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 command_noisy(qw/update-index --refresh/);
498 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
499 'Cannot dcommit with a dirty index. Commit your changes first, '
500 . "or stash them with `git stash'.\n";
501 $head ||= 'HEAD';
502 my $old_head;
504 if ($head ne 'HEAD') {
505 $old_head = eval {
506 command_oneline([qw/symbolic-ref -q HEAD/])
507 };
508 if ($old_head) {
509 $old_head =~ s{^refs/heads/}{};
510 } else {
511 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
512 }
513 command(['checkout', $head], STDERR => 0);
514 }
515 my @refs;
517 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
518 unless ($gs) {
519 die "Unable to determine upstream SVN information from ",
520 "$head history.\nPerhaps the repository is empty.";
521 }
522 if (defined $_commit_url) {
524 $url = $_commit_url;
525 } else {
526 $url = eval { command_oneline('config', '--get',
527 "svn-remote.$gs->{repo_id}.commiturl") };
528 if (!$url) {
529 $url = $gs->full_url
530 }
531 }
532 my $last_rev = $_revision if defined $_revision;
534 if ($url) {
535 print "Committing to $url ...\n";
536 }
537 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
538 if ($_no_rebase && scalar(@$linear_refs) > 1) {
539 warn "Attempting to commit more than one change while ",
540 "--no-rebase is enabled.\n",
541 "If these changes depend on each other, re-running ",
542 "without --no-rebase may be required."
543 }
544 my $expect_url = $url;
545 Git::SVN::remove_username($expect_url);
546 while (1) {
547 my $d = shift @$linear_refs or last;
548 unless (defined $last_rev) {
549 (undef, $last_rev, undef) = cmt_metadata("$d~1");
550 unless (defined $last_rev) {
551 fatal "Unable to extract revision information ",
552 "from commit $d~1";
553 }
554 }
555 if ($_dry_run) {
556 print "diff-tree $d~1 $d\n";
557 } else {
558 my $cmt_rev;
559 my %ed_opts = ( r => $last_rev,
560 log => get_commit_entry($d)->{log},
561 ra => Git::SVN::Ra->new($url),
562 config => SVN::Core::config_get_config(
563 $Git::SVN::Ra::config_dir
564 ),
565 tree_a => "$d~1",
566 tree_b => $d,
567 editor_cb => sub {
568 print "Committed r$_[0]\n";
569 $cmt_rev = $_[0];
570 },
571 svn_path => '');
572 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
573 print "No changes\n$d~1 == $d\n";
574 } elsif ($parents->{$d} && @{$parents->{$d}}) {
575 $gs->{inject_parents_dcommit}->{$cmt_rev} =
576 $parents->{$d};
577 }
578 $_fetch_all ? $gs->fetch_all : $gs->fetch;
579 $last_rev = $cmt_rev;
580 next if $_no_rebase;
581 # we always want to rebase against the current HEAD,
583 # not any head that was passed to us
584 my @diff = command('diff-tree', $d,
585 $gs->refname, '--');
586 my @finish;
587 if (@diff) {
588 @finish = rebase_cmd();
589 print STDERR "W: $d and ", $gs->refname,
590 " differ, using @finish:\n",
591 join("\n", @diff), "\n";
592 } else {
593 print "No changes between current HEAD and ",
594 $gs->refname,
595 "\nResetting to the latest ",
596 $gs->refname, "\n";
597 @finish = qw/reset --mixed/;
598 }
599 command_noisy(@finish, $gs->refname);
600 if (@diff) {
601 @refs = ();
602 my ($url_, $rev_, $uuid_, $gs_) =
603 working_head_info('HEAD', \@refs);
604 my ($linear_refs_, $parents_) =
605 linearize_history($gs_, \@refs);
606 if (scalar(@$linear_refs) !=
607 scalar(@$linear_refs_)) {
608 fatal "# of revisions changed ",
609 "\nbefore:\n",
610 join("\n", @$linear_refs),
611 "\n\nafter:\n",
612 join("\n", @$linear_refs_), "\n",
613 'If you are attempting to commit ',
614 "merges, try running:\n\t",
615 'git rebase --interactive',
616 '--preserve-merges ',
617 $gs->refname,
618 "\nBefore dcommitting";
619 }
620 if ($url_ ne $expect_url) {
621 if ($url_ eq $gs->metadata_url) {
622 print
623 "Accepting rewritten URL:",
624 " $url_\n";
625 } else {
626 fatal
627 "URL mismatch after rebase:",
628 " $url_ != $expect_url";
629 }
630 }
631 if ($uuid_ ne $uuid) {
632 fatal "uuid mismatch after rebase: ",
633 "$uuid_ != $uuid";
634 }
635 # remap parents
636 my (%p, @l, $i);
637 for ($i = 0; $i < scalar @$linear_refs; $i++) {
638 my $new = $linear_refs_->[$i] or next;
639 $p{$new} =
640 $parents->{$linear_refs->[$i]};
641 push @l, $new;
642 }
643 $parents = \%p;
644 $linear_refs = \@l;
645 }
646 }
647 }
648 if ($old_head) {
650 my $new_head = command_oneline(qw/rev-parse HEAD/);
651 my $new_is_symbolic = eval {
652 command_oneline(qw/symbolic-ref -q HEAD/);
653 };
654 if ($new_is_symbolic) {
655 print "dcommitted the branch ", $head, "\n";
656 } else {
657 print "dcommitted on a detached HEAD because you gave ",
658 "a revision argument.\n",
659 "The rewritten commit is: ", $new_head, "\n";
660 }
661 command(['checkout', $old_head], STDERR => 0);
662 }
663 unlink $gs->{index};
665}
666sub cmd_branch {
668 my ($branch_name, $head) = @_;
669 unless (defined $branch_name && length $branch_name) {
671 die(($_tag ? "tag" : "branch") . " name required\n");
672 }
673 $head ||= 'HEAD';
674 my (undef, $rev, undef, $gs) = working_head_info($head);
676 my $src = $gs->full_url;
677 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
679 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
680 my $glob;
681 if ($#{$allglobs} == 0) {
682 $glob = $allglobs->[0];
683 } else {
684 unless(defined $_branch_dest) {
685 die "Multiple ",
686 $_tag ? "tag" : "branch",
687 " paths defined for Subversion repository.\n",
688 "You must specify where you want to create the ",
689 $_tag ? "tag" : "branch",
690 " with the --destination argument.\n";
691 }
692 foreach my $g (@{$allglobs}) {
693 # SVN::Git::Editor could probably be moved to Git.pm..
694 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
695 if ($_branch_dest =~ /$re/) {
696 $glob = $g;
697 last;
698 }
699 }
700 unless (defined $glob) {
701 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
702 foreach my $g (@{$allglobs}) {
703 $g->{path}->{left} =~ /$dest_re/ or next;
704 if (defined $glob) {
705 die "Ambiguous destination: ",
706 $_branch_dest, "\nmatches both '",
707 $glob->{path}->{left}, "' and '",
708 $g->{path}->{left}, "'\n";
709 }
710 $glob = $g;
711 }
712 unless (defined $glob) {
713 die "Unknown ",
714 $_tag ? "tag" : "branch",
715 " destination $_branch_dest\n";
716 }
717 }
718 }
719 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
720 my $url;
721 if (defined $_commit_url) {
722 $url = $_commit_url;
723 } else {
724 $url = eval { command_oneline('config', '--get',
725 "svn-remote.$gs->{repo_id}.commiturl") };
726 if (!$url) {
727 $url = $remote->{url};
728 }
729 }
730 my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
731 if ($dst =~ /^https:/ && $src =~ /^http:/) {
733 $src=~s/^http:/https:/;
734 }
735 ::_req_svn();
737 my $ctx = SVN::Client->new(
739 auth => Git::SVN::Ra::_auth_providers(),
740 log_msg => sub {
741 ${ $_[0] } = defined $_message
742 ? $_message
743 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
744 . $branch_name;
745 },
746 );
747 eval {
749 $ctx->ls($dst, 'HEAD', 0);
750 } and die "branch ${branch_name} already exists\n";
751 print "Copying ${src} at r${rev} to ${dst}...\n";
753 $ctx->copy($src, $rev, $dst)
754 unless $_dry_run;
755 $gs->fetch_all;
757}
758sub cmd_find_rev {
760 my $revision_or_hash = shift or die "SVN or git revision required ",
761 "as a command-line argument\n";
762 my $result;
763 if ($revision_or_hash =~ /^r\d+$/) {
764 my $head = shift;
765 $head ||= 'HEAD';
766 my @refs;
767 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
768 unless ($gs) {
769 die "Unable to determine upstream SVN information from ",
770 "$head history\n";
771 }
772 my $desired_revision = substr($revision_or_hash, 1);
773 $result = $gs->rev_map_get($desired_revision, $uuid);
774 } else {
775 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
776 $result = $rev;
777 }
778 print "$result\n" if $result;
779}
780sub cmd_rebase {
782 command_noisy(qw/update-index --refresh/);
783 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
784 unless ($gs) {
785 die "Unable to determine upstream SVN information from ",
786 "working tree history\n";
787 }
788 if ($_dry_run) {
789 print "Remote Branch: " . $gs->refname . "\n";
790 print "SVN URL: " . $url . "\n";
791 return;
792 }
793 if (command(qw/diff-index HEAD --/)) {
794 print STDERR "Cannot rebase with uncommited changes:\n";
795 command_noisy('status');
796 exit 1;
797 }
798 unless ($_local) {
799 # rebase will checkout for us, so no need to do it explicitly
800 $_no_checkout = 'true';
801 $_fetch_all ? $gs->fetch_all : $gs->fetch;
802 }
803 command_noisy(rebase_cmd(), $gs->refname);
804 $gs->mkemptydirs;
805}
806sub cmd_show_ignore {
808 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
809 $gs ||= Git::SVN->new;
810 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
811 $gs->prop_walk($gs->{path}, $r, sub {
812 my ($gs, $path, $props) = @_;
813 print STDOUT "\n# $path\n";
814 my $s = $props->{'svn:ignore'} or return;
815 $s =~ s/[\r\n]+/\n/g;
816 $s =~ s/^\n+//;
817 chomp $s;
818 $s =~ s#^#$path#gm;
819 print STDOUT "$s\n";
820 });
821}
822sub cmd_show_externals {
824 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
825 $gs ||= Git::SVN->new;
826 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
827 $gs->prop_walk($gs->{path}, $r, sub {
828 my ($gs, $path, $props) = @_;
829 print STDOUT "\n# $path\n";
830 my $s = $props->{'svn:externals'} or return;
831 $s =~ s/[\r\n]+/\n/g;
832 chomp $s;
833 $s =~ s#^#$path#gm;
834 print STDOUT "$s\n";
835 });
836}
837sub cmd_create_ignore {
839 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
840 $gs ||= Git::SVN->new;
841 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
842 $gs->prop_walk($gs->{path}, $r, sub {
843 my ($gs, $path, $props) = @_;
844 # $path is of the form /path/to/dir/
845 $path = '.' . $path;
846 # SVN can have attributes on empty directories,
847 # which git won't track
848 mkpath([$path]) unless -d $path;
849 my $ignore = $path . '.gitignore';
850 my $s = $props->{'svn:ignore'} or return;
851 open(GITIGNORE, '>', $ignore)
852 or fatal("Failed to open `$ignore' for writing: $!");
853 $s =~ s/[\r\n]+/\n/g;
854 $s =~ s/^\n+//;
855 chomp $s;
856 # Prefix all patterns so that the ignore doesn't apply
857 # to sub-directories.
858 $s =~ s#^#/#gm;
859 print GITIGNORE "$s\n";
860 close(GITIGNORE)
861 or fatal("Failed to close `$ignore': $!");
862 command_noisy('add', '-f', $ignore);
863 });
864}
865sub cmd_mkdirs {
867 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
868 $gs ||= Git::SVN->new;
869 $gs->mkemptydirs($_revision);
870}
871sub canonicalize_path {
873 my ($path) = @_;
874 my $dot_slash_added = 0;
875 if (substr($path, 0, 1) ne "/") {
876 $path = "./" . $path;
877 $dot_slash_added = 1;
878 }
879 # File::Spec->canonpath doesn't collapse x/../y into y (for a
880 # good reason), so let's do this manually.
881 $path =~ s#/+#/#g;
882 $path =~ s#/\.(?:/|$)#/#g;
883 $path =~ s#/[^/]+/\.\.##g;
884 $path =~ s#/$##g;
885 $path =~ s#^\./## if $dot_slash_added;
886 $path =~ s#^/##;
887 $path =~ s#^\.$##;
888 return $path;
889}
890sub canonicalize_url {
892 my ($url) = @_;
893 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
894 return $url;
895}
896# get_svnprops(PATH)
898# ------------------
899# Helper for cmd_propget and cmd_proplist below.
900sub get_svnprops {
901 my $path = shift;
902 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
903 $gs ||= Git::SVN->new;
904 # prefix THE PATH by the sub-directory from which the user
906 # invoked us.
907 $path = $cmd_dir_prefix . $path;
908 fatal("No such file or directory: $path") unless -e $path;
909 my $is_dir = -d $path ? 1 : 0;
910 $path = $gs->{path} . '/' . $path;
911 # canonicalize the path (otherwise libsvn will abort or fail to
913 # find the file)
914 $path = canonicalize_path($path);
915 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
917 my $props;
918 if ($is_dir) {
919 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
920 }
921 else {
922 (undef, $props) = $gs->ra->get_file($path, $r, undef);
923 }
924 return $props;
925}
926# cmd_propget (PROP, PATH)
928# ------------------------
929# Print the SVN property PROP for PATH.
930sub cmd_propget {
931 my ($prop, $path) = @_;
932 $path = '.' if not defined $path;
933 usage(1) if not defined $prop;
934 my $props = get_svnprops($path);
935 if (not defined $props->{$prop}) {
936 fatal("`$path' does not have a `$prop' SVN property.");
937 }
938 print $props->{$prop} . "\n";
939}
940# cmd_proplist (PATH)
942# -------------------
943# Print the list of SVN properties for PATH.
944sub cmd_proplist {
945 my $path = shift;
946 $path = '.' if not defined $path;
947 my $props = get_svnprops($path);
948 print "Properties on '$path':\n";
949 foreach (sort keys %{$props}) {
950 print " $_\n";
951 }
952}
953sub cmd_multi_init {
955 my $url = shift;
956 unless (defined $_trunk || @_branches || @_tags) {
957 usage(1);
958 }
959 $_prefix = '' unless defined $_prefix;
961 if (defined $url) {
962 $url = canonicalize_url($url);
963 init_subdir(@_);
964 }
965 do_git_init_db();
966 if (defined $_trunk) {
967 $_trunk =~ s#^/+##;
968 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
969 # try both old-style and new-style lookups:
970 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
971 unless ($gs_trunk) {
972 my ($trunk_url, $trunk_path) =
973 complete_svn_url($url, $_trunk);
974 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
975 undef, $trunk_ref);
976 }
977 }
978 return unless @_branches || @_tags;
979 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
980 foreach my $path (@_branches) {
981 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
982 }
983 foreach my $path (@_tags) {
984 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
985 }
986}
987sub cmd_multi_fetch {
989 $Git::SVN::no_reuse_existing = undef;
990 my $remotes = Git::SVN::read_all_remotes();
991 foreach my $repo_id (sort keys %$remotes) {
992 if ($remotes->{$repo_id}->{url}) {
993 Git::SVN::fetch_all($repo_id, $remotes);
994 }
995 }
996}
997# this command is special because it requires no metadata
999sub cmd_commit_diff {
1000 my ($ta, $tb, $url) = @_;
1001 my $usage = "Usage: $0 commit-diff -r<revision> ".
1002 "<tree-ish> <tree-ish> [<URL>]";
1003 fatal($usage) if (!defined $ta || !defined $tb);
1004 my $svn_path = '';
1005 if (!defined $url) {
1006 my $gs = eval { Git::SVN->new };
1007 if (!$gs) {
1008 fatal("Needed URL or usable git-svn --id in ",
1009 "the command-line\n", $usage);
1010 }
1011 $url = $gs->{url};
1012 $svn_path = $gs->{path};
1013 }
1014 unless (defined $_revision) {
1015 fatal("-r|--revision is a required argument\n", $usage);
1016 }
1017 if (defined $_message && defined $_file) {
1018 fatal("Both --message/-m and --file/-F specified ",
1019 "for the commit message.\n",
1020 "I have no idea what you mean");
1021 }
1022 if (defined $_file) {
1023 $_message = file_to_s($_file);
1024 } else {
1025 $_message ||= get_commit_entry($tb)->{log};
1026 }
1027 my $ra ||= Git::SVN::Ra->new($url);
1028 my $r = $_revision;
1029 if ($r eq 'HEAD') {
1030 $r = $ra->get_latest_revnum;
1031 } elsif ($r !~ /^\d+$/) {
1032 die "revision argument: $r not understood by git-svn\n";
1033 }
1034 my %ed_opts = ( r => $r,
1035 log => $_message,
1036 ra => $ra,
1037 tree_a => $ta,
1038 tree_b => $tb,
1039 editor_cb => sub { print "Committed r$_[0]\n" },
1040 svn_path => $svn_path );
1041 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1042 print "No changes\n$ta == $tb\n";
1043 }
1044}
1045sub escape_uri_only {
1047 my ($uri) = @_;
1048 my @tmp;
1049 foreach (split m{/}, $uri) {
1050 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1051 push @tmp, $_;
1052 }
1053 join('/', @tmp);
1054}
1055sub escape_url {
1057 my ($url) = @_;
1058 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1059 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1060 $url = "$scheme://$domain$uri";
1061 }
1062 $url;
1063}
1064sub cmd_info {
1066 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1067 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1068 if (exists $_[1]) {
1069 die "Too many arguments specified\n";
1070 }
1071 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1073 if (!$file_type && !$diff_status) {
1075 print STDERR "svn: '$path' is not under version control\n";
1076 exit 1;
1077 }
1078 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1080 unless ($gs) {
1081 die "Unable to determine upstream SVN information from ",
1082 "working tree history\n";
1083 }
1084 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1086 $path = "." if $path eq "";
1087 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1089 if ($_url) {
1091 print escape_url($full_url), "\n";
1092 return;
1093 }
1094 my $result = "Path: $path\n";
1096 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1097 $result .= "URL: " . escape_url($full_url) . "\n";
1098 eval {
1100 my $repos_root = $gs->repos_root;
1101 Git::SVN::remove_username($repos_root);
1102 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1103 };
1104 if ($@) {
1105 $result .= "Repository Root: (offline)\n";
1106 }
1107 ::_req_svn();
1108 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1109 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1110 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1111 $result .= "Node Kind: " .
1113 ($file_type eq "dir" ? "directory" : "file") . "\n";
1114 my $schedule = $diff_status eq "A"
1116 ? "add"
1117 : ($diff_status eq "D" ? "delete" : "normal");
1118 $result .= "Schedule: $schedule\n";
1119 if ($diff_status eq "A") {
1121 print $result, "\n";
1122 return;
1123 }
1124 my ($lc_author, $lc_rev, $lc_date_utc);
1126 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1127 my $log = command_output_pipe(@args);
1128 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1129 while (<$log>) {
1130 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1131 $lc_author = $1;
1132 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1133 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1134 (undef, $lc_rev, undef) = ::extract_metadata($1);
1135 }
1136 }
1137 close $log;
1138 Git::SVN::Log::set_local_timezone();
1140 $result .= "Last Changed Author: $lc_author\n";
1142 $result .= "Last Changed Rev: $lc_rev\n";
1143 $result .= "Last Changed Date: " .
