![[tokkee]](http://tokkee.org/images/avatar.png){kind=avatar}
1 #!/usr/bin/env perl
2 # Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3 # License: GPL v2 or later
4 use 5.008;
5 use warnings;
6 use strict;
7 use vars qw/ $AUTHOR $VERSION
8 $sha1 $sha1_short $_revision $_repository
9 $_q $_authors $_authors_prog %users/;
10 $AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
11 $VERSION = '@@GIT_VERSION@@';
13 # From which subdir have we been invoked?
14 my $cmd_dir_prefix = eval {
15 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
16 } || '';
18 my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
19 $ENV{GIT_DIR} ||= '.git';
20 $Git::SVN::default_repo_id = 'svn';
21 $Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
22 $Git::SVN::Ra::_log_window_size = 100;
23 $Git::SVN::_minimize_url = 'unset';
25 if (! exists $ENV{SVN_SSH} && exists $ENV{GIT_SSH}) {
26 $ENV{SVN_SSH} = $ENV{GIT_SSH};
27 }
29 if (exists $ENV{SVN_SSH} && $^O eq 'msys') {
30 $ENV{SVN_SSH} =~ s/\\/\\\\/g;
31 $ENV{SVN_SSH} =~ s/(.*)/"$1"/;
32 }
34 $Git::SVN::Log::TZ = $ENV{TZ};
35 $ENV{TZ} = 'UTC';
36 $| = 1; # unbuffer STDOUT
38 sub fatal (@) { print STDERR "@_\n"; exit 1 }
39 sub _req_svn {
40 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
41 require SVN::Ra;
42 require SVN::Delta;
43 if ($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 use Memoize; # core since 5.8.0, Jul 2002
63 BEGIN {
64 # import functions from Git into our packages, en masse
65 no strict 'refs';
66 foreach (qw/command command_oneline command_noisy command_output_pipe
67 command_input_pipe command_close_pipe
68 command_bidi_pipe command_close_bidi_pipe/) {
69 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
70 Git::SVN::Migration Git::SVN::Log Git::SVN),
71 __PACKAGE__) {
72 *{"${package}::$_"} = \&{"Git::$_"};
73 }
74 }
75 Memoize::memoize 'Git::config';
76 Memoize::memoize 'Git::config_bool';
77 }
79 my ($SVN);
81 $sha1 = qr/[a-f\d]{40}/;
82 $sha1_short = qr/[a-f\d]{4,40}/;
83 my ($_stdin, $_help, $_edit,
84 $_message, $_file, $_branch_dest,
85 $_template, $_shared,
86 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
87 $_merge, $_strategy, $_dry_run, $_local,
88 $_prefix, $_no_checkout, $_url, $_verbose,
89 $_git_format, $_commit_url, $_tag, $_merge_info);
90 $Git::SVN::_follow_parent = 1;
91 $SVN::Git::Fetcher::_placeholder_filename = ".gitignore";
92 $_q ||= 0;
93 my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
94 'config-dir=s' => \$Git::SVN::Ra::config_dir,
95 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
96 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
97 my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
98 'authors-file|A=s' => \$_authors,
99 'authors-prog=s' => \$_authors_prog,
100 'repack:i' => \$Git::SVN::_repack,
101 'noMetadata' => \$Git::SVN::_no_metadata,
102 'useSvmProps' => \$Git::SVN::_use_svm_props,
103 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
104 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
105 'no-checkout' => \$_no_checkout,
106 'quiet|q+' => \$_q,
107 'repack-flags|repack-args|repack-opts=s' =>
108 \$Git::SVN::_repack_flags,
109 'use-log-author' => \$Git::SVN::_use_log_author,
110 'add-author-from' => \$Git::SVN::_add_author_from,
111 'localtime' => \$Git::SVN::_localtime,
112 %remote_opts );
114 my ($_trunk, @_tags, @_branches, $_stdlayout);
115 my %icv;
116 my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
117 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
118 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
119 'stdlayout|s' => \$_stdlayout,
120 'minimize-url|m!' => \$Git::SVN::_minimize_url,
121 'no-metadata' => sub { $icv{noMetadata} = 1 },
122 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
123 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
124 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
125 'rewrite-uuid=s' => sub { $icv{rewriteUUID} = $_[1] },
126 %remote_opts );
127 my %cmt_opts = ( 'edit|e' => \$_edit,
128 'rmdir' => \$SVN::Git::Editor::_rmdir,
129 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
130 'l=i' => \$SVN::Git::Editor::_rename_limit,
131 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
132 );
134 my %cmd = (
135 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
136 { 'revision|r=s' => \$_revision,
137 'fetch-all|all' => \$_fetch_all,
138 'parent|p' => \$_fetch_parent,
139 %fc_opts } ],
140 clone => [ \&cmd_clone, "Initialize and fetch revisions",
141 { 'revision|r=s' => \$_revision,
142 'preserve-empty-dirs' =>
143 \$SVN::Git::Fetcher::_preserve_empty_dirs,
144 'placeholder-filename=s' =>
145 \$SVN::Git::Fetcher::_placeholder_filename,
146 %fc_opts, %init_opts } ],
147 init => [ \&cmd_init, "Initialize a repo for tracking" .
148 " (requires URL argument)",
149 \%init_opts ],
150 'multi-init' => [ \&cmd_multi_init,
151 "Deprecated alias for ".
152 "'$0 init -T<trunk> -b<branches> -t<tags>'",
153 \%init_opts ],
154 dcommit => [ \&cmd_dcommit,
155 'Commit several diffs to merge with upstream',
156 { 'merge|m|M' => \$_merge,
157 'strategy|s=s' => \$_strategy,
158 'verbose|v' => \$_verbose,
159 'dry-run|n' => \$_dry_run,
160 'fetch-all|all' => \$_fetch_all,
161 'commit-url=s' => \$_commit_url,
162 'revision|r=i' => \$_revision,
163 'no-rebase' => \$_no_rebase,
164 'mergeinfo=s' => \$_merge_info,
165 %cmt_opts, %fc_opts } ],
166 branch => [ \&cmd_branch,
167 'Create a branch in the SVN repository',
168 { 'message|m=s' => \$_message,
169 'destination|d=s' => \$_branch_dest,
170 'dry-run|n' => \$_dry_run,
171 'tag|t' => \$_tag,
172 'username=s' => \$Git::SVN::Prompt::_username,
173 'commit-url=s' => \$_commit_url } ],
174 tag => [ sub { $_tag = 1; cmd_branch(@_) },
175 'Create a tag in the SVN repository',
176 { 'message|m=s' => \$_message,
177 'destination|d=s' => \$_branch_dest,
178 'dry-run|n' => \$_dry_run,
179 'username=s' => \$Git::SVN::Prompt::_username,
180 'commit-url=s' => \$_commit_url } ],
181 'set-tree' => [ \&cmd_set_tree,
182 "Set an SVN repository to a git tree-ish",
183 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
184 'create-ignore' => [ \&cmd_create_ignore,
185 'Create a .gitignore per svn:ignore',
186 { 'revision|r=i' => \$_revision
187 } ],
188 'mkdirs' => [ \&cmd_mkdirs ,
189 "recreate empty directories after a checkout",
190 { 'revision|r=i' => \$_revision } ],
191 'propget' => [ \&cmd_propget,
192 'Print the value of a property on a file or directory',
193 { 'revision|r=i' => \$_revision } ],
194 'proplist' => [ \&cmd_proplist,
195 'List all properties of a file or directory',
196 { 'revision|r=i' => \$_revision } ],
197 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
198 { 'revision|r=i' => \$_revision
199 } ],
200 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
201 { 'revision|r=i' => \$_revision
202 } ],
203 'multi-fetch' => [ \&cmd_multi_fetch,
204 "Deprecated alias for $0 fetch --all",
205 { 'revision|r=s' => \$_revision, %fc_opts } ],
206 'migrate' => [ sub { },
207 # no-op, we automatically run this anyways,
208 'Migrate configuration/metadata/layout from
209 previous versions of git-svn',
210 { 'minimize' => \$Git::SVN::Migration::_minimize,
211 %remote_opts } ],
212 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
213 { 'limit=i' => \$Git::SVN::Log::limit,
214 'revision|r=s' => \$_revision,
215 'verbose|v' => \$Git::SVN::Log::verbose,
216 'incremental' => \$Git::SVN::Log::incremental,
217 'oneline' => \$Git::SVN::Log::oneline,
218 'show-commit' => \$Git::SVN::Log::show_commit,
219 'non-recursive' => \$Git::SVN::Log::non_recursive,
220 'authors-file|A=s' => \$_authors,
221 'color' => \$Git::SVN::Log::color,
222 'pager=s' => \$Git::SVN::Log::pager
223 } ],
224 'find-rev' => [ \&cmd_find_rev,
225 "Translate between SVN revision numbers and tree-ish",
226 {} ],
227 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
228 { 'merge|m|M' => \$_merge,
229 'verbose|v' => \$_verbose,
230 'strategy|s=s' => \$_strategy,
231 'local|l' => \$_local,
232 'fetch-all|all' => \$_fetch_all,
233 'dry-run|n' => \$_dry_run,
234 %fc_opts } ],
235 'commit-diff' => [ \&cmd_commit_diff,
236 'Commit a diff between two trees',
237 { 'message|m=s' => \$_message,
238 'file|F=s' => \$_file,
239 'revision|r=s' => \$_revision,
240 %cmt_opts } ],
241 'info' => [ \&cmd_info,
242 "Show info about the latest SVN revision
243 on the current branch",
244 { 'url' => \$_url, } ],
245 'blame' => [ \&Git::SVN::Log::cmd_blame,
246 "Show what revision and author last modified each line of a file",
247 { 'git-format' => \$_git_format } ],
248 'reset' => [ \&cmd_reset,
249 "Undo fetches back to the specified SVN revision",
250 { 'revision|r=s' => \$_revision,
251 'parent|p' => \$_fetch_parent } ],
252 'gc' => [ \&cmd_gc,
253 "Compress unhandled.log files in .git/svn and remove " .
254 "index files in .git/svn",
255 {} ],
256 );
258 my $cmd;
259 for (my $i = 0; $i < @ARGV; $i++) {
260 if (defined $cmd{$ARGV[$i]}) {
261 $cmd = $ARGV[$i];
262 splice @ARGV, $i, 1;
263 last;
264 } elsif ($ARGV[$i] eq 'help') {
265 $cmd = $ARGV[$i+1];
266 usage(0);
267 }
268 };
270 # make sure we're always running at the top-level working directory
271 unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
272 unless (-d $ENV{GIT_DIR}) {
273 if ($git_dir_user_set) {
274 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
275 "but it is not a directory\n";
276 }
277 my $git_dir = delete $ENV{GIT_DIR};
278 my $cdup = undef;
279 git_cmd_try {
280 $cdup = command_oneline(qw/rev-parse --show-cdup/);
281 $git_dir = '.' unless ($cdup);
282 chomp $cdup if ($cdup);
283 $cdup = "." unless ($cdup && length $cdup);
284 } "Already at toplevel, but $git_dir not found\n";
285 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
286 unless (-d $git_dir) {
287 die "$git_dir still not found after going to ",
288 "'$cdup'\n";
289 }
290 $ENV{GIT_DIR} = $git_dir;
291 }
292 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
293 }
295 my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
297 read_git_config(\%opts);
298 if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
299 Getopt::Long::Configure('pass_through');
300 }
301 my $rv = GetOptions(%opts, 'h|H' => \$_help, 'version|V' => \$_version,
302 'minimize-connections' => \$Git::SVN::Migration::_minimize,
303 'id|i=s' => \$Git::SVN::default_ref_id,
304 'svn-remote|remote|R=s' => sub {
305 $Git::SVN::no_reuse_existing = 1;
306 $Git::SVN::default_repo_id = $_[1] });
307 exit 1 if (!$rv && $cmd && $cmd ne 'log');
309 usage(0) if $_help;
310 version() if $_version;
311 usage(1) unless defined $cmd;
312 load_authors() if $_authors;
313 if (defined $_authors_prog) {
314 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
315 }
317 unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
318 Git::SVN::Migration::migration_check();
319 }
320 Git::SVN::init_vars();
321 eval {
322 Git::SVN::verify_remotes_sanity();
323 $cmd{$cmd}->[0]->(@ARGV);
324 };
325 fatal $@ if $@;
326 post_fetch_checkout();
327 exit 0;
329 ####################### primary functions ######################
330 sub usage {
331 my $exit = shift || 0;
332 my $fd = $exit ? \*STDERR : \*STDOUT;
333 print $fd <<"";
334 git-svn - bidirectional operations between a single Subversion tree and git
335 Usage: git svn <command> [options] [arguments]\n
337 print $fd "Available commands:\n" unless $cmd;
339 foreach (sort keys %cmd) {
340 next if $cmd && $cmd ne $_;
341 next if /^multi-/; # don't show deprecated commands
342 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
343 foreach (sort keys %{$cmd{$_}->[2]}) {
344 # mixed-case options are for .git/config only
345 next if /[A-Z]/ && /^[a-z]+$/i;
346 # prints out arguments as they should be passed:
347 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
348 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
349 "--$_" : "-$_" }
350 split /\|/,$_)," $x\n";
351 }
352 }
353 print $fd <<"";
354 \nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
355 arbitrary identifier if you're tracking multiple SVN branches/repositories in
356 one git repository and want to keep them separate. See git-svn(1) for more
357 information.
359 exit $exit;
360 }
362 sub version {
363 ::_req_svn();
364 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
365 exit 0;
366 }
368 sub do_git_init_db {
369 unless (-d $ENV{GIT_DIR}) {
370 my @init_db = ('init');
371 push @init_db, "--template=$_template" if defined $_template;
372 if (defined $_shared) {
373 if ($_shared =~ /[a-z]/) {
374 push @init_db, "--shared=$_shared";
375 } else {
376 push @init_db, "--shared";
377 }
378 }
379 command_noisy(@init_db);
380 $_repository = Git->repository(Repository => ".git");
381 }
382 my $set;
383 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
384 foreach my $i (keys %icv) {
385 die "'$set' and '$i' cannot both be set\n" if $set;
386 next unless defined $icv{$i};
387 command_noisy('config', "$pfx.$i", $icv{$i});
388 $set = $i;
389 }
390 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
391 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
392 if defined $$ignore_regex;
394 if (defined $SVN::Git::Fetcher::_preserve_empty_dirs) {
395 my $fname = \$SVN::Git::Fetcher::_placeholder_filename;
396 command_noisy('config', "$pfx.preserve-empty-dirs", 'true');
397 command_noisy('config', "$pfx.placeholder-filename", $$fname);
398 }
399 }
401 sub init_subdir {
402 my $repo_path = shift or return;
403 mkpath([$repo_path]) unless -d $repo_path;
404 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
405 $ENV{GIT_DIR} = '.git';
406 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
407 }
409 sub cmd_clone {
410 my ($url, $path) = @_;
411 if (!defined $path &&
412 (defined $_trunk || @_branches || @_tags ||
413 defined $_stdlayout) &&
414 $url !~ m#^[a-z\+]+://#) {
415 $path = $url;
416 }
417 $path = basename($url) if !defined $path || !length $path;
418 my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
419 cmd_init($url, $path);
420 command_oneline('config', 'svn.authorsfile', $authors_absolute)
421 if $_authors;
422 Git::SVN::fetch_all($Git::SVN::default_repo_id);
423 }
425 sub cmd_init {
426 if (defined $_stdlayout) {
427 $_trunk = 'trunk' if (!defined $_trunk);
428 @_tags = 'tags' if (! @_tags);
429 @_branches = 'branches' if (! @_branches);
430 }
431 if (defined $_trunk || @_branches || @_tags) {
432 return cmd_multi_init(@_);
433 }
434 my $url = shift or die "SVN repository location required ",
435 "as a command-line argument\n";
436 $url = canonicalize_url($url);
437 init_subdir(@_);
438 do_git_init_db();
440 if ($Git::SVN::_minimize_url eq 'unset') {
441 $Git::SVN::_minimize_url = 0;
442 }
444 Git::SVN->init($url);
445 }
447 sub cmd_fetch {
448 if (grep /^\d+=./, @_) {
449 die "'<rev>=<commit>' fetch arguments are ",
450 "no longer supported.\n";
451 }
452 my ($remote) = @_;
453 if (@_ > 1) {
454 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
455 }
456 $Git::SVN::no_reuse_existing = undef;
457 if ($_fetch_parent) {
458 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
459 unless ($gs) {
460 die "Unable to determine upstream SVN information from ",
461 "working tree history\n";
462 }
463 # just fetch, don't checkout.
464 $_no_checkout = 'true';
465 $_fetch_all ? $gs->fetch_all : $gs->fetch;
466 } elsif ($_fetch_all) {
467 cmd_multi_fetch();
468 } else {
469 $remote ||= $Git::SVN::default_repo_id;
470 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
471 }
472 }
474 sub cmd_set_tree {
475 my (@commits) = @_;
476 if ($_stdin || !@commits) {
477 print "Reading from stdin...\n";
478 @commits = ();
479 while (<STDIN>) {
480 if (/\b($sha1_short)\b/o) {
481 unshift @commits, $1;
482 }
483 }
484 }
485 my @revs;
486 foreach my $c (@commits) {
487 my @tmp = command('rev-parse',$c);
488 if (scalar @tmp == 1) {
489 push @revs, $tmp[0];
490 } elsif (scalar @tmp > 1) {
491 push @revs, reverse(command('rev-list',@tmp));
492 } else {
493 fatal "Failed to rev-parse $c";
494 }
495 }
496 my $gs = Git::SVN->new;
497 my ($r_last, $cmt_last) = $gs->last_rev_commit;
498 $gs->fetch;
499 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
500 fatal "There are new revisions that were fetched ",
501 "and need to be merged (or acknowledged) ",
502 "before committing.\nlast rev: $r_last\n",
503 " current: $gs->{last_rev}";
504 }
505 $gs->set_tree($_) foreach @revs;
506 print "Done committing ",scalar @revs," revisions to SVN\n";
507 unlink $gs->{index};
508 }
510 sub split_merge_info_range {
511 my ($range) = @_;
512 if ($range =~ /(\d+)-(\d+)/) {
513 return (int($1), int($2));
514 } else {
515 return (int($range), int($range));
516 }
517 }
519 sub combine_ranges {
520 my ($in) = @_;
522 my @fnums = ();
523 my @arr = split(/,/, $in);
524 for my $element (@arr) {
525 my ($start, $end) = split_merge_info_range($element);
526 push @fnums, $start;
527 }
529 my @sorted = @arr [ sort {
530 $fnums[$a] <=> $fnums[$b]
531 } 0..$#arr ];
533 my @return = ();
534 my $last = -1;
535 my $first = -1;
536 for my $element (@sorted) {
537 my ($start, $end) = split_merge_info_range($element);
539 if ($last == -1) {
540 $first = $start;
541 $last = $end;
542 next;
543 }
544 if ($start <= $last+1) {
545 if ($end > $last) {
546 $last = $end;
547 }
548 next;
549 }
550 if ($first == $last) {
551 push @return, "$first";
552 } else {
553 push @return, "$first-$last";
554 }
555 $first = $start;
556 $last = $end;
557 }
559 if ($first != -1) {
560 if ($first == $last) {
561 push @return, "$first";
562 } else {
563 push @return, "$first-$last";
564 }
565 }
567 return join(',', @return);
568 }
570 sub merge_revs_into_hash {
571 my ($hash, $minfo) = @_;
572 my @lines = split(' ', $minfo);
574 for my $line (@lines) {
575 my ($branchpath, $revs) = split(/:/, $line);
577 if (exists($hash->{$branchpath})) {
578 # Merge the two revision sets
579 my $combined = "$hash->{$branchpath},$revs";
580 $hash->{$branchpath} = combine_ranges($combined);
581 } else {
582 # Just do range combining for consolidation
583 $hash->{$branchpath} = combine_ranges($revs);
584 }
585 }
586 }
588 sub merge_merge_info {
589 my ($mergeinfo_one, $mergeinfo_two) = @_;
590 my %result_hash = ();
592 merge_revs_into_hash(\%result_hash, $mergeinfo_one);
593 merge_revs_into_hash(\%result_hash, $mergeinfo_two);
595 my $result = '';
596 # Sort below is for consistency's sake
597 for my $branchname (sort keys(%result_hash)) {
598 my $revlist = $result_hash{$branchname};
599 $result .= "$branchname:$revlist\n"
600 }
601 return $result;
602 }
604 sub populate_merge_info {
605 my ($d, $gs, $uuid, $linear_refs, $rewritten_parent) = @_;
607 my %parentshash;
608 read_commit_parents(\%parentshash, $d);
609 my @parents = @{$parentshash{$d}};
610 if ($#parents > 0) {
611 # Merge commit
612 my $all_parents_ok = 1;
613 my $aggregate_mergeinfo = '';
614 my $rooturl = $gs->repos_root;
616 if (defined($rewritten_parent)) {
617 # Replace first parent with newly-rewritten version
618 shift @parents;
619 unshift @parents, $rewritten_parent;
620 }
622 foreach my $parent (@parents) {
623 my ($branchurl, $svnrev, $paruuid) =
624 cmt_metadata($parent);
626 unless (defined($svnrev)) {
627 # Should have been caught be preflight check
628 fatal "merge commit $d has ancestor $parent, but that change "
629 ."does not have git-svn metadata!";
630 }
631 unless ($branchurl =~ /^$rooturl(.*)/) {
632 fatal "commit $parent git-svn metadata changed mid-run!";
633 }
634 my $branchpath = $1;
636 my $ra = Git::SVN::Ra->new($branchurl);
637 my (undef, undef, $props) =
638 $ra->get_dir(canonicalize_path("."), $svnrev);
639 my $par_mergeinfo = $props->{'svn:mergeinfo'};
640 unless (defined $par_mergeinfo) {
641 $par_mergeinfo = '';
642 }
643 # Merge previous mergeinfo values
644 $aggregate_mergeinfo =
645 merge_merge_info($aggregate_mergeinfo,
646 $par_mergeinfo, 0);
648 next if $parent eq $parents[0]; # Skip first parent
649 # Add new changes being placed in tree by merge
650 my @cmd = (qw/rev-list --reverse/,
651 $parent, qw/--not/);
652 foreach my $par (@parents) {
653 unless ($par eq $parent) {
654 push @cmd, $par;
655 }
656 }
657 my @revsin = ();
658 my ($revlist, $ctx) = command_output_pipe(@cmd);
659 while (<$revlist>) {
660 my $irev = $_;
661 chomp $irev;
662 my (undef, $csvnrev, undef) =
663 cmt_metadata($irev);
664 unless (defined $csvnrev) {
665 # A child is missing SVN annotations...
666 # this might be OK, or might not be.
667 warn "W:child $irev is merged into revision "
668 ."$d but does not have git-svn metadata. "
669 ."This means git-svn cannot determine the "
670 ."svn revision numbers to place into the "
671 ."svn:mergeinfo property. You must ensure "
672 ."a branch is entirely committed to "
673 ."SVN before merging it in order for "
674 ."svn:mergeinfo population to function "
675 ."properly";
676 }
677 push @revsin, $csvnrev;
678 }
679 command_close_pipe($revlist, $ctx);
681 last unless $all_parents_ok;
683 # We now have a list of all SVN revnos which are
684 # merged by this particular parent. Integrate them.
685 next if $#revsin == -1;
686 my $newmergeinfo = "$branchpath:" . join(',', @revsin);
687 $aggregate_mergeinfo =
688 merge_merge_info($aggregate_mergeinfo,
689 $newmergeinfo, 1);
690 }
691 if ($all_parents_ok and $aggregate_mergeinfo) {
692 return $aggregate_mergeinfo;
693 }
694 }
696 return undef;
697 }
699 sub cmd_dcommit {
700 my $head = shift;
701 command_noisy(qw/update-index --refresh/);
702 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
703 'Cannot dcommit with a dirty index. Commit your changes first, '
704 . "or stash them with `git stash'.\n";
705 $head ||= 'HEAD';
707 my $old_head;
708 if ($head ne 'HEAD') {
709 $old_head = eval {
710 command_oneline([qw/symbolic-ref -q HEAD/])
711 };
712 if ($old_head) {
713 $old_head =~ s{^refs/heads/}{};
714 } else {
715 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
716 }
717 command(['checkout', $head], STDERR => 0);
718 }
720 my @refs;
721 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
722 unless ($gs) {
723 die "Unable to determine upstream SVN information from ",
724 "$head history.\nPerhaps the repository is empty.";
725 }
727 if (defined $_commit_url) {
728 $url = $_commit_url;
729 } else {
730 $url = eval { command_oneline('config', '--get',
731 "svn-remote.$gs->{repo_id}.commiturl") };
732 if (!$url) {
733 $url = $gs->full_pushurl
734 }
735 }
737 my $last_rev = $_revision if defined $_revision;
738 if ($url) {
739 print "Committing to $url ...\n";
740 }
741 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
742 if ($_no_rebase && scalar(@$linear_refs) > 1) {
743 warn "Attempting to commit more than one change while ",
744 "--no-rebase is enabled.\n",
745 "If these changes depend on each other, re-running ",
746 "without --no-rebase may be required."
747 }
748 my $expect_url = $url;
750 my $push_merge_info = eval {
751 command_oneline(qw/config --get svn.pushmergeinfo/)
752 };
753 if (not defined($push_merge_info)
754 or $push_merge_info eq "false"
755 or $push_merge_info eq "no"
756 or $push_merge_info eq "never") {
757 $push_merge_info = 0;
758 }
760 unless (defined($_merge_info) || ! $push_merge_info) {
761 # Preflight check of changes to ensure no issues with mergeinfo
762 # This includes check for uncommitted-to-SVN parents
763 # (other than the first parent, which we will handle),
764 # information from different SVN repos, and paths
765 # which are not underneath this repository root.
766 my $rooturl = $gs->repos_root;
767 foreach my $d (@$linear_refs) {
768 my %parentshash;
769 read_commit_parents(\%parentshash, $d);
770 my @realparents = @{$parentshash{$d}};
771 if ($#realparents > 0) {
772 # Merge commit
773 shift @realparents; # Remove/ignore first parent
774 foreach my $parent (@realparents) {
775 my ($branchurl, $svnrev, $paruuid) = cmt_metadata($parent);
776 unless (defined $paruuid) {
777 # A parent is missing SVN annotations...
778 # abort the whole operation.
779 fatal "$parent is merged into revision $d, "
780 ."but does not have git-svn metadata. "
781 ."Either dcommit the branch or use a "
782 ."local cherry-pick, FF merge, or rebase "
783 ."instead of an explicit merge commit.";
784 }
786 unless ($paruuid eq $uuid) {
787 # Parent has SVN metadata from different repository
788 fatal "merge parent $parent for change $d has "
789 ."git-svn uuid $paruuid, while current change "
790 ."has uuid $uuid!";
791 }
793 unless ($branchurl =~ /^$rooturl(.*)/) {
794 # This branch is very strange indeed.
