx;
3340
3341 sub type_diff_paths {
3342 my ($ra, $types, $path, $rev) = @_;
3343 my @p = split m#/+#, $path;
3344 my $c = shift @p;
3345 unless (defined $types->{$c}) {
3346 $types->{$c} = $ra->check_path($c, $rev);
3347 }
3348 while (@p) {
3349 $c .= '/' . shift @p;
3350 next if defined $types->{$c};
3351 $types->{$c} = $ra->check_path($c, $rev);
3352 }
3353 }
3354
3355 foreach my $m (@$mods) {
3356 foreach my $f (qw/file_a file_b/) {
3357 next unless defined $m->{$f};
3358 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3359 if (length $pfx.$dir && ! defined $types{$dir}) {
3360 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3361 }
3362 }
3363 }
3364 \%types;
3365 }
3366
3367 sub split_path {
3368 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3369 }
3370
3371 sub repo_path {
3372 my ($self, $path) = @_;
3373 $self->{path_prefix}.(defined $path ? $path : '');
3374 }
3375
3376 sub url_path {
3377 my ($self, $path) = @_;
3378 if ($self->{url} =~ m#^https?://#) {
3379 $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3380 }
3381 $self->{url} . '/' . $self->repo_path($path);
3382 }
3383
3384 sub rmdirs {
3385 my ($self) = @_;
3386 my $rm = $self->{rm};
3387 delete $rm->{''}; # we never delete the url we're tracking
3388 return unless %$rm;
3389
3390 foreach (keys %$rm) {
3391 my @d = split m#/#, $_;
3392 my $c = shift @d;
3393 $rm->{$c} = 1;
3394 while (@d) {
3395 $c .= '/' . shift @d;
3396 $rm->{$c} = 1;
3397 }
3398 }
3399 delete $rm->{$self->{svn_path}};
3400 delete $rm->{''}; # we never delete the url we're tracking
3401 return unless %$rm;
3402
3403 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3404 $self->{tree_b});
3405 local $/ = "\0";
3406 while (<$fh>) {
3407 chomp;
3408 my @dn = split m#/#, $_;
3409 while (pop @dn) {
3410 delete $rm->{join '/', @dn};
3411 }
3412 unless (%$rm) {
3413 close $fh;
3414 return;
3415 }
3416 }
3417 command_close_pipe($fh, $ctx);
3418
3419 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3420 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3421 $self->close_directory($bat->{$d}, $p);
3422 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3423 print "\tD+\t$d/\n" unless $::_q;
3424 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3425 delete $bat->{$d};
3426 }
3427 }
3428
3429 sub open_or_add_dir {
3430 my ($self, $full_path, $baton) = @_;
3431 my $t = $self->{types}->{$full_path};
3432 if (!defined $t) {
3433 die "$full_path not known in r$self->{r} or we have a bug!\n";
3434 }
3435 {
3436 no warnings 'once';
3437 # SVN::Node::none and SVN::Node::file are used only once,
3438 # so we're shutting up Perl's warnings about them.
3439 if ($t == $SVN::Node::none) {
3440 return $self->add_directory($full_path, $baton,
3441 undef, -1, $self->{pool});
3442 } elsif ($t == $SVN::Node::dir) {
3443 return $self->open_directory($full_path, $baton,
3444 $self->{r}, $self->{pool});
3445 } # no warnings 'once'
3446 print STDERR "$full_path already exists in repository at ",
3447 "r$self->{r} and it is not a directory (",
3448 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3449 } # no warnings 'once'
3450 exit 1;
3451 }
3452
3453 sub ensure_path {
3454 my ($self, $path) = @_;
3455 my $bat = $self->{bat};
3456 my $repo_path = $self->repo_path($path);
3457 return $bat->{''} unless (length $repo_path);
3458 my @p = split m#/+#, $repo_path;
3459 my $c = shift @p;
3460 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3461 while (@p) {
3462 my $c0 = $c;
3463 $c .= '/' . shift @p;
3464 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3465 }
3466 return $bat->{$c};
3467 }
3468
3469 sub A {
3470 my ($self, $m) = @_;
3471 my ($dir, $file) = split_path($m->{file_b});
3472 my $pbat = $self->ensure_path($dir);
3473 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3474 undef, -1);
3475 print "\tA\t$m->{file_b}\n" unless $::_q;
3476 $self->chg_file($fbat, $m);
3477 $self->close_file($fbat,undef,$self->{pool});
3478 }
3479
3480 sub C {
3481 my ($self, $m) = @_;
3482 my ($dir, $file) = split_path($m->{file_b});
3483 my $pbat = $self->ensure_path($dir);
3484 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3485 $self->url_path($m->{file_a}), $self->{r});
3486 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3487 $self->chg_file($fbat, $m);
3488 $self->close_file($fbat,undef,$self->{pool});
3489 }
3490
3491 sub delete_entry {
3492 my ($self, $path, $pbat) = @_;
3493 my $rpath = $self->repo_path($path);
3494 my ($dir, $file) = split_path($rpath);
3495 $self->{rm}->{$dir} = 1;
3496 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3497 }
3498
3499 sub R {
3500 my ($self, $m) = @_;
3501 my ($dir, $file) = split_path($m->{file_b});
3502 my $pbat = $self->ensure_path($dir);
3503 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3504 $self->url_path($m->{file_a}), $self->{r});
3505 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3506 $self->chg_file($fbat, $m);
3507 $self->close_file($fbat,undef,$self->{pool});
3508
3509 ($dir, $file) = split_path($m->{file_a});
3510 $pbat = $self->ensure_path($dir);
3511 $self->delete_entry($m->{file_a}, $pbat);
3512 }
3513
3514 sub M {
3515 my ($self, $m) = @_;
3516 my ($dir, $file) = split_path($m->{file_b});
3517 my $pbat = $self->ensure_path($dir);
3518 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3519 $pbat,$self->{r},$self->{pool});
