Merge branch 'maint'
[git/git.git] / git-svn.perl
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision $_repository
8 $_q $_authors $_authors_prog %users/;
9 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10 $VERSION = '@@GIT_VERSION@@';
11
12 # From which subdir have we been invoked?
13 my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15 } || '';
16
17 my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
18 $ENV{GIT_DIR} ||= '.git';
19 $Git::SVN::default_repo_id = 'svn';
20 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
21 $Git::SVN::Ra::_log_window_size = 100;
22 $Git::SVN::_minimize_url = 'unset';
23
24 if (! exists $ENV{SVN_SSH}) {
25 if (exists $ENV{GIT_SSH}) {
26 $ENV{SVN_SSH} = $ENV{GIT_SSH};
27 if ($^O eq 'msys') {
28 $ENV{SVN_SSH} =~ s/\\/\\\\/g;
29 }
30 }
31 }
32
33 $Git::SVN::Log::TZ = $ENV{TZ};
34 $ENV{TZ} = 'UTC';
35 $| = 1; # unbuffer STDOUT
36
37 sub fatal (@) { print STDERR "@_\n"; exit 1 }
38 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
39 require SVN::Ra;
40 require SVN::Delta;
41 if ($SVN::Core::VERSION lt '1.1.0') {
42 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
43 }
44 my $can_compress = eval { require Compress::Zlib; 1};
45 push @Git::SVN::Ra::ISA, 'SVN::Ra';
46 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
47 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
48 use Carp qw/croak/;
49 use Digest::MD5;
50 use IO::File qw//;
51 use File::Basename qw/dirname basename/;
52 use File::Path qw/mkpath/;
53 use File::Spec;
54 use File::Find;
55 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
56 use IPC::Open3;
57 use Git;
58
59 BEGIN {
60 # import functions from Git into our packages, en masse
61 no strict 'refs';
62 foreach (qw/command command_oneline command_noisy command_output_pipe
63 command_input_pipe command_close_pipe
64 command_bidi_pipe command_close_bidi_pipe/) {
65 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
66 Git::SVN::Migration Git::SVN::Log Git::SVN),
67 __PACKAGE__) {
68 *{"${package}::$_"} = \&{"Git::$_"};
69 }
70 }
71 }
72
73 my ($SVN);
74
75 $sha1 = qr/[a-f\d]{40}/;
76 $sha1_short = qr/[a-f\d]{4,40}/;
77 my ($_stdin, $_help, $_edit,
78 $_message, $_file, $_branch_dest,
79 $_template, $_shared,
80 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
81 $_merge, $_strategy, $_dry_run, $_local,
82 $_prefix, $_no_checkout, $_url, $_verbose,
83 $_git_format, $_commit_url, $_tag);
84 $Git::SVN::_follow_parent = 1;
85 $_q ||= 0;
86 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
87 'config-dir=s' => \$Git::SVN::Ra::config_dir,
88 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
89 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
90 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
91 'authors-file|A=s' => \$_authors,
92 'authors-prog=s' => \$_authors_prog,
93 'repack:i' => \$Git::SVN::_repack,
94 'noMetadata' => \$Git::SVN::_no_metadata,
95 'useSvmProps' => \$Git::SVN::_use_svm_props,
96 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
97 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
98 'no-checkout' => \$_no_checkout,
99 'quiet|q+' => \$_q,
100 'repack-flags|repack-args|repack-opts=s' =>
101 \$Git::SVN::_repack_flags,
102 'use-log-author' => \$Git::SVN::_use_log_author,
103 'add-author-from' => \$Git::SVN::_add_author_from,
104 'localtime' => \$Git::SVN::_localtime,
105 %remote_opts );
106
107 my ($_trunk, @_tags, @_branches, $_stdlayout);
108 my %icv;
109 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
110 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
111 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
112 'stdlayout|s' => \$_stdlayout,
113 'minimize-url|m!' => \$Git::SVN::_minimize_url,
114 'no-metadata' => sub { $icv{noMetadata} = 1 },
115 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
116 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
117 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
118 %remote_opts );
119 my %cmt_opts = ( 'edit|e' => \$_edit,
120 'rmdir' => \$SVN::Git::Editor::_rmdir,
121 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
122 'l=i' => \$SVN::Git::Editor::_rename_limit,
123 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
124 );
125
126 my %cmd = (
127 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
128 { 'revision|r=s' => \$_revision,
129 'fetch-all|all' => \$_fetch_all,
130 'parent|p' => \$_fetch_parent,
131 %fc_opts } ],
132 clone => [ \&cmd_clone, "Initialize and fetch revisions",
133 { 'revision|r=s' => \$_revision,
134 %fc_opts, %init_opts } ],
135 init => [ \&cmd_init, "Initialize a repo for tracking" .
136 " (requires URL argument)",
137 \%init_opts ],
138 'multi-init' => [ \&cmd_multi_init,
139 "Deprecated alias for ".
140 "'$0 init -T<trunk> -b<branches> -t<tags>'",
141 \%init_opts ],
142 dcommit => [ \&cmd_dcommit,
143 'Commit several diffs to merge with upstream',
144 { 'merge|m|M' => \$_merge,
145 'strategy|s=s' => \$_strategy,
146 'verbose|v' => \$_verbose,
147 'dry-run|n' => \$_dry_run,
148 'fetch-all|all' => \$_fetch_all,
149 'commit-url=s' => \$_commit_url,
150 'revision|r=i' => \$_revision,
151 'no-rebase' => \$_no_rebase,
152 %cmt_opts, %fc_opts } ],
153 branch => [ \&cmd_branch,
154 'Create a branch in the SVN repository',
155 { 'message|m=s' => \$_message,
156 'destination|d=s' => \$_branch_dest,
157 'dry-run|n' => \$_dry_run,
158 'tag|t' => \$_tag } ],
159 tag => [ sub { $_tag = 1; cmd_branch(@_) },
160 'Create a tag in the SVN repository',
161 { 'message|m=s' => \$_message,
162 'destination|d=s' => \$_branch_dest,
163 'dry-run|n' => \$_dry_run } ],
164 'set-tree' => [ \&cmd_set_tree,
165 "Set an SVN repository to a git tree-ish",
166 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
167 'create-ignore' => [ \&cmd_create_ignore,
168 'Create a .gitignore per svn:ignore',
169 { 'revision|r=i' => \$_revision
170 } ],
171 'propget' => [ \&cmd_propget,
172 'Print the value of a property on a file or directory',
173 { 'revision|r=i' => \$_revision } ],
174 'proplist' => [ \&cmd_proplist,
175 'List all properties of a file or directory',
176 { 'revision|r=i' => \$_revision } ],
177 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
178 { 'revision|r=i' => \$_revision
179 } ],
180 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
181 { 'revision|r=i' => \$_revision
182 } ],
183 'multi-fetch' => [ \&cmd_multi_fetch,
184 "Deprecated alias for $0 fetch --all",
185 { 'revision|r=s' => \$_revision, %fc_opts } ],
186 'migrate' => [ sub { },
187 # no-op, we automatically run this anyways,
188 'Migrate configuration/metadata/layout from
189 previous versions of git-svn',
190 { 'minimize' => \$Git::SVN::Migration::_minimize,
191 %remote_opts } ],
192 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
193 { 'limit=i' => \$Git::SVN::Log::limit,
194 'revision|r=s' => \$_revision,
195 'verbose|v' => \$Git::SVN::Log::verbose,
196 'incremental' => \$Git::SVN::Log::incremental,
197 'oneline' => \$Git::SVN::Log::oneline,
198 'show-commit' => \$Git::SVN::Log::show_commit,
199 'non-recursive' => \$Git::SVN::Log::non_recursive,
200 'authors-file|A=s' => \$_authors,
201 'color' => \$Git::SVN::Log::color,
202 'pager=s' => \$Git::SVN::Log::pager
203 } ],
204 'find-rev' => [ \&cmd_find_rev,
205 "Translate between SVN revision numbers and tree-ish",
206 {} ],
207 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
208 { 'merge|m|M' => \$_merge,
209 'verbose|v' => \$_verbose,
210 'strategy|s=s' => \$_strategy,
211 'local|l' => \$_local,
212 'fetch-all|all' => \$_fetch_all,
213 'dry-run|n' => \$_dry_run,
214 %fc_opts } ],
215 'commit-diff' => [ \&cmd_commit_diff,
216 'Commit a diff between two trees',
217 { 'message|m=s' => \$_message,
218 'file|F=s' => \$_file,
219 'revision|r=s' => \$_revision,
220 %cmt_opts } ],
221 'info' => [ \&cmd_info,
222 "Show info about the latest SVN revision
223 on the current branch",
224 { 'url' => \$_url, } ],
225 'blame' => [ \&Git::SVN::Log::cmd_blame,
226 "Show what revision and author last modified each line of a file",
227 { 'git-format' => \$_git_format } ],
228 'reset' => [ \&cmd_reset,
229 "Undo fetches back to the specified SVN revision",
230 { 'revision|r=s' => \$_revision,
231 'parent|p' => \$_fetch_parent } ],
232 'gc' => [ \&cmd_gc,
233 "Compress unhandled.log files in .git/svn and remove " .
234 "index files in .git/svn",
235 {} ],
236 );
237
238 my $cmd;
239 for (my $i = 0; $i < @ARGV; $i++) {
240 if (defined $cmd{$ARGV[$i]}) {
241 $cmd = $ARGV[$i];
242 splice @ARGV, $i, 1;
243 last;
244 } elsif ($ARGV[$i] eq 'help') {
245 $cmd = $ARGV[$i+1];
246 usage(0);
247 }
248 };
249
250 # make sure we're always running at the top-level working directory
251 unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
252 unless (-d $ENV{GIT_DIR}) {
253 if ($git_dir_user_set) {
254 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
255 "but it is not a directory\n";
256 }
257 my $git_dir = delete $ENV{GIT_DIR};
258 my $cdup = undef;
259 git_cmd_try {
260 $cdup = command_oneline(qw/rev-parse --show-cdup/);
261 $git_dir = '.' unless ($cdup);
262 chomp $cdup if ($cdup);
263 $cdup = "." unless ($cdup && length $cdup);
264 } "Already at toplevel, but $git_dir not found\n";
265 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
266 unless (-d $git_dir) {
267 die "$git_dir still not found after going to ",
268 "'$cdup'\n";
269 }
270 $ENV{GIT_DIR} = $git_dir;
271 }
272 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
273 }
274
275 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
276
277 read_repo_config(\%opts);
278 if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
279 Getopt::Long::Configure('pass_through');
280 }
281 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
282 'minimize-connections' => \$Git::SVN::Migration::_minimize,
283 'id|i=s' => \$Git::SVN::default_ref_id,
284 'svn-remote|remote|R=s' => sub {
285 $Git::SVN::no_reuse_existing = 1;
286 $Git::SVN::default_repo_id = $_[1] });
287 exit 1 if (!$rv && $cmd && $cmd ne 'log');
288
289 usage(0) if $_help;
290 version() if $_version;
291 usage(1) unless defined $cmd;
292 load_authors() if $_authors;
293 if (defined $_authors_prog) {
294 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
295 }
296
297 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
298 Git::SVN::Migration::migration_check();
299 }
300 Git::SVN::init_vars();
301 eval {
302 Git::SVN::verify_remotes_sanity();
303 $cmd{$cmd}->[0]->(@ARGV);
304 };
305 fatal $@ if $@;
306 post_fetch_checkout();
307 exit 0;
308
309 ####################### primary functions ######################
310 sub usage {
311 my $exit = shift || 0;
312 my $fd = $exit ? \*STDERR : \*STDOUT;
313 print $fd <<"";
314 git-svn - bidirectional operations between a single Subversion tree and git
315 Usage: git svn <command> [options] [arguments]\n
316
317 print $fd "Available commands:\n" unless $cmd;
318
319 foreach (sort keys %cmd) {
320 next if $cmd && $cmd ne $_;
321 next if /^multi-/; # don't show deprecated commands
322 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
323 foreach (sort keys %{$cmd{$_}->[2]}) {
324 # mixed-case options are for .git/config only
325 next if /[A-Z]/ && /^[a-z]+$/i;
326 # prints out arguments as they should be passed:
327 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
328 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
329 "--$_" : "-$_" }
330 split /\|/,$_)," $x\n";
331 }
332 }
333 print $fd <<"";
334 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
335 arbitrary identifier if you're tracking multiple SVN branches/repositories in
336 one git repository and want to keep them separate. See git-svn(1) for more
337 information.