1144 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1145 if ($file_type ne "dir") {
1147 my $text_last_updated_date =
1148 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1149 $result .=
1150 "Text Last Updated: " .
1151 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1152 "\n";
1153 my $checksum;
1154 if ($diff_status eq "D") {
1155 my ($fh, $ctx) =
1156 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1157 if ($file_type eq "link") {
1158 my $file_name = <$fh>;
1159 $checksum = md5sum("link $file_name");
1160 } else {
1161 $checksum = md5sum($fh);
1162 }
1163 command_close_pipe($fh, $ctx);
1164 } elsif ($file_type eq "link") {
1165 my $file_name =
1166 command(qw(cat-file blob), "HEAD:$path");
1167 $checksum =
1168 md5sum("link " . $file_name);
1169 } else {
1170 open FILE, "<", $path or die $!;
1171 $checksum = md5sum(\*FILE);
1172 close FILE or die $!;
1173 }
1174 $result .= "Checksum: " . $checksum . "\n";
1175 }
1176 print $result, "\n";
1178}
1179sub cmd_reset {
1181 my $target = shift || $_revision or die "SVN revision required\n";
1182 $target = $1 if $target =~ /^r(\d+)$/;
1183 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1184 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1185 unless ($gs) {
1186 die "Unable to determine upstream SVN information from ".
1187 "history\n";
1188 }
1189 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1190 die "Cannot find SVN revision $target\n" unless defined($c);
1191 $gs->rev_map_set($r, $c, 'reset', $uuid);
1192 print "r$r = $c ($gs->{ref_id})\n";
1193}
1194sub cmd_gc {
1196 if (!$can_compress) {
1197 warn "Compress::Zlib could not be found; unhandled.log " .
1198 "files will not be compressed.\n";
1199 }
1200 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1201}
1202########################### utility functions #########################
1204sub rebase_cmd {
1206 my @cmd = qw/rebase/;
1207 push @cmd, '-v' if $_verbose;
1208 push @cmd, qw/--merge/ if $_merge;
1209 push @cmd, "--strategy=$_strategy" if $_strategy;
1210 @cmd;
1211}
1212sub post_fetch_checkout {
1214 return if $_no_checkout;
1215 my $gs = $Git::SVN::_head or return;
1216 return if verify_ref('refs/heads/master^0');
1217 # look for "trunk" ref if it exists
1219 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1220 my $fetch = $remote->{fetch};
1221 if ($fetch) {
1222 foreach my $p (keys %$fetch) {
1223 basename($fetch->{$p}) eq 'trunk' or next;
1224 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1225 last;
1226 }
1227 }
1228 my $valid_head = verify_ref('HEAD^0');
1230 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1231 return if ($valid_head || !verify_ref('HEAD^0'));
1232 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1234 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1235 return if -f $index;
1236 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1238 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1239 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1240 print STDERR "Checked out HEAD:\n ",
1241 $gs->full_url, " r", $gs->last_rev, "\n";
1242 $gs->mkemptydirs($gs->last_rev);
1243}
1244sub complete_svn_url {
1246 my ($url, $path) = @_;
1247 $path =~ s#/+$##;
1248 if ($path !~ m#^[a-z\+]+://#) {
1249 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1250 fatal("E: '$path' is not a complete URL ",
1251 "and a separate URL is not specified");
1252 }
1253 return ($url, $path);
1254 }
1255 return ($path, '');
1256}
1257sub complete_url_ls_init {
1259 my ($ra, $repo_path, $switch, $pfx) = @_;
1260 unless ($repo_path) {
1261 print STDERR "W: $switch not specified\n";
1262 return;
1263 }
1264 $repo_path =~ s#/+$##;
1265 if ($repo_path =~ m#^[a-z\+]+://#) {
1266 $ra = Git::SVN::Ra->new($repo_path);
1267 $repo_path = '';
1268 } else {
1269 $repo_path =~ s#^/+##;
1270 unless ($ra) {
1271 fatal("E: '$repo_path' is not a complete URL ",
1272 "and a separate URL is not specified");
1273 }
1274 }
1275 my $url = $ra->{url};
1276 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1277 my $k = "svn-remote.$gs->{repo_id}.url";
1278 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1279 if ($orig_url && ($orig_url ne $gs->{url})) {
1280 die "$k already set: $orig_url\n",
1281 "wanted to set to: $gs->{url}\n";
1282 }
1283 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1284 my $remote_path = "$gs->{path}/$repo_path";
1285 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1286 $remote_path =~ s#/+#/#g;
1287 $remote_path =~ s#^/##g;
1288 $remote_path .= "/*" if $remote_path !~ /\*/;
1289 my ($n) = ($switch =~ /^--(\w+)/);
1290 if (length $pfx && $pfx !~ m#/$#) {
1291 die "--prefix='$pfx' must have a trailing slash '/'\n";
1292 }
1293 command_noisy('config',
1294 '--add',
1295 "svn-remote.$gs->{repo_id}.$n",
1296 "$remote_path:refs/remotes/$pfx*" .
1297 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1298}
1299sub verify_ref {
1301 my ($ref) = @_;
1302 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1303 { STDERR => 0 }); };
1304}
1305sub get_tree_from_treeish {
1307 my ($treeish) = @_;
1308 # $treeish can be a symbolic ref, too:
1309 my $type = command_oneline(qw/cat-file -t/, $treeish);
1310 my $expected;
1311 while ($type eq 'tag') {
1312 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1313 }
1314 if ($type eq 'commit') {
1315 $expected = (grep /^tree /, command(qw/cat-file commit/,
1316 $treeish))[0];
1317 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1318 die "Unable to get tree from $treeish\n" unless $expected;
1319 } elsif ($type eq 'tree') {
1320 $expected = $treeish;
1321 } else {
1322 die "$treeish is a $type, expected tree, tag or commit\n";
1323 }
1324 return $expected;
1325}
1326sub get_commit_entry {
1328 my ($treeish) = shift;
1329 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1330 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1331 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1332 open my $log_fh, '>', $commit_editmsg or croak $!;
1333 my $type = command_oneline(qw/cat-file -t/, $treeish);
1335 if ($type eq 'commit' || $type eq 'tag') {
1336 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1337 $type, $treeish);
1338 my $in_msg = 0;
1339 my $author;
1340 my $saw_from = 0;
1341 my $msgbuf = "";
1342 while (<$msg_fh>) {
1343 if (!$in_msg) {
1344 $in_msg = 1 if (/^\s*$/);
1345 $author = $1 if (/^author (.*>)/);
1346 } elsif (/^git-svn-id: /) {
1347 # skip this for now, we regenerate the
1348 # correct one on re-fetch anyways
1349 # TODO: set *:merge properties or like...
1350 } else {
1351 if (/^From:/ || /^Signed-off-by:/) {
1352 $saw_from = 1;
1353 }
1354 $msgbuf .= $_;
1355 }
1356 }
1357 $msgbuf =~ s/\s+$//s;
1358 if ($Git::SVN::_add_author_from && defined($author)
1359 && !$saw_from) {
1360 $msgbuf .= "\n\nFrom: $author";
1361 }
1362 print $log_fh $msgbuf or croak $!;
1363 command_close_pipe($msg_fh, $ctx);
1364 }
1365 close $log_fh or croak $!;
1366 if ($_edit || ($type eq 'tree')) {
1368 chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1369 system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1370 }
1371 rename $commit_editmsg, $commit_msg or croak $!;
1372 {
1373 require Encode;
1374 # SVN requires messages to be UTF-8 when entering the repo
1375 local $/;
1376 open $log_fh, '<', $commit_msg or croak $!;
1377 binmode $log_fh;
1378 chomp($log_entry{log} = <$log_fh>);
1379 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1381 my $msg = $log_entry{log};
1382 eval { $msg = Encode::decode($enc, $msg, 1) };
1384 if ($@) {
1385 die "Could not decode as $enc:\n", $msg,
1386 "\nPerhaps you need to set i18n.commitencoding\n";
1387 }
1388 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1390 die "Could not encode as UTF-8:\n$msg\n" if $@;
1391 $log_entry{log} = $msg;
1393 close $log_fh or croak $!;
1395 }
1396 unlink $commit_msg;
1397 \%log_entry;
1398}
1399sub s_to_file {
1401 my ($str, $file, $mode) = @_;
1402 open my $fd,'>',$file or croak $!;
1403 print $fd $str,"\n" or croak $!;
1404 close $fd or croak $!;
1405 chmod ($mode &~ umask, $file) if (defined $mode);
1406}
1407sub file_to_s {
1409 my $file = shift;
1410 open my $fd,'<',$file or croak "$!: file: $file\n";
1411 local $/;
1412 my $ret = <$fd>;
1413 close $fd or croak $!;
1414 $ret =~ s/\s*$//s;
1415 return $ret;
1416}
1417# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1419sub load_authors {
1420 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1421 my $log = $cmd eq 'log';
1422 while (<$authors>) {
1423 chomp;
1424 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1425 my ($user, $name, $email) = ($1, $2, $3);
1426 if ($log) {
1427 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1428 } else {
1429 $users{$user} = [$name, $email];
1430 }
1431 }
1432 close $authors or croak $!;
1433}
1434# convert GetOpt::Long specs for use by git-config
1436sub read_git_config {
1437 my $opts = shift;
1438 my @config_only;
1439 foreach my $o (keys %$opts) {
1440 # if we have mixedCase and a long option-only, then
1441 # it's a config-only variable that we don't need for
1442 # the command-line.
1443 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1444 my $v = $opts->{$o};
1445 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1446 $key =~ s/-//g;
1447 my $arg = 'git config';
1448 $arg .= ' --int' if ($o =~ /[:=]i$/);
1449 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1450 if (ref $v eq 'ARRAY') {
1451 chomp(my @tmp = `$arg --get-all svn.$key`);
1452 @$v = @tmp if @tmp;
1453 } else {
1454 chomp(my $tmp = `$arg --get svn.$key`);
1455 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1456 $$v = $tmp;
1457 }
1458 }
1459 }
1460 delete @$opts{@config_only} if @config_only;
1461}
1462sub extract_metadata {
1464 my $id = shift or return (undef, undef, undef);
1465 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1466 \s([a-f\d\-]+)$/ix);
1467 if (!defined $rev || !$uuid || !$url) {
1468 # some of the original repositories I made had
1469 # identifiers like this:
1470 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1471 }
1472 return ($url, $rev, $uuid);
1473}
1474sub cmt_metadata {
1476 return extract_metadata((grep(/^git-svn-id: /,
1477 command(qw/cat-file commit/, shift)))[-1]);
1478}
1479sub cmt_sha2rev_batch {
1481 my %s2r;
1482 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1483 my $list = shift;
1484 foreach my $sha (@{$list}) {
1486 my $first = 1;
1487 my $size = 0;
1488 print $out $sha, "\n";
1489 while (my $line = <$in>) {
1491 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1492 last;
1493 } elsif ($first &&
1494 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1495 $first = 0;
1496 $size = $1;
1497 next;
1498 } elsif ($line =~ /^(git-svn-id: )/) {
1499 my (undef, $rev, undef) =
1500 extract_metadata($line);
1501 $s2r{$sha} = $rev;
1502 }
1503 $size -= length($line);
1505 last if ($size == 0);
1506 }
1507 }
1508 command_close_bidi_pipe($pid, $in, $out, $ctx);
1510 return \%s2r;
1512}
1513sub working_head_info {
1515 my ($head, $refs) = @_;
1516 my @args = qw/log --no-color --no-decorate --first-parent
1517 --pretty=medium/;
1518 my ($fh, $ctx) = command_output_pipe(@args, $head);
1519 my $hash;
1520 my %max;
1521 while (<$fh>) {
1522 if ( m{^commit ($::sha1)$} ) {
1523 unshift @$refs, $hash if $hash and $refs;
1524 $hash = $1;
1525 next;
1526 }
1527 next unless s{^\s*(git-svn-id:)}{$1};
1528 my ($url, $rev, $uuid) = extract_metadata($_);
1529 if (defined $url && defined $rev) {
1530 next if $max{$url} and $max{$url} < $rev;
1531 if (my $gs = Git::SVN->find_by_url($url)) {
1532 my $c = $gs->rev_map_get($rev, $uuid);
1533 if ($c && $c eq $hash) {
1534 close $fh; # break the pipe
1535 return ($url, $rev, $uuid, $gs);
1536 } else {
1537 $max{$url} ||= $gs->rev_map_max;
1538 }
1539 }
1540 }
1541 }
1542 command_close_pipe($fh, $ctx);
1543 (undef, undef, undef, undef);
1544}
1545sub read_commit_parents {
1547 my ($parents, $c) = @_;
1548 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1549 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1550 @{$parents->{$c}} = split(/ /, $p);
1551}
1552sub linearize_history {
1554 my ($gs, $refs) = @_;
1555 my %parents;
1556 foreach my $c (@$refs) {
1557 read_commit_parents(\%parents, $c);
1558 }
1559 my @linear_refs;
1561 my %skip = ();
1562 my $last_svn_commit = $gs->last_commit;
1563 foreach my $c (reverse @$refs) {
1564 next if $c eq $last_svn_commit;
1565 last if $skip{$c};
1566 unshift @linear_refs, $c;
1568 $skip{$c} = 1;
1569 # we only want the first parent to diff against for linear
1571 # history, we save the rest to inject when we finalize the
1572 # svn commit
1573 my $fp_a = verify_ref("$c~1");
1574 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1575 if (!$fp_a || !$fp_b) {
1576 die "Commit $c\n",
1577 "has no parent commit, and therefore ",
1578 "nothing to diff against.\n",
1579 "You should be working from a repository ",
1580 "originally created by git-svn\n";
1581 }
1582 if ($fp_a ne $fp_b) {
1583 die "$c~1 = $fp_a, however parsing commit $c ",
1584 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1585 }
1586 foreach my $p (@{$parents{$c}}) {
1588 $skip{$p} = 1;
1589 }
1590 }
1591 (\@linear_refs, \%parents);
1592}
1593sub find_file_type_and_diff_status {
1595 my ($path) = @_;
1596 return ('dir', '') if $path eq '';
1597 my $diff_output =
1599 command_oneline(qw(diff --cached --name-status --), $path) || "";
1600 my $diff_status = (split(' ', $diff_output))[0] || "";
1601 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1603 return (undef, undef) if !$diff_status && !$ls_tree;
1605 if ($diff_status eq "A") {
1607 return ("link", $diff_status) if -l $path;
1608 return ("dir", $diff_status) if -d $path;
1609 return ("file", $diff_status);
1610 }
1611 my $mode = (split(' ', $ls_tree))[0] || "";
1613 return ("link", $diff_status) if $mode eq "120000";
1615 return ("dir", $diff_status) if $mode eq "040000";
1616 return ("file", $diff_status);
1617}
1618sub md5sum {
1620 my $arg = shift;
1621 my $ref = ref $arg;
1622 my $md5 = Digest::MD5->new();
1623 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1624 $md5->addfile($arg) or croak $!;
1625 } elsif ($ref eq 'SCALAR') {
1626 $md5->add($$arg) or croak $!;
1627 } elsif (!$ref) {
1628 $md5->add($arg) or croak $!;
1629 } else {
1630 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1631 }
1632 return $md5->hexdigest();
1633}
1634sub gc_directory {
1636 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1637 my $out_filename = $_ . ".gz";
1638 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1639 binmode $in_fh;
1640 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1641 die "Unable to open $out_filename: $!\n";
1642 my $res;
1644 while ($res = sysread($in_fh, my $str, 1024)) {
1645 $gz->gzwrite($str) or
1646 die "Unable to write: ".$gz->gzerror()."!\n";
1647 }
1648 unlink $_ or die "unlink $File::Find::name: $!\n";
1649 } elsif (-f $_ && basename($_) eq "index") {
1650 unlink $_ or die "unlink $_: $!\n";
1651 }
1652}
1653package Git::SVN;
1655use strict;
1656use warnings;
1657use Fcntl qw/:DEFAULT :seek/;
1658use constant rev_map_fmt => 'NH40';
1659use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1660 $_repack $_repack_flags $_use_svm_props $_head
1661 $_use_svnsync_props $no_reuse_existing $_minimize_url
1662 $_use_log_author $_add_author_from $_localtime/;
1663use Carp qw/croak/;
1664use File::Path qw/mkpath/;
1665use File::Copy qw/copy/;
1666use IPC::Open3;
1667use Memoize; # core since 5.8.0, Jul 2002
1668use Memoize::Storable;
1669my ($_gc_nr, $_gc_period);
1671# properties that we do not log:
1673my %SKIP_PROP;
1674BEGIN {
1675 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1676 svn:special svn:executable
1677 svn:entry:committed-rev
1678 svn:entry:last-author
1679 svn:entry:uuid
1680 svn:entry:committed-date/;
1681 # some options are read globally, but can be overridden locally
1683 # per [svn-remote "..."] section. Command-line options will *NOT*
1684 # override options set in an [svn-remote "..."] section
1685 no strict 'refs';
1686 for my $option (qw/follow_parent no_metadata use_svm_props
1687 use_svnsync_props/) {
1688 my $key = $option;
1689 $key =~ tr/_//d;
1690 my $prop = "-$option";
1691 *$option = sub {
1692 my ($self) = @_;
1693 return $self->{$prop} if exists $self->{$prop};
1694 my $k = "svn-remote.$self->{repo_id}.$key";
1695 eval { command_oneline(qw/config --get/, $k) };
1696 if ($@) {
1697 $self->{$prop} = ${"Git::SVN::_$option"};
1698 } else {
1699 my $v = command_oneline(qw/config --bool/,$k);
1700 $self->{$prop} = $v eq 'false' ? 0 : 1;
1701 }
1702 return $self->{$prop};
1703 }
1704 }
1705}
1706my (%LOCKFILES, %INDEX_FILES);
1709END {
1710 unlink keys %LOCKFILES if %LOCKFILES;
1711 unlink keys %INDEX_FILES if %INDEX_FILES;
1712}
1713sub resolve_local_globs {
1715 my ($url, $fetch, $glob_spec) = @_;
1716 return unless defined $glob_spec;
1717 my $ref = $glob_spec->{ref};
1718 my $path = $glob_spec->{path};
1719 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1720 next unless m#^$ref->{regex}$#;
1721 my $p = $1;
1722 my $pathname = desanitize_refname($path->full_path($p));
1723 my $refname = desanitize_refname($ref->full_path($p));
1724 if (my $existing = $fetch->{$pathname}) {
1725 if ($existing ne $refname) {
1726 die "Refspec conflict:\n",
1727 "existing: $existing\n",
1728 " globbed: $refname\n";
1729 }
1730 my $u = (::cmt_metadata("$refname"))[0];
1731 $u =~ s!^\Q$url\E(/|$)!! or die
1732 "$refname: '$url' not found in '$u'\n";
1733 if ($pathname ne $u) {
1734 warn "W: Refspec glob conflict ",
1735 "(ref: $refname):\n",
1736 "expected path: $pathname\n",
1737 " real path: $u\n",
1738 "Continuing ahead with $u\n";
1739 next;
1740 }
1741 } else {
1742 $fetch->{$pathname} = $refname;
1743 }
1744 }
1745}
1746sub parse_revision_argument {
1748 my ($base, $head) = @_;
1749 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1750 return ($base, $head);
1751 }
1752 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1753 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1754 return ($head, $head) if ($::_revision eq 'HEAD');