795 fatal "merge parent $parent for $d is on branch "
796 ."$branchurl, which is not under the "
797 ."git-svn root $rooturl!";
798 }
799 }
800 }
801 }
802 }
804 my $rewritten_parent;
805 Git::SVN::remove_username($expect_url);
806 if (defined($_merge_info)) {
807 $_merge_info =~ tr{ }{\n};
808 }
809 while (1) {
810 my $d = shift @$linear_refs or last;
811 unless (defined $last_rev) {
812 (undef, $last_rev, undef) = cmt_metadata("$d~1");
813 unless (defined $last_rev) {
814 fatal "Unable to extract revision information ",
815 "from commit $d~1";
816 }
817 }
818 if ($_dry_run) {
819 print "diff-tree $d~1 $d\n";
820 } else {
821 my $cmt_rev;
823 unless (defined($_merge_info) || ! $push_merge_info) {
824 $_merge_info = populate_merge_info($d, $gs,
825 $uuid,
826 $linear_refs,
827 $rewritten_parent);
828 }
830 my %ed_opts = ( r => $last_rev,
831 log => get_commit_entry($d)->{log},
832 ra => Git::SVN::Ra->new($url),
833 config => SVN::Core::config_get_config(
834 $Git::SVN::Ra::config_dir
835 ),
836 tree_a => "$d~1",
837 tree_b => $d,
838 editor_cb => sub {
839 print "Committed r$_[0]\n";
840 $cmt_rev = $_[0];
841 },
842 mergeinfo => $_merge_info,
843 svn_path => '');
844 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
845 print "No changes\n$d~1 == $d\n";
846 } elsif ($parents->{$d} && @{$parents->{$d}}) {
847 $gs->{inject_parents_dcommit}->{$cmt_rev} =
848 $parents->{$d};
849 }
850 $_fetch_all ? $gs->fetch_all : $gs->fetch;
851 $last_rev = $cmt_rev;
852 next if $_no_rebase;
854 # we always want to rebase against the current HEAD,
855 # not any head that was passed to us
856 my @diff = command('diff-tree', $d,
857 $gs->refname, '--');
858 my @finish;
859 if (@diff) {
860 @finish = rebase_cmd();
861 print STDERR "W: $d and ", $gs->refname,
862 " differ, using @finish:\n",
863 join("\n", @diff), "\n";
864 } else {
865 print "No changes between current HEAD and ",
866 $gs->refname,
867 "\nResetting to the latest ",
868 $gs->refname, "\n";
869 @finish = qw/reset --mixed/;
870 }
871 command_noisy(@finish, $gs->refname);
873 $rewritten_parent = command_oneline(qw/rev-parse HEAD/);
875 if (@diff) {
876 @refs = ();
877 my ($url_, $rev_, $uuid_, $gs_) =
878 working_head_info('HEAD', \@refs);
879 my ($linear_refs_, $parents_) =
880 linearize_history($gs_, \@refs);
881 if (scalar(@$linear_refs) !=
882 scalar(@$linear_refs_)) {
883 fatal "# of revisions changed ",
884 "\nbefore:\n",
885 join("\n", @$linear_refs),
886 "\n\nafter:\n",
887 join("\n", @$linear_refs_), "\n",
888 'If you are attempting to commit ',
889 "merges, try running:\n\t",
890 'git rebase --interactive',
891 '--preserve-merges ',
892 $gs->refname,
893 "\nBefore dcommitting";
894 }
895 if ($url_ ne $expect_url) {
896 if ($url_ eq $gs->metadata_url) {
897 print
898 "Accepting rewritten URL:",
899 " $url_\n";
900 } else {
901 fatal
902 "URL mismatch after rebase:",
903 " $url_ != $expect_url";
904 }
905 }
906 if ($uuid_ ne $uuid) {
907 fatal "uuid mismatch after rebase: ",
908 "$uuid_ != $uuid";
909 }
910 # remap parents
911 my (%p, @l, $i);
912 for ($i = 0; $i < scalar @$linear_refs; $i++) {
913 my $new = $linear_refs_->[$i] or next;
914 $p{$new} =
915 $parents->{$linear_refs->[$i]};
916 push @l, $new;
917 }
918 $parents = \%p;
919 $linear_refs = \@l;
920 }
921 }
922 }
924 if ($old_head) {
925 my $new_head = command_oneline(qw/rev-parse HEAD/);
926 my $new_is_symbolic = eval {
927 command_oneline(qw/symbolic-ref -q HEAD/);
928 };
929 if ($new_is_symbolic) {
930 print "dcommitted the branch ", $head, "\n";
931 } else {
932 print "dcommitted on a detached HEAD because you gave ",
933 "a revision argument.\n",
934 "The rewritten commit is: ", $new_head, "\n";
935 }
936 command(['checkout', $old_head], STDERR => 0);
937 }
939 unlink $gs->{index};
940 }
942 sub cmd_branch {
943 my ($branch_name, $head) = @_;
945 unless (defined $branch_name && length $branch_name) {
946 die(($_tag ? "tag" : "branch") . " name required\n");
947 }
948 $head ||= 'HEAD';
950 my (undef, $rev, undef, $gs) = working_head_info($head);
951 my $src = $gs->full_pushurl;
953 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
954 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
955 my $glob;
956 if ($#{$allglobs} == 0) {
957 $glob = $allglobs->[0];
958 } else {
959 unless(defined $_branch_dest) {
960 die "Multiple ",
961 $_tag ? "tag" : "branch",
962 " paths defined for Subversion repository.\n",
963 "You must specify where you want to create the ",
964 $_tag ? "tag" : "branch",
965 " with the --destination argument.\n";
966 }
967 foreach my $g (@{$allglobs}) {
968 # SVN::Git::Editor could probably be moved to Git.pm..
969 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
970 if ($_branch_dest =~ /$re/) {
971 $glob = $g;
972 last;
973 }
974 }
975 unless (defined $glob) {
976 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
977 foreach my $g (@{$allglobs}) {
978 $g->{path}->{left} =~ /$dest_re/ or next;
979 if (defined $glob) {
980 die "Ambiguous destination: ",
981 $_branch_dest, "\nmatches both '",
982 $glob->{path}->{left}, "' and '",
983 $g->{path}->{left}, "'\n";
984 }
985 $glob = $g;
986 }
987 unless (defined $glob) {
988 die "Unknown ",
989 $_tag ? "tag" : "branch",
990 " destination $_branch_dest\n";
991 }
992 }
993 }
994 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
995 my $url;
996 if (defined $_commit_url) {
997 $url = $_commit_url;
998 } else {
999 $url = eval { command_oneline('config', '--get',
1000 "svn-remote.$gs->{repo_id}.commiturl") };
1001 if (!$url) {
1002 $url = $remote->{pushurl} || $remote->{url};
1003 }
1004 }
1005 my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
1007 if ($dst =~ /^https:/ && $src =~ /^http:/) {
1008 $src=~s/^http:/https:/;
1009 }
1011 ::_req_svn();
1013 my $ctx = SVN::Client->new(
1014 auth => Git::SVN::Ra::_auth_providers(),
1015 log_msg => sub {
1016 ${ $_[0] } = defined $_message
1017 ? $_message
1018 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
1019 . $branch_name;
1020 },
1021 );
1023 eval {
1024 $ctx->ls($dst, 'HEAD', 0);
1025 } and die "branch ${branch_name} already exists\n";
1027 print "Copying ${src} at r${rev} to ${dst}...\n";
1028 $ctx->copy($src, $rev, $dst)
1029 unless $_dry_run;
1031 $gs->fetch_all;
1032 }
1034 sub cmd_find_rev {
1035 my $revision_or_hash = shift or die "SVN or git revision required ",
1036 "as a command-line argument\n";
1037 my $result;
1038 if ($revision_or_hash =~ /^r\d+$/) {
1039 my $head = shift;
1040 $head ||= 'HEAD';
1041 my @refs;
1042 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
1043 unless ($gs) {
1044 die "Unable to determine upstream SVN information from ",
1045 "$head history\n";
1046 }
1047 my $desired_revision = substr($revision_or_hash, 1);
1048 $result = $gs->rev_map_get($desired_revision, $uuid);
1049 } else {
1050 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
1051 $result = $rev;
1052 }
1053 print "$result\n" if $result;
1054 }
1056 sub auto_create_empty_directories {
1057 my ($gs) = @_;
1058 my $var = eval { command_oneline('config', '--get', '--bool',
1059 "svn-remote.$gs->{repo_id}.automkdirs") };
1060 # By default, create empty directories by consulting the unhandled log,
1061 # but allow setting it to 'false' to skip it.
1062 return !($var && $var eq 'false');
1063 }
1065 sub cmd_rebase {
1066 command_noisy(qw/update-index --refresh/);
1067 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1068 unless ($gs) {
1069 die "Unable to determine upstream SVN information from ",
1070 "working tree history\n";
1071 }
1072 if ($_dry_run) {
1073 print "Remote Branch: " . $gs->refname . "\n";
1074 print "SVN URL: " . $url . "\n";
1075 return;
1076 }
1077 if (command(qw/diff-index HEAD --/)) {
1078 print STDERR "Cannot rebase with uncommited changes:\n";
1079 command_noisy('status');
1080 exit 1;
1081 }
1082 unless ($_local) {
1083 # rebase will checkout for us, so no need to do it explicitly
1084 $_no_checkout = 'true';
1085 $_fetch_all ? $gs->fetch_all : $gs->fetch;
1086 }
1087 command_noisy(rebase_cmd(), $gs->refname);
1088 if (auto_create_empty_directories($gs)) {
1089 $gs->mkemptydirs;
1090 }
1091 }
1093 sub cmd_show_ignore {
1094 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1095 $gs ||= Git::SVN->new;
1096 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1097 $gs->prop_walk($gs->{path}, $r, sub {
1098 my ($gs, $path, $props) = @_;
1099 print STDOUT "\n# $path\n";
1100 my $s = $props->{'svn:ignore'} or return;
1101 $s =~ s/[\r\n]+/\n/g;
1102 $s =~ s/^\n+//;
1103 chomp $s;
1104 $s =~ s#^#$path#gm;
1105 print STDOUT "$s\n";
1106 });
1107 }
1109 sub cmd_show_externals {
1110 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1111 $gs ||= Git::SVN->new;
1112 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1113 $gs->prop_walk($gs->{path}, $r, sub {
1114 my ($gs, $path, $props) = @_;
1115 print STDOUT "\n# $path\n";
1116 my $s = $props->{'svn:externals'} or return;
1117 $s =~ s/[\r\n]+/\n/g;
1118 chomp $s;
1119 $s =~ s#^#$path#gm;
1120 print STDOUT "$s\n";
1121 });
1122 }
1124 sub cmd_create_ignore {
1125 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1126 $gs ||= Git::SVN->new;
1127 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1128 $gs->prop_walk($gs->{path}, $r, sub {
1129 my ($gs, $path, $props) = @_;
1130 # $path is of the form /path/to/dir/
1131 $path = '.' . $path;
1132 # SVN can have attributes on empty directories,
1133 # which git won't track
1134 mkpath([$path]) unless -d $path;
1135 my $ignore = $path . '.gitignore';
1136 my $s = $props->{'svn:ignore'} or return;
1137 open(GITIGNORE, '>', $ignore)
1138 or fatal("Failed to open `$ignore' for writing: $!");
1139 $s =~ s/[\r\n]+/\n/g;
1140 $s =~ s/^\n+//;
1141 chomp $s;
1142 # Prefix all patterns so that the ignore doesn't apply
1143 # to sub-directories.
1144 $s =~ s#^#/#gm;
1145 print GITIGNORE "$s\n";
1146 close(GITIGNORE)
1147 or fatal("Failed to close `$ignore': $!");
1148 command_noisy('add', '-f', $ignore);
1149 });
1150 }
1152 sub cmd_mkdirs {
1153 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1154 $gs ||= Git::SVN->new;
1155 $gs->mkemptydirs($_revision);
1156 }
1158 sub canonicalize_path {
1159 my ($path) = @_;
1160 my $dot_slash_added = 0;
1161 if (substr($path, 0, 1) ne "/") {
1162 $path = "./" . $path;
1163 $dot_slash_added = 1;
1164 }
1165 # File::Spec->canonpath doesn't collapse x/../y into y (for a
1166 # good reason), so let's do this manually.
1167 $path =~ s#/+#/#g;
1168 $path =~ s#/\.(?:/|$)#/#g;
1169 $path =~ s#/[^/]+/\.\.##g;
1170 $path =~ s#/$##g;
1171 $path =~ s#^\./## if $dot_slash_added;
1172 $path =~ s#^/##;
1173 $path =~ s#^\.$##;
1174 return $path;
1175 }
1177 sub canonicalize_url {
1178 my ($url) = @_;
1179 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
1180 return $url;
1181 }
1183 # get_svnprops(PATH)
1184 # ------------------
1185 # Helper for cmd_propget and cmd_proplist below.
1186 sub get_svnprops {
1187 my $path = shift;
1188 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1189 $gs ||= Git::SVN->new;
1191 # prefix THE PATH by the sub-directory from which the user
1192 # invoked us.
1193 $path = $cmd_dir_prefix . $path;
1194 fatal("No such file or directory: $path") unless -e $path;
1195 my $is_dir = -d $path ? 1 : 0;
1196 $path = $gs->{path} . '/' . $path;
1198 # canonicalize the path (otherwise libsvn will abort or fail to
1199 # find the file)
1200 $path = canonicalize_path($path);
1202 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
1203 my $props;
1204 if ($is_dir) {
1205 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
1206 }
1207 else {
1208 (undef, $props) = $gs->ra->get_file($path, $r, undef);
1209 }
1210 return $props;
1211 }
1213 # cmd_propget (PROP, PATH)
1214 # ------------------------
1215 # Print the SVN property PROP for PATH.
1216 sub cmd_propget {
1217 my ($prop, $path) = @_;
1218 $path = '.' if not defined $path;
1219 usage(1) if not defined $prop;
1220 my $props = get_svnprops($path);
1221 if (not defined $props->{$prop}) {
1222 fatal("`$path' does not have a `$prop' SVN property.");
1223 }
1224 print $props->{$prop} . "\n";
1225 }
1227 # cmd_proplist (PATH)
1228 # -------------------
1229 # Print the list of SVN properties for PATH.
1230 sub cmd_proplist {
1231 my $path = shift;
1232 $path = '.' if not defined $path;
1233 my $props = get_svnprops($path);
1234 print "Properties on '$path':\n";
1235 foreach (sort keys %{$props}) {
1236 print " $_\n";
1237 }
1238 }
1240 sub cmd_multi_init {
1241 my $url = shift;
1242 unless (defined $_trunk || @_branches || @_tags) {
1243 usage(1);
1244 }
1246 $_prefix = '' unless defined $_prefix;
1247 if (defined $url) {
1248 $url = canonicalize_url($url);
1249 init_subdir(@_);
1250 }
1251 do_git_init_db();
1252 if (defined $_trunk) {
1253 $_trunk =~ s#^/+##;
1254 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
1255 # try both old-style and new-style lookups:
1256 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
1257 unless ($gs_trunk) {
1258 my ($trunk_url, $trunk_path) =
1259 complete_svn_url($url, $_trunk);
1260 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
1261 undef, $trunk_ref);
1262 }
1263 }
1264 return unless @_branches || @_tags;
1265 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
1266 foreach my $path (@_branches) {
1267 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
1268 }
1269 foreach my $path (@_tags) {
1270 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
1271 }
1272 }
1274 sub cmd_multi_fetch {
1275 $Git::SVN::no_reuse_existing = undef;
1276 my $remotes = Git::SVN::read_all_remotes();
1277 foreach my $repo_id (sort keys %$remotes) {
1278 if ($remotes->{$repo_id}->{url}) {
1279 Git::SVN::fetch_all($repo_id, $remotes);
1280 }
1281 }
1282 }
1284 # this command is special because it requires no metadata
1285 sub cmd_commit_diff {
1286 my ($ta, $tb, $url) = @_;
1287 my $usage = "Usage: $0 commit-diff -r<revision> ".
1288 "<tree-ish> <tree-ish> [<URL>]";
1289 fatal($usage) if (!defined $ta || !defined $tb);
1290 my $svn_path = '';
1291 if (!defined $url) {
1292 my $gs = eval { Git::SVN->new };
1293 if (!$gs) {
1294 fatal("Needed URL or usable git-svn --id in ",
1295 "the command-line\n", $usage);
1296 }
1297 $url = $gs->{url};
1298 $svn_path = $gs->{path};
1299 }
1300 unless (defined $_revision) {
1301 fatal("-r|--revision is a required argument\n", $usage);
1302 }
1303 if (defined $_message && defined $_file) {
1304 fatal("Both --message/-m and --file/-F specified ",
1305 "for the commit message.\n",
1306 "I have no idea what you mean");
1307 }
1308 if (defined $_file) {
1309 $_message = file_to_s($_file);
1310 } else {
1311 $_message ||= get_commit_entry($tb)->{log};
1312 }
1313 my $ra ||= Git::SVN::Ra->new($url);
1314 my $r = $_revision;
1315 if ($r eq 'HEAD') {
1316 $r = $ra->get_latest_revnum;
1317 } elsif ($r !~ /^\d+$/) {
1318 die "revision argument: $r not understood by git-svn\n";
1319 }
1320 my %ed_opts = ( r => $r,
1321 log => $_message,
1322 ra => $ra,
1323 tree_a => $ta,
1324 tree_b => $tb,
1325 editor_cb => sub { print "Committed r$_[0]\n" },
1326 svn_path => $svn_path );
1327 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1328 print "No changes\n$ta == $tb\n";
1329 }
1330 }
1332 sub escape_uri_only {
1333 my ($uri) = @_;
1334 my @tmp;
1335 foreach (split m{/}, $uri) {
1336 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1337 push @tmp, $_;
1338 }
1339 join('/', @tmp);
1340 }
1342 sub escape_url {
1343 my ($url) = @_;
1344 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1345 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1346 $url = "$scheme://$domain$uri";
1347 }
1348 $url;
1349 }
1351 sub cmd_info {
1352 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1353 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1354 if (exists $_[1]) {
1355 die "Too many arguments specified\n";
1356 }
1358 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1360 if (!$file_type && !$diff_status) {
1361 print STDERR "svn: '$path' is not under version control\n";
1362 exit 1;
1363 }
1365 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1366 unless ($gs) {
1367 die "Unable to determine upstream SVN information from ",
1368 "working tree history\n";
1369 }
1371 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1372 $path = "." if $path eq "";
1374 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1376 if ($_url) {
1377 print escape_url($full_url), "\n";
1378 return;
1379 }
1381 my $result = "Path: $path\n";
1382 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1383 $result .= "URL: " . escape_url($full_url) . "\n";
1385 eval {
1386 my $repos_root = $gs->repos_root;
1387 Git::SVN::remove_username($repos_root);
1388 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1389 };
1390 if ($@) {
1391 $result .= "Repository Root: (offline)\n";
1392 }
1393 ::_req_svn();
1394 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1395 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1396 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1398 $result .= "Node Kind: " .
1399 ($file_type eq "dir" ? "directory" : "file") . "\n";
1401 my $schedule = $diff_status eq "A"
1402 ? "add"
1403 : ($diff_status eq "D" ? "delete" : "normal");
1404 $result .= "Schedule: $schedule\n";
1406 if ($diff_status eq "A") {
1407 print $result, "\n";
1408 return;
1409 }
1411 my ($lc_author, $lc_rev, $lc_date_utc);
1412 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1413 my $log = command_output_pipe(@args);
1414 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1415 while (<$log>) {
1416 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1417 $lc_author = $1;
1418 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1419 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1420 (undef, $lc_rev, undef) = ::extract_metadata($1);
1421 }
1422 }
1423 close $log;
1425 Git::SVN::Log::set_local_timezone();
1427 $result .= "Last Changed Author: $lc_author\n";
1428 $result .= "Last Changed Rev: $lc_rev\n";
1429 $result .= "Last Changed Date: " .
1430 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1432 if ($file_type ne "dir") {
1433 my $text_last_updated_date =
1434 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1435 $result .=
1436 "Text Last Updated: " .
1437 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1438 "\n";
1439 my $checksum;
1440 if ($diff_status eq "D") {
1441 my ($fh, $ctx) =
1442 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1443 if ($file_type eq "link") {
1444 my $file_name = <$fh>;
1445 $checksum = md5sum("link $file_name");
1446 } else {
1447 $checksum = md5sum($fh);
1448 }
1449 command_close_pipe($fh, $ctx);
1450 } elsif ($file_type eq "link") {
1451 my $file_name =
1452 command(qw(cat-file blob), "HEAD:$path");
1453 $checksum =
1454 md5sum("link " . $file_name);
1455 } else {
1456 open FILE, "<", $path or die $!;
1457 $checksum = md5sum(\*FILE);
1458 close FILE or die $!;
1459 }
1460 $result .= "Checksum: " . $checksum . "\n";
1461 }
1463 print $result, "\n";
1464 }
1466 sub cmd_reset {
1467 my $target = shift || $_revision or die "SVN revision required\n";
1468 $target = $1 if $target =~ /^r(\d+)$/;
1469 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1470 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1471 unless ($gs) {
1472 die "Unable to determine upstream SVN information from ".
1473 "history\n";
1474 }
1475 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1476 die "Cannot find SVN revision $target\n" unless defined($c);
1477 $gs->rev_map_set($r, $c, 'reset', $uuid);
1478 print "r$r = $c ($gs->{ref_id})\n";
1479 }
1481 sub cmd_gc {
1482 if (!$can_compress) {
1483 warn "Compress::Zlib could not be found; unhandled.log " .
1484 "files will not be compressed.\n";
1485 }
1486 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1487 }
1489 ########################### utility functions #########################
1491 sub rebase_cmd {
1492 my @cmd = qw/rebase/;
1493 push @cmd, '-v' if $_verbose;
1494 push @cmd, qw/--merge/ if $_merge;
1495 push @cmd, "--strategy=$_strategy" if $_strategy;
1496 @cmd;
1497 }
1499 sub post_fetch_checkout {
1500 return if $_no_checkout;
1501 my $gs = $Git::SVN::_head or return;
1502 return if verify_ref('refs/heads/master^0');
1504 # look for "trunk" ref if it exists
1505 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1506 my $fetch = $remote->{fetch};
1507 if ($fetch) {
1508 foreach my $p (keys %$fetch) {
1509 basename($fetch->{$p}) eq 'trunk' or next;
1510 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1511 last;
1512 }
1513 }
1515 my $valid_head = verify_ref('HEAD^0');
1516 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1517 return if ($valid_head || !verify_ref('HEAD^0'));
1519 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1520 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1521 return if -f $index;
1523 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1524 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1525 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1526 print STDERR "Checked out HEAD:\n ",
1527 $gs->full_url, " r", $gs->last_rev, "\n";
1528 if (auto_create_empty_directories($gs)) {
1529 $gs->mkemptydirs($gs->last_rev);
1530 }
1531 }
1533 sub complete_svn_url {
1534 my ($url, $path) = @_;
1535 $path =~ s#/+$##;
1536 if ($path !~ m#^[a-z\+]+://#) {
1537 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1538 fatal("E: '$path' is not a complete URL ",
1539 "and a separate URL is not specified");
1540 }
1541 return ($url, $path);
1542 }
1543 return ($path, '');
1544 }
1546 sub complete_url_ls_init {
1547 my ($ra, $repo_path, $switch, $pfx) = @_;
1548 unless ($repo_path) {
1549 print STDERR "W: $switch not specified\n";
1550 return;
1551 }
1552 $repo_path =~ s#/+$##;
1553 if ($repo_path =~ m#^[a-z\+]+://#) {
1554 $ra = Git::SVN::Ra->new($repo_path);
1555 $repo_path = '';
1556 } else {
1557 $repo_path =~ s#^/+##;
1558 unless ($ra) {
1559 fatal("E: '$repo_path' is not a complete URL ",
1560 "and a separate URL is not specified");
1561 }
1562 }
1563 my $url = $ra->{url};
1564 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1565 my $k = "svn-remote.$gs->{repo_id}.url";
1566 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1567 if ($orig_url && ($orig_url ne $gs->{url})) {
1568 die "$k already set: $orig_url\n",
1569 "wanted to set to: $gs->{url}\n";
1570 }
1571 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1572 my $remote_path = "$gs->{path}/$repo_path";
1573 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1574 $remote_path =~ s#/+#/#g;
1575 $remote_path =~ s#^/##g;
1576 $remote_path .= "/*" if $remote_path !~ /\*/;
1577 my ($n) = ($switch =~ /^--(\w+)/);
1578 if (length $pfx && $pfx !~ m#/$#) {
1579 die "--prefix='$pfx' must have a trailing slash '/'\n";
1580 }
1581 command_noisy('config',
1582 '--add',
1583 "svn-remote.$gs->{repo_id}.$n",
1584 "$remote_path:refs/remotes/$pfx*" .
1585 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1586 }
1588 sub verify_ref {
1589 my ($ref) = @_;
1590 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1591 { STDERR => 0 }); };
1592 }
1594 sub get_tree_from_treeish {
1595 my ($treeish) = @_;
1596 # $treeish can be a symbolic ref, too:
1597 my $type = command_oneline(qw/cat-file -t/, $treeish);
1598 my $expected;
1599 while ($type eq 'tag') {
1600 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1601 }
1602 if ($type eq 'commit') {
1603 $expected = (grep /^tree /, command(qw/cat-file commit/,
1604 $treeish))[0];
1605 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1606 die "Unable to get tree from $treeish\n" unless $expected;
1607 } elsif ($type eq 'tree') {
1608 $expected = $treeish;
1609 } else {
1610 die "$treeish is a $type, expected tree, tag or commit\n";
1611 }
1612 return $expected;
1613 }
1615 sub get_commit_entry {
1616 my ($treeish) = shift;
1617 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1618 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1619 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1620 open my $log_fh, '>', $commit_editmsg or croak $!;
1622 my $type = command_oneline(qw/cat-file -t/, $treeish);
1623 if ($type eq 'commit' || $type eq 'tag') {
1624 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1625 $type, $treeish);
1626 my $in_msg = 0;
1627 my $author;
1628 my $saw_from = 0;
1629 my $msgbuf = "";
1630 while (<$msg_fh>) {
1631 if (!$in_msg) {
1632 $in_msg = 1 if (/^\s*$/);
1633 $author = $1 if (/^author (.*>)/);
1634 } elsif (/^git-svn-id: /) {
1635 # skip this for now, we regenerate the
1636 # correct one on re-fetch anyways
1637 # TODO: set *:merge properties or like...
