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 $_trunk =~ s#^/+##;
967 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
968 # try both old-style and new-style lookups:
969 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
970 unless ($gs_trunk) {
971 my ($trunk_url, $trunk_path) =
972 complete_svn_url($url, $_trunk);
973 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
974 undef, $trunk_ref);
975 }
976 }
977 return unless @_branches || @_tags;
978 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
979 foreach my $path (@_branches) {
980 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
981 }
982 foreach my $path (@_tags) {
983 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
984 }
985}
986sub cmd_multi_fetch {
988 $Git::SVN::no_reuse_existing = undef;
989 my $remotes = Git::SVN::read_all_remotes();
990 foreach my $repo_id (sort keys %$remotes) {
991 if ($remotes->{$repo_id}->{url}) {
992 Git::SVN::fetch_all($repo_id, $remotes);
993 }
994 }
995}
996# this command is special because it requires no metadata
998sub cmd_commit_diff {
999 my ($ta, $tb, $url) = @_;
1000 my $usage = "Usage: $0 commit-diff -r<revision> ".
1001 "<tree-ish> <tree-ish> [<URL>]";
1002 fatal($usage) if (!defined $ta || !defined $tb);
1003 my $svn_path = '';
1004 if (!defined $url) {
1005 my $gs = eval { Git::SVN->new };
1006 if (!$gs) {
1007 fatal("Needed URL or usable git-svn --id in ",
1008 "the command-line\n", $usage);
1009 }
1010 $url = $gs->{url};
1011 $svn_path = $gs->{path};
1012 }
1013 unless (defined $_revision) {
1014 fatal("-r|--revision is a required argument\n", $usage);
1015 }
1016 if (defined $_message && defined $_file) {
1017 fatal("Both --message/-m and --file/-F specified ",
1018 "for the commit message.\n",
1019 "I have no idea what you mean");
1020 }
1021 if (defined $_file) {
1022 $_message = file_to_s($_file);
1023 } else {
1024 $_message ||= get_commit_entry($tb)->{log};
1025 }
1026 my $ra ||= Git::SVN::Ra->new($url);
1027 my $r = $_revision;
1028 if ($r eq 'HEAD') {
1029 $r = $ra->get_latest_revnum;
1030 } elsif ($r !~ /^\d+$/) {
1031 die "revision argument: $r not understood by git-svn\n";
1032 }
1033 my %ed_opts = ( r => $r,
1034 log => $_message,
1035 ra => $ra,
1036 tree_a => $ta,
1037 tree_b => $tb,
1038 editor_cb => sub { print "Committed r$_[0]\n" },
1039 svn_path => $svn_path );
1040 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1041 print "No changes\n$ta == $tb\n";
1042 }
1043}
1044sub escape_uri_only {
1046 my ($uri) = @_;
1047 my @tmp;
1048 foreach (split m{/}, $uri) {
1049 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1050 push @tmp, $_;
1051 }
1052 join('/', @tmp);
1053}
1054sub escape_url {
1056 my ($url) = @_;
1057 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1058 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1059 $url = "$scheme://$domain$uri";
1060 }
1061 $url;
1062}
1063sub cmd_info {
1065 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1066 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1067 if (exists $_[1]) {
1068 die "Too many arguments specified\n";
1069 }
1070 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1072 if (!$file_type && !$diff_status) {
1074 print STDERR "svn: '$path' is not under version control\n";
1075 exit 1;
1076 }
1077 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1079 unless ($gs) {
1080 die "Unable to determine upstream SVN information from ",
1081 "working tree history\n";
1082 }
1083 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1085 $path = "." if $path eq "";
1086 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1088 if ($_url) {
1090 print escape_url($full_url), "\n";
1091 return;
1092 }
1093 my $result = "Path: $path\n";
1095 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1096 $result .= "URL: " . escape_url($full_url) . "\n";
1097 eval {
1099 my $repos_root = $gs->repos_root;
1100 Git::SVN::remove_username($repos_root);
1101 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1102 };
1103 if ($@) {
1104 $result .= "Repository Root: (offline)\n";
1105 }
1106 ::_req_svn();
1107 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1108 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1109 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1110 $result .= "Node Kind: " .
1112 ($file_type eq "dir" ? "directory" : "file") . "\n";
1113 my $schedule = $diff_status eq "A"
1115 ? "add"
1116 : ($diff_status eq "D" ? "delete" : "normal");
1117 $result .= "Schedule: $schedule\n";
1118 if ($diff_status eq "A") {
1120 print $result, "\n";
1121 return;
1122 }
1123 my ($lc_author, $lc_rev, $lc_date_utc);
1125 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1126 my $log = command_output_pipe(@args);
1127 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1128 while (<$log>) {
1129 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1130 $lc_author = $1;
1131 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1132 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1133 (undef, $lc_rev, undef) = ::extract_metadata($1);
1134 }
1135 }
1136 close $log;
1137 Git::SVN::Log::set_local_timezone();
1139 $result .= "Last Changed Author: $lc_author\n";
1141 $result .= "Last Changed Rev: $lc_rev\n";
1142 $result .= "Last Changed Date: " .
1143 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1144 if ($file_type ne "dir") {
1146 my $text_last_updated_date =
1147 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1148 $result .=
1149 "Text Last Updated: " .
1150 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1151 "\n";
1152 my $checksum;
1153 if ($diff_status eq "D") {
1154 my ($fh, $ctx) =
1155 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1156 if ($file_type eq "link") {
1157 my $file_name = <$fh>;
1158 $checksum = md5sum("link $file_name");
1159 } else {
1160 $checksum = md5sum($fh);
1161 }
1162 command_close_pipe($fh, $ctx);
1163 } elsif ($file_type eq "link") {
1164 my $file_name =
1165 command(qw(cat-file blob), "HEAD:$path");
1166 $checksum =
1167 md5sum("link " . $file_name);
1168 } else {
1169 open FILE, "<", $path or die $!;
1170 $checksum = md5sum(\*FILE);
1171 close FILE or die $!;
1172 }
1173 $result .= "Checksum: " . $checksum . "\n";
1174 }
1175 print $result, "\n";
1177}
1178sub cmd_reset {
1180 my $target = shift || $_revision or die "SVN revision required\n";
1181 $target = $1 if $target =~ /^r(\d+)$/;
1182 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1183 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1184 unless ($gs) {
1185 die "Unable to determine upstream SVN information from ".
1186 "history\n";
1187 }
1188 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1189 die "Cannot find SVN revision $target\n" unless defined($c);
1190 $gs->rev_map_set($r, $c, 'reset', $uuid);
1191 print "r$r = $c ($gs->{ref_id})\n";
1192}
1193sub cmd_gc {
1195 if (!$can_compress) {
1196 warn "Compress::Zlib could not be found; unhandled.log " .
1197 "files will not be compressed.\n";
1198 }
1199 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1200}
1201########################### utility functions #########################
1203sub rebase_cmd {
1205 my @cmd = qw/rebase/;
1206 push @cmd, '-v' if $_verbose;
1207 push @cmd, qw/--merge/ if $_merge;
1208 push @cmd, "--strategy=$_strategy" if $_strategy;
1209 @cmd;
1210}
1211sub post_fetch_checkout {
1213 return if $_no_checkout;
1214 my $gs = $Git::SVN::_head or return;
1215 return if verify_ref('refs/heads/master^0');
1216 # look for "trunk" ref if it exists
1218 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1219 my $fetch = $remote->{fetch};
1220 if ($fetch) {
1221 foreach my $p (keys %$fetch) {
1222 basename($fetch->{$p}) eq 'trunk' or next;
1223 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1224 last;
1225 }
1226 }
1227 my $valid_head = verify_ref('HEAD^0');
1229 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1230 return if ($valid_head || !verify_ref('HEAD^0'));
1231 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1233 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1234 return if -f $index;
1235 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1237 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1238 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1239 print STDERR "Checked out HEAD:\n ",
1240 $gs->full_url, " r", $gs->last_rev, "\n";
1241 $gs->mkemptydirs($gs->last_rev);
1242}
1243sub complete_svn_url {
1245 my ($url, $path) = @_;
1246 $path =~ s#/+$##;
1247 if ($path !~ m#^[a-z\+]+://#) {
1248 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1249 fatal("E: '$path' is not a complete URL ",
1250 "and a separate URL is not specified");
1251 }
1252 return ($url, $path);
1253 }
1254 return ($path, '');
1255}
1256sub complete_url_ls_init {
1258 my ($ra, $repo_path, $switch, $pfx) = @_;
1259 unless ($repo_path) {
1260 print STDERR "W: $switch not specified\n";
1261 return;
1262 }
1263 $repo_path =~ s#/+$##;
1264 if ($repo_path =~ m#^[a-z\+]+://#) {
1265 $ra = Git::SVN::Ra->new($repo_path);
1266 $repo_path = '';
1267 } else {
1268 $repo_path =~ s#^/+##;
1269 unless ($ra) {
1270 fatal("E: '$repo_path' is not a complete URL ",
1271 "and a separate URL is not specified");
1272 }
1273 }
1274 my $url = $ra->{url};
1275 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1276 my $k = "svn-remote.$gs->{repo_id}.url";
1277 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1278 if ($orig_url && ($orig_url ne $gs->{url})) {
1279 die "$k already set: $orig_url\n",
1280 "wanted to set to: $gs->{url}\n";
1281 }
1282 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1283 my $remote_path = "$gs->{path}/$repo_path";
1284 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1285 $remote_path =~ s#/+#/#g;
1286 $remote_path =~ s#^/##g;
1287 $remote_path .= "/*" if $remote_path !~ /\*/;
1288 my ($n) = ($switch =~ /^--(\w+)/);
1289 if (length $pfx && $pfx !~ m#/$#) {
1290 die "--prefix='$pfx' must have a trailing slash '/'\n";
1291 }
1292 command_noisy('config',
1293 '--add',
1294 "svn-remote.$gs->{repo_id}.$n",
1295 "$remote_path:refs/remotes/$pfx*" .
1296 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1297}
1298sub verify_ref {
1300 my ($ref) = @_;
1301 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1302 { STDERR => 0 }); };
1303}
1304sub get_tree_from_treeish {
1306 my ($treeish) = @_;
1307 # $treeish can be a symbolic ref, too:
1308 my $type = command_oneline(qw/cat-file -t/, $treeish);
1309 my $expected;
1310 while ($type eq 'tag') {
1311 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1312 }
1313 if ($type eq 'commit') {
1314 $expected = (grep /^tree /, command(qw/cat-file commit/,
1315 $treeish))[0];
1316 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1317 die "Unable to get tree from $treeish\n" unless $expected;
1318 } elsif ($type eq 'tree') {
1319 $expected = $treeish;
1320 } else {
1321 die "$treeish is a $type, expected tree, tag or commit\n";
1322 }
1323 return $expected;
1324}
1325sub get_commit_entry {
1327 my ($treeish) = shift;
1328 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1329 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1330 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1331 open my $log_fh, '>', $commit_editmsg or croak $!;
1332 my $type = command_oneline(qw/cat-file -t/, $treeish);
1334 if ($type eq 'commit' || $type eq 'tag') {
1335 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1336 $type, $treeish);
1337 my $in_msg = 0;
1338 my $author;
1339 my $saw_from = 0;
1340 my $msgbuf = "";
1341 while (<$msg_fh>) {
1342 if (!$in_msg) {
1343 $in_msg = 1 if (/^\s*$/);
1344 $author = $1 if (/^author (.*>)/);
1345 } elsif (/^git-svn-id: /) {
1346 # skip this for now, we regenerate the
1347 # correct one on re-fetch anyways
1348 # TODO: set *:merge properties or like...
1349 } else {
1350 if (/^From:/ || /^Signed-off-by:/) {
1351 $saw_from = 1;
1352 }
1353 $msgbuf .= $_;
1354 }
1355 }
1356 $msgbuf =~ s/\s+$//s;
1357 if ($Git::SVN::_add_author_from && defined($author)
1358 && !$saw_from) {
1359 $msgbuf .= "\n\nFrom: $author";
1360 }
1361 print $log_fh $msgbuf or croak $!;
1362 command_close_pipe($msg_fh, $ctx);
1363 }
1364 close $log_fh or croak $!;
1365 if ($_edit || ($type eq 'tree')) {
1367 chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1368 system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1369 }
1370 rename $commit_editmsg, $commit_msg or croak $!;
1371 {
1372 require Encode;
1373 # SVN requires messages to be UTF-8 when entering the repo
1374 local $/;
1375 open $log_fh, '<', $commit_msg or croak $!;
1376 binmode $log_fh;
1377 chomp($log_entry{log} = <$log_fh>);
1378 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1380 my $msg = $log_entry{log};
1381 eval { $msg = Encode::decode($enc, $msg, 1) };
1383 if ($@) {
1384 die "Could not decode as $enc:\n", $msg,
1385 "\nPerhaps you need to set i18n.commitencoding\n";
1386 }
1387 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1389 die "Could not encode as UTF-8:\n$msg\n" if $@;
1390 $log_entry{log} = $msg;
1392 close $log_fh or croak $!;
1394 }
1395 unlink $commit_msg;
1396 \%log_entry;
1397}
1398sub s_to_file {
1400 my ($str, $file, $mode) = @_;
1401 open my $fd,'>',$file or croak $!;
1402 print $fd $str,"\n" or croak $!;
1403 close $fd or croak $!;
1404 chmod ($mode &~ umask, $file) if (defined $mode);
1405}
1406sub file_to_s {
1408 my $file = shift;
1409 open my $fd,'<',$file or croak "$!: file: $file\n";
1410 local $/;
1411 my $ret = <$fd>;
1412 close $fd or croak $!;
1413 $ret =~ s/\s*$//s;
1414 return $ret;
1415}
1416# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1418sub load_authors {
1419 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1420 my $log = $cmd eq 'log';
1421 while (<$authors>) {
1422 chomp;
1423 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1424 my ($user, $name, $email) = ($1, $2, $3);
1425 if ($log) {
1426 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1427 } else {
1428 $users{$user} = [$name, $email];
1429 }
1430 }
1431 close $authors or croak $!;
1432}
1433# convert GetOpt::Long specs for use by git-config
1435sub read_git_config {
1436 my $opts = shift;
1437 my @config_only;
1438 foreach my $o (keys %$opts) {
1439 # if we have mixedCase and a long option-only, then
1440 # it's a config-only variable that we don't need for
1441 # the command-line.
1442 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1443 my $v = $opts->{$o};
1444 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1445 $key =~ s/-//g;
1446 my $arg = 'git config';
1447 $arg .= ' --int' if ($o =~ /[:=]i$/);
1448 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1449 if (ref $v eq 'ARRAY') {
1450 chomp(my @tmp = `$arg --get-all svn.$key`);
1451 @$v = @tmp if @tmp;
1452 } else {
1453 chomp(my $tmp = `$arg --get svn.$key`);
1454 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1455 $$v = $tmp;
1456 }
1457 }
1458 }
1459 delete @$opts{@config_only} if @config_only;
1460}
1461sub extract_metadata {
1463 my $id = shift or return (undef, undef, undef);
1464 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1465 \s([a-f\d\-]+)$/ix);
1466 if (!defined $rev || !$uuid || !$url) {
1467 # some of the original repositories I made had
1468 # identifiers like this:
1469 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1470 }
1471 return ($url, $rev, $uuid);
1472}
1473sub cmt_metadata {
1475 return extract_metadata((grep(/^git-svn-id: /,
1476 command(qw/cat-file commit/, shift)))[-1]);
1477}
1478sub cmt_sha2rev_batch {
1480 my %s2r;
1481 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1482 my $list = shift;
1483 foreach my $sha (@{$list}) {
1485 my $first = 1;
1486 my $size = 0;
1487 print $out $sha, "\n";
1488 while (my $line = <$in>) {
1490 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1491 last;
1492 } elsif ($first &&
1493 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1494 $first = 0;
1495 $size = $1;
1496 next;
1497 } elsif ($line =~ /^(git-svn-id: )/) {
1498 my (undef, $rev, undef) =
1499 extract_metadata($line);
1500 $s2r{$sha} = $rev;
1501 }
1502 $size -= length($line);
1504 last if ($size == 0);
1505 }
1506 }
1507 command_close_bidi_pipe($pid, $in, $out, $ctx);
1509 return \%s2r;
1511}
1512sub working_head_info {
1514 my ($head, $refs) = @_;
1515 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1516 my ($fh, $ctx) = command_output_pipe(@args, $head);
1517 my $hash;
1518 my %max;
1519 while (<$fh>) {
1520 if ( m{^commit ($::sha1)$} ) {
1521 unshift @$refs, $hash if $hash and $refs;
1522 $hash = $1;
1523 next;
1524 }
1525 next unless s{^\s*(git-svn-id:)}{$1};
1526 my ($url, $rev, $uuid) = extract_metadata($_);
1527 if (defined $url && defined $rev) {
1528 next if $max{$url} and $max{$url} < $rev;
1529 if (my $gs = Git::SVN->find_by_url($url)) {
1530 my $c = $gs->rev_map_get($rev, $uuid);
1531 if ($c && $c eq $hash) {
1532 close $fh; # break the pipe
1533 return ($url, $rev, $uuid, $gs);
1534 } else {
1535 $max{$url} ||= $gs->rev_map_max;
1536 }
1537 }
1538 }
1539 }
1540 command_close_pipe($fh, $ctx);
1541 (undef, undef, undef, undef);
1542}
1543sub read_commit_parents {
1545 my ($parents, $c) = @_;
1546 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1547 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1548 @{$parents->{$c}} = split(/ /, $p);
1549}
1550sub linearize_history {
1552 my ($gs, $refs) = @_;
1553 my %parents;
1554 foreach my $c (@$refs) {
1555 read_commit_parents(\%parents, $c);
1556 }
1557 my @linear_refs;
1559 my %skip = ();
1560 my $last_svn_commit = $gs->last_commit;
1561 foreach my $c (reverse @$refs) {
1562 next if $c eq $last_svn_commit;
1563 last if $skip{$c};
1564 unshift @linear_refs, $c;
1566 $skip{$c} = 1;
1567 # we only want the first parent to diff against for linear
1569 # history, we save the rest to inject when we finalize the
1570 # svn commit
1571 my $fp_a = verify_ref("$c~1");
1572 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1573 if (!$fp_a || !$fp_b) {
1574 die "Commit $c\n",
1575 "has no parent commit, and therefore ",
1576 "nothing to diff against.\n",
1577 "You should be working from a repository ",
1578 "originally created by git-svn\n";
1579 }
1580 if ($fp_a ne $fp_b) {
1581 die "$c~1 = $fp_a, however parsing commit $c ",
1582 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1583 }
1584 foreach my $p (@{$parents{$c}}) {
1586 $skip{$p} = 1;
1587 }
1588 }
1589 (\@linear_refs, \%parents);
1590}
1591sub find_file_type_and_diff_status {
1593 my ($path) = @_;
1594 return ('dir', '') if $path eq '';
1595 my $diff_output =
1597 command_oneline(qw(diff --cached --name-status --), $path) || "";
1598 my $diff_status = (split(' ', $diff_output))[0] || "";
1599 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1601 return (undef, undef) if !$diff_status && !$ls_tree;
1603 if ($diff_status eq "A") {
1605 return ("link", $diff_status) if -l $path;
1606 return ("dir", $diff_status) if -d $path;
1607 return ("file", $diff_status);
1608 }
1609 my $mode = (split(' ', $ls_tree))[0] || "";
1611 return ("link", $diff_status) if $mode eq "120000";
1613 return ("dir", $diff_status) if $mode eq "040000";
1614 return ("file", $diff_status);
1615}
1616sub md5sum {
1618 my $arg = shift;
1619 my $ref = ref $arg;
1620 my $md5 = Digest::MD5->new();
1621 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1622 $md5->addfile($arg) or croak $!;
1623 } elsif ($ref eq 'SCALAR') {
1624 $md5->add($$arg) or croak $!;
1625 } elsif (!$ref) {
1626 $md5->add($arg) or croak $!;
1627 } else {
1628 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1629 }
1630 return $md5->hexdigest();
1631}
1632sub gc_directory {