3520 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3521 $self->chg_file($fbat, $m);
3522 $self->close_file($fbat,undef,$self->{pool});
3523 }
3524
3525 sub T { shift->M(@_) }
3526
3527 sub change_file_prop {
3528 my ($self, $fbat, $pname, $pval) = @_;
3529 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3530 }
3531
3532 sub chg_file {
3533 my ($self, $fbat, $m) = @_;
3534 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3535 $self->change_file_prop($fbat,'svn:executable','*');
3536 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3537 $self->change_file_prop($fbat,'svn:executable',undef);
3538 }
3539 my $fh = IO::File->new_tmpfile or croak $!;
3540 if ($m->{mode_b} =~ /^120/) {
3541 print $fh 'link ' or croak $!;
3542 $self->change_file_prop($fbat,'svn:special','*');
3543 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3544 $self->change_file_prop($fbat,'svn:special',undef);
3545 }
3546 defined(my $pid = fork) or croak $!;
3547 if (!$pid) {
3548 open STDOUT, '>&', $fh or croak $!;
3549 exec qw/git-cat-file blob/, $m->{sha1_b} or croak $!;
3550 }
3551 waitpid $pid, 0;
3552 croak $? if $?;
3553 $fh->flush == 0 or croak $!;
3554 seek $fh, 0, 0 or croak $!;
3555
3556 my $exp = ::md5sum($fh);
3557 seek $fh, 0, 0 or croak $!;
3558
3559 my $pool = SVN::Pool->new;
3560 my $atd = $self->apply_textdelta($fbat, undef, $pool);
3561 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3562 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3563 $pool->clear;
3564
3565 close $fh or croak $!;
3566 }
3567
3568 sub D {
3569 my ($self, $m) = @_;
3570 my ($dir, $file) = split_path($m->{file_b});
3571 my $pbat = $self->ensure_path($dir);
3572 print "\tD\t$m->{file_b}\n" unless $::_q;
3573 $self->delete_entry($m->{file_b}, $pbat);
3574 }
3575
3576 sub close_edit {
3577 my ($self) = @_;
3578 my ($p,$bat) = ($self->{pool}, $self->{bat});
3579 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3580 next if $_ eq '';
3581 $self->close_directory($bat->{$_}, $p);
3582 }
3583 $self->close_directory($bat->{''}, $p);
3584 $self->SUPER::close_edit($p);
3585 $p->clear;
3586 }
3587
3588 sub abort_edit {
3589 my ($self) = @_;
3590 $self->SUPER::abort_edit($self->{pool});
3591 }
3592
3593 sub DESTROY {
3594 my $self = shift;
3595 $self->SUPER::DESTROY(@_);
3596 $self->{pool}->clear;
3597 }
3598
3599 # this drives the editor
3600 sub apply_diff {
3601 my ($self) = @_;
3602 my $mods = $self->{mods};
3603 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3604 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3605 my $f = $m->{chg};
3606 if (defined $o{$f}) {
3607 $self->$f($m);
3608 } else {
3609 fatal("Invalid change type: $f");
3610 }
3611 }
3612 $self->rmdirs if $_rmdir;
3613 if (@$mods == 0) {
3614 $self->abort_edit;
3615 } else {
3616 $self->close_edit;
3617 }
3618 return scalar @$mods;
3619 }
3620
3621 package Git::SVN::Ra;
3622 use vars qw/@ISA $config_dir $_log_window_size/;
3623 use strict;
3624 use warnings;
3625 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3626
3627 BEGIN {
3628 # enforce temporary pool usage for some simple functions
3629 no strict 'refs';
3630 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3631 my $SUPER = "SUPER::$f";
3632 *$f = sub {
3633 my $self = shift;
3634 my $pool = SVN::Pool->new;
3635 my @ret = $self->$SUPER(@_,$pool);
3636 $pool->clear;
3637 wantarray ? @ret : $ret[0];
3638 };
3639 }
3640 }
3641
3642 sub _auth_providers () {
3643 [
3644 SVN::Client::get_simple_provider(),
3645 SVN::Client::get_ssl_server_trust_file_provider(),
3646 SVN::Client::get_simple_prompt_provider(
3647 \&Git::SVN::Prompt::simple, 2),
3648 SVN::Client::get_ssl_client_cert_file_provider(),
3649 SVN::Client::get_ssl_client_cert_prompt_provider(
3650 \&Git::SVN::Prompt::ssl_client_cert, 2),
3651 SVN::Client::get_ssl_client_cert_pw_file_provider(),
3652 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3653 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3654 SVN::Client::get_username_provider(),
3655 SVN::Client::get_ssl_server_trust_prompt_provider(
3656 \&Git::SVN::Prompt::ssl_server_trust),
3657 SVN::Client::get_username_prompt_provider(
3658 \&Git::SVN::Prompt::username, 2)
3659 ]
3660 }
3661
3662 sub escape_uri_only {
3663 my ($uri) = @_;
3664 my @tmp;
3665 foreach (split m{/}, $uri) {
3666 s/([^\w.-])/sprintf("%%%02X",ord($1))/eg;
3667 push @tmp, $_;
3668 }
3669 join('/', @tmp);
3670 }
3671
3672 sub escape_url {
3673 my ($url) = @_;
3674 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3675 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3676 $url = "$scheme://$domain$uri";
3677 }
3678 $url;
3679 }
3680
3681 sub new {
3682 my ($class, $url) = @_;
3683 $url =~ s!/+$!!;
3684 return $RA if ($RA && $RA->{url} eq $url);
3685
3686 SVN::_Core::svn_config_ensure($config_dir, undef);
3687 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3688 my $config = SVN::Core::config_get_config($config_dir);
3689 $RA = undef;
3690 my $dont_store_passwords = 1;
3691 my $conf_t = ${$config}{'config'};
3692 {
3693 no warnings 'once';
3694 # The usage of $SVN::_Core::SVN_CONFIG_* variables
3695 # produces warnings that variables are used only once.
3696 # I had not found the better way to shut them up, so
3697 # the warnings of type 'once' are disabled in this block.
