1#!/usr/bin/env perl
2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3# License: GPL v2 or later
4use warnings;
5use strict;
6use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision $_repository
8 $_q $_authors $_authors_prog %users/;
9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10$VERSION = '@@GIT_VERSION@@';
11# From which subdir have we been invoked?
13my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15} || '';
16my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
18$ENV{GIT_DIR} ||= '.git';
19$Git::SVN::default_repo_id = 'svn';
20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
21$Git::SVN::Ra::_log_window_size = 100;
22$Git::SVN::_minimize_url = 'unset';
23if (! exists $ENV{SVN_SSH}) {
25 if (exists $ENV{GIT_SSH}) {
26 $ENV{SVN_SSH} = $ENV{GIT_SSH};
27 if ($^O eq 'msys') {
28 $ENV{SVN_SSH} =~ s/\\/\\\\/g;
29 $ENV{SVN_SSH} =~ s/(.*)/"$1"/;
30 }
31 }
32}
33$Git::SVN::Log::TZ = $ENV{TZ};
35$ENV{TZ} = 'UTC';
36$| = 1; # unbuffer STDOUT
37sub fatal (@) { print STDERR "@_\n"; exit 1 }
39sub _req_svn {
40 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
41 require SVN::Ra;
42 require SVN::Delta;
43 if ($SVN::Core::VERSION lt '1.1.0') {
44 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
45 }
46}
47my $can_compress = eval { require Compress::Zlib; 1};
48push @Git::SVN::Ra::ISA, 'SVN::Ra';
49push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
50push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
51use Carp qw/croak/;
52use Digest::MD5;
53use IO::File qw//;
54use File::Basename qw/dirname basename/;
55use File::Path qw/mkpath/;
56use File::Spec;
57use File::Find;
58use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
59use IPC::Open3;
60use Git;
61BEGIN {
63 # import functions from Git into our packages, en masse
64 no strict 'refs';
65 foreach (qw/command command_oneline command_noisy command_output_pipe
66 command_input_pipe command_close_pipe
67 command_bidi_pipe command_close_bidi_pipe/) {
68 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
69 Git::SVN::Migration Git::SVN::Log Git::SVN),
70 __PACKAGE__) {
71 *{"${package}::$_"} = \&{"Git::$_"};
72 }
73 }
74}
75my ($SVN);
77$sha1 = qr/[a-f\d]{40}/;
79$sha1_short = qr/[a-f\d]{4,40}/;
80my ($_stdin, $_help, $_edit,
81 $_message, $_file, $_branch_dest,
82 $_template, $_shared,
83 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
84 $_merge, $_strategy, $_dry_run, $_local,
85 $_prefix, $_no_checkout, $_url, $_verbose,
86 $_git_format, $_commit_url, $_tag);
87$Git::SVN::_follow_parent = 1;
88$_q ||= 0;
89my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
90 'config-dir=s' => \$Git::SVN::Ra::config_dir,
91 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
92 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
93my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
94 'authors-file|A=s' => \$_authors,
95 'authors-prog=s' => \$_authors_prog,
96 'repack:i' => \$Git::SVN::_repack,
97 'noMetadata' => \$Git::SVN::_no_metadata,
98 'useSvmProps' => \$Git::SVN::_use_svm_props,
99 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
100 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
101 'no-checkout' => \$_no_checkout,
102 'quiet|q+' => \$_q,
103 'repack-flags|repack-args|repack-opts=s' =>
104 \$Git::SVN::_repack_flags,
105 'use-log-author' => \$Git::SVN::_use_log_author,
106 'add-author-from' => \$Git::SVN::_add_author_from,
107 'localtime' => \$Git::SVN::_localtime,
108 %remote_opts );
109my ($_trunk, @_tags, @_branches, $_stdlayout);
111my %icv;
112my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
113 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
114 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
115 'stdlayout|s' => \$_stdlayout,
116 'minimize-url|m!' => \$Git::SVN::_minimize_url,
117 'no-metadata' => sub { $icv{noMetadata} = 1 },
118 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
119 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
120 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
121 'rewrite-uuid=s' => sub { $icv{rewriteUUID} = $_[1] },
122 %remote_opts );
123my %cmt_opts = ( 'edit|e' => \$_edit,
124 'rmdir' => \$SVN::Git::Editor::_rmdir,
125 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
126 'l=i' => \$SVN::Git::Editor::_rename_limit,
127 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
128);
129my %cmd = (
131 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
132 { 'revision|r=s' => \$_revision,
133 'fetch-all|all' => \$_fetch_all,
134 'parent|p' => \$_fetch_parent,
135 %fc_opts } ],
136 clone => [ \&cmd_clone, "Initialize and fetch revisions",
137 { 'revision|r=s' => \$_revision,
138 %fc_opts, %init_opts } ],
139 init => [ \&cmd_init, "Initialize a repo for tracking" .
140 " (requires URL argument)",
141 \%init_opts ],
142 'multi-init' => [ \&cmd_multi_init,
143 "Deprecated alias for ".
144 "'$0 init -T<trunk> -b<branches> -t<tags>'",
145 \%init_opts ],
146 dcommit => [ \&cmd_dcommit,
147 'Commit several diffs to merge with upstream',
148 { 'merge|m|M' => \$_merge,
149 'strategy|s=s' => \$_strategy,
150 'verbose|v' => \$_verbose,
151 'dry-run|n' => \$_dry_run,
152 'fetch-all|all' => \$_fetch_all,
153 'commit-url=s' => \$_commit_url,
154 'revision|r=i' => \$_revision,
155 'no-rebase' => \$_no_rebase,
156 %cmt_opts, %fc_opts } ],
157 branch => [ \&cmd_branch,
158 'Create a branch in the SVN repository',
159 { 'message|m=s' => \$_message,
160 'destination|d=s' => \$_branch_dest,
161 'dry-run|n' => \$_dry_run,
162 'tag|t' => \$_tag,
163 'username=s' => \$Git::SVN::Prompt::_username,
164 'commit-url=s' => \$_commit_url } ],
165 tag => [ sub { $_tag = 1; cmd_branch(@_) },
166 'Create a tag in the SVN repository',
167 { 'message|m=s' => \$_message,
168 'destination|d=s' => \$_branch_dest,
169 'dry-run|n' => \$_dry_run,
170 'username=s' => \$Git::SVN::Prompt::_username,
171 'commit-url=s' => \$_commit_url } ],
172 'set-tree' => [ \&cmd_set_tree,
173 "Set an SVN repository to a git tree-ish",
174 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
175 'create-ignore' => [ \&cmd_create_ignore,
176 'Create a .gitignore per svn:ignore',
177 { 'revision|r=i' => \$_revision
178 } ],
179 'mkdirs' => [ \&cmd_mkdirs ,
180 "recreate empty directories after a checkout",
181 { 'revision|r=i' => \$_revision } ],
182 'propget' => [ \&cmd_propget,
183 'Print the value of a property on a file or directory',
184 { 'revision|r=i' => \$_revision } ],
185 'proplist' => [ \&cmd_proplist,
186 'List all properties of a file or directory',
187 { 'revision|r=i' => \$_revision } ],
188 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
189 { 'revision|r=i' => \$_revision
190 } ],
191 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
192 { 'revision|r=i' => \$_revision
193 } ],
194 'multi-fetch' => [ \&cmd_multi_fetch,
195 "Deprecated alias for $0 fetch --all",
196 { 'revision|r=s' => \$_revision, %fc_opts } ],
197 'migrate' => [ sub { },
198 # no-op, we automatically run this anyways,
199 'Migrate configuration/metadata/layout from
200 previous versions of git-svn',
201 { 'minimize' => \$Git::SVN::Migration::_minimize,
202 %remote_opts } ],
203 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
204 { 'limit=i' => \$Git::SVN::Log::limit,
205 'revision|r=s' => \$_revision,
206 'verbose|v' => \$Git::SVN::Log::verbose,
207 'incremental' => \$Git::SVN::Log::incremental,
208 'oneline' => \$Git::SVN::Log::oneline,
209 'show-commit' => \$Git::SVN::Log::show_commit,
210 'non-recursive' => \$Git::SVN::Log::non_recursive,
211 'authors-file|A=s' => \$_authors,
212 'color' => \$Git::SVN::Log::color,
213 'pager=s' => \$Git::SVN::Log::pager
214 } ],
215 'find-rev' => [ \&cmd_find_rev,
216 "Translate between SVN revision numbers and tree-ish",
217 {} ],
218 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
219 { 'merge|m|M' => \$_merge,
220 'verbose|v' => \$_verbose,
221 'strategy|s=s' => \$_strategy,
222 'local|l' => \$_local,
223 'fetch-all|all' => \$_fetch_all,
224 'dry-run|n' => \$_dry_run,
225 %fc_opts } ],
226 'commit-diff' => [ \&cmd_commit_diff,
227 'Commit a diff between two trees',
228 { 'message|m=s' => \$_message,
229 'file|F=s' => \$_file,
230 'revision|r=s' => \$_revision,
231 %cmt_opts } ],
232 'info' => [ \&cmd_info,
233 "Show info about the latest SVN revision
234 on the current branch",
235 { 'url' => \$_url, } ],
236 'blame' => [ \&Git::SVN::Log::cmd_blame,
237 "Show what revision and author last modified each line of a file",
238 { 'git-format' => \$_git_format } ],
239 'reset' => [ \&cmd_reset,
240 "Undo fetches back to the specified SVN revision",
241 { 'revision|r=s' => \$_revision,
242 'parent|p' => \$_fetch_parent } ],
243 'gc' => [ \&cmd_gc,
244 "Compress unhandled.log files in .git/svn and remove " .
245 "index files in .git/svn",
246 {} ],
247);
248my $cmd;
250for (my $i = 0; $i < @ARGV; $i++) {
251 if (defined $cmd{$ARGV[$i]}) {
252 $cmd = $ARGV[$i];
253 splice @ARGV, $i, 1;
254 last;
255 } elsif ($ARGV[$i] eq 'help') {
256 $cmd = $ARGV[$i+1];
257 usage(0);
258 }
259};
260# make sure we're always running at the top-level working directory
262unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
263 unless (-d $ENV{GIT_DIR}) {
264 if ($git_dir_user_set) {
265 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
266 "but it is not a directory\n";
267 }
268 my $git_dir = delete $ENV{GIT_DIR};
269 my $cdup = undef;
270 git_cmd_try {
271 $cdup = command_oneline(qw/rev-parse --show-cdup/);
272 $git_dir = '.' unless ($cdup);
273 chomp $cdup if ($cdup);
274 $cdup = "." unless ($cdup && length $cdup);
275 } "Already at toplevel, but $git_dir not found\n";
276 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
277 unless (-d $git_dir) {
278 die "$git_dir still not found after going to ",
279 "'$cdup'\n";
280 }
281 $ENV{GIT_DIR} = $git_dir;
282 }
283 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
284}
285my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
287read_git_config(\%opts);
289if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
290 Getopt::Long::Configure('pass_through');
291}
292my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
293 'minimize-connections' => \$Git::SVN::Migration::_minimize,
294 'id|i=s' => \$Git::SVN::default_ref_id,
295 'svn-remote|remote|R=s' => sub {
296 $Git::SVN::no_reuse_existing = 1;
297 $Git::SVN::default_repo_id = $_[1] });
298exit 1 if (!$rv && $cmd && $cmd ne 'log');
299usage(0) if $_help;
301version() if $_version;
302usage(1) unless defined $cmd;
303load_authors() if $_authors;
304if (defined $_authors_prog) {
305 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
306}
307unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
309 Git::SVN::Migration::migration_check();
310}
311Git::SVN::init_vars();
312eval {
313 Git::SVN::verify_remotes_sanity();
314 $cmd{$cmd}->[0]->(@ARGV);
315};
316fatal $@ if $@;
317post_fetch_checkout();
318exit 0;
319####################### primary functions ######################
321sub usage {
322 my $exit = shift || 0;
323 my $fd = $exit ? \*STDERR : \*STDOUT;
324 print $fd <<"";
325git-svn - bidirectional operations between a single Subversion tree and git
326Usage: git svn <command> [options] [arguments]\n
327 print $fd "Available commands:\n" unless $cmd;
329 foreach (sort keys %cmd) {
331 next if $cmd && $cmd ne $_;
332 next if /^multi-/; # don't show deprecated commands
333 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
334 foreach (sort keys %{$cmd{$_}->[2]}) {
335 # mixed-case options are for .git/config only
336 next if /[A-Z]/ && /^[a-z]+$/i;
337 # prints out arguments as they should be passed:
338 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
339 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
340 "--$_" : "-$_" }
341 split /\|/,$_)," $x\n";
342 }
343 }
344 print $fd <<"";
345\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
346arbitrary identifier if you're tracking multiple SVN branches/repositories in
347one git repository and want to keep them separate. See git-svn(1) for more
348information.
349 exit $exit;
351}
352sub version {
354 ::_req_svn();
355 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
356 exit 0;
357}
358sub do_git_init_db {
360 unless (-d $ENV{GIT_DIR}) {
361 my @init_db = ('init');
362 push @init_db, "--template=$_template" if defined $_template;
363 if (defined $_shared) {
364 if ($_shared =~ /[a-z]/) {
365 push @init_db, "--shared=$_shared";
366 } else {
367 push @init_db, "--shared";
368 }
369 }
370 command_noisy(@init_db);
371 $_repository = Git->repository(Repository => ".git");
372 }
373 my $set;
374 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
375 foreach my $i (keys %icv) {
376 die "'$set' and '$i' cannot both be set\n" if $set;
377 next unless defined $icv{$i};
378 command_noisy('config', "$pfx.$i", $icv{$i});
379 $set = $i;
380 }
381 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
382 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
383 if defined $$ignore_regex;
384}
385sub init_subdir {
387 my $repo_path = shift or return;
388 mkpath([$repo_path]) unless -d $repo_path;
389 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
390 $ENV{GIT_DIR} = '.git';
391 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
392}
393sub cmd_clone {
395 my ($url, $path) = @_;
396 if (!defined $path &&
397 (defined $_trunk || @_branches || @_tags ||
398 defined $_stdlayout) &&
399 $url !~ m#^[a-z\+]+://#) {
400 $path = $url;
401 }
402 $path = basename($url) if !defined $path || !length $path;
403 my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
404 cmd_init($url, $path);
405 command_oneline('config', 'svn.authorsfile', $authors_absolute)
406 if $_authors;
407 Git::SVN::fetch_all($Git::SVN::default_repo_id);
408}
409sub cmd_init {
411 if (defined $_stdlayout) {
412 $_trunk = 'trunk' if (!defined $_trunk);
413 @_tags = 'tags' if (! @_tags);
414 @_branches = 'branches' if (! @_branches);
415 }
416 if (defined $_trunk || @_branches || @_tags) {
417 return cmd_multi_init(@_);
418 }
419 my $url = shift or die "SVN repository location required ",
420 "as a command-line argument\n";
421 $url = canonicalize_url($url);
422 init_subdir(@_);
423 do_git_init_db();
424 if ($Git::SVN::_minimize_url eq 'unset') {
426 $Git::SVN::_minimize_url = 0;
427 }
428 Git::SVN->init($url);
430}
431sub cmd_fetch {
433 if (grep /^\d+=./, @_) {
434 die "'<rev>=<commit>' fetch arguments are ",
435 "no longer supported.\n";
436 }
437 my ($remote) = @_;
438 if (@_ > 1) {
439 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
440 }
441 $Git::SVN::no_reuse_existing = undef;
442 if ($_fetch_parent) {
443 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
444 unless ($gs) {
445 die "Unable to determine upstream SVN information from ",
446 "working tree history\n";
447 }
448 # just fetch, don't checkout.
449 $_no_checkout = 'true';
450 $_fetch_all ? $gs->fetch_all : $gs->fetch;
451 } elsif ($_fetch_all) {
452 cmd_multi_fetch();
453 } else {
454 $remote ||= $Git::SVN::default_repo_id;
455 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
456 }
457}
458sub cmd_set_tree {
460 my (@commits) = @_;
461 if ($_stdin || !@commits) {
462 print "Reading from stdin...\n";
463 @commits = ();
464 while (<STDIN>) {
465 if (/\b($sha1_short)\b/o) {
466 unshift @commits, $1;
467 }
468 }
469 }
470 my @revs;
471 foreach my $c (@commits) {
472 my @tmp = command('rev-parse',$c);
473 if (scalar @tmp == 1) {
474 push @revs, $tmp[0];
475 } elsif (scalar @tmp > 1) {
476 push @revs, reverse(command('rev-list',@tmp));
477 } else {
478 fatal "Failed to rev-parse $c";
479 }
480 }
481 my $gs = Git::SVN->new;
482 my ($r_last, $cmt_last) = $gs->last_rev_commit;
483 $gs->fetch;
484 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
485 fatal "There are new revisions that were fetched ",
486 "and need to be merged (or acknowledged) ",
487 "before committing.\nlast rev: $r_last\n",
488 " current: $gs->{last_rev}";
489 }
490 $gs->set_tree($_) foreach @revs;
491 print "Done committing ",scalar @revs," revisions to SVN\n";
492 unlink $gs->{index};
493}
494sub cmd_dcommit {
496 my $head = shift;
497 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
498 'Cannot dcommit with a dirty index. Commit your changes first, '
499 . "or stash them with `git stash'.\n";
500 $head ||= 'HEAD';
501 my $old_head;
503 if ($head ne 'HEAD') {
504 $old_head = eval {
505 command_oneline([qw/symbolic-ref -q HEAD/])
506 };
507 if ($old_head) {
508 $old_head =~ s{^refs/heads/}{};
509 } else {
510 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
511 }
512 command(['checkout', $head], STDERR => 0);
513 }
514 my @refs;
516 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
517 unless ($gs) {
518 die "Unable to determine upstream SVN information from ",
519 "$head history.\nPerhaps the repository is empty.";
520 }
521 if (defined $_commit_url) {
523 $url = $_commit_url;
524 } else {
525 $url = eval { command_oneline('config', '--get',
526 "svn-remote.$gs->{repo_id}.commiturl") };
527 if (!$url) {
528 $url = $gs->full_url
529 }
530 }
531 my $last_rev = $_revision if defined $_revision;
533 if ($url) {
534 print "Committing to $url ...\n";
535 }
536 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
537 if ($_no_rebase && scalar(@$linear_refs) > 1) {
538 warn "Attempting to commit more than one change while ",
539 "--no-rebase is enabled.\n",
540 "If these changes depend on each other, re-running ",
541 "without --no-rebase may be required."
542 }
543 my $expect_url = $url;
544 Git::SVN::remove_username($expect_url);
545 while (1) {
546 my $d = shift @$linear_refs or last;
547 unless (defined $last_rev) {
548 (undef, $last_rev, undef) = cmt_metadata("$d~1");
549 unless (defined $last_rev) {
550 fatal "Unable to extract revision information ",
551 "from commit $d~1";
552 }
553 }
554 if ($_dry_run) {
555 print "diff-tree $d~1 $d\n";
556 } else {
557 my $cmt_rev;
558 my %ed_opts = ( r => $last_rev,
559 log => get_commit_entry($d)->{log},
560 ra => Git::SVN::Ra->new($url),
561 config => SVN::Core::config_get_config(
562 $Git::SVN::Ra::config_dir
563 ),
564 tree_a => "$d~1",
565 tree_b => $d,
566 editor_cb => sub {
567 print "Committed r$_[0]\n";
568 $cmt_rev = $_[0];
569 },
570 svn_path => '');
571 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
572 print "No changes\n$d~1 == $d\n";
573 } elsif ($parents->{$d} && @{$parents->{$d}}) {
574 $gs->{inject_parents_dcommit}->{$cmt_rev} =
575 $parents->{$d};
576 }
577 $_fetch_all ? $gs->fetch_all : $gs->fetch;
578 $last_rev = $cmt_rev;
579 next if $_no_rebase;
580 # we always want to rebase against the current HEAD,
582 # not any head that was passed to us
583 my @diff = command('diff-tree', $d,
584 $gs->refname, '--');
585 my @finish;
586 if (@diff) {
587 @finish = rebase_cmd();
588 print STDERR "W: $d and ", $gs->refname,
589 " differ, using @finish:\n",
590 join("\n", @diff), "\n";
591 } else {
592 print "No changes between current HEAD and ",
593 $gs->refname,
594 "\nResetting to the latest ",
595 $gs->refname, "\n";
596 @finish = qw/reset --mixed/;
597 }
598 command_noisy(@finish, $gs->refname);
599 if (@diff) {
600 @refs = ();
601 my ($url_, $rev_, $uuid_, $gs_) =
602 working_head_info('HEAD', \@refs);
603 my ($linear_refs_, $parents_) =
604 linearize_history($gs_, \@refs);
605 if (scalar(@$linear_refs) !=
606 scalar(@$linear_refs_)) {
607 fatal "# of revisions changed ",
608 "\nbefore:\n",
609 join("\n", @$linear_refs),
610 "\n\nafter:\n",
611 join("\n", @$linear_refs_), "\n",
612 'If you are attempting to commit ',
613 "merges, try running:\n\t",
614 'git rebase --interactive',
615 '--preserve-merges ',
616 $gs->refname,
617 "\nBefore dcommitting";
618 }
619 if ($url_ ne $expect_url) {
620 if ($url_ eq $gs->metadata_url) {
621 print
622 "Accepting rewritten URL:",
623 " $url_\n";
624 } else {
625 fatal
626 "URL mismatch after rebase:",
627 " $url_ != $expect_url";
628 }
629 }
630 if ($uuid_ ne $uuid) {
631 fatal "uuid mismatch after rebase: ",
632 "$uuid_ != $uuid";
633 }
634 # remap parents
635 my (%p, @l, $i);
636 for ($i = 0; $i < scalar @$linear_refs; $i++) {
637 my $new = $linear_refs_->[$i] or next;
638 $p{$new} =
639 $parents->{$linear_refs->[$i]};
640 push @l, $new;
641 }
642 $parents = \%p;
643 $linear_refs = \@l;
644 }
645 }
646 }
647 if ($old_head) {
649 my $new_head = command_oneline(qw/rev-parse HEAD/);
650 my $new_is_symbolic = eval {
651 command_oneline(qw/symbolic-ref -q HEAD/);
652 };
653 if ($new_is_symbolic) {
654 print "dcommitted the branch ", $head, "\n";
655 } else {
656 print "dcommitted on a detached HEAD because you gave ",
657 "a revision argument.\n",
658 "The rewritten commit is: ", $new_head, "\n";
659 }
660 command(['checkout', $old_head], STDERR => 0);
661 }
662 unlink $gs->{index};
664}
665sub cmd_branch {
667 my ($branch_name, $head) = @_;
668 unless (defined $branch_name && length $branch_name) {
670 die(($_tag ? "tag" : "branch") . " name required\n");
671 }
672 $head ||= 'HEAD';
673 my (undef, $rev, undef, $gs) = working_head_info($head);
675 my $src = $gs->full_url;
676 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
678 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
679 my $glob;
680 if ($#{$allglobs} == 0) {
681 $glob = $allglobs->[0];
682 } else {
683 unless(defined $_branch_dest) {
684 die "Multiple ",
685 $_tag ? "tag" : "branch",
686 " paths defined for Subversion repository.\n",
687 "You must specify where you want to create the ",
688 $_tag ? "tag" : "branch",
689 " with the --destination argument.\n";
690 }
691 foreach my $g (@{$allglobs}) {
692 # SVN::Git::Editor could probably be moved to Git.pm..
