?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Revert "git-svn: always initialize with core.autocrlf=false"
[alt-git.git] / git-svn.perl
blob7bca381c4bc5faf2a605546a0ceda6599036f4cb
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use warnings;
5 use strict;
6 use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision $_repository
8 $_q $_authors $_authors_prog %users/;
9 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10 $VERSION = '@@GIT_VERSION@@';
11
12 # From which subdir have we been invoked?
13 my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15 } || '';
16
17 my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
18 $ENV{GIT_DIR} ||= '.git';
19 $Git::SVN::default_repo_id = 'svn';
20 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
21 $Git::SVN::Ra::_log_window_size = 100;
22 $Git::SVN::_minimize_url = 'unset';
23
24 if (! exists $ENV{SVN_SSH}) {
25 if (exists $ENV{GIT_SSH}) {
26 $ENV{SVN_SSH} = $ENV{GIT_SSH};
27 if ($^O eq 'msys') {
28 $ENV{SVN_SSH} =~ s/\\/\\\\/g;
29 $ENV{SVN_SSH} =~ s/(.*)/"$1"/;
30 }
31 }
32 }
33
34 $Git::SVN::Log::TZ = $ENV{TZ};
35 $ENV{TZ} = 'UTC';
36 $| = 1; # unbuffer STDOUT
37
38 sub fatal (@) { print STDERR "@_\n"; exit 1 }
39 sub _req_svn {
40 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
41 require SVN::Ra;
42 require SVN::Delta;
43 if ($SVN::Core::VERSION lt '1.1.0') {
44 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
45 }
46 }
47 my $can_compress = eval { require Compress::Zlib; 1};
48 push @Git::SVN::Ra::ISA, 'SVN::Ra';
49 push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
50 push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
51 use Carp qw/croak/;
52 use Digest::MD5;
53 use IO::File qw//;
54 use File::Basename qw/dirname basename/;
55 use File::Path qw/mkpath/;
56 use File::Spec;
57 use File::Find;
58 use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
59 use IPC::Open3;
60 use Git;
61
62 BEGIN {
63 # import functions from Git into our packages, en masse
64 no strict 'refs';
65 foreach (qw/command command_oneline command_noisy command_output_pipe
66 command_input_pipe command_close_pipe
67 command_bidi_pipe command_close_bidi_pipe/) {
68 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
69 Git::SVN::Migration Git::SVN::Log Git::SVN),
70 __PACKAGE__) {
71 *{"${package}::$_"} = \&{"Git::$_"};
72 }
73 }
74 }
75
76 my ($SVN);
77
78 $sha1 = qr/[a-f\d]{40}/;
79 $sha1_short = qr/[a-f\d]{4,40}/;
80 my ($_stdin, $_help, $_edit,
81 $_message, $_file, $_branch_dest,
82 $_template, $_shared,
83 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
84 $_merge, $_strategy, $_dry_run, $_local,
85 $_prefix, $_no_checkout, $_url, $_verbose,
86 $_git_format, $_commit_url, $_tag);
87 $Git::SVN::_follow_parent = 1;
88 $_q ||= 0;
89 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
90 'config-dir=s' => \$Git::SVN::Ra::config_dir,
91 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
92 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
93 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
94 'authors-file|A=s' => \$_authors,
95 'authors-prog=s' => \$_authors_prog,
96 'repack:i' => \$Git::SVN::_repack,
97 'noMetadata' => \$Git::SVN::_no_metadata,
98 'useSvmProps' => \$Git::SVN::_use_svm_props,
99 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
100 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
101 'no-checkout' => \$_no_checkout,
102 'quiet|q+' => \$_q,
103 'repack-flags|repack-args|repack-opts=s' =>
104 \$Git::SVN::_repack_flags,
105 'use-log-author' => \$Git::SVN::_use_log_author,
106 'add-author-from' => \$Git::SVN::_add_author_from,
107 'localtime' => \$Git::SVN::_localtime,
108 %remote_opts );
109
110 my ($_trunk, @_tags, @_branches, $_stdlayout);
111 my %icv;
112 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
113 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
114 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
115 'stdlayout|s' => \$_stdlayout,
116 'minimize-url|m!' => \$Git::SVN::_minimize_url,
117 'no-metadata' => sub { $icv{noMetadata} = 1 },
118 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
119 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
120 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
121 'rewrite-uuid=s' => sub { $icv{rewriteUUID} = $_[1] },
122 %remote_opts );
123 my %cmt_opts = ( 'edit|e' => \$_edit,
124 'rmdir' => \$SVN::Git::Editor::_rmdir,
125 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
126 'l=i' => \$SVN::Git::Editor::_rename_limit,
127 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
128 );
129
130 my %cmd = (
131 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
132 { 'revision|r=s' => \$_revision,
133 'fetch-all|all' => \$_fetch_all,
134 'parent|p' => \$_fetch_parent,
135 %fc_opts } ],
136 clone => [ \&cmd_clone, "Initialize and fetch revisions",
137 { 'revision|r=s' => \$_revision,
138 %fc_opts, %init_opts } ],
139 init => [ \&cmd_init, "Initialize a repo for tracking" .
140 " (requires URL argument)",
141 \%init_opts ],
142 'multi-init' => [ \&cmd_multi_init,
143 "Deprecated alias for ".
144 "'$0 init -T<trunk> -b<branches> -t<tags>'",
145 \%init_opts ],
146 dcommit => [ \&cmd_dcommit,
147 'Commit several diffs to merge with upstream',
148 { 'merge|m|M' => \$_merge,
149 'strategy|s=s' => \$_strategy,
150 'verbose|v' => \$_verbose,
151 'dry-run|n' => \$_dry_run,
152 'fetch-all|all' => \$_fetch_all,
153 'commit-url=s' => \$_commit_url,
154 'revision|r=i' => \$_revision,
155 'no-rebase' => \$_no_rebase,
156 %cmt_opts, %fc_opts } ],
157 branch => [ \&cmd_branch,
158 'Create a branch in the SVN repository',
159 { 'message|m=s' => \$_message,
160 'destination|d=s' => \$_branch_dest,
161 'dry-run|n' => \$_dry_run,
162 'tag|t' => \$_tag,
163 'username=s' => \$Git::SVN::Prompt::_username,
164 'commit-url=s' => \$_commit_url } ],
165 tag => [ sub { $_tag = 1; cmd_branch(@_) },
166 'Create a tag in the SVN repository',
167 { 'message|m=s' => \$_message,
168 'destination|d=s' => \$_branch_dest,
169 'dry-run|n' => \$_dry_run,
170 'username=s' => \$Git::SVN::Prompt::_username,
171 'commit-url=s' => \$_commit_url } ],
172 'set-tree' => [ \&cmd_set_tree,
173 "Set an SVN repository to a git tree-ish",
174 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
175 'create-ignore' => [ \&cmd_create_ignore,
176 'Create a .gitignore per svn:ignore',
177 { 'revision|r=i' => \$_revision
178 } ],
179 'mkdirs' => [ \&cmd_mkdirs ,
180 "recreate empty directories after a checkout",
181 { 'revision|r=i' => \$_revision } ],
182 'propget' => [ \&cmd_propget,
183 'Print the value of a property on a file or directory',
184 { 'revision|r=i' => \$_revision } ],
185 'proplist' => [ \&cmd_proplist,
186 'List all properties of a file or directory',
187 { 'revision|r=i' => \$_revision } ],
188 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
189 { 'revision|r=i' => \$_revision
190 } ],
191 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
192 { 'revision|r=i' => \$_revision
193 } ],
194 'multi-fetch' => [ \&cmd_multi_fetch,
195 "Deprecated alias for $0 fetch --all",
196 { 'revision|r=s' => \$_revision, %fc_opts } ],
197 'migrate' => [ sub { },
198 # no-op, we automatically run this anyways,
199 'Migrate configuration/metadata/layout from
200 previous versions of git-svn',
201 { 'minimize' => \$Git::SVN::Migration::_minimize,
202 %remote_opts } ],
203 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
204 { 'limit=i' => \$Git::SVN::Log::limit,
205 'revision|r=s' => \$_revision,
206 'verbose|v' => \$Git::SVN::Log::verbose,
207 'incremental' => \$Git::SVN::Log::incremental,
208 'oneline' => \$Git::SVN::Log::oneline,
209 'show-commit' => \$Git::SVN::Log::show_commit,
210 'non-recursive' => \$Git::SVN::Log::non_recursive,
211 'authors-file|A=s' => \$_authors,
212 'color' => \$Git::SVN::Log::color,
213 'pager=s' => \$Git::SVN::Log::pager
214 } ],
215 'find-rev' => [ \&cmd_find_rev,
216 "Translate between SVN revision numbers and tree-ish",
217 {} ],
218 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
219 { 'merge|m|M' => \$_merge,
220 'verbose|v' => \$_verbose,
221 'strategy|s=s' => \$_strategy,
222 'local|l' => \$_local,
223 'fetch-all|all' => \$_fetch_all,
224 'dry-run|n' => \$_dry_run,
225 %fc_opts } ],
226 'commit-diff' => [ \&cmd_commit_diff,
227 'Commit a diff between two trees',
228 { 'message|m=s' => \$_message,
229 'file|F=s' => \$_file,
230 'revision|r=s' => \$_revision,
231 %cmt_opts } ],
232 'info' => [ \&cmd_info,
233 "Show info about the latest SVN revision
234 on the current branch",
235 { 'url' => \$_url, } ],
236 'blame' => [ \&Git::SVN::Log::cmd_blame,
237 "Show what revision and author last modified each line of a file",
238 { 'git-format' => \$_git_format } ],
239 'reset' => [ \&cmd_reset,
240 "Undo fetches back to the specified SVN revision",
241 { 'revision|r=s' => \$_revision,
242 'parent|p' => \$_fetch_parent } ],
243 'gc' => [ \&cmd_gc,
244 "Compress unhandled.log files in .git/svn and remove " .
245 "index files in .git/svn",
246 {} ],
247 );
248
249 my $cmd;
250 for (my $i = 0; $i < @ARGV; $i++) {
251 if (defined $cmd{$ARGV[$i]}) {
252 $cmd = $ARGV[$i];
253 splice @ARGV, $i, 1;
254 last;
255 } elsif ($ARGV[$i] eq 'help') {
256 $cmd = $ARGV[$i+1];
257 usage(0);
258 }
259 };
260
261 # make sure we're always running at the top-level working directory
262 unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
263 unless (-d $ENV{GIT_DIR}) {
264 if ($git_dir_user_set) {
265 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
266 "but it is not a directory\n";
267 }
268 my $git_dir = delete $ENV{GIT_DIR};
269 my $cdup = undef;
270 git_cmd_try {
271 $cdup = command_oneline(qw/rev-parse --show-cdup/);
272 $git_dir = '.' unless ($cdup);
273 chomp $cdup if ($cdup);
274 $cdup = "." unless ($cdup && length $cdup);
275 } "Already at toplevel, but $git_dir not found\n";
276 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
277 unless (-d $git_dir) {
278 die "$git_dir still not found after going to ",
279 "'$cdup'\n";
280 }
281 $ENV{GIT_DIR} = $git_dir;
282 }
283 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
284 }
285
286 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
287
288 read_git_config(\%opts);
289 if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
290 Getopt::Long::Configure('pass_through');
291 }
292 my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
293 'minimize-connections' => \$Git::SVN::Migration::_minimize,
294 'id|i=s' => \$Git::SVN::default_ref_id,
295 'svn-remote|remote|R=s' => sub {
296 $Git::SVN::no_reuse_existing = 1;
297 $Git::SVN::default_repo_id = $_[1] });
298 exit 1 if (!$rv && $cmd && $cmd ne 'log');
299
300 usage(0) if $_help;
301 version() if $_version;
302 usage(1) unless defined $cmd;
303 load_authors() if $_authors;
304 if (defined $_authors_prog) {
305 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
306 }
307
308 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
309 Git::SVN::Migration::migration_check();
310 }
311 Git::SVN::init_vars();
312 eval {
313 Git::SVN::verify_remotes_sanity();
314 $cmd{$cmd}->[0]->(@ARGV);
315 };
316 fatal $@ if $@;
317 post_fetch_checkout();
318 exit 0;
319
320 ####################### primary functions ######################
321 sub usage {
322 my $exit = shift || 0;
323 my $fd = $exit ? \*STDERR : \*STDOUT;
324 print $fd <<"";
325 git-svn - bidirectional operations between a single Subversion tree and git
326 Usage: git svn <command> [options] [arguments]\n
327
328 print $fd "Available commands:\n" unless $cmd;
329
330 foreach (sort keys %cmd) {
331 next if $cmd && $cmd ne $_;
332 next if /^multi-/; # don't show deprecated commands
333 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
334 foreach (sort keys %{$cmd{$_}->[2]}) {
335 # mixed-case options are for .git/config only
336 next if /[A-Z]/ && /^[a-z]+$/i;
337 # prints out arguments as they should be passed:
338 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
339 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
340 "--$_" : "-$_" }
341 split /\|/,$_)," $x\n";
342 }
343 }
344 print $fd <<"";
345 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
346 arbitrary identifier if you're tracking multiple SVN branches/repositories in
347 one git repository and want to keep them separate. See git-svn(1) for more
348 information.
349
350 exit $exit;
351 }
352
353 sub version {
354 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
355 exit 0;
356 }
357
358 sub do_git_init_db {
359 unless (-d $ENV{GIT_DIR}) {
360 my @init_db = ('init');
361 push @init_db, "--template=$_template" if defined $_template;
362 if (defined $_shared) {
363 if ($_shared =~ /[a-z]/) {
364 push @init_db, "--shared=$_shared";
365 } else {
366 push @init_db, "--shared";
367 }
368 }
369 command_noisy(@init_db);
370 $_repository = Git->repository(Repository => ".git");
371 }
372 my $set;
373 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
374 foreach my $i (keys %icv) {
375 die "'$set' and '$i' cannot both be set\n" if $set;
376 next unless defined $icv{$i};
377 command_noisy('config', "$pfx.$i", $icv{$i});
378 $set = $i;
379 }
380 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
381 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
382 if defined $$ignore_regex;
383 }
384
385 sub init_subdir {
386 my $repo_path = shift or return;
387 mkpath([$repo_path]) unless -d $repo_path;
388 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
389 $ENV{GIT_DIR} = '.git';
390 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
391 }
392
393 sub cmd_clone {
394 my ($url, $path) = @_;
395 if (!defined $path &&
396 (defined $_trunk || @_branches || @_tags ||
397 defined $_stdlayout) &&
398 $url !~ m#^[a-z\+]+://#) {
399 $path = $url;
400 }
401 $path = basename($url) if !defined $path || !length $path;
402 my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
403 cmd_init($url, $path);
404 command_oneline('config', 'svn.authorsfile', $authors_absolute)
405 if $_authors;
406 Git::SVN::fetch_all($Git::SVN::default_repo_id);
407 }
408
409 sub cmd_init {
410 if (defined $_stdlayout) {
411 $_trunk = 'trunk' if (!defined $_trunk);
412 @_tags = 'tags' if (! @_tags);
413 @_branches = 'branches' if (! @_branches);
414 }
415 if (defined $_trunk || @_branches || @_tags) {
416 return cmd_multi_init(@_);
417 }
418 my $url = shift or die "SVN repository location required ",
419 "as a command-line argument\n";
420 $url = canonicalize_url($url);
421 init_subdir(@_);
422 do_git_init_db();
423
424 if ($Git::SVN::_minimize_url eq 'unset') {
425 $Git::SVN::_minimize_url = 0;
426 }
427
428 Git::SVN->init($url);
429 }
430
431 sub cmd_fetch {
432 if (grep /^\d+=./, @_) {
433 die "'<rev>=<commit>' fetch arguments are ",
434 "no longer supported.\n";
435 }
436 my ($remote) = @_;
437 if (@_ > 1) {
438 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
439 }
440 $Git::SVN::no_reuse_existing = undef;
441 if ($_fetch_parent) {
442 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
443 unless ($gs) {
444 die "Unable to determine upstream SVN information from ",
445 "working tree history\n";
446 }
447 # just fetch, don't checkout.
448 $_no_checkout = 'true';
449 $_fetch_all ? $gs->fetch_all : $gs->fetch;
450 } elsif ($_fetch_all) {
451 cmd_multi_fetch();
452 } else {
453 $remote ||= $Git::SVN::default_repo_id;
454 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
455 }
456 }
457
458 sub cmd_set_tree {
459 my (@commits) = @_;
460 if ($_stdin || !@commits) {
461 print "Reading from stdin...\n";
462 @commits = ();
463 while (<STDIN>) {
464 if (/\b($sha1_short)\b/o) {
465 unshift @commits, $1;
466 }
467 }
468 }
469 my @revs;
470 foreach my $c (@commits) {
471 my @tmp = command('rev-parse',$c);
472 if (scalar @tmp == 1) {
473 push @revs, $tmp[0];
474 } elsif (scalar @tmp > 1) {
475 push @revs, reverse(command('rev-list',@tmp));
476 } else {
477 fatal "Failed to rev-parse $c";
478 }
479 }
480 my $gs = Git::SVN->new;
481 my ($r_last, $cmt_last) = $gs->last_rev_commit;
482 $gs->fetch;
483 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
484 fatal "There are new revisions that were fetched ",
485 "and need to be merged (or acknowledged) ",
486 "before committing.\nlast rev: $r_last\n",
487 " current: $gs->{last_rev}";
488 }
489 $gs->set_tree($_) foreach @revs;
490 print "Done committing ",scalar @revs," revisions to SVN\n";
491 unlink $gs->{index};
492 }
493
494 sub cmd_dcommit {
495 my $head = shift;
496 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
497 'Cannot dcommit with a dirty index. Commit your changes first, '
498 . "or stash them with `git stash'.\n";
499 $head ||= 'HEAD';
500
501 my $old_head;
502 if ($head ne 'HEAD') {
503 $old_head = eval {
504 command_oneline([qw/symbolic-ref -q HEAD/])
505 };
506 if ($old_head) {
507 $old_head =~ s{^refs/heads/}{};
508 } else {
509 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
510 }
511 command(['checkout', $head], STDERR => 0);
512 }
513
514 my @refs;
515 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
516 unless ($gs) {
517 die "Unable to determine upstream SVN information from ",
518 "$head history.\nPerhaps the repository is empty.";
519 }
520
521 if (defined $_commit_url) {
522 $url = $_commit_url;
523 } else {
524 $url = eval { command_oneline('config', '--get',
525 "svn-remote.$gs->{repo_id}.commiturl") };
526 if (!$url) {
527 $url = $gs->full_url
528 }
529 }
530
531 my $last_rev = $_revision if defined $_revision;
532 if ($url) {
533 print "Committing to $url ...\n";
534 }
535 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
536 if ($_no_rebase && scalar(@$linear_refs) > 1) {
537 warn "Attempting to commit more than one change while ",
538 "--no-rebase is enabled.\n",
539 "If these changes depend on each other, re-running ",
540 "without --no-rebase may be required."
541 }
542 my $expect_url = $url;
543 Git::SVN::remove_username($expect_url);
544 while (1) {
545 my $d = shift @$linear_refs or last;
546 unless (defined $last_rev) {
547 (undef, $last_rev, undef) = cmt_metadata("$d~1");
548 unless (defined $last_rev) {
549 fatal "Unable to extract revision information ",
550 "from commit $d~1";
551 }
552 }
553 if ($_dry_run) {
554 print "diff-tree $d~1 $d\n";
555 } else {
556 my $cmt_rev;
557 my %ed_opts = ( r => $last_rev,
558 log => get_commit_entry($d)->{log},
559 ra => Git::SVN::Ra->new($url),
560 config => SVN::Core::config_get_config(
561 $Git::SVN::Ra::config_dir
562 ),
563 tree_a => "$d~1",
564 tree_b => $d,
565 editor_cb => sub {
566 print "Committed r$_[0]\n";
567 $cmt_rev = $_[0];
568 },
569 svn_path => '');
570 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
571 print "No changes\n$d~1 == $d\n";
572 } elsif ($parents->{$d} && @{$parents->{$d}}) {
573 $gs->{inject_parents_dcommit}->{$cmt_rev} =
574 $parents->{$d};
575 }
576 $_fetch_all ? $gs->fetch_all : $gs->fetch;
577 $last_rev = $cmt_rev;
578 next if $_no_rebase;
579
580 # we always want to rebase against the current HEAD,
581 # not any head that was passed to us
582 my @diff = command('diff-tree', $d,
583 $gs->refname, '--');
584 my @finish;
585 if (@diff) {
586 @finish = rebase_cmd();
587 print STDERR "W: $d and ", $gs->refname,
588 " differ, using @finish:\n",
589 join("\n", @diff), "\n";
590 } else {
591 print "No changes between current HEAD and ",
592 $gs->refname,
593 "\nResetting to the latest ",
594 $gs->refname, "\n";
595 @finish = qw/reset --mixed/;
596 }
597 command_noisy(@finish, $gs->refname);
598 if (@diff) {
599 @refs = ();
600 my ($url_, $rev_, $uuid_, $gs_) =
601 working_head_info('HEAD', \@refs);
602 my ($linear_refs_, $parents_) =
603 linearize_history($gs_, \@refs);
604 if (scalar(@$linear_refs) !=
605 scalar(@$linear_refs_)) {
606 fatal "# of revisions changed ",
607 "\nbefore:\n",
608 join("\n", @$linear_refs),
609 "\n\nafter:\n",
610 join("\n", @$linear_refs_), "\n",
611 'If you are attempting to commit ',
612 "merges, try running:\n\t",
613 'git rebase --interactive',
614 '--preserve-merges ',
615 $gs->refname,
616 "\nBefore dcommitting";
617 }
618 if ($url_ ne $expect_url) {
619 if ($url_ eq $gs->metadata_url) {
620 print
621 "Accepting rewritten URL:",
622 " $url_\n";
623 } else {
624 fatal
625 "URL mismatch after rebase:",
626 " $url_ != $expect_url";
627 }
628 }
629 if ($uuid_ ne $uuid) {
630 fatal "uuid mismatch after rebase: ",
631 "$uuid_ != $uuid";
632 }
633 # remap parents
634 my (%p, @l, $i);
635 for ($i = 0; $i < scalar @$linear_refs; $i++) {
636 my $new = $linear_refs_->[$i] or next;
637 $p{$new} =
638 $parents->{$linear_refs->[$i]};
639 push @l, $new;
640 }
641 $parents = \%p;
642 $linear_refs = \@l;
643 }
644 }
645 }
646
647 if ($old_head) {
648 my $new_head = command_oneline(qw/rev-parse HEAD/);
649 my $new_is_symbolic = eval {
650 command_oneline(qw/symbolic-ref -q HEAD/);
651 };
652 if ($new_is_symbolic) {
653 print "dcommitted the branch ", $head, "\n";
654 } else {
655 print "dcommitted on a detached HEAD because you gave ",
656 "a revision argument.\n",
657 "The rewritten commit is: ", $new_head, "\n";
658 }
659 command(['checkout', $old_head], STDERR => 0);
660 }
661
662 unlink $gs->{index};
663 }
664
665 sub cmd_branch {
666 my ($branch_name, $head) = @_;
667
668 unless (defined $branch_name && length $branch_name) {
669 die(($_tag ? "tag" : "branch") . " name required\n");
670 }
671 $head ||= 'HEAD';
672
673 my (undef, $rev, undef, $gs) = working_head_info($head);
674 my $src = $gs->full_url;
675
676 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
677 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
678 my $glob;
679 if ($#{$allglobs} == 0) {
680 $glob = $allglobs->[0];
681 } else {
682 unless(defined $_branch_dest) {
683 die "Multiple ",
684 $_tag ? "tag" : "branch",
685 " paths defined for Subversion repository.\n",
686 "You must specify where you want to create the ",
687 $_tag ? "tag" : "branch",
688 " with the --destination argument.\n";
689 }
690 foreach my $g (@{$allglobs}) {
691 # SVN::Git::Editor could probably be moved to Git.pm..