1638 } else {
1639 if (/^From:/ || /^Signed-off-by:/) {
1640 $saw_from = 1;
1641 }
1642 $msgbuf .= $_;
1643 }
1644 }
1645 $msgbuf =~ s/\s+$//s;
1646 if ($Git::SVN::_add_author_from && defined($author)
1647 && !$saw_from) {
1648 $msgbuf .= "\n\nFrom: $author";
1649 }
1650 print $log_fh $msgbuf or croak $!;
1651 command_close_pipe($msg_fh, $ctx);
1652 }
1653 close $log_fh or croak $!;
1655 if ($_edit || ($type eq 'tree')) {
1656 chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1657 system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1658 }
1659 rename $commit_editmsg, $commit_msg or croak $!;
1660 {
1661 require Encode;
1662 # SVN requires messages to be UTF-8 when entering the repo
1663 local $/;
1664 open $log_fh, '<', $commit_msg or croak $!;
1665 binmode $log_fh;
1666 chomp($log_entry{log} = <$log_fh>);
1668 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1669 my $msg = $log_entry{log};
1671 eval { $msg = Encode::decode($enc, $msg, 1) };
1672 if ($@) {
1673 die "Could not decode as $enc:\n", $msg,
1674 "\nPerhaps you need to set i18n.commitencoding\n";
1675 }
1677 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1678 die "Could not encode as UTF-8:\n$msg\n" if $@;
1680 $log_entry{log} = $msg;
1682 close $log_fh or croak $!;
1683 }
1684 unlink $commit_msg;
1685 \%log_entry;
1686 }
1688 sub s_to_file {
1689 my ($str, $file, $mode) = @_;
1690 open my $fd,'>',$file or croak $!;
1691 print $fd $str,"\n" or croak $!;
1692 close $fd or croak $!;
1693 chmod ($mode &~ umask, $file) if (defined $mode);
1694 }
1696 sub file_to_s {
1697 my $file = shift;
1698 open my $fd,'<',$file or croak "$!: file: $file\n";
1699 local $/;
1700 my $ret = <$fd>;
1701 close $fd or croak $!;
1702 $ret =~ s/\s*$//s;
1703 return $ret;
1704 }
1706 # '<svn username> = real-name <email address>' mapping based on git-svnimport:
1707 sub load_authors {
1708 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1709 my $log = $cmd eq 'log';
1710 while (<$authors>) {
1711 chomp;
1712 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1713 my ($user, $name, $email) = ($1, $2, $3);
1714 if ($log) {
1715 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1716 } else {
1717 $users{$user} = [$name, $email];
1718 }
1719 }
1720 close $authors or croak $!;
1721 }
1723 # convert GetOpt::Long specs for use by git-config
1724 sub read_git_config {
1725 my $opts = shift;
1726 my @config_only;
1727 foreach my $o (keys %$opts) {
1728 # if we have mixedCase and a long option-only, then
1729 # it's a config-only variable that we don't need for
1730 # the command-line.
1731 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1732 my $v = $opts->{$o};
1733 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1734 $key =~ s/-//g;
1735 my $arg = 'git config';
1736 $arg .= ' --int' if ($o =~ /[:=]i$/);
1737 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1738 if (ref $v eq 'ARRAY') {
1739 chomp(my @tmp = `$arg --get-all svn.$key`);
1740 @$v = @tmp if @tmp;
1741 } else {
1742 chomp(my $tmp = `$arg --get svn.$key`);
1743 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1744 $$v = $tmp;
1745 }
1746 }
1747 }
1748 delete @$opts{@config_only} if @config_only;
1749 }
1751 sub extract_metadata {
1752 my $id = shift or return (undef, undef, undef);
1753 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1754 \s([a-f\d\-]+)$/ix);
1755 if (!defined $rev || !$uuid || !$url) {
1756 # some of the original repositories I made had
1757 # identifiers like this:
1758 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1759 }
1760 return ($url, $rev, $uuid);
1761 }
1763 sub cmt_metadata {
1764 return extract_metadata((grep(/^git-svn-id: /,
1765 command(qw/cat-file commit/, shift)))[-1]);
1766 }
1768 sub cmt_sha2rev_batch {
1769 my %s2r;
1770 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1771 my $list = shift;
1773 foreach my $sha (@{$list}) {
1774 my $first = 1;
1775 my $size = 0;
1776 print $out $sha, "\n";
1778 while (my $line = <$in>) {
1779 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1780 last;
1781 } elsif ($first &&
1782 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1783 $first = 0;
1784 $size = $1;
1785 next;
1786 } elsif ($line =~ /^(git-svn-id: )/) {
1787 my (undef, $rev, undef) =
1788 extract_metadata($line);
1789 $s2r{$sha} = $rev;
1790 }
1792 $size -= length($line);
1793 last if ($size == 0);
1794 }
1795 }
1797 command_close_bidi_pipe($pid, $in, $out, $ctx);
1799 return \%s2r;
1800 }
1802 sub working_head_info {
1803 my ($head, $refs) = @_;
1804 my @args = qw/log --no-color --no-decorate --first-parent
1805 --pretty=medium/;
1806 my ($fh, $ctx) = command_output_pipe(@args, $head);
1807 my $hash;
1808 my %max;
1809 while (<$fh>) {
1810 if ( m{^commit ($::sha1)$} ) {
1811 unshift @$refs, $hash if $hash and $refs;
1812 $hash = $1;
1813 next;
1814 }
1815 next unless s{^\s*(git-svn-id:)}{$1};
1816 my ($url, $rev, $uuid) = extract_metadata($_);
1817 if (defined $url && defined $rev) {
1818 next if $max{$url} and $max{$url} < $rev;
1819 if (my $gs = Git::SVN->find_by_url($url)) {
1820 my $c = $gs->rev_map_get($rev, $uuid);
1821 if ($c && $c eq $hash) {
1822 close $fh; # break the pipe
1823 return ($url, $rev, $uuid, $gs);
1824 } else {
1825 $max{$url} ||= $gs->rev_map_max;
1826 }
1827 }
1828 }
1829 }
1830 command_close_pipe($fh, $ctx);
1831 (undef, undef, undef, undef);
1832 }
1834 sub read_commit_parents {
1835 my ($parents, $c) = @_;
1836 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1837 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1838 @{$parents->{$c}} = split(/ /, $p);
1839 }
1841 sub linearize_history {
1842 my ($gs, $refs) = @_;
1843 my %parents;
1844 foreach my $c (@$refs) {
1845 read_commit_parents(\%parents, $c);
1846 }
1848 my @linear_refs;
1849 my %skip = ();
1850 my $last_svn_commit = $gs->last_commit;
1851 foreach my $c (reverse @$refs) {
1852 next if $c eq $last_svn_commit;
1853 last if $skip{$c};
1855 unshift @linear_refs, $c;
1856 $skip{$c} = 1;
1858 # we only want the first parent to diff against for linear
1859 # history, we save the rest to inject when we finalize the
1860 # svn commit
1861 my $fp_a = verify_ref("$c~1");
1862 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1863 if (!$fp_a || !$fp_b) {
1864 die "Commit $c\n",
1865 "has no parent commit, and therefore ",
1866 "nothing to diff against.\n",
1867 "You should be working from a repository ",
1868 "originally created by git-svn\n";
1869 }
1870 if ($fp_a ne $fp_b) {
1871 die "$c~1 = $fp_a, however parsing commit $c ",
1872 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1873 }
1875 foreach my $p (@{$parents{$c}}) {
1876 $skip{$p} = 1;
1877 }
1878 }
1879 (\@linear_refs, \%parents);
1880 }
1882 sub find_file_type_and_diff_status {
1883 my ($path) = @_;
1884 return ('dir', '') if $path eq '';
1886 my $diff_output =
1887 command_oneline(qw(diff --cached --name-status --), $path) || "";
1888 my $diff_status = (split(' ', $diff_output))[0] || "";
1890 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1892 return (undef, undef) if !$diff_status && !$ls_tree;
1894 if ($diff_status eq "A") {
1895 return ("link", $diff_status) if -l $path;
1896 return ("dir", $diff_status) if -d $path;
1897 return ("file", $diff_status);
1898 }
1900 my $mode = (split(' ', $ls_tree))[0] || "";
1902 return ("link", $diff_status) if $mode eq "120000";
1903 return ("dir", $diff_status) if $mode eq "040000";
1904 return ("file", $diff_status);
1905 }
1907 sub md5sum {
1908 my $arg = shift;
1909 my $ref = ref $arg;
1910 my $md5 = Digest::MD5->new();
1911 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1912 $md5->addfile($arg) or croak $!;
1913 } elsif ($ref eq 'SCALAR') {
1914 $md5->add($$arg) or croak $!;
1915 } elsif (!$ref) {
1916 $md5->add($arg) or croak $!;
1917 } else {
1918 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1919 }
1920 return $md5->hexdigest();
1921 }
1923 sub gc_directory {
1924 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1925 my $out_filename = $_ . ".gz";
1926 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1927 binmode $in_fh;
1928 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1929 die "Unable to open $out_filename: $!\n";
1931 my $res;
1932 while ($res = sysread($in_fh, my $str, 1024)) {
1933 $gz->gzwrite($str) or
1934 die "Unable to write: ".$gz->gzerror()."!\n";
1935 }
1936 unlink $_ or die "unlink $File::Find::name: $!\n";
1937 } elsif (-f $_ && basename($_) eq "index") {
1938 unlink $_ or die "unlink $_: $!\n";
1939 }
1940 }
1942 package Git::SVN;
1943 use strict;
1944 use warnings;
1945 use Fcntl qw/:DEFAULT :seek/;
1946 use constant rev_map_fmt => 'NH40';
1947 use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1948 $_repack $_repack_flags $_use_svm_props $_head
1949 $_use_svnsync_props $no_reuse_existing $_minimize_url
1950 $_use_log_author $_add_author_from $_localtime/;
1951 use Carp qw/croak/;
1952 use File::Path qw/mkpath/;
1953 use File::Copy qw/copy/;
1954 use IPC::Open3;
1955 use Memoize; # core since 5.8.0, Jul 2002
1956 use Memoize::Storable;
1958 my ($_gc_nr, $_gc_period);
1960 # properties that we do not log:
1961 my %SKIP_PROP;
1962 BEGIN {
1963 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1964 svn:special svn:executable
1965 svn:entry:committed-rev
1966 svn:entry:last-author
1967 svn:entry:uuid
1968 svn:entry:committed-date/;
1970 # some options are read globally, but can be overridden locally
1971 # per [svn-remote "..."] section. Command-line options will *NOT*
1972 # override options set in an [svn-remote "..."] section
1973 no strict 'refs';
1974 for my $option (qw/follow_parent no_metadata use_svm_props
1975 use_svnsync_props/) {
1976 my $key = $option;
1977 $key =~ tr/_//d;
1978 my $prop = "-$option";
1979 *$option = sub {
1980 my ($self) = @_;
1981 return $self->{$prop} if exists $self->{$prop};
1982 my $k = "svn-remote.$self->{repo_id}.$key";
1983 eval { command_oneline(qw/config --get/, $k) };
1984 if ($@) {
1985 $self->{$prop} = ${"Git::SVN::_$option"};
1986 } else {
1987 my $v = command_oneline(qw/config --bool/,$k);
1988 $self->{$prop} = $v eq 'false' ? 0 : 1;
1989 }
1990 return $self->{$prop};
1991 }
1992 }
1993 }
1996 my (%LOCKFILES, %INDEX_FILES);
1997 END {
1998 unlink keys %LOCKFILES if %LOCKFILES;
1999 unlink keys %INDEX_FILES if %INDEX_FILES;
2000 }
2002 sub resolve_local_globs {
2003 my ($url, $fetch, $glob_spec) = @_;
2004 return unless defined $glob_spec;
2005 my $ref = $glob_spec->{ref};
2006 my $path = $glob_spec->{path};
2007 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
2008 next unless m#^$ref->{regex}$#;
2009 my $p = $1;
2010 my $pathname = desanitize_refname($path->full_path($p));
2011 my $refname = desanitize_refname($ref->full_path($p));
2012 if (my $existing = $fetch->{$pathname}) {
2013 if ($existing ne $refname) {
2014 die "Refspec conflict:\n",
2015 "existing: $existing\n",
2016 " globbed: $refname\n";
2017 }
2018 my $u = (::cmt_metadata("$refname"))[0];
2019 $u =~ s!^\Q$url\E(/|$)!! or die
2020 "$refname: '$url' not found in '$u'\n";
2021 if ($pathname ne $u) {
2022 warn "W: Refspec glob conflict ",
2023 "(ref: $refname):\n",
2024 "expected path: $pathname\n",
2025 " real path: $u\n",
2026 "Continuing ahead with $u\n";
2027 next;
2028 }
2029 } else {
2030 $fetch->{$pathname} = $refname;
2031 }
2032 }
2033 }
2035 sub parse_revision_argument {
2036 my ($base, $head) = @_;
2037 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
2038 return ($base, $head);
2039 }
2040 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
2041 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
2042 return ($head, $head) if ($::_revision eq 'HEAD');
2043 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
2044 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
2045 die "revision argument: $::_revision not understood by git-svn\n";
2046 }
2048 sub fetch_all {
2049 my ($repo_id, $remotes) = @_;
2050 if (ref $repo_id) {
2051 my $gs = $repo_id;
2052 $repo_id = undef;
2053 $repo_id = $gs->{repo_id};
2054 }
2055 $remotes ||= read_all_remotes();
2056 my $remote = $remotes->{$repo_id} or
2057 die "[svn-remote \"$repo_id\"] unknown\n";
2058 my $fetch = $remote->{fetch};
2059 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
2060 my (@gs, @globs);
2061 my $ra = Git::SVN::Ra->new($url);
2062 my $uuid = $ra->get_uuid;
2063 my $head = $ra->get_latest_revnum;
2065 # ignore errors, $head revision may not even exist anymore
2066 eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
2067 warn "W: $@\n" if $@;
2069 my $base = defined $fetch ? $head : 0;
2071 # read the max revs for wildcard expansion (branches/*, tags/*)
2072 foreach my $t (qw/branches tags/) {
2073 defined $remote->{$t} or next;
2074 push @globs, @{$remote->{$t}};
2076 my $max_rev = eval { tmp_config(qw/--int --get/,
2077 "svn-remote.$repo_id.${t}-maxRev") };
2078 if (defined $max_rev && ($max_rev < $base)) {
2079 $base = $max_rev;
2080 } elsif (!defined $max_rev) {
2081 $base = 0;
2082 }
2083 }
2085 if ($fetch) {
2086 foreach my $p (sort keys %$fetch) {
2087 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
2088 my $lr = $gs->rev_map_max;
2089 if (defined $lr) {
2090 $base = $lr if ($lr < $base);
2091 }
2092 push @gs, $gs;
2093 }
2094 }
2096 ($base, $head) = parse_revision_argument($base, $head);
2097 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
2098 }
2100 sub read_all_remotes {
2101 my $r = {};
2102 my $use_svm_props = eval { command_oneline(qw/config --bool
2103 svn.useSvmProps/) };
2104 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
2105 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
2106 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
2107 if (m!^(.+)\.fetch=$svn_refspec$!) {
2108 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
2109 die("svn-remote.$remote: remote ref '$remote_ref' "
2110 . "must start with 'refs/'\n")
2111 unless $remote_ref =~ m{^refs/};
2112 $local_ref = uri_decode($local_ref);
2113 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
2114 $r->{$remote}->{svm} = {} if $use_svm_props;
2115 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
2116 $r->{$1}->{svm} = {};
2117 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
2118 $r->{$1}->{url} = $2;
2119 } elsif (m!^(.+)\.pushurl=\s*(.*)\s*$!) {
2120 $r->{$1}->{pushurl} = $2;
2121 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
2122 my ($remote, $t, $local_ref, $remote_ref) =
2123 ($1, $2, $3, $4);
2124 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
2125 . "must start with 'refs/'\n")
2126 unless $remote_ref =~ m{^refs/};
2127 $local_ref = uri_decode($local_ref);
2128 my $rs = {
2129 t => $t,
2130 remote => $remote,
2131 path => Git::SVN::GlobSpec->new($local_ref, 1),
2132 ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
2133 if (length($rs->{ref}->{right}) != 0) {
2134 die "The '*' glob character must be the last ",
2135 "character of '$remote_ref'\n";
2136 }
2137 push @{ $r->{$remote}->{$t} }, $rs;
2138 }
2139 }
2141 map {
2142 if (defined $r->{$_}->{svm}) {
2143 my $svm;
2144 eval {
2145 my $section = "svn-remote.$_";
2146 $svm = {
2147 source => tmp_config('--get',
2148 "$section.svm-source"),
2149 replace => tmp_config('--get',
2150 "$section.svm-replace"),
2151 }
2152 };
2153 $r->{$_}->{svm} = $svm;
2154 }
2155 } keys %$r;
2157 $r;
2158 }
2160 sub init_vars {
2161 $_gc_nr = $_gc_period = 1000;
2162 if (defined $_repack || defined $_repack_flags) {
2163 warn "Repack options are obsolete; they have no effect.\n";
2164 }
2165 }
2167 sub verify_remotes_sanity {
2168 return unless -d $ENV{GIT_DIR};
2169 my %seen;
2170 foreach (command(qw/config -l/)) {
2171 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
2172 if ($seen{$1}) {
2173 die "Remote ref refs/remote/$1 is tracked by",
2174 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
2175 "Please resolve this ambiguity in ",
2176 "your git configuration file before ",
2177 "continuing\n";
2178 }
2179 $seen{$1} = $_;
2180 }
2181 }
2182 }
2184 sub find_existing_remote {
2185 my ($url, $remotes) = @_;
2186 return undef if $no_reuse_existing;
2187 my $existing;
2188 foreach my $repo_id (keys %$remotes) {
2189 my $u = $remotes->{$repo_id}->{url} or next;
2190 next if $u ne $url;
2191 $existing = $repo_id;
2192 last;
2193 }
2194 $existing;
2195 }
2197 sub init_remote_config {
2198 my ($self, $url, $no_write) = @_;
2199 $url =~ s!/+$!!; # strip trailing slash
2200 my $r = read_all_remotes();
2201 my $existing = find_existing_remote($url, $r);
2202 if ($existing) {
2203 unless ($no_write) {
2204 print STDERR "Using existing ",
2205 "[svn-remote \"$existing\"]\n";
2206 }
2207 $self->{repo_id} = $existing;
2208 } elsif ($_minimize_url) {
2209 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
2210 $existing = find_existing_remote($min_url, $r);
2211 if ($existing) {
2212 unless ($no_write) {
2213 print STDERR "Using existing ",
2214 "[svn-remote \"$existing\"]\n";
2215 }
2216 $self->{repo_id} = $existing;
2217 }
2218 if ($min_url ne $url) {
2219 unless ($no_write) {
2220 print STDERR "Using higher level of URL: ",
2221 "$url => $min_url\n";
2222 }
2223 my $old_path = $self->{path};
2224 $self->{path} = $url;
2225 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
2226 if (length $old_path) {
2227 $self->{path} .= "/$old_path";
2228 }
2229 $url = $min_url;
2230 }
2231 }
2232 my $orig_url;
2233 if (!$existing) {
2234 # verify that we aren't overwriting anything:
2235 $orig_url = eval {
2236 command_oneline('config', '--get',
2237 "svn-remote.$self->{repo_id}.url")
2238 };
2239 if ($orig_url && ($orig_url ne $url)) {
2240 die "svn-remote.$self->{repo_id}.url already set: ",
2241 "$orig_url\nwanted to set to: $url\n";
2242 }
2243 }
2244 my ($xrepo_id, $xpath) = find_ref($self->refname);
2245 if (!$no_write && defined $xpath) {
2246 die "svn-remote.$xrepo_id.fetch already set to track ",
2247 "$xpath:", $self->refname, "\n";
2248 }
2249 unless ($no_write) {
2250 command_noisy('config',
2251 "svn-remote.$self->{repo_id}.url", $url);
2252 $self->{path} =~ s{^/}{};
2253 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
2254 command_noisy('config', '--add',
2255 "svn-remote.$self->{repo_id}.fetch",
2256 "$self->{path}:".$self->refname);
2257 }
2258 $self->{url} = $url;
2259 }
2261 sub find_by_url { # repos_root and, path are optional
2262 my ($class, $full_url, $repos_root, $path) = @_;
2264 return undef unless defined $full_url;
2265 remove_username($full_url);
2266 remove_username($repos_root) if defined $repos_root;
2267 my $remotes = read_all_remotes();
2268 if (defined $full_url && defined $repos_root && !defined $path) {
2269 $path = $full_url;
2270 $path =~ s#^\Q$repos_root\E(?:/|$)##;
2271 }
2272 foreach my $repo_id (keys %$remotes) {
2273 my $u = $remotes->{$repo_id}->{url} or next;
2274 remove_username($u);
2275 next if defined $repos_root && $repos_root ne $u;
2277 my $fetch = $remotes->{$repo_id}->{fetch} || {};
2278 foreach my $t (qw/branches tags/) {
2279 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
2280 resolve_local_globs($u, $fetch, $globspec);
2281 }
2282 }
2283 my $p = $path;
2284 my $rwr = rewrite_root({repo_id => $repo_id});
2285 my $svm = $remotes->{$repo_id}->{svm}
2286 if defined $remotes->{$repo_id}->{svm};
2287 unless (defined $p) {
2288 $p = $full_url;
2289 my $z = $u;
2290 my $prefix = '';
2291 if ($rwr) {
2292 $z = $rwr;
2293 remove_username($z);
2294 } elsif (defined $svm) {
2295 $z = $svm->{source};
2296 $prefix = $svm->{replace};
2297 $prefix =~ s#^\Q$u\E(?:/|$)##;
2298 $prefix =~ s#/$##;
2299 }
2300 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2301 }
2302 foreach my $f (keys %$fetch) {
2303 next if $f ne $p;
2304 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2305 }
2306 }
2307 undef;
2308 }
2310 sub init {
2311 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2312 my $self = _new($class, $repo_id, $ref_id, $path);
2313 if (defined $url) {
2314 $self->init_remote_config($url, $no_write);
2315 }
2316 $self;
2317 }
2319 sub find_ref {
2320 my ($ref_id) = @_;
2321 foreach (command(qw/config -l/)) {
2322 next unless m!^svn-remote\.(.+)\.fetch=
2323 \s*(.*?)\s*:\s*(.+?)\s*$!x;
2324 my ($repo_id, $path, $ref) = ($1, $2, $3);
2325 if ($ref eq $ref_id) {
2326 $path = '' if ($path =~ m#^\./?#);
2327 return ($repo_id, $path);
2328 }
2329 }
2330 (undef, undef, undef);
2331 }
2333 sub new {
2334 my ($class, $ref_id, $repo_id, $path) = @_;
2335 if (defined $ref_id && !defined $repo_id && !defined $path) {
2336 ($repo_id, $path) = find_ref($ref_id);
2337 if (!defined $repo_id) {
2338 die "Could not find a \"svn-remote.*.fetch\" key ",
2339 "in the repository configuration matching: ",
2340 "$ref_id\n";
2341 }
2342 }
2343 my $self = _new($class, $repo_id, $ref_id, $path);
2344 if (!defined $self->{path} || !length $self->{path}) {
2345 my $fetch = command_oneline('config', '--get',
2346 "svn-remote.$repo_id.fetch",
2347 ":$ref_id\$") or
2348 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2349 "\":$ref_id\$\" in config\n";
2350 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2351 }
2352 $self->{path} =~ s{/+}{/}g;
2353 $self->{path} =~ s{\A/}{};
2354 $self->{path} =~ s{/\z}{};
2355 $self->{url} = command_oneline('config', '--get',
2356 "svn-remote.$repo_id.url") or
2357 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2358 $self->{pushurl} = eval { command_oneline('config', '--get',
2359 "svn-remote.$repo_id.pushurl") };
2360 $self->rebuild;
2361 $self;
2362 }
2364 sub refname {
2365 my ($refname) = $_[0]->{ref_id} ;
2367 # It cannot end with a slash /, we'll throw up on this because
2368 # SVN can't have directories with a slash in their name, either:
2369 if ($refname =~ m{/$}) {
2370 die "ref: '$refname' ends with a trailing slash, this is ",
2371 "not permitted by git nor Subversion\n";
2372 }
2374 # It cannot have ASCII control character space, tilde ~, caret ^,
2375 # colon :, question-mark ?, asterisk *, space, or open bracket [
2376 # anywhere.
2377 #
2378 # Additionally, % must be escaped because it is used for escaping
2379 # and we want our escaped refname to be reversible
2380 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2382 # no slash-separated component can begin with a dot .
2383 # /.* becomes /%2E*
2384 $refname =~ s{/\.}{/%2E}g;
2386 # It cannot have two consecutive dots .. anywhere
2387 # .. becomes %2E%2E
2388 $refname =~ s{\.\.}{%2E%2E}g;
2390 # trailing dots and .lock are not allowed
2391 # .$ becomes %2E and .lock becomes %2Elock
2392 $refname =~ s{\.(?=$|lock$)}{%2E};
2394 # the sequence @{ is used to access the reflog
2395 # @{ becomes %40{
2396 $refname =~ s{\@\{}{%40\{}g;
2398 return $refname;
2399 }
2401 sub desanitize_refname {
2402 my ($refname) = @_;
2403 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2404 return $refname;
2405 }
2407 sub svm_uuid {
2408 my ($self) = @_;
2409 return $self->{svm}->{uuid} if $self->svm;
2410 $self->ra;
2411 unless ($self->{svm}) {
2412 die "SVM UUID not cached, and reading remotely failed\n";
2413 }
2414 $self->{svm}->{uuid};
2415 }
2417 sub svm {
2418 my ($self) = @_;
2419 return $self->{svm} if $self->{svm};
2420 my $svm;
2421 # see if we have it in our config, first:
2422 eval {
2423 my $section = "svn-remote.$self->{repo_id}";
2424 $svm = {
2425 source => tmp_config('--get', "$section.svm-source"),
2426 uuid => tmp_config('--get', "$section.svm-uuid"),
2427 replace => tmp_config('--get', "$section.svm-replace"),
2428 }
2429 };
2430 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2431 $self->{svm} = $svm;
2432 }
2433 $self->{svm};
2434 }
2436 sub _set_svm_vars {
2437 my ($self, $ra) = @_;
2438 return $ra if $self->svm;
2440 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2441 "(svm:source, svm:uuid) ",
2442 "from the following URLs:\n" );
2443 sub read_svm_props {
2444 my ($self, $ra, $path, $r) = @_;
2445 my $props = ($ra->get_dir($path, $r))[2];
2446 my $src = $props->{'svm:source'};
2447 my $uuid = $props->{'svm:uuid'};
2448 return undef if (!$src || !$uuid);
2450 chomp($src, $uuid);
2452 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2453 or die "doesn't look right - svm:uuid is '$uuid'\n";
2455 # the '!' is used to mark the repos_root!/relative/path
2456 $src =~ s{/?!/?}{/};
2457 $src =~ s{/+$}{}; # no trailing slashes please
2458 # username is of no interest
2459 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2461 my $replace = $ra->{url};
2462 $replace .= "/$path" if length $path;
2464 my $section = "svn-remote.$self->{repo_id}";
2465 tmp_config("$section.svm-source", $src);
2466 tmp_config("$section.svm-replace", $replace);
2467 tmp_config("$section.svm-uuid", $uuid);
2468 $self->{svm} = {
2469 source => $src,
2470 uuid => $uuid,
2471 replace => $replace
2472 };
2473 }
2475 my $r = $ra->get_latest_revnum;
2476 my $path = $self->{path};
2477 my %tried;
2478 while (length $path) {
2479 unless ($tried{"$self->{url}/$path"}) {
2480 return $ra if $self->read_svm_props($ra, $path, $r);
2481 $tried{"$self->{url}/$path"} = 1;
2482 }
2483 $path =~ s#/?[^/]+$##;
2484 }
2485 die "Path: '$path' should be ''\n" if $path ne '';
2486 return $ra if $self->read_svm_props($ra, $path, $r);
2487 $tried{"$self->{url}/$path"} = 1;
2489 if ($ra->{repos_root} eq $self->{url}) {
2490 die @err, (map { " $_\n" } keys %tried), "\n";
2491 }
2493 # nope, make sure we're connected to the repository root:
2494 my $ok;
2495 my @tried_b;
2496 $path = $ra->{svn_path};
2497 $ra = Git::SVN::Ra->new($ra->{repos_root});
2498 while (length $path) {
2499 unless ($tried{"$ra->{url}/$path"}) {
2500 $ok = $self->read_svm_props($ra, $path, $r);
2501 last if $ok;
2502 $tried{"$ra->{url}/$path"} = 1;
2503 }
2504 $path =~ s#/?[^/]+$##;
2505 }
2506 die "Path: '$path' should be ''\n" if $path ne '';
2507 $ok ||= $self->read_svm_props($ra, $path, $r);
2508 $tried{"$ra->{url}/$path"} = 1;
2509 if (!$ok) {
2510 die @err, (map { " $_\n" } keys %tried), "\n";
2511 }
2512 Git::SVN::Ra->new($self->{url});
2513 }
2515 sub svnsync {
2516 my ($self) = @_;
2517 return $self->{svnsync} if $self->{svnsync};
2519 if ($self->no_metadata) {
2520 die "Can't have both 'noMetadata' and ",
2521 "'useSvnsyncProps' options set!\n";
2522 }
2523 if ($self->rewrite_root) {
2524 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2525 "options set!\n";
2526 }
2527 if ($self->rewrite_uuid) {
2528 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2529 "options set!\n";
2530 }
2532 my $svnsync;
2533 # see if we have it in our config, first:
2534 eval {
2535 my $section = "svn-remote.$self->{repo_id}";
2537 my $url = tmp_config('--get', "$section.svnsync-url");
2538 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2539 die "doesn't look right - svn:sync-from-url is '$url'\n";
2541 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2542 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2543 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2545 $svnsync = { url => $url, uuid => $uuid }
2546 };
2547 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2548 return $self->{svnsync} = $svnsync;
2549 }
2551 my $err = "useSvnsyncProps set, but failed to read " .