693 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
694 if ($_branch_dest =~ /$re/) {
695 $glob = $g;
696 last;
697 }
698 }
699 unless (defined $glob) {
700 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
701 foreach my $g (@{$allglobs}) {
702 $g->{path}->{left} =~ /$dest_re/ or next;
703 if (defined $glob) {
704 die "Ambiguous destination: ",
705 $_branch_dest, "\nmatches both '",
706 $glob->{path}->{left}, "' and '",
707 $g->{path}->{left}, "'\n";
708 }
709 $glob = $g;
710 }
711 unless (defined $glob) {
712 die "Unknown ",
713 $_tag ? "tag" : "branch",
714 " destination $_branch_dest\n";
715 }
716 }
717 }
718 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
719 my $url;
720 if (defined $_commit_url) {
721 $url = $_commit_url;
722 } else {
723 $url = eval { command_oneline('config', '--get',
724 "svn-remote.$gs->{repo_id}.commiturl") };
725 if (!$url) {
726 $url = $remote->{url};
727 }
728 }
729 my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
730 if ($dst =~ /^https:/ && $src =~ /^http:/) {
732 $src=~s/^http:/https:/;
733 }
734 ::_req_svn();
736 my $ctx = SVN::Client->new(
738 auth => Git::SVN::Ra::_auth_providers(),
739 log_msg => sub {
740 ${ $_[0] } = defined $_message
741 ? $_message
742 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
743 . $branch_name;
744 },
745 );
746 eval {
748 $ctx->ls($dst, 'HEAD', 0);
749 } and die "branch ${branch_name} already exists\n";
750 print "Copying ${src} at r${rev} to ${dst}...\n";
752 $ctx->copy($src, $rev, $dst)
753 unless $_dry_run;
754 $gs->fetch_all;
756}
757sub cmd_find_rev {
759 my $revision_or_hash = shift or die "SVN or git revision required ",
760 "as a command-line argument\n";
761 my $result;
762 if ($revision_or_hash =~ /^r\d+$/) {
763 my $head = shift;
764 $head ||= 'HEAD';
765 my @refs;
766 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
767 unless ($gs) {
768 die "Unable to determine upstream SVN information from ",
769 "$head history\n";
770 }
771 my $desired_revision = substr($revision_or_hash, 1);
772 $result = $gs->rev_map_get($desired_revision, $uuid);
773 } else {
774 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
775 $result = $rev;
776 }
777 print "$result\n" if $result;
778}
779sub cmd_rebase {
781 command_noisy(qw/update-index --refresh/);
782 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
783 unless ($gs) {
784 die "Unable to determine upstream SVN information from ",
785 "working tree history\n";
786 }
787 if ($_dry_run) {
788 print "Remote Branch: " . $gs->refname . "\n";
789 print "SVN URL: " . $url . "\n";
790 return;
791 }
792 if (command(qw/diff-index HEAD --/)) {
793 print STDERR "Cannot rebase with uncommited changes:\n";
794 command_noisy('status');
795 exit 1;
796 }
797 unless ($_local) {
798 # rebase will checkout for us, so no need to do it explicitly
799 $_no_checkout = 'true';
800 $_fetch_all ? $gs->fetch_all : $gs->fetch;
801 }
802 command_noisy(rebase_cmd(), $gs->refname);
803 $gs->mkemptydirs;
804}
805sub cmd_show_ignore {
807 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
808 $gs ||= Git::SVN->new;
809 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
810 $gs->prop_walk($gs->{path}, $r, sub {
811 my ($gs, $path, $props) = @_;
812 print STDOUT "\n# $path\n";
813 my $s = $props->{'svn:ignore'} or return;
814 $s =~ s/[\r\n]+/\n/g;
815 $s =~ s/^\n+//;
816 chomp $s;
817 $s =~ s#^#$path#gm;
818 print STDOUT "$s\n";
819 });
820}
821sub cmd_show_externals {
823 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
824 $gs ||= Git::SVN->new;
825 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
826 $gs->prop_walk($gs->{path}, $r, sub {
827 my ($gs, $path, $props) = @_;
828 print STDOUT "\n# $path\n";
829 my $s = $props->{'svn:externals'} or return;
830 $s =~ s/[\r\n]+/\n/g;
831 chomp $s;
832 $s =~ s#^#$path#gm;
833 print STDOUT "$s\n";
834 });
835}
836sub cmd_create_ignore {
838 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
839 $gs ||= Git::SVN->new;
840 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
841 $gs->prop_walk($gs->{path}, $r, sub {
842 my ($gs, $path, $props) = @_;
843 # $path is of the form /path/to/dir/
844 $path = '.' . $path;
845 # SVN can have attributes on empty directories,
846 # which git won't track
847 mkpath([$path]) unless -d $path;
848 my $ignore = $path . '.gitignore';
849 my $s = $props->{'svn:ignore'} or return;
850 open(GITIGNORE, '>', $ignore)
851 or fatal("Failed to open `$ignore' for writing: $!");
852 $s =~ s/[\r\n]+/\n/g;
853 $s =~ s/^\n+//;
854 chomp $s;
855 # Prefix all patterns so that the ignore doesn't apply
856 # to sub-directories.
857 $s =~ s#^#/#gm;
858 print GITIGNORE "$s\n";
859 close(GITIGNORE)
860 or fatal("Failed to close `$ignore': $!");
861 command_noisy('add', '-f', $ignore);
862 });
863}
864sub cmd_mkdirs {
866 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
867 $gs ||= Git::SVN->new;
868 $gs->mkemptydirs($_revision);
869}
870sub canonicalize_path {
872 my ($path) = @_;
873 my $dot_slash_added = 0;
874 if (substr($path, 0, 1) ne "/") {
875 $path = "./" . $path;
876 $dot_slash_added = 1;
877 }
878 # File::Spec->canonpath doesn't collapse x/../y into y (for a
879 # good reason), so let's do this manually.
880 $path =~ s#/+#/#g;
881 $path =~ s#/\.(?:/|$)#/#g;
882 $path =~ s#/[^/]+/\.\.##g;
883 $path =~ s#/$##g;
884 $path =~ s#^\./## if $dot_slash_added;
885 $path =~ s#^/##;
886 $path =~ s#^\.$##;
887 return $path;
888}
889sub canonicalize_url {
891 my ($url) = @_;
892 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
893 return $url;
894}
895# get_svnprops(PATH)
897# ------------------
898# Helper for cmd_propget and cmd_proplist below.
899sub get_svnprops {
900 my $path = shift;
901 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
902 $gs ||= Git::SVN->new;
903 # prefix THE PATH by the sub-directory from which the user
905 # invoked us.
906 $path = $cmd_dir_prefix . $path;
907 fatal("No such file or directory: $path") unless -e $path;
908 my $is_dir = -d $path ? 1 : 0;
909 $path = $gs->{path} . '/' . $path;
910 # canonicalize the path (otherwise libsvn will abort or fail to
912 # find the file)
913 $path = canonicalize_path($path);
914 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
916 my $props;
917 if ($is_dir) {
918 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
919 }
920 else {
921 (undef, $props) = $gs->ra->get_file($path, $r, undef);
922 }
923 return $props;
924}
925# cmd_propget (PROP, PATH)
927# ------------------------
928# Print the SVN property PROP for PATH.
929sub cmd_propget {
930 my ($prop, $path) = @_;
931 $path = '.' if not defined $path;
932 usage(1) if not defined $prop;
933 my $props = get_svnprops($path);
934 if (not defined $props->{$prop}) {
935 fatal("`$path' does not have a `$prop' SVN property.");
936 }
937 print $props->{$prop} . "\n";
938}
939# cmd_proplist (PATH)
941# -------------------
942# Print the list of SVN properties for PATH.
943sub cmd_proplist {
944 my $path = shift;
945 $path = '.' if not defined $path;
946 my $props = get_svnprops($path);
947 print "Properties on '$path':\n";
948 foreach (sort keys %{$props}) {
949 print " $_\n";
950 }
951}
952sub cmd_multi_init {
954 my $url = shift;
955 unless (defined $_trunk || @_branches || @_tags) {
956 usage(1);
957 }
958 $_prefix = '' unless defined $_prefix;
960 if (defined $url) {
961 $url = canonicalize_url($url);
962 init_subdir(@_);
963 }
964 do_git_init_db();
965 if (defined $_trunk) {
966 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
967 # try both old-style and new-style lookups:
968 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
969 unless ($gs_trunk) {
970 my ($trunk_url, $trunk_path) =
971 complete_svn_url($url, $_trunk);
972 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
973 undef, $trunk_ref);
974 }
975 }
976 return unless @_branches || @_tags;
977 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
978 foreach my $path (@_branches) {
979 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
980 }
981 foreach my $path (@_tags) {
982 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
983 }
984}
985sub cmd_multi_fetch {
987 $Git::SVN::no_reuse_existing = undef;
988 my $remotes = Git::SVN::read_all_remotes();
989 foreach my $repo_id (sort keys %$remotes) {
990 if ($remotes->{$repo_id}->{url}) {
991 Git::SVN::fetch_all($repo_id, $remotes);
992 }
993 }
994}
995# this command is special because it requires no metadata
997sub cmd_commit_diff {
998 my ($ta, $tb, $url) = @_;
999 my $usage = "Usage: $0 commit-diff -r<revision> ".
1000 "<tree-ish> <tree-ish> [<URL>]";
1001 fatal($usage) if (!defined $ta || !defined $tb);
1002 my $svn_path = '';
1003 if (!defined $url) {
1004 my $gs = eval { Git::SVN->new };
1005 if (!$gs) {
1006 fatal("Needed URL or usable git-svn --id in ",
1007 "the command-line\n", $usage);
1008 }
1009 $url = $gs->{url};
1010 $svn_path = $gs->{path};
1011 }
1012 unless (defined $_revision) {
1013 fatal("-r|--revision is a required argument\n", $usage);
1014 }
1015 if (defined $_message && defined $_file) {
1016 fatal("Both --message/-m and --file/-F specified ",
1017 "for the commit message.\n",
1018 "I have no idea what you mean");
1019 }
1020 if (defined $_file) {
1021 $_message = file_to_s($_file);
1022 } else {
1023 $_message ||= get_commit_entry($tb)->{log};
1024 }
1025 my $ra ||= Git::SVN::Ra->new($url);
1026 my $r = $_revision;
1027 if ($r eq 'HEAD') {
1028 $r = $ra->get_latest_revnum;
1029 } elsif ($r !~ /^\d+$/) {
1030 die "revision argument: $r not understood by git-svn\n";
1031 }
1032 my %ed_opts = ( r => $r,
1033 log => $_message,
1034 ra => $ra,
1035 tree_a => $ta,
1036 tree_b => $tb,
1037 editor_cb => sub { print "Committed r$_[0]\n" },
1038 svn_path => $svn_path );
1039 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1040 print "No changes\n$ta == $tb\n";
1041 }
1042}
1043sub escape_uri_only {
1045 my ($uri) = @_;
1046 my @tmp;
1047 foreach (split m{/}, $uri) {
1048 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1049 push @tmp, $_;
1050 }
1051 join('/', @tmp);
1052}
1053sub escape_url {
1055 my ($url) = @_;
1056 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1057 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1058 $url = "$scheme://$domain$uri";
1059 }
1060 $url;
1061}
1062sub cmd_info {
1064 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1065 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1066 if (exists $_[1]) {
1067 die "Too many arguments specified\n";
1068 }
1069 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1071 if (!$file_type && !$diff_status) {
1073 print STDERR "svn: '$path' is not under version control\n";
1074 exit 1;
1075 }
1076 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1078 unless ($gs) {
1079 die "Unable to determine upstream SVN information from ",
1080 "working tree history\n";
1081 }
1082 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1084 $path = "." if $path eq "";
1085 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1087 if ($_url) {
1089 print escape_url($full_url), "\n";
1090 return;
1091 }
1092 my $result = "Path: $path\n";
1094 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1095 $result .= "URL: " . escape_url($full_url) . "\n";
1096 eval {
1098 my $repos_root = $gs->repos_root;
1099 Git::SVN::remove_username($repos_root);
1100 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1101 };
1102 if ($@) {
1103 $result .= "Repository Root: (offline)\n";
1104 }
1105 ::_req_svn();
1106 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1107 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1108 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1109 $result .= "Node Kind: " .
1111 ($file_type eq "dir" ? "directory" : "file") . "\n";
1112 my $schedule = $diff_status eq "A"
1114 ? "add"
1115 : ($diff_status eq "D" ? "delete" : "normal");
1116 $result .= "Schedule: $schedule\n";
1117 if ($diff_status eq "A") {
1119 print $result, "\n";
1120 return;
1121 }
1122 my ($lc_author, $lc_rev, $lc_date_utc);
1124 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1125 my $log = command_output_pipe(@args);
1126 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1127 while (<$log>) {
1128 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1129 $lc_author = $1;
1130 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1131 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1132 (undef, $lc_rev, undef) = ::extract_metadata($1);
1133 }
1134 }
1135 close $log;
1136 Git::SVN::Log::set_local_timezone();
1138 $result .= "Last Changed Author: $lc_author\n";
1140 $result .= "Last Changed Rev: $lc_rev\n";
1141 $result .= "Last Changed Date: " .
1142 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1143 if ($file_type ne "dir") {
1145 my $text_last_updated_date =
1146 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1147 $result .=
1148 "Text Last Updated: " .
1149 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1150 "\n";
1151 my $checksum;
1152 if ($diff_status eq "D") {
1153 my ($fh, $ctx) =
1154 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1155 if ($file_type eq "link") {
1156 my $file_name = <$fh>;
1157 $checksum = md5sum("link $file_name");
1158 } else {
1159 $checksum = md5sum($fh);
1160 }
1161 command_close_pipe($fh, $ctx);
1162 } elsif ($file_type eq "link") {
1163 my $file_name =
1164 command(qw(cat-file blob), "HEAD:$path");
1165 $checksum =
1166 md5sum("link " . $file_name);
1167 } else {
1168 open FILE, "<", $path or die $!;
1169 $checksum = md5sum(\*FILE);
1170 close FILE or die $!;
1171 }
1172 $result .= "Checksum: " . $checksum . "\n";
1173 }
1174 print $result, "\n";
1176}
1177sub cmd_reset {
1179 my $target = shift || $_revision or die "SVN revision required\n";
1180 $target = $1 if $target =~ /^r(\d+)$/;
1181 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1182 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1183 unless ($gs) {
1184 die "Unable to determine upstream SVN information from ".
1185 "history\n";
1186 }
1187 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1188 die "Cannot find SVN revision $target\n" unless defined($c);
1189 $gs->rev_map_set($r, $c, 'reset', $uuid);
1190 print "r$r = $c ($gs->{ref_id})\n";
1191}
1192sub cmd_gc {
1194 if (!$can_compress) {
1195 warn "Compress::Zlib could not be found; unhandled.log " .
1196 "files will not be compressed.\n";
1197 }
1198 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1199}
1200########################### utility functions #########################
1202sub rebase_cmd {
1204 my @cmd = qw/rebase/;
1205 push @cmd, '-v' if $_verbose;
1206 push @cmd, qw/--merge/ if $_merge;
1207 push @cmd, "--strategy=$_strategy" if $_strategy;
1208 @cmd;
1209}
1210sub post_fetch_checkout {
1212 return if $_no_checkout;
1213 my $gs = $Git::SVN::_head or return;
1214 return if verify_ref('refs/heads/master^0');
1215 # look for "trunk" ref if it exists
1217 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1218 my $fetch = $remote->{fetch};
1219 if ($fetch) {
1220 foreach my $p (keys %$fetch) {
1221 basename($fetch->{$p}) eq 'trunk' or next;
1222 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1223 last;
1224 }
1225 }
1226 my $valid_head = verify_ref('HEAD^0');
1228 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1229 return if ($valid_head || !verify_ref('HEAD^0'));
1230 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1232 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1233 return if -f $index;
1234 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1236 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1237 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1238 print STDERR "Checked out HEAD:\n ",
1239 $gs->full_url, " r", $gs->last_rev, "\n";
1240 $gs->mkemptydirs($gs->last_rev);
1241}
1242sub complete_svn_url {
1244 my ($url, $path) = @_;
1245 $path =~ s#/+$##;
1246 if ($path !~ m#^[a-z\+]+://#) {
1247 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1248 fatal("E: '$path' is not a complete URL ",
1249 "and a separate URL is not specified");
1250 }
1251 return ($url, $path);
1252 }
1253 return ($path, '');
1254}
1255sub complete_url_ls_init {
1257 my ($ra, $repo_path, $switch, $pfx) = @_;
1258 unless ($repo_path) {
1259 print STDERR "W: $switch not specified\n";
1260 return;
1261 }
1262 $repo_path =~ s#/+$##;
1263 if ($repo_path =~ m#^[a-z\+]+://#) {
1264 $ra = Git::SVN::Ra->new($repo_path);
1265 $repo_path = '';
1266 } else {
1267 $repo_path =~ s#^/+##;
1268 unless ($ra) {
1269 fatal("E: '$repo_path' is not a complete URL ",
1270 "and a separate URL is not specified");
1271 }
1272 }
1273 my $url = $ra->{url};
1274 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1275 my $k = "svn-remote.$gs->{repo_id}.url";
1276 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1277 if ($orig_url && ($orig_url ne $gs->{url})) {
1278 die "$k already set: $orig_url\n",
1279 "wanted to set to: $gs->{url}\n";
1280 }
1281 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1282 my $remote_path = "$gs->{path}/$repo_path";
1283 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1284 $remote_path =~ s#/+#/#g;
1285 $remote_path =~ s#^/##g;
1286 $remote_path .= "/*" if $remote_path !~ /\*/;
1287 my ($n) = ($switch =~ /^--(\w+)/);
1288 if (length $pfx && $pfx !~ m#/$#) {
1289 die "--prefix='$pfx' must have a trailing slash '/'\n";
1290 }
1291 command_noisy('config',
1292 '--add',
1293 "svn-remote.$gs->{repo_id}.$n",
1294 "$remote_path:refs/remotes/$pfx*" .
1295 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1296}
1297sub verify_ref {
1299 my ($ref) = @_;
1300 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1301 { STDERR => 0 }); };
1302}
1303sub get_tree_from_treeish {
1305 my ($treeish) = @_;
1306 # $treeish can be a symbolic ref, too:
1307 my $type = command_oneline(qw/cat-file -t/, $treeish);
1308 my $expected;
1309 while ($type eq 'tag') {
1310 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1311 }
1312 if ($type eq 'commit') {
1313 $expected = (grep /^tree /, command(qw/cat-file commit/,
1314 $treeish))[0];
1315 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1316 die "Unable to get tree from $treeish\n" unless $expected;
1317 } elsif ($type eq 'tree') {
1318 $expected = $treeish;
1319 } else {
1320 die "$treeish is a $type, expected tree, tag or commit\n";
1321 }
1322 return $expected;
1323}
1324sub get_commit_entry {
1326 my ($treeish) = shift;
1327 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1328 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1329 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1330 open my $log_fh, '>', $commit_editmsg or croak $!;
1331 my $type = command_oneline(qw/cat-file -t/, $treeish);
1333 if ($type eq 'commit' || $type eq 'tag') {
1334 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1335 $type, $treeish);
1336 my $in_msg = 0;
1337 my $author;
1338 my $saw_from = 0;
1339 my $msgbuf = "";
1340 while (<$msg_fh>) {
1341 if (!$in_msg) {
1342 $in_msg = 1 if (/^\s*$/);
1343 $author = $1 if (/^author (.*>)/);
1344 } elsif (/^git-svn-id: /) {
1345 # skip this for now, we regenerate the
1346 # correct one on re-fetch anyways
1347 # TODO: set *:merge properties or like...
1348 } else {
1349 if (/^From:/ || /^Signed-off-by:/) {
1350 $saw_from = 1;
1351 }
1352 $msgbuf .= $_;
1353 }
1354 }
1355 $msgbuf =~ s/\s+$//s;
1356 if ($Git::SVN::_add_author_from && defined($author)
1357 && !$saw_from) {
1358 $msgbuf .= "\n\nFrom: $author";
1359 }
1360 print $log_fh $msgbuf or croak $!;
1361 command_close_pipe($msg_fh, $ctx);
1362 }
1363 close $log_fh or croak $!;
1364 if ($_edit || ($type eq 'tree')) {
1366 chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1367 system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1368 }
1369 rename $commit_editmsg, $commit_msg or croak $!;
1370 {
1371 require Encode;
1372 # SVN requires messages to be UTF-8 when entering the repo
1373 local $/;
1374 open $log_fh, '<', $commit_msg or croak $!;
1375 binmode $log_fh;
1376 chomp($log_entry{log} = <$log_fh>);
1377 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1379 my $msg = $log_entry{log};
1380 eval { $msg = Encode::decode($enc, $msg, 1) };
1382 if ($@) {
1383 die "Could not decode as $enc:\n", $msg,
1384 "\nPerhaps you need to set i18n.commitencoding\n";
1385 }
1386 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1388 die "Could not encode as UTF-8:\n$msg\n" if $@;
1389 $log_entry{log} = $msg;
1391 close $log_fh or croak $!;
1393 }
1394 unlink $commit_msg;
1395 \%log_entry;
1396}
1397sub s_to_file {
1399 my ($str, $file, $mode) = @_;
1400 open my $fd,'>',$file or croak $!;
1401 print $fd $str,"\n" or croak $!;
1402 close $fd or croak $!;
1403 chmod ($mode &~ umask, $file) if (defined $mode);
1404}
1405sub file_to_s {
1407 my $file = shift;
1408 open my $fd,'<',$file or croak "$!: file: $file\n";
1409 local $/;
1410 my $ret = <$fd>;
1411 close $fd or croak $!;
1412 $ret =~ s/\s*$//s;
1413 return $ret;
1414}
1415# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1417sub load_authors {
1418 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1419 my $log = $cmd eq 'log';
1420 while (<$authors>) {
1421 chomp;
1422 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1423 my ($user, $name, $email) = ($1, $2, $3);
1424 if ($log) {
1425 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1426 } else {
1427 $users{$user} = [$name, $email];
1428 }
1429 }
1430 close $authors or croak $!;
1431}
1432# convert GetOpt::Long specs for use by git-config
1434sub read_git_config {
1435 my $opts = shift;
1436 my @config_only;
1437 foreach my $o (keys %$opts) {
1438 # if we have mixedCase and a long option-only, then
1439 # it's a config-only variable that we don't need for
1440 # the command-line.