3698 if (SVN::_Core::svn_config_get_bool($conf_t,
3699 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3700 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3701 1) == 0) {
3702 SVN::_Core::svn_auth_set_parameter($baton,
3703 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3704 bless (\$dont_store_passwords, "_p_void"));
3705 }
3706 if (SVN::_Core::svn_config_get_bool($conf_t,
3707 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3708 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3709 1) == 0) {
3710 $Git::SVN::Prompt::_no_auth_cache = 1;
3711 }
3712 } # no warnings 'once'
3713 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3714 config => $config,
3715 pool => SVN::Pool->new,
3716 auth_provider_callbacks => $callbacks);
3717 $self->{url} = $url;
3718 $self->{svn_path} = $url;
3719 $self->{repos_root} = $self->get_repos_root;
3720 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3721 $self->{cache} = { check_path => { r => 0, data => {} },
3722 get_dir => { r => 0, data => {} } };
3723 $RA = bless $self, $class;
3724 }
3725
3726 sub check_path {
3727 my ($self, $path, $r) = @_;
3728 my $cache = $self->{cache}->{check_path};
3729 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3730 return $cache->{data}->{$path};
3731 }
3732 my $pool = SVN::Pool->new;
3733 my $t = $self->SUPER::check_path($path, $r, $pool);
3734 $pool->clear;
3735 if ($r != $cache->{r}) {
3736 %{$cache->{data}} = ();
3737 $cache->{r} = $r;
3738 }
3739 $cache->{data}->{$path} = $t;
3740 }
3741
3742 sub get_dir {
3743 my ($self, $dir, $r) = @_;
3744 my $cache = $self->{cache}->{get_dir};
3745 if ($r == $cache->{r}) {
3746 if (my $x = $cache->{data}->{$dir}) {
3747 return wantarray ? @$x : $x->[0];
3748 }
3749 }
3750 my $pool = SVN::Pool->new;
3751 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3752 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3753 $pool->clear;
3754 if ($r != $cache->{r}) {
3755 %{$cache->{data}} = ();
3756 $cache->{r} = $r;
3757 }
3758 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3759 wantarray ? (\%dirents, $r, $props) : \%dirents;
3760 }
3761
3762 sub DESTROY {
3763 # do not call the real DESTROY since we store ourselves in $RA
3764 }
3765
3766 sub get_log {
3767 my ($self, @args) = @_;
3768 my $pool = SVN::Pool->new;
3769 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3770 my $ret = $self->SUPER::get_log(@args, $pool);
3771 $pool->clear;
3772 $ret;
3773 }
3774
3775 sub trees_match {
3776 my ($self, $url1, $rev1, $url2, $rev2) = @_;
3777 my $ctx = SVN::Client->new(auth => _auth_providers);
3778 my $out = IO::File->new_tmpfile;
3779
3780 # older SVN (1.1.x) doesn't take $pool as the last parameter for
3781 # $ctx->diff(), so we'll create a default one
3782 my $pool = SVN::Pool->new_default_sub;
3783
3784 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3785 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3786 $out->flush;
3787 my $ret = (($out->stat)[7] == 0);
3788 close $out or croak $!;
3789
3790 $ret;
3791 }
3792
3793 sub get_commit_editor {
3794 my ($self, $log, $cb, $pool) = @_;
3795 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3796 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3797 }
3798
3799 sub gs_do_update {
3800 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3801 my $new = ($rev_a == $rev_b);
3802 my $path = $gs->{path};
3803
3804 if ($new && -e $gs->{index}) {
3805 unlink $gs->{index} or die
3806 "Couldn't unlink index: $gs->{index}: $!\n";
3807 }
3808 my $pool = SVN::Pool->new;
3809 $editor->set_path_strip($path);
3810 my (@pc) = split m#/#, $path;
3811 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3812 1, $editor, $pool);
3813 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3814
3815 # Since we can't rely on svn_ra_reparent being available, we'll
3816 # just have to do some magic with set_path to make it so
3817 # we only want a partial path.
3818 my $sp = '';
3819 my $final = join('/', @pc);
3820 while (@pc) {
3821 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3822 $sp .= '/' if length $sp;
3823 $sp .= shift @pc;
3824 }
3825 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3826
3827 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3828
3829 $reporter->finish_report($pool);
3830 $pool->clear;
3831 $editor->{git_commit_ok};
3832 }
3833
3834 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3835 # svn_ra_reparent didn't work before 1.4)
3836 sub gs_do_switch {
3837 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3838 my $path = $gs->{path};
3839 my $pool = SVN::Pool->new;
3840
3841 my $full_url = $self->{url};
3842 my $old_url = $full_url;
3843 $full_url .= '/' . escape_uri_only($path) if length $path;
3844 my ($ra, $reparented);
3845 if ($old_url ne $full_url) {
3846 if ($old_url !~ m#^svn(\+ssh)?://#) {
3847 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3848 $pool);
3849 $self->{url} = $full_url;
3850 $reparented = 1;
3851 } else {
3852 $_[0] = undef;
3853 $self = undef;
3854 $RA = undef;
3855 $ra = Git::SVN::Ra->new($full_url);
3856 $ra_invalid = 1;
3857 }
3858 }
3859 $ra ||= $self;
3860 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3861 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3862 $reporter->set_path('', $rev_a, 0, @lock, $pool);
3863 $reporter->finish_report($pool);
3864
3865 if ($reparented) {
3866 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3867 $self->{url} = $old_url;
3868 }
3869
3870 $pool->clear;
3871 $editor->{git_commit_ok};
3872 }
3873
3874 sub longest_common_path {
3875 my ($gsv, $globs) = @_;
3876 my %common;
3877 my $common_max = scalar @$gsv;
3878
3879 foreach my $gs (@$gsv) {
3880 my @tmp = split m#/#, $gs->{path};
3881 my $p = '';
3882 foreach (@tmp) {
3883 $p .= length($p) ? "/$_" : $_;
3884 $common{$p} ||= 0;
3885 $common{$p}++;
3886 }
3887 }
3888 $globs ||= [];
3889 $common_max += scalar @$globs;
3890 foreach my $glob (@$globs) {
3891 my @tmp = split m#/#, $glob->{path}->{left};
3892 my $p = '';
3893 foreach (@tmp) {
3894 $p .= length($p) ? "/$_" : $_;
3895 $common{$p} ||= 0;
3896 $common{$p}++;
3897 }
3898 }
3899
3900 my $longest_path = '';
3901 foreach (sort {length $b <=> length $a} keys %common) {
3902 if ($common{$_} == $common_max) {
3903 $longest_path = $_;
3904 last;
3905 }
3906 }
3907 $longest_path;
3908 }
3909
3910 sub gs_fetch_loop_common {
3911 my ($self, $base, $head, $gsv, $globs) = @_;
3912 return if ($base > $head);
3913 my $inc = $_log_window_size;
3914 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
3915 my $longest_path = longest_common_path($gsv, $globs);
3916 my $ra_url = $self->{url};
3917 while (1) {
3918 my %revs;
3919 my $err;
3920 my $err_handler = $SVN::Error::handler;
3921 $SVN::Error::handler = sub {
3922 ($err) = @_;
3923 skip_unknown_revs($err);
3924 };
3925 sub _cb {
3926 my ($paths, $r, $author, $date, $log) = @_;
3927 [ dup_changed_paths($paths),
3928 { author => $author, date => $date, log => $log } ];
3929 }
3930 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
3931 sub { $revs{$_[1]} = _cb(@_) });
3932 if ($err && $max >= $head) {
3933 print STDERR "Path '$longest_path' ",
3934 "was probably deleted:\n",
3935 $err->expanded_message,
3936 "\nWill attempt to follow ",
3937 "revisions r$min .. r$max ",
3938 "committed before the deletion\n";
3939 my $hi = $max;
3940 while (--$hi >= $min) {
3941 my $ok;
3942 $self->get_log([$longest_path], $min, $hi,
3943 0, 1, 1, sub {
3944 $ok ||= $_[1];
3945 $revs{$_[1]} = _cb(@_) });
3946 if ($ok) {
3947 print STDERR "r$min .. r$ok OK\n";
3948 last;
3949 }
3950 }
3951 }
3952 $SVN::Error::handler = $err_handler;
3953
3954 my %exists = map { $_->{path} => $_ } @$gsv;
3955 foreach my $r (sort {$a <=> $b} keys %revs) {
3956 my ($paths, $logged) = @{$revs{$r}};
3957
3958 foreach my $gs ($self->match_globs(\%exists, $paths,
3959 $globs, $r)) {
3960 if ($gs->rev_map_max >= $r) {
3961 next;
3962 }
3963 next unless $gs->match_paths($paths, $r);
3964 $gs->{logged_rev_props} = $logged;
3965 if (my $last_commit = $gs->last_commit) {
3966 $gs->assert_index_clean($last_commit);
3967 }
3968 my $log_entry = $gs->do_fetch($paths, $r);
3969 if ($log_entry) {
3970 $gs->do_git_commit($log_entry);
3971 }
3972 $INDEX_FILES{$gs->{index}} = 1;
3973 }
3974 foreach my $g (@$globs) {
3975 my $k = "svn-remote.$g->{remote}." .