2552 "svnsync property: svn:sync-from-";
2553 my $rp = $self->ra->rev_proplist(0);
2555 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2556 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2557 die "doesn't look right - svn:sync-from-url is '$url'\n";
2559 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2560 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2561 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2563 my $section = "svn-remote.$self->{repo_id}";
2564 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2565 tmp_config('--add', "$section.svnsync-url", $url);
2566 return $self->{svnsync} = { url => $url, uuid => $uuid };
2567 }
2569 # this allows us to memoize our SVN::Ra UUID locally and avoid a
2570 # remote lookup (useful for 'git svn log').
2571 sub ra_uuid {
2572 my ($self) = @_;
2573 unless ($self->{ra_uuid}) {
2574 my $key = "svn-remote.$self->{repo_id}.uuid";
2575 my $uuid = eval { tmp_config('--get', $key) };
2576 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2577 $self->{ra_uuid} = $uuid;
2578 } else {
2579 die "ra_uuid called without URL\n" unless $self->{url};
2580 $self->{ra_uuid} = $self->ra->get_uuid;
2581 tmp_config('--add', $key, $self->{ra_uuid});
2582 }
2583 }
2584 $self->{ra_uuid};
2585 }
2587 sub _set_repos_root {
2588 my ($self, $repos_root) = @_;
2589 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2590 $repos_root ||= $self->ra->{repos_root};
2591 tmp_config($k, $repos_root);
2592 $repos_root;
2593 }
2595 sub repos_root {
2596 my ($self) = @_;
2597 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2598 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2599 }
2601 sub ra {
2602 my ($self) = shift;
2603 my $ra = Git::SVN::Ra->new($self->{url});
2604 $self->_set_repos_root($ra->{repos_root});
2605 if ($self->use_svm_props && !$self->{svm}) {
2606 if ($self->no_metadata) {
2607 die "Can't have both 'noMetadata' and ",
2608 "'useSvmProps' options set!\n";
2609 } elsif ($self->use_svnsync_props) {
2610 die "Can't have both 'useSvnsyncProps' and ",
2611 "'useSvmProps' options set!\n";
2612 }
2613 $ra = $self->_set_svm_vars($ra);
2614 $self->{-want_revprops} = 1;
2615 }
2616 $ra;
2617 }
2619 # prop_walk(PATH, REV, SUB)
2620 # -------------------------
2621 # Recursively traverse PATH at revision REV and invoke SUB for each
2622 # directory that contains a SVN property. SUB will be invoked as
2623 # follows: &SUB(gs, path, props); where `gs' is this instance of
2624 # Git::SVN, `path' the path to the directory where the properties
2625 # `props' were found. The `path' will be relative to point of checkout,
2626 # that is, if url://repo/trunk is the current Git branch, and that
2627 # directory contains a sub-directory `d', SUB will be invoked with `/d/'
2628 # as `path' (note the trailing `/').
2629 sub prop_walk {
2630 my ($self, $path, $rev, $sub) = @_;
2632 $path =~ s#^/##;
2633 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2634 $path =~ s#^/*#/#g;
2635 my $p = $path;
2636 # Strip the irrelevant part of the path.
2637 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2638 # Ensure the path is terminated by a `/'.
2639 $p =~ s#/*$#/#;
2641 # The properties contain all the internal SVN stuff nobody
2642 # (usually) cares about.
2643 my $interesting_props = 0;
2644 foreach (keys %{$props}) {
2645 # If it doesn't start with `svn:', it must be a
2646 # user-defined property.
2647 ++$interesting_props and next if $_ !~ /^svn:/;
2648 # FIXME: Fragile, if SVN adds new public properties,
2649 # this needs to be updated.
2650 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2651 |eol-style|mime-type
2652 |externals|needs-lock)$/x;
2653 }
2654 &$sub($self, $p, $props) if $interesting_props;
2656 foreach (sort keys %$dirent) {
2657 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2658 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2659 }
2660 }
2662 sub last_rev { ($_[0]->last_rev_commit)[0] }
2663 sub last_commit { ($_[0]->last_rev_commit)[1] }
2665 # returns the newest SVN revision number and newest commit SHA1
2666 sub last_rev_commit {
2667 my ($self) = @_;
2668 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2669 return ($self->{last_rev}, $self->{last_commit});
2670 }
2671 my $c = ::verify_ref($self->refname.'^0');
2672 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2673 my $rev = (::cmt_metadata($c))[1];
2674 if (defined $rev) {
2675 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2676 return ($rev, $c);
2677 }
2678 }
2679 my $map_path = $self->map_path;
2680 unless (-e $map_path) {
2681 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2682 return (undef, undef);
2683 }
2684 my ($rev, $commit) = $self->rev_map_max(1);
2685 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2686 return ($rev, $commit);
2687 }
2689 sub get_fetch_range {
2690 my ($self, $min, $max) = @_;
2691 $max ||= $self->ra->get_latest_revnum;
2692 $min ||= $self->rev_map_max;
2693 (++$min, $max);
2694 }
2696 sub tmp_config {
2697 my (@args) = @_;
2698 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2699 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2700 if (! -f $config && -f $old_def_config) {
2701 rename $old_def_config, $config or
2702 die "Failed rename $old_def_config => $config: $!\n";
2703 }
2704 my $old_config = $ENV{GIT_CONFIG};
2705 $ENV{GIT_CONFIG} = $config;
2706 $@ = undef;
2707 my @ret = eval {
2708 unless (-f $config) {
2709 mkfile($config);
2710 open my $fh, '>', $config or
2711 die "Can't open $config: $!\n";
2712 print $fh "; This file is used internally by ",
2713 "git-svn\n" or die
2714 "Couldn't write to $config: $!\n";
2715 print $fh "; You should not have to edit it\n" or
2716 die "Couldn't write to $config: $!\n";
2717 close $fh or die "Couldn't close $config: $!\n";
2718 }
2719 command('config', @args);
2720 };
2721 my $err = $@;
2722 if (defined $old_config) {
2723 $ENV{GIT_CONFIG} = $old_config;
2724 } else {
2725 delete $ENV{GIT_CONFIG};
2726 }
2727 die $err if $err;
2728 wantarray ? @ret : $ret[0];
2729 }
2731 sub tmp_index_do {
2732 my ($self, $sub) = @_;
2733 my $old_index = $ENV{GIT_INDEX_FILE};
2734 $ENV{GIT_INDEX_FILE} = $self->{index};
2735 $@ = undef;
2736 my @ret = eval {
2737 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2738 mkpath([$dir]) unless -d $dir;
2739 &$sub;
2740 };
2741 my $err = $@;
2742 if (defined $old_index) {
2743 $ENV{GIT_INDEX_FILE} = $old_index;
2744 } else {
2745 delete $ENV{GIT_INDEX_FILE};
2746 }
2747 die $err if $err;
2748 wantarray ? @ret : $ret[0];
2749 }
2751 sub assert_index_clean {
2752 my ($self, $treeish) = @_;
2754 $self->tmp_index_do(sub {
2755 command_noisy('read-tree', $treeish) unless -e $self->{index};
2756 my $x = command_oneline('write-tree');
2757 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2758 /^tree ($::sha1)/mo);
2759 return if $y eq $x;
2761 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2762 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2763 command_noisy('read-tree', $treeish);
2764 $x = command_oneline('write-tree');
2765 if ($y ne $x) {
2766 ::fatal "trees ($treeish) $y != $x\n",
2767 "Something is seriously wrong...";
2768 }
2769 });
2770 }
2772 sub get_commit_parents {
2773 my ($self, $log_entry) = @_;
2774 my (%seen, @ret, @tmp);
2775 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2776 if (my $ip = $self->{inject_parents}) {
2777 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2778 push @tmp, $commit;
2779 }
2780 }
2781 if (my $cur = ::verify_ref($self->refname.'^0')) {
2782 push @tmp, $cur;
2783 }
2784 if (my $ipd = $self->{inject_parents_dcommit}) {
2785 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2786 push @tmp, @$commit;
2787 }
2788 }
2789 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2790 while (my $p = shift @tmp) {
2791 next if $seen{$p};
2792 $seen{$p} = 1;
2793 push @ret, $p;
2794 }
2795 @ret;
2796 }
2798 sub rewrite_root {
2799 my ($self) = @_;
2800 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2801 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2802 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2803 if ($rwr) {
2804 $rwr =~ s#/+$##;
2805 if ($rwr !~ m#^[a-z\+]+://#) {
2806 die "$rwr is not a valid URL (key: $k)\n";
2807 }
2808 }
2809 $self->{-rewrite_root} = $rwr;
2810 }
2812 sub rewrite_uuid {
2813 my ($self) = @_;
2814 return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2815 my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2816 my $rwid = eval { command_oneline(qw/config --get/, $k) };
2817 if ($rwid) {
2818 $rwid =~ s#/+$##;
2819 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2820 die "$rwid is not a valid UUID (key: $k)\n";
2821 }
2822 }
2823 $self->{-rewrite_uuid} = $rwid;
2824 }
2826 sub metadata_url {
2827 my ($self) = @_;
2828 ($self->rewrite_root || $self->{url}) .
2829 (length $self->{path} ? '/' . $self->{path} : '');
2830 }
2832 sub full_url {
2833 my ($self) = @_;
2834 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2835 }
2837 sub full_pushurl {
2838 my ($self) = @_;
2839 if ($self->{pushurl}) {
2840 return $self->{pushurl} . (length $self->{path} ? '/' .
2841 $self->{path} : '');
2842 } else {
2843 return $self->full_url;
2844 }
2845 }
2847 sub set_commit_header_env {
2848 my ($log_entry) = @_;
2849 my %env;
2850 foreach my $ned (qw/NAME EMAIL DATE/) {
2851 foreach my $ac (qw/AUTHOR COMMITTER/) {
2852 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2853 }
2854 }
2856 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2857 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2858 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2860 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2861 ? $log_entry->{commit_name}
2862 : $log_entry->{name};
2863 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2864 ? $log_entry->{commit_email}
2865 : $log_entry->{email};
2866 \%env;
2867 }
2869 sub restore_commit_header_env {
2870 my ($env) = @_;
2871 foreach my $ned (qw/NAME EMAIL DATE/) {
2872 foreach my $ac (qw/AUTHOR COMMITTER/) {
2873 my $k = "GIT_${ac}_${ned}";
2874 if (defined $env->{$k}) {
2875 $ENV{$k} = $env->{$k};
2876 } else {
2877 delete $ENV{$k};
2878 }
2879 }
2880 }
2881 }
2883 sub gc {
2884 command_noisy('gc', '--auto');
2885 };
2887 sub do_git_commit {
2888 my ($self, $log_entry) = @_;
2889 my $lr = $self->last_rev;
2890 if (defined $lr && $lr >= $log_entry->{revision}) {
2891 die "Last fetched revision of ", $self->refname,
2892 " was r$lr, but we are about to fetch: ",
2893 "r$log_entry->{revision}!\n";
2894 }
2895 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2896 croak "$log_entry->{revision} = $c already exists! ",
2897 "Why are we refetching it?\n";
2898 }
2899 my $old_env = set_commit_header_env($log_entry);
2900 my $tree = $log_entry->{tree};
2901 if (!defined $tree) {
2902 $tree = $self->tmp_index_do(sub {
2903 command_oneline('write-tree') });
2904 }
2905 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2907 my @exec = ('git', 'commit-tree', $tree);
2908 foreach ($self->get_commit_parents($log_entry)) {
2909 push @exec, '-p', $_;
2910 }
2911 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2912 or croak $!;
2913 binmode $msg_fh;
2915 # we always get UTF-8 from SVN, but we may want our commits in
2916 # a different encoding.
2917 if (my $enc = Git::config('i18n.commitencoding')) {
2918 require Encode;
2919 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2920 }
2921 print $msg_fh $log_entry->{log} or croak $!;
2922 restore_commit_header_env($old_env);
2923 unless ($self->no_metadata) {
2924 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2925 or croak $!;
2926 }
2927 $msg_fh->flush == 0 or croak $!;
2928 close $msg_fh or croak $!;
2929 chomp(my $commit = do { local $/; <$out_fh> });
2930 close $out_fh or croak $!;
2931 waitpid $pid, 0;
2932 croak $? if $?;
2933 if ($commit !~ /^$::sha1$/o) {
2934 die "Failed to commit, invalid sha1: $commit\n";
2935 }
2937 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2939 $self->{last_rev} = $log_entry->{revision};
2940 $self->{last_commit} = $commit;
2941 print "r$log_entry->{revision}" unless $::_q > 1;
2942 if (defined $log_entry->{svm_revision}) {
2943 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2944 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2945 0, $self->svm_uuid);
2946 }
2947 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2948 if (--$_gc_nr == 0) {
2949 $_gc_nr = $_gc_period;
2950 gc();
2951 }
2952 return $commit;
2953 }
2955 sub match_paths {
2956 my ($self, $paths, $r) = @_;
2957 return 1 if $self->{path} eq '';
2958 if (my $path = $paths->{"/$self->{path}"}) {
2959 return ($path->{action} eq 'D') ? 0 : 1;
2960 }
2961 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2962 if (grep /$self->{path_regex}/, keys %$paths) {
2963 return 1;
2964 }
2965 my $c = '';
2966 foreach (split m#/#, $self->{path}) {
2967 $c .= "/$_";
2968 next unless ($paths->{$c} &&
2969 ($paths->{$c}->{action} =~ /^[AR]$/));
2970 if ($self->ra->check_path($self->{path}, $r) ==
2971 $SVN::Node::dir) {
2972 return 1;
2973 }
2974 }
2975 return 0;
2976 }
2978 sub find_parent_branch {
2979 my ($self, $paths, $rev) = @_;
2980 return undef unless $self->follow_parent;
2981 unless (defined $paths) {
2982 my $err_handler = $SVN::Error::handler;
2983 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2984 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2985 sub { $paths = $_[0] });
2986 $SVN::Error::handler = $err_handler;
2987 }
2988 return undef unless defined $paths;
2990 # look for a parent from another branch:
2991 my @b_path_components = split m#/#, $self->{path};
2992 my @a_path_components;
2993 my $i;
2994 while (@b_path_components) {
2995 $i = $paths->{'/'.join('/', @b_path_components)};
2996 last if $i && defined $i->{copyfrom_path};
2997 unshift(@a_path_components, pop(@b_path_components));
2998 }
2999 return undef unless defined $i && defined $i->{copyfrom_path};
3000 my $branch_from = $i->{copyfrom_path};
3001 if (@a_path_components) {
3002 print STDERR "branch_from: $branch_from => ";
3003 $branch_from .= '/'.join('/', @a_path_components);
3004 print STDERR $branch_from, "\n";
3005 }
3006 my $r = $i->{copyfrom_rev};
3007 my $repos_root = $self->ra->{repos_root};
3008 my $url = $self->ra->{url};
3009 my $new_url = $url . $branch_from;
3010 print STDERR "Found possible branch point: ",
3011 "$new_url => ", $self->full_url, ", $r\n"
3012 unless $::_q > 1;
3013 $branch_from =~ s#^/##;
3014 my $gs = $self->other_gs($new_url, $url,
3015 $branch_from, $r, $self->{ref_id});
3016 my ($r0, $parent) = $gs->find_rev_before($r, 1);
3017 {
3018 my ($base, $head);
3019 if (!defined $r0 || !defined $parent) {
3020 ($base, $head) = parse_revision_argument(0, $r);
3021 } else {
3022 if ($r0 < $r) {
3023 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
3024 0, 1, sub { $base = $_[1] - 1 });
3025 }
3026 }
3027 if (defined $base && $base <= $r) {
3028 $gs->fetch($base, $r);
3029 }
3030 ($r0, $parent) = $gs->find_rev_before($r, 1);
3031 }
3032 if (defined $r0 && defined $parent) {
3033 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
3034 unless $::_q > 1;
3035 my $ed;
3036 if ($self->ra->can_do_switch) {
3037 $self->assert_index_clean($parent);
3038 print STDERR "Following parent with do_switch\n"
3039 unless $::_q > 1;
3040 # do_switch works with svn/trunk >= r22312, but that
3041 # is not included with SVN 1.4.3 (the latest version
3042 # at the moment), so we can't rely on it
3043 $self->{last_rev} = $r0;
3044 $self->{last_commit} = $parent;
3045 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
3046 $gs->ra->gs_do_switch($r0, $rev, $gs,
3047 $self->full_url, $ed)
3048 or die "SVN connection failed somewhere...\n";
3049 } elsif ($self->ra->trees_match($new_url, $r0,
3050 $self->full_url, $rev)) {
3051 print STDERR "Trees match:\n",
3052 " $new_url\@$r0\n",
3053 " ${\$self->full_url}\@$rev\n",
3054 "Following parent with no changes\n"
3055 unless $::_q > 1;
3056 $self->tmp_index_do(sub {
3057 command_noisy('read-tree', $parent);
3058 });
3059 $self->{last_commit} = $parent;
3060 } else {
3061 print STDERR "Following parent with do_update\n"
3062 unless $::_q > 1;
3063 $ed = SVN::Git::Fetcher->new($self);
3064 $self->ra->gs_do_update($rev, $rev, $self, $ed)
3065 or die "SVN connection failed somewhere...\n";
3066 }
3067 print STDERR "Successfully followed parent\n" unless $::_q > 1;
3068 return $self->make_log_entry($rev, [$parent], $ed);
3069 }
3070 return undef;
3071 }
3073 sub do_fetch {
3074 my ($self, $paths, $rev) = @_;
3075 my $ed;
3076 my ($last_rev, @parents);
3077 if (my $lc = $self->last_commit) {
3078 # we can have a branch that was deleted, then re-added
3079 # under the same name but copied from another path, in
3080 # which case we'll have multiple parents (we don't
3081 # want to break the original ref, nor lose copypath info):
3082 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
3083 push @{$log_entry->{parents}}, $lc;
3084 return $log_entry;
3085 }
3086 $ed = SVN::Git::Fetcher->new($self);
3087 $last_rev = $self->{last_rev};
3088 $ed->{c} = $lc;
3089 @parents = ($lc);
3090 } else {
3091 $last_rev = $rev;
3092 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
3093 return $log_entry;
3094 }
3095 $ed = SVN::Git::Fetcher->new($self);
3096 }
3097 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
3098 die "SVN connection failed somewhere...\n";
3099 }
3100 $self->make_log_entry($rev, \@parents, $ed);
3101 }
3103 sub mkemptydirs {
3104 my ($self, $r) = @_;
3106 sub scan {
3107 my ($r, $empty_dirs, $line) = @_;
3108 if (defined $r && $line =~ /^r(\d+)$/) {
3109 return 0 if $1 > $r;
3110 } elsif ($line =~ /^ \+empty_dir: (.+)$/) {
3111 $empty_dirs->{$1} = 1;
3112 } elsif ($line =~ /^ \-empty_dir: (.+)$/) {
3113 my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
3114 delete @$empty_dirs{@d};
3115 }
3116 1; # continue
3117 };
3119 my %empty_dirs = ();
3120 my $gz_file = "$self->{dir}/unhandled.log.gz";
3121 if (-f $gz_file) {
3122 if (!$can_compress) {
3123 warn "Compress::Zlib could not be found; ",
3124 "empty directories in $gz_file will not be read\n";
3125 } else {
3126 my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
3127 die "Unable to open $gz_file: $!\n";
3128 my $line;
3129 while ($gz->gzreadline($line) > 0) {
3130 scan($r, \%empty_dirs, $line) or last;
3131 }
3132 $gz->gzclose;
3133 }
3134 }
3136 if (open my $fh, '<', "$self->{dir}/unhandled.log") {
3137 binmode $fh or croak "binmode: $!";
3138 while (<$fh>) {
3139 scan($r, \%empty_dirs, $_) or last;
3140 }
3141 close $fh;
3142 }
3144 my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
3145 foreach my $d (sort keys %empty_dirs) {
3146 $d = uri_decode($d);
3147 $d =~ s/$strip//;
3148 next unless length($d);
3149 next if -d $d;
3150 if (-e $d) {
3151 warn "$d exists but is not a directory\n";
3152 } else {
3153 print "creating empty directory: $d\n";
3154 mkpath([$d]);
3155 }
3156 }
3157 }
3159 sub get_untracked {
3160 my ($self, $ed) = @_;
3161 my @out;
3162 my $h = $ed->{empty};
3163 foreach (sort keys %$h) {
3164 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
3165 push @out, " $act: " . uri_encode($_);
3166 warn "W: $act: $_\n";
3167 }
3168 foreach my $t (qw/dir_prop file_prop/) {
3169 $h = $ed->{$t} or next;
3170 foreach my $path (sort keys %$h) {
3171 my $ppath = $path eq '' ? '.' : $path;
3172 foreach my $prop (sort keys %{$h->{$path}}) {
3173 next if $SKIP_PROP{$prop};
3174 my $v = $h->{$path}->{$prop};
3175 my $t_ppath_prop = "$t: " .
3176 uri_encode($ppath) . ' ' .
3177 uri_encode($prop);
3178 if (defined $v) {
3179 push @out, " +$t_ppath_prop " .
3180 uri_encode($v);
3181 } else {
3182 push @out, " -$t_ppath_prop";
3183 }
3184 }
3185 }
3186 }
3187 foreach my $t (qw/absent_file absent_directory/) {
3188 $h = $ed->{$t} or next;
3189 foreach my $parent (sort keys %$h) {
3190 foreach my $path (sort @{$h->{$parent}}) {
3191 push @out, " $t: " .
3192 uri_encode("$parent/$path");
3193 warn "W: $t: $parent/$path ",
3194 "Insufficient permissions?\n";
3195 }
3196 }
3197 }
3198 \@out;
3199 }
3201 # parse_svn_date(DATE)
3202 # --------------------
3203 # Given a date (in UTC) from Subversion, return a string in the format
3204 # "<TZ Offset> <local date/time>" that Git will use.
3205 #
3206 # By default the parsed date will be in UTC; if $Git::SVN::_localtime
3207 # is true we'll convert it to the local timezone instead.
3208 sub parse_svn_date {
3209 my $date = shift || return '+0000 1970-01-01 00:00:00';
3210 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
3211 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
3212 croak "Unable to parse date: $date\n";
3213 my $parsed_date; # Set next.
3215 if ($Git::SVN::_localtime) {
3216 # Translate the Subversion datetime to an epoch time.
3217 # Begin by switching ourselves to $date's timezone, UTC.
3218 my $old_env_TZ = $ENV{TZ};
3219 $ENV{TZ} = 'UTC';
3221 my $epoch_in_UTC =
3222 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
3224 # Determine our local timezone (including DST) at the
3225 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
3226 # value of TZ, if any, at the time we were run.
3227 if (defined $Git::SVN::Log::TZ) {
3228 $ENV{TZ} = $Git::SVN::Log::TZ;
3229 } else {
3230 delete $ENV{TZ};
3231 }
3233 my $our_TZ =
3234 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
3236 # This converts $epoch_in_UTC into our local timezone.
3237 my ($sec, $min, $hour, $mday, $mon, $year,
3238 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
3240 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
3241 $our_TZ, $year + 1900, $mon + 1,
3242 $mday, $hour, $min, $sec);
3244 # Reset us to the timezone in effect when we entered
3245 # this routine.
3246 if (defined $old_env_TZ) {
3247 $ENV{TZ} = $old_env_TZ;
3248 } else {
3249 delete $ENV{TZ};
3250 }
3251 } else {
3252 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
3253 }
3255 return $parsed_date;
3256 }
3258 sub other_gs {
3259 my ($self, $new_url, $url,
3260 $branch_from, $r, $old_ref_id) = @_;
3261 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
3262 unless ($gs) {
3263 my $ref_id = $old_ref_id;
3264 $ref_id =~ s/\@\d+-*$//;
3265 $ref_id .= "\@$r";
3266 # just grow a tail if we're not unique enough :x
3267 $ref_id .= '-' while find_ref($ref_id);
3268 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
3269 if ($u =~ s#^\Q$url\E(/|$)##) {
3270 $p = $u;
3271 $u = $url;
3272 $repo_id = $self->{repo_id};
3273 }
3274 while (1) {
3275 # It is possible to tag two different subdirectories at
3276 # the same revision. If the url for an existing ref
3277 # does not match, we must either find a ref with a
3278 # matching url or create a new ref by growing a tail.
3279 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
3280 my (undef, $max_commit) = $gs->rev_map_max(1);
3281 last if (!$max_commit);
3282 my ($url) = ::cmt_metadata($max_commit);
3283 last if ($url eq $gs->metadata_url);
3284 $ref_id .= '-';
3285 }
3286 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
3287 }
3288 $gs
3289 }
3291 sub call_authors_prog {
3292 my ($orig_author) = @_;
3293 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
3294 my $author = `$::_authors_prog $orig_author`;
3295 if ($? != 0) {
3296 die "$::_authors_prog failed with exit code $?\n"
3297 }
3298 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
3299 my ($name, $email) = ($1, $2);
3300 $email = undef if length $2 == 0;
3301 return [$name, $email];
3302 } else {
3303 die "Author: $orig_author: $::_authors_prog returned "
3304 . "invalid author format: $author\n";
3305 }
3306 }
3308 sub check_author {
3309 my ($author) = @_;
3310 if (!defined $author || length $author == 0) {
3311 $author = '(no author)';
3312 }
3313 if (!defined $::users{$author}) {
3314 if (defined $::_authors_prog) {
3315 $::users{$author} = call_authors_prog($author);
3316 } elsif (defined $::_authors) {
3317 die "Author: $author not defined in $::_authors file\n";
3318 }
3319 }
3320 $author;
3321 }
3323 sub find_extra_svk_parents {
3324 my ($self, $ed, $tickets, $parents) = @_;
3325 # aha! svk:merge property changed...
3326 my @tickets = split "\n", $tickets;
3327 my @known_parents;
3328 for my $ticket ( @tickets ) {
3329 my ($uuid, $path, $rev) = split /:/, $ticket;
3330 if ( $uuid eq $self->ra_uuid ) {
3331 my $url = $self->{url};
3332 my $repos_root = $url;
3333 my $branch_from = $path;
3334 $branch_from =~ s{^/}{};
3335 my $gs = $self->other_gs($repos_root."/".$branch_from,
3336 $url,
3337 $branch_from,
3338 $rev,
3339 $self->{ref_id});
3340 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3341 # wahey! we found it, but it might be
3342 # an old one (!)