1755 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1756 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1757 die "revision argument: $::_revision not understood by git-svn\n";
1758}
1759sub fetch_all {
1761 my ($repo_id, $remotes) = @_;
1762 if (ref $repo_id) {
1763 my $gs = $repo_id;
1764 $repo_id = undef;
1765 $repo_id = $gs->{repo_id};
1766 }
1767 $remotes ||= read_all_remotes();
1768 my $remote = $remotes->{$repo_id} or
1769 die "[svn-remote \"$repo_id\"] unknown\n";
1770 my $fetch = $remote->{fetch};
1771 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1772 my (@gs, @globs);
1773 my $ra = Git::SVN::Ra->new($url);
1774 my $uuid = $ra->get_uuid;
1775 my $head = $ra->get_latest_revnum;
1776 # ignore errors, $head revision may not even exist anymore
1778 eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
1779 warn "W: $@\n" if $@;
1780 my $base = defined $fetch ? $head : 0;
1782 # read the max revs for wildcard expansion (branches/*, tags/*)
1784 foreach my $t (qw/branches tags/) {
1785 defined $remote->{$t} or next;
1786 push @globs, @{$remote->{$t}};
1787 my $max_rev = eval { tmp_config(qw/--int --get/,
1789 "svn-remote.$repo_id.${t}-maxRev") };
1790 if (defined $max_rev && ($max_rev < $base)) {
1791 $base = $max_rev;
1792 } elsif (!defined $max_rev) {
1793 $base = 0;
1794 }
1795 }
1796 if ($fetch) {
1798 foreach my $p (sort keys %$fetch) {
1799 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1800 my $lr = $gs->rev_map_max;
1801 if (defined $lr) {
1802 $base = $lr if ($lr < $base);
1803 }
1804 push @gs, $gs;
1805 }
1806 }
1807 ($base, $head) = parse_revision_argument($base, $head);
1809 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1810}
1811sub read_all_remotes {
1813 my $r = {};
1814 my $use_svm_props = eval { command_oneline(qw/config --bool
1815 svn.useSvmProps/) };
1816 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1817 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1818 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1819 if (m!^(.+)\.fetch=$svn_refspec$!) {
1820 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1821 die("svn-remote.$remote: remote ref '$remote_ref' "
1822 . "must start with 'refs/'\n")
1823 unless $remote_ref =~ m{^refs/};
1824 $local_ref = uri_decode($local_ref);
1825 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1826 $r->{$remote}->{svm} = {} if $use_svm_props;
1827 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1828 $r->{$1}->{svm} = {};
1829 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1830 $r->{$1}->{url} = $2;
1831 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1832 my ($remote, $t, $local_ref, $remote_ref) =
1833 ($1, $2, $3, $4);
1834 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1835 . "must start with 'refs/'\n")
1836 unless $remote_ref =~ m{^refs/};
1837 $local_ref = uri_decode($local_ref);
1838 my $rs = {
1839 t => $t,
1840 remote => $remote,
1841 path => Git::SVN::GlobSpec->new($local_ref, 1),
1842 ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
1843 if (length($rs->{ref}->{right}) != 0) {
1844 die "The '*' glob character must be the last ",
1845 "character of '$remote_ref'\n";
1846 }
1847 push @{ $r->{$remote}->{$t} }, $rs;
1848 }
1849 }
1850 map {
1852 if (defined $r->{$_}->{svm}) {
1853 my $svm;
1854 eval {
1855 my $section = "svn-remote.$_";
1856 $svm = {
1857 source => tmp_config('--get',
1858 "$section.svm-source"),
1859 replace => tmp_config('--get',
1860 "$section.svm-replace"),
1861 }
1862 };
1863 $r->{$_}->{svm} = $svm;
1864 }
1865 } keys %$r;
1866 $r;
1868}
1869sub init_vars {
1871 $_gc_nr = $_gc_period = 1000;
1872 if (defined $_repack || defined $_repack_flags) {
1873 warn "Repack options are obsolete; they have no effect.\n";
1874 }
1875}
1876sub verify_remotes_sanity {
1878 return unless -d $ENV{GIT_DIR};
1879 my %seen;
1880 foreach (command(qw/config -l/)) {
1881 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1882 if ($seen{$1}) {
1883 die "Remote ref refs/remote/$1 is tracked by",
1884 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1885 "Please resolve this ambiguity in ",
1886 "your git configuration file before ",
1887 "continuing\n";
1888 }
1889 $seen{$1} = $_;
1890 }
1891 }
1892}
1893sub find_existing_remote {
1895 my ($url, $remotes) = @_;
1896 return undef if $no_reuse_existing;
1897 my $existing;
1898 foreach my $repo_id (keys %$remotes) {
1899 my $u = $remotes->{$repo_id}->{url} or next;
1900 next if $u ne $url;
1901 $existing = $repo_id;
1902 last;
1903 }
1904 $existing;
1905}
1906sub init_remote_config {
1908 my ($self, $url, $no_write) = @_;
1909 $url =~ s!/+$!!; # strip trailing slash
1910 my $r = read_all_remotes();
1911 my $existing = find_existing_remote($url, $r);
1912 if ($existing) {
1913 unless ($no_write) {
1914 print STDERR "Using existing ",
1915 "[svn-remote \"$existing\"]\n";
1916 }
1917 $self->{repo_id} = $existing;
1918 } elsif ($_minimize_url) {
1919 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1920 $existing = find_existing_remote($min_url, $r);
1921 if ($existing) {
1922 unless ($no_write) {
1923 print STDERR "Using existing ",
1924 "[svn-remote \"$existing\"]\n";
1925 }
1926 $self->{repo_id} = $existing;
1927 }
1928 if ($min_url ne $url) {
1929 unless ($no_write) {
1930 print STDERR "Using higher level of URL: ",
1931 "$url => $min_url\n";
1932 }
1933 my $old_path = $self->{path};
1934 $self->{path} = $url;
1935 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1936 if (length $old_path) {
1937 $self->{path} .= "/$old_path";
1938 }
1939 $url = $min_url;
1940 }
1941 }
1942 my $orig_url;
1943 if (!$existing) {
1944 # verify that we aren't overwriting anything:
1945 $orig_url = eval {
1946 command_oneline('config', '--get',
1947 "svn-remote.$self->{repo_id}.url")
1948 };
1949 if ($orig_url && ($orig_url ne $url)) {
1950 die "svn-remote.$self->{repo_id}.url already set: ",
1951 "$orig_url\nwanted to set to: $url\n";
1952 }
1953 }
1954 my ($xrepo_id, $xpath) = find_ref($self->refname);
1955 if (!$no_write && defined $xpath) {
1956 die "svn-remote.$xrepo_id.fetch already set to track ",
1957 "$xpath:", $self->refname, "\n";
1958 }
1959 unless ($no_write) {
1960 command_noisy('config',
1961 "svn-remote.$self->{repo_id}.url", $url);
1962 $self->{path} =~ s{^/}{};
1963 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1964 command_noisy('config', '--add',
1965 "svn-remote.$self->{repo_id}.fetch",
1966 "$self->{path}:".$self->refname);
1967 }
1968 $self->{url} = $url;
1969}
1970sub find_by_url { # repos_root and, path are optional
1972 my ($class, $full_url, $repos_root, $path) = @_;
1973 return undef unless defined $full_url;
1975 remove_username($full_url);
1976 remove_username($repos_root) if defined $repos_root;
1977 my $remotes = read_all_remotes();
1978 if (defined $full_url && defined $repos_root && !defined $path) {
1979 $path = $full_url;
1980 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1981 }
1982 foreach my $repo_id (keys %$remotes) {
1983 my $u = $remotes->{$repo_id}->{url} or next;
1984 remove_username($u);
1985 next if defined $repos_root && $repos_root ne $u;
1986 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1988 foreach my $t (qw/branches tags/) {
1989 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1990 resolve_local_globs($u, $fetch, $globspec);
1991 }
1992 }
1993 my $p = $path;
1994 my $rwr = rewrite_root({repo_id => $repo_id});
1995 my $svm = $remotes->{$repo_id}->{svm}
1996 if defined $remotes->{$repo_id}->{svm};
1997 unless (defined $p) {
1998 $p = $full_url;
1999 my $z = $u;
2000 my $prefix = '';
2001 if ($rwr) {
2002 $z = $rwr;
2003 remove_username($z);
2004 } elsif (defined $svm) {
2005 $z = $svm->{source};
2006 $prefix = $svm->{replace};
2007 $prefix =~ s#^\Q$u\E(?:/|$)##;
2008 $prefix =~ s#/$##;
2009 }
2010 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2011 }
2012 foreach my $f (keys %$fetch) {
2013 next if $f ne $p;
2014 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2015 }
2016 }
2017 undef;
2018}
2019sub init {
2021 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2022 my $self = _new($class, $repo_id, $ref_id, $path);
2023 if (defined $url) {
2024 $self->init_remote_config($url, $no_write);
2025 }
2026 $self;
2027}
2028sub find_ref {
2030 my ($ref_id) = @_;
2031 foreach (command(qw/config -l/)) {
2032 next unless m!^svn-remote\.(.+)\.fetch=
2033 \s*(.*?)\s*:\s*(.+?)\s*$!x;
2034 my ($repo_id, $path, $ref) = ($1, $2, $3);
2035 if ($ref eq $ref_id) {
2036 $path = '' if ($path =~ m#^\./?#);
2037 return ($repo_id, $path);
2038 }
2039 }
2040 (undef, undef, undef);
2041}
2042sub new {
2044 my ($class, $ref_id, $repo_id, $path) = @_;
2045 if (defined $ref_id && !defined $repo_id && !defined $path) {
2046 ($repo_id, $path) = find_ref($ref_id);
2047 if (!defined $repo_id) {
2048 die "Could not find a \"svn-remote.*.fetch\" key ",
2049 "in the repository configuration matching: ",
2050 "$ref_id\n";
2051 }
2052 }
2053 my $self = _new($class, $repo_id, $ref_id, $path);
2054 if (!defined $self->{path} || !length $self->{path}) {
2055 my $fetch = command_oneline('config', '--get',
2056 "svn-remote.$repo_id.fetch",
2057 ":$ref_id\$") or
2058 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2059 "\":$ref_id\$\" in config\n";
2060 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2061 }
2062 $self->{path} =~ s{/+}{/}g;
2063 $self->{path} =~ s{\A/}{};
2064 $self->{path} =~ s{/\z}{};
2065 $self->{url} = command_oneline('config', '--get',
2066 "svn-remote.$repo_id.url") or
2067 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2068 $self->rebuild;
2069 $self;
2070}
2071sub refname {
2073 my ($refname) = $_[0]->{ref_id} ;
2074 # It cannot end with a slash /, we'll throw up on this because
2076 # SVN can't have directories with a slash in their name, either:
2077 if ($refname =~ m{/$}) {
2078 die "ref: '$refname' ends with a trailing slash, this is ",
2079 "not permitted by git nor Subversion\n";
2080 }
2081 # It cannot have ASCII control character space, tilde ~, caret ^,
2083 # colon :, question-mark ?, asterisk *, space, or open bracket [
2084 # anywhere.
2085 #
2086 # Additionally, % must be escaped because it is used for escaping
2087 # and we want our escaped refname to be reversible
2088 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2089 # no slash-separated component can begin with a dot .
2091 # /.* becomes /%2E*
2092 $refname =~ s{/\.}{/%2E}g;
2093 # It cannot have two consecutive dots .. anywhere
2095 # .. becomes %2E%2E
2096 $refname =~ s{\.\.}{%2E%2E}g;
2097 # trailing dots and .lock are not allowed
2099 # .$ becomes %2E and .lock becomes %2Elock
2100 $refname =~ s{\.(?=$|lock$)}{%2E};
2101 # the sequence @{ is used to access the reflog
2103 # @{ becomes %40{
2104 $refname =~ s{\@\{}{%40\{}g;
2105 return $refname;
2107}
2108sub desanitize_refname {
2110 my ($refname) = @_;
2111 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2112 return $refname;
2113}
2114sub svm_uuid {
2116 my ($self) = @_;
2117 return $self->{svm}->{uuid} if $self->svm;
2118 $self->ra;
2119 unless ($self->{svm}) {
2120 die "SVM UUID not cached, and reading remotely failed\n";
2121 }
2122 $self->{svm}->{uuid};
2123}
2124sub svm {
2126 my ($self) = @_;
2127 return $self->{svm} if $self->{svm};
2128 my $svm;
2129 # see if we have it in our config, first:
2130 eval {
2131 my $section = "svn-remote.$self->{repo_id}";
2132 $svm = {
2133 source => tmp_config('--get', "$section.svm-source"),
2134 uuid => tmp_config('--get', "$section.svm-uuid"),
2135 replace => tmp_config('--get', "$section.svm-replace"),
2136 }
2137 };
2138 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2139 $self->{svm} = $svm;
2140 }
2141 $self->{svm};
2142}
2143sub _set_svm_vars {
2145 my ($self, $ra) = @_;
2146 return $ra if $self->svm;
2147 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2149 "(svm:source, svm:uuid) ",
2150 "from the following URLs:\n" );
2151 sub read_svm_props {
2152 my ($self, $ra, $path, $r) = @_;
2153 my $props = ($ra->get_dir($path, $r))[2];
2154 my $src = $props->{'svm:source'};
2155 my $uuid = $props->{'svm:uuid'};
2156 return undef if (!$src || !$uuid);
2157 chomp($src, $uuid);
2159 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2161 or die "doesn't look right - svm:uuid is '$uuid'\n";
2162 # the '!' is used to mark the repos_root!/relative/path
2164 $src =~ s{/?!/?}{/};
2165 $src =~ s{/+$}{}; # no trailing slashes please
2166 # username is of no interest
2167 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2168 my $replace = $ra->{url};
2170 $replace .= "/$path" if length $path;
2171 my $section = "svn-remote.$self->{repo_id}";
2173 tmp_config("$section.svm-source", $src);
2174 tmp_config("$section.svm-replace", $replace);
2175 tmp_config("$section.svm-uuid", $uuid);
2176 $self->{svm} = {
2177 source => $src,
2178 uuid => $uuid,
2179 replace => $replace
2180 };
2181 }
2182 my $r = $ra->get_latest_revnum;
2184 my $path = $self->{path};
2185 my %tried;
2186 while (length $path) {
2187 unless ($tried{"$self->{url}/$path"}) {
2188 return $ra if $self->read_svm_props($ra, $path, $r);
2189 $tried{"$self->{url}/$path"} = 1;
2190 }
2191 $path =~ s#/?[^/]+$##;
2192 }
2193 die "Path: '$path' should be ''\n" if $path ne '';
2194 return $ra if $self->read_svm_props($ra, $path, $r);
2195 $tried{"$self->{url}/$path"} = 1;
2196 if ($ra->{repos_root} eq $self->{url}) {
2198 die @err, (map { " $_\n" } keys %tried), "\n";
2199 }
2200 # nope, make sure we're connected to the repository root:
2202 my $ok;
2203 my @tried_b;
2204 $path = $ra->{svn_path};
2205 $ra = Git::SVN::Ra->new($ra->{repos_root});
2206 while (length $path) {
2207 unless ($tried{"$ra->{url}/$path"}) {
2208 $ok = $self->read_svm_props($ra, $path, $r);
2209 last if $ok;
2210 $tried{"$ra->{url}/$path"} = 1;
2211 }
2212 $path =~ s#/?[^/]+$##;
2213 }
2214 die "Path: '$path' should be ''\n" if $path ne '';
2215 $ok ||= $self->read_svm_props($ra, $path, $r);
2216 $tried{"$ra->{url}/$path"} = 1;
2217 if (!$ok) {
2218 die @err, (map { " $_\n" } keys %tried), "\n";
2219 }
2220 Git::SVN::Ra->new($self->{url});
2221}
2222sub svnsync {
2224 my ($self) = @_;
2225 return $self->{svnsync} if $self->{svnsync};
2226 if ($self->no_metadata) {
2228 die "Can't have both 'noMetadata' and ",
2229 "'useSvnsyncProps' options set!\n";
2230 }
2231 if ($self->rewrite_root) {
2232 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2233 "options set!\n";
2234 }
2235 if ($self->rewrite_uuid) {
2236 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2237 "options set!\n";
2238 }
2239 my $svnsync;
2241 # see if we have it in our config, first:
2242 eval {
2243 my $section = "svn-remote.$self->{repo_id}";
2244 my $url = tmp_config('--get', "$section.svnsync-url");
2246 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2247 die "doesn't look right - svn:sync-from-url is '$url'\n";
2248 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2250 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2251 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2252 $svnsync = { url => $url, uuid => $uuid }
2254 };
2255 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2256 return $self->{svnsync} = $svnsync;
2257 }
2258 my $err = "useSvnsyncProps set, but failed to read " .
2260 "svnsync property: svn:sync-from-";
2261 my $rp = $self->ra->rev_proplist(0);
2262 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2264 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2265 die "doesn't look right - svn:sync-from-url is '$url'\n";
2266 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2268 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2269 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2270 my $section = "svn-remote.$self->{repo_id}";
2272 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2273 tmp_config('--add', "$section.svnsync-url", $url);
2274 return $self->{svnsync} = { url => $url, uuid => $uuid };
2275}
2276# this allows us to memoize our SVN::Ra UUID locally and avoid a
2278# remote lookup (useful for 'git svn log').
2279sub ra_uuid {
2280 my ($self) = @_;
2281 unless ($self->{ra_uuid}) {
2282 my $key = "svn-remote.$self->{repo_id}.uuid";
2283 my $uuid = eval { tmp_config('--get', $key) };
2284 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2285 $self->{ra_uuid} = $uuid;
2286 } else {
2287 die "ra_uuid called without URL\n" unless $self->{url};
2288 $self->{ra_uuid} = $self->ra->get_uuid;
2289 tmp_config('--add', $key, $self->{ra_uuid});
2290 }
2291 }
2292 $self->{ra_uuid};
2293}
2294sub _set_repos_root {
2296 my ($self, $repos_root) = @_;
2297 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2298 $repos_root ||= $self->ra->{repos_root};
2299 tmp_config($k, $repos_root);
2300 $repos_root;
2301}
2302sub repos_root {
2304 my ($self) = @_;
2305 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2306 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2307}
2308sub ra {
2310 my ($self) = shift;
2311 my $ra = Git::SVN::Ra->new($self->{url});
2312 $self->_set_repos_root($ra->{repos_root});
2313 if ($self->use_svm_props && !$self->{svm}) {
2314 if ($self->no_metadata) {
2315 die "Can't have both 'noMetadata' and ",
2316 "'useSvmProps' options set!\n";
2317 } elsif ($self->use_svnsync_props) {
2318 die "Can't have both 'useSvnsyncProps' and ",
2319 "'useSvmProps' options set!\n";
2320 }
2321 $ra = $self->_set_svm_vars($ra);
2322 $self->{-want_revprops} = 1;
2323 }
2324 $ra;
2325}
2326# prop_walk(PATH, REV, SUB)
2328# -------------------------
2329# Recursively traverse PATH at revision REV and invoke SUB for each
2330# directory that contains a SVN property. SUB will be invoked as
2331# follows: &SUB(gs, path, props); where `gs' is this instance of
2332# Git::SVN, `path' the path to the directory where the properties
2333# `props' were found. The `path' will be relative to point of checkout,
2334# that is, if url://repo/trunk is the current Git branch, and that
2335# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2336# as `path' (note the trailing `/').