1441 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1442 my $v = $opts->{$o};
1443 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1444 $key =~ s/-//g;
1445 my $arg = 'git config';
1446 $arg .= ' --int' if ($o =~ /[:=]i$/);
1447 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1448 if (ref $v eq 'ARRAY') {
1449 chomp(my @tmp = `$arg --get-all svn.$key`);
1450 @$v = @tmp if @tmp;
1451 } else {
1452 chomp(my $tmp = `$arg --get svn.$key`);
1453 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1454 $$v = $tmp;
1455 }
1456 }
1457 }
1458 delete @$opts{@config_only} if @config_only;
1459}
1460sub extract_metadata {
1462 my $id = shift or return (undef, undef, undef);
1463 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1464 \s([a-f\d\-]+)$/ix);
1465 if (!defined $rev || !$uuid || !$url) {
1466 # some of the original repositories I made had
1467 # identifiers like this:
1468 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1469 }
1470 return ($url, $rev, $uuid);
1471}
1472sub cmt_metadata {
1474 return extract_metadata((grep(/^git-svn-id: /,
1475 command(qw/cat-file commit/, shift)))[-1]);
1476}
1477sub cmt_sha2rev_batch {
1479 my %s2r;
1480 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1481 my $list = shift;
1482 foreach my $sha (@{$list}) {
1484 my $first = 1;
1485 my $size = 0;
1486 print $out $sha, "\n";
1487 while (my $line = <$in>) {
1489 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1490 last;
1491 } elsif ($first &&
1492 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1493 $first = 0;
1494 $size = $1;
1495 next;
1496 } elsif ($line =~ /^(git-svn-id: )/) {
1497 my (undef, $rev, undef) =
1498 extract_metadata($line);
1499 $s2r{$sha} = $rev;
1500 }
1501 $size -= length($line);
1503 last if ($size == 0);
1504 }
1505 }
1506 command_close_bidi_pipe($pid, $in, $out, $ctx);
1508 return \%s2r;
1510}
1511sub working_head_info {
1513 my ($head, $refs) = @_;
1514 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1515 my ($fh, $ctx) = command_output_pipe(@args, $head);
1516 my $hash;
1517 my %max;
1518 while (<$fh>) {
1519 if ( m{^commit ($::sha1)$} ) {
1520 unshift @$refs, $hash if $hash and $refs;
1521 $hash = $1;
1522 next;
1523 }
1524 next unless s{^\s*(git-svn-id:)}{$1};
1525 my ($url, $rev, $uuid) = extract_metadata($_);
1526 if (defined $url && defined $rev) {
1527 next if $max{$url} and $max{$url} < $rev;
1528 if (my $gs = Git::SVN->find_by_url($url)) {
1529 my $c = $gs->rev_map_get($rev, $uuid);
1530 if ($c && $c eq $hash) {
1531 close $fh; # break the pipe
1532 return ($url, $rev, $uuid, $gs);
1533 } else {
1534 $max{$url} ||= $gs->rev_map_max;
1535 }
1536 }
1537 }
1538 }
1539 command_close_pipe($fh, $ctx);
1540 (undef, undef, undef, undef);
1541}
1542sub read_commit_parents {
1544 my ($parents, $c) = @_;
1545 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1546 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1547 @{$parents->{$c}} = split(/ /, $p);
1548}
1549sub linearize_history {
1551 my ($gs, $refs) = @_;
1552 my %parents;
1553 foreach my $c (@$refs) {
1554 read_commit_parents(\%parents, $c);
1555 }
1556 my @linear_refs;
1558 my %skip = ();
1559 my $last_svn_commit = $gs->last_commit;
1560 foreach my $c (reverse @$refs) {
1561 next if $c eq $last_svn_commit;
1562 last if $skip{$c};
1563 unshift @linear_refs, $c;
1565 $skip{$c} = 1;
1566 # we only want the first parent to diff against for linear
1568 # history, we save the rest to inject when we finalize the
1569 # svn commit
1570 my $fp_a = verify_ref("$c~1");
1571 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1572 if (!$fp_a || !$fp_b) {
1573 die "Commit $c\n",
1574 "has no parent commit, and therefore ",
1575 "nothing to diff against.\n",
1576 "You should be working from a repository ",
1577 "originally created by git-svn\n";
1578 }
1579 if ($fp_a ne $fp_b) {
1580 die "$c~1 = $fp_a, however parsing commit $c ",
1581 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1582 }
1583 foreach my $p (@{$parents{$c}}) {
1585 $skip{$p} = 1;
1586 }
1587 }
1588 (\@linear_refs, \%parents);
1589}
1590sub find_file_type_and_diff_status {
1592 my ($path) = @_;
1593 return ('dir', '') if $path eq '';
1594 my $diff_output =
1596 command_oneline(qw(diff --cached --name-status --), $path) || "";
1597 my $diff_status = (split(' ', $diff_output))[0] || "";
1598 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1600 return (undef, undef) if !$diff_status && !$ls_tree;
1602 if ($diff_status eq "A") {
1604 return ("link", $diff_status) if -l $path;
1605 return ("dir", $diff_status) if -d $path;
1606 return ("file", $diff_status);
1607 }
1608 my $mode = (split(' ', $ls_tree))[0] || "";
1610 return ("link", $diff_status) if $mode eq "120000";
1612 return ("dir", $diff_status) if $mode eq "040000";
1613 return ("file", $diff_status);
1614}
1615sub md5sum {
1617 my $arg = shift;
1618 my $ref = ref $arg;
1619 my $md5 = Digest::MD5->new();
1620 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1621 $md5->addfile($arg) or croak $!;
1622 } elsif ($ref eq 'SCALAR') {
1623 $md5->add($$arg) or croak $!;
1624 } elsif (!$ref) {
1625 $md5->add($arg) or croak $!;
1626 } else {
1627 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1628 }
1629 return $md5->hexdigest();
1630}
1631sub gc_directory {
1633 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1634 my $out_filename = $_ . ".gz";
1635 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1636 binmode $in_fh;
1637 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1638 die "Unable to open $out_filename: $!\n";
1639 my $res;
1641 while ($res = sysread($in_fh, my $str, 1024)) {
1642 $gz->gzwrite($str) or
1643 die "Unable to write: ".$gz->gzerror()."!\n";
1644 }
1645 unlink $_ or die "unlink $File::Find::name: $!\n";
1646 } elsif (-f $_ && basename($_) eq "index") {
1647 unlink $_ or die "unlink $_: $!\n";
1648 }
1649}
1650package Git::SVN;
1652use strict;
1653use warnings;
1654use Fcntl qw/:DEFAULT :seek/;
1655use constant rev_map_fmt => 'NH40';
1656use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1657 $_repack $_repack_flags $_use_svm_props $_head
1658 $_use_svnsync_props $no_reuse_existing $_minimize_url
1659 $_use_log_author $_add_author_from $_localtime/;
1660use Carp qw/croak/;
1661use File::Path qw/mkpath/;
1662use File::Copy qw/copy/;
1663use IPC::Open3;
1664use Memoize; # core since 5.8.0, Jul 2002
1665use Memoize::Storable;
1666my ($_gc_nr, $_gc_period);
1668# properties that we do not log:
1670my %SKIP_PROP;
1671BEGIN {
1672 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1673 svn:special svn:executable
1674 svn:entry:committed-rev
1675 svn:entry:last-author
1676 svn:entry:uuid
1677 svn:entry:committed-date/;
1678 # some options are read globally, but can be overridden locally
1680 # per [svn-remote "..."] section. Command-line options will *NOT*
1681 # override options set in an [svn-remote "..."] section
1682 no strict 'refs';
1683 for my $option (qw/follow_parent no_metadata use_svm_props
1684 use_svnsync_props/) {
1685 my $key = $option;
1686 $key =~ tr/_//d;
1687 my $prop = "-$option";
1688 *$option = sub {
1689 my ($self) = @_;
1690 return $self->{$prop} if exists $self->{$prop};
1691 my $k = "svn-remote.$self->{repo_id}.$key";
1692 eval { command_oneline(qw/config --get/, $k) };
1693 if ($@) {
1694 $self->{$prop} = ${"Git::SVN::_$option"};
1695 } else {
1696 my $v = command_oneline(qw/config --bool/,$k);
1697 $self->{$prop} = $v eq 'false' ? 0 : 1;
1698 }
1699 return $self->{$prop};
1700 }
1701 }
1702}
1703my (%LOCKFILES, %INDEX_FILES);
1706END {
1707 unlink keys %LOCKFILES if %LOCKFILES;
1708 unlink keys %INDEX_FILES if %INDEX_FILES;
1709}
1710sub resolve_local_globs {
1712 my ($url, $fetch, $glob_spec) = @_;
1713 return unless defined $glob_spec;
1714 my $ref = $glob_spec->{ref};
1715 my $path = $glob_spec->{path};
1716 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1717 next unless m#^$ref->{regex}$#;
1718 my $p = $1;
1719 my $pathname = desanitize_refname($path->full_path($p));
1720 my $refname = desanitize_refname($ref->full_path($p));
1721 if (my $existing = $fetch->{$pathname}) {
1722 if ($existing ne $refname) {
1723 die "Refspec conflict:\n",
1724 "existing: $existing\n",
1725 " globbed: $refname\n";
1726 }
1727 my $u = (::cmt_metadata("$refname"))[0];
1728 $u =~ s!^\Q$url\E(/|$)!! or die
1729 "$refname: '$url' not found in '$u'\n";
1730 if ($pathname ne $u) {
1731 warn "W: Refspec glob conflict ",
1732 "(ref: $refname):\n",
1733 "expected path: $pathname\n",
1734 " real path: $u\n",
1735 "Continuing ahead with $u\n";
1736 next;
1737 }
1738 } else {
1739 $fetch->{$pathname} = $refname;
1740 }
1741 }
1742}
1743sub parse_revision_argument {
1745 my ($base, $head) = @_;
1746 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1747 return ($base, $head);
1748 }
1749 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1750 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1751 return ($head, $head) if ($::_revision eq 'HEAD');
1752 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1753 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1754 die "revision argument: $::_revision not understood by git-svn\n";
1755}
1756sub fetch_all {
1758 my ($repo_id, $remotes) = @_;
1759 if (ref $repo_id) {
1760 my $gs = $repo_id;
1761 $repo_id = undef;
1762 $repo_id = $gs->{repo_id};
1763 }
1764 $remotes ||= read_all_remotes();
1765 my $remote = $remotes->{$repo_id} or
1766 die "[svn-remote \"$repo_id\"] unknown\n";
1767 my $fetch = $remote->{fetch};
1768 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1769 my (@gs, @globs);
1770 my $ra = Git::SVN::Ra->new($url);
1771 my $uuid = $ra->get_uuid;
1772 my $head = $ra->get_latest_revnum;
1773 # ignore errors, $head revision may not even exist anymore
1775 eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
1776 warn "W: $@\n" if $@;
1777 my $base = defined $fetch ? $head : 0;
1779 # read the max revs for wildcard expansion (branches/*, tags/*)
1781 foreach my $t (qw/branches tags/) {
1782 defined $remote->{$t} or next;
1783 push @globs, @{$remote->{$t}};
1784 my $max_rev = eval { tmp_config(qw/--int --get/,
1786 "svn-remote.$repo_id.${t}-maxRev") };
1787 if (defined $max_rev && ($max_rev < $base)) {
1788 $base = $max_rev;
1789 } elsif (!defined $max_rev) {
1790 $base = 0;
1791 }
1792 }
1793 if ($fetch) {
1795 foreach my $p (sort keys %$fetch) {
1796 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1797 my $lr = $gs->rev_map_max;
1798 if (defined $lr) {
1799 $base = $lr if ($lr < $base);
1800 }
1801 push @gs, $gs;
1802 }
1803 }
1804 ($base, $head) = parse_revision_argument($base, $head);
1806 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1807}
1808sub read_all_remotes {
1810 my $r = {};
1811 my $use_svm_props = eval { command_oneline(qw/config --bool
1812 svn.useSvmProps/) };
1813 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1814 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1815 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1816 if (m!^(.+)\.fetch=$svn_refspec$!) {
1817 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1818 die("svn-remote.$remote: remote ref '$remote_ref' "
1819 . "must start with 'refs/'\n")
1820 unless $remote_ref =~ m{^refs/};
1821 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1822 $r->{$remote}->{svm} = {} if $use_svm_props;
1823 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1824 $r->{$1}->{svm} = {};
1825 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1826 $r->{$1}->{url} = $2;
1827 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1828 my ($remote, $t, $local_ref, $remote_ref) =
1829 ($1, $2, $3, $4);
1830 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1831 . "must start with 'refs/'\n")
1832 unless $remote_ref =~ m{^refs/};
1833 my $rs = {
1834 t => $t,
1835 remote => $remote,
1836 path => Git::SVN::GlobSpec->new($local_ref, 1),
1837 ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
1838 if (length($rs->{ref}->{right}) != 0) {
1839 die "The '*' glob character must be the last ",
1840 "character of '$remote_ref'\n";
1841 }
1842 push @{ $r->{$remote}->{$t} }, $rs;
1843 }
1844 }
1845 map {
1847 if (defined $r->{$_}->{svm}) {
1848 my $svm;
1849 eval {
1850 my $section = "svn-remote.$_";
1851 $svm = {
1852 source => tmp_config('--get',
1853 "$section.svm-source"),
1854 replace => tmp_config('--get',
1855 "$section.svm-replace"),
1856 }
1857 };
1858 $r->{$_}->{svm} = $svm;
1859 }
1860 } keys %$r;
1861 $r;
1863}
1864sub init_vars {
1866 $_gc_nr = $_gc_period = 1000;
1867 if (defined $_repack || defined $_repack_flags) {
1868 warn "Repack options are obsolete; they have no effect.\n";
1869 }
1870}
1871sub verify_remotes_sanity {
1873 return unless -d $ENV{GIT_DIR};
1874 my %seen;
1875 foreach (command(qw/config -l/)) {
1876 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1877 if ($seen{$1}) {
1878 die "Remote ref refs/remote/$1 is tracked by",
1879 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1880 "Please resolve this ambiguity in ",
1881 "your git configuration file before ",
1882 "continuing\n";
1883 }
1884 $seen{$1} = $_;
1885 }
1886 }
1887}
1888sub find_existing_remote {
1890 my ($url, $remotes) = @_;
1891 return undef if $no_reuse_existing;
1892 my $existing;
1893 foreach my $repo_id (keys %$remotes) {
1894 my $u = $remotes->{$repo_id}->{url} or next;
1895 next if $u ne $url;
1896 $existing = $repo_id;
1897 last;
1898 }
1899 $existing;
1900}
1901sub init_remote_config {
1903 my ($self, $url, $no_write) = @_;
1904 $url =~ s!/+$!!; # strip trailing slash
1905 my $r = read_all_remotes();
1906 my $existing = find_existing_remote($url, $r);
1907 if ($existing) {
1908 unless ($no_write) {
1909 print STDERR "Using existing ",
1910 "[svn-remote \"$existing\"]\n";
1911 }
1912 $self->{repo_id} = $existing;
1913 } elsif ($_minimize_url) {
1914 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1915 $existing = find_existing_remote($min_url, $r);
1916 if ($existing) {
1917 unless ($no_write) {
1918 print STDERR "Using existing ",
1919 "[svn-remote \"$existing\"]\n";
1920 }
1921 $self->{repo_id} = $existing;
1922 }
1923 if ($min_url ne $url) {
1924 unless ($no_write) {
1925 print STDERR "Using higher level of URL: ",
1926 "$url => $min_url\n";
1927 }
1928 my $old_path = $self->{path};
1929 $self->{path} = $url;
1930 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1931 if (length $old_path) {
1932 $self->{path} .= "/$old_path";
1933 }
1934 $url = $min_url;
1935 }
1936 }
1937 my $orig_url;
1938 if (!$existing) {
1939 # verify that we aren't overwriting anything:
1940 $orig_url = eval {
1941 command_oneline('config', '--get',
1942 "svn-remote.$self->{repo_id}.url")
1943 };
1944 if ($orig_url && ($orig_url ne $url)) {
1945 die "svn-remote.$self->{repo_id}.url already set: ",
1946 "$orig_url\nwanted to set to: $url\n";
1947 }
1948 }
1949 my ($xrepo_id, $xpath) = find_ref($self->refname);
1950 if (!$no_write && defined $xpath) {
1951 die "svn-remote.$xrepo_id.fetch already set to track ",
1952 "$xpath:", $self->refname, "\n";
1953 }
1954 unless ($no_write) {
1955 command_noisy('config',
1956 "svn-remote.$self->{repo_id}.url", $url);
1957 $self->{path} =~ s{^/}{};
1958 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1959 command_noisy('config', '--add',
1960 "svn-remote.$self->{repo_id}.fetch",
1961 "$self->{path}:".$self->refname);
1962 }
1963 $self->{url} = $url;
1964}
1965sub find_by_url { # repos_root and, path are optional
1967 my ($class, $full_url, $repos_root, $path) = @_;
1968 return undef unless defined $full_url;
1970 remove_username($full_url);
1971 remove_username($repos_root) if defined $repos_root;
1972 my $remotes = read_all_remotes();
1973 if (defined $full_url && defined $repos_root && !defined $path) {
1974 $path = $full_url;
1975 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1976 }
1977 foreach my $repo_id (keys %$remotes) {
1978 my $u = $remotes->{$repo_id}->{url} or next;
1979 remove_username($u);
1980 next if defined $repos_root && $repos_root ne $u;
1981 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1983 foreach my $t (qw/branches tags/) {
1984 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1985 resolve_local_globs($u, $fetch, $globspec);
1986 }
1987 }
1988 my $p = $path;
1989 my $rwr = rewrite_root({repo_id => $repo_id});
1990 my $svm = $remotes->{$repo_id}->{svm}
1991 if defined $remotes->{$repo_id}->{svm};
1992 unless (defined $p) {
1993 $p = $full_url;
1994 my $z = $u;
1995 my $prefix = '';
1996 if ($rwr) {
1997 $z = $rwr;
1998 remove_username($z);
1999 } elsif (defined $svm) {
2000 $z = $svm->{source};
2001 $prefix = $svm->{replace};
2002 $prefix =~ s#^\Q$u\E(?:/|$)##;
2003 $prefix =~ s#/$##;
2004 }
2005 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2006 }
2007 foreach my $f (keys %$fetch) {
2008 next if $f ne $p;
2009 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2010 }
2011 }
2012 undef;
2013}
2014sub init {
2016 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2017 my $self = _new($class, $repo_id, $ref_id, $path);
2018 if (defined $url) {
2019 $self->init_remote_config($url, $no_write);
2020 }
2021 $self;
2022}
2023sub find_ref {
2025 my ($ref_id) = @_;
2026 foreach (command(qw/config -l/)) {
2027 next unless m!^svn-remote\.(.+)\.fetch=
2028 \s*(.*?)\s*:\s*(.+?)\s*$!x;
2029 my ($repo_id, $path, $ref) = ($1, $2, $3);
2030 if ($ref eq $ref_id) {
2031 $path = '' if ($path =~ m#^\./?#);
2032 return ($repo_id, $path);
2033 }
2034 }
2035 (undef, undef, undef);
2036}
2037sub new {
2039 my ($class, $ref_id, $repo_id, $path) = @_;
2040 if (defined $ref_id && !defined $repo_id && !defined $path) {
2041 ($repo_id, $path) = find_ref($ref_id);
2042 if (!defined $repo_id) {
2043 die "Could not find a \"svn-remote.*.fetch\" key ",
2044 "in the repository configuration matching: ",
2045 "$ref_id\n";
2046 }
2047 }
2048 my $self = _new($class, $repo_id, $ref_id, $path);
2049 if (!defined $self->{path} || !length $self->{path}) {
2050 my $fetch = command_oneline('config', '--get',
2051 "svn-remote.$repo_id.fetch",
2052 ":$ref_id\$") or
2053 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2054 "\":$ref_id\$\" in config\n";
2055 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2056 }
2057 $self->{url} = command_oneline('config', '--get',
2058 "svn-remote.$repo_id.url") or
2059 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2060 $self->rebuild;
2061 $self;
2062}
2063sub refname {
2065 my ($refname) = $_[0]->{ref_id} ;
2066 # It cannot end with a slash /, we'll throw up on this because
2068 # SVN can't have directories with a slash in their name, either:
2069 if ($refname =~ m{/$}) {
2070 die "ref: '$refname' ends with a trailing slash, this is ",
2071 "not permitted by git nor Subversion\n";
2072 }
2073 # It cannot have ASCII control character space, tilde ~, caret ^,
2075 # colon :, question-mark ?, asterisk *, space, or open bracket [
2076 # anywhere.
2077 #
2078 # Additionally, % must be escaped because it is used for escaping
2079 # and we want our escaped refname to be reversible
2080 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2081 # no slash-separated component can begin with a dot .
2083 # /.* becomes /%2E*
2084 $refname =~ s{/\.}{/%2E}g;
2085 # It cannot have two consecutive dots .. anywhere
2087 # .. becomes %2E%2E
2088 $refname =~ s{\.\.}{%2E%2E}g;
2089 # trailing dots and .lock are not allowed
2091 # .$ becomes %2E and .lock becomes %2Elock
2092 $refname =~ s{\.(?=$|lock$)}{%2E};
2093 # the sequence @{ is used to access the reflog
2095 # @{ becomes %40{
2096 $refname =~ s{\@\{}{%40\{}g;
2097 return $refname;
2099}
2100sub desanitize_refname {
2102 my ($refname) = @_;
2103 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2104 return $refname;
2105}
2106sub svm_uuid {
2108 my ($self) = @_;
2109 return $self->{svm}->{uuid} if $self->svm;
2110 $self->ra;
2111 unless ($self->{svm}) {
2112 die "SVM UUID not cached, and reading remotely failed\n";
2113 }
2114 $self->{svm}->{uuid};
2115}
2116sub svm {
2118 my ($self) = @_;
2119 return $self->{svm} if $self->{svm};
2120 my $svm;
2121 # see if we have it in our config, first:
2122 eval {
2123 my $section = "svn-remote.$self->{repo_id}";
2124 $svm = {
2125 source => tmp_config('--get', "$section.svm-source"),
2126 uuid => tmp_config('--get', "$section.svm-uuid"),
2127 replace => tmp_config('--get', "$section.svm-replace"),
2128 }
2129 };
2130 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2131 $self->{svm} = $svm;
2132 }
2133 $self->{svm};
2134}
2135sub _set_svm_vars {
2137 my ($self, $ra) = @_;
2138 return $ra if $self->svm;
2139 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2141 "(svm:source, svm:uuid) ",
2142 "from the following URLs:\n" );
2143 sub read_svm_props {
2144 my ($self, $ra, $path, $r) = @_;
2145 my $props = ($ra->get_dir($path, $r))[2];
2146 my $src = $props->{'svm:source'};
2147 my $uuid = $props->{'svm:uuid'};
2148 return undef if (!$src || !$uuid);
2149 chomp($src, $uuid);
2151 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2153 or die "doesn't look right - svm:uuid is '$uuid'\n";
2154 # the '!' is used to mark the repos_root!/relative/path
2156 $src =~ s{/?!/?}{/};
2157 $src =~ s{/+$}{}; # no trailing slashes please
2158 # username is of no interest
2159 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2160 my $replace = $ra->{url};
2162 $replace .= "/$path" if length $path;
2163 my $section = "svn-remote.$self->{repo_id}";
2165 tmp_config("$section.svm-source", $src);
2166 tmp_config("$section.svm-replace", $replace);
2167 tmp_config("$section.svm-uuid", $uuid);
2168 $self->{svm} = {
2169 source => $src,
2170 uuid => $uuid,
2171 replace => $replace
2172 };
2173 }
2174 my $r = $ra->get_latest_revnum;
2176 my $path = $self->{path};
2177 my %tried;
2178 while (length $path) {
2179 unless ($tried{"$self->{url}/$path"}) {
2180 return $ra if $self->read_svm_props($ra, $path, $r);
2181 $tried{"$self->{url}/$path"} = 1;
2182 }
2183 $path =~ s#/?[^/]+$##;
2184 }
2185 die "Path: '$path' should be ''\n" if $path ne '';
2186 return $ra if $self->read_svm_props($ra, $path, $r);
2187 $tried{"$self->{url}/$path"} = 1;
2188 if ($ra->{repos_root} eq $self->{url}) {
2190 die @err, (map { " $_\n" } keys %tried), "\n";
2191 }
2192 # nope, make sure we're connected to the repository root:
2194 my $ok;
2195 my @tried_b;
2196 $path = $ra->{svn_path};
2197 $ra = Git::SVN::Ra->new($ra->{repos_root});
2198 while (length $path) {
2199 unless ($tried{"$ra->{url}/$path"}) {
2200 $ok = $self->read_svm_props($ra, $path, $r);
2201 last if $ok;
2202 $tried{"$ra->{url}/$path"} = 1;
2203 }
2204 $path =~ s#/?[^/]+$##;
2205 }
2206 die "Path: '$path' should be ''\n" if $path ne '';
2207 $ok ||= $self->read_svm_props($ra, $path, $r);
2208 $tried{"$ra->{url}/$path"} = 1;
2209 if (!$ok) {
2210 die @err, (map { " $_\n" } keys %tried), "\n";
2211 }
2212 Git::SVN::Ra->new($self->{url});
2213}
2214sub svnsync {
2216 my ($self) = @_;
2217 return $self->{svnsync} if $self->{svnsync};
2218 if ($self->no_metadata) {
2220 die "Can't have both 'noMetadata' and ",
2221 "'useSvnsyncProps' options set!\n";
2222 }
2223 if ($self->rewrite_root) {
2224 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2225 "options set!\n";
2226 }
2227 if ($self->rewrite_uuid) {
2228 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2229 "options set!\n";
2230 }
2231 my $svnsync;
2233 # see if we have it in our config, first:
2234 eval {
2235 my $section = "svn-remote.$self->{repo_id}";
2236 my $url = tmp_config('--get', "$section.svnsync-url");
2238 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2239 die "doesn't look right - svn:sync-from-url is '$url'\n";
2240 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2242 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2243 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2244 $svnsync = { url => $url, uuid => $uuid }
2246 };
2247 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2248 return $self->{svnsync} = $svnsync;
2249 }
2250 my $err = "useSvnsyncProps set, but failed to read " .
2252 "svnsync property: svn:sync-from-";
2253 my $rp = $self->ra->rev_proplist(0);
2254 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2256 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2257 die "doesn't look right - svn:sync-from-url is '$url'\n";
2258 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2260 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2261 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2262 my $section = "svn-remote.$self->{repo_id}";
2264 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2265 tmp_config('--add', "$section.svnsync-url", $url);
2266 return $self->{svnsync} = { url => $url, uuid => $uuid };
2267}
2268# this allows us to memoize our SVN::Ra UUID locally and avoid a
2270# remote lookup (useful for 'git svn log').
2271sub ra_uuid {
2272 my ($self) = @_;
2273 unless ($self->{ra_uuid}) {
2274 my $key = "svn-remote.$self->{repo_id}.uuid";
2275 my $uuid = eval { tmp_config('--get', $key) };
2276 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2277 $self->{ra_uuid} = $uuid;
2278 } else {
2279 die "ra_uuid called without URL\n" unless $self->{url};
2280 $self->{ra_uuid} = $self->ra->get_uuid;
2281 tmp_config('--add', $key, $self->{ra_uuid});
2282 }
2283 }
2284 $self->{ra_uuid};
2285}
2286sub _set_repos_root {
2288 my ($self, $repos_root) = @_;
2289 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2290 $repos_root ||= $self->ra->{repos_root};
2291 tmp_config($k, $repos_root);
2292 $repos_root;
2293}
2294sub repos_root {
2296 my ($self) = @_;
2297 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2298 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2299}
2300sub ra {
2302 my ($self) = shift;
2303 my $ra = Git::SVN::Ra->new($self->{url});
2304 $self->_set_repos_root($ra->{repos_root});
2305 if ($self->use_svm_props && !$self->{svm}) {
2306 if ($self->no_metadata) {
2307 die "Can't have both 'noMetadata' and ",
2308 "'useSvmProps' options set!\n";
2309 } elsif ($self->use_svnsync_props) {
2310 die "Can't have both 'useSvnsyncProps' and ",
2311 "'useSvmProps' options set!\n";
2312 }
2313 $ra = $self->_set_svm_vars($ra);
2314 $self->{-want_revprops} = 1;
2315 }
2316 $ra;
2317}
2318# prop_walk(PATH, REV, SUB)
2320# -------------------------
2321# Recursively traverse PATH at revision REV and invoke SUB for each
2322# directory that contains a SVN property. SUB will be invoked as
2323# follows: &SUB(gs, path, props); where `gs' is this instance of
2324# Git::SVN, `path' the path to the directory where the properties
2325# `props' were found. The `path' will be relative to point of checkout,
2326# that is, if url://repo/trunk is the current Git branch, and that
2327# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2328# as `path' (note the trailing `/').