1634 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1635 my $out_filename = $_ . ".gz";
1636 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1637 binmode $in_fh;
1638 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1639 die "Unable to open $out_filename: $!\n";
1640 my $res;
1642 while ($res = sysread($in_fh, my $str, 1024)) {
1643 $gz->gzwrite($str) or
1644 die "Unable to write: ".$gz->gzerror()."!\n";
1645 }
1646 unlink $_ or die "unlink $File::Find::name: $!\n";
1647 } elsif (-f $_ && basename($_) eq "index") {
1648 unlink $_ or die "unlink $_: $!\n";
1649 }
1650}
1651package Git::SVN;
1653use strict;
1654use warnings;
1655use Fcntl qw/:DEFAULT :seek/;
1656use constant rev_map_fmt => 'NH40';
1657use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1658 $_repack $_repack_flags $_use_svm_props $_head
1659 $_use_svnsync_props $no_reuse_existing $_minimize_url
1660 $_use_log_author $_add_author_from $_localtime/;
1661use Carp qw/croak/;
1662use File::Path qw/mkpath/;
1663use File::Copy qw/copy/;
1664use IPC::Open3;
1665use Memoize; # core since 5.8.0, Jul 2002
1666use Memoize::Storable;
1667my ($_gc_nr, $_gc_period);
1669# properties that we do not log:
1671my %SKIP_PROP;
1672BEGIN {
1673 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1674 svn:special svn:executable
1675 svn:entry:committed-rev
1676 svn:entry:last-author
1677 svn:entry:uuid
1678 svn:entry:committed-date/;
1679 # some options are read globally, but can be overridden locally
1681 # per [svn-remote "..."] section. Command-line options will *NOT*
1682 # override options set in an [svn-remote "..."] section
1683 no strict 'refs';
1684 for my $option (qw/follow_parent no_metadata use_svm_props
1685 use_svnsync_props/) {
1686 my $key = $option;
1687 $key =~ tr/_//d;
1688 my $prop = "-$option";
1689 *$option = sub {
1690 my ($self) = @_;
1691 return $self->{$prop} if exists $self->{$prop};
1692 my $k = "svn-remote.$self->{repo_id}.$key";
1693 eval { command_oneline(qw/config --get/, $k) };
1694 if ($@) {
1695 $self->{$prop} = ${"Git::SVN::_$option"};
1696 } else {
1697 my $v = command_oneline(qw/config --bool/,$k);
1698 $self->{$prop} = $v eq 'false' ? 0 : 1;
1699 }
1700 return $self->{$prop};
1701 }
1702 }
1703}
1704my (%LOCKFILES, %INDEX_FILES);
1707END {
1708 unlink keys %LOCKFILES if %LOCKFILES;
1709 unlink keys %INDEX_FILES if %INDEX_FILES;
1710}
1711sub resolve_local_globs {
1713 my ($url, $fetch, $glob_spec) = @_;
1714 return unless defined $glob_spec;
1715 my $ref = $glob_spec->{ref};
1716 my $path = $glob_spec->{path};
1717 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1718 next unless m#^$ref->{regex}$#;
1719 my $p = $1;
1720 my $pathname = desanitize_refname($path->full_path($p));
1721 my $refname = desanitize_refname($ref->full_path($p));
1722 if (my $existing = $fetch->{$pathname}) {
1723 if ($existing ne $refname) {
1724 die "Refspec conflict:\n",
1725 "existing: $existing\n",
1726 " globbed: $refname\n";
1727 }
1728 my $u = (::cmt_metadata("$refname"))[0];
1729 $u =~ s!^\Q$url\E(/|$)!! or die
1730 "$refname: '$url' not found in '$u'\n";
1731 if ($pathname ne $u) {
1732 warn "W: Refspec glob conflict ",
1733 "(ref: $refname):\n",
1734 "expected path: $pathname\n",
1735 " real path: $u\n",
1736 "Continuing ahead with $u\n";
1737 next;
1738 }
1739 } else {
1740 $fetch->{$pathname} = $refname;
1741 }
1742 }
1743}
1744sub parse_revision_argument {
1746 my ($base, $head) = @_;
1747 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1748 return ($base, $head);
1749 }
1750 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1751 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1752 return ($head, $head) if ($::_revision eq 'HEAD');
1753 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1754 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1755 die "revision argument: $::_revision not understood by git-svn\n";
1756}
1757sub fetch_all {
1759 my ($repo_id, $remotes) = @_;
1760 if (ref $repo_id) {
1761 my $gs = $repo_id;
1762 $repo_id = undef;
1763 $repo_id = $gs->{repo_id};
1764 }
1765 $remotes ||= read_all_remotes();
1766 my $remote = $remotes->{$repo_id} or
1767 die "[svn-remote \"$repo_id\"] unknown\n";
1768 my $fetch = $remote->{fetch};
1769 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1770 my (@gs, @globs);
1771 my $ra = Git::SVN::Ra->new($url);
1772 my $uuid = $ra->get_uuid;
1773 my $head = $ra->get_latest_revnum;
1774 # ignore errors, $head revision may not even exist anymore
1776 eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
1777 warn "W: $@\n" if $@;
1778 my $base = defined $fetch ? $head : 0;
1780 # read the max revs for wildcard expansion (branches/*, tags/*)
1782 foreach my $t (qw/branches tags/) {
1783 defined $remote->{$t} or next;
1784 push @globs, @{$remote->{$t}};
1785 my $max_rev = eval { tmp_config(qw/--int --get/,
1787 "svn-remote.$repo_id.${t}-maxRev") };
1788 if (defined $max_rev && ($max_rev < $base)) {
1789 $base = $max_rev;
1790 } elsif (!defined $max_rev) {
1791 $base = 0;
1792 }
1793 }
1794 if ($fetch) {
1796 foreach my $p (sort keys %$fetch) {
1797 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1798 my $lr = $gs->rev_map_max;
1799 if (defined $lr) {
1800 $base = $lr if ($lr < $base);
1801 }
1802 push @gs, $gs;
1803 }
1804 }
1805 ($base, $head) = parse_revision_argument($base, $head);
1807 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1808}
1809sub read_all_remotes {
1811 my $r = {};
1812 my $use_svm_props = eval { command_oneline(qw/config --bool
1813 svn.useSvmProps/) };
1814 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1815 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1816 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1817 if (m!^(.+)\.fetch=$svn_refspec$!) {
1818 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1819 die("svn-remote.$remote: remote ref '$remote_ref' "
1820 . "must start with 'refs/'\n")
1821 unless $remote_ref =~ m{^refs/};
1822 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1823 $r->{$remote}->{svm} = {} if $use_svm_props;
1824 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1825 $r->{$1}->{svm} = {};
1826 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1827 $r->{$1}->{url} = $2;
1828 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1829 my ($remote, $t, $local_ref, $remote_ref) =
1830 ($1, $2, $3, $4);
1831 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1832 . "must start with 'refs/'\n")
1833 unless $remote_ref =~ m{^refs/};
1834 my $rs = {
1835 t => $t,
1836 remote => $remote,
1837 path => Git::SVN::GlobSpec->new($local_ref, 1),
1838 ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
1839 if (length($rs->{ref}->{right}) != 0) {
1840 die "The '*' glob character must be the last ",
1841 "character of '$remote_ref'\n";
1842 }
1843 push @{ $r->{$remote}->{$t} }, $rs;
1844 }
1845 }
1846 map {
1848 if (defined $r->{$_}->{svm}) {
1849 my $svm;
1850 eval {
1851 my $section = "svn-remote.$_";
1852 $svm = {
1853 source => tmp_config('--get',
1854 "$section.svm-source"),
1855 replace => tmp_config('--get',
1856 "$section.svm-replace"),
1857 }
1858 };
1859 $r->{$_}->{svm} = $svm;
1860 }
1861 } keys %$r;
1862 $r;
1864}
1865sub init_vars {
1867 $_gc_nr = $_gc_period = 1000;
1868 if (defined $_repack || defined $_repack_flags) {
1869 warn "Repack options are obsolete; they have no effect.\n";
1870 }
1871}
1872sub verify_remotes_sanity {
1874 return unless -d $ENV{GIT_DIR};
1875 my %seen;
1876 foreach (command(qw/config -l/)) {
1877 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1878 if ($seen{$1}) {
1879 die "Remote ref refs/remote/$1 is tracked by",
1880 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1881 "Please resolve this ambiguity in ",
1882 "your git configuration file before ",
1883 "continuing\n";
1884 }
1885 $seen{$1} = $_;
1886 }
1887 }
1888}
1889sub find_existing_remote {
1891 my ($url, $remotes) = @_;
1892 return undef if $no_reuse_existing;
1893 my $existing;
1894 foreach my $repo_id (keys %$remotes) {
1895 my $u = $remotes->{$repo_id}->{url} or next;
1896 next if $u ne $url;
1897 $existing = $repo_id;
1898 last;
1899 }
1900 $existing;
1901}
1902sub init_remote_config {
1904 my ($self, $url, $no_write) = @_;
1905 $url =~ s!/+$!!; # strip trailing slash
1906 my $r = read_all_remotes();
1907 my $existing = find_existing_remote($url, $r);
1908 if ($existing) {
1909 unless ($no_write) {
1910 print STDERR "Using existing ",
1911 "[svn-remote \"$existing\"]\n";
1912 }
1913 $self->{repo_id} = $existing;
1914 } elsif ($_minimize_url) {
1915 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1916 $existing = find_existing_remote($min_url, $r);
1917 if ($existing) {
1918 unless ($no_write) {
1919 print STDERR "Using existing ",
1920 "[svn-remote \"$existing\"]\n";
1921 }
1922 $self->{repo_id} = $existing;
1923 }
1924 if ($min_url ne $url) {
1925 unless ($no_write) {
1926 print STDERR "Using higher level of URL: ",
1927 "$url => $min_url\n";
1928 }
1929 my $old_path = $self->{path};
1930 $self->{path} = $url;
1931 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1932 if (length $old_path) {
1933 $self->{path} .= "/$old_path";
1934 }
1935 $url = $min_url;
1936 }
1937 }
1938 my $orig_url;
1939 if (!$existing) {
1940 # verify that we aren't overwriting anything:
1941 $orig_url = eval {
1942 command_oneline('config', '--get',
1943 "svn-remote.$self->{repo_id}.url")
1944 };
1945 if ($orig_url && ($orig_url ne $url)) {
1946 die "svn-remote.$self->{repo_id}.url already set: ",
1947 "$orig_url\nwanted to set to: $url\n";
1948 }
1949 }
1950 my ($xrepo_id, $xpath) = find_ref($self->refname);
1951 if (!$no_write && defined $xpath) {
1952 die "svn-remote.$xrepo_id.fetch already set to track ",
1953 "$xpath:", $self->refname, "\n";
1954 }
1955 unless ($no_write) {
1956 command_noisy('config',
1957 "svn-remote.$self->{repo_id}.url", $url);
1958 $self->{path} =~ s{^/}{};
1959 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1960 command_noisy('config', '--add',
1961 "svn-remote.$self->{repo_id}.fetch",
1962 "$self->{path}:".$self->refname);
1963 }
1964 $self->{url} = $url;
1965}
1966sub find_by_url { # repos_root and, path are optional
1968 my ($class, $full_url, $repos_root, $path) = @_;
1969 return undef unless defined $full_url;
1971 remove_username($full_url);
1972 remove_username($repos_root) if defined $repos_root;
1973 my $remotes = read_all_remotes();
1974 if (defined $full_url && defined $repos_root && !defined $path) {
1975 $path = $full_url;
1976 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1977 }
1978 foreach my $repo_id (keys %$remotes) {
1979 my $u = $remotes->{$repo_id}->{url} or next;
1980 remove_username($u);
1981 next if defined $repos_root && $repos_root ne $u;
1982 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1984 foreach my $t (qw/branches tags/) {
1985 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1986 resolve_local_globs($u, $fetch, $globspec);
1987 }
1988 }
1989 my $p = $path;
1990 my $rwr = rewrite_root({repo_id => $repo_id});
1991 my $svm = $remotes->{$repo_id}->{svm}
1992 if defined $remotes->{$repo_id}->{svm};
1993 unless (defined $p) {
1994 $p = $full_url;
1995 my $z = $u;
1996 my $prefix = '';
1997 if ($rwr) {
1998 $z = $rwr;
1999 remove_username($z);
2000 } elsif (defined $svm) {
2001 $z = $svm->{source};
2002 $prefix = $svm->{replace};
2003 $prefix =~ s#^\Q$u\E(?:/|$)##;
2004 $prefix =~ s#/$##;
2005 }
2006 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2007 }
2008 foreach my $f (keys %$fetch) {
2009 next if $f ne $p;
2010 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2011 }
2012 }
2013 undef;
2014}
2015sub init {
2017 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2018 my $self = _new($class, $repo_id, $ref_id, $path);
2019 if (defined $url) {
2020 $self->init_remote_config($url, $no_write);
2021 }
2022 $self;
2023}
2024sub find_ref {
2026 my ($ref_id) = @_;
2027 foreach (command(qw/config -l/)) {
2028 next unless m!^svn-remote\.(.+)\.fetch=
2029 \s*(.*?)\s*:\s*(.+?)\s*$!x;
2030 my ($repo_id, $path, $ref) = ($1, $2, $3);
2031 if ($ref eq $ref_id) {
2032 $path = '' if ($path =~ m#^\./?#);
2033 return ($repo_id, $path);
2034 }
2035 }
2036 (undef, undef, undef);
2037}
2038sub new {
2040 my ($class, $ref_id, $repo_id, $path) = @_;
2041 if (defined $ref_id && !defined $repo_id && !defined $path) {
2042 ($repo_id, $path) = find_ref($ref_id);
2043 if (!defined $repo_id) {
2044 die "Could not find a \"svn-remote.*.fetch\" key ",
2045 "in the repository configuration matching: ",
2046 "$ref_id\n";
2047 }
2048 }
2049 my $self = _new($class, $repo_id, $ref_id, $path);
2050 if (!defined $self->{path} || !length $self->{path}) {
2051 my $fetch = command_oneline('config', '--get',
2052 "svn-remote.$repo_id.fetch",
2053 ":$ref_id\$") or
2054 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2055 "\":$ref_id\$\" in config\n";
2056 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2057 }
2058 $self->{path} =~ s{/+}{/}g;
2059 $self->{path} =~ s{\A/}{};
2060 $self->{path} =~ s{/\z}{};
2061 $self->{url} = command_oneline('config', '--get',
2062 "svn-remote.$repo_id.url") or
2063 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2064 $self->rebuild;
2065 $self;
2066}
2067sub refname {
2069 my ($refname) = $_[0]->{ref_id} ;
2070 # It cannot end with a slash /, we'll throw up on this because
2072 # SVN can't have directories with a slash in their name, either:
2073 if ($refname =~ m{/$}) {
2074 die "ref: '$refname' ends with a trailing slash, this is ",
2075 "not permitted by git nor Subversion\n";
2076 }
2077 # It cannot have ASCII control character space, tilde ~, caret ^,
2079 # colon :, question-mark ?, asterisk *, space, or open bracket [
2080 # anywhere.
2081 #
2082 # Additionally, % must be escaped because it is used for escaping
2083 # and we want our escaped refname to be reversible
2084 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2085 # no slash-separated component can begin with a dot .
2087 # /.* becomes /%2E*
2088 $refname =~ s{/\.}{/%2E}g;
2089 # It cannot have two consecutive dots .. anywhere
2091 # .. becomes %2E%2E
2092 $refname =~ s{\.\.}{%2E%2E}g;
2093 # trailing dots and .lock are not allowed
2095 # .$ becomes %2E and .lock becomes %2Elock
2096 $refname =~ s{\.(?=$|lock$)}{%2E};
2097 # the sequence @{ is used to access the reflog
2099 # @{ becomes %40{
2100 $refname =~ s{\@\{}{%40\{}g;
2101 return $refname;
2103}
2104sub desanitize_refname {
2106 my ($refname) = @_;
2107 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2108 return $refname;
2109}
2110sub svm_uuid {
2112 my ($self) = @_;
2113 return $self->{svm}->{uuid} if $self->svm;
2114 $self->ra;
2115 unless ($self->{svm}) {
2116 die "SVM UUID not cached, and reading remotely failed\n";
2117 }
2118 $self->{svm}->{uuid};
2119}
2120sub svm {
2122 my ($self) = @_;
2123 return $self->{svm} if $self->{svm};
2124 my $svm;
2125 # see if we have it in our config, first:
2126 eval {
2127 my $section = "svn-remote.$self->{repo_id}";
2128 $svm = {
2129 source => tmp_config('--get', "$section.svm-source"),
2130 uuid => tmp_config('--get', "$section.svm-uuid"),
2131 replace => tmp_config('--get', "$section.svm-replace"),
2132 }
2133 };
2134 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2135 $self->{svm} = $svm;
2136 }
2137 $self->{svm};
2138}
2139sub _set_svm_vars {
2141 my ($self, $ra) = @_;
2142 return $ra if $self->svm;
2143 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2145 "(svm:source, svm:uuid) ",
2146 "from the following URLs:\n" );
2147 sub read_svm_props {
2148 my ($self, $ra, $path, $r) = @_;
2149 my $props = ($ra->get_dir($path, $r))[2];
2150 my $src = $props->{'svm:source'};
2151 my $uuid = $props->{'svm:uuid'};
2152 return undef if (!$src || !$uuid);
2153 chomp($src, $uuid);
2155 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2157 or die "doesn't look right - svm:uuid is '$uuid'\n";
2158 # the '!' is used to mark the repos_root!/relative/path
2160 $src =~ s{/?!/?}{/};
2161 $src =~ s{/+$}{}; # no trailing slashes please
2162 # username is of no interest
2163 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2164 my $replace = $ra->{url};
2166 $replace .= "/$path" if length $path;
2167 my $section = "svn-remote.$self->{repo_id}";
2169 tmp_config("$section.svm-source", $src);
2170 tmp_config("$section.svm-replace", $replace);
2171 tmp_config("$section.svm-uuid", $uuid);
2172 $self->{svm} = {
2173 source => $src,
2174 uuid => $uuid,
2175 replace => $replace
2176 };
2177 }
2178 my $r = $ra->get_latest_revnum;
2180 my $path = $self->{path};
2181 my %tried;
2182 while (length $path) {
2183 unless ($tried{"$self->{url}/$path"}) {
2184 return $ra if $self->read_svm_props($ra, $path, $r);
2185 $tried{"$self->{url}/$path"} = 1;
2186 }
2187 $path =~ s#/?[^/]+$##;
2188 }
2189 die "Path: '$path' should be ''\n" if $path ne '';
2190 return $ra if $self->read_svm_props($ra, $path, $r);
2191 $tried{"$self->{url}/$path"} = 1;
2192 if ($ra->{repos_root} eq $self->{url}) {
2194 die @err, (map { " $_\n" } keys %tried), "\n";
2195 }
2196 # nope, make sure we're connected to the repository root:
2198 my $ok;
2199 my @tried_b;
2200 $path = $ra->{svn_path};
2201 $ra = Git::SVN::Ra->new($ra->{repos_root});
2202 while (length $path) {
2203 unless ($tried{"$ra->{url}/$path"}) {
2204 $ok = $self->read_svm_props($ra, $path, $r);
2205 last if $ok;
2206 $tried{"$ra->{url}/$path"} = 1;
2207 }
2208 $path =~ s#/?[^/]+$##;
2209 }
2210 die "Path: '$path' should be ''\n" if $path ne '';
2211 $ok ||= $self->read_svm_props($ra, $path, $r);
2212 $tried{"$ra->{url}/$path"} = 1;
2213 if (!$ok) {
2214 die @err, (map { " $_\n" } keys %tried), "\n";
2215 }
2216 Git::SVN::Ra->new($self->{url});
2217}
2218sub svnsync {
2220 my ($self) = @_;
2221 return $self->{svnsync} if $self->{svnsync};
2222 if ($self->no_metadata) {
2224 die "Can't have both 'noMetadata' and ",
2225 "'useSvnsyncProps' options set!\n";
2226 }
2227 if ($self->rewrite_root) {
2228 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2229 "options set!\n";
2230 }
2231 if ($self->rewrite_uuid) {
2232 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2233 "options set!\n";
2234 }
2235 my $svnsync;
2237 # see if we have it in our config, first:
2238 eval {
2239 my $section = "svn-remote.$self->{repo_id}";
2240 my $url = tmp_config('--get', "$section.svnsync-url");
2242 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2243 die "doesn't look right - svn:sync-from-url is '$url'\n";
2244 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2246 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2247 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2248 $svnsync = { url => $url, uuid => $uuid }
2250 };
2251 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2252 return $self->{svnsync} = $svnsync;
2253 }
2254 my $err = "useSvnsyncProps set, but failed to read " .
2256 "svnsync property: svn:sync-from-";
2257 my $rp = $self->ra->rev_proplist(0);
2258 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2260 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2261 die "doesn't look right - svn:sync-from-url is '$url'\n";
2262 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2264 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2265 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2266 my $section = "svn-remote.$self->{repo_id}";
2268 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2269 tmp_config('--add', "$section.svnsync-url", $url);
2270 return $self->{svnsync} = { url => $url, uuid => $uuid };
2271}
2272# this allows us to memoize our SVN::Ra UUID locally and avoid a
2274# remote lookup (useful for 'git svn log').