2337sub prop_walk {
2338 my ($self, $path, $rev, $sub) = @_;
2339 $path =~ s#^/##;
2341 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2342 $path =~ s#^/*#/#g;
2343 my $p = $path;
2344 # Strip the irrelevant part of the path.
2345 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2346 # Ensure the path is terminated by a `/'.
2347 $p =~ s#/*$#/#;
2348 # The properties contain all the internal SVN stuff nobody
2350 # (usually) cares about.
2351 my $interesting_props = 0;
2352 foreach (keys %{$props}) {
2353 # If it doesn't start with `svn:', it must be a
2354 # user-defined property.
2355 ++$interesting_props and next if $_ !~ /^svn:/;
2356 # FIXME: Fragile, if SVN adds new public properties,
2357 # this needs to be updated.
2358 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2359 |eol-style|mime-type
2360 |externals|needs-lock)$/x;
2361 }
2362 &$sub($self, $p, $props) if $interesting_props;
2363 foreach (sort keys %$dirent) {
2365 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2366 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2367 }
2368}
2369sub last_rev { ($_[0]->last_rev_commit)[0] }
2371sub last_commit { ($_[0]->last_rev_commit)[1] }
2372# returns the newest SVN revision number and newest commit SHA1
2374sub last_rev_commit {
2375 my ($self) = @_;
2376 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2377 return ($self->{last_rev}, $self->{last_commit});
2378 }
2379 my $c = ::verify_ref($self->refname.'^0');
2380 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2381 my $rev = (::cmt_metadata($c))[1];
2382 if (defined $rev) {
2383 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2384 return ($rev, $c);
2385 }
2386 }
2387 my $map_path = $self->map_path;
2388 unless (-e $map_path) {
2389 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2390 return (undef, undef);
2391 }
2392 my ($rev, $commit) = $self->rev_map_max(1);
2393 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2394 return ($rev, $commit);
2395}
2396sub get_fetch_range {
2398 my ($self, $min, $max) = @_;
2399 $max ||= $self->ra->get_latest_revnum;
2400 $min ||= $self->rev_map_max;
2401 (++$min, $max);
2402}
2403sub tmp_config {
2405 my (@args) = @_;
2406 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2407 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2408 if (! -f $config && -f $old_def_config) {
2409 rename $old_def_config, $config or
2410 die "Failed rename $old_def_config => $config: $!\n";
2411 }
2412 my $old_config = $ENV{GIT_CONFIG};
2413 $ENV{GIT_CONFIG} = $config;
2414 $@ = undef;
2415 my @ret = eval {
2416 unless (-f $config) {
2417 mkfile($config);
2418 open my $fh, '>', $config or
2419 die "Can't open $config: $!\n";
2420 print $fh "; This file is used internally by ",
2421 "git-svn\n" or die
2422 "Couldn't write to $config: $!\n";
2423 print $fh "; You should not have to edit it\n" or
2424 die "Couldn't write to $config: $!\n";
2425 close $fh or die "Couldn't close $config: $!\n";
2426 }
2427 command('config', @args);
2428 };
2429 my $err = $@;
2430 if (defined $old_config) {
2431 $ENV{GIT_CONFIG} = $old_config;
2432 } else {
2433 delete $ENV{GIT_CONFIG};
2434 }
2435 die $err if $err;
2436 wantarray ? @ret : $ret[0];
2437}
2438sub tmp_index_do {
2440 my ($self, $sub) = @_;
2441 my $old_index = $ENV{GIT_INDEX_FILE};
2442 $ENV{GIT_INDEX_FILE} = $self->{index};
2443 $@ = undef;
2444 my @ret = eval {
2445 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2446 mkpath([$dir]) unless -d $dir;
2447 &$sub;
2448 };
2449 my $err = $@;
2450 if (defined $old_index) {
2451 $ENV{GIT_INDEX_FILE} = $old_index;
2452 } else {
2453 delete $ENV{GIT_INDEX_FILE};
2454 }
2455 die $err if $err;
2456 wantarray ? @ret : $ret[0];
2457}
2458sub assert_index_clean {
2460 my ($self, $treeish) = @_;
2461 $self->tmp_index_do(sub {
2463 command_noisy('read-tree', $treeish) unless -e $self->{index};
2464 my $x = command_oneline('write-tree');
2465 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2466 /^tree ($::sha1)/mo);
2467 return if $y eq $x;
2468 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2470 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2471 command_noisy('read-tree', $treeish);
2472 $x = command_oneline('write-tree');
2473 if ($y ne $x) {
2474 ::fatal "trees ($treeish) $y != $x\n",
2475 "Something is seriously wrong...";
2476 }
2477 });
2478}
2479sub get_commit_parents {
2481 my ($self, $log_entry) = @_;
2482 my (%seen, @ret, @tmp);
2483 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2484 if (my $ip = $self->{inject_parents}) {
2485 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2486 push @tmp, $commit;
2487 }
2488 }
2489 if (my $cur = ::verify_ref($self->refname.'^0')) {
2490 push @tmp, $cur;
2491 }
2492 if (my $ipd = $self->{inject_parents_dcommit}) {
2493 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2494 push @tmp, @$commit;
2495 }
2496 }
2497 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2498 while (my $p = shift @tmp) {
2499 next if $seen{$p};
2500 $seen{$p} = 1;
2501 push @ret, $p;
2502 }
2503 @ret;
2504}
2505sub rewrite_root {
2507 my ($self) = @_;
2508 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2509 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2510 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2511 if ($rwr) {
2512 $rwr =~ s#/+$##;
2513 if ($rwr !~ m#^[a-z\+]+://#) {
2514 die "$rwr is not a valid URL (key: $k)\n";
2515 }
2516 }
2517 $self->{-rewrite_root} = $rwr;
2518}
2519sub rewrite_uuid {
2521 my ($self) = @_;
2522 return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2523 my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2524 my $rwid = eval { command_oneline(qw/config --get/, $k) };
2525 if ($rwid) {
2526 $rwid =~ s#/+$##;
2527 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2528 die "$rwid is not a valid UUID (key: $k)\n";
2529 }
2530 }
2531 $self->{-rewrite_uuid} = $rwid;
2532}
2533sub metadata_url {
2535 my ($self) = @_;
2536 ($self->rewrite_root || $self->{url}) .
2537 (length $self->{path} ? '/' . $self->{path} : '');
2538}
2539sub full_url {
2541 my ($self) = @_;
2542 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2543}
2544sub set_commit_header_env {
2547 my ($log_entry) = @_;
2548 my %env;
2549 foreach my $ned (qw/NAME EMAIL DATE/) {
2550 foreach my $ac (qw/AUTHOR COMMITTER/) {
2551 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2552 }
2553 }
2554 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2556 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2557 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2558 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2560 ? $log_entry->{commit_name}
2561 : $log_entry->{name};
2562 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2563 ? $log_entry->{commit_email}
2564 : $log_entry->{email};
2565 \%env;
2566}
2567sub restore_commit_header_env {
2569 my ($env) = @_;
2570 foreach my $ned (qw/NAME EMAIL DATE/) {
2571 foreach my $ac (qw/AUTHOR COMMITTER/) {
2572 my $k = "GIT_${ac}_${ned}";
2573 if (defined $env->{$k}) {
2574 $ENV{$k} = $env->{$k};
2575 } else {
2576 delete $ENV{$k};
2577 }
2578 }
2579 }
2580}
2581sub gc {
2583 command_noisy('gc', '--auto');
2584};
2585sub do_git_commit {
2587 my ($self, $log_entry) = @_;
2588 my $lr = $self->last_rev;
2589 if (defined $lr && $lr >= $log_entry->{revision}) {
2590 die "Last fetched revision of ", $self->refname,
2591 " was r$lr, but we are about to fetch: ",
2592 "r$log_entry->{revision}!\n";
2593 }
2594 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2595 croak "$log_entry->{revision} = $c already exists! ",
2596 "Why are we refetching it?\n";
2597 }
2598 my $old_env = set_commit_header_env($log_entry);
2599 my $tree = $log_entry->{tree};
2600 if (!defined $tree) {
2601 $tree = $self->tmp_index_do(sub {
2602 command_oneline('write-tree') });
2603 }
2604 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2605 my @exec = ('git', 'commit-tree', $tree);
2607 foreach ($self->get_commit_parents($log_entry)) {
2608 push @exec, '-p', $_;
2609 }
2610 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2611 or croak $!;
2612 binmode $msg_fh;
2613 # we always get UTF-8 from SVN, but we may want our commits in
2615 # a different encoding.
2616 if (my $enc = Git::config('i18n.commitencoding')) {
2617 require Encode;
2618 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2619 }
2620 print $msg_fh $log_entry->{log} or croak $!;
2621 restore_commit_header_env($old_env);
2622 unless ($self->no_metadata) {
2623 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2624 or croak $!;
2625 }
2626 $msg_fh->flush == 0 or croak $!;
2627 close $msg_fh or croak $!;
2628 chomp(my $commit = do { local $/; <$out_fh> });
2629 close $out_fh or croak $!;
2630 waitpid $pid, 0;
2631 croak $? if $?;
2632 if ($commit !~ /^$::sha1$/o) {
2633 die "Failed to commit, invalid sha1: $commit\n";
2634 }
2635 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2637 $self->{last_rev} = $log_entry->{revision};
2639 $self->{last_commit} = $commit;
2640 print "r$log_entry->{revision}" unless $::_q > 1;
2641 if (defined $log_entry->{svm_revision}) {
2642 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2643 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2644 0, $self->svm_uuid);
2645 }
2646 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2647 if (--$_gc_nr == 0) {
2648 $_gc_nr = $_gc_period;
2649 gc();
2650 }
2651 return $commit;
2652}
2653sub match_paths {
2655 my ($self, $paths, $r) = @_;
2656 return 1 if $self->{path} eq '';
2657 if (my $path = $paths->{"/$self->{path}"}) {
2658 return ($path->{action} eq 'D') ? 0 : 1;
2659 }
2660 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2661 if (grep /$self->{path_regex}/, keys %$paths) {
2662 return 1;
2663 }
2664 my $c = '';
2665 foreach (split m#/#, $self->{path}) {
2666 $c .= "/$_";
2667 next unless ($paths->{$c} &&
2668 ($paths->{$c}->{action} =~ /^[AR]$/));
2669 if ($self->ra->check_path($self->{path}, $r) ==
2670 $SVN::Node::dir) {
2671 return 1;
2672 }
2673 }
2674 return 0;
2675}
2676sub find_parent_branch {
2678 my ($self, $paths, $rev) = @_;
2679 return undef unless $self->follow_parent;
2680 unless (defined $paths) {
2681 my $err_handler = $SVN::Error::handler;
2682 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2683 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2684 sub { $paths = $_[0] });
2685 $SVN::Error::handler = $err_handler;
2686 }
2687 return undef unless defined $paths;
2688 # look for a parent from another branch:
2690 my @b_path_components = split m#/#, $self->{path};
2691 my @a_path_components;
2692 my $i;
2693 while (@b_path_components) {
2694 $i = $paths->{'/'.join('/', @b_path_components)};
2695 last if $i && defined $i->{copyfrom_path};
2696 unshift(@a_path_components, pop(@b_path_components));
2697 }
2698 return undef unless defined $i && defined $i->{copyfrom_path};
2699 my $branch_from = $i->{copyfrom_path};
2700 if (@a_path_components) {
2701 print STDERR "branch_from: $branch_from => ";
2702 $branch_from .= '/'.join('/', @a_path_components);
2703 print STDERR $branch_from, "\n";
2704 }
2705 my $r = $i->{copyfrom_rev};
2706 my $repos_root = $self->ra->{repos_root};
2707 my $url = $self->ra->{url};
2708 my $new_url = $url . $branch_from;
2709 print STDERR "Found possible branch point: ",
2710 "$new_url => ", $self->full_url, ", $r\n"
2711 unless $::_q > 1;
2712 $branch_from =~ s#^/##;
2713 my $gs = $self->other_gs($new_url, $url,
2714 $branch_from, $r, $self->{ref_id});
2715 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2716 {
2717 my ($base, $head);
2718 if (!defined $r0 || !defined $parent) {
2719 ($base, $head) = parse_revision_argument(0, $r);
2720 } else {
2721 if ($r0 < $r) {
2722 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2723 0, 1, sub { $base = $_[1] - 1 });
2724 }
2725 }
2726 if (defined $base && $base <= $r) {
2727 $gs->fetch($base, $r);
2728 }
2729 ($r0, $parent) = $gs->find_rev_before($r, 1);
2730 }
2731 if (defined $r0 && defined $parent) {
2732 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2733 unless $::_q > 1;
2734 my $ed;
2735 if ($self->ra->can_do_switch) {
2736 $self->assert_index_clean($parent);
2737 print STDERR "Following parent with do_switch\n"
2738 unless $::_q > 1;
2739 # do_switch works with svn/trunk >= r22312, but that
2740 # is not included with SVN 1.4.3 (the latest version
2741 # at the moment), so we can't rely on it
2742 $self->{last_rev} = $r0;
2743 $self->{last_commit} = $parent;
2744 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2745 $gs->ra->gs_do_switch($r0, $rev, $gs,
2746 $self->full_url, $ed)
2747 or die "SVN connection failed somewhere...\n";
2748 } elsif ($self->ra->trees_match($new_url, $r0,
2749 $self->full_url, $rev)) {
2750 print STDERR "Trees match:\n",
2751 " $new_url\@$r0\n",
2752 " ${\$self->full_url}\@$rev\n",
2753 "Following parent with no changes\n"
2754 unless $::_q > 1;
2755 $self->tmp_index_do(sub {
2756 command_noisy('read-tree', $parent);
2757 });
2758 $self->{last_commit} = $parent;
2759 } else {
2760 print STDERR "Following parent with do_update\n"
2761 unless $::_q > 1;
2762 $ed = SVN::Git::Fetcher->new($self);
2763 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2764 or die "SVN connection failed somewhere...\n";
2765 }
2766 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2767 return $self->make_log_entry($rev, [$parent], $ed);
2768 }
2769 return undef;
2770}
2771sub do_fetch {
2773 my ($self, $paths, $rev) = @_;
2774 my $ed;
2775 my ($last_rev, @parents);
2776 if (my $lc = $self->last_commit) {
2777 # we can have a branch that was deleted, then re-added
2778 # under the same name but copied from another path, in
2779 # which case we'll have multiple parents (we don't
2780 # want to break the original ref, nor lose copypath info):
2781 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2782 push @{$log_entry->{parents}}, $lc;
2783 return $log_entry;
2784 }
2785 $ed = SVN::Git::Fetcher->new($self);
2786 $last_rev = $self->{last_rev};
2787 $ed->{c} = $lc;
2788 @parents = ($lc);
2789 } else {
2790 $last_rev = $rev;
2791 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2792 return $log_entry;
2793 }
2794 $ed = SVN::Git::Fetcher->new($self);
2795 }
2796 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2797 die "SVN connection failed somewhere...\n";
2798 }
2799 $self->make_log_entry($rev, \@parents, $ed);
2800}
2801sub mkemptydirs {
2803 my ($self, $r) = @_;
2804 sub scan {
2806 my ($r, $empty_dirs, $line) = @_;
2807 if (defined $r && $line =~ /^r(\d+)$/) {
2808 return 0 if $1 > $r;
2809 } elsif ($line =~ /^ \+empty_dir: (.+)$/) {
2810 $empty_dirs->{$1} = 1;
2811 } elsif ($line =~ /^ \-empty_dir: (.+)$/) {
2812 my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
2813 delete @$empty_dirs{@d};
2814 }
2815 1; # continue
2816 };
2817 my %empty_dirs = ();
2819 my $gz_file = "$self->{dir}/unhandled.log.gz";
2820 if (-f $gz_file) {
2821 if (!$can_compress) {
2822 warn "Compress::Zlib could not be found; ",
2823 "empty directories in $gz_file will not be read\n";
2824 } else {
2825 my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
2826 die "Unable to open $gz_file: $!\n";
2827 my $line;
2828 while ($gz->gzreadline($line) > 0) {
2829 scan($r, \%empty_dirs, $line) or last;
2830 }
2831 $gz->gzclose;
2832 }
2833 }
2834 if (open my $fh, '<', "$self->{dir}/unhandled.log") {
2836 binmode $fh or croak "binmode: $!";
2837 while (<$fh>) {
2838 scan($r, \%empty_dirs, $_) or last;
2839 }
2840 close $fh;
2841 }
2842 my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2844 foreach my $d (sort keys %empty_dirs) {
2845 $d = uri_decode($d);
2846 $d =~ s/$strip//;
2847 next unless length($d);
2848 next if -d $d;
2849 if (-e $d) {
2850 warn "$d exists but is not a directory\n";
2851 } else {
2852 print "creating empty directory: $d\n";
2853 mkpath([$d]);
2854 }
2855 }
2856}
2857sub get_untracked {
2859 my ($self, $ed) = @_;
2860 my @out;
2861 my $h = $ed->{empty};
2862 foreach (sort keys %$h) {
2863 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2864 push @out, " $act: " . uri_encode($_);
2865 warn "W: $act: $_\n";
2866 }
2867 foreach my $t (qw/dir_prop file_prop/) {
2868 $h = $ed->{$t} or next;
2869 foreach my $path (sort keys %$h) {
2870 my $ppath = $path eq '' ? '.' : $path;
2871 foreach my $prop (sort keys %{$h->{$path}}) {
2872 next if $SKIP_PROP{$prop};
2873 my $v = $h->{$path}->{$prop};
2874 my $t_ppath_prop = "$t: " .
2875 uri_encode($ppath) . ' ' .
2876 uri_encode($prop);
2877 if (defined $v) {
2878 push @out, " +$t_ppath_prop " .
2879 uri_encode($v);
2880 } else {
2881 push @out, " -$t_ppath_prop";
2882 }
2883 }
2884 }
2885 }
2886 foreach my $t (qw/absent_file absent_directory/) {
2887 $h = $ed->{$t} or next;
2888 foreach my $parent (sort keys %$h) {
2889 foreach my $path (sort @{$h->{$parent}}) {
2890 push @out, " $t: " .
2891 uri_encode("$parent/$path");
2892 warn "W: $t: $parent/$path ",
2893 "Insufficient permissions?\n";
2894 }
2895 }
2896 }
2897 \@out;
2898}
2899# parse_svn_date(DATE)
2901# --------------------
2902# Given a date (in UTC) from Subversion, return a string in the format
2903# "<TZ Offset> <local date/time>" that Git will use.
2904#
2905# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2906# is true we'll convert it to the local timezone instead.
2907sub parse_svn_date {
2908 my $date = shift || return '+0000 1970-01-01 00:00:00';
2909 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2910 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2911 croak "Unable to parse date: $date\n";
2912 my $parsed_date; # Set next.