3343 push @known_parents, [ $rev, $commit ];
3344 }
3345 }
3346 }
3347 # Ordering matters; highest-numbered commit merge tickets
3348 # first, as they may account for later merge ticket additions
3349 # or changes.
3350 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3351 for my $parent ( @known_parents ) {
3352 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3353 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3354 my $new;
3355 while ( <$msg_fh> ) {
3356 $new=1;last;
3357 }
3358 command_close_pipe($msg_fh, $ctx);
3359 if ( $new ) {
3360 print STDERR
3361 "Found merge parent (svk:merge ticket): $parent\n";
3362 push @$parents, $parent;
3363 }
3364 }
3365 }
3367 sub lookup_svn_merge {
3368 my $uuid = shift;
3369 my $url = shift;
3370 my $merge = shift;
3372 my ($source, $revs) = split ":", $merge;
3373 my $path = $source;
3374 $path =~ s{^/}{};
3375 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3376 if ( !$gs ) {
3377 warn "Couldn't find revmap for $url$source\n";
3378 return;
3379 }
3380 my @ranges = split ",", $revs;
3381 my ($tip, $tip_commit);
3382 my @merged_commit_ranges;
3383 # find the tip
3384 for my $range ( @ranges ) {
3385 my ($bottom, $top) = split "-", $range;
3386 $top ||= $bottom;
3387 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3388 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3390 unless ($top_commit and $bottom_commit) {
3391 warn "W:unknown path/rev in svn:mergeinfo "
3392 ."dirprop: $source:$range\n";
3393 next;
3394 }
3396 if (scalar(command('rev-parse', "$bottom_commit^@"))) {
3397 push @merged_commit_ranges,
3398 "$bottom_commit^..$top_commit";
3399 } else {
3400 push @merged_commit_ranges, "$top_commit";
3401 }
3403 if ( !defined $tip or $top > $tip ) {
3404 $tip = $top;
3405 $tip_commit = $top_commit;
3406 }
3407 }
3408 return ($tip_commit, @merged_commit_ranges);
3409 }
3411 sub _rev_list {
3412 my ($msg_fh, $ctx) = command_output_pipe(
3413 "rev-list", @_,
3414 );
3415 my @rv;
3416 while ( <$msg_fh> ) {
3417 chomp;
3418 push @rv, $_;
3419 }
3420 command_close_pipe($msg_fh, $ctx);
3421 @rv;
3422 }
3424 sub check_cherry_pick {
3425 my $base = shift;
3426 my $tip = shift;
3427 my $parents = shift;
3428 my @ranges = @_;
3429 my %commits = map { $_ => 1 }
3430 _rev_list("--no-merges", $tip, "--not", $base, @$parents, "--");
3431 for my $range ( @ranges ) {
3432 delete @commits{_rev_list($range, "--")};
3433 }
3434 for my $commit (keys %commits) {
3435 if (has_no_changes($commit)) {
3436 delete $commits{$commit};
3437 }
3438 }
3439 return (keys %commits);
3440 }
3442 sub has_no_changes {
3443 my $commit = shift;
3445 my @revs = split / /, command_oneline(
3446 qw(rev-list --parents -1 -m), $commit);
3448 # Commits with no parents, e.g. the start of a partial branch,
3449 # have changes by definition.
3450 return 1 if (@revs < 2);
3452 # Commits with multiple parents, e.g a merge, have no changes
3453 # by definition.
3454 return 0 if (@revs > 2);
3456 return (command_oneline("rev-parse", "$commit^{tree}") eq
3457 command_oneline("rev-parse", "$commit~1^{tree}"));
3458 }
3460 # The GIT_DIR environment variable is not always set until after the command
3461 # line arguments are processed, so we can't memoize in a BEGIN block.
3462 {
3463 my $memoized = 0;
3465 sub memoize_svn_mergeinfo_functions {
3466 return if $memoized;
3467 $memoized = 1;
3469 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3470 mkpath([$cache_path]) unless -d $cache_path;
3472 tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3473 "$cache_path/lookup_svn_merge.db", 'nstore';
3474 memoize 'lookup_svn_merge',
3475 SCALAR_CACHE => 'FAULT',
3476 LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3477 ;
3479 tie my %check_cherry_pick_cache => 'Memoize::Storable',
3480 "$cache_path/check_cherry_pick.db", 'nstore';
3481 memoize 'check_cherry_pick',
3482 SCALAR_CACHE => 'FAULT',
3483 LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3484 ;
3486 tie my %has_no_changes_cache => 'Memoize::Storable',
3487 "$cache_path/has_no_changes.db", 'nstore';
3488 memoize 'has_no_changes',
3489 SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3490 LIST_CACHE => 'FAULT',
3491 ;
3492 }
3494 sub unmemoize_svn_mergeinfo_functions {
3495 return if not $memoized;
3496 $memoized = 0;
3498 Memoize::unmemoize 'lookup_svn_merge';
3499 Memoize::unmemoize 'check_cherry_pick';
3500 Memoize::unmemoize 'has_no_changes';
3501 }
3503 Memoize::memoize 'Git::SVN::repos_root';
3504 }
3506 END {
3507 # Force cache writeout explicitly instead of waiting for
3508 # global destruction to avoid segfault in Storable:
3509 # http://rt.cpan.org/Public/Bug/Display.html?id=36087
3510 unmemoize_svn_mergeinfo_functions();
3511 }
3513 sub parents_exclude {
3514 my $parents = shift;
3515 my @commits = @_;
3516 return unless @commits;
3518 my @excluded;
3519 my $excluded;
3520 do {
3521 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3522 $excluded = command_oneline(@cmd);
3523 if ( $excluded ) {
3524 my @new;
3525 my $found;
3526 for my $commit ( @commits ) {
3527 if ( $commit eq $excluded ) {
3528 push @excluded, $commit;
3529 $found++;
3530 last;
3531 }
3532 else {
3533 push @new, $commit;
3534 }
3535 }
3536 die "saw commit '$excluded' in rev-list output, "
3537 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3538 unless $found;
3539 @commits = @new;
3540 }
3541 }
3542 while ($excluded and @commits);
3544 return @excluded;
3545 }
3548 # note: this function should only be called if the various dirprops
3549 # have actually changed
3550 sub find_extra_svn_parents {
3551 my ($self, $ed, $mergeinfo, $parents) = @_;
3552 # aha! svk:merge property changed...
3554 memoize_svn_mergeinfo_functions();
3556 # We first search for merged tips which are not in our
3557 # history. Then, we figure out which git revisions are in
3558 # that tip, but not this revision. If all of those revisions
3559 # are now marked as merge, we can add the tip as a parent.
3560 my @merges = split "\n", $mergeinfo;
3561 my @merge_tips;
3562 my $url = $self->{url};
3563 my $uuid = $self->ra_uuid;
3564 my %ranges;
3565 for my $merge ( @merges ) {
3566 my ($tip_commit, @ranges) =
3567 lookup_svn_merge( $uuid, $url, $merge );
3568 unless (!$tip_commit or
3569 grep { $_ eq $tip_commit } @$parents ) {
3570 push @merge_tips, $tip_commit;
3571 $ranges{$tip_commit} = \@ranges;
3572 } else {
3573 push @merge_tips, undef;
3574 }
3575 }
3577 my %excluded = map { $_ => 1 }
3578 parents_exclude($parents, grep { defined } @merge_tips);
3580 # check merge tips for new parents
3581 my @new_parents;
3582 for my $merge_tip ( @merge_tips ) {
3583 my $spec = shift @merges;
3584 next unless $merge_tip and $excluded{$merge_tip};
3586 my $ranges = $ranges{$merge_tip};
3588 # check out 'new' tips
3589 my $merge_base;
3590 eval {
3591 $merge_base = command_oneline(
3592 "merge-base",
3593 @$parents, $merge_tip,
3594 );
3595 };
3596 if ($@) {
3597 die "An error occurred during merge-base"
3598 unless $@->isa("Git::Error::Command");
3600 warn "W: Cannot find common ancestor between ".
3601 "@$parents and $merge_tip. Ignoring merge info.\n";
3602 next;
3603 }
3605 # double check that there are no missing non-merge commits
3606 my (@incomplete) = check_cherry_pick(
3607 $merge_base, $merge_tip,
3608 $parents,
3609 @$ranges,
3610 );
3612 if ( @incomplete ) {
3613 warn "W:svn cherry-pick ignored ($spec) - missing "
3614 .@incomplete." commit(s) (eg $incomplete[0])\n";
3615 } else {
3616 warn
3617 "Found merge parent (svn:mergeinfo prop): ",
3618 $merge_tip, "\n";
3619 push @new_parents, $merge_tip;
3620 }
3621 }
3623 # cater for merges which merge commits from multiple branches
3624 if ( @new_parents > 1 ) {
3625 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3626 for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3627 next if $i == $j;
3628 next unless $new_parents[$i];
3629 next unless $new_parents[$j];
3630 my $revs = command_oneline(
3631 "rev-list", "-1",
3632 "$new_parents[$i]..$new_parents[$j]",
3633 );
3634 if ( !$revs ) {
3635 undef($new_parents[$j]);
3636 }
3637 }
3638 }
3639 }
3640 push @$parents, grep { defined } @new_parents;
3641 }
3643 sub make_log_entry {
3644 my ($self, $rev, $parents, $ed) = @_;
3645 my $untracked = $self->get_untracked($ed);
3647 my @parents = @$parents;
3648 my $ps = $ed->{path_strip} || "";
3649 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3650 my $props = $ed->{dir_prop}{$path};
3651 if ( $props->{"svk:merge"} ) {
3652 $self->find_extra_svk_parents
3653 ($ed, $props->{"svk:merge"}, \@parents);
3654 }
3655 if ( $props->{"svn:mergeinfo"} ) {
3656 $self->find_extra_svn_parents
3657 ($ed,
3658 $props->{"svn:mergeinfo"},
3659 \@parents);
3660 }
3661 }
3663 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3664 print $un "r$rev\n" or croak $!;
3665 print $un $_, "\n" foreach @$untracked;
3666 my %log_entry = ( parents => \@parents, revision => $rev,
3667 log => '');
3669 my $headrev;
3670 my $logged = delete $self->{logged_rev_props};
3671 if (!$logged || $self->{-want_revprops}) {
3672 my $rp = $self->ra->rev_proplist($rev);
3673 foreach (sort keys %$rp) {
3674 my $v = $rp->{$_};
3675 if (/^svn:(author|date|log)$/) {
3676 $log_entry{$1} = $v;
3677 } elsif ($_ eq 'svm:headrev') {
3678 $headrev = $v;
3679 } else {
3680 print $un " rev_prop: ", uri_encode($_), ' ',
3681 uri_encode($v), "\n";
3682 }
3683 }
3684 } else {
3685 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3686 }
3687 close $un or croak $!;
3689 $log_entry{date} = parse_svn_date($log_entry{date});
3690 $log_entry{log} .= "\n";
3691 my $author = $log_entry{author} = check_author($log_entry{author});
3692 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3693 : ($author, undef);
3695 my ($commit_name, $commit_email) = ($name, $email);
3696 if ($_use_log_author) {
3697 my $name_field;
3698 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3699 $name_field = $1;
3700 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3701 $name_field = $1;
3702 }
3703 if (!defined $name_field) {
3704 if (!defined $email) {
3705 $email = $name;
3706 }
3707 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3708 ($name, $email) = ($1, $2);
3709 } elsif ($name_field =~ /(.*)@/) {
3710 ($name, $email) = ($1, $name_field);
3711 } else {
3712 ($name, $email) = ($name_field, $name_field);
3713 }
3714 }
3715 if (defined $headrev && $self->use_svm_props) {
3716 if ($self->rewrite_root) {
3717 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3718 "options set!\n";
3719 }
3720 if ($self->rewrite_uuid) {
3721 die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3722 "options set!\n";
3723 }
3724 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3725 # we don't want "SVM: initializing mirror for junk" ...
3726 return undef if $r == 0;
3727 my $svm = $self->svm;
3728 if ($uuid ne $svm->{uuid}) {
3729 die "UUID mismatch on SVM path:\n",
3730 "expected: $svm->{uuid}\n",
3731 " got: $uuid\n";
3732 }
3733 my $full_url = $self->full_url;
3734 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3735 die "Failed to replace '$svm->{replace}' with ",
3736 "'$svm->{source}' in $full_url\n";
3737 # throw away username for storing in records
3738 remove_username($full_url);
3739 $log_entry{metadata} = "$full_url\@$r $uuid";
3740 $log_entry{svm_revision} = $r;
3741 $email ||= "$author\@$uuid";
3742 $commit_email ||= "$author\@$uuid";
3743 } elsif ($self->use_svnsync_props) {
3744 my $full_url = $self->svnsync->{url};
3745 $full_url .= "/$self->{path}" if length $self->{path};
3746 remove_username($full_url);
3747 my $uuid = $self->svnsync->{uuid};
3748 $log_entry{metadata} = "$full_url\@$rev $uuid";
3749 $email ||= "$author\@$uuid";
3750 $commit_email ||= "$author\@$uuid";
3751 } else {
3752 my $url = $self->metadata_url;
3753 remove_username($url);
3754 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3755 $log_entry{metadata} = "$url\@$rev " . $uuid;
3756 $email ||= "$author\@" . $uuid;
3757 $commit_email ||= "$author\@" . $uuid;
3758 }
3759 $log_entry{name} = $name;
3760 $log_entry{email} = $email;
3761 $log_entry{commit_name} = $commit_name;
3762 $log_entry{commit_email} = $commit_email;
3763 \%log_entry;
3764 }
3766 sub fetch {
3767 my ($self, $min_rev, $max_rev, @parents) = @_;
3768 my ($last_rev, $last_commit) = $self->last_rev_commit;
3769 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3770 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3771 }
3773 sub set_tree_cb {
3774 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3775 $self->{inject_parents} = { $rev => $tree };
3776 $self->fetch(undef, undef);
3777 }
3779 sub set_tree {
3780 my ($self, $tree) = (shift, shift);
3781 my $log_entry = ::get_commit_entry($tree);
3782 unless ($self->{last_rev}) {
3783 ::fatal("Must have an existing revision to commit");
3784 }
3785 my %ed_opts = ( r => $self->{last_rev},
3786 log => $log_entry->{log},
3787 ra => $self->ra,
3788 tree_a => $self->{last_commit},
3789 tree_b => $tree,
3790 editor_cb => sub {
3791 $self->set_tree_cb($log_entry, $tree, @_) },
3792 svn_path => $self->{path} );
3793 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3794 print "No changes\nr$self->{last_rev} = $tree\n";
3795 }
3796 }
3798 sub rebuild_from_rev_db {
3799 my ($self, $path) = @_;
3800 my $r = -1;
3801 open my $fh, '<', $path or croak "open: $!";
3802 binmode $fh or croak "binmode: $!";
3803 while (<$fh>) {
3804 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3805 chomp($_);
3806 ++$r;
3807 next if $_ eq ('0' x 40);
3808 $self->rev_map_set($r, $_);
3809 print "r$r = $_\n";
3810 }
3811 close $fh or croak "close: $!";
3812 unlink $path or croak "unlink: $!";
3813 }
3815 sub rebuild {
3816 my ($self) = @_;
3817 my $map_path = $self->map_path;
3818 my $partial = (-e $map_path && ! -z $map_path);
3819 return unless ::verify_ref($self->refname.'^0');
3820 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3821 my $rev_db = $self->rev_db_path;
3822 $self->rebuild_from_rev_db($rev_db);
3823 if ($self->use_svm_props) {
3824 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3825 $self->rebuild_from_rev_db($svm_rev_db);
3826 }
3827 $self->unlink_rev_db_symlink;
3828 return;
3829 }
3830 print "Rebuilding $map_path ...\n" if (!$partial);
3831 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3832 (undef, undef));
3833 my ($log, $ctx) =
3834 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3835 ($head ? "$head.." : "") . $self->refname,
3836 '--');
3837 my $metadata_url = $self->metadata_url;
3838 remove_username($metadata_url);
3839 my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3840 my $c;
3841 while (<$log>) {
3842 if ( m{^commit ($::sha1)$} ) {
3843 $c = $1;
3844 next;
3845 }
3846 next unless s{^\s*(git-svn-id:)}{$1};
3847 my ($url, $rev, $uuid) = ::extract_metadata($_);
3848 remove_username($url);
3850 # ignore merges (from set-tree)
3851 next if (!defined $rev || !$uuid);
3853 # if we merged or otherwise started elsewhere, this is
3854 # how we break out of it
3855 if (($uuid ne $svn_uuid) ||
3856 ($metadata_url && $url && ($url ne $metadata_url))) {
3857 next;
3858 }
3859 if ($partial && $head) {
3860 print "Partial-rebuilding $map_path ...\n";
3861 print "Currently at $base_rev = $head\n";
3862 $head = undef;
3863 }
3865 $self->rev_map_set($rev, $c);
3866 print "r$rev = $c\n";
3867 }
3868 command_close_pipe($log, $ctx);
3869 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3870 my $rev_db_path = $self->rev_db_path;
3871 if (-f $self->rev_db_path) {
3872 unlink $self->rev_db_path or croak "unlink: $!";
3873 }
3874 $self->unlink_rev_db_symlink;
3875 }
3877 # rev_map:
3878 # Tie::File seems to be prone to offset errors if revisions get sparse,
3879 # it's not that fast, either. Tie::File is also not in Perl 5.6. So
3880 # one of my favorite modules is out :< Next up would be one of the DBM
3881 # modules, but I'm not sure which is most portable...
3882 #
3883 # This is the replacement for the rev_db format, which was too big
3884 # and inefficient for large repositories with a lot of sparse history
3885 # (mainly tags)
3886 #
3887 # The format is this:
3888 # - 24 bytes for every record,
3889 # * 4 bytes for the integer representing an SVN revision number
3890 # * 20 bytes representing the sha1 of a git commit
3891 # - No empty padding records like the old format
3892 # (except the last record, which can be overwritten)
3893 # - new records are written append-only since SVN revision numbers
3894 # increase monotonically
3895 # - lookups on SVN revision number are done via a binary search
3896 # - Piping the file to xxd -c24 is a good way of dumping it for
3897 # viewing or editing (piped back through xxd -r), should the need
3898 # ever arise.
3899 # - The last record can be padding revision with an all-zero sha1
3900 # This is used to optimize fetch performance when using multiple
3901 # "fetch" directives in .git/config
3902 #
3903 # These files are disposable unless noMetadata or useSvmProps is set
3905 sub _rev_map_set {
3906 my ($fh, $rev, $commit) = @_;
3908 binmode $fh or croak "binmode: $!";
3909 my $size = (stat($fh))[7];
3910 ($size % 24) == 0 or croak "inconsistent size: $size";
3912 my $wr_offset = 0;
3913 if ($size > 0) {
3914 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3915 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3916 $read == 24 or croak "read only $read bytes (!= 24)";
3917 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3918 if ($last_commit eq ('0' x40)) {
3919 if ($size >= 48) {
3920 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3921 $read = sysread($fh, $buf, 24) or
3922 croak "read: $!";
3923 $read == 24 or
3924 croak "read only $read bytes (!= 24)";
3925 ($last_rev, $last_commit) =
3926 unpack(rev_map_fmt, $buf);
3927 if ($last_commit eq ('0' x40)) {
3928 croak "inconsistent .rev_map\n";
3929 }
3930 }
3931 if ($last_rev >= $rev) {
3932 croak "last_rev is higher!: $last_rev >= $rev";
3933 }
3934 $wr_offset = -24;
3935 }
3936 }
3937 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3938 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3939 croak "write: $!";
3940 }
3942 sub _rev_map_reset {
3943 my ($fh, $rev, $commit) = @_;
3944 my $c = _rev_map_get($fh, $rev);
3945 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3946 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3947 truncate $fh, $offset or croak "truncate: $!";
3948 }
3950 sub mkfile {
3951 my ($path) = @_;
3952 unless (-e $path) {
3953 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3954 mkpath([$dir]) unless -d $dir;
3955 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3956 close $fh or die "Couldn't close (create) $path: $!\n";
3957 }
3958 }
3960 sub rev_map_set {
3961 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3962 defined $commit or die "missing arg3\n";
3963 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3964 my $db = $self->map_path($uuid);
3965 my $db_lock = "$db.lock";
3966 my $sig;
3967 $update_ref ||= 0;
3968 if ($update_ref) {
3969 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3970 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3971 }
3972 mkfile($db);
3974 $LOCKFILES{$db_lock} = 1;
3975 my $sync;
3976 # both of these options make our .rev_db file very, very important
3977 # and we can't afford to lose it because rebuild() won't work
3978 if ($self->use_svm_props || $self->no_metadata) {
3979 $sync = 1;
3980 copy($db, $db_lock) or die "rev_map_set(@_): ",
3981 "Failed to copy: ",
3982 "$db => $db_lock ($!)\n";
3983 } else {
3984 rename $db, $db_lock or die "rev_map_set(@_): ",
3985 "Failed to rename: ",
3986 "$db => $db_lock ($!)\n";
3987 }
3989 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3990 or croak "Couldn't open $db_lock: $!\n";
3991 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3992 _rev_map_set($fh, $rev, $commit);
3993 if ($sync) {
3994 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3995 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3996 }
3997 close $fh or croak $!;
3998 if ($update_ref) {
3999 $_head = $self;
4000 my $note = "";
4001 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
4002 command_noisy('update-ref', '-m', "r$rev$note",
4003 $self->refname, $commit);
4004 }
4005 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
4006 "$db_lock => $db ($!)\n";
4007 delete $LOCKFILES{$db_lock};
4008 if ($update_ref) {
4009 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
4010 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
4011 kill $sig, $$ if defined $sig;
4012 }
4013 }
4015 # If want_commit, this will return an array of (rev, commit) where
4016 # commit _must_ be a valid commit in the archive.
4017 # Otherwise, it'll return the max revision (whether or not the
4018 # commit is valid or just a 0x40 placeholder).
4019 sub rev_map_max {
4020 my ($self, $want_commit) = @_;
4021 $self->rebuild;
4022 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
4023 $want_commit ? ($r, $c) : $r;
4024 }
4026 sub rev_map_max_norebuild {
4027 my ($self, $want_commit) = @_;
4028 my $map_path = $self->map_path;
4029 stat $map_path or return $want_commit ? (0, undef) : 0;
4030 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
4031 binmode $fh or croak "binmode: $!";
4032 my $size = (stat($fh))[7];
4033 ($size % 24) == 0 or croak "inconsistent size: $size";
4035 if ($size == 0) {
4036 close $fh or croak "close: $!";
4037 return $want_commit ? (0, undef) : 0;
4038 }
4040 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
4041 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
4042 my ($r, $c) = unpack(rev_map_fmt, $buf);
4043 if ($want_commit && $c eq ('0' x40)) {
4044 if ($size < 48) {
4045 return $want_commit ? (0, undef) : 0;
4046 }
4047 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
4048 sysread($fh, $buf, 24) == 24 or croak "read: $!";
4049 ($r, $c) = unpack(rev_map_fmt, $buf);
4050 if ($c eq ('0'x40)) {
4051 croak "Penultimate record is all-zeroes in $map_path";
4052 }
4053 }
4054 close $fh or croak "close: $!";
4055 $want_commit ? ($r, $c) : $r;
4056 }
4058 sub rev_map_get {
4059 my ($self, $rev, $uuid) = @_;
4060 my $map_path = $self->map_path($uuid);
4061 return undef unless -e $map_path;
4063 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
4064 my $c = _rev_map_get($fh, $rev);
4065 close($fh) or croak "close: $!";
4066 $c
4067 }
4069 sub _rev_map_get {
4070 my ($fh, $rev) = @_;
4072 binmode $fh or croak "binmode: $!";
4073 my $size = (stat($fh))[7];
4074 ($size % 24) == 0 or croak "inconsistent size: $size";
4076 if ($size == 0) {
4077 return undef;
4078 }
4080 my ($l, $u) = (0, $size - 24);
4081 my ($r, $c, $buf);
4083 while ($l <= $u) {
4084 my $i = int(($l/24 + $u/24) / 2) * 24;
4085 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
4086 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
4087 my ($r, $c) = unpack(rev_map_fmt, $buf);
4089 if ($r < $rev) {
4090 $l = $i + 24;
4091 } elsif ($r > $rev) {
4092 $u = $i - 24;
4093 } else { # $r == $rev
4094 return $c eq ('0' x 40) ? undef : $c;
4095 }
4096 }
4097 undef;
4098 }
4100 # Finds the first svn revision that exists on (if $eq_ok is true) or
4101 # before $rev for the current branch. It will not search any lower
4102 # than $min_rev. Returns the git commit hash and svn revision number
4103 # if found, else (undef, undef).
4104 sub find_rev_before {
4105 my ($self, $rev, $eq_ok, $min_rev) = @_;
4106 --$rev unless $eq_ok;
4107 $min_rev ||= 1;
4108 my $max_rev = $self->rev_map_max;
4109 $rev = $max_rev if ($rev > $max_rev);
4110 while ($rev >= $min_rev) {
4111 if (my $c = $self->rev_map_get($rev)) {
4112 return ($rev, $c);
4113 }
4114 --$rev;
4115 }
4116 return (undef, undef);
4117 }
4119 # Finds the first svn revision that exists on (if $eq_ok is true) or
4120 # after $rev for the current branch. It will not search any higher
4121 # than $max_rev. Returns the git commit hash and svn revision number
4122 # if found, else (undef, undef).
4123 sub find_rev_after {
4124 my ($self, $rev, $eq_ok, $max_rev) = @_;
4125 ++$rev unless $eq_ok;
4126 $max_rev ||= $self->rev_map_max;
4127 while ($rev <= $max_rev) {
4128 if (my $c = $self->rev_map_get($rev)) {
4129 return ($rev, $c);
4130 }
4131 ++$rev;
4132 }
4133 return (undef, undef);
4134 }
4136 sub _new {
4137 my ($class, $repo_id, $ref_id, $path) = @_;
4138 unless (defined $repo_id && length $repo_id) {
4139 $repo_id = $Git::SVN::default_repo_id;
4140 }
4141 unless (defined $ref_id && length $ref_id) {
4142 $_prefix = '' unless defined($_prefix);
4143 $_[2] = $ref_id =
4144 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
4145 }
4146 $_[1] = $repo_id;
4147 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
4149 # Older repos imported by us used $GIT_DIR/svn/foo instead of
4150 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
4151 if ($ref_id =~ m{^refs/remotes/(.*)}) {
4152 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
4153 if (-d $old_dir && ! -d $dir) {
4154 $dir = $old_dir;
4155 }
4156 }
4158 $_[3] = $path = '' unless (defined $path);
4159 mkpath([$dir]);
4160 bless {
4161 ref_id => $ref_id, dir => $dir, index => "$dir/index",
4162 path => $path, config => "$ENV{GIT_DIR}/svn/config",
4163 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
4164 }
4166 # for read-only access of old .rev_db formats
4167 sub unlink_rev_db_symlink {
4168 my ($self) = @_;
4169 my $link = $self->rev_db_path;
4170 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
4171 if (-l $link) {
4172 unlink $link or croak "unlink: $link failed!";
4173 }
4174 }
4176 sub rev_db_path {
4177 my ($self, $uuid) = @_;
4178 my $db_path = $self->map_path($uuid);
4179 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
4180 or croak "map_path: $db_path does not contain '/.rev_map.' !";
4181 $db_path;
4182 }
4184 # the new replacement for .rev_db
4185 sub map_path {
4186 my ($self, $uuid) = @_;
4187 $uuid ||= $self->ra_uuid;
4188 "$self->{map_root}.$uuid";
4189 }
4191 sub uri_encode {
4192 my ($f) = @_;
4193 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
4194 $f
4195 }
4197 sub uri_decode {
4198 my ($f) = @_;
4199 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
4200 $f
4201 }
4203 sub remove_username {
4204 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
4205 }
4207 package Git::SVN::Prompt;
4208 use strict;
4209 use warnings;
4210 require SVN::Core;
4211 use vars qw/$_no_auth_cache $_username/;
4213 sub simple {
4214 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
4215 $may_save = undef if $_no_auth_cache;
4216 $default_username = $_username if defined $_username;
4217 if (defined $default_username && length $default_username) {
4218 if (defined $realm && length $realm) {
4219 print STDERR "Authentication realm: $realm\n";
4220 STDERR->flush;
4221 }
4222 $cred->username($default_username);
4223 } else {
4224 username($cred, $realm, $may_save, $pool);
4225 }
4226 $cred->password(_read_password("Password for '" .