2329sub prop_walk {
2330 my ($self, $path, $rev, $sub) = @_;
2331 $path =~ s#^/##;
2333 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2334 $path =~ s#^/*#/#g;
2335 my $p = $path;
2336 # Strip the irrelevant part of the path.
2337 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2338 # Ensure the path is terminated by a `/'.
2339 $p =~ s#/*$#/#;
2340 # The properties contain all the internal SVN stuff nobody
2342 # (usually) cares about.
2343 my $interesting_props = 0;
2344 foreach (keys %{$props}) {
2345 # If it doesn't start with `svn:', it must be a
2346 # user-defined property.
2347 ++$interesting_props and next if $_ !~ /^svn:/;
2348 # FIXME: Fragile, if SVN adds new public properties,
2349 # this needs to be updated.
2350 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2351 |eol-style|mime-type
2352 |externals|needs-lock)$/x;
2353 }
2354 &$sub($self, $p, $props) if $interesting_props;
2355 foreach (sort keys %$dirent) {
2357 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2358 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2359 }
2360}
2361sub last_rev { ($_[0]->last_rev_commit)[0] }
2363sub last_commit { ($_[0]->last_rev_commit)[1] }
2364# returns the newest SVN revision number and newest commit SHA1
2366sub last_rev_commit {
2367 my ($self) = @_;
2368 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2369 return ($self->{last_rev}, $self->{last_commit});
2370 }
2371 my $c = ::verify_ref($self->refname.'^0');
2372 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2373 my $rev = (::cmt_metadata($c))[1];
2374 if (defined $rev) {
2375 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2376 return ($rev, $c);
2377 }
2378 }
2379 my $map_path = $self->map_path;
2380 unless (-e $map_path) {
2381 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2382 return (undef, undef);
2383 }
2384 my ($rev, $commit) = $self->rev_map_max(1);
2385 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2386 return ($rev, $commit);
2387}
2388sub get_fetch_range {
2390 my ($self, $min, $max) = @_;
2391 $max ||= $self->ra->get_latest_revnum;
2392 $min ||= $self->rev_map_max;
2393 (++$min, $max);
2394}
2395sub tmp_config {
2397 my (@args) = @_;
2398 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2399 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2400 if (! -f $config && -f $old_def_config) {
2401 rename $old_def_config, $config or
2402 die "Failed rename $old_def_config => $config: $!\n";
2403 }
2404 my $old_config = $ENV{GIT_CONFIG};
2405 $ENV{GIT_CONFIG} = $config;
2406 $@ = undef;
2407 my @ret = eval {
2408 unless (-f $config) {
2409 mkfile($config);
2410 open my $fh, '>', $config or
2411 die "Can't open $config: $!\n";
2412 print $fh "; This file is used internally by ",
2413 "git-svn\n" or die
2414 "Couldn't write to $config: $!\n";
2415 print $fh "; You should not have to edit it\n" or
2416 die "Couldn't write to $config: $!\n";
2417 close $fh or die "Couldn't close $config: $!\n";
2418 }
2419 command('config', @args);
2420 };
2421 my $err = $@;
2422 if (defined $old_config) {
2423 $ENV{GIT_CONFIG} = $old_config;
2424 } else {
2425 delete $ENV{GIT_CONFIG};
2426 }
2427 die $err if $err;
2428 wantarray ? @ret : $ret[0];
2429}
2430sub tmp_index_do {
2432 my ($self, $sub) = @_;
2433 my $old_index = $ENV{GIT_INDEX_FILE};
2434 $ENV{GIT_INDEX_FILE} = $self->{index};
2435 $@ = undef;
2436 my @ret = eval {
2437 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2438 mkpath([$dir]) unless -d $dir;
2439 &$sub;
2440 };
2441 my $err = $@;
2442 if (defined $old_index) {
2443 $ENV{GIT_INDEX_FILE} = $old_index;
2444 } else {
2445 delete $ENV{GIT_INDEX_FILE};
2446 }
2447 die $err if $err;
2448 wantarray ? @ret : $ret[0];
2449}
2450sub assert_index_clean {
2452 my ($self, $treeish) = @_;
2453 $self->tmp_index_do(sub {
2455 command_noisy('read-tree', $treeish) unless -e $self->{index};
2456 my $x = command_oneline('write-tree');
2457 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2458 /^tree ($::sha1)/mo);
2459 return if $y eq $x;
2460 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2462 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2463 command_noisy('read-tree', $treeish);
2464 $x = command_oneline('write-tree');
2465 if ($y ne $x) {
2466 ::fatal "trees ($treeish) $y != $x\n",
2467 "Something is seriously wrong...";
2468 }
2469 });
2470}
2471sub get_commit_parents {
2473 my ($self, $log_entry) = @_;
2474 my (%seen, @ret, @tmp);
2475 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2476 if (my $ip = $self->{inject_parents}) {
2477 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2478 push @tmp, $commit;
2479 }
2480 }
2481 if (my $cur = ::verify_ref($self->refname.'^0')) {
2482 push @tmp, $cur;
2483 }
2484 if (my $ipd = $self->{inject_parents_dcommit}) {
2485 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2486 push @tmp, @$commit;
2487 }
2488 }
2489 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2490 while (my $p = shift @tmp) {
2491 next if $seen{$p};
2492 $seen{$p} = 1;
2493 push @ret, $p;
2494 }
2495 @ret;
2496}
2497sub rewrite_root {
2499 my ($self) = @_;
2500 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2501 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2502 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2503 if ($rwr) {
2504 $rwr =~ s#/+$##;
2505 if ($rwr !~ m#^[a-z\+]+://#) {
2506 die "$rwr is not a valid URL (key: $k)\n";
2507 }
2508 }
2509 $self->{-rewrite_root} = $rwr;
2510}
2511sub rewrite_uuid {
2513 my ($self) = @_;
2514 return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2515 my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2516 my $rwid = eval { command_oneline(qw/config --get/, $k) };
2517 if ($rwid) {
2518 $rwid =~ s#/+$##;
2519 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2520 die "$rwid is not a valid UUID (key: $k)\n";
2521 }
2522 }
2523 $self->{-rewrite_uuid} = $rwid;
2524}
2525sub metadata_url {
2527 my ($self) = @_;
2528 ($self->rewrite_root || $self->{url}) .
2529 (length $self->{path} ? '/' . $self->{path} : '');
2530}
2531sub full_url {
2533 my ($self) = @_;
2534 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2535}
2536sub set_commit_header_env {
2539 my ($log_entry) = @_;
2540 my %env;
2541 foreach my $ned (qw/NAME EMAIL DATE/) {
2542 foreach my $ac (qw/AUTHOR COMMITTER/) {
2543 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2544 }
2545 }
2546 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2548 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2549 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2550 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2552 ? $log_entry->{commit_name}
2553 : $log_entry->{name};
2554 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2555 ? $log_entry->{commit_email}
2556 : $log_entry->{email};
2557 \%env;
2558}
2559sub restore_commit_header_env {
2561 my ($env) = @_;
2562 foreach my $ned (qw/NAME EMAIL DATE/) {
2563 foreach my $ac (qw/AUTHOR COMMITTER/) {
2564 my $k = "GIT_${ac}_${ned}";
2565 if (defined $env->{$k}) {
2566 $ENV{$k} = $env->{$k};
2567 } else {
2568 delete $ENV{$k};
2569 }
2570 }
2571 }
2572}
2573sub gc {
2575 command_noisy('gc', '--auto');
2576};
2577sub do_git_commit {
2579 my ($self, $log_entry) = @_;
2580 my $lr = $self->last_rev;
2581 if (defined $lr && $lr >= $log_entry->{revision}) {
2582 die "Last fetched revision of ", $self->refname,
2583 " was r$lr, but we are about to fetch: ",
2584 "r$log_entry->{revision}!\n";
2585 }
2586 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2587 croak "$log_entry->{revision} = $c already exists! ",
2588 "Why are we refetching it?\n";
2589 }
2590 my $old_env = set_commit_header_env($log_entry);
2591 my $tree = $log_entry->{tree};
2592 if (!defined $tree) {
2593 $tree = $self->tmp_index_do(sub {
2594 command_oneline('write-tree') });
2595 }
2596 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2597 my @exec = ('git', 'commit-tree', $tree);
2599 foreach ($self->get_commit_parents($log_entry)) {
2600 push @exec, '-p', $_;
2601 }
2602 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2603 or croak $!;
2604 binmode $msg_fh;
2605 # we always get UTF-8 from SVN, but we may want our commits in
2607 # a different encoding.
2608 if (my $enc = Git::config('i18n.commitencoding')) {
2609 require Encode;
2610 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2611 }
2612 print $msg_fh $log_entry->{log} or croak $!;
2613 restore_commit_header_env($old_env);
2614 unless ($self->no_metadata) {
2615 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2616 or croak $!;
2617 }
2618 $msg_fh->flush == 0 or croak $!;
2619 close $msg_fh or croak $!;
2620 chomp(my $commit = do { local $/; <$out_fh> });
2621 close $out_fh or croak $!;
2622 waitpid $pid, 0;
2623 croak $? if $?;
2624 if ($commit !~ /^$::sha1$/o) {
2625 die "Failed to commit, invalid sha1: $commit\n";
2626 }
2627 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2629 $self->{last_rev} = $log_entry->{revision};
2631 $self->{last_commit} = $commit;
2632 print "r$log_entry->{revision}" unless $::_q > 1;
2633 if (defined $log_entry->{svm_revision}) {
2634 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2635 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2636 0, $self->svm_uuid);
2637 }
2638 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2639 if (--$_gc_nr == 0) {
2640 $_gc_nr = $_gc_period;
2641 gc();
2642 }
2643 return $commit;
2644}
2645sub match_paths {
2647 my ($self, $paths, $r) = @_;
2648 return 1 if $self->{path} eq '';
2649 if (my $path = $paths->{"/$self->{path}"}) {
2650 return ($path->{action} eq 'D') ? 0 : 1;
2651 }
2652 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2653 if (grep /$self->{path_regex}/, keys %$paths) {
2654 return 1;
2655 }
2656 my $c = '';
2657 foreach (split m#/#, $self->{path}) {
2658 $c .= "/$_";
2659 next unless ($paths->{$c} &&
2660 ($paths->{$c}->{action} =~ /^[AR]$/));
2661 if ($self->ra->check_path($self->{path}, $r) ==
2662 $SVN::Node::dir) {
2663 return 1;
2664 }
2665 }
2666 return 0;
2667}
2668sub find_parent_branch {
2670 my ($self, $paths, $rev) = @_;
2671 return undef unless $self->follow_parent;
2672 unless (defined $paths) {
2673 my $err_handler = $SVN::Error::handler;
2674 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2675 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2676 sub { $paths = $_[0] });
2677 $SVN::Error::handler = $err_handler;
2678 }
2679 return undef unless defined $paths;
2680 # look for a parent from another branch:
2682 my @b_path_components = split m#/#, $self->{path};
2683 my @a_path_components;
2684 my $i;
2685 while (@b_path_components) {
2686 $i = $paths->{'/'.join('/', @b_path_components)};
2687 last if $i && defined $i->{copyfrom_path};
2688 unshift(@a_path_components, pop(@b_path_components));
2689 }
2690 return undef unless defined $i && defined $i->{copyfrom_path};
2691 my $branch_from = $i->{copyfrom_path};
2692 if (@a_path_components) {
2693 print STDERR "branch_from: $branch_from => ";
2694 $branch_from .= '/'.join('/', @a_path_components);
2695 print STDERR $branch_from, "\n";
2696 }
2697 my $r = $i->{copyfrom_rev};
2698 my $repos_root = $self->ra->{repos_root};
2699 my $url = $self->ra->{url};
2700 my $new_url = $url . $branch_from;
2701 print STDERR "Found possible branch point: ",
2702 "$new_url => ", $self->full_url, ", $r\n"
2703 unless $::_q > 1;
2704 $branch_from =~ s#^/##;
2705 my $gs = $self->other_gs($new_url, $url,
2706 $branch_from, $r, $self->{ref_id});
2707 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2708 {
2709 my ($base, $head);
2710 if (!defined $r0 || !defined $parent) {
2711 ($base, $head) = parse_revision_argument(0, $r);
2712 } else {
2713 if ($r0 < $r) {
2714 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2715 0, 1, sub { $base = $_[1] - 1 });
2716 }
2717 }
2718 if (defined $base && $base <= $r) {
2719 $gs->fetch($base, $r);
2720 }
2721 ($r0, $parent) = $gs->find_rev_before($r, 1);
2722 }
2723 if (defined $r0 && defined $parent) {
2724 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2725 unless $::_q > 1;
2726 my $ed;
2727 if ($self->ra->can_do_switch) {
2728 $self->assert_index_clean($parent);
2729 print STDERR "Following parent with do_switch\n"
2730 unless $::_q > 1;
2731 # do_switch works with svn/trunk >= r22312, but that
2732 # is not included with SVN 1.4.3 (the latest version
2733 # at the moment), so we can't rely on it
2734 $self->{last_rev} = $r0;
2735 $self->{last_commit} = $parent;
2736 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2737 $gs->ra->gs_do_switch($r0, $rev, $gs,
2738 $self->full_url, $ed)
2739 or die "SVN connection failed somewhere...\n";
2740 } elsif ($self->ra->trees_match($new_url, $r0,
2741 $self->full_url, $rev)) {
2742 print STDERR "Trees match:\n",
2743 " $new_url\@$r0\n",
2744 " ${\$self->full_url}\@$rev\n",
2745 "Following parent with no changes\n"
2746 unless $::_q > 1;
2747 $self->tmp_index_do(sub {
2748 command_noisy('read-tree', $parent);
2749 });
2750 $self->{last_commit} = $parent;
2751 } else {
2752 print STDERR "Following parent with do_update\n"
2753 unless $::_q > 1;
2754 $ed = SVN::Git::Fetcher->new($self);
2755 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2756 or die "SVN connection failed somewhere...\n";
2757 }
2758 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2759 return $self->make_log_entry($rev, [$parent], $ed);
2760 }
2761 return undef;
2762}
2763sub do_fetch {
2765 my ($self, $paths, $rev) = @_;
2766 my $ed;
2767 my ($last_rev, @parents);
2768 if (my $lc = $self->last_commit) {
2769 # we can have a branch that was deleted, then re-added
2770 # under the same name but copied from another path, in
2771 # which case we'll have multiple parents (we don't
2772 # want to break the original ref, nor lose copypath info):
2773 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2774 push @{$log_entry->{parents}}, $lc;
2775 return $log_entry;
2776 }
2777 $ed = SVN::Git::Fetcher->new($self);
2778 $last_rev = $self->{last_rev};
2779 $ed->{c} = $lc;
2780 @parents = ($lc);
2781 } else {
2782 $last_rev = $rev;
2783 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2784 return $log_entry;
2785 }
2786 $ed = SVN::Git::Fetcher->new($self);
2787 }
2788 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2789 die "SVN connection failed somewhere...\n";
2790 }
2791 $self->make_log_entry($rev, \@parents, $ed);
2792}
2793sub mkemptydirs {
2795 my ($self, $r) = @_;
2796 sub scan {
2798 my ($r, $empty_dirs, $line) = @_;
2799 if (defined $r && $line =~ /^r(\d+)$/) {
2800 return 0 if $1 > $r;
2801 } elsif ($line =~ /^ \+empty_dir: (.+)$/) {
2802 $empty_dirs->{$1} = 1;
2803 } elsif ($line =~ /^ \-empty_dir: (.+)$/) {
2804 my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
2805 delete @$empty_dirs{@d};
2806 }
2807 1; # continue
2808 };
2809 my %empty_dirs = ();
2811 my $gz_file = "$self->{dir}/unhandled.log.gz";
2812 if (-f $gz_file) {
2813 if (!$can_compress) {
2814 warn "Compress::Zlib could not be found; ",
2815 "empty directories in $gz_file will not be read\n";
2816 } else {
2817 my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
2818 die "Unable to open $gz_file: $!\n";
2819 my $line;
2820 while ($gz->gzreadline($line) > 0) {
2821 scan($r, \%empty_dirs, $line) or last;
2822 }
2823 $gz->gzclose;
2824 }
2825 }
2826 if (open my $fh, '<', "$self->{dir}/unhandled.log") {
2828 binmode $fh or croak "binmode: $!";
2829 while (<$fh>) {
2830 scan($r, \%empty_dirs, $_) or last;
2831 }
2832 close $fh;
2833 }
2834 my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2836 foreach my $d (sort keys %empty_dirs) {
2837 $d = uri_decode($d);
2838 $d =~ s/$strip//;
2839 next unless length($d);
2840 next if -d $d;
2841 if (-e $d) {
2842 warn "$d exists but is not a directory\n";
2843 } else {
2844 print "creating empty directory: $d\n";
2845 mkpath([$d]);
2846 }
2847 }
2848}
2849sub get_untracked {
2851 my ($self, $ed) = @_;
2852 my @out;
2853 my $h = $ed->{empty};
2854 foreach (sort keys %$h) {
2855 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2856 push @out, " $act: " . uri_encode($_);
2857 warn "W: $act: $_\n";
2858 }
2859 foreach my $t (qw/dir_prop file_prop/) {
2860 $h = $ed->{$t} or next;
2861 foreach my $path (sort keys %$h) {
2862 my $ppath = $path eq '' ? '.' : $path;
2863 foreach my $prop (sort keys %{$h->{$path}}) {
2864 next if $SKIP_PROP{$prop};
2865 my $v = $h->{$path}->{$prop};
2866 my $t_ppath_prop = "$t: " .
2867 uri_encode($ppath) . ' ' .
2868 uri_encode($prop);
2869 if (defined $v) {
2870 push @out, " +$t_ppath_prop " .
2871 uri_encode($v);
2872 } else {
2873 push @out, " -$t_ppath_prop";
2874 }
2875 }
2876 }
2877 }
2878 foreach my $t (qw/absent_file absent_directory/) {
2879 $h = $ed->{$t} or next;
2880 foreach my $parent (sort keys %$h) {
2881 foreach my $path (sort @{$h->{$parent}}) {
2882 push @out, " $t: " .
2883 uri_encode("$parent/$path");
2884 warn "W: $t: $parent/$path ",
2885 "Insufficient permissions?\n";
2886 }
2887 }
2888 }
2889 \@out;
2890}
2891# parse_svn_date(DATE)
2893# --------------------
2894# Given a date (in UTC) from Subversion, return a string in the format
2895# "<TZ Offset> <local date/time>" that Git will use.
2896#
2897# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2898# is true we'll convert it to the local timezone instead.
2899sub parse_svn_date {
2900 my $date = shift || return '+0000 1970-01-01 00:00:00';
2901 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2902 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2903 croak "Unable to parse date: $date\n";
2904 my $parsed_date; # Set next.
2905 if ($Git::SVN::_localtime) {
2907 # Translate the Subversion datetime to an epoch time.
2908 # Begin by switching ourselves to $date's timezone, UTC.
2909 my $old_env_TZ = $ENV{TZ};
2910 $ENV{TZ} = 'UTC';
2911 my $epoch_in_UTC =
2913 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2914 # Determine our local timezone (including DST) at the
2916 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2917 # value of TZ, if any, at the time we were run.
2918 if (defined $Git::SVN::Log::TZ) {
2919 $ENV{TZ} = $Git::SVN::Log::TZ;
2920 } else {
2921 delete $ENV{TZ};
2922 }
2923 my $our_TZ =
2925 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2926 # This converts $epoch_in_UTC into our local timezone.
2928 my ($sec, $min, $hour, $mday, $mon, $year,
2929 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2930 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2932 $our_TZ, $year + 1900, $mon + 1,
2933 $mday, $hour, $min, $sec);
2934 # Reset us to the timezone in effect when we entered
2936 # this routine.
2937 if (defined $old_env_TZ) {
2938 $ENV{TZ} = $old_env_TZ;
2939 } else {
2940 delete $ENV{TZ};
2941 }
2942 } else {
2943 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2944 }
2945 return $parsed_date;
2947}
2948sub other_gs {
2950 my ($self, $new_url, $url,
2951 $branch_from, $r, $old_ref_id) = @_;
2952 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2953 unless ($gs) {
2954 my $ref_id = $old_ref_id;
2955 $ref_id =~ s/\@\d+$//;
2956 $ref_id .= "\@$r";
2957 # just grow a tail if we're not unique enough :x
2958 $ref_id .= '-' while find_ref($ref_id);
2959 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2960 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2961 if ($u =~ s#^\Q$url\E(/|$)##) {
2962 $p = $u;
2963 $u = $url;
2964 $repo_id = $self->{repo_id};
2965 }
2966 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2967 }
2968 $gs
2969}
2970sub call_authors_prog {
2972 my ($orig_author) = @_;
2973 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2974 my $author = `$::_authors_prog $orig_author`;
2975 if ($? != 0) {
2976 die "$::_authors_prog failed with exit code $?\n"
2977 }
2978 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2979 my ($name, $email) = ($1, $2);
2980 $email = undef if length $2 == 0;
2981 return [$name, $email];
2982 } else {
2983 die "Author: $orig_author: $::_authors_prog returned "
2984 . "invalid author format: $author\n";
2985 }
2986}
2987sub check_author {
2989 my ($author) = @_;
2990 if (!defined $author || length $author == 0) {
2991 $author = '(no author)';
2992 }
2993 if (!defined $::users{$author}) {
2994 if (defined $::_authors_prog) {
2995 $::users{$author} = call_authors_prog($author);
2996 } elsif (defined $::_authors) {
2997 die "Author: $author not defined in $::_authors file\n";
2998 }
2999 }
3000 $author;
3001}
3002sub find_extra_svk_parents {
3004 my ($self, $ed, $tickets, $parents) = @_;
3005 # aha! svk:merge property changed...
3006 my @tickets = split "\n", $tickets;
3007 my @known_parents;
3008 for my $ticket ( @tickets ) {
3009 my ($uuid, $path, $rev) = split /:/, $ticket;
3010 if ( $uuid eq $self->ra_uuid ) {
3011 my $url = $self->{url};
3012 my $repos_root = $url;
3013 my $branch_from = $path;
3014 $branch_from =~ s{^/}{};
3015 my $gs = $self->other_gs($repos_root."/".$branch_from,
3016 $url,
3017 $branch_from,
3018 $rev,
3019 $self->{ref_id});
3020 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3021 # wahey! we found it, but it might be
3022 # an old one (!)
3023 push @known_parents, [ $rev, $commit ];
3024 }
3025 }
3026 }
3027 # Ordering matters; highest-numbered commit merge tickets
3028 # first, as they may account for later merge ticket additions
3029 # or changes.
3030 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3031 for my $parent ( @known_parents ) {
3032 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3033 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3034 my $new;
3035 while ( <$msg_fh> ) {
3036 $new=1;last;
3037 }
3038 command_close_pipe($msg_fh, $ctx);
3039 if ( $new ) {
3040 print STDERR
3041 "Found merge parent (svk:merge ticket): $parent\n";
3042 push @$parents, $parent;
3043 }
3044 }
3045}
3046sub lookup_svn_merge {
3048 my $uuid = shift;
3049 my $url = shift;
3050 my $merge = shift;
3051 my ($source, $revs) = split ":", $merge;
3053 my $path = $source;
3054 $path =~ s{^/}{};
3055 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3056 if ( !$gs ) {
3057 warn "Couldn't find revmap for $url$source\n";
3058 return;
3059 }
3060 my @ranges = split ",", $revs;
3061 my ($tip, $tip_commit);
3062 my @merged_commit_ranges;
3063 # find the tip
3064 for my $range ( @ranges ) {
3065 my ($bottom, $top) = split "-", $range;
3066 $top ||= $bottom;
3067 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3068 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3069 unless ($top_commit and $bottom_commit) {
3071 warn "W:unknown path/rev in svn:mergeinfo "
3072 ."dirprop: $source:$range\n";
3073 next;
3074 }
3075 push @merged_commit_ranges,
3077 "$bottom_commit^..$top_commit";
3078 if ( !defined $tip or $top > $tip ) {
3080 $tip = $top;
3081 $tip_commit = $top_commit;
3082 }
3083 }
3084 return ($tip_commit, @merged_commit_ranges);
3085}
3086sub _rev_list {
3088 my ($msg_fh, $ctx) = command_output_pipe(
3089 "rev-list", @_,
3090 );
3091 my @rv;
3092 while ( <$msg_fh> ) {
3093 chomp;
3094 push @rv, $_;
3095 }
3096 command_close_pipe($msg_fh, $ctx);
3097 @rv;
3098}
3099sub check_cherry_pick {
3101 my $base = shift;
3102 my $tip = shift;
3103 my @ranges = @_;
3104 my %commits = map { $_ => 1 }
3105 _rev_list("--no-merges", $tip, "--not", $base);
3106 for my $range ( @ranges ) {
3107 delete @commits{_rev_list($range)};
3108 }
3109 for my $commit (keys %commits) {
3110 if (has_no_changes($commit)) {
3111 delete $commits{$commit};
3112 }
3113 }
3114 return (keys %commits);
3115}
3116sub has_no_changes {
3118 my $commit = shift;
3119 my @revs = split / /, command_oneline(
3121 qw(rev-list --parents -1 -m), $commit);
3122 # Commits with no parents, e.g. the start of a partial branch,
3124 # have changes by definition.