3976 "$g->{t}-maxRev";
3977 Git::SVN::tmp_config($k, $r);
3978 }
3979 if ($ra_invalid) {
3980 $_[0] = undef;
3981 $self = undef;
3982 $RA = undef;
3983 $self = Git::SVN::Ra->new($ra_url);
3984 $ra_invalid = undef;
3985 }
3986 }
3987 # pre-fill the .rev_db since it'll eventually get filled in
3988 # with '0' x40 if something new gets committed
3989 foreach my $gs (@$gsv) {
3990 next if $gs->rev_map_max >= $max;
3991 next if defined $gs->rev_map_get($max);
3992 $gs->rev_map_set($max, 0 x40);
3993 }
3994 foreach my $g (@$globs) {
3995 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
3996 Git::SVN::tmp_config($k, $max);
3997 }
3998 last if $max >= $head;
3999 $min = $max + 1;
4000 $max += $inc;
4001 $max = $head if ($max > $head);
4002 }
4003 Git::SVN::gc();
4004 }
4005
4006 sub match_globs {
4007 my ($self, $exists, $paths, $globs, $r) = @_;
4008
4009 sub get_dir_check {
4010 my ($self, $exists, $g, $r) = @_;
4011 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
4012 return unless scalar @x == 3;
4013 my $dirents = $x[0];
4014 foreach my $de (keys %$dirents) {
4015 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4016 my $p = $g->{path}->full_path($de);
4017 next if $exists->{$p};
4018 next if (length $g->{path}->{right} &&
4019 ($self->check_path($p, $r) !=
4020 $SVN::Node::dir));
4021 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4022 $g->{ref}->full_path($de), 1);
4023 }
4024 }
4025 foreach my $g (@$globs) {
4026 if (my $path = $paths->{"/$g->{path}->{left}"}) {
4027 if ($path->{action} =~ /^[AR]$/) {
4028 get_dir_check($self, $exists, $g, $r);
4029 }
4030 }
4031 foreach (keys %$paths) {
4032 if (/$g->{path}->{left_regex}/ &&
4033 !/$g->{path}->{regex}/) {
4034 next if $paths->{$_}->{action} !~ /^[AR]$/;
4035 get_dir_check($self, $exists, $g, $r);
4036 }
4037 next unless /$g->{path}->{regex}/;
4038 my $p = $1;
4039 my $pathname = $g->{path}->full_path($p);
4040 next if $exists->{$pathname};
4041 next if ($self->check_path($pathname, $r) !=
4042 $SVN::Node::dir);
4043 $exists->{$pathname} = Git::SVN->init(
4044 $self->{url}, $pathname, undef,
4045 $g->{ref}->full_path($p), 1);
4046 }
4047 my $c = '';
4048 foreach (split m#/#, $g->{path}->{left}) {
4049 $c .= "/$_";
4050 next unless ($paths->{$c} &&
4051 ($paths->{$c}->{action} =~ /^[AR]$/));
4052 get_dir_check($self, $exists, $g, $r);
4053 }
4054 }
4055 values %$exists;
4056 }
4057
4058 sub minimize_url {
4059 my ($self) = @_;
4060 return $self->{url} if ($self->{url} eq $self->{repos_root});
4061 my $url = $self->{repos_root};
4062 my @components = split(m!/!, $self->{svn_path});
4063 my $c = '';
4064 do {
4065 $url .= "/$c" if length $c;
4066 eval { (ref $self)->new($url)->get_latest_revnum };
4067 } while ($@ && ($c = shift @components));
4068 $url;
4069 }
4070
4071 sub can_do_switch {
4072 my $self = shift;
4073 unless (defined $can_do_switch) {
4074 my $pool = SVN::Pool->new;
4075 my $rep = eval {
4076 $self->do_switch(1, '', 0, $self->{url},
4077 SVN::Delta::Editor->new, $pool);
4078 };
4079 if ($@) {
4080 $can_do_switch = 0;
4081 } else {
4082 $rep->abort_report($pool);
4083 $can_do_switch = 1;
4084 }
4085 $pool->clear;
4086 }
4087 $can_do_switch;
4088 }
4089
4090 sub skip_unknown_revs {
4091 my ($err) = @_;
4092 my $errno = $err->apr_err();
4093 # Maybe the branch we're tracking didn't
4094 # exist when the repo started, so it's
4095 # not an error if it doesn't, just continue
4096 #
4097 # Wonderfully consistent library, eh?
4098 # 160013 - svn:// and file://
4099 # 175002 - http(s)://
4100 # 175007 - http(s):// (this repo required authorization, too...)
4101 # More codes may be discovered later...