2275sub ra_uuid {
2276 my ($self) = @_;
2277 unless ($self->{ra_uuid}) {
2278 my $key = "svn-remote.$self->{repo_id}.uuid";
2279 my $uuid = eval { tmp_config('--get', $key) };
2280 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2281 $self->{ra_uuid} = $uuid;
2282 } else {
2283 die "ra_uuid called without URL\n" unless $self->{url};
2284 $self->{ra_uuid} = $self->ra->get_uuid;
2285 tmp_config('--add', $key, $self->{ra_uuid});
2286 }
2287 }
2288 $self->{ra_uuid};
2289}
2290sub _set_repos_root {
2292 my ($self, $repos_root) = @_;
2293 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2294 $repos_root ||= $self->ra->{repos_root};
2295 tmp_config($k, $repos_root);
2296 $repos_root;
2297}
2298sub repos_root {
2300 my ($self) = @_;
2301 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2302 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2303}
2304sub ra {
2306 my ($self) = shift;
2307 my $ra = Git::SVN::Ra->new($self->{url});
2308 $self->_set_repos_root($ra->{repos_root});
2309 if ($self->use_svm_props && !$self->{svm}) {
2310 if ($self->no_metadata) {
2311 die "Can't have both 'noMetadata' and ",
2312 "'useSvmProps' options set!\n";
2313 } elsif ($self->use_svnsync_props) {
2314 die "Can't have both 'useSvnsyncProps' and ",
2315 "'useSvmProps' options set!\n";
2316 }
2317 $ra = $self->_set_svm_vars($ra);
2318 $self->{-want_revprops} = 1;
2319 }
2320 $ra;
2321}
2322# prop_walk(PATH, REV, SUB)
2324# -------------------------
2325# Recursively traverse PATH at revision REV and invoke SUB for each
2326# directory that contains a SVN property. SUB will be invoked as
2327# follows: &SUB(gs, path, props); where `gs' is this instance of
2328# Git::SVN, `path' the path to the directory where the properties
2329# `props' were found. The `path' will be relative to point of checkout,
2330# that is, if url://repo/trunk is the current Git branch, and that
2331# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2332# as `path' (note the trailing `/').
2333sub prop_walk {
2334 my ($self, $path, $rev, $sub) = @_;
2335 $path =~ s#^/##;
2337 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2338 $path =~ s#^/*#/#g;
2339 my $p = $path;
2340 # Strip the irrelevant part of the path.
2341 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2342 # Ensure the path is terminated by a `/'.
2343 $p =~ s#/*$#/#;
2344 # The properties contain all the internal SVN stuff nobody
2346 # (usually) cares about.
2347 my $interesting_props = 0;
2348 foreach (keys %{$props}) {
2349 # If it doesn't start with `svn:', it must be a
2350 # user-defined property.
2351 ++$interesting_props and next if $_ !~ /^svn:/;
2352 # FIXME: Fragile, if SVN adds new public properties,
2353 # this needs to be updated.
2354 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2355 |eol-style|mime-type
2356 |externals|needs-lock)$/x;
2357 }
2358 &$sub($self, $p, $props) if $interesting_props;
2359 foreach (sort keys %$dirent) {
2361 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2362 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2363 }
2364}
2365sub last_rev { ($_[0]->last_rev_commit)[0] }
2367sub last_commit { ($_[0]->last_rev_commit)[1] }
2368# returns the newest SVN revision number and newest commit SHA1
2370sub last_rev_commit {
2371 my ($self) = @_;
2372 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2373 return ($self->{last_rev}, $self->{last_commit});
2374 }
2375 my $c = ::verify_ref($self->refname.'^0');
2376 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2377 my $rev = (::cmt_metadata($c))[1];
2378 if (defined $rev) {
2379 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2380 return ($rev, $c);
2381 }
2382 }
2383 my $map_path = $self->map_path;
2384 unless (-e $map_path) {
2385 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2386 return (undef, undef);
2387 }
2388 my ($rev, $commit) = $self->rev_map_max(1);
2389 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2390 return ($rev, $commit);
2391}
2392sub get_fetch_range {
2394 my ($self, $min, $max) = @_;
2395 $max ||= $self->ra->get_latest_revnum;
2396 $min ||= $self->rev_map_max;
2397 (++$min, $max);
2398}
2399sub tmp_config {
2401 my (@args) = @_;
2402 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2403 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2404 if (! -f $config && -f $old_def_config) {
2405 rename $old_def_config, $config or
2406 die "Failed rename $old_def_config => $config: $!\n";
2407 }
2408 my $old_config = $ENV{GIT_CONFIG};
2409 $ENV{GIT_CONFIG} = $config;
2410 $@ = undef;
2411 my @ret = eval {
2412 unless (-f $config) {
2413 mkfile($config);
2414 open my $fh, '>', $config or
2415 die "Can't open $config: $!\n";
2416 print $fh "; This file is used internally by ",
2417 "git-svn\n" or die
2418 "Couldn't write to $config: $!\n";
2419 print $fh "; You should not have to edit it\n" or
2420 die "Couldn't write to $config: $!\n";
2421 close $fh or die "Couldn't close $config: $!\n";
2422 }
2423 command('config', @args);
2424 };
2425 my $err = $@;
2426 if (defined $old_config) {
2427 $ENV{GIT_CONFIG} = $old_config;
2428 } else {
2429 delete $ENV{GIT_CONFIG};
2430 }
2431 die $err if $err;
2432 wantarray ? @ret : $ret[0];
2433}
2434sub tmp_index_do {
2436 my ($self, $sub) = @_;
2437 my $old_index = $ENV{GIT_INDEX_FILE};
2438 $ENV{GIT_INDEX_FILE} = $self->{index};
2439 $@ = undef;
2440 my @ret = eval {
2441 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2442 mkpath([$dir]) unless -d $dir;
2443 &$sub;
2444 };
2445 my $err = $@;
2446 if (defined $old_index) {
2447 $ENV{GIT_INDEX_FILE} = $old_index;
2448 } else {
2449 delete $ENV{GIT_INDEX_FILE};
2450 }
2451 die $err if $err;
2452 wantarray ? @ret : $ret[0];
2453}
2454sub assert_index_clean {
2456 my ($self, $treeish) = @_;
2457 $self->tmp_index_do(sub {
2459 command_noisy('read-tree', $treeish) unless -e $self->{index};
2460 my $x = command_oneline('write-tree');
2461 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2462 /^tree ($::sha1)/mo);
2463 return if $y eq $x;
2464 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2466 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2467 command_noisy('read-tree', $treeish);
2468 $x = command_oneline('write-tree');
2469 if ($y ne $x) {
2470 ::fatal "trees ($treeish) $y != $x\n",
2471 "Something is seriously wrong...";
2472 }
2473 });
2474}
2475sub get_commit_parents {
2477 my ($self, $log_entry) = @_;
2478 my (%seen, @ret, @tmp);
2479 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2480 if (my $ip = $self->{inject_parents}) {
2481 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2482 push @tmp, $commit;
2483 }
2484 }
2485 if (my $cur = ::verify_ref($self->refname.'^0')) {
2486 push @tmp, $cur;
2487 }
2488 if (my $ipd = $self->{inject_parents_dcommit}) {
2489 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2490 push @tmp, @$commit;
2491 }
2492 }
2493 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2494 while (my $p = shift @tmp) {
2495 next if $seen{$p};
2496 $seen{$p} = 1;
2497 push @ret, $p;
2498 }
2499 @ret;
2500}
2501sub rewrite_root {
2503 my ($self) = @_;
2504 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2505 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2506 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2507 if ($rwr) {
2508 $rwr =~ s#/+$##;
2509 if ($rwr !~ m#^[a-z\+]+://#) {
2510 die "$rwr is not a valid URL (key: $k)\n";
2511 }
2512 }
2513 $self->{-rewrite_root} = $rwr;
2514}
2515sub rewrite_uuid {
2517 my ($self) = @_;
2518 return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2519 my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2520 my $rwid = eval { command_oneline(qw/config --get/, $k) };
2521 if ($rwid) {
2522 $rwid =~ s#/+$##;
2523 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2524 die "$rwid is not a valid UUID (key: $k)\n";
2525 }
2526 }
2527 $self->{-rewrite_uuid} = $rwid;
2528}
2529sub metadata_url {
2531 my ($self) = @_;
2532 ($self->rewrite_root || $self->{url}) .
2533 (length $self->{path} ? '/' . $self->{path} : '');
2534}
2535sub full_url {
2537 my ($self) = @_;
2538 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2539}
2540sub set_commit_header_env {
2543 my ($log_entry) = @_;
2544 my %env;
2545 foreach my $ned (qw/NAME EMAIL DATE/) {
2546 foreach my $ac (qw/AUTHOR COMMITTER/) {
2547 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2548 }
2549 }
2550 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2552 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2553 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2554 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2556 ? $log_entry->{commit_name}
2557 : $log_entry->{name};
2558 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2559 ? $log_entry->{commit_email}
2560 : $log_entry->{email};
2561 \%env;
2562}
2563sub restore_commit_header_env {
2565 my ($env) = @_;
2566 foreach my $ned (qw/NAME EMAIL DATE/) {
2567 foreach my $ac (qw/AUTHOR COMMITTER/) {
2568 my $k = "GIT_${ac}_${ned}";
2569 if (defined $env->{$k}) {
2570 $ENV{$k} = $env->{$k};
2571 } else {
2572 delete $ENV{$k};
2573 }
2574 }
2575 }
2576}
2577sub gc {
2579 command_noisy('gc', '--auto');
2580};
2581sub do_git_commit {
2583 my ($self, $log_entry) = @_;
2584 my $lr = $self->last_rev;
2585 if (defined $lr && $lr >= $log_entry->{revision}) {
2586 die "Last fetched revision of ", $self->refname,
2587 " was r$lr, but we are about to fetch: ",
2588 "r$log_entry->{revision}!\n";
2589 }
2590 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2591 croak "$log_entry->{revision} = $c already exists! ",
2592 "Why are we refetching it?\n";
2593 }
2594 my $old_env = set_commit_header_env($log_entry);
2595 my $tree = $log_entry->{tree};
2596 if (!defined $tree) {
2597 $tree = $self->tmp_index_do(sub {
2598 command_oneline('write-tree') });
2599 }
2600 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2601 my @exec = ('git', 'commit-tree', $tree);
2603 foreach ($self->get_commit_parents($log_entry)) {
2604 push @exec, '-p', $_;
2605 }
2606 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2607 or croak $!;
2608 binmode $msg_fh;
2609 # we always get UTF-8 from SVN, but we may want our commits in
2611 # a different encoding.
2612 if (my $enc = Git::config('i18n.commitencoding')) {
2613 require Encode;
2614 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2615 }
2616 print $msg_fh $log_entry->{log} or croak $!;
2617 restore_commit_header_env($old_env);
2618 unless ($self->no_metadata) {
2619 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2620 or croak $!;
2621 }
2622 $msg_fh->flush == 0 or croak $!;
2623 close $msg_fh or croak $!;
2624 chomp(my $commit = do { local $/; <$out_fh> });
2625 close $out_fh or croak $!;
2626 waitpid $pid, 0;
2627 croak $? if $?;
2628 if ($commit !~ /^$::sha1$/o) {
2629 die "Failed to commit, invalid sha1: $commit\n";
2630 }
2631 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2633 $self->{last_rev} = $log_entry->{revision};
2635 $self->{last_commit} = $commit;
2636 print "r$log_entry->{revision}" unless $::_q > 1;
2637 if (defined $log_entry->{svm_revision}) {
2638 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2639 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2640 0, $self->svm_uuid);
2641 }
2642 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2643 if (--$_gc_nr == 0) {
2644 $_gc_nr = $_gc_period;
2645 gc();
2646 }
2647 return $commit;
2648}
2649sub match_paths {
2651 my ($self, $paths, $r) = @_;
2652 return 1 if $self->{path} eq '';
2653 if (my $path = $paths->{"/$self->{path}"}) {
2654 return ($path->{action} eq 'D') ? 0 : 1;
2655 }
2656 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2657 if (grep /$self->{path_regex}/, keys %$paths) {
2658 return 1;
2659 }
2660 my $c = '';
2661 foreach (split m#/#, $self->{path}) {
2662 $c .= "/$_";
2663 next unless ($paths->{$c} &&
2664 ($paths->{$c}->{action} =~ /^[AR]$/));
2665 if ($self->ra->check_path($self->{path}, $r) ==
2666 $SVN::Node::dir) {
2667 return 1;
2668 }
2669 }
2670 return 0;
2671}
2672sub find_parent_branch {
2674 my ($self, $paths, $rev) = @_;
2675 return undef unless $self->follow_parent;
2676 unless (defined $paths) {
2677 my $err_handler = $SVN::Error::handler;
2678 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2679 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2680 sub { $paths = $_[0] });
2681 $SVN::Error::handler = $err_handler;
2682 }
2683 return undef unless defined $paths;
2684 # look for a parent from another branch:
2686 my @b_path_components = split m#/#, $self->{path};
2687 my @a_path_components;
2688 my $i;
2689 while (@b_path_components) {
2690 $i = $paths->{'/'.join('/', @b_path_components)};
2691 last if $i && defined $i->{copyfrom_path};
2692 unshift(@a_path_components, pop(@b_path_components));
2693 }
2694 return undef unless defined $i && defined $i->{copyfrom_path};
2695 my $branch_from = $i->{copyfrom_path};
2696 if (@a_path_components) {
2697 print STDERR "branch_from: $branch_from => ";
2698 $branch_from .= '/'.join('/', @a_path_components);
2699 print STDERR $branch_from, "\n";
2700 }
2701 my $r = $i->{copyfrom_rev};
2702 my $repos_root = $self->ra->{repos_root};
2703 my $url = $self->ra->{url};
2704 my $new_url = $url . $branch_from;
2705 print STDERR "Found possible branch point: ",
2706 "$new_url => ", $self->full_url, ", $r\n"
2707 unless $::_q > 1;
2708 $branch_from =~ s#^/##;
2709 my $gs = $self->other_gs($new_url, $url,
2710 $branch_from, $r, $self->{ref_id});
2711 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2712 {
2713 my ($base, $head);
2714 if (!defined $r0 || !defined $parent) {
2715 ($base, $head) = parse_revision_argument(0, $r);
2716 } else {
2717 if ($r0 < $r) {
2718 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2719 0, 1, sub { $base = $_[1] - 1 });
2720 }
2721 }
2722 if (defined $base && $base <= $r) {
2723 $gs->fetch($base, $r);
2724 }
2725 ($r0, $parent) = $gs->find_rev_before($r, 1);
2726 }
2727 if (defined $r0 && defined $parent) {
2728 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2729 unless $::_q > 1;
2730 my $ed;
2731 if ($self->ra->can_do_switch) {
2732 $self->assert_index_clean($parent);
2733 print STDERR "Following parent with do_switch\n"
2734 unless $::_q > 1;
2735 # do_switch works with svn/trunk >= r22312, but that
2736 # is not included with SVN 1.4.3 (the latest version
2737 # at the moment), so we can't rely on it
2738 $self->{last_rev} = $r0;
2739 $self->{last_commit} = $parent;
2740 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2741 $gs->ra->gs_do_switch($r0, $rev, $gs,
2742 $self->full_url, $ed)
2743 or die "SVN connection failed somewhere...\n";
2744 } elsif ($self->ra->trees_match($new_url, $r0,
2745 $self->full_url, $rev)) {
2746 print STDERR "Trees match:\n",
2747 " $new_url\@$r0\n",
2748 " ${\$self->full_url}\@$rev\n",
2749 "Following parent with no changes\n"
2750 unless $::_q > 1;
2751 $self->tmp_index_do(sub {
2752 command_noisy('read-tree', $parent);
2753 });
2754 $self->{last_commit} = $parent;
2755 } else {
2756 print STDERR "Following parent with do_update\n"
2757 unless $::_q > 1;
2758 $ed = SVN::Git::Fetcher->new($self);
2759 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2760 or die "SVN connection failed somewhere...\n";
2761 }
2762 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2763 return $self->make_log_entry($rev, [$parent], $ed);
2764 }
2765 return undef;
2766}
2767sub do_fetch {
2769 my ($self, $paths, $rev) = @_;
2770 my $ed;
2771 my ($last_rev, @parents);
2772 if (my $lc = $self->last_commit) {
2773 # we can have a branch that was deleted, then re-added
2774 # under the same name but copied from another path, in
2775 # which case we'll have multiple parents (we don't
2776 # want to break the original ref, nor lose copypath info):
2777 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2778 push @{$log_entry->{parents}}, $lc;
2779 return $log_entry;
2780 }
2781 $ed = SVN::Git::Fetcher->new($self);
2782 $last_rev = $self->{last_rev};
2783 $ed->{c} = $lc;
2784 @parents = ($lc);
2785 } else {
2786 $last_rev = $rev;
2787 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2788 return $log_entry;
2789 }
2790 $ed = SVN::Git::Fetcher->new($self);
2791 }
2792 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2793 die "SVN connection failed somewhere...\n";
2794 }
2795 $self->make_log_entry($rev, \@parents, $ed);
2796}
2797sub mkemptydirs {
2799 my ($self, $r) = @_;
2800 sub scan {
2802 my ($r, $empty_dirs, $line) = @_;
2803 if (defined $r && $line =~ /^r(\d+)$/) {
2804 return 0 if $1 > $r;
2805 } elsif ($line =~ /^ \+empty_dir: (.+)$/) {
2806 $empty_dirs->{$1} = 1;
2807 } elsif ($line =~ /^ \-empty_dir: (.+)$/) {
2808 my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
2809 delete @$empty_dirs{@d};
2810 }
2811 1; # continue
2812 };
2813 my %empty_dirs = ();
2815 my $gz_file = "$self->{dir}/unhandled.log.gz";
2816 if (-f $gz_file) {
2817 if (!$can_compress) {
2818 warn "Compress::Zlib could not be found; ",
2819 "empty directories in $gz_file will not be read\n";
2820 } else {
2821 my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
2822 die "Unable to open $gz_file: $!\n";
2823 my $line;
2824 while ($gz->gzreadline($line) > 0) {
2825 scan($r, \%empty_dirs, $line) or last;
2826 }
2827 $gz->gzclose;
2828 }
2829 }
2830 if (open my $fh, '<', "$self->{dir}/unhandled.log") {
2832 binmode $fh or croak "binmode: $!";
2833 while (<$fh>) {
2834 scan($r, \%empty_dirs, $_) or last;
2835 }
2836 close $fh;
2837 }
2838 my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2840 foreach my $d (sort keys %empty_dirs) {
2841 $d = uri_decode($d);
2842 $d =~ s/$strip//;
2843 next unless length($d);
2844 next if -d $d;
2845 if (-e $d) {
2846 warn "$d exists but is not a directory\n";
2847 } else {
2848 print "creating empty directory: $d\n";
2849 mkpath([$d]);
2850 }
2851 }
2852}
2853sub get_untracked {
2855 my ($self, $ed) = @_;
2856 my @out;
2857 my $h = $ed->{empty};
2858 foreach (sort keys %$h) {
2859 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2860 push @out, " $act: " . uri_encode($_);
2861 warn "W: $act: $_\n";
2862 }
2863 foreach my $t (qw/dir_prop file_prop/) {
2864 $h = $ed->{$t} or next;
2865 foreach my $path (sort keys %$h) {
2866 my $ppath = $path eq '' ? '.' : $path;
2867 foreach my $prop (sort keys %{$h->{$path}}) {
2868 next if $SKIP_PROP{$prop};
2869 my $v = $h->{$path}->{$prop};
2870 my $t_ppath_prop = "$t: " .
2871 uri_encode($ppath) . ' ' .
2872 uri_encode($prop);
2873 if (defined $v) {
2874 push @out, " +$t_ppath_prop " .
2875 uri_encode($v);
2876 } else {
2877 push @out, " -$t_ppath_prop";
2878 }
2879 }
2880 }
2881 }
2882 foreach my $t (qw/absent_file absent_directory/) {
2883 $h = $ed->{$t} or next;
2884 foreach my $parent (sort keys %$h) {
2885 foreach my $path (sort @{$h->{$parent}}) {
2886 push @out, " $t: " .
2887 uri_encode("$parent/$path");
2888 warn "W: $t: $parent/$path ",
2889 "Insufficient permissions?\n";
2890 }
2891 }
2892 }
2893 \@out;
2894}
2895# parse_svn_date(DATE)
2897# --------------------
2898# Given a date (in UTC) from Subversion, return a string in the format
2899# "<TZ Offset> <local date/time>" that Git will use.
2900#
2901# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2902# is true we'll convert it to the local timezone instead.
2903sub parse_svn_date {
2904 my $date = shift || return '+0000 1970-01-01 00:00:00';
2905 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2906 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2907 croak "Unable to parse date: $date\n";
2908 my $parsed_date; # Set next.
2909 if ($Git::SVN::_localtime) {
2911 # Translate the Subversion datetime to an epoch time.
2912 # Begin by switching ourselves to $date's timezone, UTC.