338
339 exit $exit;
340 }
341
342 sub version {
343 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
344 exit 0;
345 }
346
347 sub do_git_init_db {
348 unless (-d $ENV{GIT_DIR}) {
349 my @init_db = ('init');
350 push @init_db, "--template=$_template" if defined $_template;
351 if (defined $_shared) {
352 if ($_shared =~ /[a-z]/) {
353 push @init_db, "--shared=$_shared";
354 } else {
355 push @init_db, "--shared";
356 }
357 }
358 command_noisy(@init_db);
359 $_repository = Git->repository(Repository => ".git");
360 }
361 command_noisy('config', 'core.autocrlf', 'false');
362 my $set;
363 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
364 foreach my $i (keys %icv) {
365 die "'$set' and '$i' cannot both be set\n" if $set;
366 next unless defined $icv{$i};
367 command_noisy('config', "$pfx.$i", $icv{$i});
368 $set = $i;
369 }
370 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
371 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
372 if defined $$ignore_regex;
373 }
374
375 sub init_subdir {
376 my $repo_path = shift or return;
377 mkpath([$repo_path]) unless -d $repo_path;
378 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
379 $ENV{GIT_DIR} = '.git';
380 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
381 }
382
383 sub cmd_clone {
384 my ($url, $path) = @_;
385 if (!defined $path &&
386 (defined $_trunk || @_branches || @_tags ||
387 defined $_stdlayout) &&
388 $url !~ m#^[a-z\+]+://#) {
389 $path = $url;
390 }
391 $path = basename($url) if !defined $path || !length $path;
392 cmd_init($url, $path);
393 Git::SVN::fetch_all($Git::SVN::default_repo_id);
394 command_oneline('config', 'svn.authorsfile', $_authors) if $_authors;
395 }
396
397 sub cmd_init {
398 if (defined $_stdlayout) {
399 $_trunk = 'trunk' if (!defined $_trunk);
400 @_tags = 'tags' if (! @_tags);
401 @_branches = 'branches' if (! @_branches);
402 }
403 if (defined $_trunk || @_branches || @_tags) {
404 return cmd_multi_init(@_);
405 }
406 my $url = shift or die "SVN repository location required ",
407 "as a command-line argument\n";
408 $url = canonicalize_url($url);
409 init_subdir(@_);
410 do_git_init_db();
411
412 if ($Git::SVN::_minimize_url eq 'unset') {
413 $Git::SVN::_minimize_url = 0;
414 }
415
416 Git::SVN->init($url);
417 }
418
419 sub cmd_fetch {
420 if (grep /^\d+=./, @_) {
421 die "'<rev>=<commit>' fetch arguments are ",
422 "no longer supported.\n";
423 }
424 my ($remote) = @_;
425 if (@_ > 1) {
426 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
427 }
428 if ($_fetch_parent) {
429 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
430 unless ($gs) {
431 die "Unable to determine upstream SVN information from ",
432 "working tree history\n";
433 }
434 # just fetch, don't checkout.
435 $_no_checkout = 'true';
436 $_fetch_all ? $gs->fetch_all : $gs->fetch;
437 } elsif ($_fetch_all) {
438 cmd_multi_fetch();
439 } else {
440 $remote ||= $Git::SVN::default_repo_id;
441 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
442 }
443 }
444
445 sub cmd_set_tree {
446 my (@commits) = @_;
447 if ($_stdin || !@commits) {
448 print "Reading from stdin...\n";
449 @commits = ();
450 while (<STDIN>) {
451 if (/\b($sha1_short)\b/o) {
452 unshift @commits, $1;
453 }
454 }
455 }
456 my @revs;
457 foreach my $c (@commits) {
458 my @tmp = command('rev-parse',$c);
459 if (scalar @tmp == 1) {
460 push @revs, $tmp[0];
461 } elsif (scalar @tmp > 1) {
462 push @revs, reverse(command('rev-list',@tmp));
463 } else {
464 fatal "Failed to rev-parse $c";
465 }
466 }
467 my $gs = Git::SVN->new;
468 my ($r_last, $cmt_last) = $gs->last_rev_commit;
469 $gs->fetch;
470 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
471 fatal "There are new revisions that were fetched ",
472 "and need to be merged (or acknowledged) ",
473 "before committing.\nlast rev: $r_last\n",
474 " current: $gs->{last_rev}";
475 }
476 $gs->set_tree($_) foreach @revs;
477 print "Done committing ",scalar @revs," revisions to SVN\n";
478 unlink $gs->{index};
479 }
480
481 sub cmd_dcommit {
482 my $head = shift;
483 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
484 'Cannot dcommit with a dirty index. Commit your changes first, '
485 . "or stash them with `git stash'.\n";
486 $head ||= 'HEAD';
487
488 my $old_head;
489 if ($head ne 'HEAD') {
490 $old_head = eval {
491 command_oneline([qw/symbolic-ref -q HEAD/])
492 };
493 if ($old_head) {
494 $old_head =~ s{^refs/heads/}{};
495 } else {
496 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
497 }
498 command(['checkout', $head], STDERR => 0);
499 }
500
501 my @refs;
502 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
503 unless ($gs) {
504 die "Unable to determine upstream SVN information from ",
505 "$head history.\nPerhaps the repository is empty.";
506 }
507
508 if (defined $_commit_url) {
509 $url = $_commit_url;
510 } else {
511 $url = eval { command_oneline('config', '--get',
512 "svn-remote.$gs->{repo_id}.commiturl") };
513 if (!$url) {
514 $url = $gs->full_url
515 }
516 }
517
518 my $last_rev = $_revision if defined $_revision;
519 if ($url) {
520 print "Committing to $url ...\n";
521 }
522 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
523 if ($_no_rebase && scalar(@$linear_refs) > 1) {
524 warn "Attempting to commit more than one change while ",
525 "--no-rebase is enabled.\n",
526 "If these changes depend on each other, re-running ",
527 "without --no-rebase may be required."
528 }
529 my $expect_url = $url;
530 Git::SVN::remove_username($expect_url);
531 while (1) {
532 my $d = shift @$linear_refs or last;
533 unless (defined $last_rev) {
534 (undef, $last_rev, undef) = cmt_metadata("$d~1");
535 unless (defined $last_rev) {
536 fatal "Unable to extract revision information ",
537 "from commit $d~1";
538 }
539 }
540 if ($_dry_run) {
541 print "diff-tree $d~1 $d\n";
542 } else {
543 my $cmt_rev;
544 my %ed_opts = ( r => $last_rev,
545 log => get_commit_entry($d)->{log},
546 ra => Git::SVN::Ra->new($url),
547 config => SVN::Core::config_get_config(
548 $Git::SVN::Ra::config_dir
549 ),
550 tree_a => "$d~1",
551 tree_b => $d,
552 editor_cb => sub {
553 print "Committed r$_[0]\n";
554 $cmt_rev = $_[0];
555 },
556 svn_path => '');
557 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
558 print "No changes\n$d~1 == $d\n";
559 } elsif ($parents->{$d} && @{$parents->{$d}}) {
560 $gs->{inject_parents_dcommit}->{$cmt_rev} =
561 $parents->{$d};
562 }
563 $_fetch_all ? $gs->fetch_all : $gs->fetch;
564 $last_rev = $cmt_rev;
565 next if $_no_rebase;
566
567 # we always want to rebase against the current HEAD,
568 # not any head that was passed to us
569 my @diff = command('diff-tree', $d,
570 $gs->refname, '--');
571 my @finish;
572 if (@diff) {
573 @finish = rebase_cmd();
574 print STDERR "W: $d and ", $gs->refname,
575 " differ, using @finish:\n",
576 join("\n", @diff), "\n";
577 } else {
578 print "No changes between current HEAD and ",
579 $gs->refname,
580 "\nResetting to the latest ",
581 $gs->refname, "\n";
582 @finish = qw/reset --mixed/;
583 }
584 command_noisy(@finish, $gs->refname);
585 if (@diff) {
586 @refs = ();
587 my ($url_, $rev_, $uuid_, $gs_) =
588 working_head_info('HEAD', \@refs);
589 my ($linear_refs_, $parents_) =
590 linearize_history($gs_, \@refs);
591 if (scalar(@$linear_refs) !=
592 scalar(@$linear_refs_)) {
593 fatal "# of revisions changed ",
594 "\nbefore:\n",
595 join("\n", @$linear_refs),
596 "\n\nafter:\n",
597 join("\n", @$linear_refs_), "\n",
598 'If you are attempting to commit ',
599 "merges, try running:\n\t",
600 'git rebase --interactive',
601 '--preserve-merges ',
602 $gs->refname,
603 "\nBefore dcommitting";
604 }
605 if ($url_ ne $expect_url) {
606 fatal "URL mismatch after rebase: ",
607 "$url_ != $expect_url";
608 }
609 if ($uuid_ ne $uuid) {
610 fatal "uuid mismatch after rebase: ",
611 "$uuid_ != $uuid";
612 }
613 # remap parents
614 my (%p, @l, $i);
615 for ($i = 0; $i < scalar @$linear_refs; $i++) {
616 my $new = $linear_refs_->[$i] or next;
617 $p{$new} =
618 $parents->{$linear_refs->[$i]};
619 push @l, $new;
620 }
621 $parents = \%p;
622 $linear_refs = \@l;
623 }
624 }
625 }
626
627 if ($old_head) {
628 my $new_head = command_oneline(qw/rev-parse HEAD/);
629 my $new_is_symbolic = eval {
630 command_oneline(qw/symbolic-ref -q HEAD/);
631 };
632 if ($new_is_symbolic) {
633 print "dcommitted the branch ", $head, "\n";
634 } else {
635 print "dcommitted on a detached HEAD because you gave ",
636 "a revision argument.\n",
637 "The rewritten commit is: ", $new_head, "\n";
638 }
639 command(['checkout', $old_head], STDERR => 0);
640 }
641
642 unlink $gs->{index};
643 }
644
645 sub cmd_branch {
646 my ($branch_name, $head) = @_;
647
648 unless (defined $branch_name && length $branch_name) {
649 die(($_tag ? "tag" : "branch") . " name required\n");
650 }
651 $head ||= 'HEAD';
652
653 my ($src, $rev, undef, $gs) = working_head_info($head);
654
655 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
656 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
657 my $glob;
658 if ($#{$allglobs} == 0) {
659 $glob = $allglobs->[0];
660 } else {
661 unless(defined $_branch_dest) {
662 die "Multiple ",
663 $_tag ? "tag" : "branch",
664 " paths defined for Subversion repository.\n",
665 "You must specify where you want to create the ",
666 $_tag ? "tag" : "branch",
667 " with the --destination argument.\n";
668 }
669 foreach my $g (@{$allglobs}) {
670 # SVN::Git::Editor could probably be moved to Git.pm..
671 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
672 if ($_branch_dest =~ /$re/) {
673 $glob = $g;
674 last;
675 }
676 }
677 unless (defined $glob) {
678 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
679 foreach my $g (@{$allglobs}) {
680 $g->{path}->{left} =~ /$dest_re/ or next;
681 if (defined $glob) {
682 die "Ambiguous destination: ",
683 $_branch_dest, "\nmatches both '",
684 $glob->{path}->{left}, "' and '",
685 $g->{path}->{left}, "'\n";
686 }
687 $glob = $g;
688 }
689 unless (defined $glob) {
690 die "Unknown ",
691 $_tag ? "tag" : "branch",
692 " destination $_branch_dest\n";
693 }
694 }
695 }
696 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
697 my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
698
699 my $ctx = SVN::Client->new(
700 auth => Git::SVN::Ra::_auth_providers(),
701 log_msg => sub {
702 ${ $_[0] } = defined $_message
703 ? $_message
704 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
705 . $branch_name;
706 },
707 );
708
709 eval {
710 $ctx->ls($dst, 'HEAD', 0);
711 } and die "branch ${branch_name} already exists\n";
712
713 print "Copying ${src} at r${rev} to ${dst}...\n";
714 $ctx->copy($src, $rev, $dst)
715 unless $_dry_run;
716
717 $gs->fetch_all;
718 }
719
720 sub cmd_find_rev {
721 my $revision_or_hash = shift or die "SVN or git revision required ",
722 "as a command-line argument\n";
723 my $result;
724 if ($revision_or_hash =~ /^r\d+$/) {
725 my $head = shift;
726 $head ||= 'HEAD';
727 my @refs;
728 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
729 unless ($gs) {
730 die "Unable to determine upstream SVN information from ",
731 "$head history\n";
732 }
733 my $desired_revision = substr($revision_or_hash, 1);
734 $result = $gs->rev_map_get($desired_revision, $uuid);
735 } else {
736 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
737 $result = $rev;
738 }
739 print "$result\n" if $result;
740 }
741
742 sub cmd_rebase {
743 command_noisy(qw/update-index --refresh/);
744 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
745 unless ($gs) {
746 die "Unable to determine upstream SVN information from ",
747 "working tree history\n";
748 }
749 if ($_dry_run) {
750 print "Remote Branch: " . $gs->refname . "\n";
751 print "SVN URL: " . $url . "\n";
752 return;
753 }
754 if (command(qw/diff-index HEAD --/)) {
755 print STDERR "Cannot rebase with uncommited changes:\n";
756 command_noisy('status');
757 exit 1;
758 }
759 unless ($_local) {
760 # rebase will checkout for us, so no need to do it explicitly
761 $_no_checkout = 'true';
762 $_fetch_all ? $gs->fetch_all : $gs->fetch;
763 }
764 command_noisy(rebase_cmd(), $gs->refname);
765 }
766
767 sub cmd_show_ignore {
768 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
769 $gs ||= Git::SVN->new;
770 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
771 $gs->prop_walk($gs->{path}, $r, sub {
772 my ($gs, $path, $props) = @_;
773 print STDOUT "\n# $path\n";
774 my $s = $props->{'svn:ignore'} or return;
775 $s =~ s/[\r\n]+/\n/g;
776 $s =~ s/^\n+//;
777 chomp $s;
778 $s =~ s#^#$path#gm;
779 print STDOUT "$s\n";
780 });
781 }
782
783 sub cmd_show_externals {
784 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
785 $gs ||= Git::SVN->new;
786 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
787 $gs->prop_walk($gs->{path}, $r, sub {
788 my ($gs, $path, $props) = @_;
789 print STDOUT "\n# $path\n";
790 my $s = $props->{'svn:externals'} or return;
791 $s =~ s/[\r\n]+/\n/g;
792 chomp $s;
793 $s =~ s#^#$path#gm;
794 print STDOUT "$s\n";
795 });
796 }
797
798 sub cmd_create_ignore {
799 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
800 $gs ||= Git::SVN->new;
801 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
802 $gs->prop_walk($gs->{path}, $r, sub {
803 my ($gs, $path, $props) = @_;
804 # $path is of the form /path/to/dir/
805 $path = '.' . $path;
806 # SVN can have attributes on empty directories,
807 # which git won't track
808 mkpath([$path]) unless -d $path;
809 my $ignore = $path . '.gitignore';
810 my $s = $props->{'svn:ignore'} or return;
811 open(GITIGNORE, '>', $ignore)
812 or fatal("Failed to open `$ignore' for writing: $!");
813 $s =~ s/[\r\n]+/\n/g;
814 $s =~ s/^\n+//;
815 chomp $s;
816 # Prefix all patterns so that the ignore doesn't apply
817 # to sub-directories.
