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 $self->{url} = command_oneline('config', '--get',
2059 "svn-remote.$repo_id.url") or
2060 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2061 $self->rebuild;
2062 $self;
2063}
2064sub refname {
2066 my ($refname) = $_[0]->{ref_id} ;
2067 # It cannot end with a slash /, we'll throw up on this because
2069 # SVN can't have directories with a slash in their name, either:
2070 if ($refname =~ m{/$}) {
2071 die "ref: '$refname' ends with a trailing slash, this is ",
2072 "not permitted by git nor Subversion\n";
2073 }
2074 # It cannot have ASCII control character space, tilde ~, caret ^,
2076 # colon :, question-mark ?, asterisk *, space, or open bracket [
2077 # anywhere.
2078 #
2079 # Additionally, % must be escaped because it is used for escaping
2080 # and we want our escaped refname to be reversible
2081 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2082 # no slash-separated component can begin with a dot .
2084 # /.* becomes /%2E*
2085 $refname =~ s{/\.}{/%2E}g;
2086 # It cannot have two consecutive dots .. anywhere
2088 # .. becomes %2E%2E
2089 $refname =~ s{\.\.}{%2E%2E}g;
2090 # trailing dots and .lock are not allowed
2092 # .$ becomes %2E and .lock becomes %2Elock
2093 $refname =~ s{\.(?=$|lock$)}{%2E};
2094 # the sequence @{ is used to access the reflog
2096 # @{ becomes %40{
2097 $refname =~ s{\@\{}{%40\{}g;
2098 return $refname;
2100}
2101sub desanitize_refname {
2103 my ($refname) = @_;
2104 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2105 return $refname;
2106}
2107sub svm_uuid {
2109 my ($self) = @_;
2110 return $self->{svm}->{uuid} if $self->svm;
2111 $self->ra;
2112 unless ($self->{svm}) {
2113 die "SVM UUID not cached, and reading remotely failed\n";
2114 }
2115 $self->{svm}->{uuid};
2116}
2117sub svm {
2119 my ($self) = @_;
2120 return $self->{svm} if $self->{svm};
2121 my $svm;
2122 # see if we have it in our config, first:
2123 eval {
2124 my $section = "svn-remote.$self->{repo_id}";
2125 $svm = {
2126 source => tmp_config('--get', "$section.svm-source"),
2127 uuid => tmp_config('--get', "$section.svm-uuid"),
2128 replace => tmp_config('--get', "$section.svm-replace"),
2129 }
2130 };
2131 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2132 $self->{svm} = $svm;
2133 }
2134 $self->{svm};
2135}
2136sub _set_svm_vars {
2138 my ($self, $ra) = @_;
2139 return $ra if $self->svm;
2140 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2142 "(svm:source, svm:uuid) ",
2143 "from the following URLs:\n" );
2144 sub read_svm_props {
2145 my ($self, $ra, $path, $r) = @_;
2146 my $props = ($ra->get_dir($path, $r))[2];
2147 my $src = $props->{'svm:source'};
2148 my $uuid = $props->{'svm:uuid'};
2149 return undef if (!$src || !$uuid);
2150 chomp($src, $uuid);
2152 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2154 or die "doesn't look right - svm:uuid is '$uuid'\n";
2155 # the '!' is used to mark the repos_root!/relative/path
2157 $src =~ s{/?!/?}{/};
2158 $src =~ s{/+$}{}; # no trailing slashes please
2159 # username is of no interest
2160 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2161 my $replace = $ra->{url};
2163 $replace .= "/$path" if length $path;
2164 my $section = "svn-remote.$self->{repo_id}";
2166 tmp_config("$section.svm-source", $src);
2167 tmp_config("$section.svm-replace", $replace);
2168 tmp_config("$section.svm-uuid", $uuid);
2169 $self->{svm} = {
2170 source => $src,
2171 uuid => $uuid,
2172 replace => $replace
2173 };
2174 }
2175 my $r = $ra->get_latest_revnum;
2177 my $path = $self->{path};
2178 my %tried;
2179 while (length $path) {
2180 unless ($tried{"$self->{url}/$path"}) {
2181 return $ra if $self->read_svm_props($ra, $path, $r);
2182 $tried{"$self->{url}/$path"} = 1;
2183 }
2184 $path =~ s#/?[^/]+$##;
2185 }
2186 die "Path: '$path' should be ''\n" if $path ne '';
2187 return $ra if $self->read_svm_props($ra, $path, $r);
2188 $tried{"$self->{url}/$path"} = 1;
2189 if ($ra->{repos_root} eq $self->{url}) {
2191 die @err, (map { " $_\n" } keys %tried), "\n";
2192 }
2193 # nope, make sure we're connected to the repository root:
2195 my $ok;
2196 my @tried_b;
2197 $path = $ra->{svn_path};
2198 $ra = Git::SVN::Ra->new($ra->{repos_root});
2199 while (length $path) {
2200 unless ($tried{"$ra->{url}/$path"}) {
2201 $ok = $self->read_svm_props($ra, $path, $r);
2202 last if $ok;
2203 $tried{"$ra->{url}/$path"} = 1;
2204 }
2205 $path =~ s#/?[^/]+$##;
2206 }
2207 die "Path: '$path' should be ''\n" if $path ne '';
2208 $ok ||= $self->read_svm_props($ra, $path, $r);
2209 $tried{"$ra->{url}/$path"} = 1;
2210 if (!$ok) {
2211 die @err, (map { " $_\n" } keys %tried), "\n";
2212 }
2213 Git::SVN::Ra->new($self->{url});
2214}
2215sub svnsync {
2217 my ($self) = @_;
2218 return $self->{svnsync} if $self->{svnsync};
2219 if ($self->no_metadata) {
2221 die "Can't have both 'noMetadata' and ",
2222 "'useSvnsyncProps' options set!\n";
2223 }
2224 if ($self->rewrite_root) {
2225 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2226 "options set!\n";
2227 }
2228 if ($self->rewrite_uuid) {
2229 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2230 "options set!\n";
2231 }
2232 my $svnsync;
2234 # see if we have it in our config, first:
2235 eval {
2236 my $section = "svn-remote.$self->{repo_id}";
2237 my $url = tmp_config('--get', "$section.svnsync-url");
2239 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2240 die "doesn't look right - svn:sync-from-url is '$url'\n";
2241 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2243 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2244 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2245 $svnsync = { url => $url, uuid => $uuid }
2247 };
2248 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2249 return $self->{svnsync} = $svnsync;
2250 }
2251 my $err = "useSvnsyncProps set, but failed to read " .
2253 "svnsync property: svn:sync-from-";
2254 my $rp = $self->ra->rev_proplist(0);
2255 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2257 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2258 die "doesn't look right - svn:sync-from-url is '$url'\n";
2259 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2261 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2262 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2263 my $section = "svn-remote.$self->{repo_id}";
2265 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2266 tmp_config('--add', "$section.svnsync-url", $url);
2267 return $self->{svnsync} = { url => $url, uuid => $uuid };
2268}
2269# this allows us to memoize our SVN::Ra UUID locally and avoid a
2271# remote lookup (useful for 'git svn log').
2272sub ra_uuid {
2273 my ($self) = @_;
2274 unless ($self->{ra_uuid}) {
2275 my $key = "svn-remote.$self->{repo_id}.uuid";
2276 my $uuid = eval { tmp_config('--get', $key) };
2277 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2278 $self->{ra_uuid} = $uuid;
2279 } else {
2280 die "ra_uuid called without URL\n" unless $self->{url};
2281 $self->{ra_uuid} = $self->ra->get_uuid;
2282 tmp_config('--add', $key, $self->{ra_uuid});
2283 }
2284 }
2285 $self->{ra_uuid};
2286}
2287sub _set_repos_root {
2289 my ($self, $repos_root) = @_;
2290 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2291 $repos_root ||= $self->ra->{repos_root};
2292 tmp_config($k, $repos_root);
2293 $repos_root;
2294}
2295sub repos_root {
2297 my ($self) = @_;
2298 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2299 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2300}
2301sub ra {
2303 my ($self) = shift;
2304 my $ra = Git::SVN::Ra->new($self->{url});
2305 $self->_set_repos_root($ra->{repos_root});
2306 if ($self->use_svm_props && !$self->{svm}) {
2307 if ($self->no_metadata) {
2308 die "Can't have both 'noMetadata' and ",
2309 "'useSvmProps' options set!\n";
2310 } elsif ($self->use_svnsync_props) {
2311 die "Can't have both 'useSvnsyncProps' and ",
2312 "'useSvmProps' options set!\n";
2313 }
2314 $ra = $self->_set_svm_vars($ra);
2315 $self->{-want_revprops} = 1;
2316 }
2317 $ra;
2318}
2319# prop_walk(PATH, REV, SUB)
2321# -------------------------
2322# Recursively traverse PATH at revision REV and invoke SUB for each
2323# directory that contains a SVN property. SUB will be invoked as
2324# follows: &SUB(gs, path, props); where `gs' is this instance of
2325# Git::SVN, `path' the path to the directory where the properties
2326# `props' were found. The `path' will be relative to point of checkout,
2327# that is, if url://repo/trunk is the current Git branch, and that
2328# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2329# as `path' (note the trailing `/').
2330sub prop_walk {
2331 my ($self, $path, $rev, $sub) = @_;
2332 $path =~ s#^/##;
2334 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2335 $path =~ s#^/*#/#g;
2336 my $p = $path;
2337 # Strip the irrelevant part of the path.
2338 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2339 # Ensure the path is terminated by a `/'.
2340 $p =~ s#/*$#/#;
2341 # The properties contain all the internal SVN stuff nobody
2343 # (usually) cares about.
2344 my $interesting_props = 0;
2345 foreach (keys %{$props}) {
2346 # If it doesn't start with `svn:', it must be a
2347 # user-defined property.
2348 ++$interesting_props and next if $_ !~ /^svn:/;
2349 # FIXME: Fragile, if SVN adds new public properties,
2350 # this needs to be updated.
2351 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2352 |eol-style|mime-type
2353 |externals|needs-lock)$/x;
2354 }
2355 &$sub($self, $p, $props) if $interesting_props;
2356 foreach (sort keys %$dirent) {
2358 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2359 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2360 }
2361}
2362sub last_rev { ($_[0]->last_rev_commit)[0] }
2364sub last_commit { ($_[0]->last_rev_commit)[1] }
2365# returns the newest SVN revision number and newest commit SHA1
2367sub last_rev_commit {
2368 my ($self) = @_;
2369 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2370 return ($self->{last_rev}, $self->{last_commit});
2371 }
2372 my $c = ::verify_ref($self->refname.'^0');
2373 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2374 my $rev = (::cmt_metadata($c))[1];
2375 if (defined $rev) {
2376 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2377 return ($rev, $c);
2378 }
2379 }
2380 my $map_path = $self->map_path;
2381 unless (-e $map_path) {
2382 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2383 return (undef, undef);
2384 }
2385 my ($rev, $commit) = $self->rev_map_max(1);
2386 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2387 return ($rev, $commit);
2388}
2389sub get_fetch_range {
2391 my ($self, $min, $max) = @_;
2392 $max ||= $self->ra->get_latest_revnum;
2393 $min ||= $self->rev_map_max;
2394 (++$min, $max);
2395}
2396sub tmp_config {
2398 my (@args) = @_;
2399 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2400 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2401 if (! -f $config && -f $old_def_config) {
2402 rename $old_def_config, $config or
2403 die "Failed rename $old_def_config => $config: $!\n";
2404 }
2405 my $old_config = $ENV{GIT_CONFIG};
2406 $ENV{GIT_CONFIG} = $config;
2407 $@ = undef;
2408 my @ret = eval {
2409 unless (-f $config) {
2410 mkfile($config);
2411 open my $fh, '>', $config or
2412 die "Can't open $config: $!\n";
2413 print $fh "; This file is used internally by ",
2414 "git-svn\n" or die
2415 "Couldn't write to $config: $!\n";
2416 print $fh "; You should not have to edit it\n" or
2417 die "Couldn't write to $config: $!\n";
2418 close $fh or die "Couldn't close $config: $!\n";
2419 }
2420 command('config', @args);
2421 };
2422 my $err = $@;
2423 if (defined $old_config) {
2424 $ENV{GIT_CONFIG} = $old_config;
2425 } else {
2426 delete $ENV{GIT_CONFIG};
2427 }
2428 die $err if $err;
2429 wantarray ? @ret : $ret[0];
2430}
2431sub tmp_index_do {
2433 my ($self, $sub) = @_;
2434 my $old_index = $ENV{GIT_INDEX_FILE};
2435 $ENV{GIT_INDEX_FILE} = $self->{index};
2436 $@ = undef;
2437 my @ret = eval {
2438 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2439 mkpath([$dir]) unless -d $dir;
2440 &$sub;
2441 };
2442 my $err = $@;
2443 if (defined $old_index) {
2444 $ENV{GIT_INDEX_FILE} = $old_index;
2445 } else {
2446 delete $ENV{GIT_INDEX_FILE};
2447 }
2448 die $err if $err;
2449 wantarray ? @ret : $ret[0];
2450}
2451sub assert_index_clean {
2453 my ($self, $treeish) = @_;
2454 $self->tmp_index_do(sub {
2456 command_noisy('read-tree', $treeish) unless -e $self->{index};
2457 my $x = command_oneline('write-tree');
2458 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2459 /^tree ($::sha1)/mo);
2460 return if $y eq $x;
2461 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2463 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2464 command_noisy('read-tree', $treeish);
2465 $x = command_oneline('write-tree');
2466 if ($y ne $x) {
2467 ::fatal "trees ($treeish) $y != $x\n",
2468 "Something is seriously wrong...";
2469 }
2470 });
2471}
2472sub get_commit_parents {
2474 my ($self, $log_entry) = @_;
2475 my (%seen, @ret, @tmp);
2476 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2477 if (my $ip = $self->{inject_parents}) {
2478 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2479 push @tmp, $commit;
2480 }
2481 }
2482 if (my $cur = ::verify_ref($self->refname.'^0')) {
2483 push @tmp, $cur;
2484 }
2485 if (my $ipd = $self->{inject_parents_dcommit}) {
2486 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2487 push @tmp, @$commit;
2488 }
2489 }
2490 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2491 while (my $p = shift @tmp) {
2492 next if $seen{$p};
2493 $seen{$p} = 1;
2494 push @ret, $p;
2495 }
2496 @ret;
2497}
2498sub rewrite_root {
2500 my ($self) = @_;
2501 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2502 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2503 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2504 if ($rwr) {
2505 $rwr =~ s#/+$##;
2506 if ($rwr !~ m#^[a-z\+]+://#) {
2507 die "$rwr is not a valid URL (key: $k)\n";
2508 }
2509 }
2510 $self->{-rewrite_root} = $rwr;
2511}
2512sub rewrite_uuid {
2514 my ($self) = @_;
2515 return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2516 my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2517 my $rwid = eval { command_oneline(qw/config --get/, $k) };
2518 if ($rwid) {
2519 $rwid =~ s#/+$##;
2520 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2521 die "$rwid is not a valid UUID (key: $k)\n";
2522 }
2523 }
2524 $self->{-rewrite_uuid} = $rwid;
2525}
2526sub metadata_url {
2528 my ($self) = @_;
2529 ($self->rewrite_root || $self->{url}) .
2530 (length $self->{path} ? '/' . $self->{path} : '');
2531}
2532sub full_url {
2534 my ($self) = @_;
2535 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2536}
2537sub set_commit_header_env {
2540 my ($log_entry) = @_;
2541 my %env;
2542 foreach my $ned (qw/NAME EMAIL DATE/) {
2543 foreach my $ac (qw/AUTHOR COMMITTER/) {
2544 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2545 }
2546 }
2547 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2549 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2550 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2551 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2553 ? $log_entry->{commit_name}
2554 : $log_entry->{name};
2555 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2556 ? $log_entry->{commit_email}
2557 : $log_entry->{email};
2558 \%env;
2559}
2560sub restore_commit_header_env {
2562 my ($env) = @_;
2563 foreach my $ned (qw/NAME EMAIL DATE/) {
2564 foreach my $ac (qw/AUTHOR COMMITTER/) {
2565 my $k = "GIT_${ac}_${ned}";
2566 if (defined $env->{$k}) {
2567 $ENV{$k} = $env->{$k};
2568 } else {
2569 delete $ENV{$k};
2570 }
2571 }
2572 }
2573}
2574sub gc {
2576 command_noisy('gc', '--auto');
2577};
2578sub do_git_commit {
2580 my ($self, $log_entry) = @_;
2581 my $lr = $self->last_rev;
2582 if (defined $lr && $lr >= $log_entry->{revision}) {
2583 die "Last fetched revision of ", $self->refname,
2584 " was r$lr, but we are about to fetch: ",
2585 "r$log_entry->{revision}!\n";
2586 }
2587 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2588 croak "$log_entry->{revision} = $c already exists! ",
2589 "Why are we refetching it?\n";
2590 }
2591 my $old_env = set_commit_header_env($log_entry);
2592 my $tree = $log_entry->{tree};
2593 if (!defined $tree) {
2594 $tree = $self->tmp_index_do(sub {
2595 command_oneline('write-tree') });
2596 }
2597 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2598 my @exec = ('git', 'commit-tree', $tree);
2600 foreach ($self->get_commit_parents($log_entry)) {
2601 push @exec, '-p', $_;
2602 }
2603 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2604 or croak $!;
2605 binmode $msg_fh;
2606 # we always get UTF-8 from SVN, but we may want our commits in
2608 # a different encoding.