3125 return 1 if (@revs < 2);
3126 # Commits with multiple parents, e.g a merge, have no changes
3128 # by definition.
3129 return 0 if (@revs > 2);
3130 return (command_oneline("rev-parse", "$commit^{tree}") eq
3132 command_oneline("rev-parse", "$commit~1^{tree}"));
3133}
3134# The GIT_DIR environment variable is not always set until after the command
3136# line arguments are processed, so we can't memoize in a BEGIN block.
3137{
3138 my $memoized = 0;
3139 sub memoize_svn_mergeinfo_functions {
3141 return if $memoized;
3142 $memoized = 1;
3143 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3145 mkpath([$cache_path]) unless -d $cache_path;
3146 tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3148 "$cache_path/lookup_svn_merge.db", 'nstore';
3149 memoize 'lookup_svn_merge',
3150 SCALAR_CACHE => 'FAULT',
3151 LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3152 ;
3153 tie my %check_cherry_pick_cache => 'Memoize::Storable',
3155 "$cache_path/check_cherry_pick.db", 'nstore';
3156 memoize 'check_cherry_pick',
3157 SCALAR_CACHE => 'FAULT',
3158 LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3159 ;
3160 tie my %has_no_changes_cache => 'Memoize::Storable',
3162 "$cache_path/has_no_changes.db", 'nstore';
3163 memoize 'has_no_changes',
3164 SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3165 LIST_CACHE => 'FAULT',
3166 ;
3167 }
3168}
3169sub parents_exclude {
3171 my $parents = shift;
3172 my @commits = @_;
3173 return unless @commits;
3174 my @excluded;
3176 my $excluded;
3177 do {
3178 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3179 $excluded = command_oneline(@cmd);
3180 if ( $excluded ) {
3181 my @new;
3182 my $found;
3183 for my $commit ( @commits ) {
3184 if ( $commit eq $excluded ) {
3185 push @excluded, $commit;
3186 $found++;
3187 last;
3188 }
3189 else {
3190 push @new, $commit;
3191 }
3192 }
3193 die "saw commit '$excluded' in rev-list output, "
3194 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3195 unless $found;
3196 @commits = @new;
3197 }
3198 }
3199 while ($excluded and @commits);
3200 return @excluded;
3202}
3203# note: this function should only be called if the various dirprops
3206# have actually changed
3207sub find_extra_svn_parents {
3208 my ($self, $ed, $mergeinfo, $parents) = @_;
3209 # aha! svk:merge property changed...
3210 memoize_svn_mergeinfo_functions();
3212 # We first search for merged tips which are not in our
3214 # history. Then, we figure out which git revisions are in
3215 # that tip, but not this revision. If all of those revisions
3216 # are now marked as merge, we can add the tip as a parent.
3217 my @merges = split "\n", $mergeinfo;
3218 my @merge_tips;
3219 my $url = $self->{url};
3220 my $uuid = $self->ra_uuid;
3221 my %ranges;
3222 for my $merge ( @merges ) {
3223 my ($tip_commit, @ranges) =
3224 lookup_svn_merge( $uuid, $url, $merge );
3225 unless (!$tip_commit or
3226 grep { $_ eq $tip_commit } @$parents ) {
3227 push @merge_tips, $tip_commit;
3228 $ranges{$tip_commit} = \@ranges;
3229 } else {
3230 push @merge_tips, undef;
3231 }
3232 }
3233 my %excluded = map { $_ => 1 }
3235 parents_exclude($parents, grep { defined } @merge_tips);
3236 # check merge tips for new parents
3238 my @new_parents;
3239 for my $merge_tip ( @merge_tips ) {
3240 my $spec = shift @merges;
3241 next unless $merge_tip and $excluded{$merge_tip};
3242 my $ranges = $ranges{$merge_tip};
3244 # check out 'new' tips
3246 my $merge_base;
3247 eval {
3248 $merge_base = command_oneline(
3249 "merge-base",
3250 @$parents, $merge_tip,
3251 );
3252 };
3253 if ($@) {
3254 die "An error occurred during merge-base"
3255 unless $@->isa("Git::Error::Command");
3256 warn "W: Cannot find common ancestor between ".
3258 "@$parents and $merge_tip. Ignoring merge info.\n";
3259 next;
3260 }
3261 # double check that there are no missing non-merge commits
3263 my (@incomplete) = check_cherry_pick(
3264 $merge_base, $merge_tip,
3265 @$ranges,
3266 );
3267 if ( @incomplete ) {
3269 warn "W:svn cherry-pick ignored ($spec) - missing "
3270 .@incomplete." commit(s) (eg $incomplete[0])\n";
3271 } else {
3272 warn
3273 "Found merge parent (svn:mergeinfo prop): ",
3274 $merge_tip, "\n";
3275 push @new_parents, $merge_tip;
3276 }
3277 }
3278 # cater for merges which merge commits from multiple branches
3280 if ( @new_parents > 1 ) {
3281 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3282 for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3283 next if $i == $j;
3284 next unless $new_parents[$i];
3285 next unless $new_parents[$j];
3286 my $revs = command_oneline(
3287 "rev-list", "-1",
3288 "$new_parents[$i]..$new_parents[$j]",
3289 );
3290 if ( !$revs ) {
3291 undef($new_parents[$j]);
3292 }
3293 }
3294 }
3295 }
3296 push @$parents, grep { defined } @new_parents;
3297}
3298sub make_log_entry {
3300 my ($self, $rev, $parents, $ed) = @_;
3301 my $untracked = $self->get_untracked($ed);
3302 my @parents = @$parents;
3304 my $ps = $ed->{path_strip} || "";
3305 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3306 my $props = $ed->{dir_prop}{$path};
3307 if ( $props->{"svk:merge"} ) {
3308 $self->find_extra_svk_parents
3309 ($ed, $props->{"svk:merge"}, \@parents);
3310 }
3311 if ( $props->{"svn:mergeinfo"} ) {
3312 $self->find_extra_svn_parents
3313 ($ed,
3314 $props->{"svn:mergeinfo"},
3315 \@parents);
3316 }
3317 }
3318 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3320 print $un "r$rev\n" or croak $!;
3321 print $un $_, "\n" foreach @$untracked;
3322 my %log_entry = ( parents => \@parents, revision => $rev,
3323 log => '');
3324 my $headrev;
3326 my $logged = delete $self->{logged_rev_props};
3327 if (!$logged || $self->{-want_revprops}) {
3328 my $rp = $self->ra->rev_proplist($rev);
3329 foreach (sort keys %$rp) {
3330 my $v = $rp->{$_};
3331 if (/^svn:(author|date|log)$/) {
3332 $log_entry{$1} = $v;
3333 } elsif ($_ eq 'svm:headrev') {
3334 $headrev = $v;
3335 } else {
3336 print $un " rev_prop: ", uri_encode($_), ' ',
3337 uri_encode($v), "\n";
3338 }
3339 }
3340 } else {
3341 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3342 }
3343 close $un or croak $!;
3344 $log_entry{date} = parse_svn_date($log_entry{date});
3346 $log_entry{log} .= "\n";
3347 my $author = $log_entry{author} = check_author($log_entry{author});
3348 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3349 : ($author, undef);
3350 my ($commit_name, $commit_email) = ($name, $email);
3352 if ($_use_log_author) {
3353 my $name_field;
3354 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3355 $name_field = $1;
3356 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3357 $name_field = $1;
3358 }
3359 if (!defined $name_field) {
3360 if (!defined $email) {
3361 $email = $name;
3362 }
3363 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3364 ($name, $email) = ($1, $2);
3365 } elsif ($name_field =~ /(.*)@/) {
3366 ($name, $email) = ($1, $name_field);
3367 } else {
3368 ($name, $email) = ($name_field, $name_field);
3369 }
3370 }
3371 if (defined $headrev && $self->use_svm_props) {
3372 if ($self->rewrite_root) {
3373 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3374 "options set!\n";
3375 }
3376 if ($self->rewrite_uuid) {
3377 die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3378 "options set!\n";
3379 }
3380 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3381 # we don't want "SVM: initializing mirror for junk" ...
3382 return undef if $r == 0;
3383 my $svm = $self->svm;
3384 if ($uuid ne $svm->{uuid}) {
3385 die "UUID mismatch on SVM path:\n",
3386 "expected: $svm->{uuid}\n",
3387 " got: $uuid\n";
3388 }
3389 my $full_url = $self->full_url;
3390 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3391 die "Failed to replace '$svm->{replace}' with ",
3392 "'$svm->{source}' in $full_url\n";
3393 # throw away username for storing in records
3394 remove_username($full_url);
3395 $log_entry{metadata} = "$full_url\@$r $uuid";
3396 $log_entry{svm_revision} = $r;
3397 $email ||= "$author\@$uuid";
3398 $commit_email ||= "$author\@$uuid";
3399 } elsif ($self->use_svnsync_props) {
3400 my $full_url = $self->svnsync->{url};
3401 $full_url .= "/$self->{path}" if length $self->{path};
3402 remove_username($full_url);
3403 my $uuid = $self->svnsync->{uuid};
3404 $log_entry{metadata} = "$full_url\@$rev $uuid";
3405 $email ||= "$author\@$uuid";
3406 $commit_email ||= "$author\@$uuid";
3407 } else {
3408 my $url = $self->metadata_url;
3409 remove_username($url);
3410 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3411 $log_entry{metadata} = "$url\@$rev " . $uuid;
3412 $email ||= "$author\@" . $uuid;
3413 $commit_email ||= "$author\@" . $uuid;
3414 }
3415 $log_entry{name} = $name;
3416 $log_entry{email} = $email;
3417 $log_entry{commit_name} = $commit_name;
3418 $log_entry{commit_email} = $commit_email;
3419 \%log_entry;
3420}
3421sub fetch {
3423 my ($self, $min_rev, $max_rev, @parents) = @_;
3424 my ($last_rev, $last_commit) = $self->last_rev_commit;
3425 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3426 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3427}
3428sub set_tree_cb {
3430 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3431 $self->{inject_parents} = { $rev => $tree };
3432 $self->fetch(undef, undef);
3433}
3434sub set_tree {
3436 my ($self, $tree) = (shift, shift);
3437 my $log_entry = ::get_commit_entry($tree);
3438 unless ($self->{last_rev}) {
3439 ::fatal("Must have an existing revision to commit");
3440 }
3441 my %ed_opts = ( r => $self->{last_rev},
3442 log => $log_entry->{log},
3443 ra => $self->ra,
3444 tree_a => $self->{last_commit},
3445 tree_b => $tree,
3446 editor_cb => sub {
3447 $self->set_tree_cb($log_entry, $tree, @_) },
3448 svn_path => $self->{path} );
3449 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3450 print "No changes\nr$self->{last_rev} = $tree\n";
3451 }
3452}
3453sub rebuild_from_rev_db {
3455 my ($self, $path) = @_;
3456 my $r = -1;
3457 open my $fh, '<', $path or croak "open: $!";
3458 binmode $fh or croak "binmode: $!";
3459 while (<$fh>) {
3460 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3461 chomp($_);
3462 ++$r;
3463 next if $_ eq ('0' x 40);
3464 $self->rev_map_set($r, $_);
3465 print "r$r = $_\n";
3466 }
3467 close $fh or croak "close: $!";
3468 unlink $path or croak "unlink: $!";
3469}
3470sub rebuild {
3472 my ($self) = @_;
3473 my $map_path = $self->map_path;
3474 my $partial = (-e $map_path && ! -z $map_path);
3475 return unless ::verify_ref($self->refname.'^0');
3476 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3477 my $rev_db = $self->rev_db_path;
3478 $self->rebuild_from_rev_db($rev_db);
3479 if ($self->use_svm_props) {
3480 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3481 $self->rebuild_from_rev_db($svm_rev_db);
3482 }
3483 $self->unlink_rev_db_symlink;
3484 return;
3485 }
3486 print "Rebuilding $map_path ...\n" if (!$partial);
3487 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3488 (undef, undef));
3489 my ($log, $ctx) =
3490 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3491 ($head ? "$head.." : "") . $self->refname,
3492 '--');
3493 my $metadata_url = $self->metadata_url;
3494 remove_username($metadata_url);
3495 my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3496 my $c;
3497 while (<$log>) {
3498 if ( m{^commit ($::sha1)$} ) {
3499 $c = $1;
3500 next;
3501 }
3502 next unless s{^\s*(git-svn-id:)}{$1};
3503 my ($url, $rev, $uuid) = ::extract_metadata($_);
3504 remove_username($url);
3505 # ignore merges (from set-tree)
3507 next if (!defined $rev || !$uuid);
3508 # if we merged or otherwise started elsewhere, this is
3510 # how we break out of it
3511 if (($uuid ne $svn_uuid) ||
3512 ($metadata_url && $url && ($url ne $metadata_url))) {
3513 next;
3514 }
3515 if ($partial && $head) {
3516 print "Partial-rebuilding $map_path ...\n";
3517 print "Currently at $base_rev = $head\n";
3518 $head = undef;
3519 }
3520 $self->rev_map_set($rev, $c);
3522 print "r$rev = $c\n";
3523 }
3524 command_close_pipe($log, $ctx);
3525 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3526 my $rev_db_path = $self->rev_db_path;
3527 if (-f $self->rev_db_path) {
3528 unlink $self->rev_db_path or croak "unlink: $!";
3529 }
3530 $self->unlink_rev_db_symlink;
3531}
3532# rev_map:
3534# Tie::File seems to be prone to offset errors if revisions get sparse,
3535# it's not that fast, either. Tie::File is also not in Perl 5.6. So
3536# one of my favorite modules is out :< Next up would be one of the DBM
3537# modules, but I'm not sure which is most portable...
3538#
3539# This is the replacement for the rev_db format, which was too big
3540# and inefficient for large repositories with a lot of sparse history
3541# (mainly tags)
3542#
3543# The format is this:
3544# - 24 bytes for every record,
3545# * 4 bytes for the integer representing an SVN revision number
3546# * 20 bytes representing the sha1 of a git commit
3547# - No empty padding records like the old format
3548# (except the last record, which can be overwritten)
3549# - new records are written append-only since SVN revision numbers
3550# increase monotonically
3551# - lookups on SVN revision number are done via a binary search
3552# - Piping the file to xxd -c24 is a good way of dumping it for
3553# viewing or editing (piped back through xxd -r), should the need
3554# ever arise.
3555# - The last record can be padding revision with an all-zero sha1
3556# This is used to optimize fetch performance when using multiple
3557# "fetch" directives in .git/config
3558#
3559# These files are disposable unless noMetadata or useSvmProps is set
3560sub _rev_map_set {
3562 my ($fh, $rev, $commit) = @_;
3563 binmode $fh or croak "binmode: $!";
3565 my $size = (stat($fh))[7];
3566 ($size % 24) == 0 or croak "inconsistent size: $size";
3567 my $wr_offset = 0;
3569 if ($size > 0) {
3570 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3571 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3572 $read == 24 or croak "read only $read bytes (!= 24)";
3573 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3574 if ($last_commit eq ('0' x40)) {
3575 if ($size >= 48) {
3576 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3577 $read = sysread($fh, $buf, 24) or
3578 croak "read: $!";
3579 $read == 24 or
3580 croak "read only $read bytes (!= 24)";
3581 ($last_rev, $last_commit) =
3582 unpack(rev_map_fmt, $buf);
3583 if ($last_commit eq ('0' x40)) {
3584 croak "inconsistent .rev_map\n";
3585 }
3586 }
3587 if ($last_rev >= $rev) {
3588 croak "last_rev is higher!: $last_rev >= $rev";
3589 }
3590 $wr_offset = -24;
3591 }
3592 }
3593 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3594 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3595 croak "write: $!";
3596}
3597sub _rev_map_reset {
3599 my ($fh, $rev, $commit) = @_;
3600 my $c = _rev_map_get($fh, $rev);
3601 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3602 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3603 truncate $fh, $offset or croak "truncate: $!";
3604}
3605sub mkfile {
3607 my ($path) = @_;
3608 unless (-e $path) {
3609 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3610 mkpath([$dir]) unless -d $dir;
3611 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3612 close $fh or die "Couldn't close (create) $path: $!\n";
3613 }
3614}
3615sub rev_map_set {
3617 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3618 defined $commit or die "missing arg3\n";
3619 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3620 my $db = $self->map_path($uuid);
3621 my $db_lock = "$db.lock";
3622 my $sig;
3623 $update_ref ||= 0;
3624 if ($update_ref) {
3625 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3626 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3627 }
3628 mkfile($db);
3629 $LOCKFILES{$db_lock} = 1;
3631 my $sync;
3632 # both of these options make our .rev_db file very, very important
3633 # and we can't afford to lose it because rebuild() won't work
3634 if ($self->use_svm_props || $self->no_metadata) {
3635 $sync = 1;
3636 copy($db, $db_lock) or die "rev_map_set(@_): ",
3637 "Failed to copy: ",
3638 "$db => $db_lock ($!)\n";
3639 } else {
3640 rename $db, $db_lock or die "rev_map_set(@_): ",
3641 "Failed to rename: ",
3642 "$db => $db_lock ($!)\n";
3643 }
3644 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3646 or croak "Couldn't open $db_lock: $!\n";
3647 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3648 _rev_map_set($fh, $rev, $commit);
3649 if ($sync) {
3650 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3651 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3652 }
3653 close $fh or croak $!;
3654 if ($update_ref) {
3655 $_head = $self;
3656 my $note = "";
3657 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3658 command_noisy('update-ref', '-m', "r$rev$note",
3659 $self->refname, $commit);
3660 }
3661 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3662 "$db_lock => $db ($!)\n";
3663 delete $LOCKFILES{$db_lock};
3664 if ($update_ref) {
3665 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3666 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3667 kill $sig, $$ if defined $sig;
3668 }
3669}
3670# If want_commit, this will return an array of (rev, commit) where
3672# commit _must_ be a valid commit in the archive.
3673# Otherwise, it'll return the max revision (whether or not the
3674# commit is valid or just a 0x40 placeholder).
3675sub rev_map_max {
3676 my ($self, $want_commit) = @_;
3677 $self->rebuild;
3678 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3679 $want_commit ? ($r, $c) : $r;
3680}
3681sub rev_map_max_norebuild {
3683 my ($self, $want_commit) = @_;
3684 my $map_path = $self->map_path;
3685 stat $map_path or return $want_commit ? (0, undef) : 0;
3686 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3687 binmode $fh or croak "binmode: $!";
3688 my $size = (stat($fh))[7];
3689 ($size % 24) == 0 or croak "inconsistent size: $size";
3690 if ($size == 0) {
3692 close $fh or croak "close: $!";
3693 return $want_commit ? (0, undef) : 0;
3694 }
3695 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3697 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3698 my ($r, $c) = unpack(rev_map_fmt, $buf);
3699 if ($want_commit && $c eq ('0' x40)) {
3700 if ($size < 48) {
3701 return $want_commit ? (0, undef) : 0;
3702 }
3703 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3704 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3705 ($r, $c) = unpack(rev_map_fmt, $buf);
3706 if ($c eq ('0'x40)) {
3707 croak "Penultimate record is all-zeroes in $map_path";
3708 }
3709 }
3710 close $fh or croak "close: $!";
3711 $want_commit ? ($r, $c) : $r;
3712}
3713sub rev_map_get {
3715 my ($self, $rev, $uuid) = @_;
3716 my $map_path = $self->map_path($uuid);
3717 return undef unless -e $map_path;
3718 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3720 my $c = _rev_map_get($fh, $rev);
3721 close($fh) or croak "close: $!";
3722 $c
3723}
3724sub _rev_map_get {
3726 my ($fh, $rev) = @_;
3727 binmode $fh or croak "binmode: $!";
3729 my $size = (stat($fh))[7];
3730 ($size % 24) == 0 or croak "inconsistent size: $size";
3731 if ($size == 0) {
3733 return undef;
3734 }
3735 my ($l, $u) = (0, $size - 24);
3737 my ($r, $c, $buf);
3738 while ($l <= $u) {
3740 my $i = int(($l/24 + $u/24) / 2) * 24;
3741 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3742 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3743 my ($r, $c) = unpack(rev_map_fmt, $buf);
3744 if ($r < $rev) {
3746 $l = $i + 24;
3747 } elsif ($r > $rev) {
3748 $u = $i - 24;
3749 } else { # $r == $rev
3750 return $c eq ('0' x 40) ? undef : $c;
3751 }
3752 }
3753 undef;
3754}
3755# Finds the first svn revision that exists on (if $eq_ok is true) or
3757# before $rev for the current branch. It will not search any lower
3758# than $min_rev. Returns the git commit hash and svn revision number
3759# if found, else (undef, undef).
3760sub find_rev_before {
3761 my ($self, $rev, $eq_ok, $min_rev) = @_;
3762 --$rev unless $eq_ok;
3763 $min_rev ||= 1;
3764 my $max_rev = $self->rev_map_max;
3765 $rev = $max_rev if ($rev > $max_rev);
3766 while ($rev >= $min_rev) {
3767 if (my $c = $self->rev_map_get($rev)) {
3768 return ($rev, $c);
3769 }
3770 --$rev;
3771 }
3772 return (undef, undef);
3773}
3774# Finds the first svn revision that exists on (if $eq_ok is true) or
3776# after $rev for the current branch. It will not search any higher
3777# than $max_rev. Returns the git commit hash and svn revision number
3778# if found, else (undef, undef).
3779sub find_rev_after {
3780 my ($self, $rev, $eq_ok, $max_rev) = @_;
3781 ++$rev unless $eq_ok;
3782 $max_rev ||= $self->rev_map_max;
3783 while ($rev <= $max_rev) {
3784 if (my $c = $self->rev_map_get($rev)) {
3785 return ($rev, $c);
3786 }
3787 ++$rev;
3788 }
3789 return (undef, undef);
3790}
3791sub _new {
3793 my ($class, $repo_id, $ref_id, $path) = @_;
3794 unless (defined $repo_id && length $repo_id) {
3795 $repo_id = $Git::SVN::default_repo_id;
3796 }
3797 unless (defined $ref_id && length $ref_id) {
3798 $_prefix = '' unless defined($_prefix);
3799 $_[2] = $ref_id =
3800 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3801 }
3802 $_[1] = $repo_id;
3803 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3804 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3806 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3807 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3808 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3809 if (-d $old_dir && ! -d $dir) {
3810 $dir = $old_dir;
3811 }
3812 }
3813 $_[3] = $path = '' unless (defined $path);
3815 mkpath([$dir]);
3816 bless {
3817 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3818 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3819 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3820}
3821# for read-only access of old .rev_db formats
3823sub unlink_rev_db_symlink {
3824 my ($self) = @_;
3825 my $link = $self->rev_db_path;
3826 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3827 if (-l $link) {
3828 unlink $link or croak "unlink: $link failed!";
3829 }
3830}
3831sub rev_db_path {
3833 my ($self, $uuid) = @_;
3834 my $db_path = $self->map_path($uuid);
3835 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3836 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3837 $db_path;
3838}
3839# the new replacement for .rev_db
3841sub map_path {
3842 my ($self, $uuid) = @_;
3843 $uuid ||= $self->ra_uuid;
3844 "$self->{map_root}.$uuid";
3845}
3846sub uri_encode {
3848 my ($f) = @_;
3849 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3850 $f
3851}
3852sub uri_decode {
3854 my ($f) = @_;
3855 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3856 $f
3857}
3858sub remove_username {
3860 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3861}
3862package Git::SVN::Prompt;
3864use strict;
3865use warnings;
3866require SVN::Core;
3867use vars qw/$_no_auth_cache $_username/;
3868sub simple {
3870 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3871 $may_save = undef if $_no_auth_cache;
3872 $default_username = $_username if defined $_username;
3873 if (defined $default_username && length $default_username) {
3874 if (defined $realm && length $realm) {
3875 print STDERR "Authentication realm: $realm\n";
3876 STDERR->flush;
3877 }
3878 $cred->username($default_username);
3879 } else {
3880 username($cred, $realm, $may_save, $pool);
3881 }
3882 $cred->password(_read_password("Password for '" .