818 $s =~ s#^#/#gm;
819 print GITIGNORE "$s\n";
820 close(GITIGNORE)
821 or fatal("Failed to close `$ignore': $!");
822 command_noisy('add', '-f', $ignore);
823 });
824 }
825
826 sub canonicalize_path {
827 my ($path) = @_;
828 my $dot_slash_added = 0;
829 if (substr($path, 0, 1) ne "/") {
830 $path = "./" . $path;
831 $dot_slash_added = 1;
832 }
833 # File::Spec->canonpath doesn't collapse x/../y into y (for a
834 # good reason), so let's do this manually.
835 $path =~ s#/+#/#g;
836 $path =~ s#/\.(?:/|$)#/#g;
837 $path =~ s#/[^/]+/\.\.##g;
838 $path =~ s#/$##g;
839 $path =~ s#^\./## if $dot_slash_added;
840 $path =~ s#^/##;
841 $path =~ s#^\.$##;
842 return $path;
843 }
844
845 sub canonicalize_url {
846 my ($url) = @_;
847 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
848 return $url;
849 }
850
851 # get_svnprops(PATH)
852 # ------------------
853 # Helper for cmd_propget and cmd_proplist below.
854 sub get_svnprops {
855 my $path = shift;
856 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
857 $gs ||= Git::SVN->new;
858
859 # prefix THE PATH by the sub-directory from which the user
860 # invoked us.
861 $path = $cmd_dir_prefix . $path;
862 fatal("No such file or directory: $path") unless -e $path;
863 my $is_dir = -d $path ? 1 : 0;
864 $path = $gs->{path} . '/' . $path;
865
866 # canonicalize the path (otherwise libsvn will abort or fail to
867 # find the file)
868 $path = canonicalize_path($path);
869
870 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
871 my $props;
872 if ($is_dir) {
873 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
874 }
875 else {
876 (undef, $props) = $gs->ra->get_file($path, $r, undef);
877 }
878 return $props;
879 }
880
881 # cmd_propget (PROP, PATH)
882 # ------------------------
883 # Print the SVN property PROP for PATH.
884 sub cmd_propget {
885 my ($prop, $path) = @_;
886 $path = '.' if not defined $path;
887 usage(1) if not defined $prop;
888 my $props = get_svnprops($path);
889 if (not defined $props->{$prop}) {
890 fatal("`$path' does not have a `$prop' SVN property.");
891 }
892 print $props->{$prop} . "\n";
893 }
894
895 # cmd_proplist (PATH)
896 # -------------------
897 # Print the list of SVN properties for PATH.
898 sub cmd_proplist {
899 my $path = shift;
900 $path = '.' if not defined $path;
901 my $props = get_svnprops($path);
902 print "Properties on '$path':\n";
903 foreach (sort keys %{$props}) {
904 print " $_\n";
905 }
906 }
907
908 sub cmd_multi_init {
909 my $url = shift;
910 unless (defined $_trunk || @_branches || @_tags) {
911 usage(1);
912 }
913
914 $_prefix = '' unless defined $_prefix;
915 if (defined $url) {
916 $url = canonicalize_url($url);
917 init_subdir(@_);
918 }
919 do_git_init_db();
920 if (defined $_trunk) {
921 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
922 # try both old-style and new-style lookups:
923 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
924 unless ($gs_trunk) {
925 my ($trunk_url, $trunk_path) =
926 complete_svn_url($url, $_trunk);
927 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
928 undef, $trunk_ref);
929 }
930 }
931 return unless @_branches || @_tags;
932 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
933 foreach my $path (@_branches) {
934 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
935 }
936 foreach my $path (@_tags) {
937 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
938 }
939 }
940
941 sub cmd_multi_fetch {
942 my $remotes = Git::SVN::read_all_remotes();
943 foreach my $repo_id (sort keys %$remotes) {
944 if ($remotes->{$repo_id}->{url}) {
945 Git::SVN::fetch_all($repo_id, $remotes);
946 }
947 }
948 }
949
950 # this command is special because it requires no metadata
951 sub cmd_commit_diff {
952 my ($ta, $tb, $url) = @_;
953 my $usage = "Usage: $0 commit-diff -r<revision> ".
954 "<tree-ish> <tree-ish> [<URL>]";
955 fatal($usage) if (!defined $ta || !defined $tb);
956 my $svn_path = '';
957 if (!defined $url) {
958 my $gs = eval { Git::SVN->new };
959 if (!$gs) {
960 fatal("Needed URL or usable git-svn --id in ",
961 "the command-line\n", $usage);
962 }
963 $url = $gs->{url};
964 $svn_path = $gs->{path};
965 }
966 unless (defined $_revision) {
967 fatal("-r|--revision is a required argument\n", $usage);
968 }
969 if (defined $_message && defined $_file) {
970 fatal("Both --message/-m and --file/-F specified ",
971 "for the commit message.\n",
972 "I have no idea what you mean");
973 }
974 if (defined $_file) {
975 $_message = file_to_s($_file);
976 } else {
977 $_message ||= get_commit_entry($tb)->{log};
978 }
979 my $ra ||= Git::SVN::Ra->new($url);
980 my $r = $_revision;
981 if ($r eq 'HEAD') {
982 $r = $ra->get_latest_revnum;
983 } elsif ($r !~ /^\d+$/) {
984 die "revision argument: $r not understood by git-svn\n";
985 }
986 my %ed_opts = ( r => $r,
987 log => $_message,
988 ra => $ra,
989 tree_a => $ta,
990 tree_b => $tb,
991 editor_cb => sub { print "Committed r$_[0]\n" },
992 svn_path => $svn_path );
993 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
994 print "No changes\n$ta == $tb\n";
995 }
996 }
997
998 sub escape_uri_only {
999 my ($uri) = @_;
1000 my @tmp;
1001 foreach (split m{/}, $uri) {
1002 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1003 push @tmp, $_;
1004 }
1005 join('/', @tmp);
1006 }
1007
1008 sub escape_url {
1009 my ($url) = @_;
1010 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1011 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1012 $url = "$scheme://$domain$uri";
1013 }
1014 $url;
1015 }
1016
1017 sub cmd_info {
1018 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1019 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1020 if (exists $_[1]) {
1021 die "Too many arguments specified\n";
1022 }
1023
1024 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1025
1026 if (!$file_type && !$diff_status) {
1027 print STDERR "svn: '$path' is not under version control\n";
1028 exit 1;
1029 }
1030
1031 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1032 unless ($gs) {
1033 die "Unable to determine upstream SVN information from ",
1034 "working tree history\n";
1035 }
1036
1037 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1038 $path = "." if $path eq "";
1039
1040 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1041
1042 if ($_url) {
1043 print escape_url($full_url), "\n";
1044 return;
1045 }
1046
1047 my $result = "Path: $path\n";
1048 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1049 $result .= "URL: " . escape_url($full_url) . "\n";
1050
1051 eval {
1052 my $repos_root = $gs->repos_root;
1053 Git::SVN::remove_username($repos_root);
1054 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1055 };
1056 if ($@) {
1057 $result .= "Repository Root: (offline)\n";
1058 }
1059 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1060 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1061 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1062
1063 $result .= "Node Kind: " .
1064 ($file_type eq "dir" ? "directory" : "file") . "\n";
1065
1066 my $schedule = $diff_status eq "A"
1067 ? "add"
1068 : ($diff_status eq "D" ? "delete" : "normal");
1069 $result .= "Schedule: $schedule\n";
1070
1071 if ($diff_status eq "A") {
1072 print $result, "\n";
1073 return;
1074 }
1075
1076 my ($lc_author, $lc_rev, $lc_date_utc);
1077 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1078 my $log = command_output_pipe(@args);
1079 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1080 while (<$log>) {
1081 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1082 $lc_author = $1;
1083 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1084 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1085 (undef, $lc_rev, undef) = ::extract_metadata($1);
1086 }
1087 }
1088 close $log;
1089
1090 Git::SVN::Log::set_local_timezone();
1091
1092 $result .= "Last Changed Author: $lc_author\n";
1093 $result .= "Last Changed Rev: $lc_rev\n";
1094 $result .= "Last Changed Date: " .
1095 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1096
1097 if ($file_type ne "dir") {
1098 my $text_last_updated_date =
1099 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1100 $result .=
1101 "Text Last Updated: " .
1102 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1103 "\n";
1104 my $checksum;
1105 if ($diff_status eq "D") {
1106 my ($fh, $ctx) =
1107 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1108 if ($file_type eq "link") {
1109 my $file_name = <$fh>;
1110 $checksum = md5sum("link $file_name");
1111 } else {
1112 $checksum = md5sum($fh);
1113 }
1114 command_close_pipe($fh, $ctx);
1115 } elsif ($file_type eq "link") {
1116 my $file_name =
1117 command(qw(cat-file blob), "HEAD:$path");
1118 $checksum =
1119 md5sum("link " . $file_name);
1120 } else {
1121 open FILE, "<", $path or die $!;
1122 $checksum = md5sum(\*FILE);
1123 close FILE or die $!;
1124 }
1125 $result .= "Checksum: " . $checksum . "\n";
1126 }
1127
1128 print $result, "\n";
1129 }
1130
1131 sub cmd_reset {
1132 my $target = shift || $_revision or die "SVN revision required\n";
1133 $target = $1 if $target =~ /^r(\d+)$/;
1134 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1135 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1136 unless ($gs) {
1137 die "Unable to determine upstream SVN information from ".
1138 "history\n";
1139 }
1140 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1141 $gs->rev_map_set($r, $c, 'reset', $uuid);
1142 print "r$r = $c ($gs->{ref_id})\n";
1143 }
1144
1145 sub cmd_gc {
1146 if (!$can_compress) {
1147 warn "Compress::Zlib could not be found; unhandled.log " .
1148 "files will not be compressed.\n";
1149 }
1150 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1151 }
1152
1153 ########################### utility functions #########################
1154
1155 sub rebase_cmd {
1156 my @cmd = qw/rebase/;
1157 push @cmd, '-v' if $_verbose;
1158 push @cmd, qw/--merge/ if $_merge;
1159 push @cmd, "--strategy=$_strategy" if $_strategy;
1160 @cmd;
1161 }
1162
1163 sub post_fetch_checkout {
1164 return if $_no_checkout;
1165 my $gs = $Git::SVN::_head or return;
1166 return if verify_ref('refs/heads/master^0');
1167
1168 # look for "trunk" ref if it exists
1169 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1170 my $fetch = $remote->{fetch};
1171 if ($fetch) {
1172 foreach my $p (keys %$fetch) {
1173 basename($fetch->{$p}) eq 'trunk' or next;
1174 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1175 last;
1176 }
1177 }
1178
1179 my $valid_head = verify_ref('HEAD^0');
1180 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1181 return if ($valid_head || !verify_ref('HEAD^0'));
1182
1183 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1184 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1185 return if -f $index;
1186
1187 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1188 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1189 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1190 print STDERR "Checked out HEAD:\n ",
1191 $gs->full_url, " r", $gs->last_rev, "\n";
1192 }
1193
1194 sub complete_svn_url {
1195 my ($url, $path) = @_;
1196 $path =~ s#/+$##;
1197 if ($path !~ m#^[a-z\+]+://#) {
1198 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1199 fatal("E: '$path' is not a complete URL ",
1200 "and a separate URL is not specified");
1201 }
1202 return ($url, $path);
1203 }
1204 return ($path, '');
1205 }
1206
1207 sub complete_url_ls_init {
1208 my ($ra, $repo_path, $switch, $pfx) = @_;
1209 unless ($repo_path) {
1210 print STDERR "W: $switch not specified\n";
1211 return;
1212 }
1213 $repo_path =~ s#/+$##;
1214 if ($repo_path =~ m#^[a-z\+]+://#) {
1215 $ra = Git::SVN::Ra->new($repo_path);
1216 $repo_path = '';
1217 } else {
1218 $repo_path =~ s#^/+##;
1219 unless ($ra) {
1220 fatal("E: '$repo_path' is not a complete URL ",
1221 "and a separate URL is not specified");
1222 }
1223 }
1224 my $url = $ra->{url};
1225 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1226 my $k = "svn-remote.$gs->{repo_id}.url";
1227 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1228 if ($orig_url && ($orig_url ne $gs->{url})) {
1229 die "$k already set: $orig_url\n",
1230 "wanted to set to: $gs->{url}\n";
1231 }
1232 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1233 my $remote_path = "$gs->{path}/$repo_path";
1234 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1235 $remote_path =~ s#/+#/#g;
1236 $remote_path =~ s#^/##g;
1237 $remote_path .= "/*" if $remote_path !~ /\*/;
1238 my ($n) = ($switch =~ /^--(\w+)/);
1239 if (length $pfx && $pfx !~ m#/$#) {
1240 die "--prefix='$pfx' must have a trailing slash '/'\n";
1241 }
1242 command_noisy('config',
1243 '--add',
1244 "svn-remote.$gs->{repo_id}.$n",
1245 "$remote_path:refs/remotes/$pfx*" .