2609 if (my $enc = Git::config('i18n.commitencoding')) {
2610 require Encode;
2611 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2612 }
2613 print $msg_fh $log_entry->{log} or croak $!;
2614 restore_commit_header_env($old_env);
2615 unless ($self->no_metadata) {
2616 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2617 or croak $!;
2618 }
2619 $msg_fh->flush == 0 or croak $!;
2620 close $msg_fh or croak $!;
2621 chomp(my $commit = do { local $/; <$out_fh> });
2622 close $out_fh or croak $!;
2623 waitpid $pid, 0;
2624 croak $? if $?;
2625 if ($commit !~ /^$::sha1$/o) {
2626 die "Failed to commit, invalid sha1: $commit\n";
2627 }
2628 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2630 $self->{last_rev} = $log_entry->{revision};
2632 $self->{last_commit} = $commit;
2633 print "r$log_entry->{revision}" unless $::_q > 1;
2634 if (defined $log_entry->{svm_revision}) {
2635 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2636 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2637 0, $self->svm_uuid);
2638 }
2639 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2640 if (--$_gc_nr == 0) {
2641 $_gc_nr = $_gc_period;
2642 gc();
2643 }
2644 return $commit;
2645}
2646sub match_paths {
2648 my ($self, $paths, $r) = @_;
2649 return 1 if $self->{path} eq '';
2650 if (my $path = $paths->{"/$self->{path}"}) {
2651 return ($path->{action} eq 'D') ? 0 : 1;
2652 }
2653 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2654 if (grep /$self->{path_regex}/, keys %$paths) {
2655 return 1;
2656 }
2657 my $c = '';
2658 foreach (split m#/#, $self->{path}) {
2659 $c .= "/$_";
2660 next unless ($paths->{$c} &&
2661 ($paths->{$c}->{action} =~ /^[AR]$/));
2662 if ($self->ra->check_path($self->{path}, $r) ==
2663 $SVN::Node::dir) {
2664 return 1;
2665 }
2666 }
2667 return 0;
2668}
2669sub find_parent_branch {
2671 my ($self, $paths, $rev) = @_;
2672 return undef unless $self->follow_parent;
2673 unless (defined $paths) {
2674 my $err_handler = $SVN::Error::handler;
2675 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2676 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2677 sub { $paths = $_[0] });
2678 $SVN::Error::handler = $err_handler;
2679 }
2680 return undef unless defined $paths;
2681 # look for a parent from another branch:
2683 my @b_path_components = split m#/#, $self->{path};
2684 my @a_path_components;
2685 my $i;
2686 while (@b_path_components) {
2687 $i = $paths->{'/'.join('/', @b_path_components)};
2688 last if $i && defined $i->{copyfrom_path};
2689 unshift(@a_path_components, pop(@b_path_components));
2690 }
2691 return undef unless defined $i && defined $i->{copyfrom_path};
2692 my $branch_from = $i->{copyfrom_path};
2693 if (@a_path_components) {
2694 print STDERR "branch_from: $branch_from => ";
2695 $branch_from .= '/'.join('/', @a_path_components);
2696 print STDERR $branch_from, "\n";
2697 }
2698 my $r = $i->{copyfrom_rev};
2699 my $repos_root = $self->ra->{repos_root};
2700 my $url = $self->ra->{url};
2701 my $new_url = $url . $branch_from;
2702 print STDERR "Found possible branch point: ",
2703 "$new_url => ", $self->full_url, ", $r\n"
2704 unless $::_q > 1;
2705 $branch_from =~ s#^/##;
2706 my $gs = $self->other_gs($new_url, $url,
2707 $branch_from, $r, $self->{ref_id});
2708 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2709 {
2710 my ($base, $head);
2711 if (!defined $r0 || !defined $parent) {
2712 ($base, $head) = parse_revision_argument(0, $r);
2713 } else {
2714 if ($r0 < $r) {
2715 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2716 0, 1, sub { $base = $_[1] - 1 });
2717 }
2718 }
2719 if (defined $base && $base <= $r) {
2720 $gs->fetch($base, $r);
2721 }
2722 ($r0, $parent) = $gs->find_rev_before($r, 1);
2723 }
2724 if (defined $r0 && defined $parent) {
2725 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2726 unless $::_q > 1;
2727 my $ed;
2728 if ($self->ra->can_do_switch) {
2729 $self->assert_index_clean($parent);
2730 print STDERR "Following parent with do_switch\n"
2731 unless $::_q > 1;
2732 # do_switch works with svn/trunk >= r22312, but that
2733 # is not included with SVN 1.4.3 (the latest version
2734 # at the moment), so we can't rely on it
2735 $self->{last_rev} = $r0;
2736 $self->{last_commit} = $parent;
2737 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2738 $gs->ra->gs_do_switch($r0, $rev, $gs,
2739 $self->full_url, $ed)
2740 or die "SVN connection failed somewhere...\n";
2741 } elsif ($self->ra->trees_match($new_url, $r0,
2742 $self->full_url, $rev)) {
2743 print STDERR "Trees match:\n",
2744 " $new_url\@$r0\n",
2745 " ${\$self->full_url}\@$rev\n",
2746 "Following parent with no changes\n"
2747 unless $::_q > 1;
2748 $self->tmp_index_do(sub {
2749 command_noisy('read-tree', $parent);
2750 });
2751 $self->{last_commit} = $parent;
2752 } else {
2753 print STDERR "Following parent with do_update\n"
2754 unless $::_q > 1;
2755 $ed = SVN::Git::Fetcher->new($self);
2756 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2757 or die "SVN connection failed somewhere...\n";
2758 }
2759 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2760 return $self->make_log_entry($rev, [$parent], $ed);
2761 }
2762 return undef;
2763}
2764sub do_fetch {
2766 my ($self, $paths, $rev) = @_;
2767 my $ed;
2768 my ($last_rev, @parents);
2769 if (my $lc = $self->last_commit) {
2770 # we can have a branch that was deleted, then re-added
2771 # under the same name but copied from another path, in
2772 # which case we'll have multiple parents (we don't
2773 # want to break the original ref, nor lose copypath info):
2774 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2775 push @{$log_entry->{parents}}, $lc;
2776 return $log_entry;
2777 }
2778 $ed = SVN::Git::Fetcher->new($self);
2779 $last_rev = $self->{last_rev};
2780 $ed->{c} = $lc;
2781 @parents = ($lc);
2782 } else {
2783 $last_rev = $rev;
2784 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2785 return $log_entry;
2786 }
2787 $ed = SVN::Git::Fetcher->new($self);
2788 }
2789 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2790 die "SVN connection failed somewhere...\n";
2791 }
2792 $self->make_log_entry($rev, \@parents, $ed);
2793}
2794sub mkemptydirs {
2796 my ($self, $r) = @_;
2797 sub scan {
2799 my ($r, $empty_dirs, $line) = @_;
2800 if (defined $r && $line =~ /^r(\d+)$/) {
2801 return 0 if $1 > $r;
2802 } elsif ($line =~ /^ \+empty_dir: (.+)$/) {
2803 $empty_dirs->{$1} = 1;
2804 } elsif ($line =~ /^ \-empty_dir: (.+)$/) {
2805 my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
2806 delete @$empty_dirs{@d};
2807 }
2808 1; # continue
2809 };
2810 my %empty_dirs = ();
2812 my $gz_file = "$self->{dir}/unhandled.log.gz";
2813 if (-f $gz_file) {
2814 if (!$can_compress) {
2815 warn "Compress::Zlib could not be found; ",
2816 "empty directories in $gz_file will not be read\n";
2817 } else {
2818 my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
2819 die "Unable to open $gz_file: $!\n";
2820 my $line;
2821 while ($gz->gzreadline($line) > 0) {
2822 scan($r, \%empty_dirs, $line) or last;
2823 }
2824 $gz->gzclose;
2825 }
2826 }
2827 if (open my $fh, '<', "$self->{dir}/unhandled.log") {
2829 binmode $fh or croak "binmode: $!";
2830 while (<$fh>) {
2831 scan($r, \%empty_dirs, $_) or last;
2832 }
2833 close $fh;
2834 }
2835 my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2837 foreach my $d (sort keys %empty_dirs) {
2838 $d = uri_decode($d);
2839 $d =~ s/$strip//;
2840 next unless length($d);
2841 next if -d $d;
2842 if (-e $d) {
2843 warn "$d exists but is not a directory\n";
2844 } else {
2845 print "creating empty directory: $d\n";
2846 mkpath([$d]);
2847 }
2848 }
2849}
2850sub get_untracked {
2852 my ($self, $ed) = @_;
2853 my @out;
2854 my $h = $ed->{empty};
2855 foreach (sort keys %$h) {
2856 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2857 push @out, " $act: " . uri_encode($_);
2858 warn "W: $act: $_\n";
2859 }
2860 foreach my $t (qw/dir_prop file_prop/) {
2861 $h = $ed->{$t} or next;
2862 foreach my $path (sort keys %$h) {
2863 my $ppath = $path eq '' ? '.' : $path;
2864 foreach my $prop (sort keys %{$h->{$path}}) {
2865 next if $SKIP_PROP{$prop};
2866 my $v = $h->{$path}->{$prop};
2867 my $t_ppath_prop = "$t: " .
2868 uri_encode($ppath) . ' ' .
2869 uri_encode($prop);
2870 if (defined $v) {
2871 push @out, " +$t_ppath_prop " .
2872 uri_encode($v);
2873 } else {
2874 push @out, " -$t_ppath_prop";
2875 }
2876 }
2877 }
2878 }
2879 foreach my $t (qw/absent_file absent_directory/) {
2880 $h = $ed->{$t} or next;
2881 foreach my $parent (sort keys %$h) {
2882 foreach my $path (sort @{$h->{$parent}}) {
2883 push @out, " $t: " .
2884 uri_encode("$parent/$path");
2885 warn "W: $t: $parent/$path ",
2886 "Insufficient permissions?\n";
2887 }
2888 }
2889 }
2890 \@out;
2891}
2892# parse_svn_date(DATE)
2894# --------------------
2895# Given a date (in UTC) from Subversion, return a string in the format
2896# "<TZ Offset> <local date/time>" that Git will use.
2897#
2898# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2899# is true we'll convert it to the local timezone instead.
2900sub parse_svn_date {
2901 my $date = shift || return '+0000 1970-01-01 00:00:00';
2902 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2903 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2904 croak "Unable to parse date: $date\n";
2905 my $parsed_date; # Set next.
2906 if ($Git::SVN::_localtime) {
2908 # Translate the Subversion datetime to an epoch time.
2909 # Begin by switching ourselves to $date's timezone, UTC.
2910 my $old_env_TZ = $ENV{TZ};
2911 $ENV{TZ} = 'UTC';
2912 my $epoch_in_UTC =
2914 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2915 # Determine our local timezone (including DST) at the
2917 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2918 # value of TZ, if any, at the time we were run.
2919 if (defined $Git::SVN::Log::TZ) {
2920 $ENV{TZ} = $Git::SVN::Log::TZ;
2921 } else {
2922 delete $ENV{TZ};
2923 }
2924 my $our_TZ =
2926 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2927 # This converts $epoch_in_UTC into our local timezone.
2929 my ($sec, $min, $hour, $mday, $mon, $year,
2930 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2931 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2933 $our_TZ, $year + 1900, $mon + 1,
2934 $mday, $hour, $min, $sec);
2935 # Reset us to the timezone in effect when we entered
2937 # this routine.
2938 if (defined $old_env_TZ) {
2939 $ENV{TZ} = $old_env_TZ;
2940 } else {
2941 delete $ENV{TZ};
2942 }
2943 } else {
2944 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2945 }
2946 return $parsed_date;
2948}
2949sub other_gs {
2951 my ($self, $new_url, $url,
2952 $branch_from, $r, $old_ref_id) = @_;
2953 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2954 unless ($gs) {
2955 my $ref_id = $old_ref_id;
2956 $ref_id =~ s/\@\d+$//;
2957 $ref_id .= "\@$r";
2958 # just grow a tail if we're not unique enough :x
2959 $ref_id .= '-' while find_ref($ref_id);
2960 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2961 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2962 if ($u =~ s#^\Q$url\E(/|$)##) {
2963 $p = $u;
2964 $u = $url;
2965 $repo_id = $self->{repo_id};
2966 }
2967 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2968 }
2969 $gs
2970}
2971sub call_authors_prog {
2973 my ($orig_author) = @_;
2974 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2975 my $author = `$::_authors_prog $orig_author`;
2976 if ($? != 0) {
2977 die "$::_authors_prog failed with exit code $?\n"
2978 }
2979 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2980 my ($name, $email) = ($1, $2);
2981 $email = undef if length $2 == 0;
2982 return [$name, $email];
2983 } else {
2984 die "Author: $orig_author: $::_authors_prog returned "
2985 . "invalid author format: $author\n";
2986 }
2987}
2988sub check_author {
2990 my ($author) = @_;
2991 if (!defined $author || length $author == 0) {
2992 $author = '(no author)';
2993 }
2994 if (!defined $::users{$author}) {
2995 if (defined $::_authors_prog) {
2996 $::users{$author} = call_authors_prog($author);
2997 } elsif (defined $::_authors) {
2998 die "Author: $author not defined in $::_authors file\n";
2999 }
3000 }
3001 $author;
3002}
3003sub find_extra_svk_parents {
3005 my ($self, $ed, $tickets, $parents) = @_;
3006 # aha! svk:merge property changed...
3007 my @tickets = split "\n", $tickets;
3008 my @known_parents;
3009 for my $ticket ( @tickets ) {
3010 my ($uuid, $path, $rev) = split /:/, $ticket;
3011 if ( $uuid eq $self->ra_uuid ) {
3012 my $url = $self->{url};
3013 my $repos_root = $url;
3014 my $branch_from = $path;
3015 $branch_from =~ s{^/}{};
3016 my $gs = $self->other_gs($repos_root."/".$branch_from,
3017 $url,
3018 $branch_from,
3019 $rev,
3020 $self->{ref_id});
3021 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3022 # wahey! we found it, but it might be
3023 # an old one (!)
3024 push @known_parents, [ $rev, $commit ];
3025 }
3026 }
3027 }
3028 # Ordering matters; highest-numbered commit merge tickets
3029 # first, as they may account for later merge ticket additions
3030 # or changes.