3883 $cred->username . "': ", $realm));
3884 $cred->may_save($may_save);
3885 $SVN::_Core::SVN_NO_ERROR;
3886}
3887sub ssl_server_trust {
3889 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3890 $may_save = undef if $_no_auth_cache;
3891 print STDERR "Error validating server certificate for '$realm':\n";
3892 {
3893 no warnings 'once';
3894 # All variables SVN::Auth::SSL::* are used only once,
3895 # so we're shutting up Perl warnings about this.
3896 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3897 print STDERR " - The certificate is not issued ",
3898 "by a trusted authority. Use the\n",
3899 " fingerprint to validate ",
3900 "the certificate manually!\n";
3901 }
3902 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3903 print STDERR " - The certificate hostname ",
3904 "does not match.\n";
3905 }
3906 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3907 print STDERR " - The certificate is not yet valid.\n";
3908 }
3909 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3910 print STDERR " - The certificate has expired.\n";
3911 }
3912 if ($failures & $SVN::Auth::SSL::OTHER) {
3913 print STDERR " - The certificate has ",
3914 "an unknown error.\n";
3915 }
3916 } # no warnings 'once'
3917 printf STDERR
3918 "Certificate information:\n".
3919 " - Hostname: %s\n".
3920 " - Valid: from %s until %s\n".
3921 " - Issuer: %s\n".
3922 " - Fingerprint: %s\n",
3923 map $cert_info->$_, qw(hostname valid_from valid_until
3924 issuer_dname fingerprint);
3925 my $choice;
3926prompt:
3927 print STDERR $may_save ?
3928 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3929 "(R)eject or accept (t)emporarily? ";
3930 STDERR->flush;
3931 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3932 if ($choice =~ /^t$/i) {
3933 $cred->may_save(undef);
3934 } elsif ($choice =~ /^r$/i) {
3935 return -1;
3936 } elsif ($may_save && $choice =~ /^p$/i) {
3937 $cred->may_save($may_save);
3938 } else {
3939 goto prompt;
3940 }
3941 $cred->accepted_failures($failures);
3942 $SVN::_Core::SVN_NO_ERROR;
3943}
3944sub ssl_client_cert {
3946 my ($cred, $realm, $may_save, $pool) = @_;
3947 $may_save = undef if $_no_auth_cache;
3948 print STDERR "Client certificate filename: ";
3949 STDERR->flush;
3950 chomp(my $filename = <STDIN>);
3951 $cred->cert_file($filename);
3952 $cred->may_save($may_save);
3953 $SVN::_Core::SVN_NO_ERROR;
3954}
3955sub ssl_client_cert_pw {
3957 my ($cred, $realm, $may_save, $pool) = @_;
3958 $may_save = undef if $_no_auth_cache;
3959 $cred->password(_read_password("Password: ", $realm));
3960 $cred->may_save($may_save);
3961 $SVN::_Core::SVN_NO_ERROR;
3962}
3963sub username {
3965 my ($cred, $realm, $may_save, $pool) = @_;
3966 $may_save = undef if $_no_auth_cache;
3967 if (defined $realm && length $realm) {
3968 print STDERR "Authentication realm: $realm\n";
3969 }
3970 my $username;
3971 if (defined $_username) {
3972 $username = $_username;
3973 } else {
3974 print STDERR "Username: ";
3975 STDERR->flush;
3976 chomp($username = <STDIN>);
3977 }
3978 $cred->username($username);
3979 $cred->may_save($may_save);
3980 $SVN::_Core::SVN_NO_ERROR;
3981}
3982sub _read_password {
3984 my ($prompt, $realm) = @_;
3985 my $password = '';
3986 if (exists $ENV{GIT_ASKPASS}) {
3987 open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
3988 $password = <PH>;
3989 $password =~ s/[\012\015]//; # \n\r
3990 close(PH);
3991 } else {
3992 print STDERR $prompt;
3993 STDERR->flush;
3994 require Term::ReadKey;
3995 Term::ReadKey::ReadMode('noecho');
3996 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3997 last if $key =~ /[\012\015]/; # \n\r
3998 $password .= $key;
3999 }
4000 Term::ReadKey::ReadMode('restore');
4001 print STDERR "\n";
4002 STDERR->flush;
4003 }
4004 $password;
4005}
4006package SVN::Git::Fetcher;
4008use vars qw/@ISA/;
4009use strict;
4010use warnings;
4011use Carp qw/croak/;
4012use IO::File qw//;
4013use vars qw/$_ignore_regex/;
4014# file baton members: path, mode_a, mode_b, pool, fh, blob, base
4016sub new {
4017 my ($class, $git_svn, $switch_path) = @_;
4018 my $self = SVN::Delta::Editor->new;
4019 bless $self, $class;
4020 if (exists $git_svn->{last_commit}) {
4021 $self->{c} = $git_svn->{last_commit};
4022 $self->{empty_symlinks} =
4023 _mark_empty_symlinks($git_svn, $switch_path);
4024 }
4025 $self->{ignore_regex} = eval { command_oneline('config', '--get',
4026 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
4027 $self->{empty} = {};
4028 $self->{dir_prop} = {};
4029 $self->{file_prop} = {};
4030 $self->{absent_dir} = {};
4031 $self->{absent_file} = {};
4032 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4033 $self;
4034}
4035# this uses the Ra object, so it must be called before do_{switch,update},
4037# not inside them (when the Git::SVN::Fetcher object is passed) to
4038# do_{switch,update}
4039sub _mark_empty_symlinks {
4040 my ($git_svn, $switch_path) = @_;
4041 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4042 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4043 my %ret;
4045 my ($rev, $cmt) = $git_svn->last_rev_commit;
4046 return {} unless ($rev && $cmt);
4047 # allow the warning to be printed for each revision we fetch to
4049 # ensure the user sees it. The user can also disable the workaround
4050 # on the repository even while git svn is running and the next
4051 # revision fetched will skip this expensive function.
4052 my $printed_warning;
4053 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4054 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4055 local $/ = "\0";
4056 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4057 $pfx .= '/' if length($pfx);
4058 while (<$ls>) {
4059 chomp;
4060 s/\A100644 blob $empty_blob\t//o or next;
4061 unless ($printed_warning) {
4062 print STDERR "Scanning for empty symlinks, ",
4063 "this may take a while if you have ",
4064 "many empty files\n",
4065 "You may disable this with `",
4066 "git config svn.brokenSymlinkWorkaround ",
4067 "false'.\n",
4068 "This may be done in a different ",
4069 "terminal without restarting ",
4070 "git svn\n";
4071 $printed_warning = 1;
4072 }
4073 my $path = $_;
4074 my (undef, $props) =
4075 $git_svn->ra->get_file($pfx.$path, $rev, undef);
4076 if ($props->{'svn:special'}) {
4077 $ret{$path} = 1;
4078 }
4079 }
4080 command_close_pipe($ls, $ctx);
4081 \%ret;
4082}
4083# returns true if a given path is inside a ".git" directory
4085sub in_dot_git {
4086 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4087}
4088# return value: 0 -- don't ignore, 1 -- ignore
4090sub is_path_ignored {
4091 my ($self, $path) = @_;
4092 return 1 if in_dot_git($path);
4093 return 1 if defined($self->{ignore_regex}) &&
4094 $path =~ m!$self->{ignore_regex}!;
4095 return 0 unless defined($_ignore_regex);
4096 return 1 if $path =~ m!$_ignore_regex!o;
4097 return 0;
4098}
4099sub set_path_strip {
4101 my ($self, $path) = @_;
4102 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4103}
4104sub open_root {
4106 { path => '' };
4107}
4108sub open_directory {
4110 my ($self, $path, $pb, $rev) = @_;
4111 { path => $path };
4112}
4113sub git_path {
4115 my ($self, $path) = @_;
4116 if ($self->{path_strip}) {
4117 $path =~ s!$self->{path_strip}!! or
4118 die "Failed to strip path '$path' ($self->{path_strip})\n";
4119 }
4120 $path;
4121}
4122sub delete_entry {
4124 my ($self, $path, $rev, $pb) = @_;
4125 return undef if $self->is_path_ignored($path);
4126 my $gpath = $self->git_path($path);
4128 return undef if ($gpath eq '');
4129 # remove entire directories.
4131 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4132 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4133 if ($tree) {
4134 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4135 -r --name-only -z/,
4136 $tree);
4137 local $/ = "\0";
4138 while (<$ls>) {
4139 chomp;
4140 my $rmpath = "$gpath/$_";
4141 $self->{gii}->remove($rmpath);
4142 print "\tD\t$rmpath\n" unless $::_q;
4143 }
4144 print "\tD\t$gpath/\n" unless $::_q;
4145 command_close_pipe($ls, $ctx);
4146 } else {
4147 $self->{gii}->remove($gpath);
4148 print "\tD\t$gpath\n" unless $::_q;
4149 }
4150 $self->{empty}->{$path} = 0;
4151 undef;
4152}
4153sub open_file {
4155 my ($self, $path, $pb, $rev) = @_;
4156 my ($mode, $blob);
4157 goto out if $self->is_path_ignored($path);
4159 my $gpath = $self->git_path($path);
4161 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4162 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4163 unless (defined $mode && defined $blob) {
4164 die "$path was not found in commit $self->{c} (r$rev)\n";
4165 }
4166 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4167 $mode = '120000';
4168 }
4169out:
4170 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4171 pool => SVN::Pool->new, action => 'M' };
4172}
4173sub add_file {
4175 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4176 my $mode;
4177 if (!$self->is_path_ignored($path)) {
4179 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4180 delete $self->{empty}->{$dir};
4181 $mode = '100644';
4182 }
4183 { path => $path, mode_a => $mode, mode_b => $mode,
4184 pool => SVN::Pool->new, action => 'A' };
4185}
4186sub add_directory {
4188 my ($self, $path, $cp_path, $cp_rev) = @_;
4189 goto out if $self->is_path_ignored($path);
4190 my $gpath = $self->git_path($path);
4191 if ($gpath eq '') {
4192 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4193 -r --name-only -z/,
4194 $self->{c});
4195 local $/ = "\0";
4196 while (<$ls>) {
4197 chomp;
4198 $self->{gii}->remove($_);
4199 print "\tD\t$_\n" unless $::_q;
4200 }
4201 command_close_pipe($ls, $ctx);
4202 $self->{empty}->{$path} = 0;
4203 }
4204 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4205 delete $self->{empty}->{$dir};
4206 $self->{empty}->{$path} = 1;
4207out:
4208 { path => $path };
4209}
4210sub change_dir_prop {
4212 my ($self, $db, $prop, $value) = @_;
4213 return undef if $self->is_path_ignored($db->{path});
4214 $self->{dir_prop}->{$db->{path}} ||= {};
4215 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4216 undef;
4217}
4218sub absent_directory {
4220 my ($self, $path, $pb) = @_;
4221 return undef if $self->is_path_ignored($path);
4222 $self->{absent_dir}->{$pb->{path}} ||= [];
4223 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4224 undef;
4225}
4226sub absent_file {
4228 my ($self, $path, $pb) = @_;
4229 return undef if $self->is_path_ignored($path);
4230 $self->{absent_file}->{$pb->{path}} ||= [];
4231 push @{$self->{absent_file}->{$pb->{path}}}, $path;
4232 undef;
4233}
4234sub change_file_prop {
4236 my ($self, $fb, $prop, $value) = @_;
4237 return undef if $self->is_path_ignored($fb->{path});
4238 if ($prop eq 'svn:executable') {
4239 if ($fb->{mode_b} != 120000) {
4240 $fb->{mode_b} = defined $value ? 100755 : 100644;
4241 }
4242 } elsif ($prop eq 'svn:special') {
4243 $fb->{mode_b} = defined $value ? 120000 : 100644;
4244 } else {
4245 $self->{file_prop}->{$fb->{path}} ||= {};
4246 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4247 }
4248 undef;
4249}
4250sub apply_textdelta {
4252 my ($self, $fb, $exp) = @_;
4253 return undef if $self->is_path_ignored($fb->{path});
4254 my $fh = $::_repository->temp_acquire('svn_delta');
4255 # $fh gets auto-closed() by SVN::TxDelta::apply(),
4256 # (but $base does not,) so dup() it for reading in close_file
4257 open my $dup, '<&', $fh or croak $!;
4258 my $base = $::_repository->temp_acquire('git_blob');
4259 if ($fb->{blob}) {
4261 my ($base_is_link, $size);
4262 if ($fb->{mode_a} eq '120000' &&
4264 ! $self->{empty_symlinks}->{$fb->{path}}) {
4265 print $base 'link ' or die "print $!\n";
4266 $base_is_link = 1;
4267 }
4268 retry:
4269 $size = $::_repository->cat_blob($fb->{blob}, $base);
4270 die "Failed to read object $fb->{blob}" if ($size < 0);
4271 if (defined $exp) {
4273 seek $base, 0, 0 or croak $!;
4274 my $got = ::md5sum($base);
4275 if ($got ne $exp) {
4276 my $err = "Checksum mismatch: ".
4277 "$fb->{path} $fb->{blob}\n" .
4278 "expected: $exp\n" .
4279 " got: $got\n";
4280 if ($base_is_link) {
4281 warn $err,
4282 "Retrying... (possibly ",
4283 "a bad symlink from SVN)\n";
4284 $::_repository->temp_reset($base);
4285 $base_is_link = 0;
4286 goto retry;
4287 }
4288 die $err;
4289 }
4290 }
4291 }
4292 seek $base, 0, 0 or croak $!;
4293 $fb->{fh} = $fh;
4294 $fb->{base} = $base;
4295 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4296}
4297sub close_file {
4299 my ($self, $fb, $exp) = @_;
4300 return undef if $self->is_path_ignored($fb->{path});
4301 my $hash;
4303 my $path = $self->git_path($fb->{path});
4304 if (my $fh = $fb->{fh}) {
4305 if (defined $exp) {
4306 seek($fh, 0, 0) or croak $!;
4307 my $got = ::md5sum($fh);
4308 if ($got ne $exp) {
4309 die "Checksum mismatch: $path\n",
4310 "expected: $exp\n got: $got\n";
4311 }
4312 }
4313 if ($fb->{mode_b} == 120000) {
4314 sysseek($fh, 0, 0) or croak $!;
4315 my $rd = sysread($fh, my $buf, 5);
4316 if (!defined $rd) {
4318 croak "sysread: $!\n";
4319 } elsif ($rd == 0) {
4320 warn "$path has mode 120000",
4321 " but it points to nothing\n",
4322 "converting to an empty file with mode",
4323 " 100644\n";
4324 $fb->{mode_b} = '100644';
4325 } elsif ($buf ne 'link ') {
4326 warn "$path has mode 120000",
4327 " but is not a link\n";
4328 } else {
4329 my $tmp_fh = $::_repository->temp_acquire(
4330 'svn_hash');
4331 my $res;
4332 while ($res = sysread($fh, my $str, 1024)) {
4333 my $out = syswrite($tmp_fh, $str, $res);
4334 defined($out) && $out == $res
4335 or croak("write ",
4336 Git::temp_path($tmp_fh),
4337 ": $!\n");
4338 }
4339 defined $res or croak $!;
4340 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4342 Git::temp_release($tmp_fh, 1);
4343 }
4344 }
4345 $hash = $::_repository->hash_and_insert_object(
4347 Git::temp_path($fh));
4348 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4349 Git::temp_release($fb->{base}, 1);
4351 Git::temp_release($fh, 1);
4352 } else {
4353 $hash = $fb->{blob} or die "no blob information\n";
4354 }
4355 $fb->{pool}->clear;
4356 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4357 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4358 undef;
4359}
4360sub abort_edit {
4362 my $self = shift;
4363 $self->{nr} = $self->{gii}->{nr};
4364 delete $self->{gii};
4365 $self->SUPER::abort_edit(@_);
4366}
4367sub close_edit {
4369 my $self = shift;
4370 $self->{git_commit_ok} = 1;
4371 $self->{nr} = $self->{gii}->{nr};
4372 delete $self->{gii};
4373 $self->SUPER::close_edit(@_);
4374}
4375package SVN::Git::Editor;
4377use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4378use strict;
4379use warnings;
4380use Carp qw/croak/;
4381use IO::File;
4382sub new {
4384 my ($class, $opts) = @_;
4385 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4386 die "$_ required!\n" unless (defined $opts->{$_});
4387 }
4388 my $pool = SVN::Pool->new;
4390 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4391 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4392 $opts->{r}, $mods);
4393 # $opts->{ra} functions should not be used after this:
4395 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4396 $opts->{editor_cb}, $pool);
4397 my $self = SVN::Delta::Editor->new(@ce, $pool);
4398 bless $self, $class;
4399 foreach (qw/svn_path r tree_a tree_b/) {
4400 $self->{$_} = $opts->{$_};
4401 }
4402 $self->{url} = $opts->{ra}->{url};
4403 $self->{mods} = $mods;
4404 $self->{types} = $types;
4405 $self->{pool} = $pool;
4406 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4407 $self->{rm} = { };
4408 $self->{path_prefix} = length $self->{svn_path} ?
4409 "$self->{svn_path}/" : '';
4410 $self->{config} = $opts->{config};
4411 return $self;
4412}
4413sub generate_diff {
4415 my ($tree_a, $tree_b) = @_;
4416 my @diff_tree = qw(diff-tree -z -r);
4417 if ($_cp_similarity) {
4418 push @diff_tree, "-C$_cp_similarity";
4419 } else {
4420 push @diff_tree, '-C';
4421 }
4422 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4423 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4424 push @diff_tree, $tree_a, $tree_b;
4425 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4426 local $/ = "\0";
4427 my $state = 'meta';
4428 my @mods;
4429 while (<$diff_fh>) {
4430 chomp $_; # this gets rid of the trailing "\0"
4431 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4432 ($::sha1)\s($::sha1)\s
4433 ([MTCRAD])\d*$/xo) {
4434 push @mods, { mode_a => $1, mode_b => $2,
4435 sha1_a => $3, sha1_b => $4,
4436 chg => $5 };
4437 if ($5 =~ /^(?:C|R)$/) {
4438 $state = 'file_a';
4439 } else {
4440 $state = 'file_b';
4441 }
4442 } elsif ($state eq 'file_a') {
4443 my $x = $mods[$#mods] or croak "Empty array\n";
4444 if ($x->{chg} !~ /^(?:C|R)$/) {
4445 croak "Error parsing $_, $x->{chg}\n";
4446 }
4447 $x->{file_a} = $_;
4448 $state = 'file_b';
4449 } elsif ($state eq 'file_b') {
4450 my $x = $mods[$#mods] or croak "Empty array\n";
4451 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4452 croak "Error parsing $_, $x->{chg}\n";
4453 }
4454 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4455 croak "Error parsing $_, $x->{chg}\n";
4456 }
4457 $x->{file_b} = $_;
4458 $state = 'meta';
4459 } else {
4460 croak "Error parsing $_\n";
4461 }
4462 }
4463 command_close_pipe($diff_fh, $ctx);
4464 \@mods;
4465}
4466sub check_diff_paths {
4468 my ($ra, $pfx, $rev, $mods) = @_;
4469 my %types;
4470 $pfx .= '/' if length $pfx;
4471 sub type_diff_paths {
4473 my ($ra, $types, $path, $rev) = @_;
4474 my @p = split m#/+#, $path;
4475 my $c = shift @p;
4476 unless (defined $types->{$c}) {
4477 $types->{$c} = $ra->check_path($c, $rev);
4478 }
4479 while (@p) {
4480 $c .= '/' . shift @p;
4481 next if defined $types->{$c};
4482 $types->{$c} = $ra->check_path($c, $rev);
4483 }
4484 }
4485 foreach my $m (@$mods) {
4487 foreach my $f (qw/file_a file_b/) {
4488 next unless defined $m->{$f};
4489 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4490 if (length $pfx.$dir && ! defined $types{$dir}) {
4491 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4492 }
4493 }
4494 }
4495 \%types;
4496}
4497sub split_path {
4499 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4500}
4501sub repo_path {
4503 my ($self, $path) = @_;
4504 $self->{path_prefix}.(defined $path ? $path : '');
4505}
4506sub url_path {
4508 my ($self, $path) = @_;
4509 if ($self->{url} =~ m#^https?://#) {
4510 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4511 }
4512 $self->{url} . '/' . $self->repo_path($path);
4513}
4514sub rmdirs {
4516 my ($self) = @_;
4517 my $rm = $self->{rm};
4518 delete $rm->{''}; # we never delete the url we're tracking
4519 return unless %$rm;
4520 foreach (keys %$rm) {
4522 my @d = split m#/#, $_;
4523 my $c = shift @d;
4524 $rm->{$c} = 1;
4525 while (@d) {
4526 $c .= '/' . shift @d;
4527 $rm->{$c} = 1;
4528 }
4529 }
4530 delete $rm->{$self->{svn_path}};
4531 delete $rm->{''}; # we never delete the url we're tracking
4532 return unless %$rm;
4533 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4535 $self->{tree_b});
4536 local $/ = "\0";
4537 while (<$fh>) {
4538 chomp;
4539 my @dn = split m#/#, $_;
4540 while (pop @dn) {
4541 delete $rm->{join '/', @dn};
4542 }
4543 unless (%$rm) {
4544 close $fh;
4545 return;
4546 }
4547 }
4548 command_close_pipe($fh, $ctx);
4549 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4551 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4552 $self->close_directory($bat->{$d}, $p);
4553 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4554 print "\tD+\t$d/\n" unless $::_q;
4555 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4556 delete $bat->{$d};
4557 }
4558}
4559sub open_or_add_dir {
4561 my ($self, $full_path, $baton) = @_;
4562 my $t = $self->{types}->{$full_path};
4563 if (!defined $t) {
4564 die "$full_path not known in r$self->{r} or we have a bug!\n";
4565 }
4566 {
4567 no warnings 'once';
4568 # SVN::Node::none and SVN::Node::file are used only once,
4569 # so we're shutting up Perl's warnings about them.
4570 if ($t == $SVN::Node::none) {
4571 return $self->add_directory($full_path, $baton,
4572 undef, -1, $self->{pool});
4573 } elsif ($t == $SVN::Node::dir) {
4574 return $self->open_directory($full_path, $baton,
4575 $self->{r}, $self->{pool});
4576 } # no warnings 'once'
4577 print STDERR "$full_path already exists in repository at ",
4578 "r$self->{r} and it is not a directory (",
4579 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4580 } # no warnings 'once'
4581 exit 1;
4582}
4583sub ensure_path {
4585 my ($self, $path) = @_;
4586 my $bat = $self->{bat};
4587 my $repo_path = $self->repo_path($path);
4588 return $bat->{''} unless (length $repo_path);
4589 my @p = split m#/+#, $repo_path;
4590 my $c = shift @p;
4591 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4592 while (@p) {
4593 my $c0 = $c;
4594 $c .= '/' . shift @p;
4595 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4596 }
4597 return $bat->{$c};
4598}
4599# Subroutine to convert a globbing pattern to a regular expression.
4601# From perl cookbook.
4602sub glob2pat {
4603 my $globstr = shift;
4604 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4605 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4606 return '^' . $globstr . '$';
4607}
4608sub check_autoprop {
4610 my ($self, $pattern, $properties, $file, $fbat) = @_;
4611 # Convert the globbing pattern to a regular expression.
4612 my $regex = glob2pat($pattern);
4613 # Check if the pattern matches the file name.
4614 if($file =~ m/($regex)/) {
4615 # Parse the list of properties to set.
4616 my @props = split(/;/, $properties);
4617 foreach my $prop (@props) {
4618 # Parse 'name=value' syntax and set the property.
4619 if ($prop =~ /([^=]+)=(.*)/) {
4620 my ($n,$v) = ($1,$2);
4621 for ($n, $v) {
4622 s/^\s+//; s/\s+$//;
4623 }
4624 $self->change_file_prop($fbat, $n, $v);
4625 }
4626 }
4627 }
4628}
4629sub apply_autoprops {
4631 my ($self, $file, $fbat) = @_;
4632 my $conf_t = ${$self->{config}}{'config'};
4633 no warnings 'once';
4634 # Check [miscellany]/enable-auto-props in svn configuration.