692 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
693 if ($_branch_dest =~ /$re/) {
694 $glob = $g;
695 last;
696 }
697 }
698 unless (defined $glob) {
699 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
700 foreach my $g (@{$allglobs}) {
701 $g->{path}->{left} =~ /$dest_re/ or next;
702 if (defined $glob) {
703 die "Ambiguous destination: ",
704 $_branch_dest, "\nmatches both '",
705 $glob->{path}->{left}, "' and '",
706 $g->{path}->{left}, "'\n";
707 }
708 $glob = $g;
709 }
710 unless (defined $glob) {
711 die "Unknown ",
712 $_tag ? "tag" : "branch",
713 " destination $_branch_dest\n";
714 }
715 }
716 }
717 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
718 my $url;
719 if (defined $_commit_url) {
720 $url = $_commit_url;
721 } else {
722 $url = eval { command_oneline('config', '--get',
723 "svn-remote.$gs->{repo_id}.commiturl") };
724 if (!$url) {
725 $url = $remote->{url};
726 }
727 }
728 my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
729
730 if ($dst =~ /^https:/ && $src =~ /^http:/) {
731 $src=~s/^http:/https:/;
732 }
733
734 ::_req_svn();
735
736 my $ctx = SVN::Client->new(
737 auth => Git::SVN::Ra::_auth_providers(),
738 log_msg => sub {
739 ${ $_[0] } = defined $_message
740 ? $_message
741 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
742 . $branch_name;
743 },
744 );
745
746 eval {
747 $ctx->ls($dst, 'HEAD', 0);
748 } and die "branch ${branch_name} already exists\n";
749
750 print "Copying ${src} at r${rev} to ${dst}...\n";
751 $ctx->copy($src, $rev, $dst)
752 unless $_dry_run;
753
754 $gs->fetch_all;
755 }
756
757 sub cmd_find_rev {
758 my $revision_or_hash = shift or die "SVN or git revision required ",
759 "as a command-line argument\n";
760 my $result;
761 if ($revision_or_hash =~ /^r\d+$/) {
762 my $head = shift;
763 $head ||= 'HEAD';
764 my @refs;
765 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
766 unless ($gs) {
767 die "Unable to determine upstream SVN information from ",
768 "$head history\n";
769 }
770 my $desired_revision = substr($revision_or_hash, 1);
771 $result = $gs->rev_map_get($desired_revision, $uuid);
772 } else {
773 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
774 $result = $rev;
775 }
776 print "$result\n" if $result;
777 }
778
779 sub cmd_rebase {
780 command_noisy(qw/update-index --refresh/);
781 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
782 unless ($gs) {
783 die "Unable to determine upstream SVN information from ",
784 "working tree history\n";
785 }
786 if ($_dry_run) {
787 print "Remote Branch: " . $gs->refname . "\n";
788 print "SVN URL: " . $url . "\n";
789 return;
790 }
791 if (command(qw/diff-index HEAD --/)) {
792 print STDERR "Cannot rebase with uncommited changes:\n";
793 command_noisy('status');
794 exit 1;
795 }
796 unless ($_local) {
797 # rebase will checkout for us, so no need to do it explicitly
798 $_no_checkout = 'true';
799 $_fetch_all ? $gs->fetch_all : $gs->fetch;
800 }
801 command_noisy(rebase_cmd(), $gs->refname);
802 $gs->mkemptydirs;
803 }
804
805 sub cmd_show_ignore {
806 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
807 $gs ||= Git::SVN->new;
808 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
809 $gs->prop_walk($gs->{path}, $r, sub {
810 my ($gs, $path, $props) = @_;
811 print STDOUT "\n# $path\n";
812 my $s = $props->{'svn:ignore'} or return;
813 $s =~ s/[\r\n]+/\n/g;
814 $s =~ s/^\n+//;
815 chomp $s;
816 $s =~ s#^#$path#gm;
817 print STDOUT "$s\n";
818 });
819 }
820
821 sub cmd_show_externals {
822 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
823 $gs ||= Git::SVN->new;
824 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
825 $gs->prop_walk($gs->{path}, $r, sub {
826 my ($gs, $path, $props) = @_;
827 print STDOUT "\n# $path\n";
828 my $s = $props->{'svn:externals'} or return;
829 $s =~ s/[\r\n]+/\n/g;
830 chomp $s;
831 $s =~ s#^#$path#gm;
832 print STDOUT "$s\n";
833 });
834 }
835
836 sub cmd_create_ignore {
837 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
838 $gs ||= Git::SVN->new;
839 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
840 $gs->prop_walk($gs->{path}, $r, sub {
841 my ($gs, $path, $props) = @_;
842 # $path is of the form /path/to/dir/
843 $path = '.' . $path;
844 # SVN can have attributes on empty directories,
845 # which git won't track
846 mkpath([$path]) unless -d $path;
847 my $ignore = $path . '.gitignore';
848 my $s = $props->{'svn:ignore'} or return;
849 open(GITIGNORE, '>', $ignore)
850 or fatal("Failed to open `$ignore' for writing: $!");
851 $s =~ s/[\r\n]+/\n/g;
852 $s =~ s/^\n+//;
853 chomp $s;
854 # Prefix all patterns so that the ignore doesn't apply
855 # to sub-directories.
856 $s =~ s#^#/#gm;
857 print GITIGNORE "$s\n";
858 close(GITIGNORE)
859 or fatal("Failed to close `$ignore': $!");
860 command_noisy('add', '-f', $ignore);
861 });
862 }
863
864 sub cmd_mkdirs {
865 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
866 $gs ||= Git::SVN->new;
867 $gs->mkemptydirs($_revision);
868 }
869
870 sub canonicalize_path {
871 my ($path) = @_;
872 my $dot_slash_added = 0;
873 if (substr($path, 0, 1) ne "/") {
874 $path = "./" . $path;
875 $dot_slash_added = 1;
876 }
877 # File::Spec->canonpath doesn't collapse x/../y into y (for a
878 # good reason), so let's do this manually.
879 $path =~ s#/+#/#g;
880 $path =~ s#/\.(?:/|$)#/#g;
881 $path =~ s#/[^/]+/\.\.##g;
882 $path =~ s#/$##g;
883 $path =~ s#^\./## if $dot_slash_added;
884 $path =~ s#^/##;
885 $path =~ s#^\.$##;
886 return $path;
887 }
888
889 sub canonicalize_url {
890 my ($url) = @_;
891 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
892 return $url;
893 }
894
895 # get_svnprops(PATH)
896 # ------------------
897 # Helper for cmd_propget and cmd_proplist below.
898 sub get_svnprops {
899 my $path = shift;
900 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
901 $gs ||= Git::SVN->new;
902
903 # prefix THE PATH by the sub-directory from which the user
904 # invoked us.
905 $path = $cmd_dir_prefix . $path;
906 fatal("No such file or directory: $path") unless -e $path;
907 my $is_dir = -d $path ? 1 : 0;
908 $path = $gs->{path} . '/' . $path;
909
910 # canonicalize the path (otherwise libsvn will abort or fail to
911 # find the file)
912 $path = canonicalize_path($path);
913
914 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
915 my $props;
916 if ($is_dir) {
917 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
918 }
919 else {
920 (undef, $props) = $gs->ra->get_file($path, $r, undef);
921 }
922 return $props;
923 }
924
925 # cmd_propget (PROP, PATH)
926 # ------------------------
927 # Print the SVN property PROP for PATH.
928 sub cmd_propget {
929 my ($prop, $path) = @_;
930 $path = '.' if not defined $path;
931 usage(1) if not defined $prop;
932 my $props = get_svnprops($path);
933 if (not defined $props->{$prop}) {
934 fatal("`$path' does not have a `$prop' SVN property.");
935 }
936 print $props->{$prop} . "\n";
937 }
938
939 # cmd_proplist (PATH)
940 # -------------------
941 # Print the list of SVN properties for PATH.
942 sub cmd_proplist {
943 my $path = shift;
944 $path = '.' if not defined $path;
945 my $props = get_svnprops($path);
946 print "Properties on '$path':\n";
947 foreach (sort keys %{$props}) {
948 print " $_\n";
949 }
950 }
951
952 sub cmd_multi_init {
953 my $url = shift;
954 unless (defined $_trunk || @_branches || @_tags) {
955 usage(1);
956 }
957
958 $_prefix = '' unless defined $_prefix;
959 if (defined $url) {
960 $url = canonicalize_url($url);
961 init_subdir(@_);
962 }
963 do_git_init_db();
964 if (defined $_trunk) {
965 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
966 # try both old-style and new-style lookups:
967 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
968 unless ($gs_trunk) {
969 my ($trunk_url, $trunk_path) =
970 complete_svn_url($url, $_trunk);
971 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
972 undef, $trunk_ref);
973 }
974 }
975 return unless @_branches || @_tags;
976 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
977 foreach my $path (@_branches) {
978 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
979 }
980 foreach my $path (@_tags) {
981 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
982 }
983 }
984
985 sub cmd_multi_fetch {
986 $Git::SVN::no_reuse_existing = undef;
987 my $remotes = Git::SVN::read_all_remotes();
988 foreach my $repo_id (sort keys %$remotes) {
989 if ($remotes->{$repo_id}->{url}) {
990 Git::SVN::fetch_all($repo_id, $remotes);
991 }
992 }
993 }
994
995 # this command is special because it requires no metadata
996 sub cmd_commit_diff {
997 my ($ta, $tb, $url) = @_;
998 my $usage = "Usage: $0 commit-diff -r<revision> ".
999 "<tree-ish> <tree-ish> [<URL>]";
1000 fatal($usage) if (!defined $ta || !defined $tb);
1001 my $svn_path = '';
1002 if (!defined $url) {
1003 my $gs = eval { Git::SVN->new };
1004 if (!$gs) {
1005 fatal("Needed URL or usable git-svn --id in ",
1006 "the command-line\n", $usage);
1007 }
1008 $url = $gs->{url};
1009 $svn_path = $gs->{path};
1010 }
1011 unless (defined $_revision) {
1012 fatal("-r|--revision is a required argument\n", $usage);
1013 }
1014 if (defined $_message && defined $_file) {
1015 fatal("Both --message/-m and --file/-F specified ",
1016 "for the commit message.\n",
1017 "I have no idea what you mean");
1018 }
1019 if (defined $_file) {
1020 $_message = file_to_s($_file);
1021 } else {
1022 $_message ||= get_commit_entry($tb)->{log};
1023 }
1024 my $ra ||= Git::SVN::Ra->new($url);
1025 my $r = $_revision;
1026 if ($r eq 'HEAD') {
1027 $r = $ra->get_latest_revnum;
1028 } elsif ($r !~ /^\d+$/) {
1029 die "revision argument: $r not understood by git-svn\n";
1030 }
1031 my %ed_opts = ( r => $r,
1032 log => $_message,
1033 ra => $ra,
1034 tree_a => $ta,
1035 tree_b => $tb,
1036 editor_cb => sub { print "Committed r$_[0]\n" },
1037 svn_path => $svn_path );
1038 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1039 print "No changes\n$ta == $tb\n";
1040 }
1041 }
1042
1043 sub escape_uri_only {
1044 my ($uri) = @_;
1045 my @tmp;
1046 foreach (split m{/}, $uri) {
1047 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1048 push @tmp, $_;
1049 }
1050 join('/', @tmp);
1051 }
1052
1053 sub escape_url {
1054 my ($url) = @_;
1055 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1056 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1057 $url = "$scheme://$domain$uri";
1058 }
1059 $url;
1060 }
1061
1062 sub cmd_info {
1063 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1064 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1065 if (exists $_[1]) {
1066 die "Too many arguments specified\n";
1067 }
1068
1069 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1070
1071 if (!$file_type && !$diff_status) {
1072 print STDERR "svn: '$path' is not under version control\n";
1073 exit 1;
1074 }
1075
1076 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1077 unless ($gs) {
1078 die "Unable to determine upstream SVN information from ",
1079 "working tree history\n";
1080 }
1081
1082 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1083 $path = "." if $path eq "";
1084
1085 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1086
1087 if ($_url) {
1088 print escape_url($full_url), "\n";
1089 return;
1090 }
1091
1092 my $result = "Path: $path\n";
1093 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1094 $result .= "URL: " . escape_url($full_url) . "\n";
1095
1096 eval {
1097 my $repos_root = $gs->repos_root;
1098 Git::SVN::remove_username($repos_root);
1099 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1100 };
1101 if ($@) {
1102 $result .= "Repository Root: (offline)\n";
1103 }
1104 ::_req_svn();
1105 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1106 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1107 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1108
1109 $result .= "Node Kind: " .
1110 ($file_type eq "dir" ? "directory" : "file") . "\n";
1111
1112 my $schedule = $diff_status eq "A"
1113 ? "add"
1114 : ($diff_status eq "D" ? "delete" : "normal");
1115 $result .= "Schedule: $schedule\n";
1116
1117 if ($diff_status eq "A") {
1118 print $result, "\n";
1119 return;
1120 }
1121
1122 my ($lc_author, $lc_rev, $lc_date_utc);
1123 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1124 my $log = command_output_pipe(@args);
1125 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1126 while (<$log>) {
1127 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1128 $lc_author = $1;
1129 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1130 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1131 (undef, $lc_rev, undef) = ::extract_metadata($1);
1132 }
1133 }
1134 close $log;
1135
1136 Git::SVN::Log::set_local_timezone();
1137
1138 $result .= "Last Changed Author: $lc_author\n";
1139 $result .= "Last Changed Rev: $lc_rev\n";
1140 $result .= "Last Changed Date: " .
1141 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1142
1143 if ($file_type ne "dir") {
1144 my $text_last_updated_date =
1145 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1146 $result .=
1147 "Text Last Updated: " .
1148 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1149 "\n";
1150 my $checksum;
1151 if ($diff_status eq "D") {
1152 my ($fh, $ctx) =
1153 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1154 if ($file_type eq "link") {
1155 my $file_name = <$fh>;
1156 $checksum = md5sum("link $file_name");
1157 } else {
1158 $checksum = md5sum($fh);
1159 }
1160 command_close_pipe($fh, $ctx);
1161 } elsif ($file_type eq "link") {
1162 my $file_name =
1163 command(qw(cat-file blob), "HEAD:$path");
1164 $checksum =
1165 md5sum("link " . $file_name);
1166 } else {
1167 open FILE, "<", $path or die $!;
1168 $checksum = md5sum(\*FILE);
1169 close FILE or die $!;
1170 }
1171 $result .= "Checksum: " . $checksum . "\n";
1172 }
1173
1174 print $result, "\n";
1175 }
1176
1177 sub cmd_reset {
1178 my $target = shift || $_revision or die "SVN revision required\n";
1179 $target = $1 if $target =~ /^r(\d+)$/;
1180 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1181 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1182 unless ($gs) {
1183 die "Unable to determine upstream SVN information from ".
1184 "history\n";
1185 }
1186 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1187 $gs->rev_map_set($r, $c, 'reset', $uuid);
1188 print "r$r = $c ($gs->{ref_id})\n";
1189 }
1190
1191 sub cmd_gc {
1192 if (!$can_compress) {
1193 warn "Compress::Zlib could not be found; unhandled.log " .
1194 "files will not be compressed.\n";
1195 }
1196 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1197 }
1198
1199 ########################### utility functions #########################
1200
1201 sub rebase_cmd {
1202 my @cmd = qw/rebase/;
1203 push @cmd, '-v' if $_verbose;
1204 push @cmd, qw/--merge/ if $_merge;
1205 push @cmd, "--strategy=$_strategy" if $_strategy;
1206 @cmd;
1207 }
1208
1209 sub post_fetch_checkout {
1210 return if $_no_checkout;
1211 my $gs = $Git::SVN::_head or return;
1212 return if verify_ref('refs/heads/master^0');
1213
1214 # look for "trunk" ref if it exists
1215 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1216 my $fetch = $remote->{fetch};
1217 if ($fetch) {
1218 foreach my $p (keys %$fetch) {
1219 basename($fetch->{$p}) eq 'trunk' or next;
1220 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1221 last;
1222 }
1223 }
1224
1225 my $valid_head = verify_ref('HEAD^0');
1226 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1227 return if ($valid_head || !verify_ref('HEAD^0'));
1228
1229 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1230 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1231 return if -f $index;
1232
1233 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1234 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1235 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1236 print STDERR "Checked out HEAD:\n ",
1237 $gs->full_url, " r", $gs->last_rev, "\n";
1238 $gs->mkemptydirs($gs->last_rev);
1239 }
1240
1241 sub complete_svn_url {
1242 my ($url, $path) = @_;
1243 $path =~ s#/+$##;
1244 if ($path !~ m#^[a-z\+]+://#) {
1245 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1246 fatal("E: '$path' is not a complete URL ",
1247 "and a separate URL is not specified");
1248 }
1249 return ($url, $path);
1250 }
1251 return ($path, '');
1252 }
1253
1254 sub complete_url_ls_init {
1255 my ($ra, $repo_path, $switch, $pfx) = @_;
1256 unless ($repo_path) {
1257 print STDERR "W: $switch not specified\n";
1258 return;
1259 }
1260 $repo_path =~ s#/+$##;
1261 if ($repo_path =~ m#^[a-z\+]+://#) {
1262 $ra = Git::SVN::Ra->new($repo_path);
1263 $repo_path = '';
1264 } else {
1265 $repo_path =~ s#^/+##;
1266 unless ($ra) {
1267 fatal("E: '$repo_path' is not a complete URL ",
1268 "and a separate URL is not specified");
1269 }
1270 }
1271 my $url = $ra->{url};
1272 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1273 my $k = "svn-remote.$gs->{repo_id}.url";
1274 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1275 if ($orig_url && ($orig_url ne $gs->{url})) {
1276 die "$k already set: $orig_url\n",
1277 "wanted to set to: $gs->{url}\n";
1278 }
1279 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1280 my $remote_path = "$gs->{path}/$repo_path";
1281 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1282 $remote_path =~ s#/+#/#g;
1283 $remote_path =~ s#^/##g;
1284 $remote_path .= "/*" if $remote_path !~ /\*/;
1285 my ($n) = ($switch =~ /^--(\w+)/);
1286 if (length $pfx && $pfx !~ m#/$#) {
1287 die "--prefix='$pfx' must have a trailing slash '/'\n";
1288 }
1289 command_noisy('config',
1290 '--add',
1291 "svn-remote.$gs->{repo_id}.$n",
1292 "$remote_path:refs/remotes/$pfx*" .
1293 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1294 }
1295
1296 sub verify_ref {
1297 my ($ref) = @_;
1298 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1299 { STDERR => 0 }); };
1300 }
1301
1302 sub get_tree_from_treeish {
1303 my ($treeish) = @_;
1304 # $treeish can be a symbolic ref, too:
1305 my $type = command_oneline(qw/cat-file -t/, $treeish);
1306 my $expected;
1307 while ($type eq 'tag') {
1308 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1309 }
1310 if ($type eq 'commit') {
1311 $expected = (grep /^tree /, command(qw/cat-file commit/,
1312 $treeish))[0];
1313 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1314 die "Unable to get tree from $treeish\n" unless $expected;
1315 } elsif ($type eq 'tree') {
1316 $expected = $treeish;
1317 } else {
1318 die "$treeish is a $type, expected tree, tag or commit\n";
1319 }
1320 return $expected;
1321 }
1322
1323 sub get_commit_entry {
1324 my ($treeish) = shift;
1325 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1326 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1327 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1328 open my $log_fh, '>', $commit_editmsg or croak $!;
1329
1330 my $type = command_oneline(qw/cat-file -t/, $treeish);
1331 if ($type eq 'commit' || $type eq 'tag') {
1332 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1333 $type, $treeish);
1334 my $in_msg = 0;
1335 my $author;
1336 my $saw_from = 0;
1337 my $msgbuf = "";
1338 while (<$msg_fh>) {
1339 if (!$in_msg) {
1340 $in_msg = 1 if (/^\s*$/);
1341 $author = $1 if (/^author (.*>)/);
1342 } elsif (/^git-svn-id: /) {
1343 # skip this for now, we regenerate the
1344 # correct one on re-fetch anyways
1345 # TODO: set *:merge properties or like...
1346 } else {
1347 if (/^From:/ || /^Signed-off-by:/) {
1348 $saw_from = 1;
1349 }
1350 $msgbuf .= $_;
1351 }
1352 }
1353 $msgbuf =~ s/\s+$//s;
1354 if ($Git::SVN::_add_author_from && defined($author)
1355 && !$saw_from) {
1356 $msgbuf .= "\n\nFrom: $author";
1357 }
1358 print $log_fh $msgbuf or croak $!;
1359 command_close_pipe($msg_fh, $ctx);
1360 }
1361 close $log_fh or croak $!;
1362
1363 if ($_edit || ($type eq 'tree')) {
1364 chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1365 system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1366 }
1367 rename $commit_editmsg, $commit_msg or croak $!;
1368 {
1369 require Encode;
1370 # SVN requires messages to be UTF-8 when entering the repo
1371 local $/;
1372 open $log_fh, '<', $commit_msg or croak $!;
1373 binmode $log_fh;
1374 chomp($log_entry{log} = <$log_fh>);
1375
1376 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1377 my $msg = $log_entry{log};
1378
1379 eval { $msg = Encode::decode($enc, $msg, 1) };
1380 if ($@) {
1381 die "Could not decode as $enc:\n", $msg,
1382 "\nPerhaps you need to set i18n.commitencoding\n";
1383 }
1384
1385 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1386 die "Could not encode as UTF-8:\n$msg\n" if $@;
1387
1388 $log_entry{log} = $msg;
1389
1390 close $log_fh or croak $!;
1391 }
1392 unlink $commit_msg;
1393 \%log_entry;
1394 }
1395
1396 sub s_to_file {
1397 my ($str, $file, $mode) = @_;
1398 open my $fd,'>',$file or croak $!;
1399 print $fd $str,"\n" or croak $!;
1400 close $fd or croak $!;
1401 chmod ($mode &~ umask, $file) if (defined $mode);
1402 }
1403
1404 sub file_to_s {
1405 my $file = shift;
1406 open my $fd,'<',$file or croak "$!: file: $file\n";
1407 local $/;
1408 my $ret = <$fd>;
1409 close $fd or croak $!;
1410 $ret =~ s/\s*$//s;
1411 return $ret;
1412 }
1413
1414 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1415 sub load_authors {
1416 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1417 my $log = $cmd eq 'log';
1418 while (<$authors>) {
1419 chomp;
1420 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1421 my ($user, $name, $email) = ($1, $2, $3);
1422 if ($log) {
1423 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1424 } else {
1425 $users{$user} = [$name, $email];
1426 }
1427 }
1428 close $authors or croak $!;
1429 }
1430
1431 # convert GetOpt::Long specs for use by git-config
1432 sub read_git_config {
1433 my $opts = shift;
1434 my @config_only;
1435 foreach my $o (keys %$opts) {
1436 # if we have mixedCase and a long option-only, then
1437 # it's a config-only variable that we don't need for
1438 # the command-line.