2913 if ($Git::SVN::_localtime) {
2915 # Translate the Subversion datetime to an epoch time.
2916 # Begin by switching ourselves to $date's timezone, UTC.
2917 my $old_env_TZ = $ENV{TZ};
2918 $ENV{TZ} = 'UTC';
2919 my $epoch_in_UTC =
2921 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2922 # Determine our local timezone (including DST) at the
2924 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2925 # value of TZ, if any, at the time we were run.
2926 if (defined $Git::SVN::Log::TZ) {
2927 $ENV{TZ} = $Git::SVN::Log::TZ;
2928 } else {
2929 delete $ENV{TZ};
2930 }
2931 my $our_TZ =
2933 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2934 # This converts $epoch_in_UTC into our local timezone.
2936 my ($sec, $min, $hour, $mday, $mon, $year,
2937 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2938 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2940 $our_TZ, $year + 1900, $mon + 1,
2941 $mday, $hour, $min, $sec);
2942 # Reset us to the timezone in effect when we entered
2944 # this routine.
2945 if (defined $old_env_TZ) {
2946 $ENV{TZ} = $old_env_TZ;
2947 } else {
2948 delete $ENV{TZ};
2949 }
2950 } else {
2951 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2952 }
2953 return $parsed_date;
2955}
2956sub other_gs {
2958 my ($self, $new_url, $url,
2959 $branch_from, $r, $old_ref_id) = @_;
2960 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2961 unless ($gs) {
2962 my $ref_id = $old_ref_id;
2963 $ref_id =~ s/\@\d+-*$//;
2964 $ref_id .= "\@$r";
2965 # just grow a tail if we're not unique enough :x
2966 $ref_id .= '-' while find_ref($ref_id);
2967 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2968 if ($u =~ s#^\Q$url\E(/|$)##) {
2969 $p = $u;
2970 $u = $url;
2971 $repo_id = $self->{repo_id};
2972 }
2973 while (1) {
2974 # It is possible to tag two different subdirectories at
2975 # the same revision. If the url for an existing ref
2976 # does not match, we must either find a ref with a
2977 # matching url or create a new ref by growing a tail.
2978 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2979 my (undef, $max_commit) = $gs->rev_map_max(1);
2980 last if (!$max_commit);
2981 my ($url) = ::cmt_metadata($max_commit);
2982 last if ($url eq $gs->full_url);
2983 $ref_id .= '-';
2984 }
2985 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
2986 }
2987 $gs
2988}
2989sub call_authors_prog {
2991 my ($orig_author) = @_;
2992 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
2993 my $author = `$::_authors_prog $orig_author`;
2994 if ($? != 0) {
2995 die "$::_authors_prog failed with exit code $?\n"
2996 }
2997 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2998 my ($name, $email) = ($1, $2);
2999 $email = undef if length $2 == 0;
3000 return [$name, $email];
3001 } else {
3002 die "Author: $orig_author: $::_authors_prog returned "
3003 . "invalid author format: $author\n";
3004 }
3005}
3006sub check_author {
3008 my ($author) = @_;
3009 if (!defined $author || length $author == 0) {
3010 $author = '(no author)';
3011 }
3012 if (!defined $::users{$author}) {
3013 if (defined $::_authors_prog) {
3014 $::users{$author} = call_authors_prog($author);
3015 } elsif (defined $::_authors) {
3016 die "Author: $author not defined in $::_authors file\n";
3017 }
3018 }
3019 $author;
3020}
3021sub find_extra_svk_parents {
3023 my ($self, $ed, $tickets, $parents) = @_;
3024 # aha! svk:merge property changed...
3025 my @tickets = split "\n", $tickets;
3026 my @known_parents;
3027 for my $ticket ( @tickets ) {
3028 my ($uuid, $path, $rev) = split /:/, $ticket;
3029 if ( $uuid eq $self->ra_uuid ) {
3030 my $url = $self->{url};
3031 my $repos_root = $url;
3032 my $branch_from = $path;
3033 $branch_from =~ s{^/}{};
3034 my $gs = $self->other_gs($repos_root."/".$branch_from,
3035 $url,
3036 $branch_from,
3037 $rev,
3038 $self->{ref_id});
3039 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3040 # wahey! we found it, but it might be
3041 # an old one (!)
3042 push @known_parents, [ $rev, $commit ];
3043 }
3044 }
3045 }
3046 # Ordering matters; highest-numbered commit merge tickets
3047 # first, as they may account for later merge ticket additions
3048 # or changes.
3049 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3050 for my $parent ( @known_parents ) {
3051 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3052 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3053 my $new;
3054 while ( <$msg_fh> ) {
3055 $new=1;last;
3056 }
3057 command_close_pipe($msg_fh, $ctx);
3058 if ( $new ) {
3059 print STDERR
3060 "Found merge parent (svk:merge ticket): $parent\n";
3061 push @$parents, $parent;
3062 }
3063 }
3064}
3065sub lookup_svn_merge {
3067 my $uuid = shift;
3068 my $url = shift;
3069 my $merge = shift;
3070 my ($source, $revs) = split ":", $merge;
3072 my $path = $source;
3073 $path =~ s{^/}{};
3074 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3075 if ( !$gs ) {
3076 warn "Couldn't find revmap for $url$source\n";
3077 return;
3078 }
3079 my @ranges = split ",", $revs;
3080 my ($tip, $tip_commit);
3081 my @merged_commit_ranges;
3082 # find the tip
3083 for my $range ( @ranges ) {
3084 my ($bottom, $top) = split "-", $range;
3085 $top ||= $bottom;
3086 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3087 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3088 unless ($top_commit and $bottom_commit) {
3090 warn "W:unknown path/rev in svn:mergeinfo "
3091 ."dirprop: $source:$range\n";
3092 next;
3093 }
3094 push @merged_commit_ranges,
3096 "$bottom_commit^..$top_commit";
3097 if ( !defined $tip or $top > $tip ) {
3099 $tip = $top;
3100 $tip_commit = $top_commit;
3101 }
3102 }
3103 return ($tip_commit, @merged_commit_ranges);
3104}
3105sub _rev_list {
3107 my ($msg_fh, $ctx) = command_output_pipe(
3108 "rev-list", @_,
3109 );
3110 my @rv;
3111 while ( <$msg_fh> ) {
3112 chomp;
3113 push @rv, $_;
3114 }
3115 command_close_pipe($msg_fh, $ctx);
3116 @rv;
3117}
3118sub check_cherry_pick {
3120 my $base = shift;
3121 my $tip = shift;
3122 my $parents = shift;
3123 my @ranges = @_;
3124 my %commits = map { $_ => 1 }
3125 _rev_list("--no-merges", $tip, "--not", $base, @$parents);
3126 for my $range ( @ranges ) {
3127 delete @commits{_rev_list($range)};
3128 }
3129 for my $commit (keys %commits) {
3130 if (has_no_changes($commit)) {
3131 delete $commits{$commit};
3132 }
3133 }
3134 return (keys %commits);
3135}
3136sub has_no_changes {
3138 my $commit = shift;
3139 my @revs = split / /, command_oneline(
3141 qw(rev-list --parents -1 -m), $commit);
3142 # Commits with no parents, e.g. the start of a partial branch,
3144 # have changes by definition.
3145 return 1 if (@revs < 2);
3146 # Commits with multiple parents, e.g a merge, have no changes
3148 # by definition.
3149 return 0 if (@revs > 2);
3150 return (command_oneline("rev-parse", "$commit^{tree}") eq
3152 command_oneline("rev-parse", "$commit~1^{tree}"));
3153}
3154# The GIT_DIR environment variable is not always set until after the command
3156# line arguments are processed, so we can't memoize in a BEGIN block.
3157{
3158 my $memoized = 0;
3159 sub memoize_svn_mergeinfo_functions {
3161 return if $memoized;
3162 $memoized = 1;
3163 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3165 mkpath([$cache_path]) unless -d $cache_path;
3166 tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3168 "$cache_path/lookup_svn_merge.db", 'nstore';
3169 memoize 'lookup_svn_merge',
3170 SCALAR_CACHE => 'FAULT',
3171 LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3172 ;
3173 tie my %check_cherry_pick_cache => 'Memoize::Storable',
3175 "$cache_path/check_cherry_pick.db", 'nstore';
3176 memoize 'check_cherry_pick',
3177 SCALAR_CACHE => 'FAULT',
3178 LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3179 ;
3180 tie my %has_no_changes_cache => 'Memoize::Storable',
3182 "$cache_path/has_no_changes.db", 'nstore';
3183 memoize 'has_no_changes',
3184 SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3185 LIST_CACHE => 'FAULT',
3186 ;
3187 }
3188 sub unmemoize_svn_mergeinfo_functions {
3190 return if not $memoized;
3191 $memoized = 0;
3192 Memoize::unmemoize 'lookup_svn_merge';
3194 Memoize::unmemoize 'check_cherry_pick';
3195 Memoize::unmemoize 'has_no_changes';
3196 }
3197}
3198END {
3200 # Force cache writeout explicitly instead of waiting for
3201 # global destruction to avoid segfault in Storable:
3202 # http://rt.cpan.org/Public/Bug/Display.html?id=36087
3203 unmemoize_svn_mergeinfo_functions();
3204}
3205sub parents_exclude {
3207 my $parents = shift;
3208 my @commits = @_;
3209 return unless @commits;
3210 my @excluded;
3212 my $excluded;
3213 do {
3214 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3215 $excluded = command_oneline(@cmd);
3216 if ( $excluded ) {
3217 my @new;
3218 my $found;
3219 for my $commit ( @commits ) {
3220 if ( $commit eq $excluded ) {
3221 push @excluded, $commit;
3222 $found++;
3223 last;
3224 }
3225 else {
3226 push @new, $commit;
3227 }
3228 }
3229 die "saw commit '$excluded' in rev-list output, "
3230 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3231 unless $found;
3232 @commits = @new;
3233 }
3234 }
3235 while ($excluded and @commits);
3236 return @excluded;
3238}
3239# note: this function should only be called if the various dirprops
3242# have actually changed
3243sub find_extra_svn_parents {
3244 my ($self, $ed, $mergeinfo, $parents) = @_;
3245 # aha! svk:merge property changed...
3246 memoize_svn_mergeinfo_functions();
3248 # We first search for merged tips which are not in our
3250 # history. Then, we figure out which git revisions are in
3251 # that tip, but not this revision. If all of those revisions
3252 # are now marked as merge, we can add the tip as a parent.
3253 my @merges = split "\n", $mergeinfo;
3254 my @merge_tips;
3255 my $url = $self->{url};
3256 my $uuid = $self->ra_uuid;
3257 my %ranges;
3258 for my $merge ( @merges ) {
3259 my ($tip_commit, @ranges) =
3260 lookup_svn_merge( $uuid, $url, $merge );
3261 unless (!$tip_commit or
3262 grep { $_ eq $tip_commit } @$parents ) {
3263 push @merge_tips, $tip_commit;
3264 $ranges{$tip_commit} = \@ranges;
3265 } else {
3266 push @merge_tips, undef;
3267 }
3268 }
3269 my %excluded = map { $_ => 1 }
3271 parents_exclude($parents, grep { defined } @merge_tips);
3272 # check merge tips for new parents
3274 my @new_parents;
3275 for my $merge_tip ( @merge_tips ) {
3276 my $spec = shift @merges;
3277 next unless $merge_tip and $excluded{$merge_tip};
3278 my $ranges = $ranges{$merge_tip};
3280 # check out 'new' tips
3282 my $merge_base;
3283 eval {
3284 $merge_base = command_oneline(
3285 "merge-base",
3286 @$parents, $merge_tip,
3287 );
3288 };
3289 if ($@) {
3290 die "An error occurred during merge-base"
3291 unless $@->isa("Git::Error::Command");
3292 warn "W: Cannot find common ancestor between ".
3294 "@$parents and $merge_tip. Ignoring merge info.\n";
3295 next;
3296 }
3297 # double check that there are no missing non-merge commits
3299 my (@incomplete) = check_cherry_pick(
3300 $merge_base, $merge_tip,
3301 $parents,
3302 @$ranges,
3303 );
3304 if ( @incomplete ) {
3306 warn "W:svn cherry-pick ignored ($spec) - missing "
3307 .@incomplete." commit(s) (eg $incomplete[0])\n";
3308 } else {
3309 warn
3310 "Found merge parent (svn:mergeinfo prop): ",
3311 $merge_tip, "\n";
3312 push @new_parents, $merge_tip;
3313 }
3314 }
3315 # cater for merges which merge commits from multiple branches
3317 if ( @new_parents > 1 ) {
3318 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3319 for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3320 next if $i == $j;
3321 next unless $new_parents[$i];
3322 next unless $new_parents[$j];
3323 my $revs = command_oneline(
3324 "rev-list", "-1",
3325 "$new_parents[$i]..$new_parents[$j]",
3326 );
3327 if ( !$revs ) {
3328 undef($new_parents[$j]);
3329 }
3330 }
3331 }
3332 }
3333 push @$parents, grep { defined } @new_parents;
3334}
3335sub make_log_entry {
3337 my ($self, $rev, $parents, $ed) = @_;
3338 my $untracked = $self->get_untracked($ed);
3339 my @parents = @$parents;
3341 my $ps = $ed->{path_strip} || "";
3342 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3343 my $props = $ed->{dir_prop}{$path};
3344 if ( $props->{"svk:merge"} ) {
3345 $self->find_extra_svk_parents
3346 ($ed, $props->{"svk:merge"}, \@parents);
3347 }
3348 if ( $props->{"svn:mergeinfo"} ) {
3349 $self->find_extra_svn_parents
3350 ($ed,
3351 $props->{"svn:mergeinfo"},
3352 \@parents);
3353 }
3354 }
3355 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3357 print $un "r$rev\n" or croak $!;
3358 print $un $_, "\n" foreach @$untracked;
3359 my %log_entry = ( parents => \@parents, revision => $rev,
3360 log => '');
3361 my $headrev;
3363 my $logged = delete $self->{logged_rev_props};
3364 if (!$logged || $self->{-want_revprops}) {
3365 my $rp = $self->ra->rev_proplist($rev);
3366 foreach (sort keys %$rp) {
3367 my $v = $rp->{$_};
3368 if (/^svn:(author|date|log)$/) {
3369 $log_entry{$1} = $v;
3370 } elsif ($_ eq 'svm:headrev') {
3371 $headrev = $v;
3372 } else {
3373 print $un " rev_prop: ", uri_encode($_), ' ',
3374 uri_encode($v), "\n";
3375 }
3376 }
3377 } else {
3378 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3379 }
3380 close $un or croak $!;
3381 $log_entry{date} = parse_svn_date($log_entry{date});
3383 $log_entry{log} .= "\n";
3384 my $author = $log_entry{author} = check_author($log_entry{author});
3385 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3386 : ($author, undef);
3387 my ($commit_name, $commit_email) = ($name, $email);
3389 if ($_use_log_author) {
3390 my $name_field;
3391 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3392 $name_field = $1;
3393 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3394 $name_field = $1;
3395 }
3396 if (!defined $name_field) {
3397 if (!defined $email) {
3398 $email = $name;
3399 }
3400 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3401 ($name, $email) = ($1, $2);
3402 } elsif ($name_field =~ /(.*)@/) {
3403 ($name, $email) = ($1, $name_field);
3404 } else {
3405 ($name, $email) = ($name_field, $name_field);
3406 }
3407 }
3408 if (defined $headrev && $self->use_svm_props) {
3409 if ($self->rewrite_root) {
3410 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3411 "options set!\n";
3412 }
3413 if ($self->rewrite_uuid) {
3414 die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3415 "options set!\n";
3416 }
3417 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3418 # we don't want "SVM: initializing mirror for junk" ...
3419 return undef if $r == 0;
3420 my $svm = $self->svm;
3421 if ($uuid ne $svm->{uuid}) {
3422 die "UUID mismatch on SVM path:\n",
3423 "expected: $svm->{uuid}\n",
3424 " got: $uuid\n";
3425 }
3426 my $full_url = $self->full_url;
3427 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3428 die "Failed to replace '$svm->{replace}' with ",
3429 "'$svm->{source}' in $full_url\n";
3430 # throw away username for storing in records
3431 remove_username($full_url);
3432 $log_entry{metadata} = "$full_url\@$r $uuid";
3433 $log_entry{svm_revision} = $r;
3434 $email ||= "$author\@$uuid";
3435 $commit_email ||= "$author\@$uuid";
3436 } elsif ($self->use_svnsync_props) {
3437 my $full_url = $self->svnsync->{url};
3438 $full_url .= "/$self->{path}" if length $self->{path};
3439 remove_username($full_url);
3440 my $uuid = $self->svnsync->{uuid};
3441 $log_entry{metadata} = "$full_url\@$rev $uuid";
3442 $email ||= "$author\@$uuid";
3443 $commit_email ||= "$author\@$uuid";
3444 } else {
3445 my $url = $self->metadata_url;
3446 remove_username($url);
3447 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3448 $log_entry{metadata} = "$url\@$rev " . $uuid;
3449 $email ||= "$author\@" . $uuid;
3450 $commit_email ||= "$author\@" . $uuid;
3451 }
3452 $log_entry{name} = $name;
3453 $log_entry{email} = $email;
3454 $log_entry{commit_name} = $commit_name;
3455 $log_entry{commit_email} = $commit_email;
3456 \%log_entry;
3457}
3458sub fetch {
3460 my ($self, $min_rev, $max_rev, @parents) = @_;
3461 my ($last_rev, $last_commit) = $self->last_rev_commit;
3462 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3463 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3464}
3465sub set_tree_cb {
3467 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3468 $self->{inject_parents} = { $rev => $tree };
3469 $self->fetch(undef, undef);
3470}
3471sub set_tree {
3473 my ($self, $tree) = (shift, shift);
3474 my $log_entry = ::get_commit_entry($tree);
3475 unless ($self->{last_rev}) {
3476 ::fatal("Must have an existing revision to commit");
3477 }
3478 my %ed_opts = ( r => $self->{last_rev},
3479 log => $log_entry->{log},
3480 ra => $self->ra,
3481 tree_a => $self->{last_commit},
3482 tree_b => $tree,
3483 editor_cb => sub {
3484 $self->set_tree_cb($log_entry, $tree, @_) },
3485 svn_path => $self->{path} );
3486 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3487 print "No changes\nr$self->{last_rev} = $tree\n";
3488 }
3489}
3490sub rebuild_from_rev_db {
3492 my ($self, $path) = @_;
3493 my $r = -1;
3494 open my $fh, '<', $path or croak "open: $!";
3495 binmode $fh or croak "binmode: $!";
3496 while (<$fh>) {
3497 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3498 chomp($_);
3499 ++$r;
3500 next if $_ eq ('0' x 40);
3501 $self->rev_map_set($r, $_);
3502 print "r$r = $_\n";
3503 }
3504 close $fh or croak "close: $!";
3505 unlink $path or croak "unlink: $!";
3506}
3507sub rebuild {
3509 my ($self) = @_;
3510 my $map_path = $self->map_path;
3511 my $partial = (-e $map_path && ! -z $map_path);
3512 return unless ::verify_ref($self->refname.'^0');
3513 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3514 my $rev_db = $self->rev_db_path;
3515 $self->rebuild_from_rev_db($rev_db);
3516 if ($self->use_svm_props) {
3517 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3518 $self->rebuild_from_rev_db($svm_rev_db);
3519 }
3520 $self->unlink_rev_db_symlink;
3521 return;
3522 }
3523 print "Rebuilding $map_path ...\n" if (!$partial);
3524 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3525 (undef, undef));
3526 my ($log, $ctx) =
3527 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3528 ($head ? "$head.." : "") . $self->refname,
3529 '--');
3530 my $metadata_url = $self->metadata_url;
3531 remove_username($metadata_url);
3532 my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3533 my $c;
3534 while (<$log>) {
3535 if ( m{^commit ($::sha1)$} ) {
3536 $c = $1;
3537 next;
3538 }
3539 next unless s{^\s*(git-svn-id:)}{$1};
3540 my ($url, $rev, $uuid) = ::extract_metadata($_);
3541 remove_username($url);
3542 # ignore merges (from set-tree)
3544 next if (!defined $rev || !$uuid);
3545 # if we merged or otherwise started elsewhere, this is
3547 # how we break out of it
3548 if (($uuid ne $svn_uuid) ||
3549 ($metadata_url && $url && ($url ne $metadata_url))) {
3550 next;
3551 }
3552 if ($partial && $head) {
3553 print "Partial-rebuilding $map_path ...\n";
3554 print "Currently at $base_rev = $head\n";
3555 $head = undef;
3556 }
3557 $self->rev_map_set($rev, $c);
3559 print "r$rev = $c\n";
3560 }
3561 command_close_pipe($log, $ctx);
3562 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3563 my $rev_db_path = $self->rev_db_path;
3564 if (-f $self->rev_db_path) {
3565 unlink $self->rev_db_path or croak "unlink: $!";
3566 }
3567 $self->unlink_rev_db_symlink;
3568}
3569# rev_map:
3571# Tie::File seems to be prone to offset errors if revisions get sparse,
3572# it's not that fast, either. Tie::File is also not in Perl 5.6. So
3573# one of my favorite modules is out :< Next up would be one of the DBM
3574# modules, but I'm not sure which is most portable...