4102 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4103 my $err_key = $err->expanded_message;
4104 # revision numbers change every time, filter them out
4105 $err_key =~ s/\d+/\0/g;
4106 $err_key = "$errno\0$err_key";
4107 unless ($ignored_err{$err_key}) {
4108 warn "W: Ignoring error from SVN, path probably ",
4109 "does not exist: ($errno): ",
4110 $err->expanded_message,"\n";
4111 warn "W: Do not be alarmed at the above message ",
4112 "git-svn is just searching aggressively for ",
4113 "old history.\n",
4114 "This may take a while on large repositories\n";
4115 $ignored_err{$err_key} = 1;
4116 }
4117 return;
4118 }
4119 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4120 }
4121
4122 # svn_log_changed_path_t objects passed to get_log are likely to be
4123 # overwritten even if only the refs are copied to an external variable,
4124 # so we should dup the structures in their entirety. Using an externally
4125 # passed pool (instead of our temporary and quickly cleared pool in
4126 # Git::SVN::Ra) does not help matters at all...
4127 sub dup_changed_paths {
4128 my ($paths) = @_;
4129 return undef unless $paths;
4130 my %ret;
4131 foreach my $p (keys %$paths) {
4132 my $i = $paths->{$p};
4133 my %s = map { $_ => $i->$_ }
4134 qw/copyfrom_path copyfrom_rev action/;
4135 $ret{$p} = \%s;
4136 }
4137 \%ret;
4138 }
4139
4140 package Git::SVN::Log;
4141 use strict;
4142 use warnings;
4143 use POSIX qw/strftime/;
4144 use constant commit_log_separator => ('-' x 72) . "\n";
4145 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4146 %rusers $show_commit $incremental/;
4147 my $l_fmt;
4148
4149 sub cmt_showable {
4150 my ($c) = @_;
4151 return 1 if defined $c->{r};
4152
4153 # big commit message got truncated by the 16k pretty buffer in rev-list
4154 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4155 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4156 @{$c->{l}} = ();
4157 my @log = command(qw/cat-file commit/, $c->{c});
4158
4159 # shift off the headers
4160 shift @log while ($log[0] ne '');
4161 shift @log;
4162
4163 # TODO: make $c->{l} not have a trailing newline in the future
4164 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4165
4166 (undef, $c->{r}, undef) = ::extract_metadata(
4167 (grep(/^git-svn-id: /, @log))[-1]);
4168 }
4169 return defined $c->{r};
4170 }
4171
4172 sub log_use_color {
4173 return $color || Git->repository->get_colorbool('color.diff');
4174 }
4175
4176 sub git_svn_log_cmd {
4177 my ($r_min, $r_max, @args) = @_;
4178 my $head = 'HEAD';
4179 my (@files, @log_opts);
4180 foreach my $x (@args) {
4181 if ($x eq '--' || @files) {
4182 push @files, $x;
4183 } else {
4184 if (::verify_ref("$x^0")) {
4185 $head = $x;
4186 } else {
4187 push @log_opts, $x;
4188 }
4189 }
4190 }
4191
4192 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4193 $gs ||= Git::SVN->_new;
4194 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4195 $gs->refname);
4196 push @cmd, '-r' unless $non_recursive;
4197 push @cmd, qw/--raw --name-status/ if $verbose;
4198 push @cmd, '--color' if log_use_color();
4199 push @cmd, @log_opts;
4200 if (defined $r_max && $r_max == $r_min) {
4201 push @cmd, '--max-count=1';
4202 if (my $c = $gs->rev_map_get($r_max)) {
4203 push @cmd, $c;
4204 }
4205 } elsif (defined $r_max) {
4206 if ($r_max < $r_min) {
4207 ($r_min, $r_max) = ($r_max, $r_min);
4208 }
4209 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4210 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4211 # If there are no commits in the range, both $c_max and $c_min
4212 # will be undefined. If there is at least 1 commit in the
4213 # range, both will be defined.
4214 return () if !defined $c_min || !defined $c_max;
4215 if ($c_min eq $c_max) {
4216 push @cmd, '--max-count=1', $c_min;
4217 } else {
4218 push @cmd, '--boundary', "$c_min..$c_max";
4219 }
4220 }
4221 return (@cmd, @files);
4222 }
4223
4224 # adapted from pager.c
4225 sub config_pager {
4226 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4227 if (!defined $pager) {
4228 $pager = 'less';
4229 } elsif (length $pager == 0 || $pager eq 'cat') {
4230 $pager = undef;
4231 }
4232 $ENV{GIT_PAGER_IN_USE} = defined($pager);
4233 }
4234
4235 sub run_pager {
4236 return unless -t *STDOUT && defined $pager;
4237 pipe my $rfd, my $wfd or return;
4238 defined(my $pid = fork) or ::fatal "Can't fork: $!";
4239 if (!$pid) {
4240 open STDOUT, '>&', $wfd or
4241 ::fatal "Can't redirect to stdout: $!";
4242 return;
4243 }
4244 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4245 $ENV{LESS} ||= 'FRSX';
4246 exec $pager or ::fatal "Can't run pager: $! ($pager)";
4247 }
4248
4249 sub format_svn_date {
4250 return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4251 }
4252
4253 sub parse_git_date {
4254 my ($t, $tz) = @_;
4255 # Date::Parse isn't in the standard Perl distro :(
4256 if ($tz =~ s/^\+//) {
4257 $t += tz_to_s_offset($tz);
4258 } elsif ($tz =~ s/^\-//) {
4259 $t -= tz_to_s_offset($tz);
4260 }
4261 return $t;
4262 }
4263
4264 sub set_local_timezone {
4265 if (defined $TZ) {
4266 $ENV{TZ} = $TZ;
4267 } else {
4268 delete $ENV{TZ};
4269 }
4270 }
4271
4272 sub tz_to_s_offset {
4273 my ($tz) = @_;
4274 $tz =~ s/(\d\d)$//;
4275 return ($1 * 60) + ($tz * 3600);
4276 }
4277
4278 sub get_author_info {
4279 my ($dest, $author, $t, $tz) = @_;
4280 $author =~ s/(?:^\s*|\s*$)//g;
4281 $dest->{a_raw} = $author;
4282 my $au;
4283 if ($::_authors) {