3031 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3032 for my $parent ( @known_parents ) {
3033 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3034 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3035 my $new;
3036 while ( <$msg_fh> ) {
3037 $new=1;last;
3038 }
3039 command_close_pipe($msg_fh, $ctx);
3040 if ( $new ) {
3041 print STDERR
3042 "Found merge parent (svk:merge ticket): $parent\n";
3043 push @$parents, $parent;
3044 }
3045 }
3046}
3047sub lookup_svn_merge {
3049 my $uuid = shift;
3050 my $url = shift;
3051 my $merge = shift;
3052 my ($source, $revs) = split ":", $merge;
3054 my $path = $source;
3055 $path =~ s{^/}{};
3056 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3057 if ( !$gs ) {
3058 warn "Couldn't find revmap for $url$source\n";
3059 return;
3060 }
3061 my @ranges = split ",", $revs;
3062 my ($tip, $tip_commit);
3063 my @merged_commit_ranges;
3064 # find the tip
3065 for my $range ( @ranges ) {
3066 my ($bottom, $top) = split "-", $range;
3067 $top ||= $bottom;
3068 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3069 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3070 unless ($top_commit and $bottom_commit) {
3072 warn "W:unknown path/rev in svn:mergeinfo "
3073 ."dirprop: $source:$range\n";
3074 next;
3075 }
3076 push @merged_commit_ranges,
3078 "$bottom_commit^..$top_commit";
3079 if ( !defined $tip or $top > $tip ) {
3081 $tip = $top;
3082 $tip_commit = $top_commit;
3083 }
3084 }
3085 return ($tip_commit, @merged_commit_ranges);
3086}
3087sub _rev_list {
3089 my ($msg_fh, $ctx) = command_output_pipe(
3090 "rev-list", @_,
3091 );
3092 my @rv;
3093 while ( <$msg_fh> ) {
3094 chomp;
3095 push @rv, $_;
3096 }
3097 command_close_pipe($msg_fh, $ctx);
3098 @rv;
3099}
3100sub check_cherry_pick {
3102 my $base = shift;
3103 my $tip = shift;
3104 my @ranges = @_;
3105 my %commits = map { $_ => 1 }
3106 _rev_list("--no-merges", $tip, "--not", $base);
3107 for my $range ( @ranges ) {
3108 delete @commits{_rev_list($range)};
3109 }
3110 for my $commit (keys %commits) {
3111 if (has_no_changes($commit)) {
3112 delete $commits{$commit};
3113 }
3114 }
3115 return (keys %commits);
3116}
3117sub has_no_changes {
3119 my $commit = shift;
3120 my @revs = split / /, command_oneline(
3122 qw(rev-list --parents -1 -m), $commit);
3123 # Commits with no parents, e.g. the start of a partial branch,
3125 # have changes by definition.
3126 return 1 if (@revs < 2);
3127 # Commits with multiple parents, e.g a merge, have no changes
3129 # by definition.
3130 return 0 if (@revs > 2);
3131 return (command_oneline("rev-parse", "$commit^{tree}") eq
3133 command_oneline("rev-parse", "$commit~1^{tree}"));
3134}
3135# The GIT_DIR environment variable is not always set until after the command
3137# line arguments are processed, so we can't memoize in a BEGIN block.
3138{
3139 my $memoized = 0;
3140 sub memoize_svn_mergeinfo_functions {
3142 return if $memoized;
3143 $memoized = 1;
3144 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3146 mkpath([$cache_path]) unless -d $cache_path;
3147 tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3149 "$cache_path/lookup_svn_merge.db", 'nstore';
3150 memoize 'lookup_svn_merge',
3151 SCALAR_CACHE => 'FAULT',
3152 LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3153 ;
3154 tie my %check_cherry_pick_cache => 'Memoize::Storable',
3156 "$cache_path/check_cherry_pick.db", 'nstore';
3157 memoize 'check_cherry_pick',
3158 SCALAR_CACHE => 'FAULT',
3159 LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3160 ;
3161 tie my %has_no_changes_cache => 'Memoize::Storable',
3163 "$cache_path/has_no_changes.db", 'nstore';
3164 memoize 'has_no_changes',
3165 SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3166 LIST_CACHE => 'FAULT',
3167 ;
3168 }
3169}
3170sub parents_exclude {
3172 my $parents = shift;
3173 my @commits = @_;
3174 return unless @commits;
3175 my @excluded;
3177 my $excluded;
3178 do {
3179 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3180 $excluded = command_oneline(@cmd);
3181 if ( $excluded ) {
3182 my @new;
3183 my $found;
3184 for my $commit ( @commits ) {
3185 if ( $commit eq $excluded ) {
3186 push @excluded, $commit;
3187 $found++;
3188 last;
3189 }
3190 else {
3191 push @new, $commit;
3192 }
3193 }
3194 die "saw commit '$excluded' in rev-list output, "
3195 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3196 unless $found;
3197 @commits = @new;
3198 }
3199 }
3200 while ($excluded and @commits);
3201 return @excluded;
3203}
3204# note: this function should only be called if the various dirprops
3207# have actually changed
3208sub find_extra_svn_parents {
3209 my ($self, $ed, $mergeinfo, $parents) = @_;
3210 # aha! svk:merge property changed...
3211 memoize_svn_mergeinfo_functions();
3213 # We first search for merged tips which are not in our
3215 # history. Then, we figure out which git revisions are in
3216 # that tip, but not this revision. If all of those revisions
3217 # are now marked as merge, we can add the tip as a parent.
3218 my @merges = split "\n", $mergeinfo;
3219 my @merge_tips;
3220 my $url = $self->{url};
3221 my $uuid = $self->ra_uuid;
3222 my %ranges;
3223 for my $merge ( @merges ) {
3224 my ($tip_commit, @ranges) =
3225 lookup_svn_merge( $uuid, $url, $merge );
3226 unless (!$tip_commit or
3227 grep { $_ eq $tip_commit } @$parents ) {
3228 push @merge_tips, $tip_commit;
3229 $ranges{$tip_commit} = \@ranges;
3230 } else {
3231 push @merge_tips, undef;
3232 }
3233 }
3234 my %excluded = map { $_ => 1 }
3236 parents_exclude($parents, grep { defined } @merge_tips);
3237 # check merge tips for new parents
3239 my @new_parents;
3240 for my $merge_tip ( @merge_tips ) {
3241 my $spec = shift @merges;
3242 next unless $merge_tip and $excluded{$merge_tip};
3243 my $ranges = $ranges{$merge_tip};
3245 # check out 'new' tips
3247 my $merge_base;
3248 eval {
3249 $merge_base = command_oneline(
3250 "merge-base",
3251 @$parents, $merge_tip,
3252 );
3253 };
3254 if ($@) {
3255 die "An error occurred during merge-base"
3256 unless $@->isa("Git::Error::Command");
3257 warn "W: Cannot find common ancestor between ".
3259 "@$parents and $merge_tip. Ignoring merge info.\n";
3260 next;
3261 }
3262 # double check that there are no missing non-merge commits
3264 my (@incomplete) = check_cherry_pick(
3265 $merge_base, $merge_tip,
3266 @$ranges,
3267 );
3268 if ( @incomplete ) {
3270 warn "W:svn cherry-pick ignored ($spec) - missing "
3271 .@incomplete." commit(s) (eg $incomplete[0])\n";
3272 } else {
3273 warn
3274 "Found merge parent (svn:mergeinfo prop): ",
3275 $merge_tip, "\n";
3276 push @new_parents, $merge_tip;
3277 }
3278 }
3279 # cater for merges which merge commits from multiple branches
3281 if ( @new_parents > 1 ) {
3282 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3283 for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3284 next if $i == $j;
3285 next unless $new_parents[$i];
3286 next unless $new_parents[$j];
3287 my $revs = command_oneline(
3288 "rev-list", "-1",
3289 "$new_parents[$i]..$new_parents[$j]",
3290 );
3291 if ( !$revs ) {
3292 undef($new_parents[$j]);
3293 }
3294 }
3295 }
3296 }
3297 push @$parents, grep { defined } @new_parents;
3298}
3299sub make_log_entry {
3301 my ($self, $rev, $parents, $ed) = @_;
3302 my $untracked = $self->get_untracked($ed);
3303 my @parents = @$parents;
3305 my $ps = $ed->{path_strip} || "";
3306 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3307 my $props = $ed->{dir_prop}{$path};
3308 if ( $props->{"svk:merge"} ) {
3309 $self->find_extra_svk_parents
3310 ($ed, $props->{"svk:merge"}, \@parents);
3311 }
3312 if ( $props->{"svn:mergeinfo"} ) {
3313 $self->find_extra_svn_parents
3314 ($ed,
3315 $props->{"svn:mergeinfo"},
3316 \@parents);
3317 }
3318 }
3319 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3321 print $un "r$rev\n" or croak $!;
3322 print $un $_, "\n" foreach @$untracked;
3323 my %log_entry = ( parents => \@parents, revision => $rev,
3324 log => '');
3325 my $headrev;
3327 my $logged = delete $self->{logged_rev_props};
3328 if (!$logged || $self->{-want_revprops}) {
3329 my $rp = $self->ra->rev_proplist($rev);
3330 foreach (sort keys %$rp) {
3331 my $v = $rp->{$_};
3332 if (/^svn:(author|date|log)$/) {
3333 $log_entry{$1} = $v;
3334 } elsif ($_ eq 'svm:headrev') {
3335 $headrev = $v;
3336 } else {
3337 print $un " rev_prop: ", uri_encode($_), ' ',
3338 uri_encode($v), "\n";
3339 }
3340 }
3341 } else {
3342 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3343 }
3344 close $un or croak $!;
3345 $log_entry{date} = parse_svn_date($log_entry{date});
3347 $log_entry{log} .= "\n";
3348 my $author = $log_entry{author} = check_author($log_entry{author});
3349 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3350 : ($author, undef);
3351 my ($commit_name, $commit_email) = ($name, $email);
3353 if ($_use_log_author) {
3354 my $name_field;
3355 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3356 $name_field = $1;
3357 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3358 $name_field = $1;
3359 }
3360 if (!defined $name_field) {
3361 if (!defined $email) {
3362 $email = $name;
3363 }
3364 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3365 ($name, $email) = ($1, $2);
3366 } elsif ($name_field =~ /(.*)@/) {
3367 ($name, $email) = ($1, $name_field);
3368 } else {
3369 ($name, $email) = ($name_field, $name_field);
3370 }
3371 }
3372 if (defined $headrev && $self->use_svm_props) {
3373 if ($self->rewrite_root) {
3374 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3375 "options set!\n";
3376 }
3377 if ($self->rewrite_uuid) {
3378 die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3379 "options set!\n";
3380 }
3381 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3382 # we don't want "SVM: initializing mirror for junk" ...
3383 return undef if $r == 0;
3384 my $svm = $self->svm;
3385 if ($uuid ne $svm->{uuid}) {
3386 die "UUID mismatch on SVM path:\n",
3387 "expected: $svm->{uuid}\n",
3388 " got: $uuid\n";
3389 }
3390 my $full_url = $self->full_url;
3391 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3392 die "Failed to replace '$svm->{replace}' with ",
3393 "'$svm->{source}' in $full_url\n";
3394 # throw away username for storing in records
3395 remove_username($full_url);
3396 $log_entry{metadata} = "$full_url\@$r $uuid";
3397 $log_entry{svm_revision} = $r;
3398 $email ||= "$author\@$uuid";
3399 $commit_email ||= "$author\@$uuid";
3400 } elsif ($self->use_svnsync_props) {
3401 my $full_url = $self->svnsync->{url};
3402 $full_url .= "/$self->{path}" if length $self->{path};
3403 remove_username($full_url);
3404 my $uuid = $self->svnsync->{uuid};
3405 $log_entry{metadata} = "$full_url\@$rev $uuid";
3406 $email ||= "$author\@$uuid";
3407 $commit_email ||= "$author\@$uuid";
3408 } else {
3409 my $url = $self->metadata_url;
3410 remove_username($url);
3411 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3412 $log_entry{metadata} = "$url\@$rev " . $uuid;
3413 $email ||= "$author\@" . $uuid;
3414 $commit_email ||= "$author\@" . $uuid;
3415 }
3416 $log_entry{name} = $name;
3417 $log_entry{email} = $email;
3418 $log_entry{commit_name} = $commit_name;
3419 $log_entry{commit_email} = $commit_email;
3420 \%log_entry;
3421}
3422sub fetch {
3424 my ($self, $min_rev, $max_rev, @parents) = @_;
3425 my ($last_rev, $last_commit) = $self->last_rev_commit;
3426 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3427 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3428}
3429sub set_tree_cb {
3431 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3432 $self->{inject_parents} = { $rev => $tree };
3433 $self->fetch(undef, undef);
3434}
3435sub set_tree {
3437 my ($self, $tree) = (shift, shift);
3438 my $log_entry = ::get_commit_entry($tree);
3439 unless ($self->{last_rev}) {
3440 ::fatal("Must have an existing revision to commit");
3441 }
3442 my %ed_opts = ( r => $self->{last_rev},
3443 log => $log_entry->{log},
3444 ra => $self->ra,
3445 tree_a => $self->{last_commit},
3446 tree_b => $tree,
3447 editor_cb => sub {
3448 $self->set_tree_cb($log_entry, $tree, @_) },
3449 svn_path => $self->{path} );
3450 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3451 print "No changes\nr$self->{last_rev} = $tree\n";
3452 }
3453}
3454sub rebuild_from_rev_db {
3456 my ($self, $path) = @_;
3457 my $r = -1;
3458 open my $fh, '<', $path or croak "open: $!";
3459 binmode $fh or croak "binmode: $!";
3460 while (<$fh>) {
3461 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3462 chomp($_);
3463 ++$r;
3464 next if $_ eq ('0' x 40);
3465 $self->rev_map_set($r, $_);
3466 print "r$r = $_\n";
3467 }
3468 close $fh or croak "close: $!";
3469 unlink $path or croak "unlink: $!";
3470}
3471sub rebuild {
3473 my ($self) = @_;
3474 my $map_path = $self->map_path;
3475 my $partial = (-e $map_path && ! -z $map_path);
3476 return unless ::verify_ref($self->refname.'^0');
3477 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3478 my $rev_db = $self->rev_db_path;
3479 $self->rebuild_from_rev_db($rev_db);
3480 if ($self->use_svm_props) {
3481 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3482 $self->rebuild_from_rev_db($svm_rev_db);
3483 }
3484 $self->unlink_rev_db_symlink;
3485 return;
3486 }
3487 print "Rebuilding $map_path ...\n" if (!$partial);
3488 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3489 (undef, undef));
3490 my ($log, $ctx) =
3491 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3492 ($head ? "$head.." : "") . $self->refname,
3493 '--');
3494 my $metadata_url = $self->metadata_url;
3495 remove_username($metadata_url);
3496 my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3497 my $c;
3498 while (<$log>) {
3499 if ( m{^commit ($::sha1)$} ) {
3500 $c = $1;
3501 next;
3502 }
3503 next unless s{^\s*(git-svn-id:)}{$1};
3504 my ($url, $rev, $uuid) = ::extract_metadata($_);
3505 remove_username($url);
3506 # ignore merges (from set-tree)
3508 next if (!defined $rev || !$uuid);
3509 # if we merged or otherwise started elsewhere, this is
3511 # how we break out of it
3512 if (($uuid ne $svn_uuid) ||
3513 ($metadata_url && $url && ($url ne $metadata_url))) {
3514 next;
3515 }
3516 if ($partial && $head) {
3517 print "Partial-rebuilding $map_path ...\n";
3518 print "Currently at $base_rev = $head\n";
3519 $head = undef;
3520 }
3521 $self->rev_map_set($rev, $c);
3523 print "r$rev = $c\n";
3524 }
3525 command_close_pipe($log, $ctx);
3526 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3527 my $rev_db_path = $self->rev_db_path;
3528 if (-f $self->rev_db_path) {
3529 unlink $self->rev_db_path or croak "unlink: $!";
3530 }
3531 $self->unlink_rev_db_symlink;
3532}
3533# rev_map:
3535# Tie::File seems to be prone to offset errors if revisions get sparse,
3536# it's not that fast, either. Tie::File is also not in Perl 5.6. So
3537# one of my favorite modules is out :< Next up would be one of the DBM
3538# modules, but I'm not sure which is most portable...
3539#
3540# This is the replacement for the rev_db format, which was too big
3541# and inefficient for large repositories with a lot of sparse history
3542# (mainly tags)
3543#
3544# The format is this:
3545# - 24 bytes for every record,
3546# * 4 bytes for the integer representing an SVN revision number
3547# * 20 bytes representing the sha1 of a git commit
3548# - No empty padding records like the old format
3549# (except the last record, which can be overwritten)
3550# - new records are written append-only since SVN revision numbers
3551# increase monotonically
3552# - lookups on SVN revision number are done via a binary search
3553# - Piping the file to xxd -c24 is a good way of dumping it for
3554# viewing or editing (piped back through xxd -r), should the need
3555# ever arise.