1439 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1440 my $v = $opts->{$o};
1441 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1442 $key =~ s/-//g;
1443 my $arg = 'git config';
1444 $arg .= ' --int' if ($o =~ /[:=]i$/);
1445 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1446 if (ref $v eq 'ARRAY') {
1447 chomp(my @tmp = `$arg --get-all svn.$key`);
1448 @$v = @tmp if @tmp;
1449 } else {
1450 chomp(my $tmp = `$arg --get svn.$key`);
1451 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1452 $$v = $tmp;
1453 }
1454 }
1455 }
1456 delete @$opts{@config_only} if @config_only;
1457 }
1458
1459 sub extract_metadata {
1460 my $id = shift or return (undef, undef, undef);
1461 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1462 \s([a-f\d\-]+)$/ix);
1463 if (!defined $rev || !$uuid || !$url) {
1464 # some of the original repositories I made had
1465 # identifiers like this:
1466 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1467 }
1468 return ($url, $rev, $uuid);
1469 }
1470
1471 sub cmt_metadata {
1472 return extract_metadata((grep(/^git-svn-id: /,
1473 command(qw/cat-file commit/, shift)))[-1]);
1474 }
1475
1476 sub cmt_sha2rev_batch {
1477 my %s2r;
1478 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1479 my $list = shift;
1480
1481 foreach my $sha (@{$list}) {
1482 my $first = 1;
1483 my $size = 0;
1484 print $out $sha, "\n";
1485
1486 while (my $line = <$in>) {
1487 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1488 last;
1489 } elsif ($first &&
1490 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1491 $first = 0;
1492 $size = $1;
1493 next;
1494 } elsif ($line =~ /^(git-svn-id: )/) {
1495 my (undef, $rev, undef) =
1496 extract_metadata($line);
1497 $s2r{$sha} = $rev;
1498 }
1499
1500 $size -= length($line);
1501 last if ($size == 0);
1502 }
1503 }
1504
1505 command_close_bidi_pipe($pid, $in, $out, $ctx);
1506
1507 return \%s2r;
1508 }
1509
1510 sub working_head_info {
1511 my ($head, $refs) = @_;
1512 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1513 my ($fh, $ctx) = command_output_pipe(@args, $head);
1514 my $hash;
1515 my %max;
1516 while (<$fh>) {
1517 if ( m{^commit ($::sha1)$} ) {
1518 unshift @$refs, $hash if $hash and $refs;
1519 $hash = $1;
1520 next;
1521 }
1522 next unless s{^\s*(git-svn-id:)}{$1};
1523 my ($url, $rev, $uuid) = extract_metadata($_);
1524 if (defined $url && defined $rev) {
1525 next if $max{$url} and $max{$url} < $rev;
1526 if (my $gs = Git::SVN->find_by_url($url)) {
1527 my $c = $gs->rev_map_get($rev, $uuid);
1528 if ($c && $c eq $hash) {
1529 close $fh; # break the pipe
1530 return ($url, $rev, $uuid, $gs);
1531 } else {
1532 $max{$url} ||= $gs->rev_map_max;
1533 }
1534 }
1535 }
1536 }
1537 command_close_pipe($fh, $ctx);
1538 (undef, undef, undef, undef);
1539 }
1540
1541 sub read_commit_parents {
1542 my ($parents, $c) = @_;
1543 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1544 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1545 @{$parents->{$c}} = split(/ /, $p);
1546 }
1547
1548 sub linearize_history {
1549 my ($gs, $refs) = @_;
1550 my %parents;
1551 foreach my $c (@$refs) {
1552 read_commit_parents(\%parents, $c);
1553 }
1554
1555 my @linear_refs;
1556 my %skip = ();
1557 my $last_svn_commit = $gs->last_commit;
1558 foreach my $c (reverse @$refs) {
1559 next if $c eq $last_svn_commit;
1560 last if $skip{$c};
1561
1562 unshift @linear_refs, $c;
1563 $skip{$c} = 1;
1564
1565 # we only want the first parent to diff against for linear
1566 # history, we save the rest to inject when we finalize the
1567 # svn commit
1568 my $fp_a = verify_ref("$c~1");
1569 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1570 if (!$fp_a || !$fp_b) {
1571 die "Commit $c\n",
1572 "has no parent commit, and therefore ",
1573 "nothing to diff against.\n",
1574 "You should be working from a repository ",
1575 "originally created by git-svn\n";
1576 }
1577 if ($fp_a ne $fp_b) {
1578 die "$c~1 = $fp_a, however parsing commit $c ",
1579 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1580 }
1581
1582 foreach my $p (@{$parents{$c}}) {
1583 $skip{$p} = 1;
1584 }
1585 }
1586 (\@linear_refs, \%parents);
1587 }
1588
1589 sub find_file_type_and_diff_status {
1590 my ($path) = @_;
1591 return ('dir', '') if $path eq '';
1592
1593 my $diff_output =
1594 command_oneline(qw(diff --cached --name-status --), $path) || "";
1595 my $diff_status = (split(' ', $diff_output))[0] || "";
1596
1597 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1598
1599 return (undef, undef) if !$diff_status && !$ls_tree;
1600
1601 if ($diff_status eq "A") {
1602 return ("link", $diff_status) if -l $path;
1603 return ("dir", $diff_status) if -d $path;
1604 return ("file", $diff_status);
1605 }
1606
1607 my $mode = (split(' ', $ls_tree))[0] || "";
1608
1609 return ("link", $diff_status) if $mode eq "120000";
1610 return ("dir", $diff_status) if $mode eq "040000";
1611 return ("file", $diff_status);
1612 }
1613
1614 sub md5sum {
1615 my $arg = shift;
1616 my $ref = ref $arg;
1617 my $md5 = Digest::MD5->new();
1618 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1619 $md5->addfile($arg) or croak $!;
1620 } elsif ($ref eq 'SCALAR') {
1621 $md5->add($$arg) or croak $!;
1622 } elsif (!$ref) {
1623 $md5->add($arg) or croak $!;
1624 } else {
1625 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1626 }
1627 return $md5->hexdigest();
1628 }
1629
1630 sub gc_directory {
1631 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1632 my $out_filename = $_ . ".gz";
1633 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1634 binmode $in_fh;
1635 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1636 die "Unable to open $out_filename: $!\n";
1637
1638 my $res;
1639 while ($res = sysread($in_fh, my $str, 1024)) {
1640 $gz->gzwrite($str) or
1641 die "Unable to write: ".$gz->gzerror()."!\n";
1642 }
1643 unlink $_ or die "unlink $File::Find::name: $!\n";
1644 } elsif (-f $_ && basename($_) eq "index") {
1645 unlink $_ or die "unlink $_: $!\n";
1646 }
1647 }
1648
1649 package Git::SVN;
1650 use strict;
1651 use warnings;
1652 use Fcntl qw/:DEFAULT :seek/;
1653 use constant rev_map_fmt => 'NH40';
1654 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1655 $_repack $_repack_flags $_use_svm_props $_head
1656 $_use_svnsync_props $no_reuse_existing $_minimize_url
1657 $_use_log_author $_add_author_from $_localtime/;
1658 use Carp qw/croak/;
1659 use File::Path qw/mkpath/;
1660 use File::Copy qw/copy/;
1661 use IPC::Open3;
1662 use Memoize; # core since 5.8.0, Jul 2002
1663 use Memoize::Storable;
1664
1665 my ($_gc_nr, $_gc_period);
1666
1667 # properties that we do not log:
1668 my %SKIP_PROP;
1669 BEGIN {
1670 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1671 svn:special svn:executable
1672 svn:entry:committed-rev
1673 svn:entry:last-author
1674 svn:entry:uuid
1675 svn:entry:committed-date/;
1676
1677 # some options are read globally, but can be overridden locally
1678 # per [svn-remote "..."] section. Command-line options will *NOT*
1679 # override options set in an [svn-remote "..."] section
1680 no strict 'refs';
1681 for my $option (qw/follow_parent no_metadata use_svm_props
1682 use_svnsync_props/) {
1683 my $key = $option;
1684 $key =~ tr/_//d;
1685 my $prop = "-$option";
1686 *$option = sub {
1687 my ($self) = @_;
1688 return $self->{$prop} if exists $self->{$prop};
1689 my $k = "svn-remote.$self->{repo_id}.$key";
1690 eval { command_oneline(qw/config --get/, $k) };
1691 if ($@) {
1692 $self->{$prop} = ${"Git::SVN::_$option"};
1693 } else {
1694 my $v = command_oneline(qw/config --bool/,$k);
1695 $self->{$prop} = $v eq 'false' ? 0 : 1;
1696 }
1697 return $self->{$prop};
1698 }
1699 }
1700 }
1701
1702
1703 my (%LOCKFILES, %INDEX_FILES);
1704 END {
1705 unlink keys %LOCKFILES if %LOCKFILES;
1706 unlink keys %INDEX_FILES if %INDEX_FILES;
1707 }
1708
1709 sub resolve_local_globs {
1710 my ($url, $fetch, $glob_spec) = @_;
1711 return unless defined $glob_spec;
1712 my $ref = $glob_spec->{ref};
1713 my $path = $glob_spec->{path};
1714 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1715 next unless m#^$ref->{regex}$#;
1716 my $p = $1;
1717 my $pathname = desanitize_refname($path->full_path($p));
1718 my $refname = desanitize_refname($ref->full_path($p));
1719 if (my $existing = $fetch->{$pathname}) {
1720 if ($existing ne $refname) {
1721 die "Refspec conflict:\n",
1722 "existing: $existing\n",
1723 " globbed: $refname\n";
1724 }
1725 my $u = (::cmt_metadata("$refname"))[0];
1726 $u =~ s!^\Q$url\E(/|$)!! or die
1727 "$refname: '$url' not found in '$u'\n";
1728 if ($pathname ne $u) {
1729 warn "W: Refspec glob conflict ",
1730 "(ref: $refname):\n",
1731 "expected path: $pathname\n",
1732 " real path: $u\n",
1733 "Continuing ahead with $u\n";
1734 next;
1735 }
1736 } else {
1737 $fetch->{$pathname} = $refname;
1738 }
1739 }
1740 }
1741
1742 sub parse_revision_argument {
1743 my ($base, $head) = @_;
1744 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1745 return ($base, $head);
1746 }
1747 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1748 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1749 return ($head, $head) if ($::_revision eq 'HEAD');
1750 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1751 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1752 die "revision argument: $::_revision not understood by git-svn\n";
1753 }
1754
1755 sub fetch_all {
1756 my ($repo_id, $remotes) = @_;
1757 if (ref $repo_id) {
1758 my $gs = $repo_id;
1759 $repo_id = undef;
1760 $repo_id = $gs->{repo_id};
1761 }
1762 $remotes ||= read_all_remotes();
1763 my $remote = $remotes->{$repo_id} or
1764 die "[svn-remote \"$repo_id\"] unknown\n";
1765 my $fetch = $remote->{fetch};
1766 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1767 my (@gs, @globs);
1768 my $ra = Git::SVN::Ra->new($url);
1769 my $uuid = $ra->get_uuid;
1770 my $head = $ra->get_latest_revnum;
1771
1772 # ignore errors, $head revision may not even exist anymore
1773 eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
1774 warn "W: $@\n" if $@;
1775
1776 my $base = defined $fetch ? $head : 0;
1777
1778 # read the max revs for wildcard expansion (branches/*, tags/*)
1779 foreach my $t (qw/branches tags/) {
1780 defined $remote->{$t} or next;
1781 push @globs, @{$remote->{$t}};
1782
1783 my $max_rev = eval { tmp_config(qw/--int --get/,
1784 "svn-remote.$repo_id.${t}-maxRev") };
1785 if (defined $max_rev && ($max_rev < $base)) {
1786 $base = $max_rev;
1787 } elsif (!defined $max_rev) {
1788 $base = 0;
1789 }
1790 }
1791
1792 if ($fetch) {
1793 foreach my $p (sort keys %$fetch) {
1794 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1795 my $lr = $gs->rev_map_max;
1796 if (defined $lr) {
1797 $base = $lr if ($lr < $base);
1798 }
1799 push @gs, $gs;
1800 }
1801 }
1802
1803 ($base, $head) = parse_revision_argument($base, $head);
1804 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1805 }
1806
1807 sub read_all_remotes {
1808 my $r = {};
1809 my $use_svm_props = eval { command_oneline(qw/config --bool
1810 svn.useSvmProps/) };
1811 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1812 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1813 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1814 if (m!^(.+)\.fetch=$svn_refspec$!) {
1815 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1816 die("svn-remote.$remote: remote ref '$remote_ref' "
1817 . "must start with 'refs/'\n")
1818 unless $remote_ref =~ m{^refs/};
1819 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1820 $r->{$remote}->{svm} = {} if $use_svm_props;
1821 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1822 $r->{$1}->{svm} = {};
1823 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1824 $r->{$1}->{url} = $2;
1825 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1826 my ($remote, $t, $local_ref, $remote_ref) =
1827 ($1, $2, $3, $4);
1828 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1829 . "must start with 'refs/'\n")
1830 unless $remote_ref =~ m{^refs/};
1831 my $rs = {
1832 t => $t,
1833 remote => $remote,
1834 path => Git::SVN::GlobSpec->new($local_ref, 1),
1835 ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
1836 if (length($rs->{ref}->{right}) != 0) {
1837 die "The '*' glob character must be the last ",
1838 "character of '$remote_ref'\n";
1839 }
1840 push @{ $r->{$remote}->{$t} }, $rs;
1841 }
1842 }
1843
1844 map {
1845 if (defined $r->{$_}->{svm}) {
1846 my $svm;
1847 eval {
1848 my $section = "svn-remote.$_";
1849 $svm = {
1850 source => tmp_config('--get',
1851 "$section.svm-source"),
1852 replace => tmp_config('--get',
1853 "$section.svm-replace"),
1854 }
1855 };
1856 $r->{$_}->{svm} = $svm;
1857 }
1858 } keys %$r;
1859
1860 $r;
1861 }
1862
1863 sub init_vars {
1864 $_gc_nr = $_gc_period = 1000;
1865 if (defined $_repack || defined $_repack_flags) {
1866 warn "Repack options are obsolete; they have no effect.\n";
1867 }
1868 }
1869
1870 sub verify_remotes_sanity {
1871 return unless -d $ENV{GIT_DIR};
1872 my %seen;
1873 foreach (command(qw/config -l/)) {
1874 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1875 if ($seen{$1}) {
1876 die "Remote ref refs/remote/$1 is tracked by",
1877 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1878 "Please resolve this ambiguity in ",
1879 "your git configuration file before ",
1880 "continuing\n";
1881 }
1882 $seen{$1} = $_;
1883 }
1884 }
1885 }
1886
1887 sub find_existing_remote {
1888 my ($url, $remotes) = @_;
1889 return undef if $no_reuse_existing;
1890 my $existing;
1891 foreach my $repo_id (keys %$remotes) {
1892 my $u = $remotes->{$repo_id}->{url} or next;
1893 next if $u ne $url;
1894 $existing = $repo_id;
1895 last;
1896 }
1897 $existing;
1898 }
1899
1900 sub init_remote_config {
1901 my ($self, $url, $no_write) = @_;
1902 $url =~ s!/+$!!; # strip trailing slash
1903 my $r = read_all_remotes();
1904 my $existing = find_existing_remote($url, $r);
1905 if ($existing) {
1906 unless ($no_write) {
1907 print STDERR "Using existing ",
1908 "[svn-remote \"$existing\"]\n";
1909 }
1910 $self->{repo_id} = $existing;
1911 } elsif ($_minimize_url) {
1912 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1913 $existing = find_existing_remote($min_url, $r);
1914 if ($existing) {
1915 unless ($no_write) {
1916 print STDERR "Using existing ",
1917 "[svn-remote \"$existing\"]\n";
1918 }
1919 $self->{repo_id} = $existing;
1920 }
1921 if ($min_url ne $url) {
1922 unless ($no_write) {
1923 print STDERR "Using higher level of URL: ",
1924 "$url => $min_url\n";
1925 }
1926 my $old_path = $self->{path};
1927 $self->{path} = $url;
1928 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1929 if (length $old_path) {
1930 $self->{path} .= "/$old_path";
1931 }
1932 $url = $min_url;
1933 }
1934 }
1935 my $orig_url;
1936 if (!$existing) {
1937 # verify that we aren't overwriting anything:
1938 $orig_url = eval {
1939 command_oneline('config', '--get',
1940 "svn-remote.$self->{repo_id}.url")
1941 };
1942 if ($orig_url && ($orig_url ne $url)) {
1943 die "svn-remote.$self->{repo_id}.url already set: ",
1944 "$orig_url\nwanted to set to: $url\n";
1945 }
1946 }
1947 my ($xrepo_id, $xpath) = find_ref($self->refname);
1948 if (!$no_write && defined $xpath) {
1949 die "svn-remote.$xrepo_id.fetch already set to track ",
1950 "$xpath:", $self->refname, "\n";
1951 }
1952 unless ($no_write) {
1953 command_noisy('config',
1954 "svn-remote.$self->{repo_id}.url", $url);
1955 $self->{path} =~ s{^/}{};
1956 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1957 command_noisy('config', '--add',
1958 "svn-remote.$self->{repo_id}.fetch",
1959 "$self->{path}:".$self->refname);
1960 }
1961 $self->{url} = $url;
1962 }
1963
1964 sub find_by_url { # repos_root and, path are optional
1965 my ($class, $full_url, $repos_root, $path) = @_;
1966
1967 return undef unless defined $full_url;
1968 remove_username($full_url);
1969 remove_username($repos_root) if defined $repos_root;
1970 my $remotes = read_all_remotes();
1971 if (defined $full_url && defined $repos_root && !defined $path) {
1972 $path = $full_url;
1973 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1974 }
1975 foreach my $repo_id (keys %$remotes) {
1976 my $u = $remotes->{$repo_id}->{url} or next;
1977 remove_username($u);
1978 next if defined $repos_root && $repos_root ne $u;
1979
1980 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1981 foreach my $t (qw/branches tags/) {
1982 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1983 resolve_local_globs($u, $fetch, $globspec);
1984 }
1985 }
1986 my $p = $path;
1987 my $rwr = rewrite_root({repo_id => $repo_id});
1988 my $svm = $remotes->{$repo_id}->{svm}
1989 if defined $remotes->{$repo_id}->{svm};
1990 unless (defined $p) {
1991 $p = $full_url;
1992 my $z = $u;
1993 my $prefix = '';
1994 if ($rwr) {
1995 $z = $rwr;
1996 remove_username($z);
1997 } elsif (defined $svm) {
1998 $z = $svm->{source};
1999 $prefix = $svm->{replace};
2000 $prefix =~ s#^\Q$u\E(?:/|$)##;
2001 $prefix =~ s#/$##;
2002 }
2003 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2004 }
2005 foreach my $f (keys %$fetch) {
2006 next if $f ne $p;
2007 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2008 }
2009 }
2010 undef;
2011 }
2012
2013 sub init {
2014 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2015 my $self = _new($class, $repo_id, $ref_id, $path);
2016 if (defined $url) {
2017 $self->init_remote_config($url, $no_write);
2018 }
2019 $self;
2020 }
2021
2022 sub find_ref {
2023 my ($ref_id) = @_;
2024 foreach (command(qw/config -l/)) {
2025 next unless m!^svn-remote\.(.+)\.fetch=
2026 \s*(.*?)\s*:\s*(.+?)\s*$!x;
2027 my ($repo_id, $path, $ref) = ($1, $2, $3);
2028 if ($ref eq $ref_id) {
2029 $path = '' if ($path =~ m#^\./?#);
2030 return ($repo_id, $path);
2031 }
2032 }
2033 (undef, undef, undef);
2034 }
2035
2036 sub new {
2037 my ($class, $ref_id, $repo_id, $path) = @_;
2038 if (defined $ref_id && !defined $repo_id && !defined $path) {
2039 ($repo_id, $path) = find_ref($ref_id);
2040 if (!defined $repo_id) {
2041 die "Could not find a \"svn-remote.*.fetch\" key ",
2042 "in the repository configuration matching: ",
2043 "$ref_id\n";
2044 }
2045 }
2046 my $self = _new($class, $repo_id, $ref_id, $path);
2047 if (!defined $self->{path} || !length $self->{path}) {
2048 my $fetch = command_oneline('config', '--get',
2049 "svn-remote.$repo_id.fetch",
2050 ":$ref_id\$") or
2051 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2052 "\":$ref_id\$\" in config\n";
2053 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2054 }
2055 $self->{url} = command_oneline('config', '--get',
2056 "svn-remote.$repo_id.url") or
2057 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2058 $self->rebuild;
2059 $self;
2060 }
2061
2062 sub refname {
2063 my ($refname) = $_[0]->{ref_id} ;
2064
2065 # It cannot end with a slash /, we'll throw up on this because
2066 # SVN can't have directories with a slash in their name, either:
2067 if ($refname =~ m{/$}) {
2068 die "ref: '$refname' ends with a trailing slash, this is ",
2069 "not permitted by git nor Subversion\n";
2070 }
2071
2072 # It cannot have ASCII control character space, tilde ~, caret ^,
2073 # colon :, question-mark ?, asterisk *, space, or open bracket [
2074 # anywhere.
2075 #
2076 # Additionally, % must be escaped because it is used for escaping
2077 # and we want our escaped refname to be reversible
2078 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2079
2080 # no slash-separated component can begin with a dot .
2081 # /.* becomes /%2E*
2082 $refname =~ s{/\.}{/%2E}g;
2083
2084 # It cannot have two consecutive dots .. anywhere
2085 # .. becomes %2E%2E
2086 $refname =~ s{\.\.}{%2E%2E}g;
2087
2088 return $refname;
2089 }
2090
2091 sub desanitize_refname {
2092 my ($refname) = @_;
2093 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2094 return $refname;
2095 }
2096
2097 sub svm_uuid {
2098 my ($self) = @_;
2099 return $self->{svm}->{uuid} if $self->svm;
2100 $self->ra;
2101 unless ($self->{svm}) {
2102 die "SVM UUID not cached, and reading remotely failed\n";
2103 }
2104 $self->{svm}->{uuid};
2105 }
2106
2107 sub svm {
2108 my ($self) = @_;
2109 return $self->{svm} if $self->{svm};
2110 my $svm;
2111 # see if we have it in our config, first:
2112 eval {
2113 my $section = "svn-remote.$self->{repo_id}";
2114 $svm = {
2115 source => tmp_config('--get', "$section.svm-source"),
2116 uuid => tmp_config('--get', "$section.svm-uuid"),
2117 replace => tmp_config('--get', "$section.svm-replace"),
2118 }
2119 };
2120 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2121 $self->{svm} = $svm;
2122 }
2123 $self->{svm};
2124 }
2125
2126 sub _set_svm_vars {
2127 my ($self, $ra) = @_;
2128 return $ra if $self->svm;
2129
2130 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2131 "(svm:source, svm:uuid) ",
2132 "from the following URLs:\n" );
2133 sub read_svm_props {
2134 my ($self, $ra, $path, $r) = @_;
2135 my $props = ($ra->get_dir($path, $r))[2];
2136 my $src = $props->{'svm:source'};
2137 my $uuid = $props->{'svm:uuid'};
2138 return undef if (!$src || !$uuid);
2139
2140 chomp($src, $uuid);
2141
2142 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2143 or die "doesn't look right - svm:uuid is '$uuid'\n";
2144
2145 # the '!' is used to mark the repos_root!/relative/path
2146 $src =~ s{/?!/?}{/};
2147 $src =~ s{/+$}{}; # no trailing slashes please
2148 # username is of no interest
2149 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2150
2151 my $replace = $ra->{url};
2152 $replace .= "/$path" if length $path;
2153
2154 my $section = "svn-remote.$self->{repo_id}";
2155 tmp_config("$section.svm-source", $src);
2156 tmp_config("$section.svm-replace", $replace);
2157 tmp_config("$section.svm-uuid", $uuid);
2158 $self->{svm} = {
2159 source => $src,
2160 uuid => $uuid,
2161 replace => $replace
2162 };
2163 }
2164
2165 my $r = $ra->get_latest_revnum;
2166 my $path = $self->{path};
2167 my %tried;
2168 while (length $path) {
2169 unless ($tried{"$self->{url}/$path"}) {
2170 return $ra if $self->read_svm_props($ra, $path, $r);
2171 $tried{"$self->{url}/$path"} = 1;
2172 }
2173 $path =~ s#/?[^/]+$##;
2174 }
2175 die "Path: '$path' should be ''\n" if $path ne '';
2176 return $ra if $self->read_svm_props($ra, $path, $r);
2177 $tried{"$self->{url}/$path"} = 1;
2178
2179 if ($ra->{repos_root} eq $self->{url}) {
2180 die @err, (map { " $_\n" } keys %tried), "\n";
2181 }
2182
2183 # nope, make sure we're connected to the repository root:
2184 my $ok;
2185 my @tried_b;
2186 $path = $ra->{svn_path};
2187 $ra = Git::SVN::Ra->new($ra->{repos_root});
2188 while (length $path) {
2189 unless ($tried{"$ra->{url}/$path"}) {
2190 $ok = $self->read_svm_props($ra, $path, $r);
2191 last if $ok;
2192 $tried{"$ra->{url}/$path"} = 1;
2193 }
2194 $path =~ s#/?[^/]+$##;
2195 }
2196 die "Path: '$path' should be ''\n" if $path ne '';
2197 $ok ||= $self->read_svm_props($ra, $path, $r);
2198 $tried{"$ra->{url}/$path"} = 1;
2199 if (!$ok) {
2200 die @err, (map { " $_\n" } keys %tried), "\n";
2201 }
2202 Git::SVN::Ra->new($self->{url});
2203 }
2204
2205 sub svnsync {
2206 my ($self) = @_;
2207 return $self->{svnsync} if $self->{svnsync};
2208
2209 if ($self->no_metadata) {
2210 die "Can't have both 'noMetadata' and ",
2211 "'useSvnsyncProps' options set!\n";
2212 }
2213 if ($self->rewrite_root) {
2214 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2215 "options set!\n";
2216 }
2217 if ($self->rewrite_uuid) {
2218 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2219 "options set!\n";
2220 }
2221
2222 my $svnsync;
2223 # see if we have it in our config, first:
2224 eval {
2225 my $section = "svn-remote.$self->{repo_id}";
2226
2227 my $url = tmp_config('--get', "$section.svnsync-url");
2228 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2229 die "doesn't look right - svn:sync-from-url is '$url'\n";
2230
2231 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2232 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2233 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2234
2235 $svnsync = { url => $url, uuid => $uuid }
2236 };
2237 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2238 return $self->{svnsync} = $svnsync;
2239 }
2240
2241 my $err = "useSvnsyncProps set, but failed to read " .
2242 "svnsync property: svn:sync-from-";
2243 my $rp = $self->ra->rev_proplist(0);
2244
2245 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2246 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2247 die "doesn't look right - svn:sync-from-url is '$url'\n";
2248
2249 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2250 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2251 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2252
2253 my $section = "svn-remote.$self->{repo_id}";
2254 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2255 tmp_config('--add', "$section.svnsync-url", $url);
2256 return $self->{svnsync} = { url => $url, uuid => $uuid };
2257 }
2258
2259 # this allows us to memoize our SVN::Ra UUID locally and avoid a
2260 # remote lookup (useful for 'git svn log').
2261 sub ra_uuid {
2262 my ($self) = @_;
2263 unless ($self->{ra_uuid}) {
2264 my $key = "svn-remote.$self->{repo_id}.uuid";
2265 my $uuid = eval { tmp_config('--get', $key) };
2266 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2267 $self->{ra_uuid} = $uuid;
2268 } else {
2269 die "ra_uuid called without URL\n" unless $self->{url};
2270 $self->{ra_uuid} = $self->ra->get_uuid;
2271 tmp_config('--add', $key, $self->{ra_uuid});
2272 }
2273 }
2274 $self->{ra_uuid};
2275 }
2276
2277 sub _set_repos_root {
2278 my ($self, $repos_root) = @_;
2279 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2280 $repos_root ||= $self->ra->{repos_root};
2281 tmp_config($k, $repos_root);
2282 $repos_root;
2283 }
2284
2285 sub repos_root {
2286 my ($self) = @_;
2287 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2288 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2289 }
2290
2291 sub ra {
2292 my ($self) = shift;
2293 my $ra = Git::SVN::Ra->new($self->{url});
2294 $self->_set_repos_root($ra->{repos_root});
2295 if ($self->use_svm_props && !$self->{svm}) {
2296 if ($self->no_metadata) {
2297 die "Can't have both 'noMetadata' and ",
2298 "'useSvmProps' options set!\n";
2299 } elsif ($self->use_svnsync_props) {
2300 die "Can't have both 'useSvnsyncProps' and ",
2301 "'useSvmProps' options set!\n";
2302 }
2303 $ra = $self->_set_svm_vars($ra);
2304 $self->{-want_revprops} = 1;
2305 }
2306 $ra;
2307 }
2308
2309 # prop_walk(PATH, REV, SUB)
2310 # -------------------------
2311 # Recursively traverse PATH at revision REV and invoke SUB for each
2312 # directory that contains a SVN property. SUB will be invoked as
2313 # follows: &SUB(gs, path, props); where `gs' is this instance of
2314 # Git::SVN, `path' the path to the directory where the properties
2315 # `props' were found. The `path' will be relative to point of checkout,
2316 # that is, if url://repo/trunk is the current Git branch, and that
2317 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
2318 # as `path' (note the trailing `/').