4635 if (SVN::_Core::svn_config_get_bool(
4636 $conf_t,
4637 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4638 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4639 0)) {
4640 # Auto-props are enabled. Enumerate them to look for matches.
4641 my $callback = sub {
4642 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4643 };
4644 SVN::_Core::svn_config_enumerate(
4645 $conf_t,
4646 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4647 $callback);
4648 }
4649}
4650sub A {
4652 my ($self, $m) = @_;
4653 my ($dir, $file) = split_path($m->{file_b});
4654 my $pbat = $self->ensure_path($dir);
4655 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4656 undef, -1);
4657 print "\tA\t$m->{file_b}\n" unless $::_q;
4658 $self->apply_autoprops($file, $fbat);
4659 $self->chg_file($fbat, $m);
4660 $self->close_file($fbat,undef,$self->{pool});
4661}
4662sub C {
4664 my ($self, $m) = @_;
4665 my ($dir, $file) = split_path($m->{file_b});
4666 my $pbat = $self->ensure_path($dir);
4667 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4668 $self->url_path($m->{file_a}), $self->{r});
4669 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4670 $self->chg_file($fbat, $m);
4671 $self->close_file($fbat,undef,$self->{pool});
4672}
4673sub delete_entry {
4675 my ($self, $path, $pbat) = @_;
4676 my $rpath = $self->repo_path($path);
4677 my ($dir, $file) = split_path($rpath);
4678 $self->{rm}->{$dir} = 1;
4679 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4680}
4681sub R {
4683 my ($self, $m) = @_;
4684 my ($dir, $file) = split_path($m->{file_b});
4685 my $pbat = $self->ensure_path($dir);
4686 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4687 $self->url_path($m->{file_a}), $self->{r});
4688 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4689 $self->apply_autoprops($file, $fbat);
4690 $self->chg_file($fbat, $m);
4691 $self->close_file($fbat,undef,$self->{pool});
4692 ($dir, $file) = split_path($m->{file_a});
4694 $pbat = $self->ensure_path($dir);
4695 $self->delete_entry($m->{file_a}, $pbat);
4696}
4697sub M {
4699 my ($self, $m) = @_;
4700 my ($dir, $file) = split_path($m->{file_b});
4701 my $pbat = $self->ensure_path($dir);
4702 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4703 $pbat,$self->{r},$self->{pool});
4704 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4705 $self->chg_file($fbat, $m);
4706 $self->close_file($fbat,undef,$self->{pool});
4707}
4708sub T { shift->M(@_) }
4710sub change_file_prop {
4712 my ($self, $fbat, $pname, $pval) = @_;
4713 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4714}
4715sub _chg_file_get_blob ($$$$) {
4717 my ($self, $fbat, $m, $which) = @_;
4718 my $fh = $::_repository->temp_acquire("git_blob_$which");
4719 if ($m->{"mode_$which"} =~ /^120/) {
4720 print $fh 'link ' or croak $!;
4721 $self->change_file_prop($fbat,'svn:special','*');
4722 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4723 $self->change_file_prop($fbat,'svn:special',undef);
4724 }
4725 my $blob = $m->{"sha1_$which"};
4726 return ($fh,) if ($blob =~ /^0{40}$/);
4727 my $size = $::_repository->cat_blob($blob, $fh);
4728 croak "Failed to read object $blob" if ($size < 0);
4729 $fh->flush == 0 or croak $!;
4730 seek $fh, 0, 0 or croak $!;
4731 my $exp = ::md5sum($fh);
4733 seek $fh, 0, 0 or croak $!;
4734 return ($fh, $exp);
4735}
4736sub chg_file {
4738 my ($self, $fbat, $m) = @_;
4739 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4740 $self->change_file_prop($fbat,'svn:executable','*');
4741 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4742 $self->change_file_prop($fbat,'svn:executable',undef);
4743 }
4744 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4745 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4746 my $pool = SVN::Pool->new;
4747 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4748 if (-s $fh_a) {
4749 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4750 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4751 if (defined $res) {
4752 die "Unexpected result from send_txstream: $res\n",
4753 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4754 }
4755 } else {
4756 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4757 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4758 if ($got ne $exp_b);
4759 }
4760 Git::temp_release($fh_b, 1);
4761 Git::temp_release($fh_a, 1);
4762 $pool->clear;
4763}
4764sub D {
4766 my ($self, $m) = @_;
4767 my ($dir, $file) = split_path($m->{file_b});
4768 my $pbat = $self->ensure_path($dir);
4769 print "\tD\t$m->{file_b}\n" unless $::_q;
4770 $self->delete_entry($m->{file_b}, $pbat);
4771}
4772sub close_edit {
4774 my ($self) = @_;
4775 my ($p,$bat) = ($self->{pool}, $self->{bat});
4776 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4777 next if $_ eq '';
4778 $self->close_directory($bat->{$_}, $p);
4779 }
4780 $self->close_directory($bat->{''}, $p);
4781 $self->SUPER::close_edit($p);
4782 $p->clear;
4783}
4784sub abort_edit {
4786 my ($self) = @_;
4787 $self->SUPER::abort_edit($self->{pool});
4788}
4789sub DESTROY {
4791 my $self = shift;
4792 $self->SUPER::DESTROY(@_);
4793 $self->{pool}->clear;
4794}
4795# this drives the editor
4797sub apply_diff {
4798 my ($self) = @_;
4799 my $mods = $self->{mods};
4800 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4801 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4802 my $f = $m->{chg};
4803 if (defined $o{$f}) {
4804 $self->$f($m);
4805 } else {
4806 fatal("Invalid change type: $f");
4807 }
4808 }
4809 $self->rmdirs if $_rmdir;
4810 if (@$mods == 0) {
4811 $self->abort_edit;
4812 } else {
4813 $self->close_edit;
4814 }
4815 return scalar @$mods;
4816}
4817package Git::SVN::Ra;
4819use vars qw/@ISA $config_dir $_log_window_size/;
4820use strict;
4821use warnings;
4822my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4823BEGIN {
4825 # enforce temporary pool usage for some simple functions
4826 no strict 'refs';
4827 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4828 get_file/) {
4829 my $SUPER = "SUPER::$f";
4830 *$f = sub {
4831 my $self = shift;
4832 my $pool = SVN::Pool->new;
4833 my @ret = $self->$SUPER(@_,$pool);
4834 $pool->clear;
4835 wantarray ? @ret : $ret[0];
4836 };
4837 }
4838}
4839sub _auth_providers () {
4841 [
4842 SVN::Client::get_simple_provider(),
4843 SVN::Client::get_ssl_server_trust_file_provider(),
4844 SVN::Client::get_simple_prompt_provider(
4845 \&Git::SVN::Prompt::simple, 2),
4846 SVN::Client::get_ssl_client_cert_file_provider(),
4847 SVN::Client::get_ssl_client_cert_prompt_provider(
4848 \&Git::SVN::Prompt::ssl_client_cert, 2),
4849 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4850 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4851 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4852 SVN::Client::get_username_provider(),
4853 SVN::Client::get_ssl_server_trust_prompt_provider(
4854 \&Git::SVN::Prompt::ssl_server_trust),
4855 SVN::Client::get_username_prompt_provider(
4856 \&Git::SVN::Prompt::username, 2)
4857 ]
4858}
4859sub escape_uri_only {
4861 my ($uri) = @_;
4862 my @tmp;
4863 foreach (split m{/}, $uri) {
4864 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4865 push @tmp, $_;
4866 }
4867 join('/', @tmp);
4868}
4869sub escape_url {
4871 my ($url) = @_;
4872 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4873 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4874 $url = "$scheme://$domain$uri";
4875 }
4876 $url;
4877}
4878sub new {
4880 my ($class, $url) = @_;
4881 $url =~ s!/+$!!;
4882 return $RA if ($RA && $RA->{url} eq $url);
4883 ::_req_svn();
4885 SVN::_Core::svn_config_ensure($config_dir, undef);
4887 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4888 my $config = SVN::Core::config_get_config($config_dir);
4889 $RA = undef;
4890 my $dont_store_passwords = 1;
4891 my $conf_t = ${$config}{'config'};
4892 {
4893 no warnings 'once';
4894 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4895 # produces warnings that variables are used only once.
4896 # I had not found the better way to shut them up, so
4897 # the warnings of type 'once' are disabled in this block.
4898 if (SVN::_Core::svn_config_get_bool($conf_t,
4899 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4900 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4901 1) == 0) {
4902 SVN::_Core::svn_auth_set_parameter($baton,
4903 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4904 bless (\$dont_store_passwords, "_p_void"));
4905 }
4906 if (SVN::_Core::svn_config_get_bool($conf_t,
4907 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4908 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4909 1) == 0) {
4910 $Git::SVN::Prompt::_no_auth_cache = 1;
4911 }
4912 } # no warnings 'once'
4913 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4914 config => $config,
4915 pool => SVN::Pool->new,
4916 auth_provider_callbacks => $callbacks);
4917 $self->{url} = $url;
4918 $self->{svn_path} = $url;
4919 $self->{repos_root} = $self->get_repos_root;
4920 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4921 $self->{cache} = { check_path => { r => 0, data => {} },
4922 get_dir => { r => 0, data => {} } };
4923 $RA = bless $self, $class;
4924}
4925sub check_path {
4927 my ($self, $path, $r) = @_;
4928 my $cache = $self->{cache}->{check_path};
4929 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4930 return $cache->{data}->{$path};
4931 }
4932 my $pool = SVN::Pool->new;
4933 my $t = $self->SUPER::check_path($path, $r, $pool);
4934 $pool->clear;
4935 if ($r != $cache->{r}) {
4936 %{$cache->{data}} = ();
4937 $cache->{r} = $r;
4938 }
4939 $cache->{data}->{$path} = $t;
4940}
4941sub get_dir {
4943 my ($self, $dir, $r) = @_;
4944 my $cache = $self->{cache}->{get_dir};
4945 if ($r == $cache->{r}) {
4946 if (my $x = $cache->{data}->{$dir}) {
4947 return wantarray ? @$x : $x->[0];
4948 }
4949 }
4950 my $pool = SVN::Pool->new;
4951 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4952 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4953 $pool->clear;
4954 if ($r != $cache->{r}) {
4955 %{$cache->{data}} = ();
4956 $cache->{r} = $r;
4957 }
4958 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4959 wantarray ? (\%dirents, $r, $props) : \%dirents;
4960}
4961sub DESTROY {
4963 # do not call the real DESTROY since we store ourselves in $RA
4964}
4965# get_log(paths, start, end, limit,
4967# discover_changed_paths, strict_node_history, receiver)
4968sub get_log {
4969 my ($self, @args) = @_;
4970 my $pool = SVN::Pool->new;
4971 # svn_log_changed_path_t objects passed to get_log are likely to be
4973 # overwritten even if only the refs are copied to an external variable,
4974 # so we should dup the structures in their entirety. Using an
4975 # externally passed pool (instead of our temporary and quickly cleared
4976 # pool in Git::SVN::Ra) does not help matters at all...
4977 my $receiver = pop @args;
4978 my $prefix = "/".$self->{svn_path};
4979 $prefix =~ s#/+($)##;
4980 my $prefix_regex = qr#^\Q$prefix\E#;
4981 push(@args, sub {
4982 my ($paths) = $_[0];
4983 return &$receiver(@_) unless $paths;
4984 $_[0] = ();
4985 foreach my $p (keys %$paths) {
4986 my $i = $paths->{$p};
4987 # Make path relative to our url, not repos_root
4988 $p =~ s/$prefix_regex//;
4989 my %s = map { $_ => $i->$_; }
4990 qw/copyfrom_path copyfrom_rev action/;
4991 if ($s{'copyfrom_path'}) {
4992 $s{'copyfrom_path'} =~ s/$prefix_regex//;
4993 }
4994 $_[0]{$p} = \%s;
4995 }
4996 &$receiver(@_);
4997 });
4998 # the limit parameter was not supported in SVN 1.1.x, so we
5001 # drop it. Therefore, the receiver callback passed to it
5002 # is made aware of this limitation by being wrapped if
5003 # the limit passed to is being wrapped.
5004 if ($SVN::Core::VERSION le '1.2.0') {
5005 my $limit = splice(@args, 3, 1);
5006 if ($limit > 0) {
5007 my $receiver = pop @args;
5008 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
5009 }
5010 }
5011 my $ret = $self->SUPER::get_log(@args, $pool);
5012 $pool->clear;
5013 $ret;
5014}
5015sub trees_match {
5017 my ($self, $url1, $rev1, $url2, $rev2) = @_;
5018 my $ctx = SVN::Client->new(auth => _auth_providers);
5019 my $out = IO::File->new_tmpfile;
5020 # older SVN (1.1.x) doesn't take $pool as the last parameter for
5022 # $ctx->diff(), so we'll create a default one
5023 my $pool = SVN::Pool->new_default_sub;
5024 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5026 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5027 $out->flush;
5028 my $ret = (($out->stat)[7] == 0);
5029 close $out or croak $!;
5030 $ret;
5032}
5033sub get_commit_editor {
5035 my ($self, $log, $cb, $pool) = @_;
5036 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
5037 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5038}
5039sub gs_do_update {
5041 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5042 my $new = ($rev_a == $rev_b);
5043 my $path = $gs->{path};
5044 if ($new && -e $gs->{index}) {
5046 unlink $gs->{index} or die
5047 "Couldn't unlink index: $gs->{index}: $!\n";
5048 }
5049 my $pool = SVN::Pool->new;
5050 $editor->set_path_strip($path);
5051 my (@pc) = split m#/#, $path;
5052 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5053 1, $editor, $pool);
5054 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5055 # Since we can't rely on svn_ra_reparent being available, we'll
5057 # just have to do some magic with set_path to make it so
5058 # we only want a partial path.
5059 my $sp = '';
5060 my $final = join('/', @pc);
5061 while (@pc) {
5062 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5063 $sp .= '/' if length $sp;
5064 $sp .= shift @pc;
5065 }
5066 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5067 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5069 $reporter->finish_report($pool);
5071 $pool->clear;
5072 $editor->{git_commit_ok};
5073}
5074# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5076# svn_ra_reparent didn't work before 1.4)
5077sub gs_do_switch {
5078 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5079 my $path = $gs->{path};
5080 my $pool = SVN::Pool->new;
5081 my $full_url = $self->{url};
5083 my $old_url = $full_url;
5084 $full_url .= '/' . $path if length $path;
5085 my ($ra, $reparented);
5086 if ($old_url =~ m#^svn(\+ssh)?://# ||
5088 ($full_url =~ m#^https?://# &&
5089 escape_url($full_url) ne $full_url)) {
5090 $_[0] = undef;
5091 $self = undef;
5092 $RA = undef;
5093 $ra = Git::SVN::Ra->new($full_url);
5094 $ra_invalid = 1;
5095 } elsif ($old_url ne $full_url) {
5096 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5097 $self->{url} = $full_url;
5098 $reparented = 1;
5099 }
5100 $ra ||= $self;
5102 $url_b = escape_url($url_b);
5103 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5104 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5105 $reporter->set_path('', $rev_a, 0, @lock, $pool);
5106 $reporter->finish_report($pool);
5107 if ($reparented) {
5109 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5110 $self->{url} = $old_url;
5111 }
5112 $pool->clear;
5114 $editor->{git_commit_ok};
5115}
5116sub longest_common_path {
5118 my ($gsv, $globs) = @_;
5119 my %common;
5120 my $common_max = scalar @$gsv;
5121 foreach my $gs (@$gsv) {
5123 my @tmp = split m#/#, $gs->{path};
5124 my $p = '';
5125 foreach (@tmp) {
5126 $p .= length($p) ? "/$_" : $_;
5127 $common{$p} ||= 0;
5128 $common{$p}++;
5129 }
5130 }
5131 $globs ||= [];
5132 $common_max += scalar @$globs;
5133 foreach my $glob (@$globs) {
5134 my @tmp = split m#/#, $glob->{path}->{left};
5135 my $p = '';
5136 foreach (@tmp) {
5137 $p .= length($p) ? "/$_" : $_;
5138 $common{$p} ||= 0;
5139 $common{$p}++;
5140 }
5141 }
5142 my $longest_path = '';
5144 foreach (sort {length $b <=> length $a} keys %common) {
5145 if ($common{$_} == $common_max) {
5146 $longest_path = $_;
5147 last;
5148 }
5149 }
5150 $longest_path;
5151}
5152sub gs_fetch_loop_common {
5154 my ($self, $base, $head, $gsv, $globs) = @_;
5155 return if ($base > $head);
5156 my $inc = $_log_window_size;
5157 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5158 my $longest_path = longest_common_path($gsv, $globs);
5159 my $ra_url = $self->{url};
5160 my $find_trailing_edge;
5161 while (1) {
5162 my %revs;
5163 my $err;
5164 my $err_handler = $SVN::Error::handler;
5165 $SVN::Error::handler = sub {
5166 ($err) = @_;
5167 skip_unknown_revs($err);
5168 };
5169 sub _cb {
5170 my ($paths, $r, $author, $date, $log) = @_;
5171 [ $paths,
5172 { author => $author, date => $date, log => $log } ];
5173 }
5174 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5175 sub { $revs{$_[1]} = _cb(@_) });
5176 if ($err) {
5177 print "Checked through r$max\r";
5178 } else {
5179 $find_trailing_edge = 1;
5180 }
5181 if ($err and $find_trailing_edge) {
5182 print STDERR "Path '$longest_path' ",
5183 "was probably deleted:\n",
5184 $err->expanded_message,
5185 "\nWill attempt to follow ",
5186 "revisions r$min .. r$max ",
5187 "committed before the deletion\n";
5188 my $hi = $max;
5189 while (--$hi >= $min) {
5190 my $ok;
5191 $self->get_log([$longest_path], $min, $hi,
5192 0, 1, 1, sub {
5193 $ok = $_[1];
5194 $revs{$_[1]} = _cb(@_) });
5195 if ($ok) {
5196 print STDERR "r$min .. r$ok OK\n";
5197 last;
5198 }
5199 }
5200 $find_trailing_edge = 0;
5201 }
5202 $SVN::Error::handler = $err_handler;
5203 my %exists = map { $_->{path} => $_ } @$gsv;
5205 foreach my $r (sort {$a <=> $b} keys %revs) {
5206 my ($paths, $logged) = @{$revs{$r}};
5207 foreach my $gs ($self->match_globs(\%exists, $paths,
5209 $globs, $r)) {
5210 if ($gs->rev_map_max >= $r) {
5211 next;
5212 }
5213 next unless $gs->match_paths($paths, $r);
5214 $gs->{logged_rev_props} = $logged;
5215 if (my $last_commit = $gs->last_commit) {
5216 $gs->assert_index_clean($last_commit);
5217 }
5218 my $log_entry = $gs->do_fetch($paths, $r);
5219 if ($log_entry) {
5220 $gs->do_git_commit($log_entry);
5221 }
5222 $INDEX_FILES{$gs->{index}} = 1;
5223 }
5224 foreach my $g (@$globs) {
5225 my $k = "svn-remote.$g->{remote}." .
5226 "$g->{t}-maxRev";
5227 Git::SVN::tmp_config($k, $r);
5228 }
5229 if ($ra_invalid) {
5230 $_[0] = undef;
5231 $self = undef;
5232 $RA = undef;
5233 $self = Git::SVN::Ra->new($ra_url);
5234 $ra_invalid = undef;
5235 }
5236 }
5237 # pre-fill the .rev_db since it'll eventually get filled in
5238 # with '0' x40 if something new gets committed
5239 foreach my $gs (@$gsv) {
5240 next if $gs->rev_map_max >= $max;
5241 next if defined $gs->rev_map_get($max);
5242 $gs->rev_map_set($max, 0 x40);
5243 }
5244 foreach my $g (@$globs) {
5245 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5246 Git::SVN::tmp_config($k, $max);
5247 }
5248 last if $max >= $head;
5249 $min = $max + 1;
5250 $max += $inc;
5251 $max = $head if ($max > $head);
5252 }
5253 Git::SVN::gc();
5254}
5255sub get_dir_globbed {
5257 my ($self, $left, $depth, $r) = @_;
5258 my @x = eval { $self->get_dir($left, $r) };
5260 return unless scalar @x == 3;
5261 my $dirents = $x[0];
5262 my @finalents;
5263 foreach my $de (keys %$dirents) {
5264 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5265 if ($depth > 1) {
5266 my @args = ("$left/$de", $depth - 1, $r);
5267 foreach my $dir ($self->get_dir_globbed(@args)) {
5268 push @finalents, "$de/$dir";
5269 }
5270 } else {
5271 push @finalents, $de;
5272 }
5273 }
5274 @finalents;
5275}
5276sub match_globs {
5278 my ($self, $exists, $paths, $globs, $r) = @_;
5279 sub get_dir_check {
5281 my ($self, $exists, $g, $r) = @_;
5282 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5284 $g->{path}->{depth},
5285 $r);
5286 foreach my $de (@dirs) {
5288 my $p = $g->{path}->full_path($de);
5289 next if $exists->{$p};
5290 next if (length $g->{path}->{right} &&
5291 ($self->check_path($p, $r) !=
5292 $SVN::Node::dir));
5293 next unless $p =~ /$g->{path}->{regex}/;
5294 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5295 $g->{ref}->full_path($de), 1);
5296 }
5297 }
5298 foreach my $g (@$globs) {
5299 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5300 if ($path->{action} =~ /^[AR]$/) {
5301 get_dir_check($self, $exists, $g, $r);
5302 }
5303 }
5304 foreach (keys %$paths) {
5305 if (/$g->{path}->{left_regex}/ &&
5306 !/$g->{path}->{regex}/) {
5307 next if $paths->{$_}->{action} !~ /^[AR]$/;
5308 get_dir_check($self, $exists, $g, $r);
5309 }
5310 next unless /$g->{path}->{regex}/;
5311 my $p = $1;
5312 my $pathname = $g->{path}->full_path($p);
5313 next if $exists->{$pathname};
5314 next if ($self->check_path($pathname, $r) !=
5315 $SVN::Node::dir);
5316 $exists->{$pathname} = Git::SVN->init(
5317 $self->{url}, $pathname, undef,
5318 $g->{ref}->full_path($p), 1);
5319 }
5320 my $c = '';
5321 foreach (split m#/#, $g->{path}->{left}) {
5322 $c .= "/$_";
5323 next unless ($paths->{$c} &&
5324 ($paths->{$c}->{action} =~ /^[AR]$/));
5325 get_dir_check($self, $exists, $g, $r);
5326 }
5327 }
5328 values %$exists;
5329}
5330sub minimize_url {
5332 my ($self) = @_;
5333 return $self->{url} if ($self->{url} eq $self->{repos_root});
5334 my $url = $self->{repos_root};
5335 my @components = split(m!/!, $self->{svn_path});
5336 my $c = '';
5337 do {
5338 $url .= "/$c" if length $c;
5339 eval {
5340 my $ra = (ref $self)->new($url);
5341 my $latest = $ra->get_latest_revnum;
5342 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5343 };
5344 } while ($@ && ($c = shift @components));
5345 $url;
5346}
5347sub can_do_switch {
5349 my $self = shift;
5350 unless (defined $can_do_switch) {
5351 my $pool = SVN::Pool->new;
5352 my $rep = eval {
5353 $self->do_switch(1, '', 0, $self->{url},
5354 SVN::Delta::Editor->new, $pool);
5355 };
5356 if ($@) {
5357 $can_do_switch = 0;
5358 } else {
5359 $rep->abort_report($pool);
5360 $can_do_switch = 1;
5361 }
5362 $pool->clear;
5363 }
5364 $can_do_switch;
5365}
5366sub skip_unknown_revs {
5368 my ($err) = @_;
5369 my $errno = $err->apr_err();
5370 # Maybe the branch we're tracking didn't
5371 # exist when the repo started, so it's
5372 # not an error if it doesn't, just continue
5373 #
5374 # Wonderfully consistent library, eh?
5375 # 160013 - svn:// and file://
5376 # 175002 - http(s)://
5377 # 175007 - http(s):// (this repo required authorization, too...)
5378 # More codes may be discovered later...