4227 $cred->username . "': ", $realm));
4228 $cred->may_save($may_save);
4229 $SVN::_Core::SVN_NO_ERROR;
4230 }
4232 sub ssl_server_trust {
4233 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
4234 $may_save = undef if $_no_auth_cache;
4235 print STDERR "Error validating server certificate for '$realm':\n";
4236 {
4237 no warnings 'once';
4238 # All variables SVN::Auth::SSL::* are used only once,
4239 # so we're shutting up Perl warnings about this.
4240 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
4241 print STDERR " - The certificate is not issued ",
4242 "by a trusted authority. Use the\n",
4243 " fingerprint to validate ",
4244 "the certificate manually!\n";
4245 }
4246 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
4247 print STDERR " - The certificate hostname ",
4248 "does not match.\n";
4249 }
4250 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
4251 print STDERR " - The certificate is not yet valid.\n";
4252 }
4253 if ($failures & $SVN::Auth::SSL::EXPIRED) {
4254 print STDERR " - The certificate has expired.\n";
4255 }
4256 if ($failures & $SVN::Auth::SSL::OTHER) {
4257 print STDERR " - The certificate has ",
4258 "an unknown error.\n";
4259 }
4260 } # no warnings 'once'
4261 printf STDERR
4262 "Certificate information:\n".
4263 " - Hostname: %s\n".
4264 " - Valid: from %s until %s\n".
4265 " - Issuer: %s\n".
4266 " - Fingerprint: %s\n",
4267 map $cert_info->$_, qw(hostname valid_from valid_until
4268 issuer_dname fingerprint);
4269 my $choice;
4270 prompt:
4271 print STDERR $may_save ?
4272 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
4273 "(R)eject or accept (t)emporarily? ";
4274 STDERR->flush;
4275 $choice = lc(substr(<STDIN> || 'R', 0, 1));
4276 if ($choice =~ /^t$/i) {
4277 $cred->may_save(undef);
4278 } elsif ($choice =~ /^r$/i) {
4279 return -1;
4280 } elsif ($may_save && $choice =~ /^p$/i) {
4281 $cred->may_save($may_save);
4282 } else {
4283 goto prompt;
4284 }
4285 $cred->accepted_failures($failures);
4286 $SVN::_Core::SVN_NO_ERROR;
4287 }
4289 sub ssl_client_cert {
4290 my ($cred, $realm, $may_save, $pool) = @_;
4291 $may_save = undef if $_no_auth_cache;
4292 print STDERR "Client certificate filename: ";
4293 STDERR->flush;
4294 chomp(my $filename = <STDIN>);
4295 $cred->cert_file($filename);
4296 $cred->may_save($may_save);
4297 $SVN::_Core::SVN_NO_ERROR;
4298 }
4300 sub ssl_client_cert_pw {
4301 my ($cred, $realm, $may_save, $pool) = @_;
4302 $may_save = undef if $_no_auth_cache;
4303 $cred->password(_read_password("Password: ", $realm));
4304 $cred->may_save($may_save);
4305 $SVN::_Core::SVN_NO_ERROR;
4306 }
4308 sub username {
4309 my ($cred, $realm, $may_save, $pool) = @_;
4310 $may_save = undef if $_no_auth_cache;
4311 if (defined $realm && length $realm) {
4312 print STDERR "Authentication realm: $realm\n";
4313 }
4314 my $username;
4315 if (defined $_username) {
4316 $username = $_username;
4317 } else {
4318 print STDERR "Username: ";
4319 STDERR->flush;
4320 chomp($username = <STDIN>);
4321 }
4322 $cred->username($username);
4323 $cred->may_save($may_save);
4324 $SVN::_Core::SVN_NO_ERROR;
4325 }
4327 sub _read_password {
4328 my ($prompt, $realm) = @_;
4329 my $password = '';
4330 if (exists $ENV{GIT_ASKPASS}) {
4331 open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
4332 $password = <PH>;
4333 $password =~ s/[\012\015]//; # \n\r
4334 close(PH);
4335 } else {
4336 print STDERR $prompt;
4337 STDERR->flush;
4338 require Term::ReadKey;
4339 Term::ReadKey::ReadMode('noecho');
4340 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
4341 last if $key =~ /[\012\015]/; # \n\r
4342 $password .= $key;
4343 }
4344 Term::ReadKey::ReadMode('restore');
4345 print STDERR "\n";
4346 STDERR->flush;
4347 }
4348 $password;
4349 }
4351 package SVN::Git::Fetcher;
4352 use vars qw/@ISA $_ignore_regex $_preserve_empty_dirs $_placeholder_filename
4353 @deleted_gpath %added_placeholder $repo_id/;
4354 use strict;
4355 use warnings;
4356 use Carp qw/croak/;
4357 use File::Basename qw/dirname/;
4358 use IO::File qw//;
4360 # file baton members: path, mode_a, mode_b, pool, fh, blob, base
4361 sub new {
4362 my ($class, $git_svn, $switch_path) = @_;
4363 my $self = SVN::Delta::Editor->new;
4364 bless $self, $class;
4365 if (exists $git_svn->{last_commit}) {
4366 $self->{c} = $git_svn->{last_commit};
4367 $self->{empty_symlinks} =
4368 _mark_empty_symlinks($git_svn, $switch_path);
4369 }
4371 # some options are read globally, but can be overridden locally
4372 # per [svn-remote "..."] section. Command-line options will *NOT*
4373 # override options set in an [svn-remote "..."] section
4374 $repo_id = $git_svn->{repo_id};
4375 my $k = "svn-remote.$repo_id.ignore-paths";
4376 my $v = eval { command_oneline('config', '--get', $k) };
4377 $self->{ignore_regex} = $v;
4379 $k = "svn-remote.$repo_id.preserve-empty-dirs";
4380 $v = eval { command_oneline('config', '--get', '--bool', $k) };
4381 if ($v && $v eq 'true') {
4382 $_preserve_empty_dirs = 1;
4383 $k = "svn-remote.$repo_id.placeholder-filename";
4384 $v = eval { command_oneline('config', '--get', $k) };
4385 $_placeholder_filename = $v;
4386 }
4388 # Load the list of placeholder files added during previous invocations.
4389 $k = "svn-remote.$repo_id.added-placeholder";
4390 $v = eval { command_oneline('config', '--get-all', $k) };
4391 if ($_preserve_empty_dirs && $v) {
4392 # command() prints errors to stderr, so we only call it if
4393 # command_oneline() succeeded.
4394 my @v = command('config', '--get-all', $k);
4395 $added_placeholder{ dirname($_) } = $_ foreach @v;
4396 }
4398 $self->{empty} = {};
4399 $self->{dir_prop} = {};
4400 $self->{file_prop} = {};
4401 $self->{absent_dir} = {};
4402 $self->{absent_file} = {};
4403 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4404 $self->{pathnameencoding} = Git::config('svn.pathnameencoding');
4405 $self;
4406 }
4408 # this uses the Ra object, so it must be called before do_{switch,update},
4409 # not inside them (when the Git::SVN::Fetcher object is passed) to
4410 # do_{switch,update}
4411 sub _mark_empty_symlinks {
4412 my ($git_svn, $switch_path) = @_;
4413 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4414 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4416 my %ret;
4417 my ($rev, $cmt) = $git_svn->last_rev_commit;
4418 return {} unless ($rev && $cmt);
4420 # allow the warning to be printed for each revision we fetch to
4421 # ensure the user sees it. The user can also disable the workaround
4422 # on the repository even while git svn is running and the next
4423 # revision fetched will skip this expensive function.
4424 my $printed_warning;
4425 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4426 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4427 local $/ = "\0";
4428 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4429 $pfx .= '/' if length($pfx);
4430 while (<$ls>) {
4431 chomp;
4432 s/\A100644 blob $empty_blob\t//o or next;
4433 unless ($printed_warning) {
4434 print STDERR "Scanning for empty symlinks, ",
4435 "this may take a while if you have ",
4436 "many empty files\n",
4437 "You may disable this with `",
4438 "git config svn.brokenSymlinkWorkaround ",
4439 "false'.\n",
4440 "This may be done in a different ",
4441 "terminal without restarting ",
4442 "git svn\n";
4443 $printed_warning = 1;
4444 }
4445 my $path = $_;
4446 my (undef, $props) =
4447 $git_svn->ra->get_file($pfx.$path, $rev, undef);
4448 if ($props->{'svn:special'}) {
4449 $ret{$path} = 1;
4450 }
4451 }
4452 command_close_pipe($ls, $ctx);
4453 \%ret;
4454 }
4456 # returns true if a given path is inside a ".git" directory
4457 sub in_dot_git {
4458 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4459 }
4461 # return value: 0 -- don't ignore, 1 -- ignore
4462 sub is_path_ignored {
4463 my ($self, $path) = @_;
4464 return 1 if in_dot_git($path);
4465 return 1 if defined($self->{ignore_regex}) &&
4466 $path =~ m!$self->{ignore_regex}!;
4467 return 0 unless defined($_ignore_regex);
4468 return 1 if $path =~ m!$_ignore_regex!o;
4469 return 0;
4470 }
4472 sub set_path_strip {
4473 my ($self, $path) = @_;
4474 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4475 }
4477 sub open_root {
4478 { path => '' };
4479 }
4481 sub open_directory {
4482 my ($self, $path, $pb, $rev) = @_;
4483 { path => $path };
4484 }
4486 sub git_path {
4487 my ($self, $path) = @_;
4488 if (my $enc = $self->{pathnameencoding}) {
4489 require Encode;
4490 Encode::from_to($path, 'UTF-8', $enc);
4491 }
4492 if ($self->{path_strip}) {
4493 $path =~ s!$self->{path_strip}!! or
4494 die "Failed to strip path '$path' ($self->{path_strip})\n";
4495 }
4496 $path;
4497 }
4499 sub delete_entry {
4500 my ($self, $path, $rev, $pb) = @_;
4501 return undef if $self->is_path_ignored($path);
4503 my $gpath = $self->git_path($path);
4504 return undef if ($gpath eq '');
4506 # remove entire directories.
4507 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4508 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4509 if ($tree) {
4510 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4511 -r --name-only -z/,
4512 $tree);
4513 local $/ = "\0";
4514 while (<$ls>) {
4515 chomp;
4516 my $rmpath = "$gpath/$_";
4517 $self->{gii}->remove($rmpath);
4518 print "\tD\t$rmpath\n" unless $::_q;
4519 }
4520 print "\tD\t$gpath/\n" unless $::_q;
4521 command_close_pipe($ls, $ctx);
4522 } else {
4523 $self->{gii}->remove($gpath);
4524 print "\tD\t$gpath\n" unless $::_q;
4525 }
4526 # Don't add to @deleted_gpath if we're deleting a placeholder file.
4527 push @deleted_gpath, $gpath unless $added_placeholder{dirname($path)};
4528 $self->{empty}->{$path} = 0;
4529 undef;
4530 }
4532 sub open_file {
4533 my ($self, $path, $pb, $rev) = @_;
4534 my ($mode, $blob);
4536 goto out if $self->is_path_ignored($path);
4538 my $gpath = $self->git_path($path);
4539 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4540 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4541 unless (defined $mode && defined $blob) {
4542 die "$path was not found in commit $self->{c} (r$rev)\n";
4543 }
4544 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4545 $mode = '120000';
4546 }
4547 out:
4548 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4549 pool => SVN::Pool->new, action => 'M' };
4550 }
4552 sub add_file {
4553 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4554 my $mode;
4556 if (!$self->is_path_ignored($path)) {
4557 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4558 delete $self->{empty}->{$dir};
4559 $mode = '100644';
4561 if ($added_placeholder{$dir}) {
4562 # Remove our placeholder file, if we created one.
4563 delete_entry($self, $added_placeholder{$dir})
4564 unless $path eq $added_placeholder{$dir};
4565 delete $added_placeholder{$dir}
4566 }
4567 }
4569 { path => $path, mode_a => $mode, mode_b => $mode,
4570 pool => SVN::Pool->new, action => 'A' };
4571 }
4573 sub add_directory {
4574 my ($self, $path, $cp_path, $cp_rev) = @_;
4575 goto out if $self->is_path_ignored($path);
4576 my $gpath = $self->git_path($path);
4577 if ($gpath eq '') {
4578 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4579 -r --name-only -z/,
4580 $self->{c});
4581 local $/ = "\0";
4582 while (<$ls>) {
4583 chomp;
4584 $self->{gii}->remove($_);
4585 print "\tD\t$_\n" unless $::_q;
4586 push @deleted_gpath, $gpath;
4587 }
4588 command_close_pipe($ls, $ctx);
4589 $self->{empty}->{$path} = 0;
4590 }
4591 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4592 delete $self->{empty}->{$dir};
4593 $self->{empty}->{$path} = 1;
4595 if ($added_placeholder{$dir}) {
4596 # Remove our placeholder file, if we created one.
4597 delete_entry($self, $added_placeholder{$dir});
4598 delete $added_placeholder{$dir}
4599 }
4601 out:
4602 { path => $path };
4603 }
4605 sub change_dir_prop {
4606 my ($self, $db, $prop, $value) = @_;
4607 return undef if $self->is_path_ignored($db->{path});
4608 $self->{dir_prop}->{$db->{path}} ||= {};
4609 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4610 undef;
4611 }
4613 sub absent_directory {
4614 my ($self, $path, $pb) = @_;
4615 return undef if $self->is_path_ignored($path);
4616 $self->{absent_dir}->{$pb->{path}} ||= [];
4617 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4618 undef;
4619 }
4621 sub absent_file {
4622 my ($self, $path, $pb) = @_;
4623 return undef if $self->is_path_ignored($path);
4624 $self->{absent_file}->{$pb->{path}} ||= [];
4625 push @{$self->{absent_file}->{$pb->{path}}}, $path;
4626 undef;
4627 }
4629 sub change_file_prop {
4630 my ($self, $fb, $prop, $value) = @_;
4631 return undef if $self->is_path_ignored($fb->{path});
4632 if ($prop eq 'svn:executable') {
4633 if ($fb->{mode_b} != 120000) {
4634 $fb->{mode_b} = defined $value ? 100755 : 100644;
4635 }
4636 } elsif ($prop eq 'svn:special') {
4637 $fb->{mode_b} = defined $value ? 120000 : 100644;
4638 } else {
4639 $self->{file_prop}->{$fb->{path}} ||= {};
4640 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4641 }
4642 undef;
4643 }
4645 sub apply_textdelta {
4646 my ($self, $fb, $exp) = @_;
4647 return undef if $self->is_path_ignored($fb->{path});
4648 my $fh = $::_repository->temp_acquire('svn_delta');
4649 # $fh gets auto-closed() by SVN::TxDelta::apply(),
4650 # (but $base does not,) so dup() it for reading in close_file
4651 open my $dup, '<&', $fh or croak $!;
4652 my $base = $::_repository->temp_acquire('git_blob');
4654 if ($fb->{blob}) {
4655 my ($base_is_link, $size);
4657 if ($fb->{mode_a} eq '120000' &&
4658 ! $self->{empty_symlinks}->{$fb->{path}}) {
4659 print $base 'link ' or die "print $!\n";
4660 $base_is_link = 1;
4661 }
4662 retry:
4663 $size = $::_repository->cat_blob($fb->{blob}, $base);
4664 die "Failed to read object $fb->{blob}" if ($size < 0);
4666 if (defined $exp) {
4667 seek $base, 0, 0 or croak $!;
4668 my $got = ::md5sum($base);
4669 if ($got ne $exp) {
4670 my $err = "Checksum mismatch: ".
4671 "$fb->{path} $fb->{blob}\n" .
4672 "expected: $exp\n" .
4673 " got: $got\n";
4674 if ($base_is_link) {
4675 warn $err,
4676 "Retrying... (possibly ",
4677 "a bad symlink from SVN)\n";
4678 $::_repository->temp_reset($base);
4679 $base_is_link = 0;
4680 goto retry;
4681 }
4682 die $err;
4683 }
4684 }
4685 }
4686 seek $base, 0, 0 or croak $!;
4687 $fb->{fh} = $fh;
4688 $fb->{base} = $base;
4689 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4690 }
4692 sub close_file {
4693 my ($self, $fb, $exp) = @_;
4694 return undef if $self->is_path_ignored($fb->{path});
4696 my $hash;
4697 my $path = $self->git_path($fb->{path});
4698 if (my $fh = $fb->{fh}) {
4699 if (defined $exp) {
4700 seek($fh, 0, 0) or croak $!;
4701 my $got = ::md5sum($fh);
4702 if ($got ne $exp) {
4703 die "Checksum mismatch: $path\n",
4704 "expected: $exp\n got: $got\n";
4705 }
4706 }
4707 if ($fb->{mode_b} == 120000) {
4708 sysseek($fh, 0, 0) or croak $!;
4709 my $rd = sysread($fh, my $buf, 5);
4711 if (!defined $rd) {
4712 croak "sysread: $!\n";
4713 } elsif ($rd == 0) {
4714 warn "$path has mode 120000",
4715 " but it points to nothing\n",
4716 "converting to an empty file with mode",
4717 " 100644\n";
4718 $fb->{mode_b} = '100644';
4719 } elsif ($buf ne 'link ') {
4720 warn "$path has mode 120000",
4721 " but is not a link\n";
4722 } else {
4723 my $tmp_fh = $::_repository->temp_acquire(
4724 'svn_hash');
4725 my $res;
4726 while ($res = sysread($fh, my $str, 1024)) {
4727 my $out = syswrite($tmp_fh, $str, $res);
4728 defined($out) && $out == $res
4729 or croak("write ",
4730 Git::temp_path($tmp_fh),
4731 ": $!\n");
4732 }
4733 defined $res or croak $!;
4735 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4736 Git::temp_release($tmp_fh, 1);
4737 }
4738 }
4740 $hash = $::_repository->hash_and_insert_object(
4741 Git::temp_path($fh));
4742 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4744 Git::temp_release($fb->{base}, 1);
4745 Git::temp_release($fh, 1);
4746 } else {
4747 $hash = $fb->{blob} or die "no blob information\n";
4748 }
4749 $fb->{pool}->clear;
4750 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4751 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4752 undef;
4753 }
4755 sub abort_edit {
4756 my $self = shift;
4757 $self->{nr} = $self->{gii}->{nr};
4758 delete $self->{gii};
4759 $self->SUPER::abort_edit(@_);
4760 }
4762 sub close_edit {
4763 my $self = shift;
4765 if ($_preserve_empty_dirs) {
4766 my @empty_dirs;
4768 # Any entry flagged as empty that also has an associated
4769 # dir_prop represents a newly created empty directory.
4770 foreach my $i (keys %{$self->{empty}}) {
4771 push @empty_dirs, $i if exists $self->{dir_prop}->{$i};
4772 }
4774 # Search for directories that have become empty due subsequent
4775 # file deletes.
4776 push @empty_dirs, $self->find_empty_directories();
4778 # Finally, add a placeholder file to each empty directory.
4779 $self->add_placeholder_file($_) foreach (@empty_dirs);
4781 $self->stash_placeholder_list();
4782 }
4784 $self->{git_commit_ok} = 1;
4785 $self->{nr} = $self->{gii}->{nr};
4786 delete $self->{gii};
4787 $self->SUPER::close_edit(@_);
4788 }
4790 sub find_empty_directories {
4791 my ($self) = @_;
4792 my @empty_dirs;
4793 my %dirs = map { dirname($_) => 1 } @deleted_gpath;
4795 foreach my $dir (sort keys %dirs) {
4796 next if $dir eq ".";
4798 # If there have been any additions to this directory, there is
4799 # no reason to check if it is empty.
4800 my $skip_added = 0;
4801 foreach my $t (qw/dir_prop file_prop/) {
4802 foreach my $path (keys %{ $self->{$t} }) {
4803 if (exists $self->{$t}->{dirname($path)}) {
4804 $skip_added = 1;
4805 last;
4806 }
4807 }
4808 last if $skip_added;
4809 }
4810 next if $skip_added;
4812 # Use `git ls-tree` to get the filenames of this directory
4813 # that existed prior to this particular commit.
4814 my $ls = command('ls-tree', '-z', '--name-only',
4815 $self->{c}, "$dir/");
4816 my %files = map { $_ => 1 } split(/\0/, $ls);
4818 # Remove the filenames that were deleted during this commit.
4819 delete $files{$_} foreach (@deleted_gpath);
4821 # Report the directory if there are no filenames left.
4822 push @empty_dirs, $dir unless (scalar %files);
4823 }
4824 @empty_dirs;
4825 }
4827 sub add_placeholder_file {
4828 my ($self, $dir) = @_;
4829 my $path = "$dir/$_placeholder_filename";
4830 my $gpath = $self->git_path($path);
4832 my $fh = $::_repository->temp_acquire($gpath);
4833 my $hash = $::_repository->hash_and_insert_object(Git::temp_path($fh));
4834 Git::temp_release($fh, 1);
4835 $self->{gii}->update('100644', $hash, $gpath) or croak $!;
4837 # The directory should no longer be considered empty.
4838 delete $self->{empty}->{$dir} if exists $self->{empty}->{$dir};
4840 # Keep track of any placeholder files we create.
4841 $added_placeholder{$dir} = $path;
4842 }
4844 sub stash_placeholder_list {
4845 my ($self) = @_;
4846 my $k = "svn-remote.$repo_id.added-placeholder";
4847 my $v = eval { command_oneline('config', '--get-all', $k) };
4848 command_noisy('config', '--unset-all', $k) if $v;
4849 foreach (values %added_placeholder) {
4850 command_noisy('config', '--add', $k, $_);
4851 }
4852 }
4854 package SVN::Git::Editor;
4855 use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4856 use strict;
4857 use warnings;
4858 use Carp qw/croak/;
4859 use IO::File;
4861 sub new {
4862 my ($class, $opts) = @_;
4863 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4864 die "$_ required!\n" unless (defined $opts->{$_});
4865 }
4867 my $pool = SVN::Pool->new;
4868 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4869 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4870 $opts->{r}, $mods);
4872 # $opts->{ra} functions should not be used after this:
4873 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4874 $opts->{editor_cb}, $pool);
4875 my $self = SVN::Delta::Editor->new(@ce, $pool);
4876 bless $self, $class;
4877 foreach (qw/svn_path r tree_a tree_b/) {
4878 $self->{$_} = $opts->{$_};
4879 }
4880 $self->{url} = $opts->{ra}->{url};
4881 $self->{mods} = $mods;
4882 $self->{types} = $types;
4883 $self->{pool} = $pool;
4884 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4885 $self->{rm} = { };
4886 $self->{path_prefix} = length $self->{svn_path} ?
4887 "$self->{svn_path}/" : '';
4888 $self->{config} = $opts->{config};
4889 $self->{mergeinfo} = $opts->{mergeinfo};
4890 return $self;
4891 }
4893 sub generate_diff {
4894 my ($tree_a, $tree_b) = @_;
4895 my @diff_tree = qw(diff-tree -z -r);
4896 if ($_cp_similarity) {
4897 push @diff_tree, "-C$_cp_similarity";
4898 } else {
4899 push @diff_tree, '-C';
4900 }
4901 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4902 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4903 push @diff_tree, $tree_a, $tree_b;
4904 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4905 local $/ = "\0";
4906 my $state = 'meta';
4907 my @mods;
4908 while (<$diff_fh>) {
4909 chomp $_; # this gets rid of the trailing "\0"
4910 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4911 ($::sha1)\s($::sha1)\s
4912 ([MTCRAD])\d*$/xo) {
4913 push @mods, { mode_a => $1, mode_b => $2,
4914 sha1_a => $3, sha1_b => $4,
4915 chg => $5 };
4916 if ($5 =~ /^(?:C|R)$/) {
4917 $state = 'file_a';
4918 } else {
4919 $state = 'file_b';
4920 }
4921 } elsif ($state eq 'file_a') {
4922 my $x = $mods[$#mods] or croak "Empty array\n";
4923 if ($x->{chg} !~ /^(?:C|R)$/) {
4924 croak "Error parsing $_, $x->{chg}\n";
4925 }
4926 $x->{file_a} = $_;
4927 $state = 'file_b';
4928 } elsif ($state eq 'file_b') {
4929 my $x = $mods[$#mods] or croak "Empty array\n";
4930 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4931 croak "Error parsing $_, $x->{chg}\n";
4932 }
4933 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4934 croak "Error parsing $_, $x->{chg}\n";
4935 }
4936 $x->{file_b} = $_;
4937 $state = 'meta';
4938 } else {
4939 croak "Error parsing $_\n";
4940 }
4941 }
4942 command_close_pipe($diff_fh, $ctx);
4943 \@mods;
4944 }
4946 sub check_diff_paths {
4947 my ($ra, $pfx, $rev, $mods) = @_;
4948 my %types;
4949 $pfx .= '/' if length $pfx;
4951 sub type_diff_paths {
4952 my ($ra, $types, $path, $rev) = @_;
4953 my @p = split m#/+#, $path;
4954 my $c = shift @p;
4955 unless (defined $types->{$c}) {
4956 $types->{$c} = $ra->check_path($c, $rev);
4957 }
4958 while (@p) {
4959 $c .= '/' . shift @p;
4960 next if defined $types->{$c};
4961 $types->{$c} = $ra->check_path($c, $rev);
4962 }
4963 }
4965 foreach my $m (@$mods) {
4966 foreach my $f (qw/file_a file_b/) {
4967 next unless defined $m->{$f};
4968 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4969 if (length $pfx.$dir && ! defined $types{$dir}) {
4970 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4971 }
4972 }
4973 }
4974 \%types;
4975 }
4977 sub split_path {
4978 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4979 }
4981 sub repo_path {
4982 my ($self, $path) = @_;
4983 if (my $enc = $self->{pathnameencoding}) {
4984 require Encode;
4985 Encode::from_to($path, $enc, 'UTF-8');
4986 }
4987 $self->{path_prefix}.(defined $path ? $path : '');
4988 }
4990 sub url_path {
4991 my ($self, $path) = @_;
4992 if ($self->{url} =~ m#^https?://#) {
4993 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4994 }
4995 $self->{url} . '/' . $self->repo_path($path);
4996 }
4998 sub rmdirs {
4999 my ($self) = @_;
5000 my $rm = $self->{rm};
5001 delete $rm->{''}; # we never delete the url we're tracking
5002 return unless %$rm;
5004 foreach (keys %$rm) {
5005 my @d = split m#/#, $_;
5006 my $c = shift @d;
5007 $rm->{$c} = 1;
5008 while (@d) {
5009 $c .= '/' . shift @d;
5010 $rm->{$c} = 1;
5011 }
5012 }
5013 delete $rm->{$self->{svn_path}};
5014 delete $rm->{''}; # we never delete the url we're tracking
5015 return unless %$rm;
5017 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
5018 $self->{tree_b});
5019 local $/ = "\0";
5020 while (<$fh>) {
5021 chomp;
5022 my @dn = split m#/#, $_;
5023 while (pop @dn) {
5024 delete $rm->{join '/', @dn};
5025 }
5026 unless (%$rm) {
5027 close $fh;
5028 return;
5029 }
5030 }
5031 command_close_pipe($fh, $ctx);
5033 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
5034 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
5035 $self->close_directory($bat->{$d}, $p);
5036 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
5037 print "\tD+\t$d/\n" unless $::_q;
5038 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
5039 delete $bat->{$d};
5040 }
5041 }
5043 sub open_or_add_dir {
5044 my ($self, $full_path, $baton) = @_;
5045 my $t = $self->{types}->{$full_path};
5046 if (!defined $t) {
5047 die "$full_path not known in r$self->{r} or we have a bug!\n";
5048 }
5049 {
5050 no warnings 'once';
5051 # SVN::Node::none and SVN::Node::file are used only once,
5052 # so we're shutting up Perl's warnings about them.