3575#
3576# This is the replacement for the rev_db format, which was too big
3577# and inefficient for large repositories with a lot of sparse history
3578# (mainly tags)
3579#
3580# The format is this:
3581# - 24 bytes for every record,
3582# * 4 bytes for the integer representing an SVN revision number
3583# * 20 bytes representing the sha1 of a git commit
3584# - No empty padding records like the old format
3585# (except the last record, which can be overwritten)
3586# - new records are written append-only since SVN revision numbers
3587# increase monotonically
3588# - lookups on SVN revision number are done via a binary search
3589# - Piping the file to xxd -c24 is a good way of dumping it for
3590# viewing or editing (piped back through xxd -r), should the need
3591# ever arise.
3592# - The last record can be padding revision with an all-zero sha1
3593# This is used to optimize fetch performance when using multiple
3594# "fetch" directives in .git/config
3595#
3596# These files are disposable unless noMetadata or useSvmProps is set
3597sub _rev_map_set {
3599 my ($fh, $rev, $commit) = @_;
3600 binmode $fh or croak "binmode: $!";
3602 my $size = (stat($fh))[7];
3603 ($size % 24) == 0 or croak "inconsistent size: $size";
3604 my $wr_offset = 0;
3606 if ($size > 0) {
3607 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3608 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3609 $read == 24 or croak "read only $read bytes (!= 24)";
3610 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3611 if ($last_commit eq ('0' x40)) {
3612 if ($size >= 48) {
3613 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3614 $read = sysread($fh, $buf, 24) or
3615 croak "read: $!";
3616 $read == 24 or
3617 croak "read only $read bytes (!= 24)";
3618 ($last_rev, $last_commit) =
3619 unpack(rev_map_fmt, $buf);
3620 if ($last_commit eq ('0' x40)) {
3621 croak "inconsistent .rev_map\n";
3622 }
3623 }
3624 if ($last_rev >= $rev) {
3625 croak "last_rev is higher!: $last_rev >= $rev";
3626 }
3627 $wr_offset = -24;
3628 }
3629 }
3630 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3631 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3632 croak "write: $!";
3633}
3634sub _rev_map_reset {
3636 my ($fh, $rev, $commit) = @_;
3637 my $c = _rev_map_get($fh, $rev);
3638 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3639 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3640 truncate $fh, $offset or croak "truncate: $!";
3641}
3642sub mkfile {
3644 my ($path) = @_;
3645 unless (-e $path) {
3646 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3647 mkpath([$dir]) unless -d $dir;
3648 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3649 close $fh or die "Couldn't close (create) $path: $!\n";
3650 }
3651}
3652sub rev_map_set {
3654 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3655 defined $commit or die "missing arg3\n";
3656 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3657 my $db = $self->map_path($uuid);
3658 my $db_lock = "$db.lock";
3659 my $sig;
3660 $update_ref ||= 0;
3661 if ($update_ref) {
3662 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3663 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3664 }
3665 mkfile($db);
3666 $LOCKFILES{$db_lock} = 1;
3668 my $sync;
3669 # both of these options make our .rev_db file very, very important
3670 # and we can't afford to lose it because rebuild() won't work
3671 if ($self->use_svm_props || $self->no_metadata) {
3672 $sync = 1;
3673 copy($db, $db_lock) or die "rev_map_set(@_): ",
3674 "Failed to copy: ",
3675 "$db => $db_lock ($!)\n";
3676 } else {
3677 rename $db, $db_lock or die "rev_map_set(@_): ",
3678 "Failed to rename: ",
3679 "$db => $db_lock ($!)\n";
3680 }
3681 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3683 or croak "Couldn't open $db_lock: $!\n";
3684 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3685 _rev_map_set($fh, $rev, $commit);
3686 if ($sync) {
3687 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3688 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3689 }
3690 close $fh or croak $!;
3691 if ($update_ref) {
3692 $_head = $self;
3693 my $note = "";
3694 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3695 command_noisy('update-ref', '-m', "r$rev$note",
3696 $self->refname, $commit);
3697 }
3698 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3699 "$db_lock => $db ($!)\n";
3700 delete $LOCKFILES{$db_lock};
3701 if ($update_ref) {
3702 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3703 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3704 kill $sig, $$ if defined $sig;
3705 }
3706}
3707# If want_commit, this will return an array of (rev, commit) where
3709# commit _must_ be a valid commit in the archive.
3710# Otherwise, it'll return the max revision (whether or not the
3711# commit is valid or just a 0x40 placeholder).
3712sub rev_map_max {
3713 my ($self, $want_commit) = @_;
3714 $self->rebuild;
3715 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3716 $want_commit ? ($r, $c) : $r;
3717}
3718sub rev_map_max_norebuild {
3720 my ($self, $want_commit) = @_;
3721 my $map_path = $self->map_path;
3722 stat $map_path or return $want_commit ? (0, undef) : 0;
3723 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3724 binmode $fh or croak "binmode: $!";
3725 my $size = (stat($fh))[7];
3726 ($size % 24) == 0 or croak "inconsistent size: $size";
3727 if ($size == 0) {
3729 close $fh or croak "close: $!";
3730 return $want_commit ? (0, undef) : 0;
3731 }
3732 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3734 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3735 my ($r, $c) = unpack(rev_map_fmt, $buf);
3736 if ($want_commit && $c eq ('0' x40)) {
3737 if ($size < 48) {
3738 return $want_commit ? (0, undef) : 0;
3739 }
3740 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3741 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3742 ($r, $c) = unpack(rev_map_fmt, $buf);
3743 if ($c eq ('0'x40)) {
3744 croak "Penultimate record is all-zeroes in $map_path";
3745 }
3746 }
3747 close $fh or croak "close: $!";
3748 $want_commit ? ($r, $c) : $r;
3749}
3750sub rev_map_get {
3752 my ($self, $rev, $uuid) = @_;
3753 my $map_path = $self->map_path($uuid);
3754 return undef unless -e $map_path;
3755 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3757 my $c = _rev_map_get($fh, $rev);
3758 close($fh) or croak "close: $!";
3759 $c
3760}
3761sub _rev_map_get {
3763 my ($fh, $rev) = @_;
3764 binmode $fh or croak "binmode: $!";
3766 my $size = (stat($fh))[7];
3767 ($size % 24) == 0 or croak "inconsistent size: $size";
3768 if ($size == 0) {
3770 return undef;
3771 }
3772 my ($l, $u) = (0, $size - 24);
3774 my ($r, $c, $buf);
3775 while ($l <= $u) {
3777 my $i = int(($l/24 + $u/24) / 2) * 24;
3778 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3779 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3780 my ($r, $c) = unpack(rev_map_fmt, $buf);
3781 if ($r < $rev) {
3783 $l = $i + 24;
3784 } elsif ($r > $rev) {
3785 $u = $i - 24;
3786 } else { # $r == $rev
3787 return $c eq ('0' x 40) ? undef : $c;
3788 }
3789 }
3790 undef;
3791}
3792# Finds the first svn revision that exists on (if $eq_ok is true) or
3794# before $rev for the current branch. It will not search any lower
3795# than $min_rev. Returns the git commit hash and svn revision number
3796# if found, else (undef, undef).
3797sub find_rev_before {
3798 my ($self, $rev, $eq_ok, $min_rev) = @_;
3799 --$rev unless $eq_ok;
3800 $min_rev ||= 1;
3801 my $max_rev = $self->rev_map_max;
3802 $rev = $max_rev if ($rev > $max_rev);
3803 while ($rev >= $min_rev) {
3804 if (my $c = $self->rev_map_get($rev)) {
3805 return ($rev, $c);
3806 }
3807 --$rev;
3808 }
3809 return (undef, undef);
3810}
3811# Finds the first svn revision that exists on (if $eq_ok is true) or
3813# after $rev for the current branch. It will not search any higher
3814# than $max_rev. Returns the git commit hash and svn revision number
3815# if found, else (undef, undef).
3816sub find_rev_after {
3817 my ($self, $rev, $eq_ok, $max_rev) = @_;
3818 ++$rev unless $eq_ok;
3819 $max_rev ||= $self->rev_map_max;
3820 while ($rev <= $max_rev) {
3821 if (my $c = $self->rev_map_get($rev)) {
3822 return ($rev, $c);
3823 }
3824 ++$rev;
3825 }
3826 return (undef, undef);
3827}
3828sub _new {
3830 my ($class, $repo_id, $ref_id, $path) = @_;
3831 unless (defined $repo_id && length $repo_id) {
3832 $repo_id = $Git::SVN::default_repo_id;
3833 }
3834 unless (defined $ref_id && length $ref_id) {
3835 $_prefix = '' unless defined($_prefix);
3836 $_[2] = $ref_id =
3837 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3838 }
3839 $_[1] = $repo_id;
3840 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3841 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3843 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3844 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3845 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3846 if (-d $old_dir && ! -d $dir) {
3847 $dir = $old_dir;
3848 }
3849 }
3850 $_[3] = $path = '' unless (defined $path);
3852 mkpath([$dir]);
3853 bless {
3854 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3855 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3856 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3857}
3858# for read-only access of old .rev_db formats
3860sub unlink_rev_db_symlink {
3861 my ($self) = @_;
3862 my $link = $self->rev_db_path;
3863 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3864 if (-l $link) {
3865 unlink $link or croak "unlink: $link failed!";
3866 }
3867}
3868sub rev_db_path {
3870 my ($self, $uuid) = @_;
3871 my $db_path = $self->map_path($uuid);
3872 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3873 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3874 $db_path;
3875}
3876# the new replacement for .rev_db
3878sub map_path {
3879 my ($self, $uuid) = @_;
3880 $uuid ||= $self->ra_uuid;
3881 "$self->{map_root}.$uuid";
3882}
3883sub uri_encode {
3885 my ($f) = @_;
3886 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3887 $f
3888}
3889sub uri_decode {
3891 my ($f) = @_;
3892 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3893 $f
3894}
3895sub remove_username {
3897 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3898}
3899package Git::SVN::Prompt;
3901use strict;
3902use warnings;
3903require SVN::Core;
3904use vars qw/$_no_auth_cache $_username/;
3905sub simple {
3907 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3908 $may_save = undef if $_no_auth_cache;
3909 $default_username = $_username if defined $_username;
3910 if (defined $default_username && length $default_username) {
3911 if (defined $realm && length $realm) {
3912 print STDERR "Authentication realm: $realm\n";
3913 STDERR->flush;
3914 }
3915 $cred->username($default_username);
3916 } else {
3917 username($cred, $realm, $may_save, $pool);
3918 }
3919 $cred->password(_read_password("Password for '" .
3920 $cred->username . "': ", $realm));
3921 $cred->may_save($may_save);
3922 $SVN::_Core::SVN_NO_ERROR;
3923}
3924sub ssl_server_trust {
3926 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3927 $may_save = undef if $_no_auth_cache;
3928 print STDERR "Error validating server certificate for '$realm':\n";
3929 {
3930 no warnings 'once';
3931 # All variables SVN::Auth::SSL::* are used only once,
3932 # so we're shutting up Perl warnings about this.
3933 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3934 print STDERR " - The certificate is not issued ",
3935 "by a trusted authority. Use the\n",
3936 " fingerprint to validate ",
3937 "the certificate manually!\n";
3938 }
3939 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3940 print STDERR " - The certificate hostname ",
3941 "does not match.\n";
3942 }
3943 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3944 print STDERR " - The certificate is not yet valid.\n";
3945 }
3946 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3947 print STDERR " - The certificate has expired.\n";
3948 }
3949 if ($failures & $SVN::Auth::SSL::OTHER) {
3950 print STDERR " - The certificate has ",
3951 "an unknown error.\n";
3952 }
3953 } # no warnings 'once'
3954 printf STDERR
3955 "Certificate information:\n".
3956 " - Hostname: %s\n".
3957 " - Valid: from %s until %s\n".
3958 " - Issuer: %s\n".
3959 " - Fingerprint: %s\n",
3960 map $cert_info->$_, qw(hostname valid_from valid_until
3961 issuer_dname fingerprint);
3962 my $choice;
3963prompt:
3964 print STDERR $may_save ?
3965 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3966 "(R)eject or accept (t)emporarily? ";
3967 STDERR->flush;
3968 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3969 if ($choice =~ /^t$/i) {
3970 $cred->may_save(undef);
3971 } elsif ($choice =~ /^r$/i) {
3972 return -1;
3973 } elsif ($may_save && $choice =~ /^p$/i) {
3974 $cred->may_save($may_save);
3975 } else {
3976 goto prompt;
3977 }
3978 $cred->accepted_failures($failures);
3979 $SVN::_Core::SVN_NO_ERROR;
3980}
3981sub ssl_client_cert {
3983 my ($cred, $realm, $may_save, $pool) = @_;
3984 $may_save = undef if $_no_auth_cache;
3985 print STDERR "Client certificate filename: ";
3986 STDERR->flush;
3987 chomp(my $filename = <STDIN>);
3988 $cred->cert_file($filename);
3989 $cred->may_save($may_save);
3990 $SVN::_Core::SVN_NO_ERROR;
3991}
3992sub ssl_client_cert_pw {
3994 my ($cred, $realm, $may_save, $pool) = @_;
3995 $may_save = undef if $_no_auth_cache;
3996 $cred->password(_read_password("Password: ", $realm));
3997 $cred->may_save($may_save);
3998 $SVN::_Core::SVN_NO_ERROR;
3999}
4000sub username {
4002 my ($cred, $realm, $may_save, $pool) = @_;
4003 $may_save = undef if $_no_auth_cache;
4004 if (defined $realm && length $realm) {
4005 print STDERR "Authentication realm: $realm\n";
4006 }
4007 my $username;
4008 if (defined $_username) {
4009 $username = $_username;
4010 } else {
4011 print STDERR "Username: ";
4012 STDERR->flush;
4013 chomp($username = <STDIN>);
4014 }
4015 $cred->username($username);
4016 $cred->may_save($may_save);
4017 $SVN::_Core::SVN_NO_ERROR;
4018}
4019sub _read_password {
4021 my ($prompt, $realm) = @_;
4022 my $password = '';
4023 if (exists $ENV{GIT_ASKPASS}) {
4024 open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
4025 $password = <PH>;
4026 $password =~ s/[\012\015]//; # \n\r
4027 close(PH);
4028 } else {
4029 print STDERR $prompt;
4030 STDERR->flush;
4031 require Term::ReadKey;
4032 Term::ReadKey::ReadMode('noecho');
4033 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
4034 last if $key =~ /[\012\015]/; # \n\r
4035 $password .= $key;
4036 }
4037 Term::ReadKey::ReadMode('restore');
4038 print STDERR "\n";
4039 STDERR->flush;
4040 }
4041 $password;
4042}
4043package SVN::Git::Fetcher;
4045use vars qw/@ISA/;
4046use strict;
4047use warnings;
4048use Carp qw/croak/;
4049use IO::File qw//;
4050use vars qw/$_ignore_regex/;
4051# file baton members: path, mode_a, mode_b, pool, fh, blob, base
4053sub new {
4054 my ($class, $git_svn, $switch_path) = @_;
4055 my $self = SVN::Delta::Editor->new;
4056 bless $self, $class;
4057 if (exists $git_svn->{last_commit}) {
4058 $self->{c} = $git_svn->{last_commit};
4059 $self->{empty_symlinks} =
4060 _mark_empty_symlinks($git_svn, $switch_path);
4061 }
4062 $self->{ignore_regex} = eval { command_oneline('config', '--get',
4063 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
4064 $self->{empty} = {};
4065 $self->{dir_prop} = {};
4066 $self->{file_prop} = {};
4067 $self->{absent_dir} = {};
4068 $self->{absent_file} = {};
4069 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4070 $self->{pathnameencoding} = Git::config('svn.pathnameencoding');
4071 $self;
4072}
4073# this uses the Ra object, so it must be called before do_{switch,update},
4075# not inside them (when the Git::SVN::Fetcher object is passed) to
4076# do_{switch,update}
4077sub _mark_empty_symlinks {
4078 my ($git_svn, $switch_path) = @_;
4079 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4080 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4081 my %ret;
4083 my ($rev, $cmt) = $git_svn->last_rev_commit;
4084 return {} unless ($rev && $cmt);
4085 # allow the warning to be printed for each revision we fetch to
4087 # ensure the user sees it. The user can also disable the workaround
4088 # on the repository even while git svn is running and the next
4089 # revision fetched will skip this expensive function.