5379 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5380 my $err_key = $err->expanded_message;
5381 # revision numbers change every time, filter them out
5382 $err_key =~ s/\d+/\0/g;
5383 $err_key = "$errno\0$err_key";
5384 unless ($ignored_err{$err_key}) {
5385 warn "W: Ignoring error from SVN, path probably ",
5386 "does not exist: ($errno): ",
5387 $err->expanded_message,"\n";
5388 warn "W: Do not be alarmed at the above message ",
5389 "git-svn is just searching aggressively for ",
5390 "old history.\n",
5391 "This may take a while on large repositories\n";
5392 $ignored_err{$err_key} = 1;
5393 }
5394 return;
5395 }
5396 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5397}
5398package Git::SVN::Log;
5400use strict;
5401use warnings;
5402use POSIX qw/strftime/;
5403use Time::Local;
5404use constant commit_log_separator => ('-' x 72) . "\n";
5405use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5406 %rusers $show_commit $incremental/;
5407my $l_fmt;
5408sub cmt_showable {
5410 my ($c) = @_;
5411 return 1 if defined $c->{r};
5412 # big commit message got truncated by the 16k pretty buffer in rev-list
5414 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5415 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5416 @{$c->{l}} = ();
5417 my @log = command(qw/cat-file commit/, $c->{c});
5418 # shift off the headers
5420 shift @log while ($log[0] ne '');
5421 shift @log;
5422 # TODO: make $c->{l} not have a trailing newline in the future
5424 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5425 (undef, $c->{r}, undef) = ::extract_metadata(
5427 (grep(/^git-svn-id: /, @log))[-1]);
5428 }
5429 return defined $c->{r};
5430}
5431sub log_use_color {
5433 return $color || Git->repository->get_colorbool('color.diff');
5434}
5435sub git_svn_log_cmd {
5437 my ($r_min, $r_max, @args) = @_;
5438 my $head = 'HEAD';
5439 my (@files, @log_opts);
5440 foreach my $x (@args) {
5441 if ($x eq '--' || @files) {
5442 push @files, $x;
5443 } else {
5444 if (::verify_ref("$x^0")) {
5445 $head = $x;
5446 } else {
5447 push @log_opts, $x;
5448 }
5449 }
5450 }
5451 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5453 $gs ||= Git::SVN->_new;
5454 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5455 $gs->refname);
5456 push @cmd, '-r' unless $non_recursive;
5457 push @cmd, qw/--raw --name-status/ if $verbose;
5458 push @cmd, '--color' if log_use_color();
5459 push @cmd, @log_opts;
5460 if (defined $r_max && $r_max == $r_min) {
5461 push @cmd, '--max-count=1';
5462 if (my $c = $gs->rev_map_get($r_max)) {
5463 push @cmd, $c;
5464 }
5465 } elsif (defined $r_max) {
5466 if ($r_max < $r_min) {
5467 ($r_min, $r_max) = ($r_max, $r_min);
5468 }
5469 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5470 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5471 # If there are no commits in the range, both $c_max and $c_min
5472 # will be undefined. If there is at least 1 commit in the
5473 # range, both will be defined.
5474 return () if !defined $c_min || !defined $c_max;
5475 if ($c_min eq $c_max) {
5476 push @cmd, '--max-count=1', $c_min;
5477 } else {
5478 push @cmd, '--boundary', "$c_min..$c_max";
5479 }
5480 }
5481 return (@cmd, @files);
5482}
5483# adapted from pager.c
5485sub config_pager {
5486 if (! -t *STDOUT) {
5487 $ENV{GIT_PAGER_IN_USE} = 'false';
5488 $pager = undef;
5489 return;
5490 }
5491 chomp($pager = command_oneline(qw(var GIT_PAGER)));
5492 if ($pager eq 'cat') {
5493 $pager = undef;
5494 }
5495 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5496}
5497sub run_pager {
5499 return unless defined $pager;
5500 pipe my ($rfd, $wfd) or return;
5501 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5502 if (!$pid) {
5503 open STDOUT, '>&', $wfd or
5504 ::fatal "Can't redirect to stdout: $!";
5505 return;
5506 }
5507 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5508 $ENV{LESS} ||= 'FRSX';
5509 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5510}
5511sub format_svn_date {
5513 # some systmes don't handle or mishandle %z, so be creative.
5514 my $t = shift || time;
5515 my $gm = timelocal(gmtime($t));
5516 my $sign = qw( + + - )[ $t <=> $gm ];
5517 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5518 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5519}
5520sub parse_git_date {
5522 my ($t, $tz) = @_;
5523 # Date::Parse isn't in the standard Perl distro :(
5524 if ($tz =~ s/^\+//) {
5525 $t += tz_to_s_offset($tz);
5526 } elsif ($tz =~ s/^\-//) {
5527 $t -= tz_to_s_offset($tz);
5528 }
5529 return $t;
5530}
5531sub set_local_timezone {
5533 if (defined $TZ) {
5534 $ENV{TZ} = $TZ;
5535 } else {
5536 delete $ENV{TZ};
5537 }
5538}
5539sub tz_to_s_offset {
5541 my ($tz) = @_;
5542 $tz =~ s/(\d\d)$//;
5543 return ($1 * 60) + ($tz * 3600);
5544}
5545sub get_author_info {
5547 my ($dest, $author, $t, $tz) = @_;
5548 $author =~ s/(?:^\s*|\s*$)//g;
5549 $dest->{a_raw} = $author;
5550 my $au;
5551 if ($::_authors) {
5552 $au = $rusers{$author} || undef;
5553 }
5554 if (!$au) {
5555 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5556 }
5557 $dest->{t} = $t;
5558 $dest->{tz} = $tz;
5559 $dest->{a} = $au;
5560 $dest->{t_utc} = parse_git_date($t, $tz);
5561}
5562sub process_commit {
5564 my ($c, $r_min, $r_max, $defer) = @_;
5565 if (defined $r_min && defined $r_max) {
5566 if ($r_min == $c->{r} && $r_min == $r_max) {
5567 show_commit($c);
5568 return 0;
5569 }
5570 return 1 if $r_min == $r_max;
5571 if ($r_min < $r_max) {
5572 # we need to reverse the print order
5573 return 0 if (defined $limit && --$limit < 0);
5574 push @$defer, $c;
5575 return 1;
5576 }
5577 if ($r_min != $r_max) {
5578 return 1 if ($r_min < $c->{r});
5579 return 1 if ($r_max > $c->{r});
5580 }
5581 }
5582 return 0 if (defined $limit && --$limit < 0);
5583 show_commit($c);
5584 return 1;
5585}
5586sub show_commit {
5588 my $c = shift;
5589 if ($oneline) {
5590 my $x = "\n";
5591 if (my $l = $c->{l}) {
5592 while ($l->[0] =~ /^\s*$/) { shift @$l }
5593 $x = $l->[0];
5594 }
5595 $l_fmt ||= 'A' . length($c->{r});
5596 print 'r',pack($l_fmt, $c->{r}),' | ';
5597 print "$c->{c} | " if $show_commit;
5598 print $x;
5599 } else {
5600 show_commit_normal($c);
5601 }
5602}
5603sub show_commit_changed_paths {
5605 my ($c) = @_;
5606 return unless $c->{changed};
5607 print "Changed paths:\n", @{$c->{changed}};
5608}
5609sub show_commit_normal {
5611 my ($c) = @_;
5612 print commit_log_separator, "r$c->{r} | ";
5613 print "$c->{c} | " if $show_commit;
5614 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5615 my $nr_line = 0;
5616 if (my $l = $c->{l}) {
5618 while ($l->[$#$l] eq "\n" && $#$l > 0
5619 && $l->[($#$l - 1)] eq "\n") {
5620 pop @$l;
5621 }
5622 $nr_line = scalar @$l;
5623 if (!$nr_line) {
5624 print "1 line\n\n\n";
5625 } else {
5626 if ($nr_line == 1) {
5627 $nr_line = '1 line';
5628 } else {
5629 $nr_line .= ' lines';
5630 }
5631 print $nr_line, "\n";
5632 show_commit_changed_paths($c);
5633 print "\n";
5634 print $_ foreach @$l;
5635 }
5636 } else {
5637 print "1 line\n";
5638 show_commit_changed_paths($c);
5639 print "\n";
5640 }
5642 foreach my $x (qw/raw stat diff/) {
5643 if ($c->{$x}) {
5644 print "\n";
5645 print $_ foreach @{$c->{$x}}
5646 }
5647 }
5648}
5649sub cmd_show_log {
5651 my (@args) = @_;
5652 my ($r_min, $r_max);
5653 my $r_last = -1; # prevent dupes
5654 set_local_timezone();
5655 if (defined $::_revision) {
5656 if ($::_revision =~ /^(\d+):(\d+)$/) {
5657 ($r_min, $r_max) = ($1, $2);
5658 } elsif ($::_revision =~ /^\d+$/) {
5659 $r_min = $r_max = $::_revision;
5660 } else {
5661 ::fatal "-r$::_revision is not supported, use ",
5662 "standard 'git log' arguments instead";
5663 }
5664 }
5665 config_pager();
5667 @args = git_svn_log_cmd($r_min, $r_max, @args);
5668 if (!@args) {
5669 print commit_log_separator unless $incremental || $oneline;
5670 return;
5671 }
5672 my $log = command_output_pipe(@args);
5673 run_pager();
5674 my (@k, $c, $d, $stat);
5675 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5676 while (<$log>) {
5677 if (/^${esc_color}commit -?($::sha1_short)/o) {
5678 my $cmt = $1;
5679 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5680 $r_last = $c->{r};
5681 process_commit($c, $r_min, $r_max, \@k) or
5682 goto out;
5683 }
5684 $d = undef;
5685 $c = { c => $cmt };
5686 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5687 get_author_info($c, $1, $2, $3);
5688 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5689 # ignore
5690 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5691 push @{$c->{raw}}, $_;
5692 } elsif (/^${esc_color}[ACRMDT]\t/) {
5693 # we could add $SVN->{svn_path} here, but that requires
5694 # remote access at the moment (repo_path_split)...
5695 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5696 push @{$c->{changed}}, $_;
5697 } elsif (/^${esc_color}diff /o) {
5698 $d = 1;
5699 push @{$c->{diff}}, $_;
5700 } elsif ($d) {
5701 push @{$c->{diff}}, $_;
5702 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5703 $esc_color*[\+\-]*$esc_color$/x) {
5704 $stat = 1;
5705 push @{$c->{stat}}, $_;
5706 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5707 push @{$c->{stat}}, $_;
5708 $stat = undef;
5709 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5710 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5711 } elsif (s/^${esc_color} //o) {
5712 push @{$c->{l}}, $_;
5713 }
5714 }
5715 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5716 $r_last = $c->{r};
5717 process_commit($c, $r_min, $r_max, \@k);
5718 }
5719 if (@k) {
5720 ($r_min, $r_max) = ($r_max, $r_min);
5721 process_commit($_, $r_min, $r_max) foreach reverse @k;
5722 }
5723out:
5724 close $log;
5725 print commit_log_separator unless $incremental || $oneline;
5726}
5727sub cmd_blame {
5729 my $path = pop;
5730 config_pager();
5732 run_pager();
5733 my ($fh, $ctx, $rev);
5735 if ($_git_format) {
5737 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5738 while (my $line = <$fh>) {
5739 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5740 # Uncommitted edits show up as a rev ID of
5741 # all zeros, which we can't look up with
5742 # cmt_metadata
5743 if ($1 !~ /^0+$/) {
5744 (undef, $rev, undef) =
5745 ::cmt_metadata($1);
5746 $rev = '0' if (!$rev);
5747 } else {
5748 $rev = '0';
5749 }
5750 $rev = sprintf('%-10s', $rev);
5751 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5752 }
5753 print $line;
5754 }
5755 } else {
5756 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5757 '--', $path);
5758 my ($sha1);
5759 my %authors;
5760 my @buffer;
5761 my %dsha; #distinct sha keys
5762 while (my $line = <$fh>) {
5764 push @buffer, $line;
5765 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5766 $dsha{$1} = 1;
5767 }
5768 }
5769 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5771 foreach my $line (@buffer) {
5773 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5774 $rev = $s2r->{$1};
5775 $rev = '0' if (!$rev)
5776 }
5777 elsif ($line =~ /^author (.*)/) {
5778 $authors{$rev} = $1;
5779 $authors{$rev} =~ s/\s/_/g;
5780 }
5781 elsif ($line =~ /^\t(.*)$/) {
5782 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5783 }
5784 }
5785 }
5786 command_close_pipe($fh, $ctx);
5787}
5788package Git::SVN::Migration;
5790# these version numbers do NOT correspond to actual version numbers
5791# of git nor git-svn. They are just relative.
5792#
5793# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5794#
5795# v1 layout: .git/$id/info/url, refs/remotes/$id
5796#
5797# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5798#
5799# v3 layout: .git/svn/$id, refs/remotes/$id
5800# - info/url may remain for backwards compatibility
5801# - this is what we migrate up to this layout automatically,
5802# - this will be used by git svn init on single branches
5803# v3.1 layout (auto migrated):
5804# - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5805# for backwards compatibility
5806#
5807# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5808# - this is only created for newly multi-init-ed
5809# repositories. Similar in spirit to the
5810# --use-separate-remotes option in git-clone (now default)
5811# - we do not automatically migrate to this (following
5812# the example set by core git)
5813#
5814# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5815# - newer, more-efficient format that uses 24-bytes per record
5816# with no filler space.
5817# - use xxd -c24 < .rev_map.$UUID to view and debug
5818# - This is a one-way migration, repositories updated to the
5819# new format will not be able to use old git-svn without
5820# rebuilding the .rev_db. Rebuilding the rev_db is not
5821# possible if noMetadata or useSvmProps are set; but should
5822# be no problem for users that use the (sensible) defaults.
5823use strict;
5824use warnings;
5825use Carp qw/croak/;
5826use File::Path qw/mkpath/;
5827use File::Basename qw/dirname basename/;
5828use vars qw/$_minimize/;
5829sub migrate_from_v0 {
5831 my $git_dir = $ENV{GIT_DIR};
5832 return undef unless -d $git_dir;
5833 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5834 my $migrated = 0;
5835 while (<$fh>) {
5836 chomp;
5837 my ($id, $orig_ref) = ($_, $_);
5838 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5839 next unless -f "$git_dir/$id/info/url";
5840 my $new_ref = "refs/remotes/$id";
5841 if (::verify_ref("$new_ref^0")) {
5842 print STDERR "W: $orig_ref is probably an old ",
5843 "branch used by an ancient version of ",
5844 "git-svn.\n",
5845 "However, $new_ref also exists.\n",
5846 "We will not be able ",
5847 "to use this branch until this ",
5848 "ambiguity is resolved.\n";
5849 next;
5850 }
5851 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5852 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5853 command_noisy('update-ref', $new_ref, $orig_ref);
5854 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5855 $migrated++;
5856 }
5857 command_close_pipe($fh, $ctx);
5858 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5859 $migrated;
5860}
5861sub migrate_from_v1 {
5863 my $git_dir = $ENV{GIT_DIR};
5864 my $migrated = 0;
5865 return $migrated unless -d $git_dir;
5866 my $svn_dir = "$git_dir/svn";
5867 # just in case somebody used 'svn' as their $id at some point...
5869 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5870 print STDERR "Migrating from a git-svn v1 layout...\n";
5872 mkpath([$svn_dir]);
5873 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5874 "$svn_dir\n\t(required for this version ",
5875 "($::VERSION) of git-svn) does not exist.\n";
5876 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5877 while (<$fh>) {
5878 my $x = $_;
5879 next unless $x =~ s#^refs/remotes/##;
5880 chomp $x;
5881 next unless -f "$git_dir/$x/info/url";
5882 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5883 next unless $u;
5884 my $dn = dirname("$git_dir/svn/$x");
5885 mkpath([$dn]) unless -d $dn;
5886 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5887 mkpath(["$git_dir/svn/svn"]);
5888 print STDERR " - $git_dir/$x/info => ",
5889 "$git_dir/svn/$x/info\n";
5890 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5891 croak "$!: $x";
5892 # don't worry too much about these, they probably
5893 # don't exist with repos this old (save for index,
5894 # and we can easily regenerate that)
5895 foreach my $f (qw/unhandled.log index .rev_db/) {
5896 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5897 }
5898 } else {
5899 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5900 rename "$git_dir/$x", "$git_dir/svn/$x" or
5901 croak "$!: $x";
5902 }
5903 $migrated++;
5904 }
5905 command_close_pipe($fh, $ctx);
5906 print STDERR "Done migrating from a git-svn v1 layout\n";
5907 $migrated;
5908}
5909sub read_old_urls {
5911 my ($l_map, $pfx, $path) = @_;
5912 my @dir;
5913 foreach (<$path/*>) {
5914 if (-r "$_/info/url") {
5915 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5916 my $ref_id = $pfx . basename $_;
5917 my $url = ::file_to_s("$_/info/url");
5918 $l_map->{$ref_id} = $url;
5919 } elsif (-d $_) {
5920 push @dir, $_;
5921 }
5922 }
5923 foreach (@dir) {
5924 my $x = $_;
5925 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5926 read_old_urls($l_map, $x, $_);
5927 }
5928}
5929sub migrate_from_v2 {
5931 my @cfg = command(qw/config -l/);
5932 return if grep /^svn-remote\..+\.url=/, @cfg;
5933 my %l_map;
5934 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5935 my $migrated = 0;
5936 foreach my $ref_id (sort keys %l_map) {
5938 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5939 if ($@) {
5940 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5941 }
5942 $migrated++;
5943 }
5944 $migrated;
5945}
5946sub minimize_connections {
5948 my $r = Git::SVN::read_all_remotes();
5949 my $new_urls = {};
5950 my $root_repos = {};
5951 foreach my $repo_id (keys %$r) {
5952 my $url = $r->{$repo_id}->{url} or next;
5953 my $fetch = $r->{$repo_id}->{fetch} or next;
5954 my $ra = Git::SVN::Ra->new($url);
5955 # skip existing cases where we already connect to the root
5957 if (($ra->{url} eq $ra->{repos_root}) ||
5958 ($ra->{repos_root} eq $repo_id)) {
5959 $root_repos->{$ra->{url}} = $repo_id;
5960 next;
5961 }
5962 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5964 my $root_path = $ra->{url};
5965 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5966 foreach my $path (keys %$fetch) {
5967 my $ref_id = $fetch->{$path};
5968 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5969 # make sure we can read when connecting to
5971 # a higher level of a repository
5972 my ($last_rev, undef) = $gs->last_rev_commit;
5973 if (!defined $last_rev) {
5974 $last_rev = eval {
5975 $root_ra->get_latest_revnum;
5976 };
5977 next if $@;
5978 }
5979 my $new = $root_path;
5980 $new .= length $path ? "/$path" : '';
5981 eval {
5982 $root_ra->get_log([$new], $last_rev, $last_rev,
5983 0, 0, 1, sub { });
5984 };
5985 next if $@;
5986 $new_urls->{$ra->{repos_root}}->{$new} =
5987 { ref_id => $ref_id,
5988 old_repo_id => $repo_id,
5989 old_path => $path };
5990 }
5991 }
5992 my @emptied;
5994 foreach my $url (keys %$new_urls) {
5995 # see if we can re-use an existing [svn-remote "repo_id"]
5996 # instead of creating a(n ugly) new section:
5997 my $repo_id = $root_repos->{$url} || $url;
5998 my $fetch = $new_urls->{$url};
6000 foreach my $path (keys %$fetch) {
6001 my $x = $fetch->{$path};
6002 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
6003 my $pfx = "svn-remote.$x->{old_repo_id}";
6004 my $old_fetch = quotemeta("$x->{old_path}:".
6006 "$x->{ref_id}");
6007 command_noisy(qw/config --unset/,
6008 "$pfx.fetch", '^'. $old_fetch . '$');
6009 delete $r->{$x->{old_repo_id}}->
6010 {fetch}->{$x->{old_path}};
6011 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6012 command_noisy(qw/config --unset/,
6013 "$pfx.url");
6014 push @emptied, $x->{old_repo_id}
6015 }
6016 }
6017 }
6018 if (@emptied) {
6019 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6020 print STDERR <<EOF;
6021The following [svn-remote] sections in your config file ($file) are empty
6022and can be safely removed:
6023EOF
6024 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6025 }
6026}
6027sub migration_check {
6029 migrate_from_v0();
6030 migrate_from_v1();
6031 migrate_from_v2();
6032 minimize_connections() if $_minimize;
6033}
6034package Git::IndexInfo;
6036use strict;
6037use warnings;
6038use Git qw/command_input_pipe command_close_pipe/;
6039sub new {
6041 my ($class) = @_;
6042 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6043 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6044}
6045sub remove {
6047 my ($self, $path) = @_;
6048 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6049 return ++$self->{nr};
6050 }
6051 undef;
6052}
6053sub update {
6055 my ($self, $mode, $hash, $path) = @_;
6056 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6057 return ++$self->{nr};
6058 }
6059 undef;
6060}
6061sub DESTROY {
6063 my ($self) = @_;
6064 command_close_pipe($self->{gui}, $self->{ctx});
6065}
6066package Git::SVN::GlobSpec;
6068use strict;
6069use warnings;
6070sub new {
6072 my ($class, $glob, $pattern_ok) = @_;
6073 my $re = $glob;
6074 $re =~ s!/+$!!g; # no need for trailing slashes
6075 my (@left, @right, @patterns);
6076 my $state = "left";
6077 my $die_msg = "Only one set of wildcard directories " .
6078 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6079 for my $part (split(m|/|, $glob)) {
6080 if ($part =~ /\*/ && $part ne "*") {
6081 die "Invalid pattern in '$glob': $part\n";
6082 } elsif ($pattern_ok && $part =~ /[{}]/ &&
6083 $part !~ /^\{[^{}]+\}/) {
6084 die "Invalid pattern in '$glob': $part\n";
6085 }
6086 if ($part eq "*") {
6087 die $die_msg if $state eq "right";
6088 $state = "pattern";
6089 push(@patterns, "[^/]*");
6090 } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6091 die $die_msg if $state eq "right";
6092 $state = "pattern";
6093 my $p = quotemeta($1);
6094 $p =~ s/\\,/|/g;
6095 push(@patterns, "(?:$p)");
6096 } else {
6097 if ($state eq "left") {
6098 push(@left, $part);
6099 } else {
6100 push(@right, $part);
6101 $state = "right";
6102 }
6103 }
6104 }
6105 my $depth = @patterns;
6106 if ($depth == 0) {
6107 die "One '*' is needed in glob: '$glob'\n";
6108 }
6109 my $left = join('/', @left);
6110 my $right = join('/', @right);
6111 $re = join('/', @patterns);
6112 $re = join('\/',
6113 grep(length, quotemeta($left), "($re)", quotemeta($right)));
6114 my $left_re = qr/^\/\Q$left\E(\/|$)/;
6115 bless { left => $left, right => $right, left_regex => $left_re,
6116 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6117}
6118sub full_path {
6120 my ($self, $path) = @_;
6121 return (length $self->{left} ? "$self->{left}/" : '') .
6122 $path . (length $self->{right} ? "/$self->{right}" : '');
6123}
6124__END__
6126Data structures:
6128$remotes = { # returned by read_all_remotes()
6131 'svn' => {
6132 # svn-remote.svn.url=https://svn.musicpd.org
6133 url => 'https://svn.musicpd.org',
6134 # svn-remote.svn.fetch=mpd/trunk:trunk
6135 fetch => {
6136 'mpd/trunk' => 'trunk',
6137 },
6138 # svn-remote.svn.tags=mpd/tags/*:tags/*
6139 tags => {
6140 path => {
6141 left => 'mpd/tags',
6142 right => '',
6143 regex => qr!mpd/tags/([^/]+)$!,
6144 glob => 'tags/*',
6145 },
6146 ref => {
6147 left => 'tags',
6148 right => '',
6149 regex => qr!tags/([^/]+)$!,
6150 glob => 'tags/*',
6151 },
6152 }
6153 }
6154};
6155$log_entry hashref as returned by libsvn_log_entry()
6157{
6158 log => 'whitespace-formatted log entry
6159', # trailing newline is preserved
6160 revision => '8', # integer
6161 date => '2004-02-24T17:01:44.108345Z', # commit date
6162 author => 'committer name'
6163};
6164# this is generated by generate_diff();
6167@mods = array of diff-index line hashes, each element represents one line
6168 of diff-index output
6169diff-index line ($m hash)
6171{
6172 mode_a => first column of diff-index output, no leading ':',
6173 mode_b => second column of diff-index output,
6174 sha1_b => sha1sum of the final blob,
6175 chg => change type [MCRADT],
6176 file_a => original file name of a file (iff chg is 'C' or 'R')
6177 file_b => new/current file name of a file (any chg)
6178}
6179;
6180# retval of read_url_paths{,_all}();
6182$l_map = {
6183 # repository root url
6184 'https://svn.musicpd.org' => {
6185 # repository path # GIT_SVN_ID
6186 'mpd/trunk' => 'trunk',
6187 'mpd/tags/0.11.5' => 'tags/0.11.5',
6188 },
6189}
6190Notes:
6192 I don't trust the each() function on unless I created %hash myself
6193 because the internal iterator may not have started at base.