1246 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1247 }
1248
1249 sub verify_ref {
1250 my ($ref) = @_;
1251 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1252 { STDERR => 0 }); };
1253 }
1254
1255 sub get_tree_from_treeish {
1256 my ($treeish) = @_;
1257 # $treeish can be a symbolic ref, too:
1258 my $type = command_oneline(qw/cat-file -t/, $treeish);
1259 my $expected;
1260 while ($type eq 'tag') {
1261 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1262 }
1263 if ($type eq 'commit') {
1264 $expected = (grep /^tree /, command(qw/cat-file commit/,
1265 $treeish))[0];
1266 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1267 die "Unable to get tree from $treeish\n" unless $expected;
1268 } elsif ($type eq 'tree') {
1269 $expected = $treeish;
1270 } else {
1271 die "$treeish is a $type, expected tree, tag or commit\n";
1272 }
1273 return $expected;
1274 }
1275
1276 sub get_commit_entry {
1277 my ($treeish) = shift;
1278 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1279 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1280 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1281 open my $log_fh, '>', $commit_editmsg or croak $!;
1282
1283 my $type = command_oneline(qw/cat-file -t/, $treeish);
1284 if ($type eq 'commit' || $type eq 'tag') {
1285 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1286 $type, $treeish);
1287 my $in_msg = 0;
1288 my $author;
1289 my $saw_from = 0;
1290 my $msgbuf = "";
1291 while (<$msg_fh>) {
1292 if (!$in_msg) {
1293 $in_msg = 1 if (/^\s*$/);
1294 $author = $1 if (/^author (.*>)/);
1295 } elsif (/^git-svn-id: /) {
1296 # skip this for now, we regenerate the
1297 # correct one on re-fetch anyways
1298 # TODO: set *:merge properties or like...
1299 } else {
1300 if (/^From:/ || /^Signed-off-by:/) {
1301 $saw_from = 1;
1302 }
1303 $msgbuf .= $_;
1304 }
1305 }
1306 $msgbuf =~ s/\s+$//s;
1307 if ($Git::SVN::_add_author_from && defined($author)
1308 && !$saw_from) {
1309 $msgbuf .= "\n\nFrom: $author";
1310 }
1311 print $log_fh $msgbuf or croak $!;
1312 command_close_pipe($msg_fh, $ctx);
1313 }
1314 close $log_fh or croak $!;
1315
1316 if ($_edit || ($type eq 'tree')) {
1317 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1318 # TODO: strip out spaces, comments, like git-commit.sh
1319 system($editor, $commit_editmsg);
1320 }
1321 rename $commit_editmsg, $commit_msg or croak $!;
1322 {
1323 require Encode;
1324 # SVN requires messages to be UTF-8 when entering the repo
1325 local $/;
1326 open $log_fh, '<', $commit_msg or croak $!;
1327 binmode $log_fh;
1328 chomp($log_entry{log} = <$log_fh>);
1329
1330 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1331 my $msg = $log_entry{log};
1332
1333 eval { $msg = Encode::decode($enc, $msg, 1) };
1334 if ($@) {
1335 die "Could not decode as $enc:\n", $msg,
1336 "\nPerhaps you need to set i18n.commitencoding\n";
1337 }
1338
1339 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1340 die "Could not encode as UTF-8:\n$msg\n" if $@;
1341
1342 $log_entry{log} = $msg;
1343
1344 close $log_fh or croak $!;
1345 }
1346 unlink $commit_msg;
1347 \%log_entry;
1348 }
1349
1350 sub s_to_file {
1351 my ($str, $file, $mode) = @_;
1352 open my $fd,'>',$file or croak $!;
1353 print $fd $str,"\n" or croak $!;
1354 close $fd or croak $!;
1355 chmod ($mode &~ umask, $file) if (defined $mode);
1356 }
1357
1358 sub file_to_s {
1359 my $file = shift;
1360 open my $fd,'<',$file or croak "$!: file: $file\n";
1361 local $/;
1362 my $ret = <$fd>;
1363 close $fd or croak $!;
1364 $ret =~ s/\s*$//s;
1365 return $ret;
1366 }
1367
1368 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1369 sub load_authors {
1370 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1371 my $log = $cmd eq 'log';
1372 while (<$authors>) {
1373 chomp;
1374 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1375 my ($user, $name, $email) = ($1, $2, $3);
1376 if ($log) {
1377 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1378 } else {
1379 $users{$user} = [$name, $email];
1380 }
1381 }
1382 close $authors or croak $!;
1383 }
1384
1385 # convert GetOpt::Long specs for use by git-config
1386 sub read_repo_config {
1387 return unless -d $ENV{GIT_DIR};
1388 my $opts = shift;
1389 my @config_only;
1390 foreach my $o (keys %$opts) {
1391 # if we have mixedCase and a long option-only, then
1392 # it's a config-only variable that we don't need for
1393 # the command-line.
1394 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1395 my $v = $opts->{$o};
1396 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1397 $key =~ s/-//g;
1398 my $arg = 'git config';
1399 $arg .= ' --int' if ($o =~ /[:=]i$/);
1400 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1401 if (ref $v eq 'ARRAY') {
1402 chomp(my @tmp = `$arg --get-all svn.$key`);
1403 @$v = @tmp if @tmp;
1404 } else {
1405 chomp(my $tmp = `$arg --get svn.$key`);
1406 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1407 $$v = $tmp;
1408 }
1409 }
1410 }
1411 delete @$opts{@config_only} if @config_only;
1412 }
1413
1414 sub extract_metadata {
1415 my $id = shift or return (undef, undef, undef);
1416 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1417 \s([a-f\d\-]+)$/ix);
1418 if (!defined $rev || !$uuid || !$url) {
1419 # some of the original repositories I made had
1420 # identifiers like this:
1421 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1422 }
1423 return ($url, $rev, $uuid);
1424 }
1425
1426 sub cmt_metadata {
1427 return extract_metadata((grep(/^git-svn-id: /,
1428 command(qw/cat-file commit/, shift)))[-1]);
1429 }
1430
1431 sub cmt_sha2rev_batch {
1432 my %s2r;
1433 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1434 my $list = shift;
1435
1436 foreach my $sha (@{$list}) {
1437 my $first = 1;
1438 my $size = 0;
1439 print $out $sha, "\n";
1440
1441 while (my $line = <$in>) {
1442 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1443 last;
1444 } elsif ($first &&
1445 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1446 $first = 0;
1447 $size = $1;
1448 next;
1449 } elsif ($line =~ /^(git-svn-id: )/) {
1450 my (undef, $rev, undef) =
1451 extract_metadata($line);
1452 $s2r{$sha} = $rev;
1453 }
1454
1455 $size -= length($line);
1456 last if ($size == 0);
1457 }
1458 }
1459
1460 command_close_bidi_pipe($pid, $in, $out, $ctx);
1461
1462 return \%s2r;
1463 }
1464
1465 sub working_head_info {
1466 my ($head, $refs) = @_;
1467 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1468 my ($fh, $ctx) = command_output_pipe(@args, $head);
1469 my $hash;
1470 my %max;
1471 while (<$fh>) {
1472 if ( m{^commit ($::sha1)$} ) {
1473 unshift @$refs, $hash if $hash and $refs;
1474 $hash = $1;
1475 next;
1476 }
1477 next unless s{^\s*(git-svn-id:)}{$1};
1478 my ($url, $rev, $uuid) = extract_metadata($_);
1479 if (defined $url && defined $rev) {
1480 next if $max{$url} and $max{$url} < $rev;
1481 if (my $gs = Git::SVN->find_by_url($url)) {
1482 my $c = $gs->rev_map_get($rev, $uuid);
1483 if ($c && $c eq $hash) {
1484 close $fh; # break the pipe
1485 return ($url, $rev, $uuid, $gs);
1486 } else {
1487 $max{$url} ||= $gs->rev_map_max;
1488 }
1489 }
1490 }
1491 }
1492 command_close_pipe($fh, $ctx);
1493 (undef, undef, undef, undef);
1494 }
1495
1496 sub read_commit_parents {
1497 my ($parents, $c) = @_;
1498 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1499 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1500 @{$parents->{$c}} = split(/ /, $p);
1501 }
1502
1503 sub linearize_history {
1504 my ($gs, $refs) = @_;
1505 my %parents;
1506 foreach my $c (@$refs) {
1507 read_commit_parents(\%parents, $c);
1508 }
1509
1510 my @linear_refs;
1511 my %skip = ();
1512 my $last_svn_commit = $gs->last_commit;
1513 foreach my $c (reverse @$refs) {
1514 next if $c eq $last_svn_commit;
1515 last if $skip{$c};
1516
1517 unshift @linear_refs, $c;
1518 $skip{$c} = 1;
1519
1520 # we only want the first parent to diff against for linear
1521 # history, we save the rest to inject when we finalize the
1522 # svn commit
1523 my $fp_a = verify_ref("$c~1");
1524 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1525 if (!$fp_a || !$fp_b) {
1526 die "Commit $c\n",
1527 "has no parent commit, and therefore ",
1528 "nothing to diff against.\n",
1529 "You should be working from a repository ",
1530 "originally created by git-svn\n";
1531 }
1532 if ($fp_a ne $fp_b) {
1533 die "$c~1 = $fp_a, however parsing commit $c ",
1534 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1535 }
1536
1537 foreach my $p (@{$parents{$c}}) {
1538 $skip{$p} = 1;
1539 }
1540 }
1541 (\@linear_refs, \%parents);
1542 }
1543
1544 sub find_file_type_and_diff_status {
1545 my ($path) = @_;
1546 return ('dir', '') if $path eq '';
1547
1548 my $diff_output =
1549 command_oneline(qw(diff --cached --name-status --), $path) || "";
1550 my $diff_status = (split(' ', $diff_output))[0] || "";
1551
1552 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1553
1554 return (undef, undef) if !$diff_status && !$ls_tree;
1555
1556 if ($diff_status eq "A") {
1557 return ("link", $diff_status) if -l $path;
1558 return ("dir", $diff_status) if -d $path;
1559 return ("file", $diff_status);
1560 }
1561
1562 my $mode = (split(' ', $ls_tree))[0] || "";
1563
1564 return ("link", $diff_status) if $mode eq "120000";
1565 return ("dir", $diff_status) if $mode eq "040000";
1566 return ("file", $diff_status);
1567 }
1568
1569 sub md5sum {
1570 my $arg = shift;
1571 my $ref = ref $arg;
1572 my $md5 = Digest::MD5->new();
1573 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1574 $md5->addfile($arg) or croak $!;
1575 } elsif ($ref eq 'SCALAR') {
1576 $md5->add($$arg) or croak $!;
1577 } elsif (!$ref) {
1578 $md5->add($arg) or croak $!;
1579 } else {
1580 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1581 }
1582 return $md5->hexdigest();
1583 }
1584
1585 sub gc_directory {
1586 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1587 my $out_filename = $_ . ".gz";
1588 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1589 binmode $in_fh;
1590 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1591 die "Unable to open $out_filename: $!\n";
1592
1593 my $res;
1594 while ($res = sysread($in_fh, my $str, 1024)) {
1595 $gz->gzwrite($str) or
1596 die "Unable to write: ".$gz->gzerror()."!\n";
1597 }
1598 unlink $_ or die "unlink $File::Find::name: $!\n";
1599 } elsif (-f $_ && basename($_) eq "index") {
1600 unlink $_ or die "unlink $_: $!\n";
1601 }
1602 }
1603
1604 package Git::SVN;
1605 use strict;
1606 use warnings;
1607 use Fcntl qw/:DEFAULT :seek/;
1608 use constant rev_map_fmt => 'NH40';
1609 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1610 $_repack $_repack_flags $_use_svm_props $_head
1611 $_use_svnsync_props $no_reuse_existing $_minimize_url
1612 $_use_log_author $_add_author_from $_localtime/;
1613 use Carp qw/croak/;
1614 use File::Path qw/mkpath/;
1615 use File::Copy qw/copy/;
1616 use IPC::Open3;
1617
1618 my ($_gc_nr, $_gc_period);
1619
1620 # properties that we do not log:
1621 my %SKIP_PROP;
1622 BEGIN {
1623 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1624 svn:special svn:executable
1625 svn:entry:committed-rev
1626 svn:entry:last-author
1627 svn:entry:uuid
1628 svn:entry:committed-date/;
1629
1630 # some options are read globally, but can be overridden locally
1631 # per [svn-remote "..."] section. Command-line options will *NOT*
1632 # override options set in an [svn-remote "..."] section
1633 no strict 'refs';
1634 for my $option (qw/follow_parent no_metadata use_svm_props
1635 use_svnsync_props/) {
1636 my $key = $option;
1637 $key =~ tr/_//d;
1638 my $prop = "-$option";
1639 *$option = sub {
1640 my ($self) = @_;
1641 return $self->{$prop} if exists $self->{$prop};
1642 my $k = "svn-remote.$self->{repo_id}.$key";
1643 eval { command_oneline(qw/config --get/, $k) };
1644 if ($@) {
1645 $self->{$prop} = ${"Git::SVN::_$option"};