2913 my $old_env_TZ = $ENV{TZ};
2914 $ENV{TZ} = 'UTC';
2915 my $epoch_in_UTC =
2917 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2918 # Determine our local timezone (including DST) at the
2920 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2921 # value of TZ, if any, at the time we were run.
2922 if (defined $Git::SVN::Log::TZ) {
2923 $ENV{TZ} = $Git::SVN::Log::TZ;
2924 } else {
2925 delete $ENV{TZ};
2926 }
2927 my $our_TZ =
2929 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2930 # This converts $epoch_in_UTC into our local timezone.
2932 my ($sec, $min, $hour, $mday, $mon, $year,
2933 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2934 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2936 $our_TZ, $year + 1900, $mon + 1,
2937 $mday, $hour, $min, $sec);
2938 # Reset us to the timezone in effect when we entered
2940 # this routine.
2941 if (defined $old_env_TZ) {
2942 $ENV{TZ} = $old_env_TZ;
2943 } else {
2944 delete $ENV{TZ};
2945 }
2946 } else {
2947 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2948 }
2949 return $parsed_date;
2951}
2952sub other_gs {
2954 my ($self, $new_url, $url,
2955 $branch_from, $r, $old_ref_id) = @_;
2956 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2957 unless ($gs) {
2958 my $ref_id = $old_ref_id;
2959 $ref_id =~ s/\@\d+$//;
2960 $ref_id .= "\@$r";
2961 # just grow a tail if we're not unique enough :x
2962 $ref_id .= '-' while find_ref($ref_id);
2963 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2964 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2965 if ($u =~ s#^\Q$url\E(/|$)##) {
2966 $p = $u;
2967 $u = $url;
2968 $repo_id = $self->{repo_id};
2969 }
2970 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2971 }
2972 $gs
2973}
2974sub call_authors_prog {
2976 my ($orig_author) = @_;
2977 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2978 my $author = `$::_authors_prog $orig_author`;
2979 if ($? != 0) {
2980 die "$::_authors_prog failed with exit code $?\n"
2981 }
2982 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2983 my ($name, $email) = ($1, $2);
2984 $email = undef if length $2 == 0;
2985 return [$name, $email];
2986 } else {
2987 die "Author: $orig_author: $::_authors_prog returned "
2988 . "invalid author format: $author\n";
2989 }
2990}
2991sub check_author {
2993 my ($author) = @_;
2994 if (!defined $author || length $author == 0) {
2995 $author = '(no author)';
2996 }
2997 if (!defined $::users{$author}) {
2998 if (defined $::_authors_prog) {
2999 $::users{$author} = call_authors_prog($author);
3000 } elsif (defined $::_authors) {
3001 die "Author: $author not defined in $::_authors file\n";
3002 }
3003 }
3004 $author;
3005}
3006sub find_extra_svk_parents {
3008 my ($self, $ed, $tickets, $parents) = @_;
3009 # aha! svk:merge property changed...
3010 my @tickets = split "\n", $tickets;
3011 my @known_parents;
3012 for my $ticket ( @tickets ) {
3013 my ($uuid, $path, $rev) = split /:/, $ticket;
3014 if ( $uuid eq $self->ra_uuid ) {
3015 my $url = $self->{url};
3016 my $repos_root = $url;
3017 my $branch_from = $path;
3018 $branch_from =~ s{^/}{};
3019 my $gs = $self->other_gs($repos_root."/".$branch_from,
3020 $url,
3021 $branch_from,
3022 $rev,
3023 $self->{ref_id});
3024 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3025 # wahey! we found it, but it might be
3026 # an old one (!)
3027 push @known_parents, [ $rev, $commit ];
3028 }
3029 }
3030 }
3031 # Ordering matters; highest-numbered commit merge tickets
3032 # first, as they may account for later merge ticket additions
3033 # or changes.
3034 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3035 for my $parent ( @known_parents ) {
3036 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3037 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3038 my $new;
3039 while ( <$msg_fh> ) {
3040 $new=1;last;
3041 }
3042 command_close_pipe($msg_fh, $ctx);
3043 if ( $new ) {
3044 print STDERR
3045 "Found merge parent (svk:merge ticket): $parent\n";
3046 push @$parents, $parent;
3047 }
3048 }
3049}
3050sub lookup_svn_merge {
3052 my $uuid = shift;
3053 my $url = shift;
3054 my $merge = shift;
3055 my ($source, $revs) = split ":", $merge;
3057 my $path = $source;
3058 $path =~ s{^/}{};
3059 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3060 if ( !$gs ) {
3061 warn "Couldn't find revmap for $url$source\n";
3062 return;
3063 }
3064 my @ranges = split ",", $revs;
3065 my ($tip, $tip_commit);
3066 my @merged_commit_ranges;
3067 # find the tip
3068 for my $range ( @ranges ) {
3069 my ($bottom, $top) = split "-", $range;
3070 $top ||= $bottom;
3071 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3072 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3073 unless ($top_commit and $bottom_commit) {
3075 warn "W:unknown path/rev in svn:mergeinfo "
3076 ."dirprop: $source:$range\n";
3077 next;
3078 }
3079 push @merged_commit_ranges,
3081 "$bottom_commit^..$top_commit";
3082 if ( !defined $tip or $top > $tip ) {
3084 $tip = $top;
3085 $tip_commit = $top_commit;
3086 }
3087 }
3088 return ($tip_commit, @merged_commit_ranges);
3089}
3090sub _rev_list {
3092 my ($msg_fh, $ctx) = command_output_pipe(
3093 "rev-list", @_,
3094 );
3095 my @rv;
3096 while ( <$msg_fh> ) {
3097 chomp;
3098 push @rv, $_;
3099 }
3100 command_close_pipe($msg_fh, $ctx);
3101 @rv;
3102}
3103sub check_cherry_pick {
3105 my $base = shift;
3106 my $tip = shift;
3107 my @ranges = @_;
3108 my %commits = map { $_ => 1 }
3109 _rev_list("--no-merges", $tip, "--not", $base);
3110 for my $range ( @ranges ) {
3111 delete @commits{_rev_list($range)};
3112 }
3113 for my $commit (keys %commits) {
3114 if (has_no_changes($commit)) {
3115 delete $commits{$commit};
3116 }
3117 }
3118 return (keys %commits);
3119}
3120sub has_no_changes {
3122 my $commit = shift;
3123 my @revs = split / /, command_oneline(
3125 qw(rev-list --parents -1 -m), $commit);
3126 # Commits with no parents, e.g. the start of a partial branch,
3128 # have changes by definition.
3129 return 1 if (@revs < 2);
3130 # Commits with multiple parents, e.g a merge, have no changes
3132 # by definition.
3133 return 0 if (@revs > 2);
3134 return (command_oneline("rev-parse", "$commit^{tree}") eq
3136 command_oneline("rev-parse", "$commit~1^{tree}"));
3137}
3138# The GIT_DIR environment variable is not always set until after the command
3140# line arguments are processed, so we can't memoize in a BEGIN block.
3141{
3142 my $memoized = 0;
3143 sub memoize_svn_mergeinfo_functions {
3145 return if $memoized;
3146 $memoized = 1;
3147 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3149 mkpath([$cache_path]) unless -d $cache_path;
3150 tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3152 "$cache_path/lookup_svn_merge.db", 'nstore';
3153 memoize 'lookup_svn_merge',
3154 SCALAR_CACHE => 'FAULT',
3155 LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3156 ;
3157 tie my %check_cherry_pick_cache => 'Memoize::Storable',
3159 "$cache_path/check_cherry_pick.db", 'nstore';
3160 memoize 'check_cherry_pick',
3161 SCALAR_CACHE => 'FAULT',
3162 LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3163 ;
3164 tie my %has_no_changes_cache => 'Memoize::Storable',
3166 "$cache_path/has_no_changes.db", 'nstore';
3167 memoize 'has_no_changes',
3168 SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3169 LIST_CACHE => 'FAULT',
3170 ;
3171 }
3172}
3173sub parents_exclude {
3175 my $parents = shift;
3176 my @commits = @_;
3177 return unless @commits;
3178 my @excluded;
3180 my $excluded;
3181 do {
3182 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3183 $excluded = command_oneline(@cmd);
3184 if ( $excluded ) {
3185 my @new;
3186 my $found;
3187 for my $commit ( @commits ) {
3188 if ( $commit eq $excluded ) {
3189 push @excluded, $commit;
3190 $found++;
3191 last;
3192 }
3193 else {
3194 push @new, $commit;
3195 }
3196 }
3197 die "saw commit '$excluded' in rev-list output, "
3198 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3199 unless $found;
3200 @commits = @new;
3201 }
3202 }
3203 while ($excluded and @commits);
3204 return @excluded;
3206}
3207# note: this function should only be called if the various dirprops
3210# have actually changed
3211sub find_extra_svn_parents {
3212 my ($self, $ed, $mergeinfo, $parents) = @_;
3213 # aha! svk:merge property changed...
3214 memoize_svn_mergeinfo_functions();
3216 # We first search for merged tips which are not in our
3218 # history. Then, we figure out which git revisions are in
3219 # that tip, but not this revision. If all of those revisions
3220 # are now marked as merge, we can add the tip as a parent.
3221 my @merges = split "\n", $mergeinfo;
3222 my @merge_tips;
3223 my $url = $self->{url};
3224 my $uuid = $self->ra_uuid;
3225 my %ranges;
3226 for my $merge ( @merges ) {
3227 my ($tip_commit, @ranges) =
3228 lookup_svn_merge( $uuid, $url, $merge );
3229 unless (!$tip_commit or
3230 grep { $_ eq $tip_commit } @$parents ) {
3231 push @merge_tips, $tip_commit;
3232 $ranges{$tip_commit} = \@ranges;
3233 } else {
3234 push @merge_tips, undef;
3235 }
3236 }
3237 my %excluded = map { $_ => 1 }
3239 parents_exclude($parents, grep { defined } @merge_tips);
3240 # check merge tips for new parents
3242 my @new_parents;
3243 for my $merge_tip ( @merge_tips ) {
3244 my $spec = shift @merges;
3245 next unless $merge_tip and $excluded{$merge_tip};
3246 my $ranges = $ranges{$merge_tip};
3248 # check out 'new' tips
3250 my $merge_base;
3251 eval {
3252 $merge_base = command_oneline(
3253 "merge-base",
3254 @$parents, $merge_tip,
3255 );
3256 };
3257 if ($@) {
3258 die "An error occurred during merge-base"
3259 unless $@->isa("Git::Error::Command");
3260 warn "W: Cannot find common ancestor between ".
3262 "@$parents and $merge_tip. Ignoring merge info.\n";
3263 next;
3264 }
3265 # double check that there are no missing non-merge commits
3267 my (@incomplete) = check_cherry_pick(
3268 $merge_base, $merge_tip,
3269 @$ranges,
3270 );
3271 if ( @incomplete ) {
3273 warn "W:svn cherry-pick ignored ($spec) - missing "
3274 .@incomplete." commit(s) (eg $incomplete[0])\n";
3275 } else {
3276 warn
3277 "Found merge parent (svn:mergeinfo prop): ",
3278 $merge_tip, "\n";
3279 push @new_parents, $merge_tip;
3280 }
3281 }
3282 # cater for merges which merge commits from multiple branches
3284 if ( @new_parents > 1 ) {
3285 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3286 for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3287 next if $i == $j;
3288 next unless $new_parents[$i];
3289 next unless $new_parents[$j];
3290 my $revs = command_oneline(
3291 "rev-list", "-1",
3292 "$new_parents[$i]..$new_parents[$j]",
3293 );
3294 if ( !$revs ) {
3295 undef($new_parents[$j]);
3296 }
3297 }
3298 }
3299 }
3300 push @$parents, grep { defined } @new_parents;
3301}
3302sub make_log_entry {
3304 my ($self, $rev, $parents, $ed) = @_;
3305 my $untracked = $self->get_untracked($ed);
3306 my @parents = @$parents;
3308 my $ps = $ed->{path_strip} || "";
3309 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3310 my $props = $ed->{dir_prop}{$path};
3311 if ( $props->{"svk:merge"} ) {
3312 $self->find_extra_svk_parents
3313 ($ed, $props->{"svk:merge"}, \@parents);
3314 }
3315 if ( $props->{"svn:mergeinfo"} ) {
3316 $self->find_extra_svn_parents
3317 ($ed,
3318 $props->{"svn:mergeinfo"},
3319 \@parents);
3320 }
3321 }
3322 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3324 print $un "r$rev\n" or croak $!;
3325 print $un $_, "\n" foreach @$untracked;
3326 my %log_entry = ( parents => \@parents, revision => $rev,
3327 log => '');
3328 my $headrev;
3330 my $logged = delete $self->{logged_rev_props};
3331 if (!$logged || $self->{-want_revprops}) {
3332 my $rp = $self->ra->rev_proplist($rev);
3333 foreach (sort keys %$rp) {
3334 my $v = $rp->{$_};
3335 if (/^svn:(author|date|log)$/) {
3336 $log_entry{$1} = $v;
3337 } elsif ($_ eq 'svm:headrev') {
3338 $headrev = $v;
3339 } else {
3340 print $un " rev_prop: ", uri_encode($_), ' ',
3341 uri_encode($v), "\n";
3342 }
3343 }
3344 } else {
3345 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3346 }
3347 close $un or croak $!;
3348 $log_entry{date} = parse_svn_date($log_entry{date});
3350 $log_entry{log} .= "\n";
3351 my $author = $log_entry{author} = check_author($log_entry{author});
3352 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3353 : ($author, undef);
3354 my ($commit_name, $commit_email) = ($name, $email);
3356 if ($_use_log_author) {
3357 my $name_field;
3358 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3359 $name_field = $1;
3360 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3361 $name_field = $1;
3362 }
3363 if (!defined $name_field) {
3364 if (!defined $email) {
3365 $email = $name;
3366 }
3367 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3368 ($name, $email) = ($1, $2);
3369 } elsif ($name_field =~ /(.*)@/) {
3370 ($name, $email) = ($1, $name_field);
3371 } else {
3372 ($name, $email) = ($name_field, $name_field);
3373 }
3374 }
3375 if (defined $headrev && $self->use_svm_props) {
3376 if ($self->rewrite_root) {
3377 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3378 "options set!\n";
3379 }
3380 if ($self->rewrite_uuid) {
3381 die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3382 "options set!\n";
3383 }
3384 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3385 # we don't want "SVM: initializing mirror for junk" ...
3386 return undef if $r == 0;
3387 my $svm = $self->svm;
3388 if ($uuid ne $svm->{uuid}) {
3389 die "UUID mismatch on SVM path:\n",
3390 "expected: $svm->{uuid}\n",
3391 " got: $uuid\n";
3392 }
3393 my $full_url = $self->full_url;
3394 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3395 die "Failed to replace '$svm->{replace}' with ",
3396 "'$svm->{source}' in $full_url\n";
3397 # throw away username for storing in records
3398 remove_username($full_url);
3399 $log_entry{metadata} = "$full_url\@$r $uuid";
3400 $log_entry{svm_revision} = $r;
3401 $email ||= "$author\@$uuid";
3402 $commit_email ||= "$author\@$uuid";
3403 } elsif ($self->use_svnsync_props) {
3404 my $full_url = $self->svnsync->{url};
3405 $full_url .= "/$self->{path}" if length $self->{path};
3406 remove_username($full_url);
3407 my $uuid = $self->svnsync->{uuid};
3408 $log_entry{metadata} = "$full_url\@$rev $uuid";
3409 $email ||= "$author\@$uuid";
3410 $commit_email ||= "$author\@$uuid";
3411 } else {
3412 my $url = $self->metadata_url;
3413 remove_username($url);
3414 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3415 $log_entry{metadata} = "$url\@$rev " . $uuid;
3416 $email ||= "$author\@" . $uuid;
3417 $commit_email ||= "$author\@" . $uuid;
3418 }
3419 $log_entry{name} = $name;
3420 $log_entry{email} = $email;
3421 $log_entry{commit_name} = $commit_name;
3422 $log_entry{commit_email} = $commit_email;
3423 \%log_entry;
3424}
3425sub fetch {
3427 my ($self, $min_rev, $max_rev, @parents) = @_;
3428 my ($last_rev, $last_commit) = $self->last_rev_commit;
3429 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3430 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3431}
3432sub set_tree_cb {
3434 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3435 $self->{inject_parents} = { $rev => $tree };
3436 $self->fetch(undef, undef);
3437}
3438sub set_tree {
3440 my ($self, $tree) = (shift, shift);
3441 my $log_entry = ::get_commit_entry($tree);
3442 unless ($self->{last_rev}) {
3443 ::fatal("Must have an existing revision to commit");
3444 }
3445 my %ed_opts = ( r => $self->{last_rev},
3446 log => $log_entry->{log},
3447 ra => $self->ra,
3448 tree_a => $self->{last_commit},
3449 tree_b => $tree,
3450 editor_cb => sub {
3451 $self->set_tree_cb($log_entry, $tree, @_) },
3452 svn_path => $self->{path} );
3453 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3454 print "No changes\nr$self->{last_rev} = $tree\n";
3455 }
3456}
3457sub rebuild_from_rev_db {
3459 my ($self, $path) = @_;
3460 my $r = -1;
3461 open my $fh, '<', $path or croak "open: $!";
3462 binmode $fh or croak "binmode: $!";
3463 while (<$fh>) {
3464 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3465 chomp($_);
3466 ++$r;
3467 next if $_ eq ('0' x 40);
3468 $self->rev_map_set($r, $_);
3469 print "r$r = $_\n";
3470 }
3471 close $fh or croak "close: $!";
3472 unlink $path or croak "unlink: $!";
3473}
3474sub rebuild {
3476 my ($self) = @_;
3477 my $map_path = $self->map_path;
3478 my $partial = (-e $map_path && ! -z $map_path);
3479 return unless ::verify_ref($self->refname.'^0');
3480 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3481 my $rev_db = $self->rev_db_path;
3482 $self->rebuild_from_rev_db($rev_db);
3483 if ($self->use_svm_props) {
3484 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3485 $self->rebuild_from_rev_db($svm_rev_db);
3486 }
3487 $self->unlink_rev_db_symlink;
3488 return;
3489 }
3490 print "Rebuilding $map_path ...\n" if (!$partial);
3491 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3492 (undef, undef));
3493 my ($log, $ctx) =
3494 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3495 ($head ? "$head.." : "") . $self->refname,
3496 '--');
3497 my $metadata_url = $self->metadata_url;
3498 remove_username($metadata_url);
3499 my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3500 my $c;
3501 while (<$log>) {
3502 if ( m{^commit ($::sha1)$} ) {
3503 $c = $1;
3504 next;
3505 }
3506 next unless s{^\s*(git-svn-id:)}{$1};
3507 my ($url, $rev, $uuid) = ::extract_metadata($_);
3508 remove_username($url);
3509 # ignore merges (from set-tree)
3511 next if (!defined $rev || !$uuid);
3512 # if we merged or otherwise started elsewhere, this is
3514 # how we break out of it
3515 if (($uuid ne $svn_uuid) ||
3516 ($metadata_url && $url && ($url ne $metadata_url))) {
3517 next;
3518 }
3519 if ($partial && $head) {
3520 print "Partial-rebuilding $map_path ...\n";
3521 print "Currently at $base_rev = $head\n";
3522 $head = undef;
3523 }
3524 $self->rev_map_set($rev, $c);
3526 print "r$rev = $c\n";
3527 }
3528 command_close_pipe($log, $ctx);
3529 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3530 my $rev_db_path = $self->rev_db_path;
3531 if (-f $self->rev_db_path) {
3532 unlink $self->rev_db_path or croak "unlink: $!";
3533 }
3534 $self->unlink_rev_db_symlink;
3535}
3536# rev_map:
3538# Tie::File seems to be prone to offset errors if revisions get sparse,
3539# it's not that fast, either. Tie::File is also not in Perl 5.6. So
3540# one of my favorite modules is out :< Next up would be one of the DBM
3541# modules, but I'm not sure which is most portable...
3542#
3543# This is the replacement for the rev_db format, which was too big
3544# and inefficient for large repositories with a lot of sparse history
3545# (mainly tags)
3546#
3547# The format is this:
3548# - 24 bytes for every record,
3549# * 4 bytes for the integer representing an SVN revision number
3550# * 20 bytes representing the sha1 of a git commit
3551# - No empty padding records like the old format
3552# (except the last record, which can be overwritten)
3553# - new records are written append-only since SVN revision numbers
3554# increase monotonically
3555# - lookups on SVN revision number are done via a binary search
3556# - Piping the file to xxd -c24 is a good way of dumping it for
3557# viewing or editing (piped back through xxd -r), should the need
3558# ever arise.