4090 my $printed_warning;
4091 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4092 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4093 local $/ = "\0";
4094 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4095 $pfx .= '/' if length($pfx);
4096 while (<$ls>) {
4097 chomp;
4098 s/\A100644 blob $empty_blob\t//o or next;
4099 unless ($printed_warning) {
4100 print STDERR "Scanning for empty symlinks, ",
4101 "this may take a while if you have ",
4102 "many empty files\n",
4103 "You may disable this with `",
4104 "git config svn.brokenSymlinkWorkaround ",
4105 "false'.\n",
4106 "This may be done in a different ",
4107 "terminal without restarting ",
4108 "git svn\n";
4109 $printed_warning = 1;
4110 }
4111 my $path = $_;
4112 my (undef, $props) =
4113 $git_svn->ra->get_file($pfx.$path, $rev, undef);
4114 if ($props->{'svn:special'}) {
4115 $ret{$path} = 1;
4116 }
4117 }
4118 command_close_pipe($ls, $ctx);
4119 \%ret;
4120}
4121# returns true if a given path is inside a ".git" directory
4123sub in_dot_git {
4124 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4125}
4126# return value: 0 -- don't ignore, 1 -- ignore
4128sub is_path_ignored {
4129 my ($self, $path) = @_;
4130 return 1 if in_dot_git($path);
4131 return 1 if defined($self->{ignore_regex}) &&
4132 $path =~ m!$self->{ignore_regex}!;
4133 return 0 unless defined($_ignore_regex);
4134 return 1 if $path =~ m!$_ignore_regex!o;
4135 return 0;
4136}
4137sub set_path_strip {
4139 my ($self, $path) = @_;
4140 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4141}
4142sub open_root {
4144 { path => '' };
4145}
4146sub open_directory {
4148 my ($self, $path, $pb, $rev) = @_;
4149 { path => $path };
4150}
4151sub git_path {
4153 my ($self, $path) = @_;
4154 if (my $enc = $self->{pathnameencoding}) {
4155 require Encode;
4156 Encode::from_to($path, 'UTF-8', $enc);
4157 }
4158 if ($self->{path_strip}) {
4159 $path =~ s!$self->{path_strip}!! or
4160 die "Failed to strip path '$path' ($self->{path_strip})\n";
4161 }
4162 $path;
4163}
4164sub delete_entry {
4166 my ($self, $path, $rev, $pb) = @_;
4167 return undef if $self->is_path_ignored($path);
4168 my $gpath = $self->git_path($path);
4170 return undef if ($gpath eq '');
4171 # remove entire directories.
4173 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4174 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4175 if ($tree) {
4176 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4177 -r --name-only -z/,
4178 $tree);
4179 local $/ = "\0";
4180 while (<$ls>) {
4181 chomp;
4182 my $rmpath = "$gpath/$_";
4183 $self->{gii}->remove($rmpath);
4184 print "\tD\t$rmpath\n" unless $::_q;
4185 }
4186 print "\tD\t$gpath/\n" unless $::_q;
4187 command_close_pipe($ls, $ctx);
4188 } else {
4189 $self->{gii}->remove($gpath);
4190 print "\tD\t$gpath\n" unless $::_q;
4191 }
4192 $self->{empty}->{$path} = 0;
4193 undef;
4194}
4195sub open_file {
4197 my ($self, $path, $pb, $rev) = @_;
4198 my ($mode, $blob);
4199 goto out if $self->is_path_ignored($path);
4201 my $gpath = $self->git_path($path);
4203 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4204 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4205 unless (defined $mode && defined $blob) {
4206 die "$path was not found in commit $self->{c} (r$rev)\n";
4207 }
4208 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4209 $mode = '120000';
4210 }
4211out:
4212 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4213 pool => SVN::Pool->new, action => 'M' };
4214}
4215sub add_file {
4217 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4218 my $mode;
4219 if (!$self->is_path_ignored($path)) {
4221 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4222 delete $self->{empty}->{$dir};
4223 $mode = '100644';
4224 }
4225 { path => $path, mode_a => $mode, mode_b => $mode,
4226 pool => SVN::Pool->new, action => 'A' };
4227}
4228sub add_directory {
4230 my ($self, $path, $cp_path, $cp_rev) = @_;
4231 goto out if $self->is_path_ignored($path);
4232 my $gpath = $self->git_path($path);
4233 if ($gpath eq '') {
4234 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4235 -r --name-only -z/,
4236 $self->{c});
4237 local $/ = "\0";
4238 while (<$ls>) {
4239 chomp;
4240 $self->{gii}->remove($_);
4241 print "\tD\t$_\n" unless $::_q;
4242 }
4243 command_close_pipe($ls, $ctx);
4244 $self->{empty}->{$path} = 0;
4245 }
4246 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4247 delete $self->{empty}->{$dir};
4248 $self->{empty}->{$path} = 1;
4249out:
4250 { path => $path };
4251}
4252sub change_dir_prop {
4254 my ($self, $db, $prop, $value) = @_;
4255 return undef if $self->is_path_ignored($db->{path});
4256 $self->{dir_prop}->{$db->{path}} ||= {};
4257 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4258 undef;
4259}
4260sub absent_directory {
4262 my ($self, $path, $pb) = @_;
4263 return undef if $self->is_path_ignored($path);
4264 $self->{absent_dir}->{$pb->{path}} ||= [];
4265 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4266 undef;
4267}
4268sub absent_file {
4270 my ($self, $path, $pb) = @_;
4271 return undef if $self->is_path_ignored($path);
4272 $self->{absent_file}->{$pb->{path}} ||= [];
4273 push @{$self->{absent_file}->{$pb->{path}}}, $path;
4274 undef;
4275}
4276sub change_file_prop {
4278 my ($self, $fb, $prop, $value) = @_;
4279 return undef if $self->is_path_ignored($fb->{path});
4280 if ($prop eq 'svn:executable') {
4281 if ($fb->{mode_b} != 120000) {
4282 $fb->{mode_b} = defined $value ? 100755 : 100644;
4283 }
4284 } elsif ($prop eq 'svn:special') {
4285 $fb->{mode_b} = defined $value ? 120000 : 100644;
4286 } else {
4287 $self->{file_prop}->{$fb->{path}} ||= {};
4288 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4289 }
4290 undef;
4291}
4292sub apply_textdelta {
4294 my ($self, $fb, $exp) = @_;
4295 return undef if $self->is_path_ignored($fb->{path});
4296 my $fh = $::_repository->temp_acquire('svn_delta');
4297 # $fh gets auto-closed() by SVN::TxDelta::apply(),
4298 # (but $base does not,) so dup() it for reading in close_file
4299 open my $dup, '<&', $fh or croak $!;
4300 my $base = $::_repository->temp_acquire('git_blob');
4301 if ($fb->{blob}) {
4303 my ($base_is_link, $size);
4304 if ($fb->{mode_a} eq '120000' &&
4306 ! $self->{empty_symlinks}->{$fb->{path}}) {
4307 print $base 'link ' or die "print $!\n";
4308 $base_is_link = 1;
4309 }
4310 retry:
4311 $size = $::_repository->cat_blob($fb->{blob}, $base);
4312 die "Failed to read object $fb->{blob}" if ($size < 0);
4313 if (defined $exp) {
4315 seek $base, 0, 0 or croak $!;
4316 my $got = ::md5sum($base);
4317 if ($got ne $exp) {
4318 my $err = "Checksum mismatch: ".
4319 "$fb->{path} $fb->{blob}\n" .
4320 "expected: $exp\n" .
4321 " got: $got\n";
4322 if ($base_is_link) {
4323 warn $err,
4324 "Retrying... (possibly ",
4325 "a bad symlink from SVN)\n";
4326 $::_repository->temp_reset($base);
4327 $base_is_link = 0;
4328 goto retry;
4329 }
4330 die $err;
4331 }
4332 }
4333 }
4334 seek $base, 0, 0 or croak $!;
4335 $fb->{fh} = $fh;
4336 $fb->{base} = $base;
4337 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4338}
4339sub close_file {
4341 my ($self, $fb, $exp) = @_;
4342 return undef if $self->is_path_ignored($fb->{path});
4343 my $hash;
4345 my $path = $self->git_path($fb->{path});
4346 if (my $fh = $fb->{fh}) {
4347 if (defined $exp) {
4348 seek($fh, 0, 0) or croak $!;
4349 my $got = ::md5sum($fh);
4350 if ($got ne $exp) {
4351 die "Checksum mismatch: $path\n",
4352 "expected: $exp\n got: $got\n";
4353 }
4354 }
4355 if ($fb->{mode_b} == 120000) {
4356 sysseek($fh, 0, 0) or croak $!;
4357 my $rd = sysread($fh, my $buf, 5);
4358 if (!defined $rd) {
4360 croak "sysread: $!\n";
4361 } elsif ($rd == 0) {
4362 warn "$path has mode 120000",
4363 " but it points to nothing\n",
4364 "converting to an empty file with mode",
4365 " 100644\n";
4366 $fb->{mode_b} = '100644';
4367 } elsif ($buf ne 'link ') {
4368 warn "$path has mode 120000",
4369 " but is not a link\n";
4370 } else {
4371 my $tmp_fh = $::_repository->temp_acquire(
4372 'svn_hash');
4373 my $res;
4374 while ($res = sysread($fh, my $str, 1024)) {
4375 my $out = syswrite($tmp_fh, $str, $res);
4376 defined($out) && $out == $res
4377 or croak("write ",
4378 Git::temp_path($tmp_fh),
4379 ": $!\n");
4380 }
4381 defined $res or croak $!;
4382 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4384 Git::temp_release($tmp_fh, 1);
4385 }
4386 }
4387 $hash = $::_repository->hash_and_insert_object(
4389 Git::temp_path($fh));
4390 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4391 Git::temp_release($fb->{base}, 1);
4393 Git::temp_release($fh, 1);
4394 } else {
4395 $hash = $fb->{blob} or die "no blob information\n";
4396 }
4397 $fb->{pool}->clear;
4398 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4399 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4400 undef;
4401}
4402sub abort_edit {
4404 my $self = shift;
4405 $self->{nr} = $self->{gii}->{nr};
4406 delete $self->{gii};
4407 $self->SUPER::abort_edit(@_);
4408}
4409sub close_edit {
4411 my $self = shift;
4412 $self->{git_commit_ok} = 1;
4413 $self->{nr} = $self->{gii}->{nr};
4414 delete $self->{gii};
4415 $self->SUPER::close_edit(@_);
4416}
4417package SVN::Git::Editor;
4419use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4420use strict;
4421use warnings;
4422use Carp qw/croak/;
4423use IO::File;
4424sub new {
4426 my ($class, $opts) = @_;
4427 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4428 die "$_ required!\n" unless (defined $opts->{$_});
4429 }
4430 my $pool = SVN::Pool->new;
4432 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4433 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4434 $opts->{r}, $mods);
4435 # $opts->{ra} functions should not be used after this:
4437 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4438 $opts->{editor_cb}, $pool);
4439 my $self = SVN::Delta::Editor->new(@ce, $pool);
4440 bless $self, $class;
4441 foreach (qw/svn_path r tree_a tree_b/) {
4442 $self->{$_} = $opts->{$_};
4443 }
4444 $self->{url} = $opts->{ra}->{url};
4445 $self->{mods} = $mods;
4446 $self->{types} = $types;
4447 $self->{pool} = $pool;
4448 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4449 $self->{rm} = { };
4450 $self->{path_prefix} = length $self->{svn_path} ?
4451 "$self->{svn_path}/" : '';
4452 $self->{config} = $opts->{config};
4453 return $self;
4454}
4455sub generate_diff {
4457 my ($tree_a, $tree_b) = @_;
4458 my @diff_tree = qw(diff-tree -z -r);
4459 if ($_cp_similarity) {
4460 push @diff_tree, "-C$_cp_similarity";
4461 } else {
4462 push @diff_tree, '-C';
4463 }
4464 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4465 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4466 push @diff_tree, $tree_a, $tree_b;
4467 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4468 local $/ = "\0";
4469 my $state = 'meta';
4470 my @mods;
4471 while (<$diff_fh>) {
4472 chomp $_; # this gets rid of the trailing "\0"
4473 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4474 ($::sha1)\s($::sha1)\s
4475 ([MTCRAD])\d*$/xo) {
4476 push @mods, { mode_a => $1, mode_b => $2,
4477 sha1_a => $3, sha1_b => $4,
4478 chg => $5 };
4479 if ($5 =~ /^(?:C|R)$/) {
4480 $state = 'file_a';
4481 } else {
4482 $state = 'file_b';
4483 }
4484 } elsif ($state eq 'file_a') {
4485 my $x = $mods[$#mods] or croak "Empty array\n";
4486 if ($x->{chg} !~ /^(?:C|R)$/) {
4487 croak "Error parsing $_, $x->{chg}\n";
4488 }
4489 $x->{file_a} = $_;
4490 $state = 'file_b';
4491 } elsif ($state eq 'file_b') {
4492 my $x = $mods[$#mods] or croak "Empty array\n";
4493 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4494 croak "Error parsing $_, $x->{chg}\n";
4495 }
4496 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4497 croak "Error parsing $_, $x->{chg}\n";
4498 }
4499 $x->{file_b} = $_;
4500 $state = 'meta';
4501 } else {
4502 croak "Error parsing $_\n";
4503 }
4504 }
4505 command_close_pipe($diff_fh, $ctx);
4506 \@mods;
4507}
4508sub check_diff_paths {
4510 my ($ra, $pfx, $rev, $mods) = @_;
4511 my %types;
4512 $pfx .= '/' if length $pfx;
4513 sub type_diff_paths {
4515 my ($ra, $types, $path, $rev) = @_;
4516 my @p = split m#/+#, $path;
4517 my $c = shift @p;
4518 unless (defined $types->{$c}) {
4519 $types->{$c} = $ra->check_path($c, $rev);
4520 }
4521 while (@p) {
4522 $c .= '/' . shift @p;
4523 next if defined $types->{$c};
4524 $types->{$c} = $ra->check_path($c, $rev);
4525 }
4526 }
4527 foreach my $m (@$mods) {
4529 foreach my $f (qw/file_a file_b/) {
4530 next unless defined $m->{$f};
4531 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4532 if (length $pfx.$dir && ! defined $types{$dir}) {
4533 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4534 }
4535 }
4536 }
4537 \%types;
4538}
4539sub split_path {
4541 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4542}
4543sub repo_path {
4545 my ($self, $path) = @_;
4546 if (my $enc = $self->{pathnameencoding}) {
4547 require Encode;
4548 Encode::from_to($path, $enc, 'UTF-8');
4549 }
4550 $self->{path_prefix}.(defined $path ? $path : '');
4551}
4552sub url_path {
4554 my ($self, $path) = @_;
4555 if ($self->{url} =~ m#^https?://#) {
4556 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4557 }
4558 $self->{url} . '/' . $self->repo_path($path);
4559}
4560sub rmdirs {
4562 my ($self) = @_;
4563 my $rm = $self->{rm};
4564 delete $rm->{''}; # we never delete the url we're tracking
4565 return unless %$rm;
4566 foreach (keys %$rm) {
4568 my @d = split m#/#, $_;
4569 my $c = shift @d;
4570 $rm->{$c} = 1;
4571 while (@d) {
4572 $c .= '/' . shift @d;
4573 $rm->{$c} = 1;
4574 }
4575 }
4576 delete $rm->{$self->{svn_path}};
4577 delete $rm->{''}; # we never delete the url we're tracking
4578 return unless %$rm;
4579 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4581 $self->{tree_b});
4582 local $/ = "\0";
4583 while (<$fh>) {
4584 chomp;
4585 my @dn = split m#/#, $_;
4586 while (pop @dn) {
4587 delete $rm->{join '/', @dn};
4588 }
4589 unless (%$rm) {
4590 close $fh;
4591 return;
4592 }
4593 }
4594 command_close_pipe($fh, $ctx);
4595 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4597 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4598 $self->close_directory($bat->{$d}, $p);
4599 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4600 print "\tD+\t$d/\n" unless $::_q;
4601 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4602 delete $bat->{$d};
4603 }
4604}
4605sub open_or_add_dir {
4607 my ($self, $full_path, $baton) = @_;
4608 my $t = $self->{types}->{$full_path};
4609 if (!defined $t) {
4610 die "$full_path not known in r$self->{r} or we have a bug!\n";
4611 }
4612 {
4613 no warnings 'once';
4614 # SVN::Node::none and SVN::Node::file are used only once,
4615 # so we're shutting up Perl's warnings about them.
4616 if ($t == $SVN::Node::none) {
4617 return $self->add_directory($full_path, $baton,
4618 undef, -1, $self->{pool});
4619 } elsif ($t == $SVN::Node::dir) {
4620 return $self->open_directory($full_path, $baton,
4621 $self->{r}, $self->{pool});
4622 } # no warnings 'once'
4623 print STDERR "$full_path already exists in repository at ",
4624 "r$self->{r} and it is not a directory (",
4625 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4626 } # no warnings 'once'
4627 exit 1;
4628}
4629sub ensure_path {
4631 my ($self, $path) = @_;
4632 my $bat = $self->{bat};
4633 my $repo_path = $self->repo_path($path);
4634 return $bat->{''} unless (length $repo_path);
4635 my @p = split m#/+#, $repo_path;
4636 my $c = shift @p;
4637 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4638 while (@p) {
4639 my $c0 = $c;
4640 $c .= '/' . shift @p;
4641 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4642 }
4643 return $bat->{$c};
4644}
4645# Subroutine to convert a globbing pattern to a regular expression.
4647# From perl cookbook.
4648sub glob2pat {
4649 my $globstr = shift;
4650 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4651 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4652 return '^' . $globstr . '$';
4653}
4654sub check_autoprop {
4656 my ($self, $pattern, $properties, $file, $fbat) = @_;
4657 # Convert the globbing pattern to a regular expression.
4658 my $regex = glob2pat($pattern);
4659 # Check if the pattern matches the file name.
4660 if($file =~ m/($regex)/) {
4661 # Parse the list of properties to set.
4662 my @props = split(/;/, $properties);
4663 foreach my $prop (@props) {
4664 # Parse 'name=value' syntax and set the property.
4665 if ($prop =~ /([^=]+)=(.*)/) {
4666 my ($n,$v) = ($1,$2);
4667 for ($n, $v) {
4668 s/^\s+//; s/\s+$//;
4669 }
4670 $self->change_file_prop($fbat, $n, $v);
4671 }
4672 }
4673 }
4674}
4675sub apply_autoprops {
4677 my ($self, $file, $fbat) = @_;
4678 my $conf_t = ${$self->{config}}{'config'};
4679 no warnings 'once';
4680 # Check [miscellany]/enable-auto-props in svn configuration.
4681 if (SVN::_Core::svn_config_get_bool(
4682 $conf_t,
4683 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4684 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4685 0)) {
4686 # Auto-props are enabled. Enumerate them to look for matches.