3556# - The last record can be padding revision with an all-zero sha1
3557# This is used to optimize fetch performance when using multiple
3558# "fetch" directives in .git/config
3559#
3560# These files are disposable unless noMetadata or useSvmProps is set
3561sub _rev_map_set {
3563 my ($fh, $rev, $commit) = @_;
3564 binmode $fh or croak "binmode: $!";
3566 my $size = (stat($fh))[7];
3567 ($size % 24) == 0 or croak "inconsistent size: $size";
3568 my $wr_offset = 0;
3570 if ($size > 0) {
3571 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3572 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3573 $read == 24 or croak "read only $read bytes (!= 24)";
3574 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3575 if ($last_commit eq ('0' x40)) {
3576 if ($size >= 48) {
3577 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3578 $read = sysread($fh, $buf, 24) or
3579 croak "read: $!";
3580 $read == 24 or
3581 croak "read only $read bytes (!= 24)";
3582 ($last_rev, $last_commit) =
3583 unpack(rev_map_fmt, $buf);
3584 if ($last_commit eq ('0' x40)) {
3585 croak "inconsistent .rev_map\n";
3586 }
3587 }
3588 if ($last_rev >= $rev) {
3589 croak "last_rev is higher!: $last_rev >= $rev";
3590 }
3591 $wr_offset = -24;
3592 }
3593 }
3594 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3595 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3596 croak "write: $!";
3597}
3598sub _rev_map_reset {
3600 my ($fh, $rev, $commit) = @_;
3601 my $c = _rev_map_get($fh, $rev);
3602 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3603 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3604 truncate $fh, $offset or croak "truncate: $!";
3605}
3606sub mkfile {
3608 my ($path) = @_;
3609 unless (-e $path) {
3610 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3611 mkpath([$dir]) unless -d $dir;
3612 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3613 close $fh or die "Couldn't close (create) $path: $!\n";
3614 }
3615}
3616sub rev_map_set {
3618 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3619 defined $commit or die "missing arg3\n";
3620 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3621 my $db = $self->map_path($uuid);
3622 my $db_lock = "$db.lock";
3623 my $sig;
3624 $update_ref ||= 0;
3625 if ($update_ref) {
3626 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3627 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3628 }
3629 mkfile($db);
3630 $LOCKFILES{$db_lock} = 1;
3632 my $sync;
3633 # both of these options make our .rev_db file very, very important
3634 # and we can't afford to lose it because rebuild() won't work
3635 if ($self->use_svm_props || $self->no_metadata) {
3636 $sync = 1;
3637 copy($db, $db_lock) or die "rev_map_set(@_): ",
3638 "Failed to copy: ",
3639 "$db => $db_lock ($!)\n";
3640 } else {
3641 rename $db, $db_lock or die "rev_map_set(@_): ",
3642 "Failed to rename: ",
3643 "$db => $db_lock ($!)\n";
3644 }
3645 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3647 or croak "Couldn't open $db_lock: $!\n";
3648 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3649 _rev_map_set($fh, $rev, $commit);
3650 if ($sync) {
3651 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3652 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3653 }
3654 close $fh or croak $!;
3655 if ($update_ref) {
3656 $_head = $self;
3657 my $note = "";
3658 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3659 command_noisy('update-ref', '-m', "r$rev$note",
3660 $self->refname, $commit);
3661 }
3662 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3663 "$db_lock => $db ($!)\n";
3664 delete $LOCKFILES{$db_lock};
3665 if ($update_ref) {
3666 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3667 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3668 kill $sig, $$ if defined $sig;
3669 }
3670}
3671# If want_commit, this will return an array of (rev, commit) where
3673# commit _must_ be a valid commit in the archive.
3674# Otherwise, it'll return the max revision (whether or not the
3675# commit is valid or just a 0x40 placeholder).
3676sub rev_map_max {
3677 my ($self, $want_commit) = @_;
3678 $self->rebuild;
3679 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3680 $want_commit ? ($r, $c) : $r;
3681}
3682sub rev_map_max_norebuild {
3684 my ($self, $want_commit) = @_;
3685 my $map_path = $self->map_path;
3686 stat $map_path or return $want_commit ? (0, undef) : 0;
3687 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3688 binmode $fh or croak "binmode: $!";
3689 my $size = (stat($fh))[7];
3690 ($size % 24) == 0 or croak "inconsistent size: $size";
3691 if ($size == 0) {
3693 close $fh or croak "close: $!";
3694 return $want_commit ? (0, undef) : 0;
3695 }
3696 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3698 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3699 my ($r, $c) = unpack(rev_map_fmt, $buf);
3700 if ($want_commit && $c eq ('0' x40)) {
3701 if ($size < 48) {
3702 return $want_commit ? (0, undef) : 0;
3703 }
3704 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3705 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3706 ($r, $c) = unpack(rev_map_fmt, $buf);
3707 if ($c eq ('0'x40)) {
3708 croak "Penultimate record is all-zeroes in $map_path";
3709 }
3710 }
3711 close $fh or croak "close: $!";
3712 $want_commit ? ($r, $c) : $r;
3713}
3714sub rev_map_get {
3716 my ($self, $rev, $uuid) = @_;
3717 my $map_path = $self->map_path($uuid);
3718 return undef unless -e $map_path;
3719 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3721 my $c = _rev_map_get($fh, $rev);
3722 close($fh) or croak "close: $!";
3723 $c
3724}
3725sub _rev_map_get {
3727 my ($fh, $rev) = @_;
3728 binmode $fh or croak "binmode: $!";
3730 my $size = (stat($fh))[7];
3731 ($size % 24) == 0 or croak "inconsistent size: $size";
3732 if ($size == 0) {
3734 return undef;
3735 }
3736 my ($l, $u) = (0, $size - 24);
3738 my ($r, $c, $buf);
3739 while ($l <= $u) {
3741 my $i = int(($l/24 + $u/24) / 2) * 24;
3742 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3743 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3744 my ($r, $c) = unpack(rev_map_fmt, $buf);
3745 if ($r < $rev) {
3747 $l = $i + 24;
3748 } elsif ($r > $rev) {
3749 $u = $i - 24;
3750 } else { # $r == $rev
3751 return $c eq ('0' x 40) ? undef : $c;
3752 }
3753 }
3754 undef;
3755}
3756# Finds the first svn revision that exists on (if $eq_ok is true) or
3758# before $rev for the current branch. It will not search any lower
3759# than $min_rev. Returns the git commit hash and svn revision number
3760# if found, else (undef, undef).
3761sub find_rev_before {
3762 my ($self, $rev, $eq_ok, $min_rev) = @_;
3763 --$rev unless $eq_ok;
3764 $min_rev ||= 1;
3765 my $max_rev = $self->rev_map_max;
3766 $rev = $max_rev if ($rev > $max_rev);
3767 while ($rev >= $min_rev) {
3768 if (my $c = $self->rev_map_get($rev)) {
3769 return ($rev, $c);
3770 }
3771 --$rev;
3772 }
3773 return (undef, undef);
3774}
3775# Finds the first svn revision that exists on (if $eq_ok is true) or
3777# after $rev for the current branch. It will not search any higher
3778# than $max_rev. Returns the git commit hash and svn revision number
3779# if found, else (undef, undef).
3780sub find_rev_after {
3781 my ($self, $rev, $eq_ok, $max_rev) = @_;
3782 ++$rev unless $eq_ok;
3783 $max_rev ||= $self->rev_map_max;
3784 while ($rev <= $max_rev) {
3785 if (my $c = $self->rev_map_get($rev)) {
3786 return ($rev, $c);
3787 }
3788 ++$rev;
3789 }
3790 return (undef, undef);
3791}
3792sub _new {
3794 my ($class, $repo_id, $ref_id, $path) = @_;
3795 unless (defined $repo_id && length $repo_id) {
3796 $repo_id = $Git::SVN::default_repo_id;
3797 }
3798 unless (defined $ref_id && length $ref_id) {
3799 $_prefix = '' unless defined($_prefix);
3800 $_[2] = $ref_id =
3801 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3802 }
3803 $_[1] = $repo_id;
3804 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3805 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3807 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3808 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3809 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3810 if (-d $old_dir && ! -d $dir) {
3811 $dir = $old_dir;
3812 }
3813 }
3814 $_[3] = $path = '' unless (defined $path);
3816 mkpath([$dir]);
3817 bless {
3818 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3819 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3820 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3821}
3822# for read-only access of old .rev_db formats
3824sub unlink_rev_db_symlink {
3825 my ($self) = @_;
3826 my $link = $self->rev_db_path;
3827 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3828 if (-l $link) {
3829 unlink $link or croak "unlink: $link failed!";
3830 }
3831}
3832sub rev_db_path {
3834 my ($self, $uuid) = @_;
3835 my $db_path = $self->map_path($uuid);
3836 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3837 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3838 $db_path;
3839}
3840# the new replacement for .rev_db
3842sub map_path {
3843 my ($self, $uuid) = @_;
3844 $uuid ||= $self->ra_uuid;
3845 "$self->{map_root}.$uuid";
3846}
3847sub uri_encode {
3849 my ($f) = @_;
3850 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3851 $f
3852}
3853sub uri_decode {
3855 my ($f) = @_;
3856 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3857 $f
3858}
3859sub remove_username {
3861 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3862}
3863package Git::SVN::Prompt;
3865use strict;
3866use warnings;
3867require SVN::Core;
3868use vars qw/$_no_auth_cache $_username/;
3869sub simple {
3871 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3872 $may_save = undef if $_no_auth_cache;
3873 $default_username = $_username if defined $_username;
3874 if (defined $default_username && length $default_username) {
3875 if (defined $realm && length $realm) {
3876 print STDERR "Authentication realm: $realm\n";
3877 STDERR->flush;
3878 }
3879 $cred->username($default_username);
3880 } else {
3881 username($cred, $realm, $may_save, $pool);
3882 }
3883 $cred->password(_read_password("Password for '" .
3884 $cred->username . "': ", $realm));
3885 $cred->may_save($may_save);
3886 $SVN::_Core::SVN_NO_ERROR;
3887}
3888sub ssl_server_trust {
3890 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3891 $may_save = undef if $_no_auth_cache;
3892 print STDERR "Error validating server certificate for '$realm':\n";
3893 {
3894 no warnings 'once';
3895 # All variables SVN::Auth::SSL::* are used only once,
3896 # so we're shutting up Perl warnings about this.
3897 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3898 print STDERR " - The certificate is not issued ",
3899 "by a trusted authority. Use the\n",
3900 " fingerprint to validate ",
3901 "the certificate manually!\n";
3902 }
3903 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3904 print STDERR " - The certificate hostname ",
3905 "does not match.\n";
3906 }
3907 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3908 print STDERR " - The certificate is not yet valid.\n";
3909 }
3910 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3911 print STDERR " - The certificate has expired.\n";
3912 }
3913 if ($failures & $SVN::Auth::SSL::OTHER) {
3914 print STDERR " - The certificate has ",
3915 "an unknown error.\n";
3916 }
3917 } # no warnings 'once'
3918 printf STDERR
3919 "Certificate information:\n".
3920 " - Hostname: %s\n".
3921 " - Valid: from %s until %s\n".
3922 " - Issuer: %s\n".
3923 " - Fingerprint: %s\n",
3924 map $cert_info->$_, qw(hostname valid_from valid_until
3925 issuer_dname fingerprint);
3926 my $choice;
3927prompt:
3928 print STDERR $may_save ?
3929 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3930 "(R)eject or accept (t)emporarily? ";
3931 STDERR->flush;
3932 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3933 if ($choice =~ /^t$/i) {
3934 $cred->may_save(undef);
3935 } elsif ($choice =~ /^r$/i) {
3936 return -1;
3937 } elsif ($may_save && $choice =~ /^p$/i) {
3938 $cred->may_save($may_save);
3939 } else {
3940 goto prompt;
3941 }
3942 $cred->accepted_failures($failures);
3943 $SVN::_Core::SVN_NO_ERROR;
3944}
3945sub ssl_client_cert {
3947 my ($cred, $realm, $may_save, $pool) = @_;
3948 $may_save = undef if $_no_auth_cache;
3949 print STDERR "Client certificate filename: ";
3950 STDERR->flush;
3951 chomp(my $filename = <STDIN>);
3952 $cred->cert_file($filename);
3953 $cred->may_save($may_save);
3954 $SVN::_Core::SVN_NO_ERROR;
3955}
3956sub ssl_client_cert_pw {
3958 my ($cred, $realm, $may_save, $pool) = @_;
3959 $may_save = undef if $_no_auth_cache;
3960 $cred->password(_read_password("Password: ", $realm));
3961 $cred->may_save($may_save);
3962 $SVN::_Core::SVN_NO_ERROR;
3963}
3964sub username {
3966 my ($cred, $realm, $may_save, $pool) = @_;
3967 $may_save = undef if $_no_auth_cache;
3968 if (defined $realm && length $realm) {
3969 print STDERR "Authentication realm: $realm\n";
3970 }
3971 my $username;
3972 if (defined $_username) {
3973 $username = $_username;
3974 } else {
3975 print STDERR "Username: ";
3976 STDERR->flush;
3977 chomp($username = <STDIN>);
3978 }
3979 $cred->username($username);
3980 $cred->may_save($may_save);
3981 $SVN::_Core::SVN_NO_ERROR;
3982}
3983sub _read_password {
3985 my ($prompt, $realm) = @_;
3986 my $password = '';
3987 if (exists $ENV{GIT_ASKPASS}) {
3988 open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
3989 $password = <PH>;
3990 $password =~ s/[\012\015]//; # \n\r
3991 close(PH);
3992 } else {
3993 print STDERR $prompt;
3994 STDERR->flush;
3995 require Term::ReadKey;
3996 Term::ReadKey::ReadMode('noecho');
3997 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3998 last if $key =~ /[\012\015]/; # \n\r
3999 $password .= $key;
4000 }
4001 Term::ReadKey::ReadMode('restore');
4002 print STDERR "\n";
4003 STDERR->flush;
4004 }
4005 $password;
4006}
4007package SVN::Git::Fetcher;
4009use vars qw/@ISA/;
4010use strict;
4011use warnings;
4012use Carp qw/croak/;
4013use IO::File qw//;
4014use vars qw/$_ignore_regex/;
4015# file baton members: path, mode_a, mode_b, pool, fh, blob, base
4017sub new {
4018 my ($class, $git_svn, $switch_path) = @_;
4019 my $self = SVN::Delta::Editor->new;
4020 bless $self, $class;
4021 if (exists $git_svn->{last_commit}) {
4022 $self->{c} = $git_svn->{last_commit};
4023 $self->{empty_symlinks} =
4024 _mark_empty_symlinks($git_svn, $switch_path);
4025 }
4026 $self->{ignore_regex} = eval { command_oneline('config', '--get',
4027 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
4028 $self->{empty} = {};
4029 $self->{dir_prop} = {};
4030 $self->{file_prop} = {};
4031 $self->{absent_dir} = {};
4032 $self->{absent_file} = {};
4033 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4034 $self;
4035}
4036# this uses the Ra object, so it must be called before do_{switch,update},
4038# not inside them (when the Git::SVN::Fetcher object is passed) to
4039# do_{switch,update}
4040sub _mark_empty_symlinks {
4041 my ($git_svn, $switch_path) = @_;
4042 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4043 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4044 my %ret;
4046 my ($rev, $cmt) = $git_svn->last_rev_commit;
4047 return {} unless ($rev && $cmt);
4048 # allow the warning to be printed for each revision we fetch to
4050 # ensure the user sees it. The user can also disable the workaround
4051 # on the repository even while git svn is running and the next
4052 # revision fetched will skip this expensive function.
4053 my $printed_warning;
4054 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4055 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4056 local $/ = "\0";
4057 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4058 $pfx .= '/' if length($pfx);
4059 while (<$ls>) {
4060 chomp;
4061 s/\A100644 blob $empty_blob\t//o or next;
4062 unless ($printed_warning) {
4063 print STDERR "Scanning for empty symlinks, ",
4064 "this may take a while if you have ",
4065 "many empty files\n",
4066 "You may disable this with `",
4067 "git config svn.brokenSymlinkWorkaround ",
4068 "false'.\n",
4069 "This may be done in a different ",
4070 "terminal without restarting ",
4071 "git svn\n";
4072 $printed_warning = 1;
4073 }
4074 my $path = $_;
4075 my (undef, $props) =
4076 $git_svn->ra->get_file($pfx.$path, $rev, undef);
4077 if ($props->{'svn:special'}) {
4078 $ret{$path} = 1;
4079 }
4080 }
4081 command_close_pipe($ls, $ctx);
4082 \%ret;
4083}
4084# returns true if a given path is inside a ".git" directory
4086sub in_dot_git {
4087 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4088}
4089# return value: 0 -- don't ignore, 1 -- ignore
4091sub is_path_ignored {
4092 my ($self, $path) = @_;
4093 return 1 if in_dot_git($path);
4094 return 1 if defined($self->{ignore_regex}) &&
4095 $path =~ m!$self->{ignore_regex}!;
4096 return 0 unless defined($_ignore_regex);
4097 return 1 if $path =~ m!$_ignore_regex!o;
4098 return 0;
4099}
4100sub set_path_strip {
4102 my ($self, $path) = @_;
4103 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4104}
4105sub open_root {
4107 { path => '' };
4108}
4109sub open_directory {
4111 my ($self, $path, $pb, $rev) = @_;
4112 { path => $path };
4113}
4114sub git_path {
4116 my ($self, $path) = @_;
4117 if ($self->{path_strip}) {
4118 $path =~ s!$self->{path_strip}!! or
4119 die "Failed to strip path '$path' ($self->{path_strip})\n";
4120 }
4121 $path;
4122}
4123sub delete_entry {
4125 my ($self, $path, $rev, $pb) = @_;
4126 return undef if $self->is_path_ignored($path);
4127 my $gpath = $self->git_path($path);
4129 return undef if ($gpath eq '');
4130 # remove entire directories.