1646 } else {
1647 my $v = command_oneline(qw/config --bool/,$k);
1648 $self->{$prop} = $v eq 'false' ? 0 : 1;
1649 }
1650 return $self->{$prop};
1651 }
1652 }
1653 }
1654
1655
1656 my (%LOCKFILES, %INDEX_FILES);
1657 END {
1658 unlink keys %LOCKFILES if %LOCKFILES;
1659 unlink keys %INDEX_FILES if %INDEX_FILES;
1660 }
1661
1662 sub resolve_local_globs {
1663 my ($url, $fetch, $glob_spec) = @_;
1664 return unless defined $glob_spec;
1665 my $ref = $glob_spec->{ref};
1666 my $path = $glob_spec->{path};
1667 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1668 next unless m#^$ref->{regex}$#;
1669 my $p = $1;
1670 my $pathname = desanitize_refname($path->full_path($p));
1671 my $refname = desanitize_refname($ref->full_path($p));
1672 if (my $existing = $fetch->{$pathname}) {
1673 if ($existing ne $refname) {
1674 die "Refspec conflict:\n",
1675 "existing: $existing\n",
1676 " globbed: $refname\n";
1677 }
1678 my $u = (::cmt_metadata("$refname"))[0];
1679 $u =~ s!^\Q$url\E(/|$)!! or die
1680 "$refname: '$url' not found in '$u'\n";
1681 if ($pathname ne $u) {
1682 warn "W: Refspec glob conflict ",
1683 "(ref: $refname):\n",
1684 "expected path: $pathname\n",
1685 " real path: $u\n",
1686 "Continuing ahead with $u\n";
1687 next;
1688 }
1689 } else {
1690 $fetch->{$pathname} = $refname;
1691 }
1692 }
1693 }
1694
1695 sub parse_revision_argument {
1696 my ($base, $head) = @_;
1697 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1698 return ($base, $head);
1699 }
1700 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1701 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1702 return ($head, $head) if ($::_revision eq 'HEAD');
1703 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1704 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1705 die "revision argument: $::_revision not understood by git-svn\n";
1706 }
1707
1708 sub fetch_all {
1709 my ($repo_id, $remotes) = @_;
1710 if (ref $repo_id) {
1711 my $gs = $repo_id;
1712 $repo_id = undef;
1713 $repo_id = $gs->{repo_id};
1714 }
1715 $remotes ||= read_all_remotes();
1716 my $remote = $remotes->{$repo_id} or
1717 die "[svn-remote \"$repo_id\"] unknown\n";
1718 my $fetch = $remote->{fetch};
1719 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1720 my (@gs, @globs);
1721 my $ra = Git::SVN::Ra->new($url);
1722 my $uuid = $ra->get_uuid;
1723 my $head = $ra->get_latest_revnum;
1724 $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1725 my $base = defined $fetch ? $head : 0;
1726
1727 # read the max revs for wildcard expansion (branches/*, tags/*)
1728 foreach my $t (qw/branches tags/) {
1729 defined $remote->{$t} or next;
1730 push @globs, @{$remote->{$t}};
1731
1732 my $max_rev = eval { tmp_config(qw/--int --get/,
1733 "svn-remote.$repo_id.${t}-maxRev") };
1734 if (defined $max_rev && ($max_rev < $base)) {
1735 $base = $max_rev;
1736 } elsif (!defined $max_rev) {
1737 $base = 0;
1738 }
1739 }
1740
1741 if ($fetch) {
1742 foreach my $p (sort keys %$fetch) {
1743 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1744 my $lr = $gs->rev_map_max;
1745 if (defined $lr) {
1746 $base = $lr if ($lr < $base);
1747 }
1748 push @gs, $gs;
1749 }
1750 }
1751
1752 ($base, $head) = parse_revision_argument($base, $head);
1753 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1754 }
1755
1756 sub read_all_remotes {
1757 my $r = {};
1758 my $use_svm_props = eval { command_oneline(qw/config --bool
1759 svn.useSvmProps/) };
1760 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1761 my $svn_refspec = qr{\s*/?(.*?)\s*:\s*(.+?)\s*};
1762 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1763 if (m!^(.+)\.fetch=$svn_refspec$!) {
1764 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1765 die("svn-remote.$remote: remote ref '$remote_ref' "
1766 . "must start with 'refs/'\n")
1767 unless $remote_ref =~ m{^refs/};
1768 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1769 $r->{$remote}->{svm} = {} if $use_svm_props;
1770 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1771 $r->{$1}->{svm} = {};
1772 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1773 $r->{$1}->{url} = $2;
1774 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1775 my ($remote, $t, $local_ref, $remote_ref) =
1776 ($1, $2, $3, $4);
1777 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1778 . "must start with 'refs/'\n")
1779 unless $remote_ref =~ m{^refs/};
1780 my $rs = {
1781 t => $t,
1782 remote => $remote,
1783 path => Git::SVN::GlobSpec->new($local_ref),
1784 ref => Git::SVN::GlobSpec->new($remote_ref) };
1785 if (length($rs->{ref}->{right}) != 0) {
1786 die "The '*' glob character must be the last ",
1787 "character of '$remote_ref'\n";
1788 }
1789 push @{ $r->{$remote}->{$t} }, $rs;
1790 }
1791 }
1792
1793 map {
1794 if (defined $r->{$_}->{svm}) {
1795 my $svm;
1796 eval {
1797 my $section = "svn-remote.$_";
1798 $svm = {
1799 source => tmp_config('--get',
1800 "$section.svm-source"),
1801 replace => tmp_config('--get',
1802 "$section.svm-replace"),
1803 }
1804 };
1805 $r->{$_}->{svm} = $svm;
1806 }
1807 } keys %$r;
1808
1809 $r;
1810 }
1811
1812 sub init_vars {
1813 $_gc_nr = $_gc_period = 1000;
1814 if (defined $_repack || defined $_repack_flags) {
1815 warn "Repack options are obsolete; they have no effect.\n";
1816 }
1817 }
1818
1819 sub verify_remotes_sanity {
1820 return unless -d $ENV{GIT_DIR};
1821 my %seen;
1822 foreach (command(qw/config -l/)) {
1823 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1824 if ($seen{$1}) {
1825 die "Remote ref refs/remote/$1 is tracked by",
1826 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1827 "Please resolve this ambiguity in ",
1828 "your git configuration file before ",
1829 "continuing\n";
1830 }
1831 $seen{$1} = $_;
1832 }
1833 }
1834 }
1835
1836 sub find_existing_remote {
1837 my ($url, $remotes) = @_;
1838 return undef if $no_reuse_existing;
1839 my $existing;
1840 foreach my $repo_id (keys %$remotes) {
1841 my $u = $remotes->{$repo_id}->{url} or next;
1842 next if $u ne $url;
1843 $existing = $repo_id;
1844 last;
1845 }
1846 $existing;
1847 }
1848
1849 sub init_remote_config {
1850 my ($self, $url, $no_write) = @_;
1851 $url =~ s!/+$!!; # strip trailing slash
1852 my $r = read_all_remotes();
1853 my $existing = find_existing_remote($url, $r);
1854 if ($existing) {
1855 unless ($no_write) {
1856 print STDERR "Using existing ",
1857 "[svn-remote \"$existing\"]\n";
1858 }
1859 $self->{repo_id} = $existing;
1860 } elsif ($_minimize_url) {
1861 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1862 $existing = find_existing_remote($min_url, $r);
1863 if ($existing) {
1864 unless ($no_write) {
1865 print STDERR "Using existing ",
1866 "[svn-remote \"$existing\"]\n";
1867 }
1868 $self->{repo_id} = $existing;
1869 }
1870 if ($min_url ne $url) {
1871 unless ($no_write) {
1872 print STDERR "Using higher level of URL: ",
1873 "$url => $min_url\n";
1874 }
1875 my $old_path = $self->{path};
1876 $self->{path} = $url;
1877 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1878 if (length $old_path) {
1879 $self->{path} .= "/$old_path";
1880 }
1881 $url = $min_url;
1882 }
1883 }
1884 my $orig_url;
1885 if (!$existing) {
1886 # verify that we aren't overwriting anything:
1887 $orig_url = eval {
1888 command_oneline('config', '--get',
1889 "svn-remote.$self->{repo_id}.url")
1890 };
1891 if ($orig_url && ($orig_url ne $url)) {
1892 die "svn-remote.$self->{repo_id}.url already set: ",
1893 "$orig_url\nwanted to set to: $url\n";
1894 }
1895 }
1896 my ($xrepo_id, $xpath) = find_ref($self->refname);
1897 if (!$no_write && defined $xpath) {
1898 die "svn-remote.$xrepo_id.fetch already set to track ",
1899 "$xpath:", $self->refname, "\n";
1900 }
1901 unless ($no_write) {
1902 command_noisy('config',
1903 "svn-remote.$self->{repo_id}.url", $url);
1904 $self->{path} =~ s{^/}{};
1905 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1906 command_noisy('config', '--add',
1907 "svn-remote.$self->{repo_id}.fetch",
1908 "$self->{path}:".$self->refname);
1909 }
1910 $self->{url} = $url;
1911 }
1912
1913 sub find_by_url { # repos_root and, path are optional
1914 my ($class, $full_url, $repos_root, $path) = @_;
1915
1916 return undef unless defined $full_url;
1917 remove_username($full_url);
1918 remove_username($repos_root) if defined $repos_root;
1919 my $remotes = read_all_remotes();
1920 if (defined $full_url && defined $repos_root && !defined $path) {
1921 $path = $full_url;
1922 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1923 }
1924 foreach my $repo_id (keys %$remotes) {
1925 my $u = $remotes->{$repo_id}->{url} or next;
1926 remove_username($u);
1927 next if defined $repos_root && $repos_root ne $u;
1928
1929 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1930 foreach my $t (qw/branches tags/) {
1931 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1932 resolve_local_globs($u, $fetch, $globspec);
1933 }
1934 }
1935 my $p = $path;
1936 my $rwr = rewrite_root({repo_id => $repo_id});
1937 my $svm = $remotes->{$repo_id}->{svm}
1938 if defined $remotes->{$repo_id}->{svm};
1939 unless (defined $p) {
1940 $p = $full_url;
1941 my $z = $u;
1942 my $prefix = '';
1943 if ($rwr) {
1944 $z = $rwr;
1945 remove_username($z);
1946 } elsif (defined $svm) {
1947 $z = $svm->{source};
1948 $prefix = $svm->{replace};
1949 $prefix =~ s#^\Q$u\E(?:/|$)##;
1950 $prefix =~ s#/$##;
1951 }
1952 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1953 }
1954 foreach my $f (keys %$fetch) {
1955 next if $f ne $p;
1956 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1957 }
1958 }
1959 undef;
1960 }
1961
1962 sub init {
1963 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1964 my $self = _new($class, $repo_id, $ref_id, $path);
1965 if (defined $url) {
1966 $self->init_remote_config($url, $no_write);
1967 }
1968 $self;
1969 }
1970
1971 sub find_ref {
1972 my ($ref_id) = @_;
1973 foreach (command(qw/config -l/)) {
1974 next unless m!^svn-remote\.(.+)\.fetch=
1975 \s*/?(.*?)\s*:\s*(.+?)\s*$!x;
1976 my ($repo_id, $path, $ref) = ($1, $2, $3);
1977 if ($ref eq $ref_id) {
1978 $path = '' if ($path =~ m#^\./?#);
1979 return ($repo_id, $path);
1980 }
1981 }
1982 (undef, undef, undef);
1983 }
1984
1985 sub new {
1986 my ($class, $ref_id, $repo_id, $path) = @_;
1987 if (defined $ref_id && !defined $repo_id && !defined $path) {
1988 ($repo_id, $path) = find_ref($ref_id);
1989 if (!defined $repo_id) {
1990 die "Could not find a \"svn-remote.*.fetch\" key ",
1991 "in the repository configuration matching: ",
1992 "$ref_id\n";
1993 }
1994 }
1995 my $self = _new($class, $repo_id, $ref_id, $path);
1996 if (!defined $self->{path} || !length $self->{path}) {
1997 my $fetch = command_oneline('config', '--get',
1998 "svn-remote.$repo_id.fetch",
1999 ":$ref_id\$") or
2000 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2001 "\":$ref_id\$\" in config\n";
2002 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2003 }
2004 $self->{url} = command_oneline('config', '--get',
2005 "svn-remote.$repo_id.url") or
2006 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2007 $self->rebuild;
2008 $self;
2009 }
2010
2011 sub refname {
2012 my ($refname) = $_[0]->{ref_id} ;
2013
2014 # It cannot end with a slash /, we'll throw up on this because
2015 # SVN can't have directories with a slash in their name, either:
2016 if ($refname =~ m{/$}) {
2017 die "ref: '$refname' ends with a trailing slash, this is ",
2018 "not permitted by git nor Subversion\n";
2019 }
2020
2021 # It cannot have ASCII control character space, tilde ~, caret ^,
2022 # colon :, question-mark ?, asterisk *, space, or open bracket [
2023 # anywhere.
2024 #
2025 # Additionally, % must be escaped because it is used for escaping
2026 # and we want our escaped refname to be reversible
2027 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2028
2029 # no slash-separated component can begin with a dot .