4284 $au = $rusers{$author} || undef;
4285 }
4286 if (!$au) {
4287 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4288 }
4289 $dest->{t} = $t;
4290 $dest->{tz} = $tz;
4291 $dest->{a} = $au;
4292 $dest->{t_utc} = parse_git_date($t, $tz);
4293 }
4294
4295 sub process_commit {
4296 my ($c, $r_min, $r_max, $defer) = @_;
4297 if (defined $r_min && defined $r_max) {
4298 if ($r_min == $c->{r} && $r_min == $r_max) {
4299 show_commit($c);
4300 return 0;
4301 }
4302 return 1 if $r_min == $r_max;
4303 if ($r_min < $r_max) {
4304 # we need to reverse the print order
4305 return 0 if (defined $limit && --$limit < 0);
4306 push @$defer, $c;
4307 return 1;
4308 }
4309 if ($r_min != $r_max) {
4310 return 1 if ($r_min < $c->{r});
4311 return 1 if ($r_max > $c->{r});
4312 }
4313 }
4314 return 0 if (defined $limit && --$limit < 0);
4315 show_commit($c);
4316 return 1;
4317 }
4318
4319 sub show_commit {
4320 my $c = shift;
4321 if ($oneline) {
4322 my $x = "\n";
4323 if (my $l = $c->{l}) {
4324 while ($l->[0] =~ /^\s*$/) { shift @$l }
4325 $x = $l->[0];
4326 }
4327 $l_fmt ||= 'A' . length($c->{r});
4328 print 'r',pack($l_fmt, $c->{r}),' | ';
4329 print "$c->{c} | " if $show_commit;
4330 print $x;
4331 } else {
4332 show_commit_normal($c);
4333 }
4334 }
4335
4336 sub show_commit_changed_paths {
4337 my ($c) = @_;
4338 return unless $c->{changed};
4339 print "Changed paths:\n", @{$c->{changed}};
4340 }
4341
4342 sub show_commit_normal {
4343 my ($c) = @_;
4344 print commit_log_separator, "r$c->{r} | ";
4345 print "$c->{c} | " if $show_commit;
4346 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4347 my $nr_line = 0;
4348
4349 if (my $l = $c->{l}) {
4350 while ($l->[$#$l] eq "\n" && $#$l > 0
4351 && $l->[($#$l - 1)] eq "\n") {
4352 pop @$l;
4353 }
4354 $nr_line = scalar @$l;
4355 if (!$nr_line) {
4356 print "1 line\n\n\n";
4357 } else {
4358 if ($nr_line == 1) {
4359 $nr_line = '1 line';
4360 } else {
4361 $nr_line .= ' lines';
4362 }
4363 print $nr_line, "\n";
4364 show_commit_changed_paths($c);
4365 print "\n";
4366 print $_ foreach @$l;
4367 }
4368 } else {
4369 print "1 line\n";
4370 show_commit_changed_paths($c);
4371 print "\n";
4372
4373 }
4374 foreach my $x (qw/raw stat diff/) {
4375 if ($c->{$x}) {
4376 print "\n";
4377 print $_ foreach @{$c->{$x}}
4378 }
4379 }
4380 }
4381
4382 sub cmd_show_log {
4383 my (@args) = @_;
4384 my ($r_min, $r_max);
4385 my $r_last = -1; # prevent dupes
4386 set_local_timezone();
4387 if (defined $::_revision) {
4388 if ($::_revision =~ /^(\d+):(\d+)$/) {
4389 ($r_min, $r_max) = ($1, $2);
4390 } elsif ($::_revision =~ /^\d+$/) {
4391 $r_min = $r_max = $::_revision;
4392 } else {
4393 ::fatal "-r$::_revision is not supported, use ",
4394 "standard 'git log' arguments instead";
4395 }
4396 }
4397
4398 config_pager();
4399 @args = git_svn_log_cmd($r_min, $r_max, @args);
4400 if (!@args) {
4401 print commit_log_separator unless $incremental || $oneline;
4402 return;
4403 }
4404 my $log = command_output_pipe(@args);
4405 run_pager();
4406 my (@k, $c, $d, $stat);
4407 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4408 while (<$log>) {
4409 if (/^${esc_color}commit -?($::sha1_short)/o) {
4410 my $cmt = $1;
4411 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4412 $r_last = $c->{r};
4413 process_commit($c, $r_min, $r_max, \@k) or
4414 goto out;
4415 }
4416 $d = undef;
4417 $c = { c => $cmt };
4418 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4419 get_author_info($c, $1, $2, $3);
4420 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4421 # ignore
4422 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4423 push @{$c->{raw}}, $_;
4424 } elsif (/^${esc_color}[ACRMDT]\t/) {
4425 # we could add $SVN->{svn_path} here, but that requires
4426 # remote access at the moment (repo_path_split)...
4427 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
4428 push @{$c->{changed}}, $_;
4429 } elsif (/^${esc_color}diff /o) {
4430 $d = 1;
4431 push @{$c->{diff}}, $_;
4432 } elsif ($d) {
4433 push @{$c->{diff}}, $_;
4434 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4435 $esc_color*[\+\-]*$esc_color$/x) {
4436 $stat = 1;
4437 push @{$c->{stat}}, $_;
4438 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4439 push @{$c->{stat}}, $_;
4440 $stat = undef;
4441 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
4442 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4443 } elsif (s/^${esc_color} //o) {
4444 push @{$c->{l}}, $_;
4445 }
4446 }
4447 if ($c && defined $c->{r} && $c->{r} != $r_last) {
4448 $r_last = $c->{r};
4449 process_commit($c, $r_min, $r_max, \@k);
4450 }
4451 if (@k) {
4452 ($r_min, $r_max) = ($r_max, $r_min);
4453 process_commit($_, $r_min, $r_max) foreach reverse @k;
4454 }
4455 out:
4456 close $log;
4457 print commit_log_separator unless $incremental || $oneline;
4458 }
4459
4460 sub cmd_blame {
4461 my $path = shift;
4462
4463 config_pager();
4464 run_pager();
4465
4466 my ($fh, $ctx) = command_output_pipe('blame', @_, $path);
4467 while (my $line = <$fh>) {
4468 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
4469 my (undef, $rev, undef) = ::cmt_metadata($1);
4470 $rev = sprintf('%-10s', $rev);
4471 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
4472 }
4473 print $line;
4474 }
4475 command_close_pipe($fh, $ctx);
4476 }
4477
4478 package Git::SVN::Migration;
4479 # these version numbers do NOT correspond to actual version numbers
4480 # of git nor git-svn. They are just relative.