3559# - The last record can be padding revision with an all-zero sha1
3560# This is used to optimize fetch performance when using multiple
3561# "fetch" directives in .git/config
3562#
3563# These files are disposable unless noMetadata or useSvmProps is set
3564sub _rev_map_set {
3566 my ($fh, $rev, $commit) = @_;
3567 binmode $fh or croak "binmode: $!";
3569 my $size = (stat($fh))[7];
3570 ($size % 24) == 0 or croak "inconsistent size: $size";
3571 my $wr_offset = 0;
3573 if ($size > 0) {
3574 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3575 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3576 $read == 24 or croak "read only $read bytes (!= 24)";
3577 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3578 if ($last_commit eq ('0' x40)) {
3579 if ($size >= 48) {
3580 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3581 $read = sysread($fh, $buf, 24) or
3582 croak "read: $!";
3583 $read == 24 or
3584 croak "read only $read bytes (!= 24)";
3585 ($last_rev, $last_commit) =
3586 unpack(rev_map_fmt, $buf);
3587 if ($last_commit eq ('0' x40)) {
3588 croak "inconsistent .rev_map\n";
3589 }
3590 }
3591 if ($last_rev >= $rev) {
3592 croak "last_rev is higher!: $last_rev >= $rev";
3593 }
3594 $wr_offset = -24;
3595 }
3596 }
3597 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3598 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3599 croak "write: $!";
3600}
3601sub _rev_map_reset {
3603 my ($fh, $rev, $commit) = @_;
3604 my $c = _rev_map_get($fh, $rev);
3605 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3606 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3607 truncate $fh, $offset or croak "truncate: $!";
3608}
3609sub mkfile {
3611 my ($path) = @_;
3612 unless (-e $path) {
3613 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3614 mkpath([$dir]) unless -d $dir;
3615 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3616 close $fh or die "Couldn't close (create) $path: $!\n";
3617 }
3618}
3619sub rev_map_set {
3621 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3622 defined $commit or die "missing arg3\n";
3623 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3624 my $db = $self->map_path($uuid);
3625 my $db_lock = "$db.lock";
3626 my $sig;
3627 $update_ref ||= 0;
3628 if ($update_ref) {
3629 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3630 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3631 }
3632 mkfile($db);
3633 $LOCKFILES{$db_lock} = 1;
3635 my $sync;
3636 # both of these options make our .rev_db file very, very important
3637 # and we can't afford to lose it because rebuild() won't work
3638 if ($self->use_svm_props || $self->no_metadata) {
3639 $sync = 1;
3640 copy($db, $db_lock) or die "rev_map_set(@_): ",
3641 "Failed to copy: ",
3642 "$db => $db_lock ($!)\n";
3643 } else {
3644 rename $db, $db_lock or die "rev_map_set(@_): ",
3645 "Failed to rename: ",
3646 "$db => $db_lock ($!)\n";
3647 }
3648 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3650 or croak "Couldn't open $db_lock: $!\n";
3651 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3652 _rev_map_set($fh, $rev, $commit);
3653 if ($sync) {
3654 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3655 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3656 }
3657 close $fh or croak $!;
3658 if ($update_ref) {
3659 $_head = $self;
3660 my $note = "";
3661 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3662 command_noisy('update-ref', '-m', "r$rev$note",
3663 $self->refname, $commit);
3664 }
3665 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3666 "$db_lock => $db ($!)\n";
3667 delete $LOCKFILES{$db_lock};
3668 if ($update_ref) {
3669 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3670 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3671 kill $sig, $$ if defined $sig;
3672 }
3673}
3674# If want_commit, this will return an array of (rev, commit) where
3676# commit _must_ be a valid commit in the archive.
3677# Otherwise, it'll return the max revision (whether or not the
3678# commit is valid or just a 0x40 placeholder).
3679sub rev_map_max {
3680 my ($self, $want_commit) = @_;
3681 $self->rebuild;
3682 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3683 $want_commit ? ($r, $c) : $r;
3684}
3685sub rev_map_max_norebuild {
3687 my ($self, $want_commit) = @_;
3688 my $map_path = $self->map_path;
3689 stat $map_path or return $want_commit ? (0, undef) : 0;
3690 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3691 binmode $fh or croak "binmode: $!";
3692 my $size = (stat($fh))[7];
3693 ($size % 24) == 0 or croak "inconsistent size: $size";
3694 if ($size == 0) {
3696 close $fh or croak "close: $!";
3697 return $want_commit ? (0, undef) : 0;
3698 }
3699 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3701 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3702 my ($r, $c) = unpack(rev_map_fmt, $buf);
3703 if ($want_commit && $c eq ('0' x40)) {
3704 if ($size < 48) {
3705 return $want_commit ? (0, undef) : 0;
3706 }
3707 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3708 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3709 ($r, $c) = unpack(rev_map_fmt, $buf);
3710 if ($c eq ('0'x40)) {
3711 croak "Penultimate record is all-zeroes in $map_path";
3712 }
3713 }
3714 close $fh or croak "close: $!";
3715 $want_commit ? ($r, $c) : $r;
3716}
3717sub rev_map_get {
3719 my ($self, $rev, $uuid) = @_;
3720 my $map_path = $self->map_path($uuid);
3721 return undef unless -e $map_path;
3722 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3724 my $c = _rev_map_get($fh, $rev);
3725 close($fh) or croak "close: $!";
3726 $c
3727}
3728sub _rev_map_get {
3730 my ($fh, $rev) = @_;
3731 binmode $fh or croak "binmode: $!";
3733 my $size = (stat($fh))[7];
3734 ($size % 24) == 0 or croak "inconsistent size: $size";
3735 if ($size == 0) {
3737 return undef;
3738 }
3739 my ($l, $u) = (0, $size - 24);
3741 my ($r, $c, $buf);
3742 while ($l <= $u) {
3744 my $i = int(($l/24 + $u/24) / 2) * 24;
3745 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3746 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3747 my ($r, $c) = unpack(rev_map_fmt, $buf);
3748 if ($r < $rev) {
3750 $l = $i + 24;
3751 } elsif ($r > $rev) {
3752 $u = $i - 24;
3753 } else { # $r == $rev
3754 return $c eq ('0' x 40) ? undef : $c;
3755 }
3756 }
3757 undef;
3758}
3759# Finds the first svn revision that exists on (if $eq_ok is true) or
3761# before $rev for the current branch. It will not search any lower
3762# than $min_rev. Returns the git commit hash and svn revision number
3763# if found, else (undef, undef).
3764sub find_rev_before {
3765 my ($self, $rev, $eq_ok, $min_rev) = @_;
3766 --$rev unless $eq_ok;
3767 $min_rev ||= 1;
3768 my $max_rev = $self->rev_map_max;
3769 $rev = $max_rev if ($rev > $max_rev);
3770 while ($rev >= $min_rev) {
3771 if (my $c = $self->rev_map_get($rev)) {
3772 return ($rev, $c);
3773 }
3774 --$rev;
3775 }
3776 return (undef, undef);
3777}
3778# Finds the first svn revision that exists on (if $eq_ok is true) or
3780# after $rev for the current branch. It will not search any higher
3781# than $max_rev. Returns the git commit hash and svn revision number
3782# if found, else (undef, undef).
3783sub find_rev_after {
3784 my ($self, $rev, $eq_ok, $max_rev) = @_;
3785 ++$rev unless $eq_ok;
3786 $max_rev ||= $self->rev_map_max;
3787 while ($rev <= $max_rev) {
3788 if (my $c = $self->rev_map_get($rev)) {
3789 return ($rev, $c);
3790 }
3791 ++$rev;
3792 }
3793 return (undef, undef);
3794}
3795sub _new {
3797 my ($class, $repo_id, $ref_id, $path) = @_;
3798 unless (defined $repo_id && length $repo_id) {
3799 $repo_id = $Git::SVN::default_repo_id;
3800 }
3801 unless (defined $ref_id && length $ref_id) {
3802 $_prefix = '' unless defined($_prefix);
3803 $_[2] = $ref_id =
3804 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3805 }
3806 $_[1] = $repo_id;
3807 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3808 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3810 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3811 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3812 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3813 if (-d $old_dir && ! -d $dir) {
3814 $dir = $old_dir;
3815 }
3816 }
3817 $_[3] = $path = '' unless (defined $path);
3819 mkpath([$dir]);
3820 bless {
3821 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3822 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3823 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3824}
3825# for read-only access of old .rev_db formats
3827sub unlink_rev_db_symlink {
3828 my ($self) = @_;
3829 my $link = $self->rev_db_path;
3830 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3831 if (-l $link) {
3832 unlink $link or croak "unlink: $link failed!";
3833 }
3834}
3835sub rev_db_path {
3837 my ($self, $uuid) = @_;
3838 my $db_path = $self->map_path($uuid);
3839 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3840 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3841 $db_path;
3842}
3843# the new replacement for .rev_db
3845sub map_path {
3846 my ($self, $uuid) = @_;
3847 $uuid ||= $self->ra_uuid;
3848 "$self->{map_root}.$uuid";
3849}
3850sub uri_encode {
3852 my ($f) = @_;
3853 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3854 $f
3855}
3856sub uri_decode {
3858 my ($f) = @_;
3859 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3860 $f
3861}
3862sub remove_username {
3864 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3865}
3866package Git::SVN::Prompt;
3868use strict;
3869use warnings;
3870require SVN::Core;
3871use vars qw/$_no_auth_cache $_username/;
3872sub simple {
3874 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3875 $may_save = undef if $_no_auth_cache;
3876 $default_username = $_username if defined $_username;
3877 if (defined $default_username && length $default_username) {
3878 if (defined $realm && length $realm) {
3879 print STDERR "Authentication realm: $realm\n";
3880 STDERR->flush;
3881 }
3882 $cred->username($default_username);
3883 } else {
3884 username($cred, $realm, $may_save, $pool);
3885 }
3886 $cred->password(_read_password("Password for '" .
3887 $cred->username . "': ", $realm));
3888 $cred->may_save($may_save);
3889 $SVN::_Core::SVN_NO_ERROR;
3890}
3891sub ssl_server_trust {
3893 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3894 $may_save = undef if $_no_auth_cache;
3895 print STDERR "Error validating server certificate for '$realm':\n";
3896 {
3897 no warnings 'once';
3898 # All variables SVN::Auth::SSL::* are used only once,
3899 # so we're shutting up Perl warnings about this.
3900 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3901 print STDERR " - The certificate is not issued ",
3902 "by a trusted authority. Use the\n",
3903 " fingerprint to validate ",
3904 "the certificate manually!\n";
3905 }
3906 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3907 print STDERR " - The certificate hostname ",
3908 "does not match.\n";
3909 }
3910 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3911 print STDERR " - The certificate is not yet valid.\n";
3912 }
3913 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3914 print STDERR " - The certificate has expired.\n";
3915 }
3916 if ($failures & $SVN::Auth::SSL::OTHER) {
3917 print STDERR " - The certificate has ",
3918 "an unknown error.\n";
3919 }
3920 } # no warnings 'once'
3921 printf STDERR
3922 "Certificate information:\n".
3923 " - Hostname: %s\n".
3924 " - Valid: from %s until %s\n".
3925 " - Issuer: %s\n".
3926 " - Fingerprint: %s\n",
3927 map $cert_info->$_, qw(hostname valid_from valid_until
3928 issuer_dname fingerprint);
3929 my $choice;
3930prompt:
3931 print STDERR $may_save ?
3932 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3933 "(R)eject or accept (t)emporarily? ";
3934 STDERR->flush;
3935 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3936 if ($choice =~ /^t$/i) {
3937 $cred->may_save(undef);
3938 } elsif ($choice =~ /^r$/i) {
3939 return -1;
3940 } elsif ($may_save && $choice =~ /^p$/i) {
3941 $cred->may_save($may_save);
3942 } else {
3943 goto prompt;
3944 }
3945 $cred->accepted_failures($failures);
3946 $SVN::_Core::SVN_NO_ERROR;
3947}
3948sub ssl_client_cert {
3950 my ($cred, $realm, $may_save, $pool) = @_;
3951 $may_save = undef if $_no_auth_cache;
3952 print STDERR "Client certificate filename: ";
3953 STDERR->flush;
3954 chomp(my $filename = <STDIN>);
3955 $cred->cert_file($filename);
3956 $cred->may_save($may_save);
3957 $SVN::_Core::SVN_NO_ERROR;
3958}
3959sub ssl_client_cert_pw {
3961 my ($cred, $realm, $may_save, $pool) = @_;
3962 $may_save = undef if $_no_auth_cache;
3963 $cred->password(_read_password("Password: ", $realm));
3964 $cred->may_save($may_save);
3965 $SVN::_Core::SVN_NO_ERROR;
3966}
3967sub username {
3969 my ($cred, $realm, $may_save, $pool) = @_;
3970 $may_save = undef if $_no_auth_cache;
3971 if (defined $realm && length $realm) {
3972 print STDERR "Authentication realm: $realm\n";
3973 }
3974 my $username;
3975 if (defined $_username) {
3976 $username = $_username;
3977 } else {
3978 print STDERR "Username: ";
3979 STDERR->flush;
3980 chomp($username = <STDIN>);
3981 }
3982 $cred->username($username);
3983 $cred->may_save($may_save);
3984 $SVN::_Core::SVN_NO_ERROR;
3985}
3986sub _read_password {
3988 my ($prompt, $realm) = @_;
3989 my $password = '';
3990 if (exists $ENV{GIT_ASKPASS}) {
3991 open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
3992 $password = <PH>;
3993 $password =~ s/[\012\015]//; # \n\r
3994 close(PH);
3995 } else {
3996 print STDERR $prompt;
3997 STDERR->flush;
3998 require Term::ReadKey;
3999 Term::ReadKey::ReadMode('noecho');
4000 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
4001 last if $key =~ /[\012\015]/; # \n\r
4002 $password .= $key;
4003 }
4004 Term::ReadKey::ReadMode('restore');
4005 print STDERR "\n";
4006 STDERR->flush;
4007 }
4008 $password;
4009}
4010package SVN::Git::Fetcher;
4012use vars qw/@ISA/;
4013use strict;
4014use warnings;
4015use Carp qw/croak/;
4016use IO::File qw//;
4017use vars qw/$_ignore_regex/;
4018# file baton members: path, mode_a, mode_b, pool, fh, blob, base
4020sub new {
4021 my ($class, $git_svn, $switch_path) = @_;
4022 my $self = SVN::Delta::Editor->new;
4023 bless $self, $class;
4024 if (exists $git_svn->{last_commit}) {
4025 $self->{c} = $git_svn->{last_commit};
4026 $self->{empty_symlinks} =
4027 _mark_empty_symlinks($git_svn, $switch_path);
4028 }
4029 $self->{ignore_regex} = eval { command_oneline('config', '--get',
4030 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
4031 $self->{empty} = {};
4032 $self->{dir_prop} = {};
4033 $self->{file_prop} = {};
4034 $self->{absent_dir} = {};
4035 $self->{absent_file} = {};
4036 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4037 $self;
4038}
4039# this uses the Ra object, so it must be called before do_{switch,update},
4041# not inside them (when the Git::SVN::Fetcher object is passed) to
4042# do_{switch,update}
4043sub _mark_empty_symlinks {
4044 my ($git_svn, $switch_path) = @_;
4045 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4046 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4047 my %ret;
4049 my ($rev, $cmt) = $git_svn->last_rev_commit;
4050 return {} unless ($rev && $cmt);
4051 # allow the warning to be printed for each revision we fetch to
4053 # ensure the user sees it. The user can also disable the workaround
4054 # on the repository even while git svn is running and the next
4055 # revision fetched will skip this expensive function.
4056 my $printed_warning;
4057 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4058 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4059 local $/ = "\0";
4060 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4061 $pfx .= '/' if length($pfx);
4062 while (<$ls>) {
4063 chomp;
4064 s/\A100644 blob $empty_blob\t//o or next;
4065 unless ($printed_warning) {
4066 print STDERR "Scanning for empty symlinks, ",
4067 "this may take a while if you have ",
4068 "many empty files\n",
4069 "You may disable this with `",
4070 "git config svn.brokenSymlinkWorkaround ",
4071 "false'.\n",
4072 "This may be done in a different ",
4073 "terminal without restarting ",
4074 "git svn\n";
4075 $printed_warning = 1;
4076 }
4077 my $path = $_;
4078 my (undef, $props) =
4079 $git_svn->ra->get_file($pfx.$path, $rev, undef);
4080 if ($props->{'svn:special'}) {
4081 $ret{$path} = 1;
4082 }
4083 }
4084 command_close_pipe($ls, $ctx);
4085 \%ret;
4086}
4087# returns true if a given path is inside a ".git" directory
4089sub in_dot_git {
4090 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4091}
4092# return value: 0 -- don't ignore, 1 -- ignore
4094sub is_path_ignored {
4095 my ($self, $path) = @_;
4096 return 1 if in_dot_git($path);
4097 return 1 if defined($self->{ignore_regex}) &&
4098 $path =~ m!$self->{ignore_regex}!;
4099 return 0 unless defined($_ignore_regex);
4100 return 1 if $path =~ m!$_ignore_regex!o;
4101 return 0;
4102}
4103sub set_path_strip {
4105 my ($self, $path) = @_;
4106 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4107}
4108sub open_root {
4110 { path => '' };
4111}
4112sub open_directory {
4114 my ($self, $path, $pb, $rev) = @_;
4115 { path => $path };
4116}
4117sub git_path {
4119 my ($self, $path) = @_;
4120 if ($self->{path_strip}) {
4121 $path =~ s!$self->{path_strip}!! or
4122 die "Failed to strip path '$path' ($self->{path_strip})\n";
4123 }
4124 $path;
4125}
4126sub delete_entry {
4128 my ($self, $path, $rev, $pb) = @_;
4129 return undef if $self->is_path_ignored($path);
4130 my $gpath = $self->git_path($path);
4132 return undef if ($gpath eq '');
4133 # remove entire directories.
4135 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4136 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4137 if ($tree) {
4138 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4139 -r --name-only -z/,
4140 $tree);
4141 local $/ = "\0";
4142 while (<$ls>) {
4143 chomp;
4144 my $rmpath = "$gpath/$_";
4145 $self->{gii}->remove($rmpath);
4146 print "\tD\t$rmpath\n" unless $::_q;
4147 }
4148 print "\tD\t$gpath/\n" unless $::_q;
4149 command_close_pipe($ls, $ctx);
4150 } else {
4151 $self->{gii}->remove($gpath);
4152 print "\tD\t$gpath\n" unless $::_q;
4153 }
4154 $self->{empty}->{$path} = 0;
4155 undef;
4156}
4157sub open_file {
4159 my ($self, $path, $pb, $rev) = @_;
4160 my ($mode, $blob);
4161 goto out if $self->is_path_ignored($path);
4163 my $gpath = $self->git_path($path);
4165 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4166 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4167 unless (defined $mode && defined $blob) {
4168 die "$path was not found in commit $self->{c} (r$rev)\n";
4169 }
4170 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4171 $mode = '120000';
4172 }
4173out:
4174 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4175 pool => SVN::Pool->new, action => 'M' };
4176}
4177sub add_file {
4179 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4180 my $mode;
4181 if (!$self->is_path_ignored($path)) {
4183 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4184 delete $self->{empty}->{$dir};
4185 $mode = '100644';
4186 }
4187 { path => $path, mode_a => $mode, mode_b => $mode,
4188 pool => SVN::Pool->new, action => 'A' };
4189}
4190sub add_directory {
4192 my ($self, $path, $cp_path, $cp_rev) = @_;
4193 goto out if $self->is_path_ignored($path);
4194 my $gpath = $self->git_path($path);
4195 if ($gpath eq '') {
4196 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4197 -r --name-only -z/,
4198 $self->{c});
4199 local $/ = "\0";
4200 while (<$ls>) {
4201 chomp;
4202 $self->{gii}->remove($_);
4203 print "\tD\t$_\n" unless $::_q;
4204 }
4205 command_close_pipe($ls, $ctx);
4206 $self->{empty}->{$path} = 0;
4207 }
4208 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4209 delete $self->{empty}->{$dir};
4210 $self->{empty}->{$path} = 1;
4211out:
4212 { path => $path };
4213}
4214sub change_dir_prop {
4216 my ($self, $db, $prop, $value) = @_;
4217 return undef if $self->is_path_ignored($db->{path});
4218 $self->{dir_prop}->{$db->{path}} ||= {};
4219 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4220 undef;
4221}
4222sub absent_directory {
4224 my ($self, $path, $pb) = @_;
4225 return undef if $self->is_path_ignored($path);
4226 $self->{absent_dir}->{$pb->{path}} ||= [];
4227 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4228 undef;
4229}
4230sub absent_file {
4232 my ($self, $path, $pb) = @_;
4233 return undef if $self->is_path_ignored($path);
4234 $self->{absent_file}->{$pb->{path}} ||= [];
4235 push @{$self->{absent_file}->{$pb->{path}}}, $path;
4236 undef;
4237}
4238sub change_file_prop {
4240 my ($self, $fb, $prop, $value) = @_;
4241 return undef if $self->is_path_ignored($fb->{path});
4242 if ($prop eq 'svn:executable') {
4243 if ($fb->{mode_b} != 120000) {
4244 $fb->{mode_b} = defined $value ? 100755 : 100644;
4245 }
4246 } elsif ($prop eq 'svn:special') {
4247 $fb->{mode_b} = defined $value ? 120000 : 100644;
4248 } else {
4249 $self->{file_prop}->{$fb->{path}} ||= {};
4250 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4251 }
4252 undef;
4253}
4254sub apply_textdelta {
4256 my ($self, $fb, $exp) = @_;
4257 return undef if $self->is_path_ignored($fb->{path});
4258 my $fh = $::_repository->temp_acquire('svn_delta');
4259 # $fh gets auto-closed() by SVN::TxDelta::apply(),
4260 # (but $base does not,) so dup() it for reading in close_file
4261 open my $dup, '<&', $fh or croak $!;
4262 my $base = $::_repository->temp_acquire('git_blob');
4263 if ($fb->{blob}) {
4265 my ($base_is_link, $size);
4266 if ($fb->{mode_a} eq '120000' &&
4268 ! $self->{empty_symlinks}->{$fb->{path}}) {
4269 print $base 'link ' or die "print $!\n";
4270 $base_is_link = 1;
4271 }
4272 retry:
4273 $size = $::_repository->cat_blob($fb->{blob}, $base);
4274 die "Failed to read object $fb->{blob}" if ($size < 0);
4275 if (defined $exp) {
4277 seek $base, 0, 0 or croak $!;
4278 my $got = ::md5sum($base);
4279 if ($got ne $exp) {
4280 my $err = "Checksum mismatch: ".