4132 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4133 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4134 if ($tree) {
4135 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4136 -r --name-only -z/,
4137 $tree);
4138 local $/ = "\0";
4139 while (<$ls>) {
4140 chomp;
4141 my $rmpath = "$gpath/$_";
4142 $self->{gii}->remove($rmpath);
4143 print "\tD\t$rmpath\n" unless $::_q;
4144 }
4145 print "\tD\t$gpath/\n" unless $::_q;
4146 command_close_pipe($ls, $ctx);
4147 } else {
4148 $self->{gii}->remove($gpath);
4149 print "\tD\t$gpath\n" unless $::_q;
4150 }
4151 $self->{empty}->{$path} = 0;
4152 undef;
4153}
4154sub open_file {
4156 my ($self, $path, $pb, $rev) = @_;
4157 my ($mode, $blob);
4158 goto out if $self->is_path_ignored($path);
4160 my $gpath = $self->git_path($path);
4162 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4163 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4164 unless (defined $mode && defined $blob) {
4165 die "$path was not found in commit $self->{c} (r$rev)\n";
4166 }
4167 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4168 $mode = '120000';
4169 }
4170out:
4171 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4172 pool => SVN::Pool->new, action => 'M' };
4173}
4174sub add_file {
4176 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4177 my $mode;
4178 if (!$self->is_path_ignored($path)) {
4180 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4181 delete $self->{empty}->{$dir};
4182 $mode = '100644';
4183 }
4184 { path => $path, mode_a => $mode, mode_b => $mode,
4185 pool => SVN::Pool->new, action => 'A' };
4186}
4187sub add_directory {
4189 my ($self, $path, $cp_path, $cp_rev) = @_;
4190 goto out if $self->is_path_ignored($path);
4191 my $gpath = $self->git_path($path);
4192 if ($gpath eq '') {
4193 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4194 -r --name-only -z/,
4195 $self->{c});
4196 local $/ = "\0";
4197 while (<$ls>) {
4198 chomp;
4199 $self->{gii}->remove($_);
4200 print "\tD\t$_\n" unless $::_q;
4201 }
4202 command_close_pipe($ls, $ctx);
4203 $self->{empty}->{$path} = 0;
4204 }
4205 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4206 delete $self->{empty}->{$dir};
4207 $self->{empty}->{$path} = 1;
4208out:
4209 { path => $path };
4210}
4211sub change_dir_prop {
4213 my ($self, $db, $prop, $value) = @_;
4214 return undef if $self->is_path_ignored($db->{path});
4215 $self->{dir_prop}->{$db->{path}} ||= {};
4216 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4217 undef;
4218}
4219sub absent_directory {
4221 my ($self, $path, $pb) = @_;
4222 return undef if $self->is_path_ignored($path);
4223 $self->{absent_dir}->{$pb->{path}} ||= [];
4224 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4225 undef;
4226}
4227sub absent_file {
4229 my ($self, $path, $pb) = @_;
4230 return undef if $self->is_path_ignored($path);
4231 $self->{absent_file}->{$pb->{path}} ||= [];
4232 push @{$self->{absent_file}->{$pb->{path}}}, $path;
4233 undef;
4234}
4235sub change_file_prop {
4237 my ($self, $fb, $prop, $value) = @_;
4238 return undef if $self->is_path_ignored($fb->{path});
4239 if ($prop eq 'svn:executable') {
4240 if ($fb->{mode_b} != 120000) {
4241 $fb->{mode_b} = defined $value ? 100755 : 100644;
4242 }
4243 } elsif ($prop eq 'svn:special') {
4244 $fb->{mode_b} = defined $value ? 120000 : 100644;
4245 } else {
4246 $self->{file_prop}->{$fb->{path}} ||= {};
4247 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4248 }
4249 undef;
4250}
4251sub apply_textdelta {
4253 my ($self, $fb, $exp) = @_;
4254 return undef if $self->is_path_ignored($fb->{path});
4255 my $fh = $::_repository->temp_acquire('svn_delta');
4256 # $fh gets auto-closed() by SVN::TxDelta::apply(),
4257 # (but $base does not,) so dup() it for reading in close_file
4258 open my $dup, '<&', $fh or croak $!;
4259 my $base = $::_repository->temp_acquire('git_blob');
4260 if ($fb->{blob}) {
4262 my ($base_is_link, $size);
4263 if ($fb->{mode_a} eq '120000' &&
4265 ! $self->{empty_symlinks}->{$fb->{path}}) {
4266 print $base 'link ' or die "print $!\n";
4267 $base_is_link = 1;
4268 }
4269 retry:
4270 $size = $::_repository->cat_blob($fb->{blob}, $base);
4271 die "Failed to read object $fb->{blob}" if ($size < 0);
4272 if (defined $exp) {
4274 seek $base, 0, 0 or croak $!;
4275 my $got = ::md5sum($base);
4276 if ($got ne $exp) {
4277 my $err = "Checksum mismatch: ".
4278 "$fb->{path} $fb->{blob}\n" .
4279 "expected: $exp\n" .
4280 " got: $got\n";
4281 if ($base_is_link) {
4282 warn $err,
4283 "Retrying... (possibly ",
4284 "a bad symlink from SVN)\n";
4285 $::_repository->temp_reset($base);
4286 $base_is_link = 0;
4287 goto retry;
4288 }
4289 die $err;
4290 }
4291 }
4292 }
4293 seek $base, 0, 0 or croak $!;
4294 $fb->{fh} = $fh;
4295 $fb->{base} = $base;
4296 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4297}
4298sub close_file {
4300 my ($self, $fb, $exp) = @_;
4301 return undef if $self->is_path_ignored($fb->{path});
4302 my $hash;
4304 my $path = $self->git_path($fb->{path});
4305 if (my $fh = $fb->{fh}) {
4306 if (defined $exp) {
4307 seek($fh, 0, 0) or croak $!;
4308 my $got = ::md5sum($fh);
4309 if ($got ne $exp) {
4310 die "Checksum mismatch: $path\n",
4311 "expected: $exp\n got: $got\n";
4312 }
4313 }
4314 if ($fb->{mode_b} == 120000) {
4315 sysseek($fh, 0, 0) or croak $!;
4316 my $rd = sysread($fh, my $buf, 5);
4317 if (!defined $rd) {
4319 croak "sysread: $!\n";
4320 } elsif ($rd == 0) {
4321 warn "$path has mode 120000",
4322 " but it points to nothing\n",
4323 "converting to an empty file with mode",
4324 " 100644\n";
4325 $fb->{mode_b} = '100644';
4326 } elsif ($buf ne 'link ') {
4327 warn "$path has mode 120000",
4328 " but is not a link\n";
4329 } else {
4330 my $tmp_fh = $::_repository->temp_acquire(
4331 'svn_hash');
4332 my $res;
4333 while ($res = sysread($fh, my $str, 1024)) {
4334 my $out = syswrite($tmp_fh, $str, $res);
4335 defined($out) && $out == $res
4336 or croak("write ",
4337 Git::temp_path($tmp_fh),
4338 ": $!\n");
4339 }
4340 defined $res or croak $!;
4341 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4343 Git::temp_release($tmp_fh, 1);
4344 }
4345 }
4346 $hash = $::_repository->hash_and_insert_object(
4348 Git::temp_path($fh));
4349 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4350 Git::temp_release($fb->{base}, 1);
4352 Git::temp_release($fh, 1);
4353 } else {
4354 $hash = $fb->{blob} or die "no blob information\n";
4355 }
4356 $fb->{pool}->clear;
4357 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4358 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4359 undef;
4360}
4361sub abort_edit {
4363 my $self = shift;
4364 $self->{nr} = $self->{gii}->{nr};
4365 delete $self->{gii};
4366 $self->SUPER::abort_edit(@_);
4367}
4368sub close_edit {
4370 my $self = shift;
4371 $self->{git_commit_ok} = 1;
4372 $self->{nr} = $self->{gii}->{nr};
4373 delete $self->{gii};
4374 $self->SUPER::close_edit(@_);
4375}
4376package SVN::Git::Editor;
4378use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4379use strict;
4380use warnings;
4381use Carp qw/croak/;
4382use IO::File;
4383sub new {
4385 my ($class, $opts) = @_;
4386 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4387 die "$_ required!\n" unless (defined $opts->{$_});
4388 }
4389 my $pool = SVN::Pool->new;
4391 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4392 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4393 $opts->{r}, $mods);
4394 # $opts->{ra} functions should not be used after this:
4396 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4397 $opts->{editor_cb}, $pool);
4398 my $self = SVN::Delta::Editor->new(@ce, $pool);
4399 bless $self, $class;
4400 foreach (qw/svn_path r tree_a tree_b/) {
4401 $self->{$_} = $opts->{$_};
4402 }
4403 $self->{url} = $opts->{ra}->{url};
4404 $self->{mods} = $mods;
4405 $self->{types} = $types;
4406 $self->{pool} = $pool;
4407 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4408 $self->{rm} = { };
4409 $self->{path_prefix} = length $self->{svn_path} ?
4410 "$self->{svn_path}/" : '';
4411 $self->{config} = $opts->{config};
4412 return $self;
4413}
4414sub generate_diff {
4416 my ($tree_a, $tree_b) = @_;
4417 my @diff_tree = qw(diff-tree -z -r);
4418 if ($_cp_similarity) {
4419 push @diff_tree, "-C$_cp_similarity";
4420 } else {
4421 push @diff_tree, '-C';
4422 }
4423 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4424 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4425 push @diff_tree, $tree_a, $tree_b;
4426 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4427 local $/ = "\0";
4428 my $state = 'meta';
4429 my @mods;
4430 while (<$diff_fh>) {
4431 chomp $_; # this gets rid of the trailing "\0"
4432 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4433 ($::sha1)\s($::sha1)\s
4434 ([MTCRAD])\d*$/xo) {
4435 push @mods, { mode_a => $1, mode_b => $2,
4436 sha1_a => $3, sha1_b => $4,
4437 chg => $5 };
4438 if ($5 =~ /^(?:C|R)$/) {
4439 $state = 'file_a';
4440 } else {
4441 $state = 'file_b';
4442 }
4443 } elsif ($state eq 'file_a') {
4444 my $x = $mods[$#mods] or croak "Empty array\n";
4445 if ($x->{chg} !~ /^(?:C|R)$/) {
4446 croak "Error parsing $_, $x->{chg}\n";
4447 }
4448 $x->{file_a} = $_;
4449 $state = 'file_b';
4450 } elsif ($state eq 'file_b') {
4451 my $x = $mods[$#mods] or croak "Empty array\n";
4452 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4453 croak "Error parsing $_, $x->{chg}\n";
4454 }
4455 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4456 croak "Error parsing $_, $x->{chg}\n";
4457 }
4458 $x->{file_b} = $_;
4459 $state = 'meta';
4460 } else {
4461 croak "Error parsing $_\n";
4462 }
4463 }
4464 command_close_pipe($diff_fh, $ctx);
4465 \@mods;
4466}
4467sub check_diff_paths {
4469 my ($ra, $pfx, $rev, $mods) = @_;
4470 my %types;
4471 $pfx .= '/' if length $pfx;
4472 sub type_diff_paths {
4474 my ($ra, $types, $path, $rev) = @_;
4475 my @p = split m#/+#, $path;
4476 my $c = shift @p;
4477 unless (defined $types->{$c}) {
4478 $types->{$c} = $ra->check_path($c, $rev);
4479 }
4480 while (@p) {
4481 $c .= '/' . shift @p;
4482 next if defined $types->{$c};
4483 $types->{$c} = $ra->check_path($c, $rev);
4484 }
4485 }
4486 foreach my $m (@$mods) {
4488 foreach my $f (qw/file_a file_b/) {
4489 next unless defined $m->{$f};
4490 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4491 if (length $pfx.$dir && ! defined $types{$dir}) {
4492 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4493 }
4494 }
4495 }
4496 \%types;
4497}
4498sub split_path {
4500 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4501}
4502sub repo_path {
4504 my ($self, $path) = @_;
4505 $self->{path_prefix}.(defined $path ? $path : '');
4506}
4507sub url_path {
4509 my ($self, $path) = @_;
4510 if ($self->{url} =~ m#^https?://#) {
4511 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4512 }
4513 $self->{url} . '/' . $self->repo_path($path);
4514}
4515sub rmdirs {
4517 my ($self) = @_;
4518 my $rm = $self->{rm};
4519 delete $rm->{''}; # we never delete the url we're tracking
4520 return unless %$rm;
4521 foreach (keys %$rm) {
4523 my @d = split m#/#, $_;
4524 my $c = shift @d;
4525 $rm->{$c} = 1;
4526 while (@d) {
4527 $c .= '/' . shift @d;
4528 $rm->{$c} = 1;
4529 }
4530 }
4531 delete $rm->{$self->{svn_path}};
4532 delete $rm->{''}; # we never delete the url we're tracking
4533 return unless %$rm;
4534 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4536 $self->{tree_b});
4537 local $/ = "\0";
4538 while (<$fh>) {
4539 chomp;
4540 my @dn = split m#/#, $_;
4541 while (pop @dn) {
4542 delete $rm->{join '/', @dn};
4543 }
4544 unless (%$rm) {
4545 close $fh;
4546 return;
4547 }
4548 }
4549 command_close_pipe($fh, $ctx);
4550 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4552 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4553 $self->close_directory($bat->{$d}, $p);
4554 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4555 print "\tD+\t$d/\n" unless $::_q;
4556 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4557 delete $bat->{$d};
4558 }
4559}
4560sub open_or_add_dir {
4562 my ($self, $full_path, $baton) = @_;
4563 my $t = $self->{types}->{$full_path};
4564 if (!defined $t) {
4565 die "$full_path not known in r$self->{r} or we have a bug!\n";
4566 }
4567 {
4568 no warnings 'once';
4569 # SVN::Node::none and SVN::Node::file are used only once,
4570 # so we're shutting up Perl's warnings about them.
4571 if ($t == $SVN::Node::none) {
4572 return $self->add_directory($full_path, $baton,
4573 undef, -1, $self->{pool});
4574 } elsif ($t == $SVN::Node::dir) {
4575 return $self->open_directory($full_path, $baton,
4576 $self->{r}, $self->{pool});
4577 } # no warnings 'once'
4578 print STDERR "$full_path already exists in repository at ",
4579 "r$self->{r} and it is not a directory (",
4580 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4581 } # no warnings 'once'
4582 exit 1;
4583}
4584sub ensure_path {
4586 my ($self, $path) = @_;
4587 my $bat = $self->{bat};
4588 my $repo_path = $self->repo_path($path);
4589 return $bat->{''} unless (length $repo_path);
4590 my @p = split m#/+#, $repo_path;
4591 my $c = shift @p;
4592 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4593 while (@p) {
4594 my $c0 = $c;
4595 $c .= '/' . shift @p;
4596 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4597 }
4598 return $bat->{$c};
4599}
4600# Subroutine to convert a globbing pattern to a regular expression.
4602# From perl cookbook.
4603sub glob2pat {
4604 my $globstr = shift;
4605 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4606 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4607 return '^' . $globstr . '$';
4608}
4609sub check_autoprop {
4611 my ($self, $pattern, $properties, $file, $fbat) = @_;
4612 # Convert the globbing pattern to a regular expression.
4613 my $regex = glob2pat($pattern);
4614 # Check if the pattern matches the file name.
4615 if($file =~ m/($regex)/) {
4616 # Parse the list of properties to set.
4617 my @props = split(/;/, $properties);
4618 foreach my $prop (@props) {
4619 # Parse 'name=value' syntax and set the property.
4620 if ($prop =~ /([^=]+)=(.*)/) {
4621 my ($n,$v) = ($1,$2);
4622 for ($n, $v) {
4623 s/^\s+//; s/\s+$//;
4624 }
4625 $self->change_file_prop($fbat, $n, $v);
4626 }
4627 }
4628 }
4629}
4630sub apply_autoprops {
4632 my ($self, $file, $fbat) = @_;
4633 my $conf_t = ${$self->{config}}{'config'};
4634 no warnings 'once';
4635 # Check [miscellany]/enable-auto-props in svn configuration.