2319 sub prop_walk {
2320 my ($self, $path, $rev, $sub) = @_;
2321
2322 $path =~ s#^/##;
2323 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2324 $path =~ s#^/*#/#g;
2325 my $p = $path;
2326 # Strip the irrelevant part of the path.
2327 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2328 # Ensure the path is terminated by a `/'.
2329 $p =~ s#/*$#/#;
2330
2331 # The properties contain all the internal SVN stuff nobody
2332 # (usually) cares about.
2333 my $interesting_props = 0;
2334 foreach (keys %{$props}) {
2335 # If it doesn't start with `svn:', it must be a
2336 # user-defined property.
2337 ++$interesting_props and next if $_ !~ /^svn:/;
2338 # FIXME: Fragile, if SVN adds new public properties,
2339 # this needs to be updated.
2340 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2341 |eol-style|mime-type
2342 |externals|needs-lock)$/x;
2343 }
2344 &$sub($self, $p, $props) if $interesting_props;
2345
2346 foreach (sort keys %$dirent) {
2347 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2348 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2349 }
2350 }
2351
2352 sub last_rev { ($_[0]->last_rev_commit)[0] }
2353 sub last_commit { ($_[0]->last_rev_commit)[1] }
2354
2355 # returns the newest SVN revision number and newest commit SHA1
2356 sub last_rev_commit {
2357 my ($self) = @_;
2358 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2359 return ($self->{last_rev}, $self->{last_commit});
2360 }
2361 my $c = ::verify_ref($self->refname.'^0');
2362 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2363 my $rev = (::cmt_metadata($c))[1];
2364 if (defined $rev) {
2365 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2366 return ($rev, $c);
2367 }
2368 }
2369 my $map_path = $self->map_path;
2370 unless (-e $map_path) {
2371 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2372 return (undef, undef);
2373 }
2374 my ($rev, $commit) = $self->rev_map_max(1);
2375 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2376 return ($rev, $commit);
2377 }
2378
2379 sub get_fetch_range {
2380 my ($self, $min, $max) = @_;
2381 $max ||= $self->ra->get_latest_revnum;
2382 $min ||= $self->rev_map_max;
2383 (++$min, $max);
2384 }
2385
2386 sub tmp_config {
2387 my (@args) = @_;
2388 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2389 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2390 if (! -f $config && -f $old_def_config) {
2391 rename $old_def_config, $config or
2392 die "Failed rename $old_def_config => $config: $!\n";
2393 }
2394 my $old_config = $ENV{GIT_CONFIG};
2395 $ENV{GIT_CONFIG} = $config;
2396 $@ = undef;
2397 my @ret = eval {
2398 unless (-f $config) {
2399 mkfile($config);
2400 open my $fh, '>', $config or
2401 die "Can't open $config: $!\n";
2402 print $fh "; This file is used internally by ",
2403 "git-svn\n" or die
2404 "Couldn't write to $config: $!\n";
2405 print $fh "; You should not have to edit it\n" or
2406 die "Couldn't write to $config: $!\n";
2407 close $fh or die "Couldn't close $config: $!\n";
2408 }
2409 command('config', @args);
2410 };
2411 my $err = $@;
2412 if (defined $old_config) {
2413 $ENV{GIT_CONFIG} = $old_config;
2414 } else {
2415 delete $ENV{GIT_CONFIG};
2416 }
2417 die $err if $err;
2418 wantarray ? @ret : $ret[0];
2419 }
2420
2421 sub tmp_index_do {
2422 my ($self, $sub) = @_;
2423 my $old_index = $ENV{GIT_INDEX_FILE};
2424 $ENV{GIT_INDEX_FILE} = $self->{index};
2425 $@ = undef;
2426 my @ret = eval {
2427 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2428 mkpath([$dir]) unless -d $dir;
2429 &$sub;
2430 };
2431 my $err = $@;
2432 if (defined $old_index) {
2433 $ENV{GIT_INDEX_FILE} = $old_index;
2434 } else {
2435 delete $ENV{GIT_INDEX_FILE};
2436 }
2437 die $err if $err;
2438 wantarray ? @ret : $ret[0];
2439 }
2440
2441 sub assert_index_clean {
2442 my ($self, $treeish) = @_;
2443
2444 $self->tmp_index_do(sub {
2445 command_noisy('read-tree', $treeish) unless -e $self->{index};
2446 my $x = command_oneline('write-tree');
2447 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2448 /^tree ($::sha1)/mo);
2449 return if $y eq $x;
2450
2451 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2452 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2453 command_noisy('read-tree', $treeish);
2454 $x = command_oneline('write-tree');
2455 if ($y ne $x) {
2456 ::fatal "trees ($treeish) $y != $x\n",
2457 "Something is seriously wrong...";
2458 }
2459 });
2460 }
2461
2462 sub get_commit_parents {
2463 my ($self, $log_entry) = @_;
2464 my (%seen, @ret, @tmp);
2465 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2466 if (my $ip = $self->{inject_parents}) {
2467 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2468 push @tmp, $commit;
2469 }
2470 }
2471 if (my $cur = ::verify_ref($self->refname.'^0')) {
2472 push @tmp, $cur;
2473 }
2474 if (my $ipd = $self->{inject_parents_dcommit}) {
2475 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2476 push @tmp, @$commit;
2477 }
2478 }
2479 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2480 while (my $p = shift @tmp) {
2481 next if $seen{$p};
2482 $seen{$p} = 1;
2483 push @ret, $p;
2484 }
2485 @ret;
2486 }
2487
2488 sub rewrite_root {
2489 my ($self) = @_;
2490 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2491 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2492 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2493 if ($rwr) {
2494 $rwr =~ s#/+$##;
2495 if ($rwr !~ m#^[a-z\+]+://#) {
2496 die "$rwr is not a valid URL (key: $k)\n";
2497 }
2498 }
2499 $self->{-rewrite_root} = $rwr;
2500 }
2501
2502 sub rewrite_uuid {
2503 my ($self) = @_;
2504 return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2505 my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2506 my $rwid = eval { command_oneline(qw/config --get/, $k) };
2507 if ($rwid) {
2508 $rwid =~ s#/+$##;
2509 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2510 die "$rwid is not a valid UUID (key: $k)\n";
2511 }
2512 }
2513 $self->{-rewrite_uuid} = $rwid;
2514 }
2515
2516 sub metadata_url {
2517 my ($self) = @_;
2518 ($self->rewrite_root || $self->{url}) .
2519 (length $self->{path} ? '/' . $self->{path} : '');
2520 }
2521
2522 sub full_url {
2523 my ($self) = @_;
2524 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2525 }
2526
2527
2528 sub set_commit_header_env {
2529 my ($log_entry) = @_;
2530 my %env;
2531 foreach my $ned (qw/NAME EMAIL DATE/) {
2532 foreach my $ac (qw/AUTHOR COMMITTER/) {
2533 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2534 }
2535 }
2536
2537 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2538 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2539 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2540
2541 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2542 ? $log_entry->{commit_name}
2543 : $log_entry->{name};
2544 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2545 ? $log_entry->{commit_email}
2546 : $log_entry->{email};
2547 \%env;
2548 }
2549
2550 sub restore_commit_header_env {
2551 my ($env) = @_;
2552 foreach my $ned (qw/NAME EMAIL DATE/) {
2553 foreach my $ac (qw/AUTHOR COMMITTER/) {
2554 my $k = "GIT_${ac}_${ned}";
2555 if (defined $env->{$k}) {
2556 $ENV{$k} = $env->{$k};
2557 } else {
2558 delete $ENV{$k};
2559 }
2560 }
2561 }
2562 }
2563
2564 sub gc {
2565 command_noisy('gc', '--auto');
2566 };
2567
2568 sub do_git_commit {
2569 my ($self, $log_entry) = @_;
2570 my $lr = $self->last_rev;
2571 if (defined $lr && $lr >= $log_entry->{revision}) {
2572 die "Last fetched revision of ", $self->refname,
2573 " was r$lr, but we are about to fetch: ",
2574 "r$log_entry->{revision}!\n";
2575 }
2576 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2577 croak "$log_entry->{revision} = $c already exists! ",
2578 "Why are we refetching it?\n";
2579 }
2580 my $old_env = set_commit_header_env($log_entry);
2581 my $tree = $log_entry->{tree};
2582 if (!defined $tree) {
2583 $tree = $self->tmp_index_do(sub {
2584 command_oneline('write-tree') });
2585 }
2586 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2587
2588 my @exec = ('git', 'commit-tree', $tree);
2589 foreach ($self->get_commit_parents($log_entry)) {
2590 push @exec, '-p', $_;
2591 }
2592 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2593 or croak $!;
2594 binmode $msg_fh;
2595
2596 # we always get UTF-8 from SVN, but we may want our commits in
2597 # a different encoding.
2598 if (my $enc = Git::config('i18n.commitencoding')) {
2599 require Encode;
2600 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2601 }
2602 print $msg_fh $log_entry->{log} or croak $!;
2603 restore_commit_header_env($old_env);
2604 unless ($self->no_metadata) {
2605 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2606 or croak $!;
2607 }
2608 $msg_fh->flush == 0 or croak $!;
2609 close $msg_fh or croak $!;
2610 chomp(my $commit = do { local $/; <$out_fh> });
2611 close $out_fh or croak $!;
2612 waitpid $pid, 0;
2613 croak $? if $?;
2614 if ($commit !~ /^$::sha1$/o) {
2615 die "Failed to commit, invalid sha1: $commit\n";
2616 }
2617
2618 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2619
2620 $self->{last_rev} = $log_entry->{revision};
2621 $self->{last_commit} = $commit;
2622 print "r$log_entry->{revision}" unless $::_q > 1;
2623 if (defined $log_entry->{svm_revision}) {
2624 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2625 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2626 0, $self->svm_uuid);
2627 }
2628 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2629 if (--$_gc_nr == 0) {
2630 $_gc_nr = $_gc_period;
2631 gc();
2632 }
2633 return $commit;
2634 }
2635
2636 sub match_paths {
2637 my ($self, $paths, $r) = @_;
2638 return 1 if $self->{path} eq '';
2639 if (my $path = $paths->{"/$self->{path}"}) {
2640 return ($path->{action} eq 'D') ? 0 : 1;
2641 }
2642 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2643 if (grep /$self->{path_regex}/, keys %$paths) {
2644 return 1;
2645 }
2646 my $c = '';
2647 foreach (split m#/#, $self->{path}) {
2648 $c .= "/$_";
2649 next unless ($paths->{$c} &&
2650 ($paths->{$c}->{action} =~ /^[AR]$/));
2651 if ($self->ra->check_path($self->{path}, $r) ==
2652 $SVN::Node::dir) {
2653 return 1;
2654 }
2655 }
2656 return 0;
2657 }
2658
2659 sub find_parent_branch {
2660 my ($self, $paths, $rev) = @_;
2661 return undef unless $self->follow_parent;
2662 unless (defined $paths) {
2663 my $err_handler = $SVN::Error::handler;
2664 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2665 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2666 sub { $paths = $_[0] });
2667 $SVN::Error::handler = $err_handler;
2668 }
2669 return undef unless defined $paths;
2670
2671 # look for a parent from another branch:
2672 my @b_path_components = split m#/#, $self->{path};
2673 my @a_path_components;
2674 my $i;
2675 while (@b_path_components) {
2676 $i = $paths->{'/'.join('/', @b_path_components)};
2677 last if $i && defined $i->{copyfrom_path};
2678 unshift(@a_path_components, pop(@b_path_components));
2679 }
2680 return undef unless defined $i && defined $i->{copyfrom_path};
2681 my $branch_from = $i->{copyfrom_path};
2682 if (@a_path_components) {
2683 print STDERR "branch_from: $branch_from => ";
2684 $branch_from .= '/'.join('/', @a_path_components);
2685 print STDERR $branch_from, "\n";
2686 }
2687 my $r = $i->{copyfrom_rev};
2688 my $repos_root = $self->ra->{repos_root};
2689 my $url = $self->ra->{url};
2690 my $new_url = $url . $branch_from;
2691 print STDERR "Found possible branch point: ",
2692 "$new_url => ", $self->full_url, ", $r\n"
2693 unless $::_q > 1;
2694 $branch_from =~ s#^/##;
2695 my $gs = $self->other_gs($new_url, $url,
2696 $branch_from, $r, $self->{ref_id});
2697 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2698 {
2699 my ($base, $head);
2700 if (!defined $r0 || !defined $parent) {
2701 ($base, $head) = parse_revision_argument(0, $r);
2702 } else {
2703 if ($r0 < $r) {
2704 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2705 0, 1, sub { $base = $_[1] - 1 });
2706 }
2707 }
2708 if (defined $base && $base <= $r) {
2709 $gs->fetch($base, $r);
2710 }
2711 ($r0, $parent) = $gs->find_rev_before($r, 1);
2712 }
2713 if (defined $r0 && defined $parent) {
2714 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2715 unless $::_q > 1;
2716 my $ed;
2717 if ($self->ra->can_do_switch) {
2718 $self->assert_index_clean($parent);
2719 print STDERR "Following parent with do_switch\n"
2720 unless $::_q > 1;
2721 # do_switch works with svn/trunk >= r22312, but that
2722 # is not included with SVN 1.4.3 (the latest version
2723 # at the moment), so we can't rely on it
2724 $self->{last_rev} = $r0;
2725 $self->{last_commit} = $parent;
2726 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2727 $gs->ra->gs_do_switch($r0, $rev, $gs,
2728 $self->full_url, $ed)
2729 or die "SVN connection failed somewhere...\n";
2730 } elsif ($self->ra->trees_match($new_url, $r0,
2731 $self->full_url, $rev)) {
2732 print STDERR "Trees match:\n",
2733 " $new_url\@$r0\n",
2734 " ${\$self->full_url}\@$rev\n",
2735 "Following parent with no changes\n"
2736 unless $::_q > 1;
2737 $self->tmp_index_do(sub {
2738 command_noisy('read-tree', $parent);
2739 });
2740 $self->{last_commit} = $parent;
2741 } else {
2742 print STDERR "Following parent with do_update\n"
2743 unless $::_q > 1;
2744 $ed = SVN::Git::Fetcher->new($self);
2745 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2746 or die "SVN connection failed somewhere...\n";
2747 }
2748 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2749 return $self->make_log_entry($rev, [$parent], $ed);
2750 }
2751 return undef;
2752 }
2753
2754 sub do_fetch {
2755 my ($self, $paths, $rev) = @_;
2756 my $ed;
2757 my ($last_rev, @parents);
2758 if (my $lc = $self->last_commit) {
2759 # we can have a branch that was deleted, then re-added
2760 # under the same name but copied from another path, in
2761 # which case we'll have multiple parents (we don't
2762 # want to break the original ref, nor lose copypath info):
2763 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2764 push @{$log_entry->{parents}}, $lc;
2765 return $log_entry;
2766 }
2767 $ed = SVN::Git::Fetcher->new($self);
2768 $last_rev = $self->{last_rev};
2769 $ed->{c} = $lc;
2770 @parents = ($lc);
2771 } else {
2772 $last_rev = $rev;
2773 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2774 return $log_entry;
2775 }
2776 $ed = SVN::Git::Fetcher->new($self);
2777 }
2778 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2779 die "SVN connection failed somewhere...\n";
2780 }
2781 $self->make_log_entry($rev, \@parents, $ed);
2782 }
2783
2784 sub mkemptydirs {
2785 my ($self, $r) = @_;
2786
2787 sub scan {
2788 my ($r, $empty_dirs, $line) = @_;
2789 if (defined $r && $line =~ /^r(\d+)$/) {
2790 return 0 if $1 > $r;
2791 } elsif ($line =~ /^ \+empty_dir: (.+)$/) {
2792 $empty_dirs->{$1} = 1;
2793 } elsif ($line =~ /^ \-empty_dir: (.+)$/) {
2794 my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
2795 delete @$empty_dirs{@d};
2796 }
2797 1; # continue
2798 };
2799
2800 my %empty_dirs = ();
2801 my $gz_file = "$self->{dir}/unhandled.log.gz";
2802 if (-f $gz_file) {
2803 if (!$can_compress) {
2804 warn "Compress::Zlib could not be found; ",
2805 "empty directories in $gz_file will not be read\n";
2806 } else {
2807 my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
2808 die "Unable to open $gz_file: $!\n";
2809 my $line;
2810 while ($gz->gzreadline($line) > 0) {
2811 scan($r, \%empty_dirs, $line) or last;
2812 }
2813 $gz->gzclose;
2814 }
2815 }
2816
2817 if (open my $fh, '<', "$self->{dir}/unhandled.log") {
2818 binmode $fh or croak "binmode: $!";
2819 while (<$fh>) {
2820 scan($r, \%empty_dirs, $_) or last;
2821 }
2822 close $fh;
2823 }
2824
2825 my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2826 foreach my $d (sort keys %empty_dirs) {
2827 $d = uri_decode($d);
2828 $d =~ s/$strip//;
2829 next if -d $d;
2830 if (-e _) {
2831 warn "$d exists but is not a directory\n";
2832 } else {
2833 print "creating empty directory: $d\n";
2834 mkpath([$d]);
2835 }
2836 }
2837 }
2838
2839 sub get_untracked {
2840 my ($self, $ed) = @_;
2841 my @out;
2842 my $h = $ed->{empty};
2843 foreach (sort keys %$h) {
2844 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2845 push @out, " $act: " . uri_encode($_);
2846 warn "W: $act: $_\n";
2847 }
2848 foreach my $t (qw/dir_prop file_prop/) {
2849 $h = $ed->{$t} or next;
2850 foreach my $path (sort keys %$h) {
2851 my $ppath = $path eq '' ? '.' : $path;
2852 foreach my $prop (sort keys %{$h->{$path}}) {
2853 next if $SKIP_PROP{$prop};
2854 my $v = $h->{$path}->{$prop};
2855 my $t_ppath_prop = "$t: " .
2856 uri_encode($ppath) . ' ' .
2857 uri_encode($prop);
2858 if (defined $v) {
2859 push @out, " +$t_ppath_prop " .
2860 uri_encode($v);
2861 } else {
2862 push @out, " -$t_ppath_prop";
2863 }
2864 }
2865 }
2866 }
2867 foreach my $t (qw/absent_file absent_directory/) {
2868 $h = $ed->{$t} or next;
2869 foreach my $parent (sort keys %$h) {
2870 foreach my $path (sort @{$h->{$parent}}) {
2871 push @out, " $t: " .
2872 uri_encode("$parent/$path");
2873 warn "W: $t: $parent/$path ",
2874 "Insufficient permissions?\n";
2875 }
2876 }
2877 }
2878 \@out;
2879 }
2880
2881 # parse_svn_date(DATE)
2882 # --------------------
2883 # Given a date (in UTC) from Subversion, return a string in the format
2884 # "<TZ Offset> <local date/time>" that Git will use.
2885 #
2886 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
2887 # is true we'll convert it to the local timezone instead.
2888 sub parse_svn_date {
2889 my $date = shift || return '+0000 1970-01-01 00:00:00';
2890 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2891 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2892 croak "Unable to parse date: $date\n";
2893 my $parsed_date; # Set next.
2894
2895 if ($Git::SVN::_localtime) {
2896 # Translate the Subversion datetime to an epoch time.
2897 # Begin by switching ourselves to $date's timezone, UTC.
2898 my $old_env_TZ = $ENV{TZ};
2899 $ENV{TZ} = 'UTC';
2900
2901 my $epoch_in_UTC =
2902 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2903
2904 # Determine our local timezone (including DST) at the
2905 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2906 # value of TZ, if any, at the time we were run.
2907 if (defined $Git::SVN::Log::TZ) {
2908 $ENV{TZ} = $Git::SVN::Log::TZ;
2909 } else {
2910 delete $ENV{TZ};
2911 }
2912
2913 my $our_TZ =
2914 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2915
2916 # This converts $epoch_in_UTC into our local timezone.
2917 my ($sec, $min, $hour, $mday, $mon, $year,
2918 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2919
2920 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2921 $our_TZ, $year + 1900, $mon + 1,
2922 $mday, $hour, $min, $sec);
2923
2924 # Reset us to the timezone in effect when we entered
2925 # this routine.
2926 if (defined $old_env_TZ) {
2927 $ENV{TZ} = $old_env_TZ;
2928 } else {
2929 delete $ENV{TZ};
2930 }
2931 } else {
2932 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2933 }
2934
2935 return $parsed_date;
2936 }
2937
2938 sub other_gs {
2939 my ($self, $new_url, $url,
2940 $branch_from, $r, $old_ref_id) = @_;
2941 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2942 unless ($gs) {
2943 my $ref_id = $old_ref_id;
2944 $ref_id =~ s/\@\d+$//;
2945 $ref_id .= "\@$r";
2946 # just grow a tail if we're not unique enough :x
2947 $ref_id .= '-' while find_ref($ref_id);
2948 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2949 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2950 if ($u =~ s#^\Q$url\E(/|$)##) {
2951 $p = $u;
2952 $u = $url;
2953 $repo_id = $self->{repo_id};
2954 }
2955 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2956 }
2957 $gs
2958 }
2959
2960 sub call_authors_prog {
2961 my ($orig_author) = @_;
2962 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2963 my $author = `$::_authors_prog $orig_author`;
2964 if ($? != 0) {
2965 die "$::_authors_prog failed with exit code $?\n"
2966 }
2967 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2968 my ($name, $email) = ($1, $2);
2969 $email = undef if length $2 == 0;
2970 return [$name, $email];
2971 } else {
2972 die "Author: $orig_author: $::_authors_prog returned "
2973 . "invalid author format: $author\n";
2974 }
2975 }
2976
2977 sub check_author {
2978 my ($author) = @_;
2979 if (!defined $author || length $author == 0) {
2980 $author = '(no author)';
2981 }
2982 if (!defined $::users{$author}) {
2983 if (defined $::_authors_prog) {
2984 $::users{$author} = call_authors_prog($author);
2985 } elsif (defined $::_authors) {
2986 die "Author: $author not defined in $::_authors file\n";
2987 }
2988 }
2989 $author;
2990 }
2991
2992 sub find_extra_svk_parents {
2993 my ($self, $ed, $tickets, $parents) = @_;
2994 # aha! svk:merge property changed...
2995 my @tickets = split "\n", $tickets;
2996 my @known_parents;
2997 for my $ticket ( @tickets ) {
2998 my ($uuid, $path, $rev) = split /:/, $ticket;
2999 if ( $uuid eq $self->ra_uuid ) {
3000 my $url = $self->rewrite_root || $self->{url};
3001 my $repos_root = $url;
3002 my $branch_from = $path;
3003 $branch_from =~ s{^/}{};
3004 my $gs = $self->other_gs($repos_root."/".$branch_from,
3005 $url,
3006 $branch_from,
3007 $rev,
3008 $self->{ref_id});
3009 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3010 # wahey! we found it, but it might be
3011 # an old one (!)
3012 push @known_parents, [ $rev, $commit ];
3013 }
3014 }
3015 }
3016 # Ordering matters; highest-numbered commit merge tickets
3017 # first, as they may account for later merge ticket additions
3018 # or changes.
3019 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3020 for my $parent ( @known_parents ) {
3021 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3022 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3023 my $new;
3024 while ( <$msg_fh> ) {
3025 $new=1;last;
3026 }
3027 command_close_pipe($msg_fh, $ctx);
3028 if ( $new ) {
3029 print STDERR
3030 "Found merge parent (svk:merge ticket): $parent\n";
3031 push @$parents, $parent;
3032 }
3033 }
3034 }
3035
3036 sub lookup_svn_merge {
3037 my $uuid = shift;
3038 my $url = shift;
3039 my $merge = shift;
3040
3041 my ($source, $revs) = split ":", $merge;
3042 my $path = $source;
3043 $path =~ s{^/}{};
3044 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3045 if ( !$gs ) {
3046 warn "Couldn't find revmap for $url$source\n";
3047 return;
3048 }
3049 my @ranges = split ",", $revs;
3050 my ($tip, $tip_commit);
3051 my @merged_commit_ranges;
3052 # find the tip
3053 for my $range ( @ranges ) {
3054 my ($bottom, $top) = split "-", $range;
3055 $top ||= $bottom;
3056 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3057 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3058
3059 unless ($top_commit and $bottom_commit) {
3060 warn "W:unknown path/rev in svn:mergeinfo "
3061 ."dirprop: $source:$range\n";
3062 next;
3063 }
3064
3065 push @merged_commit_ranges,
3066 "$bottom_commit^..$top_commit";
3067
3068 if ( !defined $tip or $top > $tip ) {
3069 $tip = $top;
3070 $tip_commit = $top_commit;
3071 }
3072 }
3073 return ($tip_commit, @merged_commit_ranges);
3074 }
3075
3076 sub _rev_list {
3077 my ($msg_fh, $ctx) = command_output_pipe(
3078 "rev-list", @_,
3079 );
3080 my @rv;
3081 while ( <$msg_fh> ) {
3082 chomp;
3083 push @rv, $_;
3084 }
3085 command_close_pipe($msg_fh, $ctx);
3086 @rv;
3087 }
3088
3089 sub check_cherry_pick {
3090 my $base = shift;
3091 my $tip = shift;
3092 my @ranges = @_;
3093 my %commits = map { $_ => 1 }
3094 _rev_list("--no-merges", $tip, "--not", $base);
3095 for my $range ( @ranges ) {
3096 delete @commits{_rev_list($range)};
3097 }
3098 for my $commit (keys %commits) {
3099 if (has_no_changes($commit)) {
3100 delete $commits{$commit};
3101 }
3102 }
3103 return (keys %commits);
3104 }
3105
3106 sub has_no_changes {
3107 my $commit = shift;
3108
3109 my @revs = split / /, command_oneline(
3110 qw(rev-list --parents -1 -m), $commit);
3111
3112 # Commits with no parents, e.g. the start of a partial branch,
3113 # have changes by definition.