4281 "$fb->{path} $fb->{blob}\n" .
4282 "expected: $exp\n" .
4283 " got: $got\n";
4284 if ($base_is_link) {
4285 warn $err,
4286 "Retrying... (possibly ",
4287 "a bad symlink from SVN)\n";
4288 $::_repository->temp_reset($base);
4289 $base_is_link = 0;
4290 goto retry;
4291 }
4292 die $err;
4293 }
4294 }
4295 }
4296 seek $base, 0, 0 or croak $!;
4297 $fb->{fh} = $fh;
4298 $fb->{base} = $base;
4299 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4300}
4301sub close_file {
4303 my ($self, $fb, $exp) = @_;
4304 return undef if $self->is_path_ignored($fb->{path});
4305 my $hash;
4307 my $path = $self->git_path($fb->{path});
4308 if (my $fh = $fb->{fh}) {
4309 if (defined $exp) {
4310 seek($fh, 0, 0) or croak $!;
4311 my $got = ::md5sum($fh);
4312 if ($got ne $exp) {
4313 die "Checksum mismatch: $path\n",
4314 "expected: $exp\n got: $got\n";
4315 }
4316 }
4317 if ($fb->{mode_b} == 120000) {
4318 sysseek($fh, 0, 0) or croak $!;
4319 my $rd = sysread($fh, my $buf, 5);
4320 if (!defined $rd) {
4322 croak "sysread: $!\n";
4323 } elsif ($rd == 0) {
4324 warn "$path has mode 120000",
4325 " but it points to nothing\n",
4326 "converting to an empty file with mode",
4327 " 100644\n";
4328 $fb->{mode_b} = '100644';
4329 } elsif ($buf ne 'link ') {
4330 warn "$path has mode 120000",
4331 " but is not a link\n";
4332 } else {
4333 my $tmp_fh = $::_repository->temp_acquire(
4334 'svn_hash');
4335 my $res;
4336 while ($res = sysread($fh, my $str, 1024)) {
4337 my $out = syswrite($tmp_fh, $str, $res);
4338 defined($out) && $out == $res
4339 or croak("write ",
4340 Git::temp_path($tmp_fh),
4341 ": $!\n");
4342 }
4343 defined $res or croak $!;
4344 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4346 Git::temp_release($tmp_fh, 1);
4347 }
4348 }
4349 $hash = $::_repository->hash_and_insert_object(
4351 Git::temp_path($fh));
4352 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4353 Git::temp_release($fb->{base}, 1);
4355 Git::temp_release($fh, 1);
4356 } else {
4357 $hash = $fb->{blob} or die "no blob information\n";
4358 }
4359 $fb->{pool}->clear;
4360 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4361 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4362 undef;
4363}
4364sub abort_edit {
4366 my $self = shift;
4367 $self->{nr} = $self->{gii}->{nr};
4368 delete $self->{gii};
4369 $self->SUPER::abort_edit(@_);
4370}
4371sub close_edit {
4373 my $self = shift;
4374 $self->{git_commit_ok} = 1;
4375 $self->{nr} = $self->{gii}->{nr};
4376 delete $self->{gii};
4377 $self->SUPER::close_edit(@_);
4378}
4379package SVN::Git::Editor;
4381use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4382use strict;
4383use warnings;
4384use Carp qw/croak/;
4385use IO::File;
4386sub new {
4388 my ($class, $opts) = @_;
4389 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4390 die "$_ required!\n" unless (defined $opts->{$_});
4391 }
4392 my $pool = SVN::Pool->new;
4394 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4395 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4396 $opts->{r}, $mods);
4397 # $opts->{ra} functions should not be used after this:
4399 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4400 $opts->{editor_cb}, $pool);
4401 my $self = SVN::Delta::Editor->new(@ce, $pool);
4402 bless $self, $class;
4403 foreach (qw/svn_path r tree_a tree_b/) {
4404 $self->{$_} = $opts->{$_};
4405 }
4406 $self->{url} = $opts->{ra}->{url};
4407 $self->{mods} = $mods;
4408 $self->{types} = $types;
4409 $self->{pool} = $pool;
4410 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4411 $self->{rm} = { };
4412 $self->{path_prefix} = length $self->{svn_path} ?
4413 "$self->{svn_path}/" : '';
4414 $self->{config} = $opts->{config};
4415 return $self;
4416}
4417sub generate_diff {
4419 my ($tree_a, $tree_b) = @_;
4420 my @diff_tree = qw(diff-tree -z -r);
4421 if ($_cp_similarity) {
4422 push @diff_tree, "-C$_cp_similarity";
4423 } else {
4424 push @diff_tree, '-C';
4425 }
4426 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4427 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4428 push @diff_tree, $tree_a, $tree_b;
4429 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4430 local $/ = "\0";
4431 my $state = 'meta';
4432 my @mods;
4433 while (<$diff_fh>) {
4434 chomp $_; # this gets rid of the trailing "\0"
4435 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4436 ($::sha1)\s($::sha1)\s
4437 ([MTCRAD])\d*$/xo) {
4438 push @mods, { mode_a => $1, mode_b => $2,
4439 sha1_a => $3, sha1_b => $4,
4440 chg => $5 };
4441 if ($5 =~ /^(?:C|R)$/) {
4442 $state = 'file_a';
4443 } else {
4444 $state = 'file_b';
4445 }
4446 } elsif ($state eq 'file_a') {
4447 my $x = $mods[$#mods] or croak "Empty array\n";
4448 if ($x->{chg} !~ /^(?:C|R)$/) {
4449 croak "Error parsing $_, $x->{chg}\n";
4450 }
4451 $x->{file_a} = $_;
4452 $state = 'file_b';
4453 } elsif ($state eq 'file_b') {
4454 my $x = $mods[$#mods] or croak "Empty array\n";
4455 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4456 croak "Error parsing $_, $x->{chg}\n";
4457 }
4458 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4459 croak "Error parsing $_, $x->{chg}\n";
4460 }
4461 $x->{file_b} = $_;
4462 $state = 'meta';
4463 } else {
4464 croak "Error parsing $_\n";
4465 }
4466 }
4467 command_close_pipe($diff_fh, $ctx);
4468 \@mods;
4469}
4470sub check_diff_paths {
4472 my ($ra, $pfx, $rev, $mods) = @_;
4473 my %types;
4474 $pfx .= '/' if length $pfx;
4475 sub type_diff_paths {
4477 my ($ra, $types, $path, $rev) = @_;
4478 my @p = split m#/+#, $path;
4479 my $c = shift @p;
4480 unless (defined $types->{$c}) {
4481 $types->{$c} = $ra->check_path($c, $rev);
4482 }
4483 while (@p) {
4484 $c .= '/' . shift @p;
4485 next if defined $types->{$c};
4486 $types->{$c} = $ra->check_path($c, $rev);
4487 }
4488 }
4489 foreach my $m (@$mods) {
4491 foreach my $f (qw/file_a file_b/) {
4492 next unless defined $m->{$f};
4493 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4494 if (length $pfx.$dir && ! defined $types{$dir}) {
4495 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4496 }
4497 }
4498 }
4499 \%types;
4500}
4501sub split_path {
4503 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4504}
4505sub repo_path {
4507 my ($self, $path) = @_;
4508 $self->{path_prefix}.(defined $path ? $path : '');
4509}
4510sub url_path {
4512 my ($self, $path) = @_;
4513 if ($self->{url} =~ m#^https?://#) {
4514 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4515 }
4516 $self->{url} . '/' . $self->repo_path($path);
4517}
4518sub rmdirs {
4520 my ($self) = @_;
4521 my $rm = $self->{rm};
4522 delete $rm->{''}; # we never delete the url we're tracking
4523 return unless %$rm;
4524 foreach (keys %$rm) {
4526 my @d = split m#/#, $_;
4527 my $c = shift @d;
4528 $rm->{$c} = 1;
4529 while (@d) {
4530 $c .= '/' . shift @d;
4531 $rm->{$c} = 1;
4532 }
4533 }
4534 delete $rm->{$self->{svn_path}};
4535 delete $rm->{''}; # we never delete the url we're tracking
4536 return unless %$rm;
4537 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4539 $self->{tree_b});
4540 local $/ = "\0";
4541 while (<$fh>) {
4542 chomp;
4543 my @dn = split m#/#, $_;
4544 while (pop @dn) {
4545 delete $rm->{join '/', @dn};
4546 }
4547 unless (%$rm) {
4548 close $fh;
4549 return;
4550 }
4551 }
4552 command_close_pipe($fh, $ctx);
4553 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4555 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4556 $self->close_directory($bat->{$d}, $p);
4557 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4558 print "\tD+\t$d/\n" unless $::_q;
4559 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4560 delete $bat->{$d};
4561 }
4562}
4563sub open_or_add_dir {
4565 my ($self, $full_path, $baton) = @_;
4566 my $t = $self->{types}->{$full_path};
4567 if (!defined $t) {
4568 die "$full_path not known in r$self->{r} or we have a bug!\n";
4569 }
4570 {
4571 no warnings 'once';
4572 # SVN::Node::none and SVN::Node::file are used only once,
4573 # so we're shutting up Perl's warnings about them.
4574 if ($t == $SVN::Node::none) {
4575 return $self->add_directory($full_path, $baton,
4576 undef, -1, $self->{pool});
4577 } elsif ($t == $SVN::Node::dir) {
4578 return $self->open_directory($full_path, $baton,
4579 $self->{r}, $self->{pool});
4580 } # no warnings 'once'
4581 print STDERR "$full_path already exists in repository at ",
4582 "r$self->{r} and it is not a directory (",
4583 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4584 } # no warnings 'once'
4585 exit 1;
4586}
4587sub ensure_path {
4589 my ($self, $path) = @_;
4590 my $bat = $self->{bat};
4591 my $repo_path = $self->repo_path($path);
4592 return $bat->{''} unless (length $repo_path);
4593 my @p = split m#/+#, $repo_path;
4594 my $c = shift @p;
4595 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4596 while (@p) {
4597 my $c0 = $c;
4598 $c .= '/' . shift @p;
4599 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4600 }
4601 return $bat->{$c};
4602}
4603# Subroutine to convert a globbing pattern to a regular expression.
4605# From perl cookbook.
4606sub glob2pat {
4607 my $globstr = shift;
4608 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4609 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4610 return '^' . $globstr . '$';
4611}
4612sub check_autoprop {
4614 my ($self, $pattern, $properties, $file, $fbat) = @_;
4615 # Convert the globbing pattern to a regular expression.
4616 my $regex = glob2pat($pattern);
4617 # Check if the pattern matches the file name.
4618 if($file =~ m/($regex)/) {
4619 # Parse the list of properties to set.
4620 my @props = split(/;/, $properties);
4621 foreach my $prop (@props) {
4622 # Parse 'name=value' syntax and set the property.
4623 if ($prop =~ /([^=]+)=(.*)/) {
4624 my ($n,$v) = ($1,$2);
4625 for ($n, $v) {
4626 s/^\s+//; s/\s+$//;
4627 }
4628 $self->change_file_prop($fbat, $n, $v);
4629 }
4630 }
4631 }
4632}
4633sub apply_autoprops {
4635 my ($self, $file, $fbat) = @_;
4636 my $conf_t = ${$self->{config}}{'config'};
4637 no warnings 'once';
4638 # Check [miscellany]/enable-auto-props in svn configuration.
4639 if (SVN::_Core::svn_config_get_bool(
4640 $conf_t,
4641 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4642 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4643 0)) {
4644 # Auto-props are enabled. Enumerate them to look for matches.
4645 my $callback = sub {
4646 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4647 };
4648 SVN::_Core::svn_config_enumerate(
4649 $conf_t,
4650 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4651 $callback);
4652 }
4653}
4654sub A {
4656 my ($self, $m) = @_;
4657 my ($dir, $file) = split_path($m->{file_b});
4658 my $pbat = $self->ensure_path($dir);
4659 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4660 undef, -1);
4661 print "\tA\t$m->{file_b}\n" unless $::_q;
4662 $self->apply_autoprops($file, $fbat);
4663 $self->chg_file($fbat, $m);
4664 $self->close_file($fbat,undef,$self->{pool});
4665}
4666sub C {
4668 my ($self, $m) = @_;
4669 my ($dir, $file) = split_path($m->{file_b});
4670 my $pbat = $self->ensure_path($dir);
4671 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4672 $self->url_path($m->{file_a}), $self->{r});
4673 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4674 $self->chg_file($fbat, $m);
4675 $self->close_file($fbat,undef,$self->{pool});
4676}
4677sub delete_entry {
4679 my ($self, $path, $pbat) = @_;
4680 my $rpath = $self->repo_path($path);
4681 my ($dir, $file) = split_path($rpath);
4682 $self->{rm}->{$dir} = 1;
4683 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4684}
4685sub R {
4687 my ($self, $m) = @_;
4688 my ($dir, $file) = split_path($m->{file_b});
4689 my $pbat = $self->ensure_path($dir);
4690 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4691 $self->url_path($m->{file_a}), $self->{r});
4692 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4693 $self->apply_autoprops($file, $fbat);
4694 $self->chg_file($fbat, $m);
4695 $self->close_file($fbat,undef,$self->{pool});
4696 ($dir, $file) = split_path($m->{file_a});
4698 $pbat = $self->ensure_path($dir);
4699 $self->delete_entry($m->{file_a}, $pbat);
4700}
4701sub M {
4703 my ($self, $m) = @_;
4704 my ($dir, $file) = split_path($m->{file_b});
4705 my $pbat = $self->ensure_path($dir);
4706 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4707 $pbat,$self->{r},$self->{pool});
4708 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4709 $self->chg_file($fbat, $m);
4710 $self->close_file($fbat,undef,$self->{pool});
4711}
4712sub T { shift->M(@_) }
4714sub change_file_prop {
4716 my ($self, $fbat, $pname, $pval) = @_;
4717 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4718}
4719sub _chg_file_get_blob ($$$$) {
4721 my ($self, $fbat, $m, $which) = @_;
4722 my $fh = $::_repository->temp_acquire("git_blob_$which");
4723 if ($m->{"mode_$which"} =~ /^120/) {
4724 print $fh 'link ' or croak $!;
4725 $self->change_file_prop($fbat,'svn:special','*');
4726 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4727 $self->change_file_prop($fbat,'svn:special',undef);
4728 }
4729 my $blob = $m->{"sha1_$which"};
4730 return ($fh,) if ($blob =~ /^0{40}$/);
4731 my $size = $::_repository->cat_blob($blob, $fh);
4732 croak "Failed to read object $blob" if ($size < 0);
4733 $fh->flush == 0 or croak $!;
4734 seek $fh, 0, 0 or croak $!;
4735 my $exp = ::md5sum($fh);
4737 seek $fh, 0, 0 or croak $!;
4738 return ($fh, $exp);
4739}
4740sub chg_file {
4742 my ($self, $fbat, $m) = @_;
4743 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4744 $self->change_file_prop($fbat,'svn:executable','*');
4745 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4746 $self->change_file_prop($fbat,'svn:executable',undef);
4747 }
4748 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4749 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4750 my $pool = SVN::Pool->new;
4751 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4752 if (-s $fh_a) {
4753 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4754 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4755 if (defined $res) {
4756 die "Unexpected result from send_txstream: $res\n",
4757 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4758 }
4759 } else {
4760 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4761 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4762 if ($got ne $exp_b);
4763 }
4764 Git::temp_release($fh_b, 1);
4765 Git::temp_release($fh_a, 1);
4766 $pool->clear;
4767}
4768sub D {
4770 my ($self, $m) = @_;
4771 my ($dir, $file) = split_path($m->{file_b});
4772 my $pbat = $self->ensure_path($dir);
4773 print "\tD\t$m->{file_b}\n" unless $::_q;
4774 $self->delete_entry($m->{file_b}, $pbat);
4775}
4776sub close_edit {
4778 my ($self) = @_;
4779 my ($p,$bat) = ($self->{pool}, $self->{bat});
4780 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4781 next if $_ eq '';
4782 $self->close_directory($bat->{$_}, $p);
4783 }
4784 $self->close_directory($bat->{''}, $p);
4785 $self->SUPER::close_edit($p);
4786 $p->clear;
4787}
4788sub abort_edit {
4790 my ($self) = @_;
4791 $self->SUPER::abort_edit($self->{pool});
4792}
4793sub DESTROY {
4795 my $self = shift;
4796 $self->SUPER::DESTROY(@_);
4797 $self->{pool}->clear;
4798}
4799# this drives the editor
4801sub apply_diff {
4802 my ($self) = @_;
4803 my $mods = $self->{mods};
4804 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4805 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4806 my $f = $m->{chg};
4807 if (defined $o{$f}) {
4808 $self->$f($m);
4809 } else {
4810 fatal("Invalid change type: $f");
4811 }
4812 }
4813 $self->rmdirs if $_rmdir;
4814 if (@$mods == 0) {
4815 $self->abort_edit;
4816 } else {
4817 $self->close_edit;
4818 }
4819 return scalar @$mods;
4820}
4821package Git::SVN::Ra;
4823use vars qw/@ISA $config_dir $_log_window_size/;
4824use strict;
4825use warnings;
4826my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4827BEGIN {
4829 # enforce temporary pool usage for some simple functions
4830 no strict 'refs';
4831 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4832 get_file/) {
4833 my $SUPER = "SUPER::$f";
4834 *$f = sub {
4835 my $self = shift;
4836 my $pool = SVN::Pool->new;
4837 my @ret = $self->$SUPER(@_,$pool);
4838 $pool->clear;
4839 wantarray ? @ret : $ret[0];
4840 };
4841 }
4842}
4843sub _auth_providers () {
4845 [
4846 SVN::Client::get_simple_provider(),
4847 SVN::Client::get_ssl_server_trust_file_provider(),
4848 SVN::Client::get_simple_prompt_provider(
4849 \&Git::SVN::Prompt::simple, 2),
4850 SVN::Client::get_ssl_client_cert_file_provider(),
4851 SVN::Client::get_ssl_client_cert_prompt_provider(
4852 \&Git::SVN::Prompt::ssl_client_cert, 2),
4853 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4854 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4855 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4856 SVN::Client::get_username_provider(),
4857 SVN::Client::get_ssl_server_trust_prompt_provider(
4858 \&Git::SVN::Prompt::ssl_server_trust),
4859 SVN::Client::get_username_prompt_provider(
4860 \&Git::SVN::Prompt::username, 2)
4861 ]
4862}
4863sub escape_uri_only {
4865 my ($uri) = @_;
4866 my @tmp;
4867 foreach (split m{/}, $uri) {
4868 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4869 push @tmp, $_;
4870 }
4871 join('/', @tmp);
4872}
4873sub escape_url {
4875 my ($url) = @_;
4876 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4877 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4878 $url = "$scheme://$domain$uri";
4879 }
4880 $url;
4881}
4882sub new {
4884 my ($class, $url) = @_;
4885 $url =~ s!/+$!!;
4886 return $RA if ($RA && $RA->{url} eq $url);
4887 ::_req_svn();
4889 SVN::_Core::svn_config_ensure($config_dir, undef);
4891 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4892 my $config = SVN::Core::config_get_config($config_dir);
4893 $RA = undef;
4894 my $dont_store_passwords = 1;
4895 my $conf_t = ${$config}{'config'};
4896 {
4897 no warnings 'once';
4898 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4899 # produces warnings that variables are used only once.