4636 if (SVN::_Core::svn_config_get_bool(
4637 $conf_t,
4638 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4639 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4640 0)) {
4641 # Auto-props are enabled. Enumerate them to look for matches.
4642 my $callback = sub {
4643 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4644 };
4645 SVN::_Core::svn_config_enumerate(
4646 $conf_t,
4647 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4648 $callback);
4649 }
4650}
4651sub A {
4653 my ($self, $m) = @_;
4654 my ($dir, $file) = split_path($m->{file_b});
4655 my $pbat = $self->ensure_path($dir);
4656 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4657 undef, -1);
4658 print "\tA\t$m->{file_b}\n" unless $::_q;
4659 $self->apply_autoprops($file, $fbat);
4660 $self->chg_file($fbat, $m);
4661 $self->close_file($fbat,undef,$self->{pool});
4662}
4663sub C {
4665 my ($self, $m) = @_;
4666 my ($dir, $file) = split_path($m->{file_b});
4667 my $pbat = $self->ensure_path($dir);
4668 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4669 $self->url_path($m->{file_a}), $self->{r});
4670 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4671 $self->chg_file($fbat, $m);
4672 $self->close_file($fbat,undef,$self->{pool});
4673}
4674sub delete_entry {
4676 my ($self, $path, $pbat) = @_;
4677 my $rpath = $self->repo_path($path);
4678 my ($dir, $file) = split_path($rpath);
4679 $self->{rm}->{$dir} = 1;
4680 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4681}
4682sub R {
4684 my ($self, $m) = @_;
4685 my ($dir, $file) = split_path($m->{file_b});
4686 my $pbat = $self->ensure_path($dir);
4687 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4688 $self->url_path($m->{file_a}), $self->{r});
4689 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4690 $self->apply_autoprops($file, $fbat);
4691 $self->chg_file($fbat, $m);
4692 $self->close_file($fbat,undef,$self->{pool});
4693 ($dir, $file) = split_path($m->{file_a});
4695 $pbat = $self->ensure_path($dir);
4696 $self->delete_entry($m->{file_a}, $pbat);
4697}
4698sub M {
4700 my ($self, $m) = @_;
4701 my ($dir, $file) = split_path($m->{file_b});
4702 my $pbat = $self->ensure_path($dir);
4703 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4704 $pbat,$self->{r},$self->{pool});
4705 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4706 $self->chg_file($fbat, $m);
4707 $self->close_file($fbat,undef,$self->{pool});
4708}
4709sub T { shift->M(@_) }
4711sub change_file_prop {
4713 my ($self, $fbat, $pname, $pval) = @_;
4714 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4715}
4716sub _chg_file_get_blob ($$$$) {
4718 my ($self, $fbat, $m, $which) = @_;
4719 my $fh = $::_repository->temp_acquire("git_blob_$which");
4720 if ($m->{"mode_$which"} =~ /^120/) {
4721 print $fh 'link ' or croak $!;
4722 $self->change_file_prop($fbat,'svn:special','*');
4723 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4724 $self->change_file_prop($fbat,'svn:special',undef);
4725 }
4726 my $blob = $m->{"sha1_$which"};
4727 return ($fh,) if ($blob =~ /^0{40}$/);
4728 my $size = $::_repository->cat_blob($blob, $fh);
4729 croak "Failed to read object $blob" if ($size < 0);
4730 $fh->flush == 0 or croak $!;
4731 seek $fh, 0, 0 or croak $!;
4732 my $exp = ::md5sum($fh);
4734 seek $fh, 0, 0 or croak $!;
4735 return ($fh, $exp);
4736}
4737sub chg_file {
4739 my ($self, $fbat, $m) = @_;
4740 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4741 $self->change_file_prop($fbat,'svn:executable','*');
4742 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4743 $self->change_file_prop($fbat,'svn:executable',undef);
4744 }
4745 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4746 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4747 my $pool = SVN::Pool->new;
4748 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4749 if (-s $fh_a) {
4750 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4751 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4752 if (defined $res) {
4753 die "Unexpected result from send_txstream: $res\n",
4754 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4755 }
4756 } else {
4757 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4758 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4759 if ($got ne $exp_b);
4760 }
4761 Git::temp_release($fh_b, 1);
4762 Git::temp_release($fh_a, 1);
4763 $pool->clear;
4764}
4765sub D {
4767 my ($self, $m) = @_;
4768 my ($dir, $file) = split_path($m->{file_b});
4769 my $pbat = $self->ensure_path($dir);
4770 print "\tD\t$m->{file_b}\n" unless $::_q;
4771 $self->delete_entry($m->{file_b}, $pbat);
4772}
4773sub close_edit {
4775 my ($self) = @_;
4776 my ($p,$bat) = ($self->{pool}, $self->{bat});
4777 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4778 next if $_ eq '';
4779 $self->close_directory($bat->{$_}, $p);
4780 }
4781 $self->close_directory($bat->{''}, $p);
4782 $self->SUPER::close_edit($p);
4783 $p->clear;
4784}
4785sub abort_edit {
4787 my ($self) = @_;
4788 $self->SUPER::abort_edit($self->{pool});
4789}
4790sub DESTROY {
4792 my $self = shift;
4793 $self->SUPER::DESTROY(@_);
4794 $self->{pool}->clear;
4795}
4796# this drives the editor
4798sub apply_diff {
4799 my ($self) = @_;
4800 my $mods = $self->{mods};
4801 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4802 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4803 my $f = $m->{chg};
4804 if (defined $o{$f}) {
4805 $self->$f($m);
4806 } else {
4807 fatal("Invalid change type: $f");
4808 }
4809 }
4810 $self->rmdirs if $_rmdir;
4811 if (@$mods == 0) {
4812 $self->abort_edit;
4813 } else {
4814 $self->close_edit;
4815 }
4816 return scalar @$mods;
4817}
4818package Git::SVN::Ra;
4820use vars qw/@ISA $config_dir $_log_window_size/;
4821use strict;
4822use warnings;
4823my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4824BEGIN {
4826 # enforce temporary pool usage for some simple functions
4827 no strict 'refs';
4828 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4829 get_file/) {
4830 my $SUPER = "SUPER::$f";
4831 *$f = sub {
4832 my $self = shift;
4833 my $pool = SVN::Pool->new;
4834 my @ret = $self->$SUPER(@_,$pool);
4835 $pool->clear;
4836 wantarray ? @ret : $ret[0];
4837 };
4838 }
4839}
4840sub _auth_providers () {
4842 [
4843 SVN::Client::get_simple_provider(),
4844 SVN::Client::get_ssl_server_trust_file_provider(),
4845 SVN::Client::get_simple_prompt_provider(
4846 \&Git::SVN::Prompt::simple, 2),
4847 SVN::Client::get_ssl_client_cert_file_provider(),
4848 SVN::Client::get_ssl_client_cert_prompt_provider(
4849 \&Git::SVN::Prompt::ssl_client_cert, 2),
4850 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4851 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4852 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4853 SVN::Client::get_username_provider(),
4854 SVN::Client::get_ssl_server_trust_prompt_provider(
4855 \&Git::SVN::Prompt::ssl_server_trust),
4856 SVN::Client::get_username_prompt_provider(
4857 \&Git::SVN::Prompt::username, 2)
4858 ]
4859}
4860sub escape_uri_only {
4862 my ($uri) = @_;
4863 my @tmp;
4864 foreach (split m{/}, $uri) {
4865 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4866 push @tmp, $_;
4867 }
4868 join('/', @tmp);
4869}
4870sub escape_url {
4872 my ($url) = @_;
4873 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4874 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4875 $url = "$scheme://$domain$uri";
4876 }
4877 $url;
4878}
4879sub new {
4881 my ($class, $url) = @_;
4882 $url =~ s!/+$!!;
4883 return $RA if ($RA && $RA->{url} eq $url);
4884 ::_req_svn();
4886 SVN::_Core::svn_config_ensure($config_dir, undef);
4888 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4889 my $config = SVN::Core::config_get_config($config_dir);
4890 $RA = undef;
4891 my $dont_store_passwords = 1;
4892 my $conf_t = ${$config}{'config'};
4893 {
4894 no warnings 'once';
4895 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4896 # produces warnings that variables are used only once.
4897 # I had not found the better way to shut them up, so
4898 # the warnings of type 'once' are disabled in this block.
4899 if (SVN::_Core::svn_config_get_bool($conf_t,
4900 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4901 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4902 1) == 0) {
4903 SVN::_Core::svn_auth_set_parameter($baton,
4904 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4905 bless (\$dont_store_passwords, "_p_void"));
4906 }
4907 if (SVN::_Core::svn_config_get_bool($conf_t,
4908 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4909 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4910 1) == 0) {
4911 $Git::SVN::Prompt::_no_auth_cache = 1;
4912 }
4913 } # no warnings 'once'
4914 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4915 config => $config,
4916 pool => SVN::Pool->new,
4917 auth_provider_callbacks => $callbacks);
4918 $self->{url} = $url;
4919 $self->{svn_path} = $url;
4920 $self->{repos_root} = $self->get_repos_root;
4921 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4922 $self->{cache} = { check_path => { r => 0, data => {} },
4923 get_dir => { r => 0, data => {} } };
4924 $RA = bless $self, $class;
4925}
4926sub check_path {
4928 my ($self, $path, $r) = @_;
4929 my $cache = $self->{cache}->{check_path};
4930 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4931 return $cache->{data}->{$path};
4932 }
4933 my $pool = SVN::Pool->new;
4934 my $t = $self->SUPER::check_path($path, $r, $pool);
4935 $pool->clear;
4936 if ($r != $cache->{r}) {
4937 %{$cache->{data}} = ();
4938 $cache->{r} = $r;
4939 }
4940 $cache->{data}->{$path} = $t;
4941}
4942sub get_dir {
4944 my ($self, $dir, $r) = @_;
4945 my $cache = $self->{cache}->{get_dir};
4946 if ($r == $cache->{r}) {
4947 if (my $x = $cache->{data}->{$dir}) {
4948 return wantarray ? @$x : $x->[0];
4949 }
4950 }
4951 my $pool = SVN::Pool->new;
4952 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4953 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4954 $pool->clear;
4955 if ($r != $cache->{r}) {
4956 %{$cache->{data}} = ();
4957 $cache->{r} = $r;
4958 }
4959 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4960 wantarray ? (\%dirents, $r, $props) : \%dirents;
4961}
4962sub DESTROY {
4964 # do not call the real DESTROY since we store ourselves in $RA
4965}
4966# get_log(paths, start, end, limit,
4968# discover_changed_paths, strict_node_history, receiver)
4969sub get_log {
4970 my ($self, @args) = @_;
4971 my $pool = SVN::Pool->new;
4972 # svn_log_changed_path_t objects passed to get_log are likely to be
4974 # overwritten even if only the refs are copied to an external variable,
4975 # so we should dup the structures in their entirety. Using an
4976 # externally passed pool (instead of our temporary and quickly cleared
4977 # pool in Git::SVN::Ra) does not help matters at all...
4978 my $receiver = pop @args;
4979 my $prefix = "/".$self->{svn_path};
4980 $prefix =~ s#/+($)##;
4981 my $prefix_regex = qr#^\Q$prefix\E#;
4982 push(@args, sub {
4983 my ($paths) = $_[0];
4984 return &$receiver(@_) unless $paths;
4985 $_[0] = ();
4986 foreach my $p (keys %$paths) {
4987 my $i = $paths->{$p};
4988 # Make path relative to our url, not repos_root
4989 $p =~ s/$prefix_regex//;
4990 my %s = map { $_ => $i->$_; }
4991 qw/copyfrom_path copyfrom_rev action/;
4992 if ($s{'copyfrom_path'}) {
4993 $s{'copyfrom_path'} =~ s/$prefix_regex//;
4994 }
4995 $_[0]{$p} = \%s;
4996 }
4997 &$receiver(@_);
4998 });
4999 # the limit parameter was not supported in SVN 1.1.x, so we
5002 # drop it. Therefore, the receiver callback passed to it
5003 # is made aware of this limitation by being wrapped if
5004 # the limit passed to is being wrapped.
5005 if ($SVN::Core::VERSION le '1.2.0') {
5006 my $limit = splice(@args, 3, 1);
5007 if ($limit > 0) {
5008 my $receiver = pop @args;
5009 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
5010 }
5011 }
5012 my $ret = $self->SUPER::get_log(@args, $pool);
5013 $pool->clear;
5014 $ret;
5015}
5016sub trees_match {
5018 my ($self, $url1, $rev1, $url2, $rev2) = @_;
5019 my $ctx = SVN::Client->new(auth => _auth_providers);
5020 my $out = IO::File->new_tmpfile;
5021 # older SVN (1.1.x) doesn't take $pool as the last parameter for
5023 # $ctx->diff(), so we'll create a default one
5024 my $pool = SVN::Pool->new_default_sub;
5025 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5027 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5028 $out->flush;
5029 my $ret = (($out->stat)[7] == 0);
5030 close $out or croak $!;
5031 $ret;
5033}
5034sub get_commit_editor {
5036 my ($self, $log, $cb, $pool) = @_;
5037 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
5038 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5039}
5040sub gs_do_update {
5042 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5043 my $new = ($rev_a == $rev_b);
5044 my $path = $gs->{path};
5045 if ($new && -e $gs->{index}) {
5047 unlink $gs->{index} or die
5048 "Couldn't unlink index: $gs->{index}: $!\n";
5049 }
5050 my $pool = SVN::Pool->new;
5051 $editor->set_path_strip($path);
5052 my (@pc) = split m#/#, $path;
5053 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5054 1, $editor, $pool);
5055 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5056 # Since we can't rely on svn_ra_reparent being available, we'll
5058 # just have to do some magic with set_path to make it so
5059 # we only want a partial path.
5060 my $sp = '';
5061 my $final = join('/', @pc);
5062 while (@pc) {
5063 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5064 $sp .= '/' if length $sp;
5065 $sp .= shift @pc;
5066 }
5067 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5068 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5070 $reporter->finish_report($pool);
5072 $pool->clear;
5073 $editor->{git_commit_ok};
5074}
5075# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5077# svn_ra_reparent didn't work before 1.4)
5078sub gs_do_switch {
5079 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5080 my $path = $gs->{path};
5081 my $pool = SVN::Pool->new;
5082 my $full_url = $self->{url};
5084 my $old_url = $full_url;
5085 $full_url .= '/' . $path if length $path;
5086 my ($ra, $reparented);
5087 if ($old_url =~ m#^svn(\+ssh)?://# ||
5089 ($full_url =~ m#^https?://# &&
5090 escape_url($full_url) ne $full_url)) {
5091 $_[0] = undef;
5092 $self = undef;
5093 $RA = undef;
5094 $ra = Git::SVN::Ra->new($full_url);
5095 $ra_invalid = 1;
5096 } elsif ($old_url ne $full_url) {
5097 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5098 $self->{url} = $full_url;
5099 $reparented = 1;
5100 }
5101 $ra ||= $self;
5103 $url_b = escape_url($url_b);
5104 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5105 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5106 $reporter->set_path('', $rev_a, 0, @lock, $pool);
5107 $reporter->finish_report($pool);
5108 if ($reparented) {
5110 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5111 $self->{url} = $old_url;
5112 }
5113 $pool->clear;
5115 $editor->{git_commit_ok};
5116}
5117sub longest_common_path {
5119 my ($gsv, $globs) = @_;
5120 my %common;
5121 my $common_max = scalar @$gsv;
5122 foreach my $gs (@$gsv) {
5124 my @tmp = split m#/#, $gs->{path};
5125 my $p = '';
5126 foreach (@tmp) {
5127 $p .= length($p) ? "/$_" : $_;
5128 $common{$p} ||= 0;
5129 $common{$p}++;
5130 }
5131 }
5132 $globs ||= [];
5133 $common_max += scalar @$globs;
5134 foreach my $glob (@$globs) {
5135 my @tmp = split m#/#, $glob->{path}->{left};
5136 my $p = '';
5137 foreach (@tmp) {
5138 $p .= length($p) ? "/$_" : $_;
5139 $common{$p} ||= 0;
5140 $common{$p}++;
5141 }
5142 }
5143 my $longest_path = '';
5145 foreach (sort {length $b <=> length $a} keys %common) {
5146 if ($common{$_} == $common_max) {
5147 $longest_path = $_;
5148 last;
5149 }
5150 }
5151 $longest_path;
5152}
5153sub gs_fetch_loop_common {
5155 my ($self, $base, $head, $gsv, $globs) = @_;
5156 return if ($base > $head);
5157 my $inc = $_log_window_size;
5158 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5159 my $longest_path = longest_common_path($gsv, $globs);
5160 my $ra_url = $self->{url};
5161 my $find_trailing_edge;
5162 while (1) {
5163 my %revs;
5164 my $err;
5165 my $err_handler = $SVN::Error::handler;
5166 $SVN::Error::handler = sub {
5167 ($err) = @_;
5168 skip_unknown_revs($err);
5169 };
5170 sub _cb {
5171 my ($paths, $r, $author, $date, $log) = @_;
5172 [ $paths,
5173 { author => $author, date => $date, log => $log } ];
5174 }
5175 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5176 sub { $revs{$_[1]} = _cb(@_) });
5177 if ($err) {
5178 print "Checked through r$max\r";
5179 } else {
5180 $find_trailing_edge = 1;
5181 }
5182 if ($err and $find_trailing_edge) {
5183 print STDERR "Path '$longest_path' ",
5184 "was probably deleted:\n",
5185 $err->expanded_message,
5186 "\nWill attempt to follow ",
5187 "revisions r$min .. r$max ",
5188 "committed before the deletion\n";
5189 my $hi = $max;
5190 while (--$hi >= $min) {
5191 my $ok;
5192 $self->get_log([$longest_path], $min, $hi,
5193 0, 1, 1, sub {
5194 $ok = $_[1];
5195 $revs{$_[1]} = _cb(@_) });
5196 if ($ok) {
5197 print STDERR "r$min .. r$ok OK\n";
5198 last;
5199 }
5200 }
5201 $find_trailing_edge = 0;
5202 }
5203 $SVN::Error::handler = $err_handler;
5204 my %exists = map { $_->{path} => $_ } @$gsv;
5206 foreach my $r (sort {$a <=> $b} keys %revs) {
5207 my ($paths, $logged) = @{$revs{$r}};
5208 foreach my $gs ($self->match_globs(\%exists, $paths,
5210 $globs, $r)) {
5211 if ($gs->rev_map_max >= $r) {
5212 next;
5213 }
5214 next unless $gs->match_paths($paths, $r);
5215 $gs->{logged_rev_props} = $logged;
5216 if (my $last_commit = $gs->last_commit) {
5217 $gs->assert_index_clean($last_commit);
5218 }
5219 my $log_entry = $gs->do_fetch($paths, $r);
5220 if ($log_entry) {
5221 $gs->do_git_commit($log_entry);
5222 }
5223 $INDEX_FILES{$gs->{index}} = 1;
5224 }
5225 foreach my $g (@$globs) {
5226 my $k = "svn-remote.$g->{remote}." .