5053 if ($t == $SVN::Node::none) {
5054 return $self->add_directory($full_path, $baton,
5055 undef, -1, $self->{pool});
5056 } elsif ($t == $SVN::Node::dir) {
5057 return $self->open_directory($full_path, $baton,
5058 $self->{r}, $self->{pool});
5059 } # no warnings 'once'
5060 print STDERR "$full_path already exists in repository at ",
5061 "r$self->{r} and it is not a directory (",
5062 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
5063 } # no warnings 'once'
5064 exit 1;
5065 }
5067 sub ensure_path {
5068 my ($self, $path) = @_;
5069 my $bat = $self->{bat};
5070 my $repo_path = $self->repo_path($path);
5071 return $bat->{''} unless (length $repo_path);
5072 my @p = split m#/+#, $repo_path;
5073 my $c = shift @p;
5074 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
5075 while (@p) {
5076 my $c0 = $c;
5077 $c .= '/' . shift @p;
5078 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
5079 }
5080 return $bat->{$c};
5081 }
5083 # Subroutine to convert a globbing pattern to a regular expression.
5084 # From perl cookbook.
5085 sub glob2pat {
5086 my $globstr = shift;
5087 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
5088 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
5089 return '^' . $globstr . '$';
5090 }
5092 sub check_autoprop {
5093 my ($self, $pattern, $properties, $file, $fbat) = @_;
5094 # Convert the globbing pattern to a regular expression.
5095 my $regex = glob2pat($pattern);
5096 # Check if the pattern matches the file name.
5097 if($file =~ m/($regex)/) {
5098 # Parse the list of properties to set.
5099 my @props = split(/;/, $properties);
5100 foreach my $prop (@props) {
5101 # Parse 'name=value' syntax and set the property.
5102 if ($prop =~ /([^=]+)=(.*)/) {
5103 my ($n,$v) = ($1,$2);
5104 for ($n, $v) {
5105 s/^\s+//; s/\s+$//;
5106 }
5107 $self->change_file_prop($fbat, $n, $v);
5108 }
5109 }
5110 }
5111 }
5113 sub apply_autoprops {
5114 my ($self, $file, $fbat) = @_;
5115 my $conf_t = ${$self->{config}}{'config'};
5116 no warnings 'once';
5117 # Check [miscellany]/enable-auto-props in svn configuration.
5118 if (SVN::_Core::svn_config_get_bool(
5119 $conf_t,
5120 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
5121 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
5122 0)) {
5123 # Auto-props are enabled. Enumerate them to look for matches.
5124 my $callback = sub {
5125 $self->check_autoprop($_[0], $_[1], $file, $fbat);
5126 };
5127 SVN::_Core::svn_config_enumerate(
5128 $conf_t,
5129 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
5130 $callback);
5131 }
5132 }
5134 sub A {
5135 my ($self, $m) = @_;
5136 my ($dir, $file) = split_path($m->{file_b});
5137 my $pbat = $self->ensure_path($dir);
5138 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
5139 undef, -1);
5140 print "\tA\t$m->{file_b}\n" unless $::_q;
5141 $self->apply_autoprops($file, $fbat);
5142 $self->chg_file($fbat, $m);
5143 $self->close_file($fbat,undef,$self->{pool});
5144 }
5146 sub C {
5147 my ($self, $m) = @_;
5148 my ($dir, $file) = split_path($m->{file_b});
5149 my $pbat = $self->ensure_path($dir);
5150 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
5151 $self->url_path($m->{file_a}), $self->{r});
5152 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
5153 $self->chg_file($fbat, $m);
5154 $self->close_file($fbat,undef,$self->{pool});
5155 }
5157 sub delete_entry {
5158 my ($self, $path, $pbat) = @_;
5159 my $rpath = $self->repo_path($path);
5160 my ($dir, $file) = split_path($rpath);
5161 $self->{rm}->{$dir} = 1;
5162 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
5163 }
5165 sub R {
5166 my ($self, $m) = @_;
5167 my ($dir, $file) = split_path($m->{file_b});
5168 my $pbat = $self->ensure_path($dir);
5169 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
5170 $self->url_path($m->{file_a}), $self->{r});
5171 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
5172 $self->apply_autoprops($file, $fbat);
5173 $self->chg_file($fbat, $m);
5174 $self->close_file($fbat,undef,$self->{pool});
5176 ($dir, $file) = split_path($m->{file_a});
5177 $pbat = $self->ensure_path($dir);
5178 $self->delete_entry($m->{file_a}, $pbat);
5179 }
5181 sub M {
5182 my ($self, $m) = @_;
5183 my ($dir, $file) = split_path($m->{file_b});
5184 my $pbat = $self->ensure_path($dir);
5185 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
5186 $pbat,$self->{r},$self->{pool});
5187 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
5188 $self->chg_file($fbat, $m);
5189 $self->close_file($fbat,undef,$self->{pool});
5190 }
5192 sub T { shift->M(@_) }
5194 sub change_file_prop {
5195 my ($self, $fbat, $pname, $pval) = @_;
5196 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
5197 }
5199 sub change_dir_prop {
5200 my ($self, $pbat, $pname, $pval) = @_;
5201 $self->SUPER::change_dir_prop($pbat, $pname, $pval, $self->{pool});
5202 }
5204 sub _chg_file_get_blob ($$$$) {
5205 my ($self, $fbat, $m, $which) = @_;
5206 my $fh = $::_repository->temp_acquire("git_blob_$which");
5207 if ($m->{"mode_$which"} =~ /^120/) {
5208 print $fh 'link ' or croak $!;
5209 $self->change_file_prop($fbat,'svn:special','*');
5210 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
5211 $self->change_file_prop($fbat,'svn:special',undef);
5212 }
5213 my $blob = $m->{"sha1_$which"};
5214 return ($fh,) if ($blob =~ /^0{40}$/);
5215 my $size = $::_repository->cat_blob($blob, $fh);
5216 croak "Failed to read object $blob" if ($size < 0);
5217 $fh->flush == 0 or croak $!;
5218 seek $fh, 0, 0 or croak $!;
5220 my $exp = ::md5sum($fh);
5221 seek $fh, 0, 0 or croak $!;
5222 return ($fh, $exp);
5223 }
5225 sub chg_file {
5226 my ($self, $fbat, $m) = @_;
5227 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
5228 $self->change_file_prop($fbat,'svn:executable','*');
5229 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
5230 $self->change_file_prop($fbat,'svn:executable',undef);
5231 }
5232 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
5233 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
5234 my $pool = SVN::Pool->new;
5235 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
5236 if (-s $fh_a) {
5237 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
5238 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
5239 if (defined $res) {
5240 die "Unexpected result from send_txstream: $res\n",
5241 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
5242 }
5243 } else {
5244 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
5245 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
5246 if ($got ne $exp_b);
5247 }
5248 Git::temp_release($fh_b, 1);
5249 Git::temp_release($fh_a, 1);
5250 $pool->clear;
5251 }
5253 sub D {
5254 my ($self, $m) = @_;
5255 my ($dir, $file) = split_path($m->{file_b});
5256 my $pbat = $self->ensure_path($dir);
5257 print "\tD\t$m->{file_b}\n" unless $::_q;
5258 $self->delete_entry($m->{file_b}, $pbat);
5259 }
5261 sub close_edit {
5262 my ($self) = @_;
5263 my ($p,$bat) = ($self->{pool}, $self->{bat});
5264 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
5265 next if $_ eq '';
5266 $self->close_directory($bat->{$_}, $p);
5267 }
5268 $self->close_directory($bat->{''}, $p);
5269 $self->SUPER::close_edit($p);
5270 $p->clear;
5271 }
5273 sub abort_edit {
5274 my ($self) = @_;
5275 $self->SUPER::abort_edit($self->{pool});
5276 }
5278 sub DESTROY {
5279 my $self = shift;
5280 $self->SUPER::DESTROY(@_);
5281 $self->{pool}->clear;
5282 }
5284 # this drives the editor
5285 sub apply_diff {
5286 my ($self) = @_;
5287 my $mods = $self->{mods};
5288 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
5289 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
5290 my $f = $m->{chg};
5291 if (defined $o{$f}) {
5292 $self->$f($m);
5293 } else {
5294 fatal("Invalid change type: $f");
5295 }
5296 }
5298 if (defined($self->{mergeinfo})) {
5299 $self->change_dir_prop($self->{bat}{''}, "svn:mergeinfo",
5300 $self->{mergeinfo});
5301 }
5302 $self->rmdirs if $_rmdir;
5303 if (@$mods == 0) {
5304 $self->abort_edit;
5305 } else {
5306 $self->close_edit;
5307 }
5308 return scalar @$mods;
5309 }
5311 package Git::SVN::Ra;
5312 use vars qw/@ISA $config_dir $_log_window_size/;
5313 use strict;
5314 use warnings;
5315 my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
5317 BEGIN {
5318 # enforce temporary pool usage for some simple functions
5319 no strict 'refs';
5320 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
5321 get_file/) {
5322 my $SUPER = "SUPER::$f";
5323 *$f = sub {
5324 my $self = shift;
5325 my $pool = SVN::Pool->new;
5326 my @ret = $self->$SUPER(@_,$pool);
5327 $pool->clear;
5328 wantarray ? @ret : $ret[0];
5329 };
5330 }
5331 }
5333 sub _auth_providers () {
5334 [
5335 SVN::Client::get_simple_provider(),
5336 SVN::Client::get_ssl_server_trust_file_provider(),
5337 SVN::Client::get_simple_prompt_provider(
5338 \&Git::SVN::Prompt::simple, 2),
5339 SVN::Client::get_ssl_client_cert_file_provider(),
5340 SVN::Client::get_ssl_client_cert_prompt_provider(
5341 \&Git::SVN::Prompt::ssl_client_cert, 2),
5342 SVN::Client::get_ssl_client_cert_pw_file_provider(),
5343 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
5344 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
5345 SVN::Client::get_username_provider(),
5346 SVN::Client::get_ssl_server_trust_prompt_provider(
5347 \&Git::SVN::Prompt::ssl_server_trust),
5348 SVN::Client::get_username_prompt_provider(
5349 \&Git::SVN::Prompt::username, 2)
5350 ]
5351 }
5353 sub escape_uri_only {
5354 my ($uri) = @_;
5355 my @tmp;
5356 foreach (split m{/}, $uri) {
5357 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
5358 push @tmp, $_;
5359 }
5360 join('/', @tmp);
5361 }
5363 sub escape_url {
5364 my ($url) = @_;
5365 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
5366 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
5367 $url = "$scheme://$domain$uri";
5368 }
5369 $url;
5370 }
5372 sub new {
5373 my ($class, $url) = @_;
5374 $url =~ s!/+$!!;
5375 return $RA if ($RA && $RA->{url} eq $url);
5377 ::_req_svn();
5379 SVN::_Core::svn_config_ensure($config_dir, undef);
5380 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
5381 my $config = SVN::Core::config_get_config($config_dir);
5382 $RA = undef;
5383 my $dont_store_passwords = 1;
5384 my $conf_t = ${$config}{'config'};
5385 {
5386 no warnings 'once';
5387 # The usage of $SVN::_Core::SVN_CONFIG_* variables
5388 # produces warnings that variables are used only once.
5389 # I had not found the better way to shut them up, so
5390 # the warnings of type 'once' are disabled in this block.
5391 if (SVN::_Core::svn_config_get_bool($conf_t,
5392 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
5393 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
5394 1) == 0) {
5395 SVN::_Core::svn_auth_set_parameter($baton,
5396 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
5397 bless (\$dont_store_passwords, "_p_void"));
5398 }
5399 if (SVN::_Core::svn_config_get_bool($conf_t,
5400 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
5401 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
5402 1) == 0) {
5403 $Git::SVN::Prompt::_no_auth_cache = 1;
5404 }
5405 } # no warnings 'once'
5406 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
5407 config => $config,
5408 pool => SVN::Pool->new,
5409 auth_provider_callbacks => $callbacks);
5410 $self->{url} = $url;
5411 $self->{svn_path} = $url;
5412 $self->{repos_root} = $self->get_repos_root;
5413 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
5414 $self->{cache} = { check_path => { r => 0, data => {} },
5415 get_dir => { r => 0, data => {} } };
5416 $RA = bless $self, $class;
5417 }
5419 sub check_path {
5420 my ($self, $path, $r) = @_;
5421 my $cache = $self->{cache}->{check_path};
5422 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
5423 return $cache->{data}->{$path};
5424 }
5425 my $pool = SVN::Pool->new;
5426 my $t = $self->SUPER::check_path($path, $r, $pool);
5427 $pool->clear;
5428 if ($r != $cache->{r}) {
5429 %{$cache->{data}} = ();
5430 $cache->{r} = $r;
5431 }
5432 $cache->{data}->{$path} = $t;
5433 }
5435 sub get_dir {
5436 my ($self, $dir, $r) = @_;
5437 my $cache = $self->{cache}->{get_dir};
5438 if ($r == $cache->{r}) {
5439 if (my $x = $cache->{data}->{$dir}) {
5440 return wantarray ? @$x : $x->[0];
5441 }
5442 }
5443 my $pool = SVN::Pool->new;
5444 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
5445 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
5446 $pool->clear;
5447 if ($r != $cache->{r}) {
5448 %{$cache->{data}} = ();
5449 $cache->{r} = $r;
5450 }
5451 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
5452 wantarray ? (\%dirents, $r, $props) : \%dirents;
5453 }
5455 sub DESTROY {
5456 # do not call the real DESTROY since we store ourselves in $RA
5457 }
5459 # get_log(paths, start, end, limit,
5460 # discover_changed_paths, strict_node_history, receiver)
5461 sub get_log {
5462 my ($self, @args) = @_;
5463 my $pool = SVN::Pool->new;
5465 # svn_log_changed_path_t objects passed to get_log are likely to be
5466 # overwritten even if only the refs are copied to an external variable,
5467 # so we should dup the structures in their entirety. Using an
5468 # externally passed pool (instead of our temporary and quickly cleared
5469 # pool in Git::SVN::Ra) does not help matters at all...
5470 my $receiver = pop @args;
5471 my $prefix = "/".$self->{svn_path};
5472 $prefix =~ s#/+($)##;
5473 my $prefix_regex = qr#^\Q$prefix\E#;
5474 push(@args, sub {
5475 my ($paths) = $_[0];
5476 return &$receiver(@_) unless $paths;
5477 $_[0] = ();
5478 foreach my $p (keys %$paths) {
5479 my $i = $paths->{$p};
5480 # Make path relative to our url, not repos_root
5481 $p =~ s/$prefix_regex//;
5482 my %s = map { $_ => $i->$_; }
5483 qw/copyfrom_path copyfrom_rev action/;
5484 if ($s{'copyfrom_path'}) {
5485 $s{'copyfrom_path'} =~ s/$prefix_regex//;
5486 }
5487 $_[0]{$p} = \%s;
5488 }
5489 &$receiver(@_);
5490 });
5493 # the limit parameter was not supported in SVN 1.1.x, so we
5494 # drop it. Therefore, the receiver callback passed to it
5495 # is made aware of this limitation by being wrapped if
5496 # the limit passed to is being wrapped.
5497 if ($SVN::Core::VERSION le '1.2.0') {
5498 my $limit = splice(@args, 3, 1);
5499 if ($limit > 0) {
5500 my $receiver = pop @args;
5501 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
5502 }
5503 }
5504 my $ret = $self->SUPER::get_log(@args, $pool);
5505 $pool->clear;
5506 $ret;
5507 }
5509 sub trees_match {
5510 my ($self, $url1, $rev1, $url2, $rev2) = @_;
5511 my $ctx = SVN::Client->new(auth => _auth_providers);
5512 my $out = IO::File->new_tmpfile;
5514 # older SVN (1.1.x) doesn't take $pool as the last parameter for
5515 # $ctx->diff(), so we'll create a default one
5516 my $pool = SVN::Pool->new_default_sub;
5518 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5519 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5520 $out->flush;
5521 my $ret = (($out->stat)[7] == 0);
5522 close $out or croak $!;
5524 $ret;
5525 }
5527 sub get_commit_editor {
5528 my ($self, $log, $cb, $pool) = @_;
5529 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
5530 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5531 }
5533 sub gs_do_update {
5534 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5535 my $new = ($rev_a == $rev_b);
5536 my $path = $gs->{path};
5538 if ($new && -e $gs->{index}) {
5539 unlink $gs->{index} or die
5540 "Couldn't unlink index: $gs->{index}: $!\n";
5541 }
5542 my $pool = SVN::Pool->new;
5543 $editor->set_path_strip($path);
5544 my (@pc) = split m#/#, $path;
5545 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5546 1, $editor, $pool);
5547 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5549 # Since we can't rely on svn_ra_reparent being available, we'll
5550 # just have to do some magic with set_path to make it so
5551 # we only want a partial path.
5552 my $sp = '';
5553 my $final = join('/', @pc);
5554 while (@pc) {
5555 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5556 $sp .= '/' if length $sp;
5557 $sp .= shift @pc;
5558 }
5559 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5561 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5563 $reporter->finish_report($pool);
5564 $pool->clear;
5565 $editor->{git_commit_ok};
5566 }
5568 # this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5569 # svn_ra_reparent didn't work before 1.4)
5570 sub gs_do_switch {
5571 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5572 my $path = $gs->{path};
5573 my $pool = SVN::Pool->new;
5575 my $full_url = $self->{url};
5576 my $old_url = $full_url;
5577 $full_url .= '/' . $path if length $path;
5578 my ($ra, $reparented);
5580 if ($old_url =~ m#^svn(\+ssh)?://# ||
5581 ($full_url =~ m#^https?://# &&
5582 escape_url($full_url) ne $full_url)) {
5583 $_[0] = undef;
5584 $self = undef;
5585 $RA = undef;
5586 $ra = Git::SVN::Ra->new($full_url);
5587 $ra_invalid = 1;
5588 } elsif ($old_url ne $full_url) {
5589 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5590 $self->{url} = $full_url;
5591 $reparented = 1;
5592 }
5594 $ra ||= $self;
5595 $url_b = escape_url($url_b);
5596 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5597 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5598 $reporter->set_path('', $rev_a, 0, @lock, $pool);
5599 $reporter->finish_report($pool);
5601 if ($reparented) {
5602 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5603 $self->{url} = $old_url;
5604 }
5606 $pool->clear;
5607 $editor->{git_commit_ok};
5608 }
5610 sub longest_common_path {
5611 my ($gsv, $globs) = @_;
5612 my %common;
5613 my $common_max = scalar @$gsv;
5615 foreach my $gs (@$gsv) {
5616 my @tmp = split m#/#, $gs->{path};
5617 my $p = '';
5618 foreach (@tmp) {
5619 $p .= length($p) ? "/$_" : $_;
5620 $common{$p} ||= 0;
5621 $common{$p}++;
5622 }
5623 }
5624 $globs ||= [];
5625 $common_max += scalar @$globs;
5626 foreach my $glob (@$globs) {
5627 my @tmp = split m#/#, $glob->{path}->{left};
5628 my $p = '';
5629 foreach (@tmp) {
5630 $p .= length($p) ? "/$_" : $_;
5631 $common{$p} ||= 0;
5632 $common{$p}++;
5633 }
5634 }
5636 my $longest_path = '';
5637 foreach (sort {length $b <=> length $a} keys %common) {
5638 if ($common{$_} == $common_max) {
5639 $longest_path = $_;
5640 last;
5641 }
5642 }
5643 $longest_path;
5644 }
5646 sub gs_fetch_loop_common {
5647 my ($self, $base, $head, $gsv, $globs) = @_;
5648 return if ($base > $head);
5649 my $inc = $_log_window_size;
5650 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5651 my $longest_path = longest_common_path($gsv, $globs);
5652 my $ra_url = $self->{url};
5653 my $find_trailing_edge;
5654 while (1) {
5655 my %revs;
5656 my $err;
5657 my $err_handler = $SVN::Error::handler;
5658 $SVN::Error::handler = sub {
5659 ($err) = @_;
5660 skip_unknown_revs($err);
5661 };
5662 sub _cb {
5663 my ($paths, $r, $author, $date, $log) = @_;
5664 [ $paths,
5665 { author => $author, date => $date, log => $log } ];
5666 }
5667 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5668 sub { $revs{$_[1]} = _cb(@_) });
5669 if ($err) {
5670 print "Checked through r$max\r";
5671 } else {
5672 $find_trailing_edge = 1;
5673 }
5674 if ($err and $find_trailing_edge) {
5675 print STDERR "Path '$longest_path' ",
5676 "was probably deleted:\n",
5677 $err->expanded_message,
5678 "\nWill attempt to follow ",
5679 "revisions r$min .. r$max ",
5680 "committed before the deletion\n";
5681 my $hi = $max;
5682 while (--$hi >= $min) {
5683 my $ok;
5684 $self->get_log([$longest_path], $min, $hi,
5685 0, 1, 1, sub {
5686 $ok = $_[1];
5687 $revs{$_[1]} = _cb(@_) });
5688 if ($ok) {
5689 print STDERR "r$min .. r$ok OK\n";
5690 last;
5691 }
5692 }
5693 $find_trailing_edge = 0;
5694 }
5695 $SVN::Error::handler = $err_handler;
5697 my %exists = map { $_->{path} => $_ } @$gsv;
5698 foreach my $r (sort {$a <=> $b} keys %revs) {
5699 my ($paths, $logged) = @{$revs{$r}};
5701 foreach my $gs ($self->match_globs(\%exists, $paths,
5702 $globs, $r)) {
5703 if ($gs->rev_map_max >= $r) {
5704 next;
5705 }
5706 next unless $gs->match_paths($paths, $r);
5707 $gs->{logged_rev_props} = $logged;
5708 if (my $last_commit = $gs->last_commit) {
5709 $gs->assert_index_clean($last_commit);
5710 }
5711 my $log_entry = $gs->do_fetch($paths, $r);
5712 if ($log_entry) {
5713 $gs->do_git_commit($log_entry);
5714 }
5715 $INDEX_FILES{$gs->{index}} = 1;
5716 }
5717 foreach my $g (@$globs) {
5718 my $k = "svn-remote.$g->{remote}." .
5719 "$g->{t}-maxRev";
5720 Git::SVN::tmp_config($k, $r);
5721 }
5722 if ($ra_invalid) {
5723 $_[0] = undef;
5724 $self = undef;
5725 $RA = undef;
5726 $self = Git::SVN::Ra->new($ra_url);
5727 $ra_invalid = undef;
5728 }
5729 }
5730 # pre-fill the .rev_db since it'll eventually get filled in
5731 # with '0' x40 if something new gets committed
5732 foreach my $gs (@$gsv) {
5733 next if $gs->rev_map_max >= $max;
5734 next if defined $gs->rev_map_get($max);
5735 $gs->rev_map_set($max, 0 x40);
5736 }
5737 foreach my $g (@$globs) {
5738 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5739 Git::SVN::tmp_config($k, $max);
5740 }
5741 last if $max >= $head;
5742 $min = $max + 1;
5743 $max += $inc;
5744 $max = $head if ($max > $head);
5745 }
5746 Git::SVN::gc();
5747 }
5749 sub get_dir_globbed {
5750 my ($self, $left, $depth, $r) = @_;
5752 my @x = eval { $self->get_dir($left, $r) };
5753 return unless scalar @x == 3;
5754 my $dirents = $x[0];
5755 my @finalents;
5756 foreach my $de (keys %$dirents) {
5757 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5758 if ($depth > 1) {
5759 my @args = ("$left/$de", $depth - 1, $r);
5760 foreach my $dir ($self->get_dir_globbed(@args)) {
5761 push @finalents, "$de/$dir";
5762 }
5763 } else {
5764 push @finalents, $de;
5765 }
5766 }
5767 @finalents;
5768 }
5770 sub match_globs {
5771 my ($self, $exists, $paths, $globs, $r) = @_;
5773 sub get_dir_check {
5774 my ($self, $exists, $g, $r) = @_;
5776 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5777 $g->{path}->{depth},
5778 $r);
5780 foreach my $de (@dirs) {
5781 my $p = $g->{path}->full_path($de);
5782 next if $exists->{$p};
5783 next if (length $g->{path}->{right} &&
5784 ($self->check_path($p, $r) !=
5785 $SVN::Node::dir));
5786 next unless $p =~ /$g->{path}->{regex}/;
5787 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5788 $g->{ref}->full_path($de), 1);
5789 }
5790 }
5791 foreach my $g (@$globs) {
5792 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5793 if ($path->{action} =~ /^[AR]$/) {
5794 get_dir_check($self, $exists, $g, $r);
5795 }
5796 }
5797 foreach (keys %$paths) {
5798 if (/$g->{path}->{left_regex}/ &&
5799 !/$g->{path}->{regex}/) {
5800 next if $paths->{$_}->{action} !~ /^[AR]$/;
5801 get_dir_check($self, $exists, $g, $r);
5802 }
5803 next unless /$g->{path}->{regex}/;
5804 my $p = $1;
5805 my $pathname = $g->{path}->full_path($p);
5806 next if $exists->{$pathname};
5807 next if ($self->check_path($pathname, $r) !=
5808 $SVN::Node::dir);
5809 $exists->{$pathname} = Git::SVN->init(
5810 $self->{url}, $pathname, undef,
5811 $g->{ref}->full_path($p), 1);
5812 }
5813 my $c = '';
5814 foreach (split m#/#, $g->{path}->{left}) {
5815 $c .= "/$_";
5816 next unless ($paths->{$c} &&
5817 ($paths->{$c}->{action} =~ /^[AR]$/));
5818 get_dir_check($self, $exists, $g, $r);
5819 }
5820 }
5821 values %$exists;
5822 }
5824 sub minimize_url {
5825 my ($self) = @_;
5826 return $self->{url} if ($self->{url} eq $self->{repos_root});
5827 my $url = $self->{repos_root};
5828 my @components = split(m!/!, $self->{svn_path});
5829 my $c = '';
5830 do {
5831 $url .= "/$c" if length $c;
5832 eval {
5833 my $ra = (ref $self)->new($url);
5834 my $latest = $ra->get_latest_revnum;
5835 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5836 };
5837 } while ($@ && ($c = shift @components));
5838 $url;
5839 }
5841 sub can_do_switch {
5842 my $self = shift;
5843 unless (defined $can_do_switch) {
5844 my $pool = SVN::Pool->new;
5845 my $rep = eval {
5846 $self->do_switch(1, '', 0, $self->{url},
5847 SVN::Delta::Editor->new, $pool);
5848 };
5849 if ($@) {
5850 $can_do_switch = 0;
5851 } else {
5852 $rep->abort_report($pool);
5853 $can_do_switch = 1;
5854 }
5855 $pool->clear;
5856 }
5857 $can_do_switch;
5858 }
5860 sub skip_unknown_revs {
5861 my ($err) = @_;
5862 my $errno = $err->apr_err();
5863 # Maybe the branch we're tracking didn't
5864 # exist when the repo started, so it's
5865 # not an error if it doesn't, just continue
5866 #
5867 # Wonderfully consistent library, eh?