4481 #
4482 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4483 #
4484 # v1 layout: .git/$id/info/url, refs/remotes/$id
4485 #
4486 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4487 #
4488 # v3 layout: .git/svn/$id, refs/remotes/$id
4489 # - info/url may remain for backwards compatibility
4490 # - this is what we migrate up to this layout automatically,
4491 # - this will be used by git svn init on single branches
4492 # v3.1 layout (auto migrated):
4493 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4494 # for backwards compatibility
4495 #
4496 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4497 # - this is only created for newly multi-init-ed
4498 # repositories. Similar in spirit to the
4499 # --use-separate-remotes option in git-clone (now default)
4500 # - we do not automatically migrate to this (following
4501 # the example set by core git)
4502 #
4503 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
4504 # - newer, more-efficient format that uses 24-bytes per record
4505 # with no filler space.
4506 # - use xxd -c24 < .rev_map.$UUID to view and debug
4507 # - This is a one-way migration, repositories updated to the
4508 # new format will not be able to use old git-svn without
4509 # rebuilding the .rev_db. Rebuilding the rev_db is not
4510 # possible if noMetadata or useSvmProps are set; but should
4511 # be no problem for users that use the (sensible) defaults.
4512 use strict;
4513 use warnings;
4514 use Carp qw/croak/;
4515 use File::Path qw/mkpath/;
4516 use File::Basename qw/dirname basename/;
4517 use vars qw/$_minimize/;
4518
4519 sub migrate_from_v0 {
4520 my $git_dir = $ENV{GIT_DIR};
4521 return undef unless -d $git_dir;
4522 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4523 my $migrated = 0;
4524 while (<$fh>) {
4525 chomp;
4526 my ($id, $orig_ref) = ($_, $_);
4527 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4528 next unless -f "$git_dir/$id/info/url";
4529 my $new_ref = "refs/remotes/$id";
4530 if (::verify_ref("$new_ref^0")) {
4531 print STDERR "W: $orig_ref is probably an old ",
4532 "branch used by an ancient version of ",
4533 "git-svn.\n",
4534 "However, $new_ref also exists.\n",
4535 "We will not be able ",
4536 "to use this branch until this ",
4537 "ambiguity is resolved.\n";
4538 next;
4539 }
4540 print STDERR "Migrating from v0 layout...\n" if !$migrated;
4541 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4542 command_noisy('update-ref', $new_ref, $orig_ref);
4543 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4544 $migrated++;
4545 }
4546 command_close_pipe($fh, $ctx);
4547 print STDERR "Done migrating from v0 layout...\n" if $migrated;
4548 $migrated;
4549 }
4550
4551 sub migrate_from_v1 {
4552 my $git_dir = $ENV{GIT_DIR};
4553 my $migrated = 0;
4554 return $migrated unless -d $git_dir;
4555 my $svn_dir = "$git_dir/svn";
4556
4557 # just in case somebody used 'svn' as their $id at some point...
4558 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4559
4560 print STDERR "Migrating from a git-svn v1 layout...\n";
4561 mkpath([$svn_dir]);
4562 print STDERR "Data from a previous version of git-svn exists, but\n\t",
4563 "$svn_dir\n\t(required for this version ",
4564 "($::VERSION) of git-svn) does not. exist\n";
4565 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4566 while (<$fh>) {
4567 my $x = $_;
4568 next unless $x =~ s#^refs/remotes/##;
4569 chomp $x;
4570 next unless -f "$git_dir/$x/info/url";
4571 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4572 next unless $u;
4573 my $dn = dirname("$git_dir/svn/$x");
4574 mkpath([$dn]) unless -d $dn;
4575 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4576 mkpath(["$git_dir/svn/svn"]);
4577 print STDERR " - $git_dir/$x/info => ",
4578 "$git_dir/svn/$x/info\n";
4579 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4580 croak "$!: $x";
4581 # don't worry too much about these, they probably
4582 # don't exist with repos this old (save for index,
4583 # and we can easily regenerate that)
4584 foreach my $f (qw/unhandled.log index .rev_db/) {
4585 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4586 }
4587 } else {
4588 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4589 rename "$git_dir/$x", "$git_dir/svn/$x" or
4590 croak "$!: $x";
4591 }
4592 $migrated++;
4593 }
4594 command_close_pipe($fh, $ctx);
4595 print STDERR "Done migrating from a git-svn v1 layout\n";
4596 $migrated;
4597 }
4598
4599 sub read_old_urls {
4600 my ($l_map, $pfx, $path) = @_;
4601 my @dir;
4602 foreach (<$path/*>) {
4603 if (-r "$_/info/url") {
4604 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4605 my $ref_id = $pfx . basename $_;
4606 my $url = ::file_to_s("$_/info/url");
4607 $l_map->{$ref_id} = $url;
4608 } elsif (-d $_) {
4609 push @dir, $_;
4610 }
4611 }
4612 foreach (@dir) {
4613 my $x = $_;
4614 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4615 read_old_urls($l_map, $x, $_);
4616 }
4617 }
4618
4619 sub migrate_from_v2 {
4620 my @cfg = command(qw/config -l/);
4621 return if grep /^svn-remote\..+\.url=/, @cfg;
4622 my %l_map;
4623 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4624 my $migrated = 0;
4625
4626 foreach my $ref_id (sort keys %l_map) {
4627 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4628 if ($@) {
4629 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4630 }
4631 $migrated++;
4632 }
4633 $migrated;
4634 }
4635
4636 sub minimize_connections {
4637 my $r = Git::SVN::read_all_remotes();
4638 my $new_urls = {};
4639 my $root_repos = {};
4640 foreach my $repo_id (keys %$r) {
4641 my $url = $r->{$repo_id}->{url} or next;
4642 my $fetch = $r->{$repo_id}->{fetch} or next;
4643 my $ra = Git::SVN::Ra->new($url);
4644
4645 # skip existing cases where we already connect to the root
4646 if (($ra->{url} eq $ra->{repos_root}) ||
4647 (Git::SVN::sanitize_remote_name($ra->{repos_root}) eq
4648 $repo_id)) {
4649 $root_repos->{$ra->{url}} = $repo_id;
4650 next;
4651 }
4652
4653 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4654 my $root_path = $ra->{url};
4655 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4656 foreach my $path (keys %$fetch) {
4657 my $ref_id = $fetch->{$path};
4658 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4659
4660 # make sure we can read when connecting to
4661 # a higher level of a repository
4662 my ($last_rev, undef) = $gs->last_rev_commit;
4663 if (!defined $last_rev) {
4664 $last_rev = eval {
4665 $root_ra->get_latest_revnum;
4666 };
4667 next if $@;
4668 }
4669 my $new = $root_path;
4670 $new .= length $path ? "/$path" : '';
4671 eval {
4672 $root_ra->get_log([$new], $last_rev, $last_rev,
4673 0, 0, 1, sub { });
4674 };
4675 next if $@;
4676 $new_urls->{$ra->{repos_root}}->{$new} =
4677 { ref_id => $ref_id,
4678 old_repo_id => $repo_id,
4679 old_path => $path };
4680 }
4681 }
4682
4683 my @emptied;
4684 foreach my $url (keys %$new_urls) {
4685 # see if we can re-use an existing [svn-remote "repo_id"]
4686 # instead of creating a(n ugly) new section:
4687 my $repo_id = $root_repos->{$url} ||
4688 Git::SVN::sanitize_remote_name($url);
4689
4690 my $fetch = $new_urls->{$url};
4691 foreach my $path (keys %$fetch) {
4692 my $x = $fetch->{$path};
4693 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4694 my $pfx = "svn-remote.$x->{old_repo_id}";
4695
4696 my $old_fetch = quotemeta("$x->{old_path}:".