5227 "$g->{t}-maxRev";
5228 Git::SVN::tmp_config($k, $r);
5229 }
5230 if ($ra_invalid) {
5231 $_[0] = undef;
5232 $self = undef;
5233 $RA = undef;
5234 $self = Git::SVN::Ra->new($ra_url);
5235 $ra_invalid = undef;
5236 }
5237 }
5238 # pre-fill the .rev_db since it'll eventually get filled in
5239 # with '0' x40 if something new gets committed
5240 foreach my $gs (@$gsv) {
5241 next if $gs->rev_map_max >= $max;
5242 next if defined $gs->rev_map_get($max);
5243 $gs->rev_map_set($max, 0 x40);
5244 }
5245 foreach my $g (@$globs) {
5246 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5247 Git::SVN::tmp_config($k, $max);
5248 }
5249 last if $max >= $head;
5250 $min = $max + 1;
5251 $max += $inc;
5252 $max = $head if ($max > $head);
5253 }
5254 Git::SVN::gc();
5255}
5256sub get_dir_globbed {
5258 my ($self, $left, $depth, $r) = @_;
5259 my @x = eval { $self->get_dir($left, $r) };
5261 return unless scalar @x == 3;
5262 my $dirents = $x[0];
5263 my @finalents;
5264 foreach my $de (keys %$dirents) {
5265 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5266 if ($depth > 1) {
5267 my @args = ("$left/$de", $depth - 1, $r);
5268 foreach my $dir ($self->get_dir_globbed(@args)) {
5269 push @finalents, "$de/$dir";
5270 }
5271 } else {
5272 push @finalents, $de;
5273 }
5274 }
5275 @finalents;
5276}
5277sub match_globs {
5279 my ($self, $exists, $paths, $globs, $r) = @_;
5280 sub get_dir_check {
5282 my ($self, $exists, $g, $r) = @_;
5283 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5285 $g->{path}->{depth},
5286 $r);
5287 foreach my $de (@dirs) {
5289 my $p = $g->{path}->full_path($de);
5290 next if $exists->{$p};
5291 next if (length $g->{path}->{right} &&
5292 ($self->check_path($p, $r) !=
5293 $SVN::Node::dir));
5294 next unless $p =~ /$g->{path}->{regex}/;
5295 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5296 $g->{ref}->full_path($de), 1);
5297 }
5298 }
5299 foreach my $g (@$globs) {
5300 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5301 if ($path->{action} =~ /^[AR]$/) {
5302 get_dir_check($self, $exists, $g, $r);
5303 }
5304 }
5305 foreach (keys %$paths) {
5306 if (/$g->{path}->{left_regex}/ &&
5307 !/$g->{path}->{regex}/) {
5308 next if $paths->{$_}->{action} !~ /^[AR]$/;
5309 get_dir_check($self, $exists, $g, $r);
5310 }
5311 next unless /$g->{path}->{regex}/;
5312 my $p = $1;
5313 my $pathname = $g->{path}->full_path($p);
5314 next if $exists->{$pathname};
5315 next if ($self->check_path($pathname, $r) !=
5316 $SVN::Node::dir);
5317 $exists->{$pathname} = Git::SVN->init(
5318 $self->{url}, $pathname, undef,
5319 $g->{ref}->full_path($p), 1);
5320 }
5321 my $c = '';
5322 foreach (split m#/#, $g->{path}->{left}) {
5323 $c .= "/$_";
5324 next unless ($paths->{$c} &&
5325 ($paths->{$c}->{action} =~ /^[AR]$/));
5326 get_dir_check($self, $exists, $g, $r);
5327 }
5328 }
5329 values %$exists;
5330}
5331sub minimize_url {
5333 my ($self) = @_;
5334 return $self->{url} if ($self->{url} eq $self->{repos_root});
5335 my $url = $self->{repos_root};
5336 my @components = split(m!/!, $self->{svn_path});
5337 my $c = '';
5338 do {
5339 $url .= "/$c" if length $c;
5340 eval {
5341 my $ra = (ref $self)->new($url);
5342 my $latest = $ra->get_latest_revnum;
5343 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5344 };
5345 } while ($@ && ($c = shift @components));
5346 $url;
5347}
5348sub can_do_switch {
5350 my $self = shift;
5351 unless (defined $can_do_switch) {
5352 my $pool = SVN::Pool->new;
5353 my $rep = eval {
5354 $self->do_switch(1, '', 0, $self->{url},
5355 SVN::Delta::Editor->new, $pool);
5356 };
5357 if ($@) {
5358 $can_do_switch = 0;
5359 } else {
5360 $rep->abort_report($pool);
5361 $can_do_switch = 1;
5362 }
5363 $pool->clear;
5364 }
5365 $can_do_switch;
5366}
5367sub skip_unknown_revs {
5369 my ($err) = @_;
5370 my $errno = $err->apr_err();
5371 # Maybe the branch we're tracking didn't
5372 # exist when the repo started, so it's
5373 # not an error if it doesn't, just continue
5374 #
5375 # Wonderfully consistent library, eh?
5376 # 160013 - svn:// and file://
5377 # 175002 - http(s)://
5378 # 175007 - http(s):// (this repo required authorization, too...)
5379 # More codes may be discovered later...
5380 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5381 my $err_key = $err->expanded_message;
5382 # revision numbers change every time, filter them out
5383 $err_key =~ s/\d+/\0/g;
5384 $err_key = "$errno\0$err_key";
5385 unless ($ignored_err{$err_key}) {
5386 warn "W: Ignoring error from SVN, path probably ",
5387 "does not exist: ($errno): ",
5388 $err->expanded_message,"\n";
5389 warn "W: Do not be alarmed at the above message ",
5390 "git-svn is just searching aggressively for ",
5391 "old history.\n",
5392 "This may take a while on large repositories\n";
5393 $ignored_err{$err_key} = 1;
5394 }
5395 return;
5396 }
5397 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5398}
5399package Git::SVN::Log;
5401use strict;
5402use warnings;
5403use POSIX qw/strftime/;
5404use Time::Local;
5405use constant commit_log_separator => ('-' x 72) . "\n";
5406use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5407 %rusers $show_commit $incremental/;
5408my $l_fmt;
5409sub cmt_showable {
5411 my ($c) = @_;
5412 return 1 if defined $c->{r};
5413 # big commit message got truncated by the 16k pretty buffer in rev-list
5415 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5416 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5417 @{$c->{l}} = ();
5418 my @log = command(qw/cat-file commit/, $c->{c});
5419 # shift off the headers
5421 shift @log while ($log[0] ne '');
5422 shift @log;
5423 # TODO: make $c->{l} not have a trailing newline in the future
5425 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5426 (undef, $c->{r}, undef) = ::extract_metadata(
5428 (grep(/^git-svn-id: /, @log))[-1]);
5429 }
5430 return defined $c->{r};
5431}
5432sub log_use_color {
5434 return $color || Git->repository->get_colorbool('color.diff');
5435}
5436sub git_svn_log_cmd {
5438 my ($r_min, $r_max, @args) = @_;
5439 my $head = 'HEAD';
5440 my (@files, @log_opts);
5441 foreach my $x (@args) {
5442 if ($x eq '--' || @files) {
5443 push @files, $x;
5444 } else {
5445 if (::verify_ref("$x^0")) {
5446 $head = $x;
5447 } else {
5448 push @log_opts, $x;
5449 }
5450 }
5451 }
5452 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5454 $gs ||= Git::SVN->_new;
5455 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5456 $gs->refname);
5457 push @cmd, '-r' unless $non_recursive;
5458 push @cmd, qw/--raw --name-status/ if $verbose;
5459 push @cmd, '--color' if log_use_color();
5460 push @cmd, @log_opts;
5461 if (defined $r_max && $r_max == $r_min) {
5462 push @cmd, '--max-count=1';
5463 if (my $c = $gs->rev_map_get($r_max)) {
5464 push @cmd, $c;
5465 }
5466 } elsif (defined $r_max) {
5467 if ($r_max < $r_min) {
5468 ($r_min, $r_max) = ($r_max, $r_min);
5469 }
5470 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5471 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5472 # If there are no commits in the range, both $c_max and $c_min
5473 # will be undefined. If there is at least 1 commit in the
5474 # range, both will be defined.
5475 return () if !defined $c_min || !defined $c_max;
5476 if ($c_min eq $c_max) {
5477 push @cmd, '--max-count=1', $c_min;
5478 } else {
5479 push @cmd, '--boundary', "$c_min..$c_max";
5480 }
5481 }
5482 return (@cmd, @files);
5483}
5484# adapted from pager.c
5486sub config_pager {
5487 if (! -t *STDOUT) {
5488 $ENV{GIT_PAGER_IN_USE} = 'false';
5489 $pager = undef;
5490 return;
5491 }
5492 chomp($pager = command_oneline(qw(var GIT_PAGER)));
5493 if ($pager eq 'cat') {
5494 $pager = undef;
5495 }
5496 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5497}
5498sub run_pager {
5500 return unless defined $pager;
5501 pipe my ($rfd, $wfd) or return;
5502 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5503 if (!$pid) {
5504 open STDOUT, '>&', $wfd or
5505 ::fatal "Can't redirect to stdout: $!";
5506 return;
5507 }
5508 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5509 $ENV{LESS} ||= 'FRSX';
5510 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5511}
5512sub format_svn_date {
5514 # some systmes don't handle or mishandle %z, so be creative.
5515 my $t = shift || time;
5516 my $gm = timelocal(gmtime($t));
5517 my $sign = qw( + + - )[ $t <=> $gm ];
5518 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5519 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5520}
5521sub parse_git_date {
5523 my ($t, $tz) = @_;
5524 # Date::Parse isn't in the standard Perl distro :(
5525 if ($tz =~ s/^\+//) {
5526 $t += tz_to_s_offset($tz);
5527 } elsif ($tz =~ s/^\-//) {
5528 $t -= tz_to_s_offset($tz);
5529 }
5530 return $t;
5531}
5532sub set_local_timezone {
5534 if (defined $TZ) {
5535 $ENV{TZ} = $TZ;
5536 } else {
5537 delete $ENV{TZ};
5538 }
5539}
5540sub tz_to_s_offset {
5542 my ($tz) = @_;
5543 $tz =~ s/(\d\d)$//;
5544 return ($1 * 60) + ($tz * 3600);
5545}
5546sub get_author_info {
5548 my ($dest, $author, $t, $tz) = @_;
5549 $author =~ s/(?:^\s*|\s*$)//g;
5550 $dest->{a_raw} = $author;
5551 my $au;
5552 if ($::_authors) {
5553 $au = $rusers{$author} || undef;
5554 }
5555 if (!$au) {
5556 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5557 }
5558 $dest->{t} = $t;
5559 $dest->{tz} = $tz;
5560 $dest->{a} = $au;
5561 $dest->{t_utc} = parse_git_date($t, $tz);
5562}
5563sub process_commit {
5565 my ($c, $r_min, $r_max, $defer) = @_;
5566 if (defined $r_min && defined $r_max) {
5567 if ($r_min == $c->{r} && $r_min == $r_max) {
5568 show_commit($c);
5569 return 0;
5570 }
5571 return 1 if $r_min == $r_max;
5572 if ($r_min < $r_max) {
5573 # we need to reverse the print order
5574 return 0 if (defined $limit && --$limit < 0);
5575 push @$defer, $c;
5576 return 1;
5577 }
5578 if ($r_min != $r_max) {
5579 return 1 if ($r_min < $c->{r});
5580 return 1 if ($r_max > $c->{r});
5581 }
5582 }
5583 return 0 if (defined $limit && --$limit < 0);
5584 show_commit($c);
5585 return 1;
5586}
5587sub show_commit {
5589 my $c = shift;
5590 if ($oneline) {
5591 my $x = "\n";
5592 if (my $l = $c->{l}) {
5593 while ($l->[0] =~ /^\s*$/) { shift @$l }
5594 $x = $l->[0];
5595 }
5596 $l_fmt ||= 'A' . length($c->{r});
5597 print 'r',pack($l_fmt, $c->{r}),' | ';
5598 print "$c->{c} | " if $show_commit;
5599 print $x;
5600 } else {
5601 show_commit_normal($c);
5602 }
5603}
5604sub show_commit_changed_paths {
5606 my ($c) = @_;
5607 return unless $c->{changed};
5608 print "Changed paths:\n", @{$c->{changed}};
5609}
5610sub show_commit_normal {
5612 my ($c) = @_;
5613 print commit_log_separator, "r$c->{r} | ";
5614 print "$c->{c} | " if $show_commit;
5615 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5616 my $nr_line = 0;
5617 if (my $l = $c->{l}) {
5619 while ($l->[$#$l] eq "\n" && $#$l > 0
5620 && $l->[($#$l - 1)] eq "\n") {
5621 pop @$l;
5622 }
5623 $nr_line = scalar @$l;
5624 if (!$nr_line) {
5625 print "1 line\n\n\n";
5626 } else {
5627 if ($nr_line == 1) {
5628 $nr_line = '1 line';
5629 } else {
5630 $nr_line .= ' lines';
5631 }
5632 print $nr_line, "\n";
5633 show_commit_changed_paths($c);
5634 print "\n";
5635 print $_ foreach @$l;
5636 }
5637 } else {
5638 print "1 line\n";
5639 show_commit_changed_paths($c);
5640 print "\n";
5641 }
5643 foreach my $x (qw/raw stat diff/) {
5644 if ($c->{$x}) {
5645 print "\n";
5646 print $_ foreach @{$c->{$x}}
5647 }
5648 }
5649}
5650sub cmd_show_log {
5652 my (@args) = @_;
5653 my ($r_min, $r_max);
5654 my $r_last = -1; # prevent dupes
5655 set_local_timezone();
5656 if (defined $::_revision) {
5657 if ($::_revision =~ /^(\d+):(\d+)$/) {
5658 ($r_min, $r_max) = ($1, $2);
5659 } elsif ($::_revision =~ /^\d+$/) {
5660 $r_min = $r_max = $::_revision;
5661 } else {
5662 ::fatal "-r$::_revision is not supported, use ",
5663 "standard 'git log' arguments instead";
5664 }
5665 }
5666 config_pager();
5668 @args = git_svn_log_cmd($r_min, $r_max, @args);
5669 if (!@args) {
5670 print commit_log_separator unless $incremental || $oneline;
5671 return;
5672 }
5673 my $log = command_output_pipe(@args);
5674 run_pager();
5675 my (@k, $c, $d, $stat);
5676 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5677 while (<$log>) {
5678 if (/^${esc_color}commit -?($::sha1_short)/o) {
5679 my $cmt = $1;
5680 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5681 $r_last = $c->{r};
5682 process_commit($c, $r_min, $r_max, \@k) or
5683 goto out;
5684 }
5685 $d = undef;
5686 $c = { c => $cmt };
5687 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5688 get_author_info($c, $1, $2, $3);
5689 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5690 # ignore
5691 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5692 push @{$c->{raw}}, $_;
5693 } elsif (/^${esc_color}[ACRMDT]\t/) {
5694 # we could add $SVN->{svn_path} here, but that requires
5695 # remote access at the moment (repo_path_split)...