5868 # 160013 - svn:// and file://
5869 # 175002 - http(s)://
5870 # 175007 - http(s):// (this repo required authorization, too...)
5871 # More codes may be discovered later...
5872 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5873 my $err_key = $err->expanded_message;
5874 # revision numbers change every time, filter them out
5875 $err_key =~ s/\d+/\0/g;
5876 $err_key = "$errno\0$err_key";
5877 unless ($ignored_err{$err_key}) {
5878 warn "W: Ignoring error from SVN, path probably ",
5879 "does not exist: ($errno): ",
5880 $err->expanded_message,"\n";
5881 warn "W: Do not be alarmed at the above message ",
5882 "git-svn is just searching aggressively for ",
5883 "old history.\n",
5884 "This may take a while on large repositories\n";
5885 $ignored_err{$err_key} = 1;
5886 }
5887 return;
5888 }
5889 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5890 }
5892 package Git::SVN::Log;
5893 use strict;
5894 use warnings;
5895 use POSIX qw/strftime/;
5896 use Time::Local;
5897 use constant commit_log_separator => ('-' x 72) . "\n";
5898 use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5899 %rusers $show_commit $incremental/;
5900 my $l_fmt;
5902 sub cmt_showable {
5903 my ($c) = @_;
5904 return 1 if defined $c->{r};
5906 # big commit message got truncated by the 16k pretty buffer in rev-list
5907 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5908 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5909 @{$c->{l}} = ();
5910 my @log = command(qw/cat-file commit/, $c->{c});
5912 # shift off the headers
5913 shift @log while ($log[0] ne '');
5914 shift @log;
5916 # TODO: make $c->{l} not have a trailing newline in the future
5917 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5919 (undef, $c->{r}, undef) = ::extract_metadata(
5920 (grep(/^git-svn-id: /, @log))[-1]);
5921 }
5922 return defined $c->{r};
5923 }
5925 sub log_use_color {
5926 return $color || Git->repository->get_colorbool('color.diff');
5927 }
5929 sub git_svn_log_cmd {
5930 my ($r_min, $r_max, @args) = @_;
5931 my $head = 'HEAD';
5932 my (@files, @log_opts);
5933 foreach my $x (@args) {
5934 if ($x eq '--' || @files) {
5935 push @files, $x;
5936 } else {
5937 if (::verify_ref("$x^0")) {
5938 $head = $x;
5939 } else {
5940 push @log_opts, $x;
5941 }
5942 }
5943 }
5945 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5946 $gs ||= Git::SVN->_new;
5947 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5948 $gs->refname);
5949 push @cmd, '-r' unless $non_recursive;
5950 push @cmd, qw/--raw --name-status/ if $verbose;
5951 push @cmd, '--color' if log_use_color();
5952 push @cmd, @log_opts;
5953 if (defined $r_max && $r_max == $r_min) {
5954 push @cmd, '--max-count=1';
5955 if (my $c = $gs->rev_map_get($r_max)) {
5956 push @cmd, $c;
5957 }
5958 } elsif (defined $r_max) {
5959 if ($r_max < $r_min) {
5960 ($r_min, $r_max) = ($r_max, $r_min);
5961 }
5962 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5963 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5964 # If there are no commits in the range, both $c_max and $c_min
5965 # will be undefined. If there is at least 1 commit in the
5966 # range, both will be defined.
5967 return () if !defined $c_min || !defined $c_max;
5968 if ($c_min eq $c_max) {
5969 push @cmd, '--max-count=1', $c_min;
5970 } else {
5971 push @cmd, '--boundary', "$c_min..$c_max";
5972 }
5973 }
5974 return (@cmd, @files);
5975 }
5977 # adapted from pager.c
5978 sub config_pager {
5979 if (! -t *STDOUT) {
5980 $ENV{GIT_PAGER_IN_USE} = 'false';
5981 $pager = undef;
5982 return;
5983 }
5984 chomp($pager = command_oneline(qw(var GIT_PAGER)));
5985 if ($pager eq 'cat') {
5986 $pager = undef;
5987 }
5988 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5989 }
5991 sub run_pager {
5992 return unless defined $pager;
5993 pipe my ($rfd, $wfd) or return;
5994 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5995 if (!$pid) {
5996 open STDOUT, '>&', $wfd or
5997 ::fatal "Can't redirect to stdout: $!";
5998 return;
5999 }
6000 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
6001 $ENV{LESS} ||= 'FRSX';
6002 exec $pager or ::fatal "Can't run pager: $! ($pager)";
6003 }
6005 sub format_svn_date {
6006 # some systmes don't handle or mishandle %z, so be creative.
6007 my $t = shift || time;
6008 my $gm = timelocal(gmtime($t));
6009 my $sign = qw( + + - )[ $t <=> $gm ];
6010 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
6011 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
6012 }
6014 sub parse_git_date {
6015 my ($t, $tz) = @_;
6016 # Date::Parse isn't in the standard Perl distro :(
6017 if ($tz =~ s/^\+//) {
6018 $t += tz_to_s_offset($tz);
6019 } elsif ($tz =~ s/^\-//) {
6020 $t -= tz_to_s_offset($tz);
6021 }
6022 return $t;
6023 }
6025 sub set_local_timezone {
6026 if (defined $TZ) {
6027 $ENV{TZ} = $TZ;
6028 } else {
6029 delete $ENV{TZ};
6030 }
6031 }
6033 sub tz_to_s_offset {
6034 my ($tz) = @_;
6035 $tz =~ s/(\d\d)$//;
6036 return ($1 * 60) + ($tz * 3600);
6037 }
6039 sub get_author_info {
6040 my ($dest, $author, $t, $tz) = @_;
6041 $author =~ s/(?:^\s*|\s*$)//g;
6042 $dest->{a_raw} = $author;
6043 my $au;
6044 if ($::_authors) {
6045 $au = $rusers{$author} || undef;
6046 }
6047 if (!$au) {
6048 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
6049 }
6050 $dest->{t} = $t;
6051 $dest->{tz} = $tz;
6052 $dest->{a} = $au;
6053 $dest->{t_utc} = parse_git_date($t, $tz);
6054 }
6056 sub process_commit {
6057 my ($c, $r_min, $r_max, $defer) = @_;
6058 if (defined $r_min && defined $r_max) {
6059 if ($r_min == $c->{r} && $r_min == $r_max) {
6060 show_commit($c);
6061 return 0;
6062 }
6063 return 1 if $r_min == $r_max;
6064 if ($r_min < $r_max) {
6065 # we need to reverse the print order
6066 return 0 if (defined $limit && --$limit < 0);
6067 push @$defer, $c;
6068 return 1;
6069 }
6070 if ($r_min != $r_max) {
6071 return 1 if ($r_min < $c->{r});
6072 return 1 if ($r_max > $c->{r});
6073 }
6074 }
6075 return 0 if (defined $limit && --$limit < 0);
6076 show_commit($c);
6077 return 1;
6078 }
6080 sub show_commit {
6081 my $c = shift;
6082 if ($oneline) {
6083 my $x = "\n";
6084 if (my $l = $c->{l}) {
6085 while ($l->[0] =~ /^\s*$/) { shift @$l }
6086 $x = $l->[0];
6087 }
6088 $l_fmt ||= 'A' . length($c->{r});
6089 print 'r',pack($l_fmt, $c->{r}),' | ';
6090 print "$c->{c} | " if $show_commit;
6091 print $x;
6092 } else {
6093 show_commit_normal($c);
6094 }
6095 }
6097 sub show_commit_changed_paths {
6098 my ($c) = @_;
6099 return unless $c->{changed};
6100 print "Changed paths:\n", @{$c->{changed}};
6101 }
6103 sub show_commit_normal {
6104 my ($c) = @_;
6105 print commit_log_separator, "r$c->{r} | ";
6106 print "$c->{c} | " if $show_commit;
6107 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
6108 my $nr_line = 0;
6110 if (my $l = $c->{l}) {
6111 while ($l->[$#$l] eq "\n" && $#$l > 0
6112 && $l->[($#$l - 1)] eq "\n") {
6113 pop @$l;
6114 }
6115 $nr_line = scalar @$l;
6116 if (!$nr_line) {
6117 print "1 line\n\n\n";
6118 } else {
6119 if ($nr_line == 1) {
6120 $nr_line = '1 line';
6121 } else {
6122 $nr_line .= ' lines';
6123 }
6124 print $nr_line, "\n";
6125 show_commit_changed_paths($c);
6126 print "\n";
6127 print $_ foreach @$l;
6128 }
6129 } else {
6130 print "1 line\n";
6131 show_commit_changed_paths($c);
6132 print "\n";
6134 }
6135 foreach my $x (qw/raw stat diff/) {
6136 if ($c->{$x}) {
6137 print "\n";
6138 print $_ foreach @{$c->{$x}}
6139 }
6140 }
6141 }
6143 sub cmd_show_log {
6144 my (@args) = @_;
6145 my ($r_min, $r_max);
6146 my $r_last = -1; # prevent dupes
6147 set_local_timezone();
6148 if (defined $::_revision) {
6149 if ($::_revision =~ /^(\d+):(\d+)$/) {
6150 ($r_min, $r_max) = ($1, $2);
6151 } elsif ($::_revision =~ /^\d+$/) {
6152 $r_min = $r_max = $::_revision;
6153 } else {
6154 ::fatal "-r$::_revision is not supported, use ",
6155 "standard 'git log' arguments instead";
6156 }
6157 }
6159 config_pager();
6160 @args = git_svn_log_cmd($r_min, $r_max, @args);
6161 if (!@args) {
6162 print commit_log_separator unless $incremental || $oneline;
6163 return;
6164 }
6165 my $log = command_output_pipe(@args);
6166 run_pager();
6167 my (@k, $c, $d, $stat);
6168 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
6169 while (<$log>) {
6170 if (/^${esc_color}commit (?:- )?($::sha1_short)/o) {
6171 my $cmt = $1;
6172 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
6173 $r_last = $c->{r};
6174 process_commit($c, $r_min, $r_max, \@k) or
6175 goto out;
6176 }
6177 $d = undef;
6178 $c = { c => $cmt };
6179 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
6180 get_author_info($c, $1, $2, $3);
6181 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
6182 # ignore
6183 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
6184 push @{$c->{raw}}, $_;
6185 } elsif (/^${esc_color}[ACRMDT]\t/) {
6186 # we could add $SVN->{svn_path} here, but that requires
6187 # remote access at the moment (repo_path_split)...
6188 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
6189 push @{$c->{changed}}, $_;
6190 } elsif (/^${esc_color}diff /o) {
6191 $d = 1;
6192 push @{$c->{diff}}, $_;
6193 } elsif ($d) {
6194 push @{$c->{diff}}, $_;
6195 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
6196 $esc_color*[\+\-]*$esc_color$/x) {
6197 $stat = 1;
6198 push @{$c->{stat}}, $_;
6199 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
6200 push @{$c->{stat}}, $_;
6201 $stat = undef;
6202 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
6203 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
6204 } elsif (s/^${esc_color} //o) {
6205 push @{$c->{l}}, $_;
6206 }
6207 }
6208 if ($c && defined $c->{r} && $c->{r} != $r_last) {
6209 $r_last = $c->{r};
6210 process_commit($c, $r_min, $r_max, \@k);
6211 }
6212 if (@k) {
6213 ($r_min, $r_max) = ($r_max, $r_min);
6214 process_commit($_, $r_min, $r_max) foreach reverse @k;
6215 }
6216 out:
6217 close $log;
6218 print commit_log_separator unless $incremental || $oneline;
6219 }
6221 sub cmd_blame {
6222 my $path = pop;
6224 config_pager();
6225 run_pager();
6227 my ($fh, $ctx, $rev);
6229 if ($_git_format) {
6230 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
6231 while (my $line = <$fh>) {
6232 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
6233 # Uncommitted edits show up as a rev ID of
6234 # all zeros, which we can't look up with
6235 # cmt_metadata
6236 if ($1 !~ /^0+$/) {
6237 (undef, $rev, undef) =
6238 ::cmt_metadata($1);
6239 $rev = '0' if (!$rev);
6240 } else {
6241 $rev = '0';
6242 }
6243 $rev = sprintf('%-10s', $rev);
6244 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
6245 }
6246 print $line;
6247 }
6248 } else {
6249 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
6250 '--', $path);
6251 my ($sha1);
6252 my %authors;
6253 my @buffer;
6254 my %dsha; #distinct sha keys
6256 while (my $line = <$fh>) {
6257 push @buffer, $line;
6258 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
6259 $dsha{$1} = 1;
6260 }
6261 }
6263 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
6265 foreach my $line (@buffer) {
6266 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
6267 $rev = $s2r->{$1};
6268 $rev = '0' if (!$rev)
6269 }
6270 elsif ($line =~ /^author (.*)/) {
6271 $authors{$rev} = $1;
6272 $authors{$rev} =~ s/\s/_/g;
6273 }
6274 elsif ($line =~ /^\t(.*)$/) {
6275 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
6276 }
6277 }
6278 }
6279 command_close_pipe($fh, $ctx);
6280 }
6282 package Git::SVN::Migration;
6283 # these version numbers do NOT correspond to actual version numbers
6284 # of git nor git-svn. They are just relative.
6285 #
6286 # v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
6287 #
6288 # v1 layout: .git/$id/info/url, refs/remotes/$id
6289 #
6290 # v2 layout: .git/svn/$id/info/url, refs/remotes/$id
6291 #
6292 # v3 layout: .git/svn/$id, refs/remotes/$id
6293 # - info/url may remain for backwards compatibility
6294 # - this is what we migrate up to this layout automatically,
6295 # - this will be used by git svn init on single branches
6296 # v3.1 layout (auto migrated):
6297 # - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
6298 # for backwards compatibility
6299 #
6300 # v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
6301 # - this is only created for newly multi-init-ed
6302 # repositories. Similar in spirit to the
6303 # --use-separate-remotes option in git-clone (now default)
6304 # - we do not automatically migrate to this (following
6305 # the example set by core git)
6306 #
6307 # v5 layout: .rev_db.$UUID => .rev_map.$UUID
6308 # - newer, more-efficient format that uses 24-bytes per record
6309 # with no filler space.
6310 # - use xxd -c24 < .rev_map.$UUID to view and debug
6311 # - This is a one-way migration, repositories updated to the
6312 # new format will not be able to use old git-svn without
6313 # rebuilding the .rev_db. Rebuilding the rev_db is not
6314 # possible if noMetadata or useSvmProps are set; but should
6315 # be no problem for users that use the (sensible) defaults.
6316 use strict;
6317 use warnings;
6318 use Carp qw/croak/;
6319 use File::Path qw/mkpath/;
6320 use File::Basename qw/dirname basename/;
6321 use vars qw/$_minimize/;
6323 sub migrate_from_v0 {
6324 my $git_dir = $ENV{GIT_DIR};
6325 return undef unless -d $git_dir;
6326 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
6327 my $migrated = 0;
6328 while (<$fh>) {
6329 chomp;
6330 my ($id, $orig_ref) = ($_, $_);
6331 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
6332 next unless -f "$git_dir/$id/info/url";
6333 my $new_ref = "refs/remotes/$id";
6334 if (::verify_ref("$new_ref^0")) {
6335 print STDERR "W: $orig_ref is probably an old ",
6336 "branch used by an ancient version of ",
6337 "git-svn.\n",
6338 "However, $new_ref also exists.\n",
6339 "We will not be able ",
6340 "to use this branch until this ",
6341 "ambiguity is resolved.\n";
6342 next;
6343 }
6344 print STDERR "Migrating from v0 layout...\n" if !$migrated;
6345 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
6346 command_noisy('update-ref', $new_ref, $orig_ref);
6347 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
6348 $migrated++;
6349 }
6350 command_close_pipe($fh, $ctx);
6351 print STDERR "Done migrating from v0 layout...\n" if $migrated;
6352 $migrated;
6353 }
6355 sub migrate_from_v1 {
6356 my $git_dir = $ENV{GIT_DIR};
6357 my $migrated = 0;
6358 return $migrated unless -d $git_dir;
6359 my $svn_dir = "$git_dir/svn";
6361 # just in case somebody used 'svn' as their $id at some point...
6362 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
6364 print STDERR "Migrating from a git-svn v1 layout...\n";
6365 mkpath([$svn_dir]);
6366 print STDERR "Data from a previous version of git-svn exists, but\n\t",
6367 "$svn_dir\n\t(required for this version ",
6368 "($::VERSION) of git-svn) does not exist.\n";
6369 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
6370 while (<$fh>) {
6371 my $x = $_;
6372 next unless $x =~ s#^refs/remotes/##;
6373 chomp $x;
6374 next unless -f "$git_dir/$x/info/url";
6375 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
6376 next unless $u;
6377 my $dn = dirname("$git_dir/svn/$x");
6378 mkpath([$dn]) unless -d $dn;
6379 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
6380 mkpath(["$git_dir/svn/svn"]);
6381 print STDERR " - $git_dir/$x/info => ",
6382 "$git_dir/svn/$x/info\n";
6383 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
6384 croak "$!: $x";
6385 # don't worry too much about these, they probably
6386 # don't exist with repos this old (save for index,
6387 # and we can easily regenerate that)
6388 foreach my $f (qw/unhandled.log index .rev_db/) {
6389 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
6390 }
6391 } else {
6392 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
6393 rename "$git_dir/$x", "$git_dir/svn/$x" or
6394 croak "$!: $x";
6395 }
6396 $migrated++;
6397 }
6398 command_close_pipe($fh, $ctx);
6399 print STDERR "Done migrating from a git-svn v1 layout\n";
6400 $migrated;
6401 }
6403 sub read_old_urls {
6404 my ($l_map, $pfx, $path) = @_;
6405 my @dir;
6406 foreach (<$path/*>) {
6407 if (-r "$_/info/url") {
6408 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
6409 my $ref_id = $pfx . basename $_;
6410 my $url = ::file_to_s("$_/info/url");
6411 $l_map->{$ref_id} = $url;
6412 } elsif (-d $_) {
6413 push @dir, $_;
6414 }
6415 }
6416 foreach (@dir) {
6417 my $x = $_;
6418 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
6419 read_old_urls($l_map, $x, $_);
6420 }
6421 }
6423 sub migrate_from_v2 {
6424 my @cfg = command(qw/config -l/);
6425 return if grep /^svn-remote\..+\.url=/, @cfg;
6426 my %l_map;
6427 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
6428 my $migrated = 0;
6430 foreach my $ref_id (sort keys %l_map) {
6431 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
6432 if ($@) {
6433 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
6434 }
6435 $migrated++;
6436 }
6437 $migrated;
6438 }
6440 sub minimize_connections {
6441 my $r = Git::SVN::read_all_remotes();
6442 my $new_urls = {};
6443 my $root_repos = {};
6444 foreach my $repo_id (keys %$r) {
6445 my $url = $r->{$repo_id}->{url} or next;
6446 my $fetch = $r->{$repo_id}->{fetch} or next;
6447 my $ra = Git::SVN::Ra->new($url);
6449 # skip existing cases where we already connect to the root
6450 if (($ra->{url} eq $ra->{repos_root}) ||
6451 ($ra->{repos_root} eq $repo_id)) {
6452 $root_repos->{$ra->{url}} = $repo_id;
6453 next;
6454 }
6456 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
6457 my $root_path = $ra->{url};
6458 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
6459 foreach my $path (keys %$fetch) {
6460 my $ref_id = $fetch->{$path};
6461 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
6463 # make sure we can read when connecting to
6464 # a higher level of a repository
6465 my ($last_rev, undef) = $gs->last_rev_commit;
6466 if (!defined $last_rev) {
6467 $last_rev = eval {
6468 $root_ra->get_latest_revnum;
6469 };
6470 next if $@;
6471 }
6472 my $new = $root_path;
6473 $new .= length $path ? "/$path" : '';
6474 eval {
6475 $root_ra->get_log([$new], $last_rev, $last_rev,
6476 0, 0, 1, sub { });
6477 };
6478 next if $@;
6479 $new_urls->{$ra->{repos_root}}->{$new} =
6480 { ref_id => $ref_id,
6481 old_repo_id => $repo_id,
6482 old_path => $path };
6483 }
6484 }
6486 my @emptied;
6487 foreach my $url (keys %$new_urls) {
6488 # see if we can re-use an existing [svn-remote "repo_id"]
6489 # instead of creating a(n ugly) new section:
6490 my $repo_id = $root_repos->{$url} || $url;
6492 my $fetch = $new_urls->{$url};
6493 foreach my $path (keys %$fetch) {
6494 my $x = $fetch->{$path};
6495 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
6496 my $pfx = "svn-remote.$x->{old_repo_id}";
6498 my $old_fetch = quotemeta("$x->{old_path}:".
6499 "$x->{ref_id}");
6500 command_noisy(qw/config --unset/,
6501 "$pfx.fetch", '^'. $old_fetch . '$');
6502 delete $r->{$x->{old_repo_id}}->
6503 {fetch}->{$x->{old_path}};
6504 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6505 command_noisy(qw/config --unset/,
6506 "$pfx.url");
6507 push @emptied, $x->{old_repo_id}
6508 }
6509 }
6510 }
6511 if (@emptied) {
6512 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6513 print STDERR <<EOF;
6514 The following [svn-remote] sections in your config file ($file) are empty
6515 and can be safely removed:
6516 EOF
6517 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6518 }
6519 }
6521 sub migration_check {
6522 migrate_from_v0();
6523 migrate_from_v1();
6524 migrate_from_v2();
6525 minimize_connections() if $_minimize;
6526 }
6528 package Git::IndexInfo;
6529 use strict;
6530 use warnings;
6531 use Git qw/command_input_pipe command_close_pipe/;
6533 sub new {
6534 my ($class) = @_;
6535 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6536 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6537 }
6539 sub remove {
6540 my ($self, $path) = @_;
6541 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6542 return ++$self->{nr};
6543 }
6544 undef;
6545 }
6547 sub update {
6548 my ($self, $mode, $hash, $path) = @_;
6549 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6550 return ++$self->{nr};
6551 }
6552 undef;
6553 }
6555 sub DESTROY {
6556 my ($self) = @_;
6557 command_close_pipe($self->{gui}, $self->{ctx});
6558 }
6560 package Git::SVN::GlobSpec;
6561 use strict;
6562 use warnings;
6564 sub new {
6565 my ($class, $glob, $pattern_ok) = @_;
6566 my $re = $glob;
6567 $re =~ s!/+$!!g; # no need for trailing slashes
6568 my (@left, @right, @patterns);
6569 my $state = "left";
6570 my $die_msg = "Only one set of wildcard directories " .
6571 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6572 for my $part (split(m|/|, $glob)) {
6573 if ($part =~ /\*/ && $part ne "*") {
6574 die "Invalid pattern in '$glob': $part\n";
6575 } elsif ($pattern_ok && $part =~ /[{}]/ &&
6576 $part !~ /^\{[^{}]+\}/) {
6577 die "Invalid pattern in '$glob': $part\n";
6578 }
6579 if ($part eq "*") {
6580 die $die_msg if $state eq "right";
6581 $state = "pattern";
6582 push(@patterns, "[^/]*");
6583 } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6584 die $die_msg if $state eq "right";
6585 $state = "pattern";
6586 my $p = quotemeta($1);
6587 $p =~ s/\\,/|/g;
6588 push(@patterns, "(?:$p)");
6589 } else {
6590 if ($state eq "left") {
6591 push(@left, $part);
6592 } else {
6593 push(@right, $part);
6594 $state = "right";
6595 }
6596 }
6597 }
6598 my $depth = @patterns;
6599 if ($depth == 0) {
6600 die "One '*' is needed in glob: '$glob'\n";
6601 }
6602 my $left = join('/', @left);
6603 my $right = join('/', @right);
6604 $re = join('/', @patterns);
6605 $re = join('\/',
6606 grep(length, quotemeta($left), "($re)", quotemeta($right)));
6607 my $left_re = qr/^\/\Q$left\E(\/|$)/;
6608 bless { left => $left, right => $right, left_regex => $left_re,
6609 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6610 }
6612 sub full_path {
6613 my ($self, $path) = @_;
6614 return (length $self->{left} ? "$self->{left}/" : '') .
6615 $path . (length $self->{right} ? "/$self->{right}" : '');
6616 }
6618 __END__
6620 Data structures:
6623 $remotes = { # returned by read_all_remotes()
6624 'svn' => {
6625 # svn-remote.svn.url=https://svn.musicpd.org
6626 url => 'https://svn.musicpd.org',
6627 # svn-remote.svn.fetch=mpd/trunk:trunk
6628 fetch => {
6629 'mpd/trunk' => 'trunk',
6630 },
6631 # svn-remote.svn.tags=mpd/tags/*:tags/*
6632 tags => {
6633 path => {
6634 left => 'mpd/tags',
6635 right => '',
6636 regex => qr!mpd/tags/([^/]+)$!,
6637 glob => 'tags/*',
6638 },
6639 ref => {
6640 left => 'tags',
6641 right => '',
6642 regex => qr!tags/([^/]+)$!,
6643 glob => 'tags/*',
6644 },
6645 }
6646 }
6647 };
6649 $log_entry hashref as returned by libsvn_log_entry()
6650 {
6651 log => 'whitespace-formatted log entry
6652 ', # trailing newline is preserved
6653 revision => '8', # integer
6654 date => '2004-02-24T17:01:44.108345Z', # commit date
6655 author => 'committer name'
6656 };
6659 # this is generated by generate_diff();
6660 @mods = array of diff-index line hashes, each element represents one line
6661 of diff-index output
6663 diff-index line ($m hash)
6664 {
6665 mode_a => first column of diff-index output, no leading ':',
6666 mode_b => second column of diff-index output,
6667 sha1_b => sha1sum of the final blob,
6668 chg => change type [MCRADT],
6669 file_a => original file name of a file (iff chg is 'C' or 'R')
6670 file_b => new/current file name of a file (any chg)
6671 }
6672 ;
6674 # retval of read_url_paths{,_all}();
6675 $l_map = {
6676 # repository root url
6677 'https://svn.musicpd.org' => {
6678 # repository path # GIT_SVN_ID
6679 'mpd/trunk' => 'trunk',
6680 'mpd/tags/0.11.5' => 'tags/0.11.5',
6681 },
6682 }
6684 Notes:
6685 I don't trust the each() function on unless I created %hash myself
6686 because the internal iterator may not have started at base.