4900 # I had not found the better way to shut them up, so
4901 # the warnings of type 'once' are disabled in this block.
4902 if (SVN::_Core::svn_config_get_bool($conf_t,
4903 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4904 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4905 1) == 0) {
4906 SVN::_Core::svn_auth_set_parameter($baton,
4907 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4908 bless (\$dont_store_passwords, "_p_void"));
4909 }
4910 if (SVN::_Core::svn_config_get_bool($conf_t,
4911 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4912 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4913 1) == 0) {
4914 $Git::SVN::Prompt::_no_auth_cache = 1;
4915 }
4916 } # no warnings 'once'
4917 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4918 config => $config,
4919 pool => SVN::Pool->new,
4920 auth_provider_callbacks => $callbacks);
4921 $self->{url} = $url;
4922 $self->{svn_path} = $url;
4923 $self->{repos_root} = $self->get_repos_root;
4924 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4925 $self->{cache} = { check_path => { r => 0, data => {} },
4926 get_dir => { r => 0, data => {} } };
4927 $RA = bless $self, $class;
4928}
4929sub check_path {
4931 my ($self, $path, $r) = @_;
4932 my $cache = $self->{cache}->{check_path};
4933 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4934 return $cache->{data}->{$path};
4935 }
4936 my $pool = SVN::Pool->new;
4937 my $t = $self->SUPER::check_path($path, $r, $pool);
4938 $pool->clear;
4939 if ($r != $cache->{r}) {
4940 %{$cache->{data}} = ();
4941 $cache->{r} = $r;
4942 }
4943 $cache->{data}->{$path} = $t;
4944}
4945sub get_dir {
4947 my ($self, $dir, $r) = @_;
4948 my $cache = $self->{cache}->{get_dir};
4949 if ($r == $cache->{r}) {
4950 if (my $x = $cache->{data}->{$dir}) {
4951 return wantarray ? @$x : $x->[0];
4952 }
4953 }
4954 my $pool = SVN::Pool->new;
4955 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4956 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4957 $pool->clear;
4958 if ($r != $cache->{r}) {
4959 %{$cache->{data}} = ();
4960 $cache->{r} = $r;
4961 }
4962 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4963 wantarray ? (\%dirents, $r, $props) : \%dirents;
4964}
4965sub DESTROY {
4967 # do not call the real DESTROY since we store ourselves in $RA
4968}
4969# get_log(paths, start, end, limit,
4971# discover_changed_paths, strict_node_history, receiver)
4972sub get_log {
4973 my ($self, @args) = @_;
4974 my $pool = SVN::Pool->new;
4975 # svn_log_changed_path_t objects passed to get_log are likely to be
4977 # overwritten even if only the refs are copied to an external variable,
4978 # so we should dup the structures in their entirety. Using an
4979 # externally passed pool (instead of our temporary and quickly cleared
4980 # pool in Git::SVN::Ra) does not help matters at all...
4981 my $receiver = pop @args;
4982 my $prefix = "/".$self->{svn_path};
4983 $prefix =~ s#/+($)##;
4984 my $prefix_regex = qr#^\Q$prefix\E#;
4985 push(@args, sub {
4986 my ($paths) = $_[0];
4987 return &$receiver(@_) unless $paths;
4988 $_[0] = ();
4989 foreach my $p (keys %$paths) {
4990 my $i = $paths->{$p};
4991 # Make path relative to our url, not repos_root
4992 $p =~ s/$prefix_regex//;
4993 my %s = map { $_ => $i->$_; }
4994 qw/copyfrom_path copyfrom_rev action/;
4995 if ($s{'copyfrom_path'}) {
4996 $s{'copyfrom_path'} =~ s/$prefix_regex//;
4997 }
4998 $_[0]{$p} = \%s;
4999 }
5000 &$receiver(@_);
5001 });
5002 # the limit parameter was not supported in SVN 1.1.x, so we
5005 # drop it. Therefore, the receiver callback passed to it
5006 # is made aware of this limitation by being wrapped if
5007 # the limit passed to is being wrapped.
5008 if ($SVN::Core::VERSION le '1.2.0') {
5009 my $limit = splice(@args, 3, 1);
5010 if ($limit > 0) {
5011 my $receiver = pop @args;
5012 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
5013 }
5014 }
5015 my $ret = $self->SUPER::get_log(@args, $pool);
5016 $pool->clear;
5017 $ret;
5018}
5019sub trees_match {
5021 my ($self, $url1, $rev1, $url2, $rev2) = @_;
5022 my $ctx = SVN::Client->new(auth => _auth_providers);
5023 my $out = IO::File->new_tmpfile;
5024 # older SVN (1.1.x) doesn't take $pool as the last parameter for
5026 # $ctx->diff(), so we'll create a default one
5027 my $pool = SVN::Pool->new_default_sub;
5028 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5030 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5031 $out->flush;
5032 my $ret = (($out->stat)[7] == 0);
5033 close $out or croak $!;
5034 $ret;
5036}
5037sub get_commit_editor {
5039 my ($self, $log, $cb, $pool) = @_;
5040 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
5041 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5042}
5043sub gs_do_update {
5045 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5046 my $new = ($rev_a == $rev_b);
5047 my $path = $gs->{path};
5048 if ($new && -e $gs->{index}) {
5050 unlink $gs->{index} or die
5051 "Couldn't unlink index: $gs->{index}: $!\n";
5052 }
5053 my $pool = SVN::Pool->new;
5054 $editor->set_path_strip($path);
5055 my (@pc) = split m#/#, $path;
5056 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5057 1, $editor, $pool);
5058 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5059 # Since we can't rely on svn_ra_reparent being available, we'll
5061 # just have to do some magic with set_path to make it so
5062 # we only want a partial path.
5063 my $sp = '';
5064 my $final = join('/', @pc);
5065 while (@pc) {
5066 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5067 $sp .= '/' if length $sp;
5068 $sp .= shift @pc;
5069 }
5070 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5071 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5073 $reporter->finish_report($pool);
5075 $pool->clear;
5076 $editor->{git_commit_ok};
5077}
5078# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5080# svn_ra_reparent didn't work before 1.4)
5081sub gs_do_switch {
5082 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5083 my $path = $gs->{path};
5084 my $pool = SVN::Pool->new;
5085 my $full_url = $self->{url};
5087 my $old_url = $full_url;
5088 $full_url .= '/' . $path if length $path;
5089 my ($ra, $reparented);
5090 if ($old_url =~ m#^svn(\+ssh)?://# ||
5092 ($full_url =~ m#^https?://# &&
5093 escape_url($full_url) ne $full_url)) {
5094 $_[0] = undef;
5095 $self = undef;
5096 $RA = undef;
5097 $ra = Git::SVN::Ra->new($full_url);
5098 $ra_invalid = 1;
5099 } elsif ($old_url ne $full_url) {
5100 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5101 $self->{url} = $full_url;
5102 $reparented = 1;
5103 }
5104 $ra ||= $self;
5106 $url_b = escape_url($url_b);
5107 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5108 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5109 $reporter->set_path('', $rev_a, 0, @lock, $pool);
5110 $reporter->finish_report($pool);
5111 if ($reparented) {
5113 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5114 $self->{url} = $old_url;
5115 }
5116 $pool->clear;
5118 $editor->{git_commit_ok};
5119}
5120sub longest_common_path {
5122 my ($gsv, $globs) = @_;
5123 my %common;
5124 my $common_max = scalar @$gsv;
5125 foreach my $gs (@$gsv) {
5127 my @tmp = split m#/#, $gs->{path};
5128 my $p = '';
5129 foreach (@tmp) {
5130 $p .= length($p) ? "/$_" : $_;
5131 $common{$p} ||= 0;
5132 $common{$p}++;
5133 }
5134 }
5135 $globs ||= [];
5136 $common_max += scalar @$globs;
5137 foreach my $glob (@$globs) {
5138 my @tmp = split m#/#, $glob->{path}->{left};
5139 my $p = '';
5140 foreach (@tmp) {
5141 $p .= length($p) ? "/$_" : $_;
5142 $common{$p} ||= 0;
5143 $common{$p}++;
5144 }
5145 }
5146 my $longest_path = '';
5148 foreach (sort {length $b <=> length $a} keys %common) {
5149 if ($common{$_} == $common_max) {
5150 $longest_path = $_;
5151 last;
5152 }
5153 }
5154 $longest_path;
5155}
5156sub gs_fetch_loop_common {
5158 my ($self, $base, $head, $gsv, $globs) = @_;
5159 return if ($base > $head);
5160 my $inc = $_log_window_size;
5161 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5162 my $longest_path = longest_common_path($gsv, $globs);
5163 my $ra_url = $self->{url};
5164 my $find_trailing_edge;
5165 while (1) {
5166 my %revs;
5167 my $err;
5168 my $err_handler = $SVN::Error::handler;
5169 $SVN::Error::handler = sub {
5170 ($err) = @_;
5171 skip_unknown_revs($err);
5172 };
5173 sub _cb {
5174 my ($paths, $r, $author, $date, $log) = @_;
5175 [ $paths,
5176 { author => $author, date => $date, log => $log } ];
5177 }
5178 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5179 sub { $revs{$_[1]} = _cb(@_) });
5180 if ($err) {
5181 print "Checked through r$max\r";
5182 } else {
5183 $find_trailing_edge = 1;
5184 }
5185 if ($err and $find_trailing_edge) {
5186 print STDERR "Path '$longest_path' ",
5187 "was probably deleted:\n",
5188 $err->expanded_message,
5189 "\nWill attempt to follow ",
5190 "revisions r$min .. r$max ",
5191 "committed before the deletion\n";
5192 my $hi = $max;
5193 while (--$hi >= $min) {
5194 my $ok;
5195 $self->get_log([$longest_path], $min, $hi,
5196 0, 1, 1, sub {
5197 $ok = $_[1];
5198 $revs{$_[1]} = _cb(@_) });
5199 if ($ok) {
5200 print STDERR "r$min .. r$ok OK\n";
5201 last;
5202 }
5203 }
5204 $find_trailing_edge = 0;
5205 }
5206 $SVN::Error::handler = $err_handler;
5207 my %exists = map { $_->{path} => $_ } @$gsv;
5209 foreach my $r (sort {$a <=> $b} keys %revs) {
5210 my ($paths, $logged) = @{$revs{$r}};
5211 foreach my $gs ($self->match_globs(\%exists, $paths,
5213 $globs, $r)) {
5214 if ($gs->rev_map_max >= $r) {
5215 next;
5216 }
5217 next unless $gs->match_paths($paths, $r);
5218 $gs->{logged_rev_props} = $logged;
5219 if (my $last_commit = $gs->last_commit) {
5220 $gs->assert_index_clean($last_commit);
5221 }
5222 my $log_entry = $gs->do_fetch($paths, $r);
5223 if ($log_entry) {
5224 $gs->do_git_commit($log_entry);
5225 }
5226 $INDEX_FILES{$gs->{index}} = 1;
5227 }
5228 foreach my $g (@$globs) {
5229 my $k = "svn-remote.$g->{remote}." .
5230 "$g->{t}-maxRev";
5231 Git::SVN::tmp_config($k, $r);
5232 }
5233 if ($ra_invalid) {
5234 $_[0] = undef;
5235 $self = undef;
5236 $RA = undef;
5237 $self = Git::SVN::Ra->new($ra_url);
5238 $ra_invalid = undef;
5239 }
5240 }
5241 # pre-fill the .rev_db since it'll eventually get filled in
5242 # with '0' x40 if something new gets committed
5243 foreach my $gs (@$gsv) {
5244 next if $gs->rev_map_max >= $max;
5245 next if defined $gs->rev_map_get($max);
5246 $gs->rev_map_set($max, 0 x40);
5247 }
5248 foreach my $g (@$globs) {
5249 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5250 Git::SVN::tmp_config($k, $max);
5251 }
5252 last if $max >= $head;
5253 $min = $max + 1;
5254 $max += $inc;
5255 $max = $head if ($max > $head);
5256 }
5257 Git::SVN::gc();
5258}
5259sub get_dir_globbed {
5261 my ($self, $left, $depth, $r) = @_;
5262 my @x = eval { $self->get_dir($left, $r) };
5264 return unless scalar @x == 3;
5265 my $dirents = $x[0];
5266 my @finalents;
5267 foreach my $de (keys %$dirents) {
5268 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5269 if ($depth > 1) {
5270 my @args = ("$left/$de", $depth - 1, $r);
5271 foreach my $dir ($self->get_dir_globbed(@args)) {
5272 push @finalents, "$de/$dir";
5273 }
5274 } else {
5275 push @finalents, $de;
5276 }
5277 }
5278 @finalents;
5279}
5280sub match_globs {
5282 my ($self, $exists, $paths, $globs, $r) = @_;
5283 sub get_dir_check {
5285 my ($self, $exists, $g, $r) = @_;
5286 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5288 $g->{path}->{depth},
5289 $r);
5290 foreach my $de (@dirs) {
5292 my $p = $g->{path}->full_path($de);
5293 next if $exists->{$p};
5294 next if (length $g->{path}->{right} &&
5295 ($self->check_path($p, $r) !=
5296 $SVN::Node::dir));
5297 next unless $p =~ /$g->{path}->{regex}/;
5298 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5299 $g->{ref}->full_path($de), 1);
5300 }
5301 }
5302 foreach my $g (@$globs) {
5303 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5304 if ($path->{action} =~ /^[AR]$/) {
5305 get_dir_check($self, $exists, $g, $r);
5306 }
5307 }
5308 foreach (keys %$paths) {
5309 if (/$g->{path}->{left_regex}/ &&
5310 !/$g->{path}->{regex}/) {
5311 next if $paths->{$_}->{action} !~ /^[AR]$/;
5312 get_dir_check($self, $exists, $g, $r);
5313 }
5314 next unless /$g->{path}->{regex}/;
5315 my $p = $1;
5316 my $pathname = $g->{path}->full_path($p);
5317 next if $exists->{$pathname};
5318 next if ($self->check_path($pathname, $r) !=
5319 $SVN::Node::dir);
5320 $exists->{$pathname} = Git::SVN->init(
5321 $self->{url}, $pathname, undef,
5322 $g->{ref}->full_path($p), 1);
5323 }
5324 my $c = '';
5325 foreach (split m#/#, $g->{path}->{left}) {
5326 $c .= "/$_";
5327 next unless ($paths->{$c} &&
5328 ($paths->{$c}->{action} =~ /^[AR]$/));
5329 get_dir_check($self, $exists, $g, $r);
5330 }
5331 }
5332 values %$exists;
5333}
5334sub minimize_url {
5336 my ($self) = @_;
5337 return $self->{url} if ($self->{url} eq $self->{repos_root});
5338 my $url = $self->{repos_root};
5339 my @components = split(m!/!, $self->{svn_path});
5340 my $c = '';
5341 do {
5342 $url .= "/$c" if length $c;
5343 eval {
5344 my $ra = (ref $self)->new($url);
5345 my $latest = $ra->get_latest_revnum;
5346 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5347 };
5348 } while ($@ && ($c = shift @components));
5349 $url;
5350}
5351sub can_do_switch {
5353 my $self = shift;
5354 unless (defined $can_do_switch) {
5355 my $pool = SVN::Pool->new;
5356 my $rep = eval {
5357 $self->do_switch(1, '', 0, $self->{url},
5358 SVN::Delta::Editor->new, $pool);
5359 };
5360 if ($@) {
5361 $can_do_switch = 0;
5362 } else {
5363 $rep->abort_report($pool);
5364 $can_do_switch = 1;
5365 }
5366 $pool->clear;
5367 }
5368 $can_do_switch;
5369}
5370sub skip_unknown_revs {
5372 my ($err) = @_;
5373 my $errno = $err->apr_err();
5374 # Maybe the branch we're tracking didn't
5375 # exist when the repo started, so it's
5376 # not an error if it doesn't, just continue
5377 #
5378 # Wonderfully consistent library, eh?
5379 # 160013 - svn:// and file://
5380 # 175002 - http(s)://
5381 # 175007 - http(s):// (this repo required authorization, too...)
5382 # More codes may be discovered later...
5383 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5384 my $err_key = $err->expanded_message;
5385 # revision numbers change every time, filter them out
5386 $err_key =~ s/\d+/\0/g;
5387 $err_key = "$errno\0$err_key";
5388 unless ($ignored_err{$err_key}) {
5389 warn "W: Ignoring error from SVN, path probably ",
5390 "does not exist: ($errno): ",
5391 $err->expanded_message,"\n";
5392 warn "W: Do not be alarmed at the above message ",
5393 "git-svn is just searching aggressively for ",
5394 "old history.\n",
5395 "This may take a while on large repositories\n";
5396 $ignored_err{$err_key} = 1;
5397 }
5398 return;
5399 }
5400 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5401}
5402package Git::SVN::Log;
5404use strict;
5405use warnings;
5406use POSIX qw/strftime/;
5407use Time::Local;
5408use constant commit_log_separator => ('-' x 72) . "\n";
5409use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5410 %rusers $show_commit $incremental/;
5411my $l_fmt;
5412sub cmt_showable {
5414 my ($c) = @_;
5415 return 1 if defined $c->{r};
5416 # big commit message got truncated by the 16k pretty buffer in rev-list
5418 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5419 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5420 @{$c->{l}} = ();
5421 my @log = command(qw/cat-file commit/, $c->{c});
5422 # shift off the headers
5424 shift @log while ($log[0] ne '');
5425 shift @log;
5426 # TODO: make $c->{l} not have a trailing newline in the future
5428 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5429 (undef, $c->{r}, undef) = ::extract_metadata(
5431 (grep(/^git-svn-id: /, @log))[-1]);
5432 }
5433 return defined $c->{r};
5434}
5435sub log_use_color {
5437 return $color || Git->repository->get_colorbool('color.diff');
5438}
5439sub git_svn_log_cmd {
5441 my ($r_min, $r_max, @args) = @_;
5442 my $head = 'HEAD';
5443 my (@files, @log_opts);
5444 foreach my $x (@args) {
5445 if ($x eq '--' || @files) {
5446 push @files, $x;
5447 } else {
5448 if (::verify_ref("$x^0")) {
5449 $head = $x;
5450 } else {
5451 push @log_opts, $x;
5452 }
5453 }
5454 }
5455 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5457 $gs ||= Git::SVN->_new;
5458 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5459 $gs->refname);
5460 push @cmd, '-r' unless $non_recursive;
5461 push @cmd, qw/--raw --name-status/ if $verbose;
5462 push @cmd, '--color' if log_use_color();
5463 push @cmd, @log_opts;
5464 if (defined $r_max && $r_max == $r_min) {
5465 push @cmd, '--max-count=1';
5466 if (my $c = $gs->rev_map_get($r_max)) {
5467 push @cmd, $c;
5468 }
5469 } elsif (defined $r_max) {
5470 if ($r_max < $r_min) {
5471 ($r_min, $r_max) = ($r_max, $r_min);
5472 }
5473 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5474 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5475 # If there are no commits in the range, both $c_max and $c_min
5476 # will be undefined. If there is at least 1 commit in the
5477 # range, both will be defined.
5478 return () if !defined $c_min || !defined $c_max;
5479 if ($c_min eq $c_max) {
5480 push @cmd, '--max-count=1', $c_min;
5481 } else {
5482 push @cmd, '--boundary', "$c_min..$c_max";
5483 }
5484 }
5485 return (@cmd, @files);
5486}
5487# adapted from pager.c
5489sub config_pager {
5490 if (! -t *STDOUT) {
5491 $ENV{GIT_PAGER_IN_USE} = 'false';
5492 $pager = undef;
5493 return;
5494 }
5495 chomp($pager = command_oneline(qw(var GIT_PAGER)));
5496 if ($pager eq 'cat') {
5497 $pager = undef;
5498 }
5499 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5500}
5501sub run_pager {
5503 return unless defined $pager;
5504 pipe my ($rfd, $wfd) or return;
5505 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5506 if (!$pid) {
5507 open STDOUT, '>&', $wfd or
5508 ::fatal "Can't redirect to stdout: $!";
5509 return;
5510 }
5511 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5512 $ENV{LESS} ||= 'FRSX';
5513 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5514}
5515sub format_svn_date {
5517 # some systmes don't handle or mishandle %z, so be creative.