2030 # /.* becomes /%2E*
2031 $refname =~ s{/\.}{/%2E}g;
2032
2033 # It cannot have two consecutive dots .. anywhere
2034 # .. becomes %2E%2E
2035 $refname =~ s{\.\.}{%2E%2E}g;
2036
2037 return $refname;
2038 }
2039
2040 sub desanitize_refname {
2041 my ($refname) = @_;
2042 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2043 return $refname;
2044 }
2045
2046 sub svm_uuid {
2047 my ($self) = @_;
2048 return $self->{svm}->{uuid} if $self->svm;
2049 $self->ra;
2050 unless ($self->{svm}) {
2051 die "SVM UUID not cached, and reading remotely failed\n";
2052 }
2053 $self->{svm}->{uuid};
2054 }
2055
2056 sub svm {
2057 my ($self) = @_;
2058 return $self->{svm} if $self->{svm};
2059 my $svm;
2060 # see if we have it in our config, first:
2061 eval {
2062 my $section = "svn-remote.$self->{repo_id}";
2063 $svm = {
2064 source => tmp_config('--get', "$section.svm-source"),
2065 uuid => tmp_config('--get', "$section.svm-uuid"),
2066 replace => tmp_config('--get', "$section.svm-replace"),
2067 }
2068 };
2069 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2070 $self->{svm} = $svm;
2071 }
2072 $self->{svm};
2073 }
2074
2075 sub _set_svm_vars {
2076 my ($self, $ra) = @_;
2077 return $ra if $self->svm;
2078
2079 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2080 "(svm:source, svm:uuid) ",
2081 "from the following URLs:\n" );
2082 sub read_svm_props {
2083 my ($self, $ra, $path, $r) = @_;
2084 my $props = ($ra->get_dir($path, $r))[2];
2085 my $src = $props->{'svm:source'};
2086 my $uuid = $props->{'svm:uuid'};
2087 return undef if (!$src || !$uuid);
2088
2089 chomp($src, $uuid);
2090
2091 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2092 or die "doesn't look right - svm:uuid is '$uuid'\n";
2093
2094 # the '!' is used to mark the repos_root!/relative/path
2095 $src =~ s{/?!/?}{/};
2096 $src =~ s{/+$}{}; # no trailing slashes please
2097 # username is of no interest
2098 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2099
2100 my $replace = $ra->{url};
2101 $replace .= "/$path" if length $path;
2102
2103 my $section = "svn-remote.$self->{repo_id}";
2104 tmp_config("$section.svm-source", $src);
2105 tmp_config("$section.svm-replace", $replace);
2106 tmp_config("$section.svm-uuid", $uuid);
2107 $self->{svm} = {
2108 source => $src,
2109 uuid => $uuid,
2110 replace => $replace
2111 };
2112 }
2113
2114 my $r = $ra->get_latest_revnum;
2115 my $path = $self->{path};
2116 my %tried;
2117 while (length $path) {
2118 unless ($tried{"$self->{url}/$path"}) {
2119 return $ra if $self->read_svm_props($ra, $path, $r);
2120 $tried{"$self->{url}/$path"} = 1;
2121 }
2122 $path =~ s#/?[^/]+$##;
2123 }
2124 die "Path: '$path' should be ''\n" if $path ne '';
2125 return $ra if $self->read_svm_props($ra, $path, $r);
2126 $tried{"$self->{url}/$path"} = 1;
2127
2128 if ($ra->{repos_root} eq $self->{url}) {
2129 die @err, (map { " $_\n" } keys %tried), "\n";
2130 }
2131
2132 # nope, make sure we're connected to the repository root:
2133 my $ok;
2134 my @tried_b;
2135 $path = $ra->{svn_path};
2136 $ra = Git::SVN::Ra->new($ra->{repos_root});
2137 while (length $path) {
2138 unless ($tried{"$ra->{url}/$path"}) {
2139 $ok = $self->read_svm_props($ra, $path, $r);
2140 last if $ok;
2141 $tried{"$ra->{url}/$path"} = 1;
2142 }
2143 $path =~ s#/?[^/]+$##;
2144 }
2145 die "Path: '$path' should be ''\n" if $path ne '';
2146 $ok ||= $self->read_svm_props($ra, $path, $r);
2147 $tried{"$ra->{url}/$path"} = 1;
2148 if (!$ok) {
2149 die @err, (map { " $_\n" } keys %tried), "\n";
2150 }
2151 Git::SVN::Ra->new($self->{url});
2152 }
2153
2154 sub svnsync {
2155 my ($self) = @_;
2156 return $self->{svnsync} if $self->{svnsync};
2157
2158 if ($self->no_metadata) {
2159 die "Can't have both 'noMetadata' and ",
2160 "'useSvnsyncProps' options set!\n";
2161 }
2162 if ($self->rewrite_root) {
2163 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2164 "options set!\n";
2165 }
2166
2167 my $svnsync;
2168 # see if we have it in our config, first:
2169 eval {
2170 my $section = "svn-remote.$self->{repo_id}";
2171
2172 my $url = tmp_config('--get', "$section.svnsync-url");
2173 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2174 die "doesn't look right - svn:sync-from-url is '$url'\n";
2175
2176 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2177 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2178 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2179
2180 $svnsync = { url => $url, uuid => $uuid }
2181 };
2182 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2183 return $self->{svnsync} = $svnsync;
2184 }
2185
2186 my $err = "useSvnsyncProps set, but failed to read " .
2187 "svnsync property: svn:sync-from-";
2188 my $rp = $self->ra->rev_proplist(0);
2189
2190 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2191 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2192 die "doesn't look right - svn:sync-from-url is '$url'\n";
2193
2194 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2195 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2196 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2197
2198 my $section = "svn-remote.$self->{repo_id}";
2199 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2200 tmp_config('--add', "$section.svnsync-url", $url);
2201 return $self->{svnsync} = { url => $url, uuid => $uuid };
2202 }
2203
2204 # this allows us to memoize our SVN::Ra UUID locally and avoid a
2205 # remote lookup (useful for 'git svn log').
2206 sub ra_uuid {
2207 my ($self) = @_;
2208 unless ($self->{ra_uuid}) {
2209 my $key = "svn-remote.$self->{repo_id}.uuid";
2210 my $uuid = eval { tmp_config('--get', $key) };
2211 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2212 $self->{ra_uuid} = $uuid;
2213 } else {
2214 die "ra_uuid called without URL\n" unless $self->{url};
2215 $self->{ra_uuid} = $self->ra->get_uuid;
2216 tmp_config('--add', $key, $self->{ra_uuid});
2217 }
2218 }
2219 $self->{ra_uuid};
2220 }
2221
2222 sub _set_repos_root {
2223 my ($self, $repos_root) = @_;
2224 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2225 $repos_root ||= $self->ra->{repos_root};
2226 tmp_config($k, $repos_root);
2227 $repos_root;
2228 }
2229
2230 sub repos_root {
2231 my ($self) = @_;
2232 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2233 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2234 }
2235
2236 sub ra {
2237 my ($self) = shift;
2238 my $ra = Git::SVN::Ra->new($self->{url});
2239 $self->_set_repos_root($ra->{repos_root});
2240 if ($self->use_svm_props && !$self->{svm}) {
2241 if ($self->no_metadata) {
2242 die "Can't have both 'noMetadata' and ",
2243 "'useSvmProps' options set!\n";
2244 } elsif ($self->use_svnsync_props) {
2245 die "Can't have both 'useSvnsyncProps' and ",
2246 "'useSvmProps' options set!\n";
2247 }
2248 $ra = $self->_set_svm_vars($ra);
2249 $self->{-want_revprops} = 1;
2250 }
2251 $ra;
2252 }
2253
2254 # prop_walk(PATH, REV, SUB)
2255 # -------------------------
2256 # Recursively traverse PATH at revision REV and invoke SUB for each
2257 # directory that contains a SVN property. SUB will be invoked as
2258 # follows: &SUB(gs, path, props); where `gs' is this instance of
2259 # Git::SVN, `path' the path to the directory where the properties
2260 # `props' were found. The `path' will be relative to point of checkout,
2261 # that is, if url://repo/trunk is the current Git branch, and that
2262 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
2263 # as `path' (note the trailing `/').
2264 sub prop_walk {
2265 my ($self, $path, $rev, $sub) = @_;
2266
2267 $path =~ s#^/##;
2268 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2269 $path =~ s#^/*#/#g;
2270 my $p = $path;
2271 # Strip the irrelevant part of the path.
2272 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2273 # Ensure the path is terminated by a `/'.
2274 $p =~ s#/*$#/#;
2275
2276 # The properties contain all the internal SVN stuff nobody
2277 # (usually) cares about.
2278 my $interesting_props = 0;
2279 foreach (keys %{$props}) {
2280 # If it doesn't start with `svn:', it must be a
2281 # user-defined property.
2282 ++$interesting_props and next if $_ !~ /^svn:/;
2283 # FIXME: Fragile, if SVN adds new public properties,
2284 # this needs to be updated.
2285 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2286 |eol-style|mime-type
2287 |externals|needs-lock)$/x;
2288 }
2289 &$sub($self, $p, $props) if $interesting_props;
2290
2291 foreach (sort keys %$dirent) {
2292 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2293 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2294 }
2295 }
2296
2297 sub last_rev { ($_[0]->last_rev_commit)[0] }
2298 sub last_commit { ($_[0]->last_rev_commit)[1] }
2299
2300 # returns the newest SVN revision number and newest commit SHA1
2301 sub last_rev_commit {
2302 my ($self) = @_;
2303 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2304 return ($self->{last_rev}, $self->{last_commit});
2305 }
2306 my $c = ::verify_ref($self->refname.'^0');
2307 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2308 my $rev = (::cmt_metadata($c))[1];
2309 if (defined $rev) {
2310 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2311 return ($rev, $c);
2312 }
2313 }
2314 my $map_path = $self->map_path;
2315 unless (-e $map_path) {
2316 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2317 return (undef, undef);
2318 }
2319 my ($rev, $commit) = $self->rev_map_max(1);
2320 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2321 return ($rev, $commit);
2322 }
2323
2324 sub get_fetch_range {
2325 my ($self, $min, $max) = @_;
2326 $max ||= $self->ra->get_latest_revnum;
2327 $min ||= $self->rev_map_max;
2328 (++$min, $max);
2329 }
2330
2331 sub tmp_config {
2332 my (@args) = @_;
2333 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2334 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2335 if (! -f $config && -f $old_def_config) {
2336 rename $old_def_config, $config or
2337 die "Failed rename $old_def_config => $config: $!\n";
2338 }
2339 my $old_config = $ENV{GIT_CONFIG};
2340 $ENV{GIT_CONFIG} = $config;
2341 $@ = undef;
2342 my @ret = eval {
2343 unless (-f $config) {
2344 mkfile($config);
2345 open my $fh, '>', $config or
2346 die "Can't open $config: $!\n";
2347 print $fh "; This file is used internally by ",
2348 "git-svn\n" or die
2349 "Couldn't write to $config: $!\n";
2350 print $fh "; You should not have to edit it\n" or
2351 die "Couldn't write to $config: $!\n";
2352 close $fh or die "Couldn't close $config: $!\n";
2353 }
2354 command('config', @args);
2355 };
2356 my $err = $@;
2357 if (defined $old_config) {
2358 $ENV{GIT_CONFIG} = $old_config;
2359 } else {
2360 delete $ENV{GIT_CONFIG};
2361 }
2362 die $err if $err;
2363 wantarray ? @ret : $ret[0];
2364 }
2365
2366 sub tmp_index_do {
2367 my ($self, $sub) = @_;
2368 my $old_index = $ENV{GIT_INDEX_FILE};
2369 $ENV{GIT_INDEX_FILE} = $self->{index};
2370 $@ = undef;
2371 my @ret = eval {
2372 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2373 mkpath([$dir]) unless -d $dir;
2374 &$sub;
2375 };
2376 my $err = $@;
2377 if (defined $old_index) {
2378 $ENV{GIT_INDEX_FILE} = $old_index;
2379 } else {
2380 delete $ENV{GIT_INDEX_FILE};
2381 }
2382 die $err if $err;
2383 wantarray ? @ret : $ret[0];
2384 }
2385
2386 sub assert_index_clean {
2387 my ($self, $treeish) = @_;
2388
2389 $self->tmp_index_do(sub {
2390 command_noisy('read-tree', $treeish) unless -e $self->{index};
2391 my $x = command_oneline('write-tree');
2392 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2393 /^tree ($::sha1)/mo);
2394 return if $y eq $x;
2395
2396 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2397 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2398 command_noisy('read-tree', $treeish);
2399 $x = command_oneline('write-tree');
2400 if ($y ne $x) {
2401 ::fatal "trees ($treeish) $y != $x\n",
2402 "Something is seriously wrong...";
2403 }
2404 });
2405 }
2406
2407 sub get_commit_parents {
2408 my ($self, $log_entry) = @_;
2409 my (%seen, @ret, @tmp);
2410 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2411 if (my $ip = $self->{inject_parents}) {
2412 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2413 push @tmp, $commit;
2414 }
2415 }
2416 if (my $cur = ::verify_ref($self->refname.'^0')) {
2417 push @tmp, $cur;
2418 }
2419 if (my $ipd = $self->{inject_parents_dcommit}) {
2420 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2421 push @tmp, @$commit;
2422 }
2423 }
2424 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2425 while (my $p = shift @tmp) {
2426 next if $seen{$p};
2427 $seen{$p} = 1;
2428 push @ret, $p;
2429 # MAXPARENT is defined to 16 in commit-tree.c:
2430 last if @ret >= 16;
2431 }
2432 if (@tmp) {
2433 die "r$log_entry->{revision}: No room for parents:\n\t",
2434 join("\n\t", @tmp), "\n";
2435 }
2436 @ret;
2437 }
2438
2439 sub rewrite_root {
2440 my ($self) = @_;
2441 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2442 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2443 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2444 if ($rwr) {
2445 $rwr =~ s#/+$##;
2446 if ($rwr !~ m#^[a-z\+]+://#) {
2447 die "$rwr is not a valid URL (key: $k)\n";
2448 }
2449 }
2450 $self->{-rewrite_root} = $rwr;
2451 }
2452
2453 sub metadata_url {
2454 my ($self) = @_;
2455 ($self->rewrite_root || $self->{url}) .
2456 (length $self->{path} ? '/' . $self->{path} : '');
2457 }
2458
2459 sub full_url {
2460 my ($self) = @_;
2461 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2462 }
2463
2464
2465 sub set_commit_header_env {
2466 my ($log_entry) = @_;
2467 my %env;
2468 foreach my $ned (qw/NAME EMAIL DATE/) {
2469 foreach my $ac (qw/AUTHOR COMMITTER/) {
2470 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2471 }
2472 }
2473
2474 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2475 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2476 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2477
2478 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2479 ? $log_entry->{commit_name}
2480 : $log_entry->{name};
2481 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2482 ? $log_entry->{commit_email}
2483 : $log_entry->{email};
2484 \%env;
2485 }
2486
2487 sub restore_commit_header_env {
2488 my ($env) = @_;
2489 foreach my $ned (qw/NAME EMAIL DATE/) {
2490 foreach my $ac (qw/AUTHOR COMMITTER/) {
2491 my $k = "GIT_${ac}_${ned}";
2492 if (defined $env->{$k}) {
2493 $ENV{$k} = $env->{$k};
2494 } else {
2495 delete $ENV{$k};
2496 }
2497 }
2498 }
2499 }
2500
2501 sub gc {
2502 command_noisy('gc', '--auto');
2503 };
2504
2505 sub do_git_commit {
2506 my ($self, $log_entry) = @_;
2507 my $lr = $self->last_rev;
2508 if (defined $lr && $lr >= $log_entry->{revision}) {
2509 die "Last fetched revision of ", $self->refname,
2510 " was r$lr, but we are about to fetch: ",
2511 "r$log_entry->{revision}!\n";
2512 }
2513 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2514 croak "$log_entry->{revision} = $c already exists! ",
2515 "Why are we refetching it?\n";
2516 }
2517 my $old_env = set_commit_header_env($log_entry);
2518 my $tree = $log_entry->{tree};
2519 if (!defined $tree) {
2520 $tree = $self->tmp_index_do(sub {
2521 command_oneline('write-tree') });
2522 }
2523 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2524
2525 my @exec = ('git', 'commit-tree', $tree);
2526 foreach ($self->get_commit_parents($log_entry)) {
2527 push @exec, '-p', $_;
2528 }
2529 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2530 or croak $!;
2531 binmode $msg_fh;
2532
2533 # we always get UTF-8 from SVN, but we may want our commits in
2534 # a different encoding.