3114 return 1 if (@revs < 2);
3115
3116 # Commits with multiple parents, e.g a merge, have no changes
3117 # by definition.
3118 return 0 if (@revs > 2);
3119
3120 return (command_oneline("rev-parse", "$commit^{tree}") eq
3121 command_oneline("rev-parse", "$commit~1^{tree}"));
3122 }
3123
3124 # The GIT_DIR environment variable is not always set until after the command
3125 # line arguments are processed, so we can't memoize in a BEGIN block.
3126 {
3127 my $memoized = 0;
3128
3129 sub memoize_svn_mergeinfo_functions {
3130 return if $memoized;
3131 $memoized = 1;
3132
3133 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3134 mkpath([$cache_path]) unless -d $cache_path;
3135
3136 tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3137 "$cache_path/lookup_svn_merge.db", 'nstore';
3138 memoize 'lookup_svn_merge',
3139 SCALAR_CACHE => 'FAULT',
3140 LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3141 ;
3142
3143 tie my %check_cherry_pick_cache => 'Memoize::Storable',
3144 "$cache_path/check_cherry_pick.db", 'nstore';
3145 memoize 'check_cherry_pick',
3146 SCALAR_CACHE => 'FAULT',
3147 LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3148 ;
3149
3150 tie my %has_no_changes_cache => 'Memoize::Storable',
3151 "$cache_path/has_no_changes.db", 'nstore';
3152 memoize 'has_no_changes',
3153 SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3154 LIST_CACHE => 'FAULT',
3155 ;
3156 }
3157 }
3158
3159 sub parents_exclude {
3160 my $parents = shift;
3161 my @commits = @_;
3162 return unless @commits;
3163
3164 my @excluded;
3165 my $excluded;
3166 do {
3167 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3168 $excluded = command_oneline(@cmd);
3169 if ( $excluded ) {
3170 my @new;
3171 my $found;
3172 for my $commit ( @commits ) {
3173 if ( $commit eq $excluded ) {
3174 push @excluded, $commit;
3175 $found++;
3176 last;
3177 }
3178 else {
3179 push @new, $commit;
3180 }
3181 }
3182 die "saw commit '$excluded' in rev-list output, "
3183 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3184 unless $found;
3185 @commits = @new;
3186 }
3187 }
3188 while ($excluded and @commits);
3189
3190 return @excluded;
3191 }
3192
3193
3194 # note: this function should only be called if the various dirprops
3195 # have actually changed
3196 sub find_extra_svn_parents {
3197 my ($self, $ed, $mergeinfo, $parents) = @_;
3198 # aha! svk:merge property changed...
3199
3200 memoize_svn_mergeinfo_functions();
3201
3202 # We first search for merged tips which are not in our
3203 # history. Then, we figure out which git revisions are in
3204 # that tip, but not this revision. If all of those revisions
3205 # are now marked as merge, we can add the tip as a parent.
3206 my @merges = split "\n", $mergeinfo;
3207 my @merge_tips;
3208 my $url = $self->rewrite_root || $self->{url};
3209 my $uuid = $self->ra_uuid;
3210 my %ranges;
3211 for my $merge ( @merges ) {
3212 my ($tip_commit, @ranges) =
3213 lookup_svn_merge( $uuid, $url, $merge );
3214 unless (!$tip_commit or
3215 grep { $_ eq $tip_commit } @$parents ) {
3216 push @merge_tips, $tip_commit;
3217 $ranges{$tip_commit} = \@ranges;
3218 } else {
3219 push @merge_tips, undef;
3220 }
3221 }
3222
3223 my %excluded = map { $_ => 1 }
3224 parents_exclude($parents, grep { defined } @merge_tips);
3225
3226 # check merge tips for new parents
3227 my @new_parents;
3228 for my $merge_tip ( @merge_tips ) {
3229 my $spec = shift @merges;
3230 next unless $merge_tip and $excluded{$merge_tip};
3231
3232 my $ranges = $ranges{$merge_tip};
3233
3234 # check out 'new' tips
3235 my $merge_base;
3236 eval {
3237 $merge_base = command_oneline(
3238 "merge-base",
3239 @$parents, $merge_tip,
3240 );
3241 };
3242 if ($@) {
3243 die "An error occurred during merge-base"
3244 unless $@->isa("Git::Error::Command");
3245
3246 warn "W: Cannot find common ancestor between ".
3247 "@$parents and $merge_tip. Ignoring merge info.\n";
3248 next;
3249 }
3250
3251 # double check that there are no missing non-merge commits
3252 my (@incomplete) = check_cherry_pick(
3253 $merge_base, $merge_tip,
3254 @$ranges,
3255 );
3256
3257 if ( @incomplete ) {
3258 warn "W:svn cherry-pick ignored ($spec) - missing "
3259 .@incomplete." commit(s) (eg $incomplete[0])\n";
3260 } else {
3261 warn
3262 "Found merge parent (svn:mergeinfo prop): ",
3263 $merge_tip, "\n";
3264 push @new_parents, $merge_tip;
3265 }
3266 }
3267
3268 # cater for merges which merge commits from multiple branches
3269 if ( @new_parents > 1 ) {
3270 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3271 for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3272 next if $i == $j;
3273 next unless $new_parents[$i];
3274 next unless $new_parents[$j];
3275 my $revs = command_oneline(
3276 "rev-list", "-1",
3277 "$new_parents[$i]..$new_parents[$j]",
3278 );
3279 if ( !$revs ) {
3280 undef($new_parents[$j]);
3281 }
3282 }
3283 }
3284 }
3285 push @$parents, grep { defined } @new_parents;
3286 }
3287
3288 sub make_log_entry {
3289 my ($self, $rev, $parents, $ed) = @_;
3290 my $untracked = $self->get_untracked($ed);
3291
3292 my @parents = @$parents;
3293 my $ps = $ed->{path_strip} || "";
3294 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3295 my $props = $ed->{dir_prop}{$path};
3296 if ( $props->{"svk:merge"} ) {
3297 $self->find_extra_svk_parents
3298 ($ed, $props->{"svk:merge"}, \@parents);
3299 }
3300 if ( $props->{"svn:mergeinfo"} ) {
3301 $self->find_extra_svn_parents
3302 ($ed,
3303 $props->{"svn:mergeinfo"},
3304 \@parents);
3305 }
3306 }
3307
3308 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3309 print $un "r$rev\n" or croak $!;
3310 print $un $_, "\n" foreach @$untracked;
3311 my %log_entry = ( parents => \@parents, revision => $rev,
3312 log => '');
3313
3314 my $headrev;
3315 my $logged = delete $self->{logged_rev_props};
3316 if (!$logged || $self->{-want_revprops}) {
3317 my $rp = $self->ra->rev_proplist($rev);
3318 foreach (sort keys %$rp) {
3319 my $v = $rp->{$_};
3320 if (/^svn:(author|date|log)$/) {
3321 $log_entry{$1} = $v;
3322 } elsif ($_ eq 'svm:headrev') {
3323 $headrev = $v;
3324 } else {
3325 print $un " rev_prop: ", uri_encode($_), ' ',
3326 uri_encode($v), "\n";
3327 }
3328 }
3329 } else {
3330 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3331 }
3332 close $un or croak $!;
3333
3334 $log_entry{date} = parse_svn_date($log_entry{date});
3335 $log_entry{log} .= "\n";
3336 my $author = $log_entry{author} = check_author($log_entry{author});
3337 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3338 : ($author, undef);
3339
3340 my ($commit_name, $commit_email) = ($name, $email);
3341 if ($_use_log_author) {
3342 my $name_field;
3343 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3344 $name_field = $1;
3345 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3346 $name_field = $1;
3347 }
3348 if (!defined $name_field) {
3349 if (!defined $email) {
3350 $email = $name;
3351 }
3352 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3353 ($name, $email) = ($1, $2);
3354 } elsif ($name_field =~ /(.*)@/) {
3355 ($name, $email) = ($1, $name_field);
3356 } else {
3357 ($name, $email) = ($name_field, $name_field);
3358 }
3359 }
3360 if (defined $headrev && $self->use_svm_props) {
3361 if ($self->rewrite_root) {
3362 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3363 "options set!\n";
3364 }
3365 if ($self->rewrite_uuid) {
3366 die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3367 "options set!\n";
3368 }
3369 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3370 # we don't want "SVM: initializing mirror for junk" ...
3371 return undef if $r == 0;
3372 my $svm = $self->svm;
3373 if ($uuid ne $svm->{uuid}) {
3374 die "UUID mismatch on SVM path:\n",
3375 "expected: $svm->{uuid}\n",
3376 " got: $uuid\n";
3377 }
3378 my $full_url = $self->full_url;
3379 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3380 die "Failed to replace '$svm->{replace}' with ",
3381 "'$svm->{source}' in $full_url\n";
3382 # throw away username for storing in records
3383 remove_username($full_url);
3384 $log_entry{metadata} = "$full_url\@$r $uuid";
3385 $log_entry{svm_revision} = $r;
3386 $email ||= "$author\@$uuid";
3387 $commit_email ||= "$author\@$uuid";
3388 } elsif ($self->use_svnsync_props) {
3389 my $full_url = $self->svnsync->{url};
3390 $full_url .= "/$self->{path}" if length $self->{path};
3391 remove_username($full_url);
3392 my $uuid = $self->svnsync->{uuid};
3393 $log_entry{metadata} = "$full_url\@$rev $uuid";
3394 $email ||= "$author\@$uuid";
3395 $commit_email ||= "$author\@$uuid";
3396 } else {
3397 my $url = $self->metadata_url;
3398 remove_username($url);
3399 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3400 $log_entry{metadata} = "$url\@$rev " . $uuid;
3401 $email ||= "$author\@" . $uuid;
3402 $commit_email ||= "$author\@" . $uuid;
3403 }
3404 $log_entry{name} = $name;
3405 $log_entry{email} = $email;
3406 $log_entry{commit_name} = $commit_name;
3407 $log_entry{commit_email} = $commit_email;
3408 \%log_entry;
3409 }
3410
3411 sub fetch {
3412 my ($self, $min_rev, $max_rev, @parents) = @_;
3413 my ($last_rev, $last_commit) = $self->last_rev_commit;
3414 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3415 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3416 }
3417
3418 sub set_tree_cb {
3419 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3420 $self->{inject_parents} = { $rev => $tree };
3421 $self->fetch(undef, undef);
3422 }
3423
3424 sub set_tree {
3425 my ($self, $tree) = (shift, shift);
3426 my $log_entry = ::get_commit_entry($tree);
3427 unless ($self->{last_rev}) {
3428 ::fatal("Must have an existing revision to commit");
3429 }
3430 my %ed_opts = ( r => $self->{last_rev},
3431 log => $log_entry->{log},
3432 ra => $self->ra,
3433 tree_a => $self->{last_commit},
3434 tree_b => $tree,
3435 editor_cb => sub {
3436 $self->set_tree_cb($log_entry, $tree, @_) },
3437 svn_path => $self->{path} );
3438 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3439 print "No changes\nr$self->{last_rev} = $tree\n";
3440 }
3441 }
3442
3443 sub rebuild_from_rev_db {
3444 my ($self, $path) = @_;
3445 my $r = -1;
3446 open my $fh, '<', $path or croak "open: $!";
3447 binmode $fh or croak "binmode: $!";
3448 while (<$fh>) {
3449 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3450 chomp($_);
3451 ++$r;
3452 next if $_ eq ('0' x 40);
3453 $self->rev_map_set($r, $_);
3454 print "r$r = $_\n";
3455 }
3456 close $fh or croak "close: $!";
3457 unlink $path or croak "unlink: $!";
3458 }
3459
3460 sub rebuild {
3461 my ($self) = @_;
3462 my $map_path = $self->map_path;
3463 my $partial = (-e $map_path && ! -z $map_path);
3464 return unless ::verify_ref($self->refname.'^0');
3465 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3466 my $rev_db = $self->rev_db_path;
3467 $self->rebuild_from_rev_db($rev_db);
3468 if ($self->use_svm_props) {
3469 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3470 $self->rebuild_from_rev_db($svm_rev_db);
3471 }
3472 $self->unlink_rev_db_symlink;
3473 return;
3474 }
3475 print "Rebuilding $map_path ...\n" if (!$partial);
3476 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3477 (undef, undef));
3478 my ($log, $ctx) =
3479 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3480 ($head ? "$head.." : "") . $self->refname,
3481 '--');
3482 my $metadata_url = $self->metadata_url;
3483 remove_username($metadata_url);
3484 my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3485 my $c;
3486 while (<$log>) {
3487 if ( m{^commit ($::sha1)$} ) {
3488 $c = $1;
3489 next;
3490 }
3491 next unless s{^\s*(git-svn-id:)}{$1};
3492 my ($url, $rev, $uuid) = ::extract_metadata($_);
3493 remove_username($url);
3494
3495 # ignore merges (from set-tree)
3496 next if (!defined $rev || !$uuid);
3497
3498 # if we merged or otherwise started elsewhere, this is
3499 # how we break out of it
3500 if (($uuid ne $svn_uuid) ||
3501 ($metadata_url && $url && ($url ne $metadata_url))) {
3502 next;
3503 }
3504 if ($partial && $head) {
3505 print "Partial-rebuilding $map_path ...\n";
3506 print "Currently at $base_rev = $head\n";
3507 $head = undef;
3508 }
3509
3510 $self->rev_map_set($rev, $c);
3511 print "r$rev = $c\n";
3512 }
3513 command_close_pipe($log, $ctx);
3514 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3515 my $rev_db_path = $self->rev_db_path;
3516 if (-f $self->rev_db_path) {
3517 unlink $self->rev_db_path or croak "unlink: $!";
3518 }
3519 $self->unlink_rev_db_symlink;
3520 }
3521
3522 # rev_map:
3523 # Tie::File seems to be prone to offset errors if revisions get sparse,
3524 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
3525 # one of my favorite modules is out :< Next up would be one of the DBM
3526 # modules, but I'm not sure which is most portable...
3527 #
3528 # This is the replacement for the rev_db format, which was too big
3529 # and inefficient for large repositories with a lot of sparse history
3530 # (mainly tags)
3531 #
3532 # The format is this:
3533 # - 24 bytes for every record,
3534 # * 4 bytes for the integer representing an SVN revision number
3535 # * 20 bytes representing the sha1 of a git commit
3536 # - No empty padding records like the old format
3537 # (except the last record, which can be overwritten)
3538 # - new records are written append-only since SVN revision numbers
3539 # increase monotonically
3540 # - lookups on SVN revision number are done via a binary search
3541 # - Piping the file to xxd -c24 is a good way of dumping it for
3542 # viewing or editing (piped back through xxd -r), should the need
3543 # ever arise.
3544 # - The last record can be padding revision with an all-zero sha1
3545 # This is used to optimize fetch performance when using multiple
3546 # "fetch" directives in .git/config
3547 #
3548 # These files are disposable unless noMetadata or useSvmProps is set
3549
3550 sub _rev_map_set {
3551 my ($fh, $rev, $commit) = @_;
3552
3553 binmode $fh or croak "binmode: $!";
3554 my $size = (stat($fh))[7];
3555 ($size % 24) == 0 or croak "inconsistent size: $size";
3556
3557 my $wr_offset = 0;
3558 if ($size > 0) {
3559 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3560 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3561 $read == 24 or croak "read only $read bytes (!= 24)";
3562 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3563 if ($last_commit eq ('0' x40)) {
3564 if ($size >= 48) {
3565 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3566 $read = sysread($fh, $buf, 24) or
3567 croak "read: $!";
3568 $read == 24 or
3569 croak "read only $read bytes (!= 24)";
3570 ($last_rev, $last_commit) =
3571 unpack(rev_map_fmt, $buf);
3572 if ($last_commit eq ('0' x40)) {
3573 croak "inconsistent .rev_map\n";
3574 }
3575 }
3576 if ($last_rev >= $rev) {
3577 croak "last_rev is higher!: $last_rev >= $rev";
3578 }
3579 $wr_offset = -24;
3580 }
3581 }
3582 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3583 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3584 croak "write: $!";
3585 }
3586
3587 sub _rev_map_reset {
3588 my ($fh, $rev, $commit) = @_;
3589 my $c = _rev_map_get($fh, $rev);
3590 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3591 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3592 truncate $fh, $offset or croak "truncate: $!";
3593 }
3594
3595 sub mkfile {
3596 my ($path) = @_;
3597 unless (-e $path) {
3598 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3599 mkpath([$dir]) unless -d $dir;
3600 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3601 close $fh or die "Couldn't close (create) $path: $!\n";
3602 }
3603 }
3604
3605 sub rev_map_set {
3606 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3607 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3608 my $db = $self->map_path($uuid);
3609 my $db_lock = "$db.lock";
3610 my $sig;
3611 $update_ref ||= 0;
3612 if ($update_ref) {
3613 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3614 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3615 }
3616 mkfile($db);
3617
3618 $LOCKFILES{$db_lock} = 1;
3619 my $sync;
3620 # both of these options make our .rev_db file very, very important
3621 # and we can't afford to lose it because rebuild() won't work
3622 if ($self->use_svm_props || $self->no_metadata) {
3623 $sync = 1;
3624 copy($db, $db_lock) or die "rev_map_set(@_): ",
3625 "Failed to copy: ",
3626 "$db => $db_lock ($!)\n";
3627 } else {
3628 rename $db, $db_lock or die "rev_map_set(@_): ",
3629 "Failed to rename: ",
3630 "$db => $db_lock ($!)\n";
3631 }
3632
3633 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3634 or croak "Couldn't open $db_lock: $!\n";
3635 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3636 _rev_map_set($fh, $rev, $commit);
3637 if ($sync) {
3638 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3639 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3640 }
3641 close $fh or croak $!;
3642 if ($update_ref) {
3643 $_head = $self;
3644 my $note = "";
3645 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3646 command_noisy('update-ref', '-m', "r$rev$note",
3647 $self->refname, $commit);
3648 }
3649 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3650 "$db_lock => $db ($!)\n";
3651 delete $LOCKFILES{$db_lock};
3652 if ($update_ref) {
3653 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3654 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3655 kill $sig, $$ if defined $sig;
3656 }
3657 }
3658
3659 # If want_commit, this will return an array of (rev, commit) where
3660 # commit _must_ be a valid commit in the archive.
3661 # Otherwise, it'll return the max revision (whether or not the
3662 # commit is valid or just a 0x40 placeholder).
3663 sub rev_map_max {
3664 my ($self, $want_commit) = @_;
3665 $self->rebuild;
3666 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3667 $want_commit ? ($r, $c) : $r;
3668 }
3669
3670 sub rev_map_max_norebuild {
3671 my ($self, $want_commit) = @_;
3672 my $map_path = $self->map_path;
3673 stat $map_path or return $want_commit ? (0, undef) : 0;
3674 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3675 binmode $fh or croak "binmode: $!";
3676 my $size = (stat($fh))[7];
3677 ($size % 24) == 0 or croak "inconsistent size: $size";
3678
3679 if ($size == 0) {
3680 close $fh or croak "close: $!";
3681 return $want_commit ? (0, undef) : 0;
3682 }
3683
3684 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3685 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3686 my ($r, $c) = unpack(rev_map_fmt, $buf);
3687 if ($want_commit && $c eq ('0' x40)) {
3688 if ($size < 48) {
3689 return $want_commit ? (0, undef) : 0;
3690 }
3691 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3692 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3693 ($r, $c) = unpack(rev_map_fmt, $buf);
3694 if ($c eq ('0'x40)) {
3695 croak "Penultimate record is all-zeroes in $map_path";
3696 }
3697 }
3698 close $fh or croak "close: $!";
3699 $want_commit ? ($r, $c) : $r;
3700 }
3701
3702 sub rev_map_get {
3703 my ($self, $rev, $uuid) = @_;
3704 my $map_path = $self->map_path($uuid);
3705 return undef unless -e $map_path;
3706
3707 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3708 my $c = _rev_map_get($fh, $rev);
3709 close($fh) or croak "close: $!";
3710 $c
3711 }
3712
3713 sub _rev_map_get {
3714 my ($fh, $rev) = @_;
3715
3716 binmode $fh or croak "binmode: $!";
3717 my $size = (stat($fh))[7];
3718 ($size % 24) == 0 or croak "inconsistent size: $size";
3719
3720 if ($size == 0) {
3721 return undef;
3722 }
3723
3724 my ($l, $u) = (0, $size - 24);
3725 my ($r, $c, $buf);
3726
3727 while ($l <= $u) {
3728 my $i = int(($l/24 + $u/24) / 2) * 24;
3729 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3730 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3731 my ($r, $c) = unpack(rev_map_fmt, $buf);
3732
3733 if ($r < $rev) {
3734 $l = $i + 24;
3735 } elsif ($r > $rev) {
3736 $u = $i - 24;
3737 } else { # $r == $rev
3738 return $c eq ('0' x 40) ? undef : $c;
3739 }
3740 }
3741 undef;
3742 }
3743
3744 # Finds the first svn revision that exists on (if $eq_ok is true) or
3745 # before $rev for the current branch. It will not search any lower
3746 # than $min_rev. Returns the git commit hash and svn revision number
3747 # if found, else (undef, undef).
3748 sub find_rev_before {
3749 my ($self, $rev, $eq_ok, $min_rev) = @_;
3750 --$rev unless $eq_ok;
3751 $min_rev ||= 1;
3752 my $max_rev = $self->rev_map_max;
3753 $rev = $max_rev if ($rev > $max_rev);
3754 while ($rev >= $min_rev) {
3755 if (my $c = $self->rev_map_get($rev)) {
3756 return ($rev, $c);
3757 }
3758 --$rev;
3759 }
3760 return (undef, undef);
3761 }
3762
3763 # Finds the first svn revision that exists on (if $eq_ok is true) or
3764 # after $rev for the current branch. It will not search any higher
3765 # than $max_rev. Returns the git commit hash and svn revision number
3766 # if found, else (undef, undef).
3767 sub find_rev_after {
3768 my ($self, $rev, $eq_ok, $max_rev) = @_;
3769 ++$rev unless $eq_ok;
3770 $max_rev ||= $self->rev_map_max;
3771 while ($rev <= $max_rev) {
3772 if (my $c = $self->rev_map_get($rev)) {
3773 return ($rev, $c);
3774 }
3775 ++$rev;
3776 }
3777 return (undef, undef);
3778 }
3779
3780 sub _new {
3781 my ($class, $repo_id, $ref_id, $path) = @_;
3782 unless (defined $repo_id && length $repo_id) {
3783 $repo_id = $Git::SVN::default_repo_id;
3784 }
3785 unless (defined $ref_id && length $ref_id) {
3786 $_prefix = '' unless defined($_prefix);
3787 $_[2] = $ref_id =
3788 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3789 }
3790 $_[1] = $repo_id;
3791 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3792
3793 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3794 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3795 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3796 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3797 if (-d $old_dir && ! -d $dir) {
3798 $dir = $old_dir;
3799 }
3800 }
3801
3802 $_[3] = $path = '' unless (defined $path);
3803 mkpath([$dir]);
3804 bless {
3805 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3806 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3807 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3808 }
3809
3810 # for read-only access of old .rev_db formats
3811 sub unlink_rev_db_symlink {
3812 my ($self) = @_;
3813 my $link = $self->rev_db_path;
3814 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3815 if (-l $link) {
3816 unlink $link or croak "unlink: $link failed!";
3817 }
3818 }
3819
3820 sub rev_db_path {
3821 my ($self, $uuid) = @_;
3822 my $db_path = $self->map_path($uuid);
3823 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3824 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3825 $db_path;
3826 }
3827
3828 # the new replacement for .rev_db
3829 sub map_path {
3830 my ($self, $uuid) = @_;
3831 $uuid ||= $self->ra_uuid;
3832 "$self->{map_root}.$uuid";
3833 }
3834
3835 sub uri_encode {
3836 my ($f) = @_;
3837 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3838 $f
3839 }
3840
3841 sub uri_decode {
3842 my ($f) = @_;
3843 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3844 $f
3845 }
3846
3847 sub remove_username {
3848 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3849 }
3850
3851 package Git::SVN::Prompt;
3852 use strict;
3853 use warnings;
3854 require SVN::Core;
3855 use vars qw/$_no_auth_cache $_username/;
3856
3857 sub simple {
3858 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3859 $may_save = undef if $_no_auth_cache;
3860 $default_username = $_username if defined $_username;
3861 if (defined $default_username && length $default_username) {
3862 if (defined $realm && length $realm) {
3863 print STDERR "Authentication realm: $realm\n";
3864 STDERR->flush;
3865 }
3866 $cred->username($default_username);
3867 } else {
3868 username($cred, $realm, $may_save, $pool);
3869 }
3870 $cred->password(_read_password("Password for '" .