5518 my $t = shift || time;
5519 my $gm = timelocal(gmtime($t));
5520 my $sign = qw( + + - )[ $t <=> $gm ];
5521 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5522 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5523}
5524sub parse_git_date {
5526 my ($t, $tz) = @_;
5527 # Date::Parse isn't in the standard Perl distro :(
5528 if ($tz =~ s/^\+//) {
5529 $t += tz_to_s_offset($tz);
5530 } elsif ($tz =~ s/^\-//) {
5531 $t -= tz_to_s_offset($tz);
5532 }
5533 return $t;
5534}
5535sub set_local_timezone {
5537 if (defined $TZ) {
5538 $ENV{TZ} = $TZ;
5539 } else {
5540 delete $ENV{TZ};
5541 }
5542}
5543sub tz_to_s_offset {
5545 my ($tz) = @_;
5546 $tz =~ s/(\d\d)$//;
5547 return ($1 * 60) + ($tz * 3600);
5548}
5549sub get_author_info {
5551 my ($dest, $author, $t, $tz) = @_;
5552 $author =~ s/(?:^\s*|\s*$)//g;
5553 $dest->{a_raw} = $author;
5554 my $au;
5555 if ($::_authors) {
5556 $au = $rusers{$author} || undef;
5557 }
5558 if (!$au) {
5559 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5560 }
5561 $dest->{t} = $t;
5562 $dest->{tz} = $tz;
5563 $dest->{a} = $au;
5564 $dest->{t_utc} = parse_git_date($t, $tz);
5565}
5566sub process_commit {
5568 my ($c, $r_min, $r_max, $defer) = @_;
5569 if (defined $r_min && defined $r_max) {
5570 if ($r_min == $c->{r} && $r_min == $r_max) {
5571 show_commit($c);
5572 return 0;
5573 }
5574 return 1 if $r_min == $r_max;
5575 if ($r_min < $r_max) {
5576 # we need to reverse the print order
5577 return 0 if (defined $limit && --$limit < 0);
5578 push @$defer, $c;
5579 return 1;
5580 }
5581 if ($r_min != $r_max) {
5582 return 1 if ($r_min < $c->{r});
5583 return 1 if ($r_max > $c->{r});
5584 }
5585 }
5586 return 0 if (defined $limit && --$limit < 0);
5587 show_commit($c);
5588 return 1;
5589}
5590sub show_commit {
5592 my $c = shift;
5593 if ($oneline) {
5594 my $x = "\n";
5595 if (my $l = $c->{l}) {
5596 while ($l->[0] =~ /^\s*$/) { shift @$l }
5597 $x = $l->[0];
5598 }
5599 $l_fmt ||= 'A' . length($c->{r});
5600 print 'r',pack($l_fmt, $c->{r}),' | ';
5601 print "$c->{c} | " if $show_commit;
5602 print $x;
5603 } else {
5604 show_commit_normal($c);
5605 }
5606}
5607sub show_commit_changed_paths {
5609 my ($c) = @_;
5610 return unless $c->{changed};
5611 print "Changed paths:\n", @{$c->{changed}};
5612}
5613sub show_commit_normal {
5615 my ($c) = @_;
5616 print commit_log_separator, "r$c->{r} | ";
5617 print "$c->{c} | " if $show_commit;
5618 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5619 my $nr_line = 0;
5620 if (my $l = $c->{l}) {
5622 while ($l->[$#$l] eq "\n" && $#$l > 0
5623 && $l->[($#$l - 1)] eq "\n") {
5624 pop @$l;
5625 }
5626 $nr_line = scalar @$l;
5627 if (!$nr_line) {
5628 print "1 line\n\n\n";
5629 } else {
5630 if ($nr_line == 1) {
5631 $nr_line = '1 line';
5632 } else {
5633 $nr_line .= ' lines';
5634 }
5635 print $nr_line, "\n";
5636 show_commit_changed_paths($c);
5637 print "\n";
5638 print $_ foreach @$l;
5639 }
5640 } else {
5641 print "1 line\n";
5642 show_commit_changed_paths($c);
5643 print "\n";
5644 }
5646 foreach my $x (qw/raw stat diff/) {
5647 if ($c->{$x}) {
5648 print "\n";
5649 print $_ foreach @{$c->{$x}}
5650 }
5651 }
5652}
5653sub cmd_show_log {
5655 my (@args) = @_;
5656 my ($r_min, $r_max);
5657 my $r_last = -1; # prevent dupes
5658 set_local_timezone();
5659 if (defined $::_revision) {
5660 if ($::_revision =~ /^(\d+):(\d+)$/) {
5661 ($r_min, $r_max) = ($1, $2);
5662 } elsif ($::_revision =~ /^\d+$/) {
5663 $r_min = $r_max = $::_revision;
5664 } else {
5665 ::fatal "-r$::_revision is not supported, use ",
5666 "standard 'git log' arguments instead";
5667 }
5668 }
5669 config_pager();
5671 @args = git_svn_log_cmd($r_min, $r_max, @args);
5672 if (!@args) {
5673 print commit_log_separator unless $incremental || $oneline;
5674 return;
5675 }
5676 my $log = command_output_pipe(@args);
5677 run_pager();
5678 my (@k, $c, $d, $stat);
5679 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5680 while (<$log>) {
5681 if (/^${esc_color}commit -?($::sha1_short)/o) {
5682 my $cmt = $1;
5683 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5684 $r_last = $c->{r};
5685 process_commit($c, $r_min, $r_max, \@k) or
5686 goto out;
5687 }
5688 $d = undef;
5689 $c = { c => $cmt };
5690 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5691 get_author_info($c, $1, $2, $3);
5692 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5693 # ignore
5694 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5695 push @{$c->{raw}}, $_;
5696 } elsif (/^${esc_color}[ACRMDT]\t/) {
5697 # we could add $SVN->{svn_path} here, but that requires
5698 # remote access at the moment (repo_path_split)...
5699 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5700 push @{$c->{changed}}, $_;
5701 } elsif (/^${esc_color}diff /o) {
5702 $d = 1;
5703 push @{$c->{diff}}, $_;
5704 } elsif ($d) {
5705 push @{$c->{diff}}, $_;
5706 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5707 $esc_color*[\+\-]*$esc_color$/x) {
5708 $stat = 1;
5709 push @{$c->{stat}}, $_;
5710 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5711 push @{$c->{stat}}, $_;
5712 $stat = undef;
5713 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5714 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5715 } elsif (s/^${esc_color} //o) {
5716 push @{$c->{l}}, $_;
5717 }
5718 }
5719 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5720 $r_last = $c->{r};
5721 process_commit($c, $r_min, $r_max, \@k);
5722 }
5723 if (@k) {
5724 ($r_min, $r_max) = ($r_max, $r_min);
5725 process_commit($_, $r_min, $r_max) foreach reverse @k;
5726 }
5727out:
5728 close $log;
5729 print commit_log_separator unless $incremental || $oneline;
5730}
5731sub cmd_blame {
5733 my $path = pop;
5734 config_pager();
5736 run_pager();
5737 my ($fh, $ctx, $rev);
5739 if ($_git_format) {
5741 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5742 while (my $line = <$fh>) {
5743 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5744 # Uncommitted edits show up as a rev ID of
5745 # all zeros, which we can't look up with
5746 # cmt_metadata
5747 if ($1 !~ /^0+$/) {
5748 (undef, $rev, undef) =
5749 ::cmt_metadata($1);
5750 $rev = '0' if (!$rev);
5751 } else {
5752 $rev = '0';
5753 }
5754 $rev = sprintf('%-10s', $rev);
5755 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5756 }
5757 print $line;
5758 }
5759 } else {
5760 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5761 '--', $path);
5762 my ($sha1);
5763 my %authors;
5764 my @buffer;
5765 my %dsha; #distinct sha keys
5766 while (my $line = <$fh>) {
5768 push @buffer, $line;
5769 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5770 $dsha{$1} = 1;
5771 }
5772 }
5773 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5775 foreach my $line (@buffer) {
5777 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5778 $rev = $s2r->{$1};
5779 $rev = '0' if (!$rev)
5780 }
5781 elsif ($line =~ /^author (.*)/) {
5782 $authors{$rev} = $1;
5783 $authors{$rev} =~ s/\s/_/g;
5784 }
5785 elsif ($line =~ /^\t(.*)$/) {
5786 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5787 }
5788 }
5789 }
5790 command_close_pipe($fh, $ctx);
5791}
5792package Git::SVN::Migration;
5794# these version numbers do NOT correspond to actual version numbers
5795# of git nor git-svn. They are just relative.
5796#
5797# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5798#
5799# v1 layout: .git/$id/info/url, refs/remotes/$id
5800#
5801# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5802#
5803# v3 layout: .git/svn/$id, refs/remotes/$id
5804# - info/url may remain for backwards compatibility
5805# - this is what we migrate up to this layout automatically,
5806# - this will be used by git svn init on single branches
5807# v3.1 layout (auto migrated):
5808# - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5809# for backwards compatibility
5810#
5811# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5812# - this is only created for newly multi-init-ed
5813# repositories. Similar in spirit to the
5814# --use-separate-remotes option in git-clone (now default)
5815# - we do not automatically migrate to this (following
5816# the example set by core git)
5817#
5818# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5819# - newer, more-efficient format that uses 24-bytes per record
5820# with no filler space.
5821# - use xxd -c24 < .rev_map.$UUID to view and debug
5822# - This is a one-way migration, repositories updated to the
5823# new format will not be able to use old git-svn without
5824# rebuilding the .rev_db. Rebuilding the rev_db is not
5825# possible if noMetadata or useSvmProps are set; but should
5826# be no problem for users that use the (sensible) defaults.
5827use strict;
5828use warnings;
5829use Carp qw/croak/;
5830use File::Path qw/mkpath/;
5831use File::Basename qw/dirname basename/;
5832use vars qw/$_minimize/;
5833sub migrate_from_v0 {
5835 my $git_dir = $ENV{GIT_DIR};
5836 return undef unless -d $git_dir;
5837 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5838 my $migrated = 0;
5839 while (<$fh>) {
5840 chomp;
5841 my ($id, $orig_ref) = ($_, $_);
5842 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5843 next unless -f "$git_dir/$id/info/url";
5844 my $new_ref = "refs/remotes/$id";
5845 if (::verify_ref("$new_ref^0")) {
5846 print STDERR "W: $orig_ref is probably an old ",
5847 "branch used by an ancient version of ",
5848 "git-svn.\n",
5849 "However, $new_ref also exists.\n",
5850 "We will not be able ",
5851 "to use this branch until this ",
5852 "ambiguity is resolved.\n";
5853 next;
5854 }
5855 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5856 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5857 command_noisy('update-ref', $new_ref, $orig_ref);
5858 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5859 $migrated++;
5860 }
5861 command_close_pipe($fh, $ctx);
5862 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5863 $migrated;
5864}
5865sub migrate_from_v1 {
5867 my $git_dir = $ENV{GIT_DIR};
5868 my $migrated = 0;
5869 return $migrated unless -d $git_dir;
5870 my $svn_dir = "$git_dir/svn";
5871 # just in case somebody used 'svn' as their $id at some point...
5873 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5874 print STDERR "Migrating from a git-svn v1 layout...\n";
5876 mkpath([$svn_dir]);
5877 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5878 "$svn_dir\n\t(required for this version ",
5879 "($::VERSION) of git-svn) does not exist.\n";
5880 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5881 while (<$fh>) {
5882 my $x = $_;
5883 next unless $x =~ s#^refs/remotes/##;
5884 chomp $x;
5885 next unless -f "$git_dir/$x/info/url";
5886 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5887 next unless $u;
5888 my $dn = dirname("$git_dir/svn/$x");
5889 mkpath([$dn]) unless -d $dn;
5890 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5891 mkpath(["$git_dir/svn/svn"]);
5892 print STDERR " - $git_dir/$x/info => ",
5893 "$git_dir/svn/$x/info\n";
5894 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5895 croak "$!: $x";
5896 # don't worry too much about these, they probably
5897 # don't exist with repos this old (save for index,
5898 # and we can easily regenerate that)
5899 foreach my $f (qw/unhandled.log index .rev_db/) {
5900 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5901 }
5902 } else {
5903 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5904 rename "$git_dir/$x", "$git_dir/svn/$x" or
5905 croak "$!: $x";
5906 }
5907 $migrated++;
5908 }
5909 command_close_pipe($fh, $ctx);
5910 print STDERR "Done migrating from a git-svn v1 layout\n";
5911 $migrated;
5912}
5913sub read_old_urls {
5915 my ($l_map, $pfx, $path) = @_;
5916 my @dir;
5917 foreach (<$path/*>) {
5918 if (-r "$_/info/url") {
5919 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5920 my $ref_id = $pfx . basename $_;
5921 my $url = ::file_to_s("$_/info/url");
5922 $l_map->{$ref_id} = $url;
5923 } elsif (-d $_) {
5924 push @dir, $_;
5925 }
5926 }
5927 foreach (@dir) {
5928 my $x = $_;
5929 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5930 read_old_urls($l_map, $x, $_);
5931 }
5932}
5933sub migrate_from_v2 {
5935 my @cfg = command(qw/config -l/);
5936 return if grep /^svn-remote\..+\.url=/, @cfg;
5937 my %l_map;
5938 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5939 my $migrated = 0;
5940 foreach my $ref_id (sort keys %l_map) {
5942 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5943 if ($@) {
5944 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5945 }
5946 $migrated++;
5947 }
5948 $migrated;
5949}
5950sub minimize_connections {
5952 my $r = Git::SVN::read_all_remotes();
5953 my $new_urls = {};
5954 my $root_repos = {};
5955 foreach my $repo_id (keys %$r) {
5956 my $url = $r->{$repo_id}->{url} or next;
5957 my $fetch = $r->{$repo_id}->{fetch} or next;
5958 my $ra = Git::SVN::Ra->new($url);
5959 # skip existing cases where we already connect to the root
5961 if (($ra->{url} eq $ra->{repos_root}) ||
5962 ($ra->{repos_root} eq $repo_id)) {
5963 $root_repos->{$ra->{url}} = $repo_id;
5964 next;
5965 }
5966 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5968 my $root_path = $ra->{url};
5969 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5970 foreach my $path (keys %$fetch) {
5971 my $ref_id = $fetch->{$path};
5972 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5973 # make sure we can read when connecting to
5975 # a higher level of a repository
5976 my ($last_rev, undef) = $gs->last_rev_commit;
5977 if (!defined $last_rev) {
5978 $last_rev = eval {
5979 $root_ra->get_latest_revnum;
5980 };
5981 next if $@;
5982 }
5983 my $new = $root_path;
5984 $new .= length $path ? "/$path" : '';
5985 eval {
5986 $root_ra->get_log([$new], $last_rev, $last_rev,
5987 0, 0, 1, sub { });
5988 };
5989 next if $@;
5990 $new_urls->{$ra->{repos_root}}->{$new} =
5991 { ref_id => $ref_id,
5992 old_repo_id => $repo_id,
5993 old_path => $path };
5994 }
5995 }
5996 my @emptied;
5998 foreach my $url (keys %$new_urls) {
5999 # see if we can re-use an existing [svn-remote "repo_id"]
6000 # instead of creating a(n ugly) new section:
6001 my $repo_id = $root_repos->{$url} || $url;
6002 my $fetch = $new_urls->{$url};
6004 foreach my $path (keys %$fetch) {
6005 my $x = $fetch->{$path};
6006 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
6007 my $pfx = "svn-remote.$x->{old_repo_id}";
6008 my $old_fetch = quotemeta("$x->{old_path}:".
6010 "$x->{ref_id}");
6011 command_noisy(qw/config --unset/,
6012 "$pfx.fetch", '^'. $old_fetch . '$');
6013 delete $r->{$x->{old_repo_id}}->
6014 {fetch}->{$x->{old_path}};
6015 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6016 command_noisy(qw/config --unset/,
6017 "$pfx.url");
6018 push @emptied, $x->{old_repo_id}
6019 }
6020 }
6021 }
6022 if (@emptied) {
6023 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6024 print STDERR <<EOF;
6025The following [svn-remote] sections in your config file ($file) are empty
6026and can be safely removed:
6027EOF
6028 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6029 }
6030}
6031sub migration_check {
6033 migrate_from_v0();
6034 migrate_from_v1();
6035 migrate_from_v2();
6036 minimize_connections() if $_minimize;
6037}
6038package Git::IndexInfo;
6040use strict;
6041use warnings;
6042use Git qw/command_input_pipe command_close_pipe/;
6043sub new {
6045 my ($class) = @_;
6046 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6047 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6048}
6049sub remove {
6051 my ($self, $path) = @_;
6052 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6053 return ++$self->{nr};
6054 }
6055 undef;
6056}
6057sub update {
6059 my ($self, $mode, $hash, $path) = @_;
6060 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6061 return ++$self->{nr};
6062 }
6063 undef;
6064}
6065sub DESTROY {
6067 my ($self) = @_;
6068 command_close_pipe($self->{gui}, $self->{ctx});
6069}
6070package Git::SVN::GlobSpec;
6072use strict;
6073use warnings;
6074sub new {
6076 my ($class, $glob, $pattern_ok) = @_;
6077 my $re = $glob;
6078 $re =~ s!/+$!!g; # no need for trailing slashes
6079 my (@left, @right, @patterns);
6080 my $state = "left";
6081 my $die_msg = "Only one set of wildcard directories " .
6082 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6083 for my $part (split(m|/|, $glob)) {
6084 if ($part =~ /\*/ && $part ne "*") {
6085 die "Invalid pattern in '$glob': $part\n";
6086 } elsif ($pattern_ok && $part =~ /[{}]/ &&
6087 $part !~ /^\{[^{}]+\}/) {
6088 die "Invalid pattern in '$glob': $part\n";
6089 }
6090 if ($part eq "*") {
6091 die $die_msg if $state eq "right";
6092 $state = "pattern";
6093 push(@patterns, "[^/]*");
6094 } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6095 die $die_msg if $state eq "right";
6096 $state = "pattern";
6097 my $p = quotemeta($1);
6098 $p =~ s/\\,/|/g;
6099 push(@patterns, "(?:$p)");
6100 } else {
6101 if ($state eq "left") {
6102 push(@left, $part);
6103 } else {
6104 push(@right, $part);
6105 $state = "right";
6106 }
6107 }
6108 }
6109 my $depth = @patterns;
6110 if ($depth == 0) {
6111 die "One '*' is needed in glob: '$glob'\n";
6112 }
6113 my $left = join('/', @left);
6114 my $right = join('/', @right);
6115 $re = join('/', @patterns);
6116 $re = join('\/',
6117 grep(length, quotemeta($left), "($re)", quotemeta($right)));
6118 my $left_re = qr/^\/\Q$left\E(\/|$)/;
6119 bless { left => $left, right => $right, left_regex => $left_re,
6120 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6121}
6122sub full_path {
6124 my ($self, $path) = @_;
6125 return (length $self->{left} ? "$self->{left}/" : '') .
6126 $path . (length $self->{right} ? "/$self->{right}" : '');
6127}
6128__END__
6130Data structures:
6132$remotes = { # returned by read_all_remotes()
6135 'svn' => {
6136 # svn-remote.svn.url=https://svn.musicpd.org
6137 url => 'https://svn.musicpd.org',
6138 # svn-remote.svn.fetch=mpd/trunk:trunk
6139 fetch => {
6140 'mpd/trunk' => 'trunk',
6141 },
6142 # svn-remote.svn.tags=mpd/tags/*:tags/*
6143 tags => {
6144 path => {
6145 left => 'mpd/tags',
6146 right => '',
6147 regex => qr!mpd/tags/([^/]+)$!,
6148 glob => 'tags/*',
6149 },
6150 ref => {
6151 left => 'tags',
6152 right => '',
6153 regex => qr!tags/([^/]+)$!,
6154 glob => 'tags/*',
6155 },
6156 }
6157 }
6158};
6159$log_entry hashref as returned by libsvn_log_entry()
6161{
6162 log => 'whitespace-formatted log entry
6163', # trailing newline is preserved
6164 revision => '8', # integer
6165 date => '2004-02-24T17:01:44.108345Z', # commit date
6166 author => 'committer name'
6167};
6168# this is generated by generate_diff();
6171@mods = array of diff-index line hashes, each element represents one line
6172 of diff-index output
6173diff-index line ($m hash)
6175{
6176 mode_a => first column of diff-index output, no leading ':',
6177 mode_b => second column of diff-index output,
6178 sha1_b => sha1sum of the final blob,
6179 chg => change type [MCRADT],
6180 file_a => original file name of a file (iff chg is 'C' or 'R')
6181 file_b => new/current file name of a file (any chg)
6182}
6183;
6184# retval of read_url_paths{,_all}();
6186$l_map = {
6187 # repository root url
6188 'https://svn.musicpd.org' => {
6189 # repository path # GIT_SVN_ID
6190 'mpd/trunk' => 'trunk',
6191 'mpd/tags/0.11.5' => 'tags/0.11.5',
6192 },
6193}
6194Notes:
6196 I don't trust the each() function on unless I created %hash myself
6197 because the internal iterator may not have started at base.