4697 "refs/remotes/$x->{ref_id}");
4698 command_noisy(qw/config --unset/,
4699 "$pfx.fetch", '^'. $old_fetch . '$');
4700 delete $r->{$x->{old_repo_id}}->
4701 {fetch}->{$x->{old_path}};
4702 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4703 command_noisy(qw/config --unset/,
4704 "$pfx.url");
4705 push @emptied, $x->{old_repo_id}
4706 }
4707 }
4708 }
4709 if (@emptied) {
4710 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4711 "$ENV{GIT_DIR}/config";
4712 print STDERR <<EOF;
4713 The following [svn-remote] sections in your config file ($file) are empty
4714 and can be safely removed:
4715 EOF
4716 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4717 }
4718 }
4719
4720 sub migration_check {
4721 migrate_from_v0();
4722 migrate_from_v1();
4723 migrate_from_v2();
4724 minimize_connections() if $_minimize;
4725 }
4726
4727 package Git::IndexInfo;
4728 use strict;
4729 use warnings;
4730 use Git qw/command_input_pipe command_close_pipe/;
4731
4732 sub new {
4733 my ($class) = @_;
4734 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4735 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4736 }
4737
4738 sub remove {
4739 my ($self, $path) = @_;
4740 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4741 return ++$self->{nr};
4742 }
4743 undef;
4744 }
4745
4746 sub update {
4747 my ($self, $mode, $hash, $path) = @_;
4748 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4749 return ++$self->{nr};
4750 }
4751 undef;
4752 }
4753
4754 sub DESTROY {
4755 my ($self) = @_;
4756 command_close_pipe($self->{gui}, $self->{ctx});
4757 }
4758
4759 package Git::SVN::GlobSpec;
4760 use strict;
4761 use warnings;
4762
4763 sub new {
4764 my ($class, $glob) = @_;
4765 my $re = $glob;
4766 $re =~ s!/+$!!g; # no need for trailing slashes
4767 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
4768 my ($left, $right) = ($1, $2);
4769 if ($nr > 1) {
4770 die "Only one '*' wildcard expansion ",
4771 "is supported (got $nr): '$glob'\n";
4772 } elsif ($nr == 0) {
4773 die "One '*' is needed for glob: '$glob'\n";
4774 }
4775 $re = quotemeta($left) . $re . quotemeta($right);
4776 if (length $left && !($left =~ s!/+$!!g)) {
4777 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4778 }
4779 if (length $right && !($right =~ s!^/+!!g)) {
4780 die "Missing leading '/' on right side of: '$glob' ($right)\n";
4781 }
4782 my $left_re = qr/^\/\Q$left\E(\/|$)/;
4783 bless { left => $left, right => $right, left_regex => $left_re,
4784 regex => qr/$re/, glob => $glob }, $class;
4785 }
4786
4787 sub full_path {
4788 my ($self, $path) = @_;
4789 return (length $self->{left} ? "$self->{left}/" : '') .
4790 $path . (length $self->{right} ? "/$self->{right}" : '');
4791 }
4792
4793 __END__
4794
4795 Data structures:
4796
4797
4798 $remotes = { # returned by read_all_remotes()
4799 'svn' => {
4800 # svn-remote.svn.url=https://svn.musicpd.org
4801 url => 'https://svn.musicpd.org',
4802 # svn-remote.svn.fetch=mpd/trunk:trunk
4803 fetch => {
4804 'mpd/trunk' => 'trunk',
4805 },
4806 # svn-remote.svn.tags=mpd/tags/*:tags/*
4807 tags => {
4808 path => {
4809 left => 'mpd/tags',
4810 right => '',
4811 regex => qr!mpd/tags/([^/]+)$!,
4812 glob => 'tags/*',
4813 },
4814 ref => {
4815 left => 'tags',
4816 right => '',
4817 regex => qr!tags/([^/]+)$!,
4818 glob => 'tags/*',
4819 },
4820 }
4821 }
4822 };
4823
4824 $log_entry hashref as returned by libsvn_log_entry()
4825 {
4826 log => 'whitespace-formatted log entry
4827 ', # trailing newline is preserved
4828 revision => '8', # integer
4829 date => '2004-02-24T17:01:44.108345Z', # commit date
4830 author => 'committer name'
4831 };
4832
4833
4834 # this is generated by generate_diff();
4835 @mods = array of diff-index line hashes, each element represents one line
4836 of diff-index output
4837
4838 diff-index line ($m hash)
4839 {
4840 mode_a => first column of diff-index output, no leading ':',
4841 mode_b => second column of diff-index output,
4842 sha1_b => sha1sum of the final blob,
4843 chg => change type [MCRADT],
4844 file_a => original file name of a file (iff chg is 'C' or 'R')
4845 file_b => new/current file name of a file (any chg)
4846 }
4847 ;
4848
4849 # retval of read_url_paths{,_all}();
4850 $l_map = {
4851 # repository root url
4852 'https://svn.musicpd.org' => {
4853 # repository path # GIT_SVN_ID
4854 'mpd/trunk' => 'trunk',
4855 'mpd/tags/0.11.5' => 'tags/0.11.5',
4856 },
4857 }
4858
4859 Notes:
4860 I don't trust the each() function on unless I created %hash myself
4861 because the internal iterator may not have started at base.