2535 if (my $enc = Git::config('i18n.commitencoding')) {
2536 require Encode;
2537 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2538 }
2539 print $msg_fh $log_entry->{log} or croak $!;
2540 restore_commit_header_env($old_env);
2541 unless ($self->no_metadata) {
2542 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2543 or croak $!;
2544 }
2545 $msg_fh->flush == 0 or croak $!;
2546 close $msg_fh or croak $!;
2547 chomp(my $commit = do { local $/; <$out_fh> });
2548 close $out_fh or croak $!;
2549 waitpid $pid, 0;
2550 croak $? if $?;
2551 if ($commit !~ /^$::sha1$/o) {
2552 die "Failed to commit, invalid sha1: $commit\n";
2553 }
2554
2555 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2556
2557 $self->{last_rev} = $log_entry->{revision};
2558 $self->{last_commit} = $commit;
2559 print "r$log_entry->{revision}" unless $::_q > 1;
2560 if (defined $log_entry->{svm_revision}) {
2561 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2562 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2563 0, $self->svm_uuid);
2564 }
2565 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2566 if (--$_gc_nr == 0) {
2567 $_gc_nr = $_gc_period;
2568 gc();
2569 }
2570 return $commit;
2571 }
2572
2573 sub match_paths {
2574 my ($self, $paths, $r) = @_;
2575 return 1 if $self->{path} eq '';
2576 if (my $path = $paths->{"/$self->{path}"}) {
2577 return ($path->{action} eq 'D') ? 0 : 1;
2578 }
2579 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2580 if (grep /$self->{path_regex}/, keys %$paths) {
2581 return 1;
2582 }
2583 my $c = '';
2584 foreach (split m#/#, $self->{path}) {
2585 $c .= "/$_";
2586 next unless ($paths->{$c} &&
2587 ($paths->{$c}->{action} =~ /^[AR]$/));
2588 if ($self->ra->check_path($self->{path}, $r) ==
2589 $SVN::Node::dir) {
2590 return 1;
2591 }
2592 }
2593 return 0;
2594 }
2595
2596 sub find_parent_branch {
2597 my ($self, $paths, $rev) = @_;
2598 return undef unless $self->follow_parent;
2599 unless (defined $paths) {
2600 my $err_handler = $SVN::Error::handler;
2601 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2602 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2603 sub { $paths = $_[0] });
2604 $SVN::Error::handler = $err_handler;
2605 }
2606 return undef unless defined $paths;
2607
2608 # look for a parent from another branch:
2609 my @b_path_components = split m#/#, $self->{path};
2610 my @a_path_components;
2611 my $i;
2612 while (@b_path_components) {
2613 $i = $paths->{'/'.join('/', @b_path_components)};
2614 last if $i && defined $i->{copyfrom_path};
2615 unshift(@a_path_components, pop(@b_path_components));
2616 }
2617 return undef unless defined $i && defined $i->{copyfrom_path};
2618 my $branch_from = $i->{copyfrom_path};
2619 if (@a_path_components) {
2620 print STDERR "branch_from: $branch_from => ";
2621 $branch_from .= '/'.join('/', @a_path_components);
2622 print STDERR $branch_from, "\n";
2623 }
2624 my $r = $i->{copyfrom_rev};
2625 my $repos_root = $self->ra->{repos_root};
2626 my $url = $self->ra->{url};
2627 my $new_url = $url . $branch_from;
2628 print STDERR "Found possible branch point: ",
2629 "$new_url => ", $self->full_url, ", $r\n";
2630 $branch_from =~ s#^/##;
2631 my $gs = $self->other_gs($new_url, $url,
2632 $branch_from, $r, $self->{ref_id});
2633 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2634 {
2635 my ($base, $head);
2636 if (!defined $r0 || !defined $parent) {
2637 ($base, $head) = parse_revision_argument(0, $r);
2638 } else {
2639 if ($r0 < $r) {
2640 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2641 0, 1, sub { $base = $_[1] - 1 });
2642 }
2643 }
2644 if (defined $base && $base <= $r) {
2645 $gs->fetch($base, $r);
2646 }
2647 ($r0, $parent) = $gs->find_rev_before($r, 1);
2648 }
2649 if (defined $r0 && defined $parent) {
2650 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2651 my $ed;
2652 if ($self->ra->can_do_switch) {
2653 $self->assert_index_clean($parent);
2654 print STDERR "Following parent with do_switch\n";
2655 # do_switch works with svn/trunk >= r22312, but that
2656 # is not included with SVN 1.4.3 (the latest version
2657 # at the moment), so we can't rely on it
2658 $self->{last_rev} = $r0;
2659 $self->{last_commit} = $parent;
2660 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2661 $gs->ra->gs_do_switch($r0, $rev, $gs,
2662 $self->full_url, $ed)
2663 or die "SVN connection failed somewhere...\n";
2664 } elsif ($self->ra->trees_match($new_url, $r0,
2665 $self->full_url, $rev)) {
2666 print STDERR "Trees match:\n",
2667 " $new_url\@$r0\n",
2668 " ${\$self->full_url}\@$rev\n",
2669 "Following parent with no changes\n";
2670 $self->tmp_index_do(sub {
2671 command_noisy('read-tree', $parent);
2672 });
2673 $self->{last_commit} = $parent;
2674 } else {
2675 print STDERR "Following parent with do_update\n";
2676 $ed = SVN::Git::Fetcher->new($self);
2677 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2678 or die "SVN connection failed somewhere...\n";
2679 }
2680 print STDERR "Successfully followed parent\n";
2681 return $self->make_log_entry($rev, [$parent], $ed);
2682 }
2683 return undef;
2684 }
2685
2686 sub do_fetch {
2687 my ($self, $paths, $rev) = @_;
2688 my $ed;
2689 my ($last_rev, @parents);
2690 if (my $lc = $self->last_commit) {
2691 # we can have a branch that was deleted, then re-added
2692 # under the same name but copied from another path, in
2693 # which case we'll have multiple parents (we don't
2694 # want to break the original ref, nor lose copypath info):
2695 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2696 push @{$log_entry->{parents}}, $lc;
2697 return $log_entry;
2698 }
2699 $ed = SVN::Git::Fetcher->new($self);
2700 $last_rev = $self->{last_rev};
2701 $ed->{c} = $lc;
2702 @parents = ($lc);
2703 } else {
2704 $last_rev = $rev;
2705 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2706 return $log_entry;
2707 }
2708 $ed = SVN::Git::Fetcher->new($self);
2709 }
2710 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2711 die "SVN connection failed somewhere...\n";
2712 }
2713 $self->make_log_entry($rev, \@parents, $ed);
2714 }
2715
2716 sub get_untracked {
2717 my ($self, $ed) = @_;
2718 my @out;
2719 my $h = $ed->{empty};
2720 foreach (sort keys %$h) {
2721 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2722 push @out, " $act: " . uri_encode($_);
2723 warn "W: $act: $_\n";
2724 }
2725 foreach my $t (qw/dir_prop file_prop/) {
2726 $h = $ed->{$t} or next;
2727 foreach my $path (sort keys %$h) {
2728 my $ppath = $path eq '' ? '.' : $path;
2729 foreach my $prop (sort keys %{$h->{$path}}) {
2730 next if $SKIP_PROP{$prop};
2731 my $v = $h->{$path}->{$prop};
2732 my $t_ppath_prop = "$t: " .
2733 uri_encode($ppath) . ' ' .
2734 uri_encode($prop);
2735 if (defined $v) {
2736 push @out, " +$t_ppath_prop " .
2737 uri_encode($v);
2738 } else {
2739 push @out, " -$t_ppath_prop";
2740 }
2741 }
2742 }
2743 }
2744 foreach my $t (qw/absent_file absent_directory/) {
2745 $h = $ed->{$t} or next;
2746 foreach my $parent (sort keys %$h) {
2747 foreach my $path (sort @{$h->{$parent}}) {
2748 push @out, " $t: " .
2749 uri_encode("$parent/$path");
2750 warn "W: $t: $parent/$path ",
2751 "Insufficient permissions?\n";
2752 }
2753 }
2754 }
2755 \@out;
2756 }
2757
2758 # parse_svn_date(DATE)
2759 # --------------------
2760 # Given a date (in UTC) from Subversion, return a string in the format
2761 # "<TZ Offset> <local date/time>" that Git will use.
2762 #
2763 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
2764 # is true we'll convert it to the local timezone instead.
2765 sub parse_svn_date {
2766 my $date = shift || return '+0000 1970-01-01 00:00:00';
2767 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2768 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2769 croak "Unable to parse date: $date\n";
2770 my $parsed_date; # Set next.
2771
2772 if ($Git::SVN::_localtime) {
2773 # Translate the Subversion datetime to an epoch time.
2774 # Begin by switching ourselves to $date's timezone, UTC.
2775 my $old_env_TZ = $ENV{TZ};
2776 $ENV{TZ} = 'UTC';
2777
2778 my $epoch_in_UTC =
2779 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2780
2781 # Determine our local timezone (including DST) at the
2782 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2783 # value of TZ, if any, at the time we were run.
2784 if (defined $Git::SVN::Log::TZ) {
2785 $ENV{TZ} = $Git::SVN::Log::TZ;
2786 } else {
2787 delete $ENV{TZ};
2788 }
2789
2790 my $our_TZ =
2791 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2792
2793 # This converts $epoch_in_UTC into our local timezone.
2794 my ($sec, $min, $hour, $mday, $mon, $year,
2795 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2796
2797 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2798 $our_TZ, $year + 1900, $mon + 1,
2799 $mday, $hour, $min, $sec);
2800
2801 # Reset us to the timezone in effect when we entered
2802 # this routine.
2803 if (defined $old_env_TZ) {
2804 $ENV{TZ} = $old_env_TZ;
2805 } else {
2806 delete $ENV{TZ};
2807 }
2808 } else {
2809 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2810 }
2811
2812 return $parsed_date;
2813 }
2814
2815 sub other_gs {
2816 my ($self, $new_url, $url,
2817 $branch_from, $r, $old_ref_id) = @_;
2818 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2819 unless ($gs) {
2820 my $ref_id = $old_ref_id;
2821 $ref_id =~ s/\@\d+$//;
2822 $ref_id .= "\@$r";
2823 # just grow a tail if we're not unique enough :x
2824 $ref_id .= '-' while find_ref($ref_id);
2825 print STDERR "Initializing parent: $ref_id\n";
2826 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2827 if ($u =~ s#^\Q$url\E(/|$)##) {
2828 $p = $u;
2829 $u = $url;
2830 $repo_id = $self->{repo_id};
2831 }
2832 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2833 }
2834 $gs
2835 }
2836
2837 sub call_authors_prog {
2838 my ($orig_author) = @_;
2839 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2840 my $author = `$::_authors_prog $orig_author`;
2841 if ($? != 0) {
2842 die "$::_authors_prog failed with exit code $?\n"
2843 }
2844 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2845 my ($name, $email) = ($1, $2);
2846 $email = undef if length $2 == 0;
2847 return [$name, $email];
2848 } else {
2849 die "Author: $orig_author: $::_authors_prog returned "
2850 . "invalid author format: $author\n";
2851 }
2852 }
2853
2854 sub check_author {
2855 my ($author) = @_;
2856 if (!defined $author || length $author == 0) {
2857 $author = '(no author)';
2858 }
2859 if (!defined $::users{$author}) {
2860 if (defined $::_authors_prog) {
2861 $::users{$author} = call_authors_prog($author);
2862 } elsif (defined $::_authors) {
2863 die "Author: $author not defined in $::_authors file\n";
2864 }
2865 }
2866 $author;
2867 }
2868
2869 sub make_log_entry {
2870 my ($self, $rev, $parents, $ed) = @_;
2871 my $untracked = $self->get_untracked($ed);
2872
2873 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2874 print $un "r$rev\n" or croak $!;
2875 print $un $_, "\n" foreach @$untracked;
2876 my %log_entry = ( parents => $parents || [], revision => $rev,
2877 log => '');
2878
2879 my $headrev;
2880 my $logged = delete $self->{logged_rev_props};
2881 if (!$logged || $self->{-want_revprops}) {
2882 my $rp = $self->ra->rev_proplist($rev);
2883 foreach (sort keys %$rp) {
2884 my $v = $rp->{$_};
2885 if (/^svn:(author|date|log)$/) {
2886 $log_entry{$1} = $v;
2887 } elsif ($_ eq 'svm:headrev') {
2888 $headrev = $v;
2889 } else {
2890 print $un " rev_prop: ", uri_encode($_), ' ',
2891 uri_encode($v), "\n";
2892 }
2893 }
2894 } else {
2895 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2896 }
2897 close $un or croak $!;
2898
2899 $log_entry{date} = parse_svn_date($log_entry{date});
2900 $log_entry{log} .= "\n";
2901 my $author = $log_entry{author} = check_author($log_entry{author});
2902 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2903 : ($author, undef);
2904
2905 my ($commit_name, $commit_email) = ($name, $email);
2906 if ($_use_log_author) {
2907 my $name_field;
2908 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2909 $name_field = $1;
2910 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2911 $name_field = $1;
2912 }
2913 if (!defined $name_field) {
2914 if (!defined $email) {
2915 $email = $name;
2916 }
2917 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2918 ($name, $email) = ($1, $2);
2919 } elsif ($name_field =~ /(.*)@/) {
2920 ($name, $email) = ($1, $name_field);
2921 } else {
2922 ($name, $email) = ($name_field, $name_field);
2923 }
2924 }
2925 if (defined $headrev && $self->use_svm_props) {
2926 if ($self->rewrite_root) {
2927 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2928 "options set!\n";
2929 }
2930 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
2931 # we don't want "SVM: initializing mirror for junk" ...