3871 $cred->username . "': ", $realm));
3872 $cred->may_save($may_save);
3873 $SVN::_Core::SVN_NO_ERROR;
3874 }
3875
3876 sub ssl_server_trust {
3877 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3878 $may_save = undef if $_no_auth_cache;
3879 print STDERR "Error validating server certificate for '$realm':\n";
3880 {
3881 no warnings 'once';
3882 # All variables SVN::Auth::SSL::* are used only once,
3883 # so we're shutting up Perl warnings about this.
3884 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3885 print STDERR " - The certificate is not issued ",
3886 "by a trusted authority. Use the\n",
3887 " fingerprint to validate ",
3888 "the certificate manually!\n";
3889 }
3890 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3891 print STDERR " - The certificate hostname ",
3892 "does not match.\n";
3893 }
3894 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3895 print STDERR " - The certificate is not yet valid.\n";
3896 }
3897 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3898 print STDERR " - The certificate has expired.\n";
3899 }
3900 if ($failures & $SVN::Auth::SSL::OTHER) {
3901 print STDERR " - The certificate has ",
3902 "an unknown error.\n";
3903 }
3904 } # no warnings 'once'
3905 printf STDERR
3906 "Certificate information:\n".
3907 " - Hostname: %s\n".
3908 " - Valid: from %s until %s\n".
3909 " - Issuer: %s\n".
3910 " - Fingerprint: %s\n",
3911 map $cert_info->$_, qw(hostname valid_from valid_until
3912 issuer_dname fingerprint);
3913 my $choice;
3914 prompt:
3915 print STDERR $may_save ?
3916 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3917 "(R)eject or accept (t)emporarily? ";
3918 STDERR->flush;
3919 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3920 if ($choice =~ /^t$/i) {
3921 $cred->may_save(undef);
3922 } elsif ($choice =~ /^r$/i) {
3923 return -1;
3924 } elsif ($may_save && $choice =~ /^p$/i) {
3925 $cred->may_save($may_save);
3926 } else {
3927 goto prompt;
3928 }
3929 $cred->accepted_failures($failures);
3930 $SVN::_Core::SVN_NO_ERROR;
3931 }
3932
3933 sub ssl_client_cert {
3934 my ($cred, $realm, $may_save, $pool) = @_;
3935 $may_save = undef if $_no_auth_cache;
3936 print STDERR "Client certificate filename: ";
3937 STDERR->flush;
3938 chomp(my $filename = <STDIN>);
3939 $cred->cert_file($filename);
3940 $cred->may_save($may_save);
3941 $SVN::_Core::SVN_NO_ERROR;
3942 }
3943
3944 sub ssl_client_cert_pw {
3945 my ($cred, $realm, $may_save, $pool) = @_;
3946 $may_save = undef if $_no_auth_cache;
3947 $cred->password(_read_password("Password: ", $realm));
3948 $cred->may_save($may_save);
3949 $SVN::_Core::SVN_NO_ERROR;
3950 }
3951
3952 sub username {
3953 my ($cred, $realm, $may_save, $pool) = @_;
3954 $may_save = undef if $_no_auth_cache;
3955 if (defined $realm && length $realm) {
3956 print STDERR "Authentication realm: $realm\n";
3957 }
3958 my $username;
3959 if (defined $_username) {
3960 $username = $_username;
3961 } else {
3962 print STDERR "Username: ";
3963 STDERR->flush;
3964 chomp($username = <STDIN>);
3965 }
3966 $cred->username($username);
3967 $cred->may_save($may_save);
3968 $SVN::_Core::SVN_NO_ERROR;
3969 }
3970
3971 sub _read_password {
3972 my ($prompt, $realm) = @_;
3973 print STDERR $prompt;
3974 STDERR->flush;
3975 require Term::ReadKey;
3976 Term::ReadKey::ReadMode('noecho');
3977 my $password = '';
3978 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3979 last if $key =~ /[\012\015]/; # \n\r
3980 $password .= $key;
3981 }
3982 Term::ReadKey::ReadMode('restore');
3983 print STDERR "\n";
3984 STDERR->flush;
3985 $password;
3986 }
3987
3988 package SVN::Git::Fetcher;
3989 use vars qw/@ISA/;
3990 use strict;
3991 use warnings;
3992 use Carp qw/croak/;
3993 use File::Temp qw/tempfile/;
3994 use IO::File qw//;
3995 use vars qw/$_ignore_regex/;
3996
3997 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
3998 sub new {
3999 my ($class, $git_svn, $switch_path) = @_;
4000 my $self = SVN::Delta::Editor->new;
4001 bless $self, $class;
4002 if (exists $git_svn->{last_commit}) {
4003 $self->{c} = $git_svn->{last_commit};
4004 $self->{empty_symlinks} =
4005 _mark_empty_symlinks($git_svn, $switch_path);
4006 }
4007 $self->{ignore_regex} = eval { command_oneline('config', '--get',
4008 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
4009 $self->{empty} = {};
4010 $self->{dir_prop} = {};
4011 $self->{file_prop} = {};
4012 $self->{absent_dir} = {};
4013 $self->{absent_file} = {};
4014 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4015 $self;
4016 }
4017
4018 # this uses the Ra object, so it must be called before do_{switch,update},
4019 # not inside them (when the Git::SVN::Fetcher object is passed) to
4020 # do_{switch,update}
4021 sub _mark_empty_symlinks {
4022 my ($git_svn, $switch_path) = @_;
4023 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4024 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4025
4026 my %ret;
4027 my ($rev, $cmt) = $git_svn->last_rev_commit;
4028 return {} unless ($rev && $cmt);
4029
4030 # allow the warning to be printed for each revision we fetch to
4031 # ensure the user sees it. The user can also disable the workaround
4032 # on the repository even while git svn is running and the next
4033 # revision fetched will skip this expensive function.
4034 my $printed_warning;
4035 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4036 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4037 local $/ = "\0";
4038 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4039 $pfx .= '/' if length($pfx);
4040 while (<$ls>) {
4041 chomp;
4042 s/\A100644 blob $empty_blob\t//o or next;
4043 unless ($printed_warning) {
4044 print STDERR "Scanning for empty symlinks, ",
4045 "this may take a while if you have ",
4046 "many empty files\n",
4047 "You may disable this with `",
4048 "git config svn.brokenSymlinkWorkaround ",
4049 "false'.\n",
4050 "This may be done in a different ",
4051 "terminal without restarting ",
4052 "git svn\n";
4053 $printed_warning = 1;
4054 }
4055 my $path = $_;
4056 my (undef, $props) =
4057 $git_svn->ra->get_file($pfx.$path, $rev, undef);
4058 if ($props->{'svn:special'}) {
4059 $ret{$path} = 1;
4060 }
4061 }
4062 command_close_pipe($ls, $ctx);
4063 \%ret;
4064 }
4065
4066 # returns true if a given path is inside a ".git" directory
4067 sub in_dot_git {
4068 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4069 }
4070
4071 # return value: 0 -- don't ignore, 1 -- ignore
4072 sub is_path_ignored {
4073 my ($self, $path) = @_;
4074 return 1 if in_dot_git($path);
4075 return 1 if defined($self->{ignore_regex}) &&
4076 $path =~ m!$self->{ignore_regex}!;
4077 return 0 unless defined($_ignore_regex);
4078 return 1 if $path =~ m!$_ignore_regex!o;
4079 return 0;
4080 }
4081
4082 sub set_path_strip {
4083 my ($self, $path) = @_;
4084 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4085 }
4086
4087 sub open_root {
4088 { path => '' };
4089 }
4090
4091 sub open_directory {
4092 my ($self, $path, $pb, $rev) = @_;
4093 { path => $path };
4094 }
4095
4096 sub git_path {
4097 my ($self, $path) = @_;
4098 if ($self->{path_strip}) {
4099 $path =~ s!$self->{path_strip}!! or
4100 die "Failed to strip path '$path' ($self->{path_strip})\n";
4101 }
4102 $path;
4103 }
4104
4105 sub delete_entry {
4106 my ($self, $path, $rev, $pb) = @_;
4107 return undef if $self->is_path_ignored($path);
4108
4109 my $gpath = $self->git_path($path);
4110 return undef if ($gpath eq '');
4111
4112 # remove entire directories.
4113 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4114 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4115 if ($tree) {
4116 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4117 -r --name-only -z/,
4118 $tree);
4119 local $/ = "\0";
4120 while (<$ls>) {
4121 chomp;
4122 my $rmpath = "$gpath/$_";
4123 $self->{gii}->remove($rmpath);
4124 print "\tD\t$rmpath\n" unless $::_q;
4125 }
4126 print "\tD\t$gpath/\n" unless $::_q;
4127 command_close_pipe($ls, $ctx);
4128 } else {
4129 $self->{gii}->remove($gpath);
4130 print "\tD\t$gpath\n" unless $::_q;
4131 }
4132 $self->{empty}->{$path} = 0;
4133 undef;
4134 }
4135
4136 sub open_file {
4137 my ($self, $path, $pb, $rev) = @_;
4138 my ($mode, $blob);
4139
4140 goto out if $self->is_path_ignored($path);
4141
4142 my $gpath = $self->git_path($path);
4143 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4144 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4145 unless (defined $mode && defined $blob) {
4146 die "$path was not found in commit $self->{c} (r$rev)\n";
4147 }
4148 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4149 $mode = '120000';
4150 }
4151 out:
4152 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4153 pool => SVN::Pool->new, action => 'M' };
4154 }
4155
4156 sub add_file {
4157 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4158 my $mode;
4159
4160 if (!$self->is_path_ignored($path)) {
4161 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4162 delete $self->{empty}->{$dir};
4163 $mode = '100644';
4164 }
4165 { path => $path, mode_a => $mode, mode_b => $mode,
4166 pool => SVN::Pool->new, action => 'A' };
4167 }
4168
4169 sub add_directory {
4170 my ($self, $path, $cp_path, $cp_rev) = @_;
4171 goto out if $self->is_path_ignored($path);
4172 my $gpath = $self->git_path($path);
4173 if ($gpath eq '') {
4174 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4175 -r --name-only -z/,
4176 $self->{c});
4177 local $/ = "\0";
4178 while (<$ls>) {
4179 chomp;
4180 $self->{gii}->remove($_);
4181 print "\tD\t$_\n" unless $::_q;
4182 }
4183 command_close_pipe($ls, $ctx);
4184 $self->{empty}->{$path} = 0;
4185 }
4186 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4187 delete $self->{empty}->{$dir};
4188 $self->{empty}->{$path} = 1;
4189 out:
4190 { path => $path };
4191 }
4192
4193 sub change_dir_prop {
4194 my ($self, $db, $prop, $value) = @_;
4195 return undef if $self->is_path_ignored($db->{path});
4196 $self->{dir_prop}->{$db->{path}} ||= {};
4197 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4198 undef;
4199 }
4200
4201 sub absent_directory {
4202 my ($self, $path, $pb) = @_;
4203 return undef if $self->is_path_ignored($path);
4204 $self->{absent_dir}->{$pb->{path}} ||= [];
4205 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4206 undef;
4207 }
4208
4209 sub absent_file {
4210 my ($self, $path, $pb) = @_;
4211 return undef if $self->is_path_ignored($path);
4212 $self->{absent_file}->{$pb->{path}} ||= [];
4213 push @{$self->{absent_file}->{$pb->{path}}}, $path;
4214 undef;
4215 }
4216
4217 sub change_file_prop {
4218 my ($self, $fb, $prop, $value) = @_;
4219 return undef if $self->is_path_ignored($fb->{path});
4220 if ($prop eq 'svn:executable') {
4221 if ($fb->{mode_b} != 120000) {
4222 $fb->{mode_b} = defined $value ? 100755 : 100644;
4223 }
4224 } elsif ($prop eq 'svn:special') {
4225 $fb->{mode_b} = defined $value ? 120000 : 100644;
4226 } else {
4227 $self->{file_prop}->{$fb->{path}} ||= {};
4228 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4229 }
4230 undef;
4231 }
4232
4233 sub apply_textdelta {
4234 my ($self, $fb, $exp) = @_;
4235 return undef if $self->is_path_ignored($fb->{path});
4236 my $fh = $::_repository->temp_acquire('svn_delta');
4237 # $fh gets auto-closed() by SVN::TxDelta::apply(),
4238 # (but $base does not,) so dup() it for reading in close_file
4239 open my $dup, '<&', $fh or croak $!;
4240 my $base = $::_repository->temp_acquire('git_blob');
4241
4242 if ($fb->{blob}) {
4243 my ($base_is_link, $size);
4244
4245 if ($fb->{mode_a} eq '120000' &&
4246 ! $self->{empty_symlinks}->{$fb->{path}}) {
4247 print $base 'link ' or die "print $!\n";
4248 $base_is_link = 1;
4249 }
4250 retry:
4251 $size = $::_repository->cat_blob($fb->{blob}, $base);
4252 die "Failed to read object $fb->{blob}" if ($size < 0);
4253
4254 if (defined $exp) {
4255 seek $base, 0, 0 or croak $!;
4256 my $got = ::md5sum($base);
4257 if ($got ne $exp) {
4258 my $err = "Checksum mismatch: ".
4259 "$fb->{path} $fb->{blob}\n" .
4260 "expected: $exp\n" .
4261 " got: $got\n";
4262 if ($base_is_link) {
4263 warn $err,
4264 "Retrying... (possibly ",
4265 "a bad symlink from SVN)\n";
4266 $::_repository->temp_reset($base);
4267 $base_is_link = 0;
4268 goto retry;
4269 }
4270 die $err;
4271 }
4272 }
4273 }
4274 seek $base, 0, 0 or croak $!;
4275 $fb->{fh} = $fh;
4276 $fb->{base} = $base;
4277 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4278 }
4279
4280 sub close_file {
4281 my ($self, $fb, $exp) = @_;
4282 return undef if $self->is_path_ignored($fb->{path});
4283
4284 my $hash;
4285 my $path = $self->git_path($fb->{path});
4286 if (my $fh = $fb->{fh}) {
4287 if (defined $exp) {
4288 seek($fh, 0, 0) or croak $!;
4289 my $got = ::md5sum($fh);
4290 if ($got ne $exp) {
4291 die "Checksum mismatch: $path\n",
4292 "expected: $exp\n got: $got\n";
4293 }
4294 }
4295 if ($fb->{mode_b} == 120000) {
4296 sysseek($fh, 0, 0) or croak $!;
4297 my $rd = sysread($fh, my $buf, 5);
4298
4299 if (!defined $rd) {
4300 croak "sysread: $!\n";
4301 } elsif ($rd == 0) {
4302 warn "$path has mode 120000",
4303 " but it points to nothing\n",
4304 "converting to an empty file with mode",
4305 " 100644\n";
4306 $fb->{mode_b} = '100644';
4307 } elsif ($buf ne 'link ') {
4308 warn "$path has mode 120000",
4309 " but is not a link\n";
4310 } else {
4311 my $tmp_fh = $::_repository->temp_acquire(
4312 'svn_hash');
4313 my $res;
4314 while ($res = sysread($fh, my $str, 1024)) {
4315 my $out = syswrite($tmp_fh, $str, $res);
4316 defined($out) && $out == $res
4317 or croak("write ",
4318 Git::temp_path($tmp_fh),
4319 ": $!\n");
4320 }
4321 defined $res or croak $!;
4322
4323 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4324 Git::temp_release($tmp_fh, 1);
4325 }
4326 }
4327
4328 $hash = $::_repository->hash_and_insert_object(
4329 Git::temp_path($fh));
4330 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4331
4332 Git::temp_release($fb->{base}, 1);
4333 Git::temp_release($fh, 1);
4334 } else {
4335 $hash = $fb->{blob} or die "no blob information\n";
4336 }
4337 $fb->{pool}->clear;
4338 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4339 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4340 undef;
4341 }
4342
4343 sub abort_edit {
4344 my $self = shift;
4345 $self->{nr} = $self->{gii}->{nr};
4346 delete $self->{gii};
4347 $self->SUPER::abort_edit(@_);
4348 }
4349
4350 sub close_edit {
4351 my $self = shift;
4352 $self->{git_commit_ok} = 1;
4353 $self->{nr} = $self->{gii}->{nr};
4354 delete $self->{gii};
4355 $self->SUPER::close_edit(@_);
4356 }
4357
4358 package SVN::Git::Editor;
4359 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4360 use strict;
4361 use warnings;
4362 use Carp qw/croak/;
4363 use IO::File;
4364
4365 sub new {
4366 my ($class, $opts) = @_;
4367 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4368 die "$_ required!\n" unless (defined $opts->{$_});
4369 }
4370
4371 my $pool = SVN::Pool->new;
4372 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4373 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4374 $opts->{r}, $mods);
4375
4376 # $opts->{ra} functions should not be used after this:
4377 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4378 $opts->{editor_cb}, $pool);
4379 my $self = SVN::Delta::Editor->new(@ce, $pool);
4380 bless $self, $class;
4381 foreach (qw/svn_path r tree_a tree_b/) {
4382 $self->{$_} = $opts->{$_};
4383 }
4384 $self->{url} = $opts->{ra}->{url};
4385 $self->{mods} = $mods;
4386 $self->{types} = $types;
4387 $self->{pool} = $pool;
4388 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4389 $self->{rm} = { };
4390 $self->{path_prefix} = length $self->{svn_path} ?
4391 "$self->{svn_path}/" : '';
4392 $self->{config} = $opts->{config};
4393 return $self;
4394 }
4395
4396 sub generate_diff {
4397 my ($tree_a, $tree_b) = @_;
4398 my @diff_tree = qw(diff-tree -z -r);
4399 if ($_cp_similarity) {
4400 push @diff_tree, "-C$_cp_similarity";
4401 } else {
4402 push @diff_tree, '-C';
4403 }
4404 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4405 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4406 push @diff_tree, $tree_a, $tree_b;
4407 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4408 local $/ = "\0";
4409 my $state = 'meta';
4410 my @mods;
4411 while (<$diff_fh>) {
4412 chomp $_; # this gets rid of the trailing "\0"
4413 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4414 ($::sha1)\s($::sha1)\s
4415 ([MTCRAD])\d*$/xo) {
4416 push @mods, { mode_a => $1, mode_b => $2,
4417 sha1_a => $3, sha1_b => $4,
4418 chg => $5 };
4419 if ($5 =~ /^(?:C|R)$/) {
4420 $state = 'file_a';
4421 } else {
4422 $state = 'file_b';
4423 }
4424 } elsif ($state eq 'file_a') {
4425 my $x = $mods[$#mods] or croak "Empty array\n";
4426 if ($x->{chg} !~ /^(?:C|R)$/) {
4427 croak "Error parsing $_, $x->{chg}\n";
4428 }
4429 $x->{file_a} = $_;
4430 $state = 'file_b';
4431 } elsif ($state eq 'file_b') {
4432 my $x = $mods[$#mods] or croak "Empty array\n";
4433 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4434 croak "Error parsing $_, $x->{chg}\n";
4435 }
4436 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4437 croak "Error parsing $_, $x->{chg}\n";
4438 }
4439 $x->{file_b} = $_;
4440 $state = 'meta';
4441 } else {
4442 croak "Error parsing $_\n";
4443 }
4444 }
4445 command_close_pipe($diff_fh, $ctx);
4446 \@mods;
4447 }
4448
4449 sub check_diff_paths {
4450 my ($ra, $pfx, $rev, $mods) = @_;
4451 my %types;
4452 $pfx .= '/' if length $pfx;
4453
4454 sub type_diff_paths {
4455 my ($ra, $types, $path, $rev) = @_;
4456 my @p = split m#/+#, $path;
4457 my $c = shift @p;
4458 unless (defined $types->{$c}) {
4459 $types->{$c} = $ra->check_path($c, $rev);
4460 }
4461 while (@p) {
4462 $c .= '/' . shift @p;
4463 next if defined $types->{$c};
4464 $types->{$c} = $ra->check_path($c, $rev);
4465 }
4466 }
4467
4468 foreach my $m (@$mods) {
4469 foreach my $f (qw/file_a file_b/) {
4470 next unless defined $m->{$f};
4471 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4472 if (length $pfx.$dir && ! defined $types{$dir}) {
4473 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4474 }
4475 }
4476 }
4477 \%types;
4478 }
4479
4480 sub split_path {
4481 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4482 }
4483
4484 sub repo_path {
4485 my ($self, $path) = @_;
4486 $self->{path_prefix}.(defined $path ? $path : '');
4487 }
4488
4489 sub url_path {
4490 my ($self, $path) = @_;
4491 if ($self->{url} =~ m#^https?://#) {
4492 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4493 }
4494 $self->{url} . '/' . $self->repo_path($path);
4495 }
4496
4497 sub rmdirs {
4498 my ($self) = @_;
4499 my $rm = $self->{rm};
4500 delete $rm->{''}; # we never delete the url we're tracking
4501 return unless %$rm;
4502
4503 foreach (keys %$rm) {
4504 my @d = split m#/#, $_;
4505 my $c = shift @d;
4506 $rm->{$c} = 1;
4507 while (@d) {
4508 $c .= '/' . shift @d;
4509 $rm->{$c} = 1;
4510 }
4511 }
4512 delete $rm->{$self->{svn_path}};
4513 delete $rm->{''}; # we never delete the url we're tracking
4514 return unless %$rm;
4515
4516 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4517 $self->{tree_b});
4518 local $/ = "\0";
4519 while (<$fh>) {
4520 chomp;
4521 my @dn = split m#/#, $_;
4522 while (pop @dn) {
4523 delete $rm->{join '/', @dn};
4524 }
4525 unless (%$rm) {
4526 close $fh;
4527 return;
4528 }
4529 }
4530 command_close_pipe($fh, $ctx);
4531
4532 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4533 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4534 $self->close_directory($bat->{$d}, $p);
4535 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4536 print "\tD+\t$d/\n" unless $::_q;
4537 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4538 delete $bat->{$d};
4539 }
4540 }
4541
4542 sub open_or_add_dir {
4543 my ($self, $full_path, $baton) = @_;
4544 my $t = $self->{types}->{$full_path};
4545 if (!defined $t) {
4546 die "$full_path not known in r$self->{r} or we have a bug!\n";
4547 }
4548 {
4549 no warnings 'once';
4550 # SVN::Node::none and SVN::Node::file are used only once,
4551 # so we're shutting up Perl's warnings about them.
4552 if ($t == $SVN::Node::none) {
4553 return $self->add_directory($full_path, $baton,
4554 undef, -1, $self->{pool});
4555 } elsif ($t == $SVN::Node::dir) {
4556 return $self->open_directory($full_path, $baton,
4557 $self->{r}, $self->{pool});
4558 } # no warnings 'once'
4559 print STDERR "$full_path already exists in repository at ",
4560 "r$self->{r} and it is not a directory (",
4561 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4562 } # no warnings 'once'
4563 exit 1;
4564 }
4565
4566 sub ensure_path {
4567 my ($self, $path) = @_;
4568 my $bat = $self->{bat};
4569 my $repo_path = $self->repo_path($path);
4570 return $bat->{''} unless (length $repo_path);
4571 my @p = split m#/+#, $repo_path;
4572 my $c = shift @p;
4573 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4574 while (@p) {
4575 my $c0 = $c;
4576 $c .= '/' . shift @p;
4577 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4578 }
4579 return $bat->{$c};
4580 }
4581
4582 # Subroutine to convert a globbing pattern to a regular expression.
4583 # From perl cookbook.
4584 sub glob2pat {
4585 my $globstr = shift;
4586 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4587 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4588 return '^' . $globstr . '$';
4589 }
4590
4591 sub check_autoprop {
4592 my ($self, $pattern, $properties, $file, $fbat) = @_;
4593 # Convert the globbing pattern to a regular expression.
4594 my $regex = glob2pat($pattern);
4595 # Check if the pattern matches the file name.
4596 if($file =~ m/($regex)/) {
4597 # Parse the list of properties to set.
4598 my @props = split(/;/, $properties);
4599 foreach my $prop (@props) {
4600 # Parse 'name=value' syntax and set the property.
4601 if ($prop =~ /([^=]+)=(.*)/) {
4602 my ($n,$v) = ($1,$2);
4603 for ($n, $v) {
4604 s/^\s+//; s/\s+$//;
4605 }
4606 $self->change_file_prop($fbat, $n, $v);
4607 }
4608 }
4609 }
4610 }
4611
4612 sub apply_autoprops {
4613 my ($self, $file, $fbat) = @_;
4614 my $conf_t = ${$self->{config}}{'config'};
4615 no warnings 'once';
4616 # Check [miscellany]/enable-auto-props in svn configuration.
4617 if (SVN::_Core::svn_config_get_bool(
4618 $conf_t,
4619 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4620 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4621 0)) {
4622 # Auto-props are enabled. Enumerate them to look for matches.