5696 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5697 push @{$c->{changed}}, $_;
5698 } elsif (/^${esc_color}diff /o) {
5699 $d = 1;
5700 push @{$c->{diff}}, $_;
5701 } elsif ($d) {
5702 push @{$c->{diff}}, $_;
5703 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5704 $esc_color*[\+\-]*$esc_color$/x) {
5705 $stat = 1;
5706 push @{$c->{stat}}, $_;
5707 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5708 push @{$c->{stat}}, $_;
5709 $stat = undef;
5710 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5711 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5712 } elsif (s/^${esc_color} //o) {
5713 push @{$c->{l}}, $_;
5714 }
5715 }
5716 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5717 $r_last = $c->{r};
5718 process_commit($c, $r_min, $r_max, \@k);
5719 }
5720 if (@k) {
5721 ($r_min, $r_max) = ($r_max, $r_min);
5722 process_commit($_, $r_min, $r_max) foreach reverse @k;
5723 }
5724out:
5725 close $log;
5726 print commit_log_separator unless $incremental || $oneline;
5727}
5728sub cmd_blame {
5730 my $path = pop;
5731 config_pager();
5733 run_pager();
5734 my ($fh, $ctx, $rev);
5736 if ($_git_format) {
5738 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5739 while (my $line = <$fh>) {
5740 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5741 # Uncommitted edits show up as a rev ID of
5742 # all zeros, which we can't look up with
5743 # cmt_metadata
5744 if ($1 !~ /^0+$/) {
5745 (undef, $rev, undef) =
5746 ::cmt_metadata($1);
5747 $rev = '0' if (!$rev);
5748 } else {
5749 $rev = '0';
5750 }
5751 $rev = sprintf('%-10s', $rev);
5752 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5753 }
5754 print $line;
5755 }
5756 } else {
5757 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5758 '--', $path);
5759 my ($sha1);
5760 my %authors;
5761 my @buffer;
5762 my %dsha; #distinct sha keys
5763 while (my $line = <$fh>) {
5765 push @buffer, $line;
5766 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5767 $dsha{$1} = 1;
5768 }
5769 }
5770 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5772 foreach my $line (@buffer) {
5774 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5775 $rev = $s2r->{$1};
5776 $rev = '0' if (!$rev)
5777 }
5778 elsif ($line =~ /^author (.*)/) {
5779 $authors{$rev} = $1;
5780 $authors{$rev} =~ s/\s/_/g;
5781 }
5782 elsif ($line =~ /^\t(.*)$/) {
5783 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5784 }
5785 }
5786 }
5787 command_close_pipe($fh, $ctx);
5788}
5789package Git::SVN::Migration;
5791# these version numbers do NOT correspond to actual version numbers
5792# of git nor git-svn. They are just relative.
5793#
5794# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5795#
5796# v1 layout: .git/$id/info/url, refs/remotes/$id
5797#
5798# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5799#
5800# v3 layout: .git/svn/$id, refs/remotes/$id
5801# - info/url may remain for backwards compatibility
5802# - this is what we migrate up to this layout automatically,
5803# - this will be used by git svn init on single branches
5804# v3.1 layout (auto migrated):
5805# - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5806# for backwards compatibility
5807#
5808# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5809# - this is only created for newly multi-init-ed
5810# repositories. Similar in spirit to the
5811# --use-separate-remotes option in git-clone (now default)
5812# - we do not automatically migrate to this (following
5813# the example set by core git)
5814#
5815# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5816# - newer, more-efficient format that uses 24-bytes per record
5817# with no filler space.
5818# - use xxd -c24 < .rev_map.$UUID to view and debug
5819# - This is a one-way migration, repositories updated to the
5820# new format will not be able to use old git-svn without
5821# rebuilding the .rev_db. Rebuilding the rev_db is not
5822# possible if noMetadata or useSvmProps are set; but should
5823# be no problem for users that use the (sensible) defaults.
5824use strict;
5825use warnings;
5826use Carp qw/croak/;
5827use File::Path qw/mkpath/;
5828use File::Basename qw/dirname basename/;
5829use vars qw/$_minimize/;
5830sub migrate_from_v0 {
5832 my $git_dir = $ENV{GIT_DIR};
5833 return undef unless -d $git_dir;
5834 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5835 my $migrated = 0;
5836 while (<$fh>) {
5837 chomp;
5838 my ($id, $orig_ref) = ($_, $_);
5839 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5840 next unless -f "$git_dir/$id/info/url";
5841 my $new_ref = "refs/remotes/$id";
5842 if (::verify_ref("$new_ref^0")) {
5843 print STDERR "W: $orig_ref is probably an old ",
5844 "branch used by an ancient version of ",
5845 "git-svn.\n",
5846 "However, $new_ref also exists.\n",
5847 "We will not be able ",
5848 "to use this branch until this ",
5849 "ambiguity is resolved.\n";
5850 next;
5851 }
5852 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5853 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5854 command_noisy('update-ref', $new_ref, $orig_ref);
5855 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5856 $migrated++;
5857 }
5858 command_close_pipe($fh, $ctx);
5859 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5860 $migrated;
5861}
5862sub migrate_from_v1 {
5864 my $git_dir = $ENV{GIT_DIR};
5865 my $migrated = 0;
5866 return $migrated unless -d $git_dir;
5867 my $svn_dir = "$git_dir/svn";
5868 # just in case somebody used 'svn' as their $id at some point...
5870 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5871 print STDERR "Migrating from a git-svn v1 layout...\n";
5873 mkpath([$svn_dir]);
5874 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5875 "$svn_dir\n\t(required for this version ",
5876 "($::VERSION) of git-svn) does not exist.\n";
5877 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5878 while (<$fh>) {
5879 my $x = $_;
5880 next unless $x =~ s#^refs/remotes/##;
5881 chomp $x;
5882 next unless -f "$git_dir/$x/info/url";
5883 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5884 next unless $u;
5885 my $dn = dirname("$git_dir/svn/$x");
5886 mkpath([$dn]) unless -d $dn;
5887 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5888 mkpath(["$git_dir/svn/svn"]);
5889 print STDERR " - $git_dir/$x/info => ",
5890 "$git_dir/svn/$x/info\n";
5891 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5892 croak "$!: $x";
5893 # don't worry too much about these, they probably
5894 # don't exist with repos this old (save for index,
5895 # and we can easily regenerate that)
5896 foreach my $f (qw/unhandled.log index .rev_db/) {
5897 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5898 }
5899 } else {
5900 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5901 rename "$git_dir/$x", "$git_dir/svn/$x" or
5902 croak "$!: $x";
5903 }
5904 $migrated++;
5905 }
5906 command_close_pipe($fh, $ctx);
5907 print STDERR "Done migrating from a git-svn v1 layout\n";
5908 $migrated;
5909}
5910sub read_old_urls {
5912 my ($l_map, $pfx, $path) = @_;
5913 my @dir;
5914 foreach (<$path/*>) {
5915 if (-r "$_/info/url") {
5916 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5917 my $ref_id = $pfx . basename $_;
5918 my $url = ::file_to_s("$_/info/url");
5919 $l_map->{$ref_id} = $url;
5920 } elsif (-d $_) {
5921 push @dir, $_;
5922 }
5923 }
5924 foreach (@dir) {
5925 my $x = $_;
5926 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5927 read_old_urls($l_map, $x, $_);
5928 }
5929}
5930sub migrate_from_v2 {
5932 my @cfg = command(qw/config -l/);
5933 return if grep /^svn-remote\..+\.url=/, @cfg;
5934 my %l_map;
5935 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5936 my $migrated = 0;
5937 foreach my $ref_id (sort keys %l_map) {
5939 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5940 if ($@) {
5941 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5942 }
5943 $migrated++;
5944 }
5945 $migrated;
5946}
5947sub minimize_connections {
5949 my $r = Git::SVN::read_all_remotes();
5950 my $new_urls = {};
5951 my $root_repos = {};
5952 foreach my $repo_id (keys %$r) {
5953 my $url = $r->{$repo_id}->{url} or next;
5954 my $fetch = $r->{$repo_id}->{fetch} or next;
5955 my $ra = Git::SVN::Ra->new($url);
5956 # skip existing cases where we already connect to the root
5958 if (($ra->{url} eq $ra->{repos_root}) ||
5959 ($ra->{repos_root} eq $repo_id)) {
5960 $root_repos->{$ra->{url}} = $repo_id;
5961 next;
5962 }
5963 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5965 my $root_path = $ra->{url};
5966 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5967 foreach my $path (keys %$fetch) {
5968 my $ref_id = $fetch->{$path};
5969 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5970 # make sure we can read when connecting to
5972 # a higher level of a repository
5973 my ($last_rev, undef) = $gs->last_rev_commit;
5974 if (!defined $last_rev) {
5975 $last_rev = eval {
5976 $root_ra->get_latest_revnum;
5977 };
5978 next if $@;
5979 }
5980 my $new = $root_path;
5981 $new .= length $path ? "/$path" : '';
5982 eval {
5983 $root_ra->get_log([$new], $last_rev, $last_rev,
5984 0, 0, 1, sub { });
5985 };
5986 next if $@;
5987 $new_urls->{$ra->{repos_root}}->{$new} =
5988 { ref_id => $ref_id,
5989 old_repo_id => $repo_id,
5990 old_path => $path };
5991 }
5992 }
5993 my @emptied;
5995 foreach my $url (keys %$new_urls) {
5996 # see if we can re-use an existing [svn-remote "repo_id"]
5997 # instead of creating a(n ugly) new section:
5998 my $repo_id = $root_repos->{$url} || $url;
5999 my $fetch = $new_urls->{$url};
6001 foreach my $path (keys %$fetch) {
6002 my $x = $fetch->{$path};
6003 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
6004 my $pfx = "svn-remote.$x->{old_repo_id}";
6005 my $old_fetch = quotemeta("$x->{old_path}:".
6007 "$x->{ref_id}");
6008 command_noisy(qw/config --unset/,
6009 "$pfx.fetch", '^'. $old_fetch . '$');
6010 delete $r->{$x->{old_repo_id}}->
6011 {fetch}->{$x->{old_path}};
6012 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6013 command_noisy(qw/config --unset/,
6014 "$pfx.url");
6015 push @emptied, $x->{old_repo_id}
6016 }
6017 }
6018 }
6019 if (@emptied) {
6020 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6021 print STDERR <<EOF;
6022The following [svn-remote] sections in your config file ($file) are empty
6023and can be safely removed:
6024EOF
6025 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6026 }
6027}
6028sub migration_check {
6030 migrate_from_v0();
6031 migrate_from_v1();
6032 migrate_from_v2();
6033 minimize_connections() if $_minimize;
6034}
6035package Git::IndexInfo;
6037use strict;
6038use warnings;
6039use Git qw/command_input_pipe command_close_pipe/;
6040sub new {
6042 my ($class) = @_;
6043 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6044 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6045}
6046sub remove {
6048 my ($self, $path) = @_;
6049 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6050 return ++$self->{nr};
6051 }
6052 undef;
6053}
6054sub update {
6056 my ($self, $mode, $hash, $path) = @_;
6057 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6058 return ++$self->{nr};
6059 }
6060 undef;
6061}
6062sub DESTROY {
6064 my ($self) = @_;
6065 command_close_pipe($self->{gui}, $self->{ctx});
6066}
6067package Git::SVN::GlobSpec;
6069use strict;
6070use warnings;
6071sub new {
6073 my ($class, $glob, $pattern_ok) = @_;
6074 my $re = $glob;
6075 $re =~ s!/+$!!g; # no need for trailing slashes
6076 my (@left, @right, @patterns);
6077 my $state = "left";
6078 my $die_msg = "Only one set of wildcard directories " .
6079 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6080 for my $part (split(m|/|, $glob)) {
6081 if ($part =~ /\*/ && $part ne "*") {
6082 die "Invalid pattern in '$glob': $part\n";
6083 } elsif ($pattern_ok && $part =~ /[{}]/ &&
6084 $part !~ /^\{[^{}]+\}/) {
6085 die "Invalid pattern in '$glob': $part\n";
6086 }
6087 if ($part eq "*") {
6088 die $die_msg if $state eq "right";
6089 $state = "pattern";
6090 push(@patterns, "[^/]*");
6091 } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6092 die $die_msg if $state eq "right";
6093 $state = "pattern";
6094 my $p = quotemeta($1);
6095 $p =~ s/\\,/|/g;
6096 push(@patterns, "(?:$p)");
6097 } else {
6098 if ($state eq "left") {
6099 push(@left, $part);
6100 } else {
6101 push(@right, $part);
6102 $state = "right";
6103 }
6104 }
6105 }
6106 my $depth = @patterns;
6107 if ($depth == 0) {
6108 die "One '*' is needed in glob: '$glob'\n";
6109 }
6110 my $left = join('/', @left);
6111 my $right = join('/', @right);
6112 $re = join('/', @patterns);
6113 $re = join('\/',
6114 grep(length, quotemeta($left), "($re)", quotemeta($right)));
6115 my $left_re = qr/^\/\Q$left\E(\/|$)/;
6116 bless { left => $left, right => $right, left_regex => $left_re,
6117 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6118}
6119sub full_path {
6121 my ($self, $path) = @_;
6122 return (length $self->{left} ? "$self->{left}/" : '') .
6123 $path . (length $self->{right} ? "/$self->{right}" : '');
6124}
6125__END__
6127Data structures:
6129$remotes = { # returned by read_all_remotes()
6132 'svn' => {
6133 # svn-remote.svn.url=https://svn.musicpd.org
6134 url => 'https://svn.musicpd.org',
6135 # svn-remote.svn.fetch=mpd/trunk:trunk
6136 fetch => {
6137 'mpd/trunk' => 'trunk',
6138 },
6139 # svn-remote.svn.tags=mpd/tags/*:tags/*
6140 tags => {
6141 path => {
6142 left => 'mpd/tags',
6143 right => '',
6144 regex => qr!mpd/tags/([^/]+)$!,
6145 glob => 'tags/*',
6146 },
6147 ref => {
6148 left => 'tags',
6149 right => '',
6150 regex => qr!tags/([^/]+)$!,
6151 glob => 'tags/*',
6152 },
6153 }
6154 }
6155};
6156$log_entry hashref as returned by libsvn_log_entry()
6158{
6159 log => 'whitespace-formatted log entry
6160', # trailing newline is preserved
6161 revision => '8', # integer
6162 date => '2004-02-24T17:01:44.108345Z', # commit date
6163 author => 'committer name'
6164};
6165# this is generated by generate_diff();
6168@mods = array of diff-index line hashes, each element represents one line
6169 of diff-index output
6170diff-index line ($m hash)
6172{
6173 mode_a => first column of diff-index output, no leading ':',
6174 mode_b => second column of diff-index output,
6175 sha1_b => sha1sum of the final blob,
6176 chg => change type [MCRADT],
6177 file_a => original file name of a file (iff chg is 'C' or 'R')
6178 file_b => new/current file name of a file (any chg)
6179}
6180;
6181# retval of read_url_paths{,_all}();
6183$l_map = {
6184 # repository root url
6185 'https://svn.musicpd.org' => {
6186 # repository path # GIT_SVN_ID
6187 'mpd/trunk' => 'trunk',
6188 'mpd/tags/0.11.5' => 'tags/0.11.5',
6189 },
6190}
6191Notes:
6193 I don't trust the each() function on unless I created %hash myself
6194 because the internal iterator may not have started at base.