2932 return undef if $r == 0;
2933 my $svm = $self->svm;
2934 if ($uuid ne $svm->{uuid}) {
2935 die "UUID mismatch on SVM path:\n",
2936 "expected: $svm->{uuid}\n",
2937 " got: $uuid\n";
2938 }
2939 my $full_url = $self->full_url;
2940 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2941 die "Failed to replace '$svm->{replace}' with ",
2942 "'$svm->{source}' in $full_url\n";
2943 # throw away username for storing in records
2944 remove_username($full_url);
2945 $log_entry{metadata} = "$full_url\@$r $uuid";
2946 $log_entry{svm_revision} = $r;
2947 $email ||= "$author\@$uuid";
2948 $commit_email ||= "$author\@$uuid";
2949 } elsif ($self->use_svnsync_props) {
2950 my $full_url = $self->svnsync->{url};
2951 $full_url .= "/$self->{path}" if length $self->{path};
2952 remove_username($full_url);
2953 my $uuid = $self->svnsync->{uuid};
2954 $log_entry{metadata} = "$full_url\@$rev $uuid";
2955 $email ||= "$author\@$uuid";
2956 $commit_email ||= "$author\@$uuid";
2957 } else {
2958 my $url = $self->metadata_url;
2959 remove_username($url);
2960 $log_entry{metadata} = "$url\@$rev " .
2961 $self->ra->get_uuid;
2962 $email ||= "$author\@" . $self->ra->get_uuid;
2963 $commit_email ||= "$author\@" . $self->ra->get_uuid;
2964 }
2965 $log_entry{name} = $name;
2966 $log_entry{email} = $email;
2967 $log_entry{commit_name} = $commit_name;
2968 $log_entry{commit_email} = $commit_email;
2969 \%log_entry;
2970 }
2971
2972 sub fetch {
2973 my ($self, $min_rev, $max_rev, @parents) = @_;
2974 my ($last_rev, $last_commit) = $self->last_rev_commit;
2975 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2976 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2977 }
2978
2979 sub set_tree_cb {
2980 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2981 $self->{inject_parents} = { $rev => $tree };
2982 $self->fetch(undef, undef);
2983 }
2984
2985 sub set_tree {
2986 my ($self, $tree) = (shift, shift);
2987 my $log_entry = ::get_commit_entry($tree);
2988 unless ($self->{last_rev}) {
2989 ::fatal("Must have an existing revision to commit");
2990 }
2991 my %ed_opts = ( r => $self->{last_rev},
2992 log => $log_entry->{log},
2993 ra => $self->ra,
2994 tree_a => $self->{last_commit},
2995 tree_b => $tree,
2996 editor_cb => sub {
2997 $self->set_tree_cb($log_entry, $tree, @_) },
2998 svn_path => $self->{path} );
2999 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3000 print "No changes\nr$self->{last_rev} = $tree\n";
3001 }
3002 }
3003
3004 sub rebuild_from_rev_db {
3005 my ($self, $path) = @_;
3006 my $r = -1;
3007 open my $fh, '<', $path or croak "open: $!";
3008 binmode $fh or croak "binmode: $!";
3009 while (<$fh>) {
3010 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3011 chomp($_);
3012 ++$r;
3013 next if $_ eq ('0' x 40);
3014 $self->rev_map_set($r, $_);
3015 print "r$r = $_\n";
3016 }
3017 close $fh or croak "close: $!";
3018 unlink $path or croak "unlink: $!";
3019 }
3020
3021 sub rebuild {
3022 my ($self) = @_;
3023 my $map_path = $self->map_path;
3024 my $partial = (-e $map_path && ! -z $map_path);
3025 return unless ::verify_ref($self->refname.'^0');
3026 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3027 my $rev_db = $self->rev_db_path;
3028 $self->rebuild_from_rev_db($rev_db);
3029 if ($self->use_svm_props) {
3030 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3031 $self->rebuild_from_rev_db($svm_rev_db);
3032 }
3033 $self->unlink_rev_db_symlink;
3034 return;
3035 }
3036 print "Rebuilding $map_path ...\n" if (!$partial);
3037 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3038 (undef, undef));
3039 my ($log, $ctx) =
3040 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3041 ($head ? "$head.." : "") . $self->refname,
3042 '--');
3043 my $metadata_url = $self->metadata_url;
3044 remove_username($metadata_url);
3045 my $svn_uuid = $self->ra_uuid;
3046 my $c;
3047 while (<$log>) {
3048 if ( m{^commit ($::sha1)$} ) {
3049 $c = $1;
3050 next;
3051 }
3052 next unless s{^\s*(git-svn-id:)}{$1};
3053 my ($url, $rev, $uuid) = ::extract_metadata($_);
3054 remove_username($url);
3055
3056 # ignore merges (from set-tree)
3057 next if (!defined $rev || !$uuid);
3058
3059 # if we merged or otherwise started elsewhere, this is
3060 # how we break out of it
3061 if (($uuid ne $svn_uuid) ||
3062 ($metadata_url && $url && ($url ne $metadata_url))) {
3063 next;
3064 }
3065 if ($partial && $head) {
3066 print "Partial-rebuilding $map_path ...\n";
3067 print "Currently at $base_rev = $head\n";
3068 $head = undef;
3069 }
3070
3071 $self->rev_map_set($rev, $c);
3072 print "r$rev = $c\n";
3073 }
3074 command_close_pipe($log, $ctx);
3075 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3076 my $rev_db_path = $self->rev_db_path;
3077 if (-f $self->rev_db_path) {
3078 unlink $self->rev_db_path or croak "unlink: $!";
3079 }
3080 $self->unlink_rev_db_symlink;
3081 }
3082
3083 # rev_map:
3084 # Tie::File seems to be prone to offset errors if revisions get sparse,
3085 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
3086 # one of my favorite modules is out :< Next up would be one of the DBM
3087 # modules, but I'm not sure which is most portable...
3088 #
3089 # This is the replacement for the rev_db format, which was too big
3090 # and inefficient for large repositories with a lot of sparse history
3091 # (mainly tags)
3092 #
3093 # The format is this:
3094 # - 24 bytes for every record,
3095 # * 4 bytes for the integer representing an SVN revision number
3096 # * 20 bytes representing the sha1 of a git commit
3097 # - No empty padding records like the old format
3098 # (except the last record, which can be overwritten)
3099 # - new records are written append-only since SVN revision numbers
3100 # increase monotonically
3101 # - lookups on SVN revision number are done via a binary search
3102 # - Piping the file to xxd -c24 is a good way of dumping it for
3103 # viewing or editing (piped back through xxd -r), should the need
3104 # ever arise.
3105 # - The last record can be padding revision with an all-zero sha1
3106 # This is used to optimize fetch performance when using multiple
3107 # "fetch" directives in .git/config
3108 #
3109 # These files are disposable unless noMetadata or useSvmProps is set
3110
3111 sub _rev_map_set {
3112 my ($fh, $rev, $commit) = @_;
3113
3114 binmode $fh or croak "binmode: $!";
3115 my $size = (stat($fh))[7];
3116 ($size % 24) == 0 or croak "inconsistent size: $size";
3117
3118 my $wr_offset = 0;
3119 if ($size > 0) {
3120 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3121 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3122 $read == 24 or croak "read only $read bytes (!= 24)";
3123 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3124 if ($last_commit eq ('0' x40)) {
3125 if ($size >= 48) {
3126 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3127 $read = sysread($fh, $buf, 24) or
3128 croak "read: $!";
3129 $read == 24 or
3130 croak "read only $read bytes (!= 24)";
3131 ($last_rev, $last_commit) =
3132 unpack(rev_map_fmt, $buf);
3133 if ($last_commit eq ('0' x40)) {
3134 croak "inconsistent .rev_map\n";
3135 }
3136 }
3137 if ($last_rev >= $rev) {
3138 croak "last_rev is higher!: $last_rev >= $rev";
3139 }
3140 $wr_offset = -24;
3141 }
3142 }
3143 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3144 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3145 croak "write: $!";
3146 }
3147
3148 sub _rev_map_reset {
3149 my ($fh, $rev, $commit) = @_;
3150 my $c = _rev_map_get($fh, $rev);
3151 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3152 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3153 truncate $fh, $offset or croak "truncate: $!";
3154 }
3155
3156 sub mkfile {
3157 my ($path) = @_;
3158 unless (-e $path) {
3159 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3160 mkpath([$dir]) unless -d $dir;
3161 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3162 close $fh or die "Couldn't close (create) $path: $!\n";
3163 }
3164 }
3165
3166 sub rev_map_set {
3167 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3168 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3169 my $db = $self->map_path($uuid);
3170 my $db_lock = "$db.lock";
3171 my $sig;
3172 $update_ref ||= 0;
3173 if ($update_ref) {
3174 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3175 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3176 }
3177 mkfile($db);
3178
3179 $LOCKFILES{$db_lock} = 1;
3180 my $sync;
3181 # both of these options make our .rev_db file very, very important
3182 # and we can't afford to lose it because rebuild() won't work
3183 if ($self->use_svm_props || $self->no_metadata) {
3184 $sync = 1;
3185 copy($db, $db_lock) or die "rev_map_set(@_): ",
3186 "Failed to copy: ",
3187 "$db => $db_lock ($!)\n";
3188 } else {
3189 rename $db, $db_lock or die "rev_map_set(@_): ",
3190 "Failed to rename: ",
3191 "$db => $db_lock ($!)\n";
3192 }
3193
3194 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3195 or croak "Couldn't open $db_lock: $!\n";
3196 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3197 _rev_map_set($fh, $rev, $commit);
3198 if ($sync) {
3199 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3200 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3201 }
3202 close $fh or croak $!;
3203 if ($update_ref) {
3204 $_head = $self;
3205 my $note = "";
3206 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3207 command_noisy('update-ref', '-m', "r$rev$note",
3208 $self->refname, $commit);
3209 }
3210 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3211 "$db_lock => $db ($!)\n";
3212 delete $LOCKFILES{$db_lock};
3213 if ($update_ref) {
3214 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3215 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3216 kill $sig, $$ if defined $sig;
3217 }
3218 }
3219
3220 # If want_commit, this will return an array of (rev, commit) where
3221 # commit _must_ be a valid commit in the archive.
3222 # Otherwise, it'll return the max revision (whether or not the
3223 # commit is valid or just a 0x40 placeholder).
3224 sub rev_map_max {
3225 my ($self, $want_commit) = @_;
3226 $self->rebuild;
3227 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3228 $want_commit ? ($r, $c) : $r;
3229 }
3230
3231 sub rev_map_max_norebuild {
3232 my ($self, $want_commit) = @_;
3233 my $map_path = $self->map_path;
3234 stat $map_path or return $want_commit ? (0, undef) : 0;
3235 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3236 binmode $fh or croak "binmode: $!";
3237 my $size = (stat($fh))[7];
3238 ($size % 24) == 0 or croak "inconsistent size: $size";
3239
3240 if ($size == 0) {
3241 close $fh or croak "close: $!";
3242 return $want_commit ? (0, undef) : 0;
3243 }
3244
3245 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3246 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3247 my ($r, $c) = unpack(rev_map_fmt, $buf);
3248 if ($want_commit && $c eq ('0' x40)) {
3249 if ($size < 48) {
3250 return $want_commit ? (0, undef) : 0;
3251 }
3252 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3253 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3254 ($r, $c) = unpack(rev_map_fmt, $buf);
3255 if ($c eq ('0'x40)) {
3256 croak "Penultimate record is all-zeroes in $map_path";
3257 }
3258 }
3259 close $fh or croak "close: $!";
3260 $want_commit ? ($r, $c) : $r;
3261 }
3262
3263 sub rev_map_get {
3264 my ($self, $rev, $uuid) = @_;
3265 my $map_path = $self->map_path($uuid);
3266 return undef unless -e $map_path;
3267
3268 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3269 my $c = _rev_map_get($fh, $rev);
3270 close($fh) or croak "close: $!";
3271 $c
3272 }
3273
3274 sub _rev_map_get {
3275 my ($fh, $rev) = @_;
3276
3277 binmode $fh or croak "binmode: $!";
3278 my $size = (stat($fh))[7];
3279 ($size % 24) == 0 or croak "inconsistent size: $size";
3280
3281 if ($size == 0) {
3282 return undef;
3283 }
3284
3285 my ($l, $u) = (0, $size - 24);
3286 my ($r, $c, $buf);
3287
3288 while ($l <= $u) {
3289 my $i = int(($l/24 + $u/24) / 2) * 24;
3290 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3291 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3292 my ($r, $c) = unpack(rev_map_fmt, $buf);
3293
3294 if ($r < $rev) {
3295 $l = $i + 24;
3296 } elsif ($r > $rev) {
3297 $u = $i - 24;
3298 } else { # $r == $rev
3299 return $c eq ('0' x 40) ? undef : $c;
3300 }
3301 }
3302 undef;
3303 }
3304
3305 # Finds the first svn revision that exists on (if $eq_ok is true) or
3306 # before $rev for the current branch. It will not search any lower
3307 # than $min_rev. Returns the git commit hash and svn revision number
3308 # if found, else (undef, undef).
3309 sub find_rev_before {
3310 my ($self, $rev, $eq_ok, $min_rev) = @_;
3311 --$rev unless $eq_ok;
3312 $min_rev ||= 1;
3313 my $max_rev = $self->rev_map_max;
3314