4623 my $callback = sub {
4624 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4625 };
4626 SVN::_Core::svn_config_enumerate(
4627 $conf_t,
4628 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4629 $callback);
4630 }
4631 }
4632
4633 sub A {
4634 my ($self, $m) = @_;
4635 my ($dir, $file) = split_path($m->{file_b});
4636 my $pbat = $self->ensure_path($dir);
4637 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4638 undef, -1);
4639 print "\tA\t$m->{file_b}\n" unless $::_q;
4640 $self->apply_autoprops($file, $fbat);
4641 $self->chg_file($fbat, $m);
4642 $self->close_file($fbat,undef,$self->{pool});
4643 }
4644
4645 sub C {
4646 my ($self, $m) = @_;
4647 my ($dir, $file) = split_path($m->{file_b});
4648 my $pbat = $self->ensure_path($dir);
4649 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4650 $self->url_path($m->{file_a}), $self->{r});
4651 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4652 $self->chg_file($fbat, $m);
4653 $self->close_file($fbat,undef,$self->{pool});
4654 }
4655
4656 sub delete_entry {
4657 my ($self, $path, $pbat) = @_;
4658 my $rpath = $self->repo_path($path);
4659 my ($dir, $file) = split_path($rpath);
4660 $self->{rm}->{$dir} = 1;
4661 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4662 }
4663
4664 sub R {
4665 my ($self, $m) = @_;
4666 my ($dir, $file) = split_path($m->{file_b});
4667 my $pbat = $self->ensure_path($dir);
4668 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4669 $self->url_path($m->{file_a}), $self->{r});
4670 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4671 $self->apply_autoprops($file, $fbat);
4672 $self->chg_file($fbat, $m);
4673 $self->close_file($fbat,undef,$self->{pool});
4674
4675 ($dir, $file) = split_path($m->{file_a});
4676 $pbat = $self->ensure_path($dir);
4677 $self->delete_entry($m->{file_a}, $pbat);
4678 }
4679
4680 sub M {
4681 my ($self, $m) = @_;
4682 my ($dir, $file) = split_path($m->{file_b});
4683 my $pbat = $self->ensure_path($dir);
4684 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4685 $pbat,$self->{r},$self->{pool});
4686 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4687 $self->chg_file($fbat, $m);
4688 $self->close_file($fbat,undef,$self->{pool});
4689 }
4690
4691 sub T { shift->M(@_) }
4692
4693 sub change_file_prop {
4694 my ($self, $fbat, $pname, $pval) = @_;
4695 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4696 }
4697
4698 sub _chg_file_get_blob ($$$$) {
4699 my ($self, $fbat, $m, $which) = @_;
4700 my $fh = $::_repository->temp_acquire("git_blob_$which");
4701 if ($m->{"mode_$which"} =~ /^120/) {
4702 print $fh 'link ' or croak $!;
4703 $self->change_file_prop($fbat,'svn:special','*');
4704 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4705 $self->change_file_prop($fbat,'svn:special',undef);
4706 }
4707 my $blob = $m->{"sha1_$which"};
4708 return ($fh,) if ($blob =~ /^0{40}$/);
4709 my $size = $::_repository->cat_blob($blob, $fh);
4710 croak "Failed to read object $blob" if ($size < 0);
4711 $fh->flush == 0 or croak $!;
4712 seek $fh, 0, 0 or croak $!;
4713
4714 my $exp = ::md5sum($fh);
4715 seek $fh, 0, 0 or croak $!;
4716 return ($fh, $exp);
4717 }
4718
4719 sub chg_file {
4720 my ($self, $fbat, $m) = @_;
4721 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4722 $self->change_file_prop($fbat,'svn:executable','*');
4723 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4724 $self->change_file_prop($fbat,'svn:executable',undef);
4725 }
4726 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4727 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4728 my $pool = SVN::Pool->new;
4729 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4730 if (-s $fh_a) {
4731 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4732 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4733 if (defined $res) {
4734 die "Unexpected result from send_txstream: $res\n",
4735 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4736 }
4737 } else {
4738 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4739 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4740 if ($got ne $exp_b);
4741 }
4742 Git::temp_release($fh_b, 1);
4743 Git::temp_release($fh_a, 1);
4744 $pool->clear;
4745 }
4746
4747 sub D {
4748 my ($self, $m) = @_;
4749 my ($dir, $file) = split_path($m->{file_b});
4750 my $pbat = $self->ensure_path($dir);
4751 print "\tD\t$m->{file_b}\n" unless $::_q;
4752 $self->delete_entry($m->{file_b}, $pbat);
4753 }
4754
4755 sub close_edit {
4756 my ($self) = @_;
4757 my ($p,$bat) = ($self->{pool}, $self->{bat});
4758 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4759 next if $_ eq '';
4760 $self->close_directory($bat->{$_}, $p);
4761 }
4762 $self->close_directory($bat->{''}, $p);
4763 $self->SUPER::close_edit($p);
4764 $p->clear;
4765 }
4766
4767 sub abort_edit {
4768 my ($self) = @_;
4769 $self->SUPER::abort_edit($self->{pool});
4770 }
4771
4772 sub DESTROY {
4773 my $self = shift;
4774 $self->SUPER::DESTROY(@_);
4775 $self->{pool}->clear;
4776 }
4777
4778 # this drives the editor
4779 sub apply_diff {
4780 my ($self) = @_;
4781 my $mods = $self->{mods};
4782 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4783 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4784 my $f = $m->{chg};
4785 if (defined $o{$f}) {
4786 $self->$f($m);
4787 } else {
4788 fatal("Invalid change type: $f");
4789 }
4790 }
4791 $self->rmdirs if $_rmdir;
4792 if (@$mods == 0) {
4793 $self->abort_edit;
4794 } else {
4795 $self->close_edit;
4796 }
4797 return scalar @$mods;
4798 }
4799
4800 package Git::SVN::Ra;
4801 use vars qw/@ISA $config_dir $_log_window_size/;
4802 use strict;
4803 use warnings;
4804 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4805
4806 BEGIN {
4807 # enforce temporary pool usage for some simple functions
4808 no strict 'refs';
4809 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4810 get_file/) {
4811 my $SUPER = "SUPER::$f";
4812 *$f = sub {
4813 my $self = shift;
4814 my $pool = SVN::Pool->new;
4815 my @ret = $self->$SUPER(@_,$pool);
4816 $pool->clear;
4817 wantarray ? @ret : $ret[0];
4818 };
4819 }
4820 }
4821
4822 sub _auth_providers () {
4823 [
4824 SVN::Client::get_simple_provider(),
4825 SVN::Client::get_ssl_server_trust_file_provider(),
4826 SVN::Client::get_simple_prompt_provider(
4827 \&Git::SVN::Prompt::simple, 2),
4828 SVN::Client::get_ssl_client_cert_file_provider(),
4829 SVN::Client::get_ssl_client_cert_prompt_provider(
4830 \&Git::SVN::Prompt::ssl_client_cert, 2),
4831 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4832 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4833 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4834 SVN::Client::get_username_provider(),
4835 SVN::Client::get_ssl_server_trust_prompt_provider(
4836 \&Git::SVN::Prompt::ssl_server_trust),
4837 SVN::Client::get_username_prompt_provider(
4838 \&Git::SVN::Prompt::username, 2)
4839 ]
4840 }
4841
4842 sub escape_uri_only {
4843 my ($uri) = @_;
4844 my @tmp;
4845 foreach (split m{/}, $uri) {
4846 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4847 push @tmp, $_;
4848 }
4849 join('/', @tmp);
4850 }
4851
4852 sub escape_url {
4853 my ($url) = @_;
4854 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4855 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4856 $url = "$scheme://$domain$uri";
4857 }
4858 $url;
4859 }
4860
4861 sub new {
4862 my ($class, $url) = @_;
4863 $url =~ s!/+$!!;
4864 return $RA if ($RA && $RA->{url} eq $url);
4865
4866 ::_req_svn();
4867
4868 SVN::_Core::svn_config_ensure($config_dir, undef);
4869 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4870 my $config = SVN::Core::config_get_config($config_dir);
4871 $RA = undef;
4872 my $dont_store_passwords = 1;
4873 my $conf_t = ${$config}{'config'};
4874 {
4875 no warnings 'once';
4876 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4877 # produces warnings that variables are used only once.
4878 # I had not found the better way to shut them up, so
4879 # the warnings of type 'once' are disabled in this block.
4880 if (SVN::_Core::svn_config_get_bool($conf_t,
4881 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4882 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4883 1) == 0) {
4884 SVN::_Core::svn_auth_set_parameter($baton,
4885 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4886 bless (\$dont_store_passwords, "_p_void"));
4887 }
4888 if (SVN::_Core::svn_config_get_bool($conf_t,
4889 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4890 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4891 1) == 0) {
4892 $Git::SVN::Prompt::_no_auth_cache = 1;
4893 }
4894 } # no warnings 'once'
4895 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4896 config => $config,
4897 pool => SVN::Pool->new,
4898 auth_provider_callbacks => $callbacks);
4899 $self->{url} = $url;
4900 $self->{svn_path} = $url;
4901 $self->{repos_root} = $self->get_repos_root;
4902 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4903 $self->{cache} = { check_path => { r => 0, data => {} },
4904 get_dir => { r => 0, data => {} } };
4905 $RA = bless $self, $class;
4906 }
4907
4908 sub check_path {
4909 my ($self, $path, $r) = @_;
4910 my $cache = $self->{cache}->{check_path};
4911 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4912 return $cache->{data}->{$path};
4913 }
4914 my $pool = SVN::Pool->new;
4915 my $t = $self->SUPER::check_path($path, $r, $pool);
4916 $pool->clear;
4917 if ($r != $cache->{r}) {
4918 %{$cache->{data}} = ();
4919 $cache->{r} = $r;
4920 }
4921 $cache->{data}->{$path} = $t;
4922 }
4923
4924 sub get_dir {
4925 my ($self, $dir, $r) = @_;
4926 my $cache = $self->{cache}->{get_dir};
4927 if ($r == $cache->{r}) {
4928 if (my $x = $cache->{data}->{$dir}) {
4929 return wantarray ? @$x : $x->[0];
4930 }
4931 }
4932 my $pool = SVN::Pool->new;
4933 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4934 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4935 $pool->clear;
4936 if ($r != $cache->{r}) {
4937 %{$cache->{data}} = ();
4938 $cache->{r} = $r;
4939 }
4940 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4941 wantarray ? (\%dirents, $r, $props) : \%dirents;
4942 }
4943
4944 sub DESTROY {
4945 # do not call the real DESTROY since we store ourselves in $RA
4946 }
4947
4948 # get_log(paths, start, end, limit,
4949 # discover_changed_paths, strict_node_history, receiver)
4950 sub get_log {
4951 my ($self, @args) = @_;
4952 my $pool = SVN::Pool->new;
4953
4954 # svn_log_changed_path_t objects passed to get_log are likely to be
4955 # overwritten even if only the refs are copied to an external variable,
4956 # so we should dup the structures in their entirety. Using an
4957 # externally passed pool (instead of our temporary and quickly cleared
4958 # pool in Git::SVN::Ra) does not help matters at all...
4959 my $receiver = pop @args;
4960 my $prefix = "/".$self->{svn_path};
4961 $prefix =~ s#/+($)##;
4962 my $prefix_regex = qr#^\Q$prefix\E#;
4963 push(@args, sub {
4964 my ($paths) = $_[0];
4965 return &$receiver(@_) unless $paths;
4966 $_[0] = ();
4967 foreach my $p (keys %$paths) {
4968 my $i = $paths->{$p};
4969 # Make path relative to our url, not repos_root
4970 $p =~ s/$prefix_regex//;
4971 my %s = map { $_ => $i->$_; }
4972 qw/copyfrom_path copyfrom_rev action/;
4973 if ($s{'copyfrom_path'}) {
4974 $s{'copyfrom_path'} =~ s/$prefix_regex//;
4975 }
4976 $_[0]{$p} = \%s;
4977 }
4978 &$receiver(@_);
4979 });
4980
4981
4982 # the limit parameter was not supported in SVN 1.1.x, so we
4983 # drop it. Therefore, the receiver callback passed to it
4984 # is made aware of this limitation by being wrapped if
4985 # the limit passed to is being wrapped.
4986 if ($SVN::Core::VERSION le '1.2.0') {
4987 my $limit = splice(@args, 3, 1);
4988 if ($limit > 0) {
4989 my $receiver = pop @args;
4990 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4991 }
4992 }
4993 my $ret = $self->SUPER::get_log(@args, $pool);
4994 $pool->clear;
4995 $ret;
4996 }
4997
4998 sub trees_match {
4999 my ($self, $url1, $rev1, $url2, $rev2) = @_;
5000 my $ctx = SVN::Client->new(auth => _auth_providers);
5001 my $out = IO::File->new_tmpfile;
5002
5003 # older SVN (1.1.x) doesn't take $pool as the last parameter for
5004 # $ctx->diff(), so we'll create a default one
5005 my $pool = SVN::Pool->new_default_sub;
5006
5007 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5008 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5009 $out->flush;
5010 my $ret = (($out->stat)[7] == 0);
5011 close $out or croak $!;
5012
5013 $ret;
5014 }
5015
5016 sub get_commit_editor {
5017 my ($self, $log, $cb, $pool) = @_;
5018 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
5019 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5020 }
5021
5022 sub gs_do_update {
5023 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5024 my $new = ($rev_a == $rev_b);
5025 my $path = $gs->{path};
5026
5027 if ($new && -e $gs->{index}) {
5028 unlink $gs->{index} or die
5029 "Couldn't unlink index: $gs->{index}: $!\n";
5030 }
5031 my $pool = SVN::Pool->new;
5032 $editor->set_path_strip($path);
5033 my (@pc) = split m#/#, $path;
5034 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5035 1, $editor, $pool);
5036 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5037
5038 # Since we can't rely on svn_ra_reparent being available, we'll
5039 # just have to do some magic with set_path to make it so
5040 # we only want a partial path.
5041 my $sp = '';
5042 my $final = join('/', @pc);
5043 while (@pc) {
5044 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5045 $sp .= '/' if length $sp;
5046 $sp .= shift @pc;
5047 }
5048 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5049
5050 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5051
5052 $reporter->finish_report($pool);
5053 $pool->clear;
5054 $editor->{git_commit_ok};
5055 }
5056
5057 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5058 # svn_ra_reparent didn't work before 1.4)
5059 sub gs_do_switch {
5060 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5061 my $path = $gs->{path};
5062 my $pool = SVN::Pool->new;
5063
5064 my $full_url = $self->{url};
5065 my $old_url = $full_url;
5066 $full_url .= '/' . $path if length $path;
5067 my ($ra, $reparented);
5068
5069 if ($old_url =~ m#^svn(\+ssh)?://# ||
5070 ($full_url =~ m#^https?://# &&
5071 escape_url($full_url) ne $full_url)) {
5072 $_[0] = undef;
5073 $self = undef;
5074 $RA = undef;
5075 $ra = Git::SVN::Ra->new($full_url);
5076 $ra_invalid = 1;
5077 } elsif ($old_url ne $full_url) {
5078 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5079 $self->{url} = $full_url;
5080 $reparented = 1;
5081 }
5082
5083 $ra ||= $self;
5084 $url_b = escape_url($url_b);
5085 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5086 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5087 $reporter->set_path('', $rev_a, 0, @lock, $pool);
5088 $reporter->finish_report($pool);
5089
5090 if ($reparented) {
5091 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5092 $self->{url} = $old_url;
5093 }
5094
5095 $pool->clear;
5096 $editor->{git_commit_ok};
5097 }
5098
5099 sub longest_common_path {
5100 my ($gsv, $globs) = @_;
5101 my %common;
5102 my $common_max = scalar @$gsv;
5103
5104 foreach my $gs (@$gsv) {
5105 my @tmp = split m#/#, $gs->{path};
5106 my $p = '';
5107 foreach (@tmp) {
5108 $p .= length($p) ? "/$_" : $_;
5109 $common{$p} ||= 0;
5110 $common{$p}++;
5111 }
5112 }
5113 $globs ||= [];
5114 $common_max += scalar @$globs;
5115 foreach my $glob (@$globs) {
5116 my @tmp = split m#/#, $glob->{path}->{left};
5117 my $p = '';
5118 foreach (@tmp) {
5119 $p .= length($p) ? "/$_" : $_;
5120 $common{$p} ||= 0;
5121 $common{$p}++;
5122 }
5123 }
5124
5125 my $longest_path = '';
5126 foreach (sort {length $b <=> length $a} keys %common) {
5127 if ($common{$_} == $common_max) {
5128 $longest_path = $_;
5129 last;
5130 }
5131 }
5132 $longest_path;
5133 }
5134
5135 sub gs_fetch_loop_common {
5136 my ($self, $base, $head, $gsv, $globs) = @_;
5137 return if ($base > $head);
5138 my $inc = $_log_window_size;
5139 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5140 my $longest_path = longest_common_path($gsv, $globs);
5141 my $ra_url = $self->{url};
5142 my $find_trailing_edge;
5143 while (1) {
5144 my %revs;
5145 my $err;
5146 my $err_handler = $SVN::Error::handler;
5147 $SVN::Error::handler = sub {
5148 ($err) = @_;
5149 skip_unknown_revs($err);
5150 };
5151 sub _cb {
5152 my ($paths, $r, $author, $date, $log) = @_;
5153 [ $paths,
5154 { author => $author, date => $date, log => $log } ];
5155 }
5156 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5157 sub { $revs{$_[1]} = _cb(@_) });
5158 if ($err) {
5159 print "Checked through r$max\r";
5160 } else {
5161 $find_trailing_edge = 1;
5162 }
5163 if ($err and $find_trailing_edge) {
5164 print STDERR "Path '$longest_path' ",
5165 "was probably deleted:\n",
5166 $err->expanded_message,
5167 "\nWill attempt to follow ",
5168 "revisions r$min .. r$max ",
5169 "committed before the deletion\n";
5170 my $hi = $max;
5171 while (--$hi >= $min) {
5172 my $ok;
5173 $self->get_log([$longest_path], $min, $hi,
5174 0, 1, 1, sub {
5175 $ok = $_[1];
5176 $revs{$_[1]} = _cb(@_) });
5177 if ($ok) {
5178 print STDERR "r$min .. r$ok OK\n";
5179 last;
5180 }
5181 }
5182 $find_trailing_edge = 0;
5183 }
5184 $SVN::Error::handler = $err_handler;
5185
5186 my %exists = map { $_->{path} => $_ } @$gsv;
5187 foreach my $r (sort {$a <=> $b} keys %revs) {
5188 my ($paths, $logged) = @{$revs{$r}};
5189
5190 foreach my $gs ($self->match_globs(\%exists, $paths,
5191 $globs, $r)) {
5192 if ($gs->rev_map_max >= $r) {
5193 next;
5194 }
5195 next unless $gs->match_paths($paths, $r);
5196 $gs->{logged_rev_props} = $logged;
5197 if (my $last_commit = $gs->last_commit) {
5198 $gs->assert_index_clean($last_commit);
5199 }
5200 my $log_entry = $gs->do_fetch($paths, $r);
5201 if ($log_entry) {
5202 $gs->do_git_commit($log_entry);
5203 }
5204 $INDEX_FILES{$gs->{index}} = 1;
5205 }
5206 foreach my $g (@$globs) {
5207 my $k = "svn-remote.$g->{remote}." .
5208 "$g->{t}-maxRev";
5209 Git::SVN::tmp_config($k, $r);
5210 }
5211 if ($ra_invalid) {
5212 $_[0] = undef;
5213 $self = undef;
5214 $RA = undef;
5215 $self = Git::SVN::Ra->new($ra_url);
5216 $ra_invalid = undef;
5217 }
5218 }
5219 # pre-fill the .rev_db since it'll eventually get filled in
5220 # with '0' x40 if something new gets committed
5221 foreach my $gs (@$gsv) {
5222 next if $gs->rev_map_max >= $max;
5223 next if defined $gs->rev_map_get($max);
5224 $gs->rev_map_set($max, 0 x40);
5225 }
5226 foreach my $g (@$globs) {
5227 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5228 Git::SVN::tmp_config($k, $max);
5229 }
5230 last if $max >= $head;
5231 $min = $max + 1;
5232 $max += $inc;
5233 $max = $head if ($max > $head);
5234 }
5235 Git::SVN::gc();
5236 }
5237
5238 sub get_dir_globbed {
5239 my ($self, $left, $depth, $r) = @_;
5240
5241 my @x = eval { $self->get_dir($left, $r) };
5242 return unless scalar @x == 3;
5243 my $dirents = $x[0];
5244 my @finalents;
5245 foreach my $de (keys %$dirents) {
5246 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5247 if ($depth > 1) {
5248 my @args = ("$left/$de", $depth - 1, $r);
5249 foreach my $dir ($self->get_dir_globbed(@args)) {
5250 push @finalents, "$de/$dir";
5251 }
5252 } else {
5253 push @finalents, $de;
5254 }
5255 }
5256 @finalents;
5257 }
5258
5259 sub match_globs {
5260 my ($self, $exists, $paths, $globs, $r) = @_;
5261
5262 sub get_dir_check {
5263 my ($self, $exists, $g, $r) = @_;
5264
5265 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5266 $g->{path}->{depth},
5267 $r);
5268
5269 foreach my $de (@dirs) {
5270 my $p = $g->{path}->full_path($de);
5271 next if $exists->{$p};
5272 next if (length $g->{path}->{right} &&
5273 ($self->check_path($p, $r) !=
5274 $SVN::Node::dir));
5275 next unless $p =~ /$g->{path}->{regex}/;
5276 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5277 $g->{ref}->full_path($de), 1);
5278 }
5279 }
5280 foreach my $g (@$globs) {
5281 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5282 if ($path->{action} =~ /^[AR]$/) {
5283 get_dir_check($self, $exists, $g, $r);
5284 }
5285 }
5286 foreach (keys %$paths) {
5287 if (/$g->{path}->{left_regex}/ &&
5288 !/$g->{path}->{regex}/) {
5289 next if $paths->{$_}->{action} !~ /^[AR]$/;
5290 get_dir_check($self, $exists, $g, $r);
5291 }
5292 next unless /$g->{path}->{regex}/;
5293 my $p = $1;
5294 my $pathname = $g->{path}->full_path($p);
5295 next if $exists->{$pathname};
5296 next if ($self->check_path($pathname, $r) !=
5297 $SVN::Node::dir);
5298 $exists->{$pathname} = Git::SVN->init(
5299 $self->{url}, $pathname, undef,
5300 $g->{ref}->full_path($p), 1);
5301 }
5302 my $c = '';
5303 foreach (split m#/#, $g->{path}->{left}) {
5304 $c .= "/$_";
5305 next unless ($paths->{$c} &&
5306 ($paths->{$c}->{action} =~ /^[AR]$/));
5307 get_dir_check($self, $exists, $g, $r);
5308 }
5309 }
5310 values %$exists;
5311 }
5312
5313 sub minimize_url {
5314 my ($self) = @_;
5315 return $self->{url} if ($self->{url} eq $self->{repos_root});
5316 my $url = $self->{repos_root};
5317 my @components = split(m!/!, $self->{svn_path});
5318 my $c = '';
5319 do {
5320 $url .= "/$c" if length $c;
5321 eval {
5322 my $ra = (ref $self)->new($url);
5323 my $latest = $ra->get_latest_revnum;
5324 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5325 };
5326 } while ($@ && ($c = shift @components));
5327 $url;
5328 }
5329
5330 sub can_do_switch {
5331 my $self = shift;
5332 unless (defined $can_do_switch) {
5333 my $pool = SVN::Pool->new;
5334 my $rep = eval {
5335 $self->do_switch(1, '', 0, $self->{url},
5336 SVN::Delta::Editor->new, $pool);
5337 };
5338 if ($@) {
5339 $can_do_switch = 0;
5340 } else {
5341 $rep->abort_report($pool);
5342 $can_do_switch = 1;
5343 }
5344 $pool->clear;
5345 }
5346 $can_do_switch;
5347 }
5348
5349 sub skip_unknown_revs {
5350 my ($err) = @_;
5351 my $errno = $err->apr_err();
5352 # Maybe the branch we're tracking didn't
5353 # exist when the repo started, so it's
5354 # not an error if it doesn't, just continue
5355 #
5356 # Wonderfully consistent library, eh?
5357 # 160013 - svn:// and file://
5358 # 175002 - http(s)://
5359 # 175007 - http(s):// (this repo required authorization, too...)
5360 # More codes may be discovered later...
5361 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5362 my $err_key = $err->expanded_message;
5363 # revision numbers change every time, filter them out
5364 $err_key =~ s/\d+/\0/g;
5365 $err_key = "$errno\0$err_key";
5366 unless ($ignored_err{$err_key}) {
5367 warn "W: Ignoring error from SVN, path probably ",
5368 "does not exist: ($errno): ",
5369 $err->expanded_message,"\n";
5370 warn "W: Do not be alarmed at the above message ",
5371 "git-svn is just searching aggressively for ",
5372 "old history.\n",
5373 "This may take a while on large repositories\n";
5374 $ignored_err{$err_key} = 1;
5375 }
5376 return;
5377 }
5378 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5379 }
5380
5381 package Git::SVN::Log;
5382 use strict;
5383 use warnings;
5384 use POSIX qw/strftime/;
5385 use Time::Local;
5386 use constant commit_log_separator => ('-' x 72) . "\n";
5387 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5388 %rusers $show_commit $incremental/;
5389 my $l_fmt;
5390
5391 sub cmt_showable {
5392 my ($c) = @_;
5393 return 1 if defined $c->{r};
5394
5395 # big commit message got truncated by the 16k pretty buffer in rev-list
5396 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5397 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5398 @{$c->{l}} = ();
5399 my @log = command(qw/cat-file commit/, $c->{c});
5400
5401 # shift off the headers
5402 shift @log while ($log[0] ne '');
5403 shift @log;
5404
5405 # TODO: make $c->{l} not have a trailing newline in the future
5406 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5407
5408 (undef, $c->{r}, undef) = ::extract_metadata(
5409 (grep(/^git-svn-id: /, @log))[-1]);
5410 }
5411 return defined $c->{r};
5412 }
5413
5414 sub log_use_color {
5415 return $color || Git->repository->get_colorbool('color.diff');
5416 }
5417
5418 sub git_svn_log_cmd {
5419 my ($r_min, $r_max, @args) = @_;
5420 my $head = 'HEAD';
5421 my (@files, @log_opts);
5422 foreach my $x (@args) {
5423 if ($x eq '--' || @files) {
5424 push @files, $x;
5425 } else {
5426 if (::verify_ref("$x^0")) {
5427 $head = $x;
5428 } else {
5429 push @log_opts, $x;
5430 }
5431 }
5432 }
5433
5434 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5435 $gs ||= Git::SVN->_new;
5436 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5437 $gs->refname);
5438 push @cmd, '-r' unless $non_recursive;
5439 push @cmd, qw/--raw --name-status/ if $verbose;
5440 push @cmd, '--color' if log_use_color();
5441 push @cmd, @log_opts;
5442 if (defined $r_max && $r_max == $r_min) {
5443 push @cmd, '--max-count=1';
5444 if (my $c = $gs->rev_map_get($r_max)) {
5445 push @cmd, $c;
5446 }
5447 } elsif (defined $r_max) {
5448 if ($r_max < $r_min) {
5449 ($r_min, $r_max) = ($r_max, $r_min);
5450 }
5451 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5452 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5453 # If there are no commits in the range, both $c_max and $c_min
5454 # will be undefined. If there is at least 1 commit in the
5455 # range, both will be defined.
5456 return () if !defined $c_min || !defined $c_max;
5457 if ($c_min eq $c_max) {
5458 push @cmd, '--max-count=1', $c_min;
5459 } else {
5460 push @cmd, '--boundary', "$c_min..$c_max";
5461 }
5462 }
5463 return (@cmd, @files);
5464 }
5465
5466 # adapted from pager.c
5467 sub config_pager {
5468 if (! -t *STDOUT) {
5469 $ENV{GIT_PAGER_IN_USE} = 'false';
5470 $pager = undef;
5471 return;
5472 }
5473 chomp($pager = command_oneline(qw(var GIT_PAGER)));
5474 if ($pager eq 'cat') {
5475 $pager = undef;
5476 }
5477 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5478 }
5479
5480 sub run_pager {
5481 return unless defined $pager;
5482 pipe my ($rfd, $wfd) or return;
5483 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5484 if (!$pid) {
5485 open STDOUT, '>&', $wfd or
5486 ::fatal "Can't redirect to stdout: $!";
5487 return;
5488 }
5489 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5490 $ENV{LESS} ||= 'FRSX';
5491 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5492 }
5493
5494 sub format_svn_date {
5495 # some systmes don't handle or mishandle %z, so be creative.
5496 my $t = shift || time;
5497 my $gm = timelocal(gmtime($t));
5498 my $sign = qw( + + - )[ $t <=> $gm ];
5499 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5500 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5501 }
5502
5503 sub parse_git_date {
5504 my ($t, $tz) = @_;
5505 # Date::Parse isn't in the standard Perl distro :(
5506 if ($tz =~ s/^\+//) {
5507 $t += tz_to_s_offset($tz);
5508 } elsif ($tz =~ s/^\-//) {
5509 $t -= tz_to_s_offset($tz);
5510 }
5511 return $t;
5512 }
5513
5514 sub set_local_timezone {
5515 if (defined $TZ) {
5516 $ENV{TZ} = $TZ;
5517 } else {
5518 delete $ENV{TZ};
5519 }
5520 }
5521
5522 sub tz_to_s_offset {
5523 my ($tz) = @_;
5524 $tz =~ s/(\d\d)$//;
5525 return ($1 * 60) + ($tz * 3600);
5526 }
5527
5528 sub get_author_info {
5529 my ($dest, $author, $t, $tz) = @_;
5530 $author =~ s/(?:^\s*|\s*$)//g;
5531 $dest->{a_raw} = $author;
5532 my $au;
5533 if ($::_authors) {
5534 $au = $rusers{$author} || undef;
5535 }
5536 if (!$au) {
5537 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5538 }
5539 $dest->{t} = $t;
5540 $dest->{tz} = $tz;
5541 $dest->{a} = $au;
5542 $dest->{t_utc} = parse_git_date($t, $tz);
5543 }
5544
5545 sub process_commit {
5546 my ($c, $r_min, $r_max, $defer) = @_;
5547 if (defined $r_min && defined $r_max) {
5548 if ($r_min == $c->{r} && $r_min == $r_max) {
5549 show_commit($c);
5550 return 0;
5551 }
5552 return 1 if $r_min == $r_max;
5553 if ($r_min < $r_max) {
5554 # we need to reverse the print order
5555 return 0 if (defined $limit && --$limit < 0);
5556 push @$defer, $c;
5557 return 1;
5558 }
5559 if ($r_min != $r_max) {
5560 return 1 if ($r_min < $c->{r});
5561 return 1 if ($r_max > $c->{r});
5562 }
5563 }
5564 return 0 if (defined $limit && --$limit < 0);
5565 show_commit($c);
5566 return 1;
5567 }
5568
5569 sub show_commit {
5570 my $c = shift;
5571 if ($oneline) {
5572 my $x = "\n";
5573 if (my $l = $c->{l}) {
5574 while ($l->[0] =~ /^\s*$/) { shift @$l }
5575 $x = $l->[0];
5576 }
5577 $l_fmt ||= 'A' . length($c->{r});
5578 print 'r',pack($l_fmt, $c->{r}),' | ';
5579 print "$c->{c} | " if $show_commit;
5580 print $x;
5581 } else {
5582 show_commit_normal($c);
5583 }
5584 }
5585
5586 sub show_commit_changed_paths {
5587 my ($c) = @_;
5588 return unless $c->{changed};
5589 print "Changed paths:\n", @{$c->{changed}};
5590 }
5591
5592 sub show_commit_normal {
5593 my ($c) = @_;
5594 print commit_log_separator, "r$c->{r} | ";
5595 print "$c->{c} | " if $show_commit;
5596 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5597 my $nr_line = 0;
5598
5599 if (my $l = $c->{l}) {
5600 while ($l->[$#$l] eq "\n" && $#$l > 0
5601 && $l->[($#$l - 1)] eq "\n") {
5602 pop @$l;
5603 }
5604 $nr_line = scalar @$l;
5605 if (!$nr_line) {
5606 print "1 line\n\n\n";
5607 } else {
5608 if ($nr_line == 1) {
5609 $nr_line = '1 line';
5610 } else {
5611 $nr_line .= ' lines';
5612 }
5613 print $nr_line, "\n";
5614 show_commit_changed_paths($c);
5615 print "\n";
5616 print $_ foreach @$l;
5617 }
5618 } else {
5619 print "1 line\n";
5620 show_commit_changed_paths($c);
5621 print "\n";
5622
5623 }
5624 foreach my $x (qw/raw stat diff/) {
5625 if ($c->{$x}) {
5626 print "\n";
5627 print $_ foreach @{$c->{$x}}
5628 }
5629 }
5630 }
5631
5632 sub cmd_show_log {
5633 my (@args) = @_;
5634 my ($r_min, $r_max);
5635 my $r_last = -1; # prevent dupes
5636 set_local_timezone();
5637 if (defined $::_revision) {
5638 if ($::_revision =~ /^(\d+):(\d+)$/) {
5639 ($r_min, $r_max) = ($1, $2);
5640 } elsif ($::_revision =~ /^\d+$/) {
5641 $r_min = $r_max = $::_revision;
5642 } else {
5643 ::fatal "-r$::_revision is not supported, use ",
5644 "standard 'git log' arguments instead";
5645 }
5646 }
5647
5648 config_pager();
5649 @args = git_svn_log_cmd($r_min, $r_max, @args);
5650 if (!@args) {
5651 print commit_log_separator unless $incremental || $oneline;
5652 return;
5653 }
5654 my $log = command_output_pipe(@args);
5655 run_pager();
5656 my (@k, $c, $d, $stat);
5657 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5658 while (<$log>) {
5659 if (/^${esc_color}commit -?($::sha1_short)/o) {
5660 my $cmt = $1;
5661 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5662 $r_last = $c->{r};
5663 process_commit($c, $r_min, $r_max, \@k) or
5664 goto out;
5665 }
5666 $d = undef;
5667 $c = { c => $cmt };
5668 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5669 get_author_info($c, $1, $2, $3);
5670 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5671 # ignore
5672 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5673 push @{$c->{raw}}, $_;
5674 } elsif (/^${esc_color}[ACRMDT]\t/) {
5675 # we could add $SVN->{svn_path} here, but that requires
5676 # remote access at the moment (repo_path_split)...
5677 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5678 push @{$c->{changed}}, $_;
5679 } elsif (/^${esc_color}diff /o) {
5680 $d = 1;
5681 push @{$c->{diff}}, $_;
5682 } elsif ($d) {
5683 push @{$c->{diff}}, $_;
5684 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5685 $esc_color*[\+\-]*$esc_color$/x) {
5686 $stat = 1;
5687 push @{$c->{stat}}, $_;
5688 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5689 push @{$c->{stat}}, $_;
5690 $stat = undef;
5691 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5692 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5693 } elsif (s/^${esc_color} //o) {
5694 push @{$c->{l}}, $_;
5695 }
5696 }
5697 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5698 $r_last = $c->{r};
5699 process_commit($c, $r_min, $r_max, \@k);
5700 }
5701 if (@k) {
5702 ($r_min, $r_max) = ($r_max, $r_min);
5703 process_commit($_, $r_min, $r_max) foreach reverse @k;
5704 }
5705 out:
5706 close $log;
5707 print commit_log_separator unless $incremental || $oneline;
5708 }
5709
5710 sub cmd_blame {
5711 my $path = pop;
5712
5713 config_pager();
5714 run_pager();
5715
5716 my ($fh, $ctx, $rev);
5717
5718 if ($_git_format) {
5719 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5720 while (my $line = <$fh>) {
5721 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5722 # Uncommitted edits show up as a rev ID of
5723 # all zeros, which we can't look up with
5724 # cmt_metadata
5725 if ($1 !~ /^0+$/) {
5726 (undef, $rev, undef) =
5727 ::cmt_metadata($1);
5728 $rev = '0' if (!$rev);
5729 } else {
5730 $rev = '0';
5731 }
5732 $rev = sprintf('%-10s', $rev);
5733 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5734 }
5735 print $line;
5736 }
5737 } else {
5738 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5739 '--', $path);
5740 my ($sha1);
5741 my %authors;
5742 my @buffer;
5743 my %dsha; #distinct sha keys
5744
5745 while (my $line = <$fh>) {
5746 push @buffer, $line;
5747 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5748 $dsha{$1} = 1;
5749 }
5750 }
5751
5752 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5753
5754 foreach my $line (@buffer) {
5755 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5756 $rev = $s2r->{$1};
5757 $rev = '0' if (!$rev)
5758 }
5759 elsif ($line =~ /^author (.*)/) {
5760 $authors{$rev} = $1;
5761 $authors{$rev} =~ s/\s/_/g;
5762 }
5763 elsif ($line =~ /^\t(.*)$/) {
5764 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5765 }
5766 }
5767 }
5768 command_close_pipe($fh, $ctx);
5769 }
5770
5771 package Git::SVN::Migration;
5772 # these version numbers do NOT correspond to actual version numbers
5773 # of git nor git-svn. They are just relative.
5774 #
5775 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5776 #
5777 # v1 layout: .git/$id/info/url, refs/remotes/$id
5778 #
5779 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5780 #
5781 # v3 layout: .git/svn/$id, refs/remotes/$id
5782 # - info/url may remain for backwards compatibility
5783 # - this is what we migrate up to this layout automatically,
5784 # - this will be used by git svn init on single branches
5785 # v3.1 layout (auto migrated):
5786 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5787 # for backwards compatibility
5788 #
5789 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5790 # - this is only created for newly multi-init-ed
5791 # repositories. Similar in spirit to the
5792 # --use-separate-remotes option in git-clone (now default)
5793 # - we do not automatically migrate to this (following
5794 # the example set by core git)
5795 #
5796 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
5797 # - newer, more-efficient format that uses 24-bytes per record
5798 # with no filler space.
5799 # - use xxd -c24 < .rev_map.$UUID to view and debug
5800 # - This is a one-way migration, repositories updated to the
5801 # new format will not be able to use old git-svn without
5802 # rebuilding the .rev_db. Rebuilding the rev_db is not
5803 # possible if noMetadata or useSvmProps are set; but should
5804 # be no problem for users that use the (sensible) defaults.
5805 use strict;
5806 use warnings;
5807 use Carp qw/croak/;
5808 use File::Path qw/mkpath/;
5809 use File::Basename qw/dirname basename/;
5810 use vars qw/$_minimize/;
5811
5812 sub migrate_from_v0 {
5813 my $git_dir = $ENV{GIT_DIR};
5814 return undef unless -d $git_dir;
5815 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5816 my $migrated = 0;
5817 while (<$fh>) {
5818 chomp;
5819 my ($id, $orig_ref) = ($_, $_);
5820 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5821 next unless -f "$git_dir/$id/info/url";
5822 my $new_ref = "refs/remotes/$id";
5823 if (::verify_ref("$new_ref^0")) {
5824 print STDERR "W: $orig_ref is probably an old ",
5825 "branch used by an ancient version of ",
5826 "git-svn.\n",
5827 "However, $new_ref also exists.\n",
5828 "We will not be able ",
5829 "to use this branch until this ",
5830 "ambiguity is resolved.\n";
5831 next;
5832 }
5833 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5834 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5835 command_noisy('update-ref', $new_ref, $orig_ref);
5836 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5837 $migrated++;
5838 }
5839 command_close_pipe($fh, $ctx);
5840 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5841 $migrated;
5842 }
5843
5844 sub migrate_from_v1 {
5845 my $git_dir = $ENV{GIT_DIR};
5846 my $migrated = 0;
5847 return $migrated unless -d $git_dir;
5848 my $svn_dir = "$git_dir/svn";
5849
5850 # just in case somebody used 'svn' as their $id at some point...
5851 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5852
5853 print STDERR "Migrating from a git-svn v1 layout...\n";
5854 mkpath([$svn_dir]);
5855 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5856 "$svn_dir\n\t(required for this version ",
5857 "($::VERSION) of git-svn) does not exist.\n";
5858 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5859 while (<$fh>) {
5860 my $x = $_;
5861 next unless $x =~ s#^refs/remotes/##;
5862 chomp $x;
5863 next unless -f "$git_dir/$x/info/url";
5864 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5865 next unless $u;
5866 my $dn = dirname("$git_dir/svn/$x");
5867 mkpath([$dn]) unless -d $dn;
5868 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5869 mkpath(["$git_dir/svn/svn"]);
5870 print STDERR " - $git_dir/$x/info => ",
5871 "$git_dir/svn/$x/info\n";
5872 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5873 croak "$!: $x";
5874 # don't worry too much about these, they probably
5875 # don't exist with repos this old (save for index,
5876 # and we can easily regenerate that)
5877 foreach my $f (qw/unhandled.log index .rev_db/) {
5878 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5879 }
5880 } else {
5881 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5882 rename "$git_dir/$x", "$git_dir/svn/$x" or
5883 croak "$!: $x";
5884 }
5885 $migrated++;
5886 }
5887 command_close_pipe($fh, $ctx);
5888 print STDERR "Done migrating from a git-svn v1 layout\n";
5889 $migrated;
5890 }
5891
5892 sub read_old_urls {
5893 my ($l_map, $pfx, $path) = @_;
5894 my @dir;
5895 foreach (<$path/*>) {
5896 if (-r "$_/info/url") {
5897 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5898 my $ref_id = $pfx . basename $_;
5899 my $url = ::file_to_s("$_/info/url");
5900 $l_map->{$ref_id} = $url;
5901 } elsif (-d $_) {
5902 push @dir, $_;
5903 }
5904 }
5905 foreach (@dir) {
5906 my $x = $_;
5907 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5908 read_old_urls($l_map, $x, $_);
5909 }
5910 }
5911
5912 sub migrate_from_v2 {
5913 my @cfg = command(qw/config -l/);
5914 return if grep /^svn-remote\..+\.url=/, @cfg;
5915 my %l_map;
5916 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5917 my $migrated = 0;
5918
5919 foreach my $ref_id (sort keys %l_map) {
5920 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5921 if ($@) {
5922 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5923 }
5924 $migrated++;
5925 }
5926 $migrated;
5927 }
5928
5929 sub minimize_connections {
5930 my $r = Git::SVN::read_all_remotes();
5931 my $new_urls = {};
5932 my $root_repos = {};
5933 foreach my $repo_id (keys %$r) {
5934 my $url = $r->{$repo_id}->{url} or next;
5935 my $fetch = $r->{$repo_id}->{fetch} or next;
5936 my $ra = Git::SVN::Ra->new($url);
5937
5938 # skip existing cases where we already connect to the root
5939 if (($ra->{url} eq $ra->{repos_root}) ||
5940 ($ra->{repos_root} eq $repo_id)) {
5941 $root_repos->{$ra->{url}} = $repo_id;
5942 next;
5943 }
5944
5945 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5946 my $root_path = $ra->{url};
5947 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5948 foreach my $path (keys %$fetch) {
5949 my $ref_id = $fetch->{$path};
5950 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5951
5952 # make sure we can read when connecting to
5953 # a higher level of a repository
5954 my ($last_rev, undef) = $gs->last_rev_commit;
5955 if (!defined $last_rev) {
5956 $last_rev = eval {
5957 $root_ra->get_latest_revnum;
5958 };
5959 next if $@;
5960 }
5961 my $new = $root_path;
5962 $new .= length $path ? "/$path" : '';
5963 eval {
5964 $root_ra->get_log([$new], $last_rev, $last_rev,
5965 0, 0, 1, sub { });
5966 };
5967 next if $@;
5968 $new_urls->{$ra->{repos_root}}->{$new} =
5969 { ref_id => $ref_id,
5970 old_repo_id => $repo_id,
5971 old_path => $path };
5972 }
5973 }
5974
5975 my @emptied;
5976 foreach my $url (keys %$new_urls) {
5977 # see if we can re-use an existing [svn-remote "repo_id"]
5978 # instead of creating a(n ugly) new section:
5979 my $repo_id = $root_repos->{$url} || $url;
5980
5981 my $fetch = $new_urls->{$url};
5982 foreach my $path (keys %$fetch) {
5983 my $x = $fetch->{$path};
5984 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5985 my $pfx = "svn-remote.$x->{old_repo_id}";
5986
5987 my $old_fetch = quotemeta("$x->{old_path}:".
5988 "$x->{ref_id}");
5989 command_noisy(qw/config --unset/,
5990 "$pfx.fetch", '^'. $old_fetch . '$');
5991 delete $r->{$x->{old_repo_id}}->
5992 {fetch}->{$x->{old_path}};
5993 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5994 command_noisy(qw/config --unset/,
5995 "$pfx.url");
5996 push @emptied, $x->{old_repo_id}
5997 }
5998 }
5999 }
6000 if (@emptied) {
6001 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6002 print STDERR <<EOF;
6003 The following [svn-remote] sections in your config file ($file) are empty
6004 and can be safely removed:
6005 EOF
6006 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6007 }
6008 }
6009
6010 sub migration_check {
6011 migrate_from_v0();
6012 migrate_from_v1();
6013 migrate_from_v2();
6014 minimize_connections() if $_minimize;
6015 }
6016
6017 package Git::IndexInfo;
6018 use strict;
6019 use warnings;
6020 use Git qw/command_input_pipe command_close_pipe/;
6021
6022 sub new {
6023 my ($class) = @_;
6024 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6025 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6026 }
6027
6028 sub remove {
6029 my ($self, $path) = @_;
6030 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6031 return ++$self->{nr};
6032 }
6033 undef;
6034 }
6035
6036 sub update {
6037 my ($self, $mode, $hash, $path) = @_;
6038 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6039 return ++$self->{nr};
6040 }
6041 undef;
6042 }
6043
6044 sub DESTROY {
6045 my ($self) = @_;
6046 command_close_pipe($self->{gui}, $self->{ctx});
6047 }
6048
6049 package Git::SVN::GlobSpec;
6050 use strict;
6051 use warnings;
6052
6053 sub new {
6054 my ($class, $glob, $pattern_ok) = @_;
6055 my $re = $glob;
6056 $re =~ s!/+$!!g; # no need for trailing slashes
6057 my (@left, @right, @patterns);
6058 my $state = "left";
6059 my $die_msg = "Only one set of wildcard directories " .
6060 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6061 for my $part (split(m|/|, $glob)) {
6062 if ($part =~ /\*/ && $part ne "*") {
6063 die "Invalid pattern in '$glob': $part\n";
6064 } elsif ($pattern_ok && $part =~ /[{}]/ &&
6065 $part !~ /^\{[^{}]+\}/) {
6066 die "Invalid pattern in '$glob': $part\n";
6067 }
6068 if ($part eq "*") {
6069 die $die_msg if $state eq "right";
6070 $state = "pattern";
6071 push(@patterns, "[^/]*");
6072 } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6073 die $die_msg if $state eq "right";
6074 $state = "pattern";
6075 my $p = quotemeta($1);
6076 $p =~ s/\\,/|/g;
6077 push(@patterns, "(?:$p)");
6078 } else {
6079 if ($state eq "left") {
6080 push(@left, $part);
6081 } else {
6082 push(@right, $part);
6083 $state = "right";
6084 }
6085 }
6086 }
6087 my $depth = @patterns;
6088 if ($depth == 0) {
6089 die "One '*' is needed in glob: '$glob'\n";
6090 }
6091 my $left = join('/', @left);
6092 my $right = join('/', @right);
6093 $re = join('/', @patterns);
6094 $re = join('\/',
6095 grep(length, quotemeta($left), "($re)", quotemeta($right)));
6096 my $left_re = qr/^\/\Q$left\E(\/|$)/;
6097 bless { left => $left, right => $right, left_regex => $left_re,
6098 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6099 }
6100
6101 sub full_path {
6102 my ($self, $path) = @_;
6103 return (length $self->{left} ? "$self->{left}/" : '') .
6104 $path . (length $self->{right} ? "/$self->{right}" : '');
6105 }
6106
6107 __END__
6108
6109 Data structures:
6110
6111
6112 $remotes = { # returned by read_all_remotes()
6113 'svn' => {
6114 # svn-remote.svn.url=https://svn.musicpd.org
6115 url => 'https://svn.musicpd.org',
6116 # svn-remote.svn.fetch=mpd/trunk:trunk
6117 fetch => {
6118 'mpd/trunk' => 'trunk',
6119 },
6120 # svn-remote.svn.tags=mpd/tags/*:tags/*
6121 tags => {
6122 path => {
6123 left => 'mpd/tags',
6124 right => '',
6125 regex => qr!mpd/tags/([^/]+)$!,
6126 glob => 'tags/*',
6127 },
6128 ref => {
6129 left => 'tags',
6130 right => '',
6131 regex => qr!tags/([^/]+)$!,
6132 glob => 'tags/*',
6133 },
6134 }
6135 }
6136 };
6137
6138 $log_entry hashref as returned by libsvn_log_entry()
6139 {
6140 log => 'whitespace-formatted log entry
6141 ', # trailing newline is preserved
6142 revision => '8', # integer
6143 date => '2004-02-24T17:01:44.108345Z', # commit date
6144 author => 'committer name'
6145 };
6146
6147
6148 # this is generated by generate_diff();
6149 @mods = array of diff-index line hashes, each element represents one line
6150 of diff-index output
6151
6152 diff-index line ($m hash)
6153 {
6154 mode_a => first column of diff-index output, no leading ':',
6155 mode_b => second column of diff-index output,
6156 sha1_b => sha1sum of the final blob,
6157 chg => change type [MCRADT],
6158 file_a => original file name of a file (iff chg is 'C' or 'R')
6159 file_b => new/current file name of a file (any chg)
6160 }
6161 ;
6162
6163 # retval of read_url_paths{,_all}();
6164 $l_map = {
6165 # repository root url
6166 'https://svn.musicpd.org' => {
6167 # repository path # GIT_SVN_ID
6168 'mpd/trunk' => 'trunk',
6169 'mpd/tags/0.11.5' => 'tags/0.11.5',
6170 },
6171 }
6172
6173 Notes:
6174 I don't trust the each() function on unless I created %hash myself
6175 because the internal iterator may not have started at base.
Alternate source for your git.git needs