4687 my $callback = sub {
4688 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4689 };
4690 SVN::_Core::svn_config_enumerate(
4691 $conf_t,
4692 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4693 $callback);
4694 }
4695}
4696sub A {
4698 my ($self, $m) = @_;
4699 my ($dir, $file) = split_path($m->{file_b});
4700 my $pbat = $self->ensure_path($dir);
4701 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4702 undef, -1);
4703 print "\tA\t$m->{file_b}\n" unless $::_q;
4704 $self->apply_autoprops($file, $fbat);
4705 $self->chg_file($fbat, $m);
4706 $self->close_file($fbat,undef,$self->{pool});
4707}
4708sub C {
4710 my ($self, $m) = @_;
4711 my ($dir, $file) = split_path($m->{file_b});
4712 my $pbat = $self->ensure_path($dir);
4713 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4714 $self->url_path($m->{file_a}), $self->{r});
4715 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4716 $self->chg_file($fbat, $m);
4717 $self->close_file($fbat,undef,$self->{pool});
4718}
4719sub delete_entry {
4721 my ($self, $path, $pbat) = @_;
4722 my $rpath = $self->repo_path($path);
4723 my ($dir, $file) = split_path($rpath);
4724 $self->{rm}->{$dir} = 1;
4725 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4726}
4727sub R {
4729 my ($self, $m) = @_;
4730 my ($dir, $file) = split_path($m->{file_b});
4731 my $pbat = $self->ensure_path($dir);
4732 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4733 $self->url_path($m->{file_a}), $self->{r});
4734 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4735 $self->apply_autoprops($file, $fbat);
4736 $self->chg_file($fbat, $m);
4737 $self->close_file($fbat,undef,$self->{pool});
4738 ($dir, $file) = split_path($m->{file_a});
4740 $pbat = $self->ensure_path($dir);
4741 $self->delete_entry($m->{file_a}, $pbat);
4742}
4743sub M {
4745 my ($self, $m) = @_;
4746 my ($dir, $file) = split_path($m->{file_b});
4747 my $pbat = $self->ensure_path($dir);
4748 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4749 $pbat,$self->{r},$self->{pool});
4750 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4751 $self->chg_file($fbat, $m);
4752 $self->close_file($fbat,undef,$self->{pool});
4753}
4754sub T { shift->M(@_) }
4756sub change_file_prop {
4758 my ($self, $fbat, $pname, $pval) = @_;
4759 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4760}
4761sub _chg_file_get_blob ($$$$) {
4763 my ($self, $fbat, $m, $which) = @_;
4764 my $fh = $::_repository->temp_acquire("git_blob_$which");
4765 if ($m->{"mode_$which"} =~ /^120/) {
4766 print $fh 'link ' or croak $!;
4767 $self->change_file_prop($fbat,'svn:special','*');
4768 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4769 $self->change_file_prop($fbat,'svn:special',undef);
4770 }
4771 my $blob = $m->{"sha1_$which"};
4772 return ($fh,) if ($blob =~ /^0{40}$/);
4773 my $size = $::_repository->cat_blob($blob, $fh);
4774 croak "Failed to read object $blob" if ($size < 0);
4775 $fh->flush == 0 or croak $!;
4776 seek $fh, 0, 0 or croak $!;
4777 my $exp = ::md5sum($fh);
4779 seek $fh, 0, 0 or croak $!;
4780 return ($fh, $exp);
4781}
4782sub chg_file {
4784 my ($self, $fbat, $m) = @_;
4785 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4786 $self->change_file_prop($fbat,'svn:executable','*');
4787 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4788 $self->change_file_prop($fbat,'svn:executable',undef);
4789 }
4790 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4791 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4792 my $pool = SVN::Pool->new;
4793 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4794 if (-s $fh_a) {
4795 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4796 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4797 if (defined $res) {
4798 die "Unexpected result from send_txstream: $res\n",
4799 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4800 }
4801 } else {
4802 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4803 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4804 if ($got ne $exp_b);
4805 }
4806 Git::temp_release($fh_b, 1);
4807 Git::temp_release($fh_a, 1);
4808 $pool->clear;
4809}
4810sub D {
4812 my ($self, $m) = @_;
4813 my ($dir, $file) = split_path($m->{file_b});
4814 my $pbat = $self->ensure_path($dir);
4815 print "\tD\t$m->{file_b}\n" unless $::_q;
4816 $self->delete_entry($m->{file_b}, $pbat);
4817}
4818sub close_edit {
4820 my ($self) = @_;
4821 my ($p,$bat) = ($self->{pool}, $self->{bat});
4822 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4823 next if $_ eq '';
4824 $self->close_directory($bat->{$_}, $p);
4825 }
4826 $self->close_directory($bat->{''}, $p);
4827 $self->SUPER::close_edit($p);
4828 $p->clear;
4829}
4830sub abort_edit {
4832 my ($self) = @_;
4833 $self->SUPER::abort_edit($self->{pool});
4834}
4835sub DESTROY {
4837 my $self = shift;
4838 $self->SUPER::DESTROY(@_);
4839 $self->{pool}->clear;
4840}
4841# this drives the editor
4843sub apply_diff {
4844 my ($self) = @_;
4845 my $mods = $self->{mods};
4846 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4847 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4848 my $f = $m->{chg};
4849 if (defined $o{$f}) {
4850 $self->$f($m);
4851 } else {
4852 fatal("Invalid change type: $f");
4853 }
4854 }
4855 $self->rmdirs if $_rmdir;
4856 if (@$mods == 0) {
4857 $self->abort_edit;
4858 } else {
4859 $self->close_edit;
4860 }
4861 return scalar @$mods;
4862}
4863package Git::SVN::Ra;
4865use vars qw/@ISA $config_dir $_log_window_size/;
4866use strict;
4867use warnings;
4868my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4869BEGIN {
4871 # enforce temporary pool usage for some simple functions
4872 no strict 'refs';
4873 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4874 get_file/) {
4875 my $SUPER = "SUPER::$f";
4876 *$f = sub {
4877 my $self = shift;
4878 my $pool = SVN::Pool->new;
4879 my @ret = $self->$SUPER(@_,$pool);
4880 $pool->clear;
4881 wantarray ? @ret : $ret[0];
4882 };
4883 }
4884}
4885sub _auth_providers () {
4887 [
4888 SVN::Client::get_simple_provider(),
4889 SVN::Client::get_ssl_server_trust_file_provider(),
4890 SVN::Client::get_simple_prompt_provider(
4891 \&Git::SVN::Prompt::simple, 2),
4892 SVN::Client::get_ssl_client_cert_file_provider(),
4893 SVN::Client::get_ssl_client_cert_prompt_provider(
4894 \&Git::SVN::Prompt::ssl_client_cert, 2),
4895 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4896 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4897 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4898 SVN::Client::get_username_provider(),
4899 SVN::Client::get_ssl_server_trust_prompt_provider(
4900 \&Git::SVN::Prompt::ssl_server_trust),
4901 SVN::Client::get_username_prompt_provider(
4902 \&Git::SVN::Prompt::username, 2)
4903 ]
4904}
4905sub escape_uri_only {
4907 my ($uri) = @_;
4908 my @tmp;
4909 foreach (split m{/}, $uri) {
4910 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4911 push @tmp, $_;
4912 }
4913 join('/', @tmp);
4914}
4915sub escape_url {
4917 my ($url) = @_;
4918 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4919 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4920 $url = "$scheme://$domain$uri";
4921 }
4922 $url;
4923}
4924sub new {
4926 my ($class, $url) = @_;
4927 $url =~ s!/+$!!;
4928 return $RA if ($RA && $RA->{url} eq $url);
4929 ::_req_svn();
4931 SVN::_Core::svn_config_ensure($config_dir, undef);
4933 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4934 my $config = SVN::Core::config_get_config($config_dir);
4935 $RA = undef;
4936 my $dont_store_passwords = 1;
4937 my $conf_t = ${$config}{'config'};
4938 {
4939 no warnings 'once';
4940 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4941 # produces warnings that variables are used only once.
4942 # I had not found the better way to shut them up, so
4943 # the warnings of type 'once' are disabled in this block.
4944 if (SVN::_Core::svn_config_get_bool($conf_t,
4945 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4946 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4947 1) == 0) {
4948 SVN::_Core::svn_auth_set_parameter($baton,
4949 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4950 bless (\$dont_store_passwords, "_p_void"));
4951 }
4952 if (SVN::_Core::svn_config_get_bool($conf_t,
4953 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4954 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4955 1) == 0) {
4956 $Git::SVN::Prompt::_no_auth_cache = 1;
4957 }
4958 } # no warnings 'once'
4959 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4960 config => $config,
4961 pool => SVN::Pool->new,
4962 auth_provider_callbacks => $callbacks);
4963 $self->{url} = $url;
4964 $self->{svn_path} = $url;
4965 $self->{repos_root} = $self->get_repos_root;
4966 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4967 $self->{cache} = { check_path => { r => 0, data => {} },
4968 get_dir => { r => 0, data => {} } };
4969 $RA = bless $self, $class;
4970}
4971sub check_path {
4973 my ($self, $path, $r) = @_;
4974 my $cache = $self->{cache}->{check_path};
4975 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4976 return $cache->{data}->{$path};
4977 }
4978 my $pool = SVN::Pool->new;
4979 my $t = $self->SUPER::check_path($path, $r, $pool);
4980 $pool->clear;
4981 if ($r != $cache->{r}) {
4982 %{$cache->{data}} = ();
4983 $cache->{r} = $r;
4984 }
4985 $cache->{data}->{$path} = $t;
4986}
4987sub get_dir {
4989 my ($self, $dir, $r) = @_;
4990 my $cache = $self->{cache}->{get_dir};
4991 if ($r == $cache->{r}) {
4992 if (my $x = $cache->{data}->{$dir}) {
4993 return wantarray ? @$x : $x->[0];
4994 }
4995 }
4996 my $pool = SVN::Pool->new;
4997 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4998 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4999 $pool->clear;
5000 if ($r != $cache->{r}) {
5001 %{$cache->{data}} = ();
5002 $cache->{r} = $r;
5003 }
5004 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
5005 wantarray ? (\%dirents, $r, $props) : \%dirents;
5006}
5007sub DESTROY {
5009 # do not call the real DESTROY since we store ourselves in $RA
5010}
5011# get_log(paths, start, end, limit,
5013# discover_changed_paths, strict_node_history, receiver)
5014sub get_log {
5015 my ($self, @args) = @_;
5016 my $pool = SVN::Pool->new;
5017 # svn_log_changed_path_t objects passed to get_log are likely to be
5019 # overwritten even if only the refs are copied to an external variable,
5020 # so we should dup the structures in their entirety. Using an
5021 # externally passed pool (instead of our temporary and quickly cleared
5022 # pool in Git::SVN::Ra) does not help matters at all...
5023 my $receiver = pop @args;
5024 my $prefix = "/".$self->{svn_path};
5025 $prefix =~ s#/+($)##;
5026 my $prefix_regex = qr#^\Q$prefix\E#;
5027 push(@args, sub {
5028 my ($paths) = $_[0];
5029 return &$receiver(@_) unless $paths;
5030 $_[0] = ();
5031 foreach my $p (keys %$paths) {
5032 my $i = $paths->{$p};
5033 # Make path relative to our url, not repos_root
5034 $p =~ s/$prefix_regex//;
5035 my %s = map { $_ => $i->$_; }
5036 qw/copyfrom_path copyfrom_rev action/;
5037 if ($s{'copyfrom_path'}) {
5038 $s{'copyfrom_path'} =~ s/$prefix_regex//;
5039 }
5040 $_[0]{$p} = \%s;
5041 }
5042 &$receiver(@_);
5043 });
5044 # the limit parameter was not supported in SVN 1.1.x, so we
5047 # drop it. Therefore, the receiver callback passed to it
5048 # is made aware of this limitation by being wrapped if
5049 # the limit passed to is being wrapped.
5050 if ($SVN::Core::VERSION le '1.2.0') {
5051 my $limit = splice(@args, 3, 1);
5052 if ($limit > 0) {
5053 my $receiver = pop @args;
5054 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
5055 }
5056 }
5057 my $ret = $self->SUPER::get_log(@args, $pool);
5058 $pool->clear;
5059 $ret;
5060}
5061sub trees_match {
5063 my ($self, $url1, $rev1, $url2, $rev2) = @_;
5064 my $ctx = SVN::Client->new(auth => _auth_providers);
5065 my $out = IO::File->new_tmpfile;
5066 # older SVN (1.1.x) doesn't take $pool as the last parameter for
5068 # $ctx->diff(), so we'll create a default one
5069 my $pool = SVN::Pool->new_default_sub;
5070 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5072 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5073 $out->flush;
5074 my $ret = (($out->stat)[7] == 0);
5075 close $out or croak $!;
5076 $ret;
5078}
5079sub get_commit_editor {
5081 my ($self, $log, $cb, $pool) = @_;
5082 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
5083 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5084}
5085sub gs_do_update {
5087 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5088 my $new = ($rev_a == $rev_b);
5089 my $path = $gs->{path};
5090 if ($new && -e $gs->{index}) {
5092 unlink $gs->{index} or die
5093 "Couldn't unlink index: $gs->{index}: $!\n";
5094 }
5095 my $pool = SVN::Pool->new;
5096 $editor->set_path_strip($path);
5097 my (@pc) = split m#/#, $path;
5098 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5099 1, $editor, $pool);
5100 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5101 # Since we can't rely on svn_ra_reparent being available, we'll
5103 # just have to do some magic with set_path to make it so
5104 # we only want a partial path.
5105 my $sp = '';
5106 my $final = join('/', @pc);
5107 while (@pc) {
5108 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5109 $sp .= '/' if length $sp;
5110 $sp .= shift @pc;
5111 }
5112 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5113 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5115 $reporter->finish_report($pool);
5117 $pool->clear;
5118 $editor->{git_commit_ok};
5119}
5120# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5122# svn_ra_reparent didn't work before 1.4)
5123sub gs_do_switch {
5124 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5125 my $path = $gs->{path};
5126 my $pool = SVN::Pool->new;
5127 my $full_url = $self->{url};
5129 my $old_url = $full_url;
5130 $full_url .= '/' . $path if length $path;
5131 my ($ra, $reparented);
5132 if ($old_url =~ m#^svn(\+ssh)?://# ||
5134 ($full_url =~ m#^https?://# &&
5135 escape_url($full_url) ne $full_url)) {
5136 $_[0] = undef;
5137 $self = undef;
5138 $RA = undef;
5139 $ra = Git::SVN::Ra->new($full_url);
5140 $ra_invalid = 1;
5141 } elsif ($old_url ne $full_url) {
5142 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5143 $self->{url} = $full_url;
5144 $reparented = 1;
5145 }
5146 $ra ||= $self;
5148 $url_b = escape_url($url_b);
5149 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5150 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5151 $reporter->set_path('', $rev_a, 0, @lock, $pool);
5152 $reporter->finish_report($pool);
5153 if ($reparented) {
5155 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5156 $self->{url} = $old_url;
5157 }
5158 $pool->clear;
5160 $editor->{git_commit_ok};
5161}
5162sub longest_common_path {
5164 my ($gsv, $globs) = @_;
5165 my %common;
5166 my $common_max = scalar @$gsv;
5167 foreach my $gs (@$gsv) {
5169 my @tmp = split m#/#, $gs->{path};
5170 my $p = '';
5171 foreach (@tmp) {
5172 $p .= length($p) ? "/$_" : $_;
5173 $common{$p} ||= 0;
5174 $common{$p}++;
5175 }
5176 }
5177 $globs ||= [];
5178 $common_max += scalar @$globs;
5179 foreach my $glob (@$globs) {
5180 my @tmp = split m#/#, $glob->{path}->{left};
5181 my $p = '';
5182 foreach (@tmp) {
5183 $p .= length($p) ? "/$_" : $_;
5184 $common{$p} ||= 0;
5185 $common{$p}++;
5186 }
5187 }
5188 my $longest_path = '';
5190 foreach (sort {length $b <=> length $a} keys %common) {
5191 if ($common{$_} == $common_max) {
5192 $longest_path = $_;
5193 last;
5194 }
5195 }
5196 $longest_path;
5197}
5198sub gs_fetch_loop_common {
5200 my ($self, $base, $head, $gsv, $globs) = @_;
5201 return if ($base > $head);
5202 my $inc = $_log_window_size;
5203 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5204 my $longest_path = longest_common_path($gsv, $globs);
5205 my $ra_url = $self->{url};
5206 my $find_trailing_edge;
5207 while (1) {
5208 my %revs;
5209 my $err;
5210 my $err_handler = $SVN::Error::handler;
5211 $SVN::Error::handler = sub {
5212 ($err) = @_;
5213 skip_unknown_revs($err);
5214 };
5215 sub _cb {
5216 my ($paths, $r, $author, $date, $log) = @_;
5217 [ $paths,
5218 { author => $author, date => $date, log => $log } ];
5219 }
5220 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5221 sub { $revs{$_[1]} = _cb(@_) });
5222 if ($err) {
5223 print "Checked through r$max\r";
5224 } else {
5225 $find_trailing_edge = 1;
5226 }
5227 if ($err and $find_trailing_edge) {
5228 print STDERR "Path '$longest_path' ",
5229 "was probably deleted:\n",
5230 $err->expanded_message,
5231 "\nWill attempt to follow ",
5232 "revisions r$min .. r$max ",
5233 "committed before the deletion\n";
5234 my $hi = $max;
5235 while (--$hi >= $min) {
5236 my $ok;
5237 $self->get_log([$longest_path], $min, $hi,
5238 0, 1, 1, sub {
5239 $ok = $_[1];
5240 $revs{$_[1]} = _cb(@_) });
5241 if ($ok) {
5242 print STDERR "r$min .. r$ok OK\n";
5243 last;
5244 }
5245 }
5246 $find_trailing_edge = 0;
5247 }
5248 $SVN::Error::handler = $err_handler;
5249 my %exists = map { $_->{path} => $_ } @$gsv;
5251 foreach my $r (sort {$a <=> $b} keys %revs) {
5252 my ($paths, $logged) = @{$revs{$r}};
5253 foreach my $gs ($self->match_globs(\%exists, $paths,
5255 $globs, $r)) {
5256 if ($gs->rev_map_max >= $r) {
5257 next;
5258 }
5259 next unless $gs->match_paths($paths, $r);
5260 $gs->{logged_rev_props} = $logged;
5261 if (my $last_commit = $gs->last_commit) {
5262 $gs->assert_index_clean($last_commit);
5263 }
5264 my $log_entry = $gs->do_fetch($paths, $r);
5265 if ($log_entry) {
5266 $gs->do_git_commit($log_entry);
5267 }
5268 $INDEX_FILES{$gs->{index}} = 1;
5269 }
5270 foreach my $g (@$globs) {
5271 my $k = "svn-remote.$g->{remote}." .
5272 "$g->{t}-maxRev";
5273 Git::SVN::tmp_config($k, $r);
5274 }
5275 if ($ra_invalid) {
5276 $_[0] = undef;
5277 $self = undef;
5278 $RA = undef;
5279 $self = Git::SVN::Ra->new($ra_url);
5280 $ra_invalid = undef;
5281 }
5282 }
5283 # pre-fill the .rev_db since it'll eventually get filled in
5284 # with '0' x40 if something new gets committed
5285 foreach my $gs (@$gsv) {
5286 next if $gs->rev_map_max >= $max;
5287 next if defined $gs->rev_map_get($max);
5288 $gs->rev_map_set($max, 0 x40);
5289 }
5290 foreach my $g (@$globs) {
5291 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5292 Git::SVN::tmp_config($k, $max);
5293 }
5294 last if $max >= $head;
5295 $min = $max + 1;
5296 $max += $inc;
5297 $max = $head if ($max > $head);
5298 }
5299 Git::SVN::gc();
5300}
5301sub get_dir_globbed {
5303 my ($self, $left, $depth, $r) = @_;
5304 my @x = eval { $self->get_dir($left, $r) };
5306 return unless scalar @x == 3;
5307 my $dirents = $x[0];
5308 my @finalents;
5309 foreach my $de (keys %$dirents) {
5310 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5311 if ($depth > 1) {
5312 my @args = ("$left/$de", $depth - 1, $r);
5313 foreach my $dir ($self->get_dir_globbed(@args)) {
5314 push @finalents, "$de/$dir";
5315 }
5316 } else {
5317 push @finalents, $de;
5318 }
5319 }
5320 @finalents;
5321}
5322sub match_globs {
5324 my ($self, $exists, $paths, $globs, $r) = @_;
5325 sub get_dir_check {
5327 my ($self, $exists, $g, $r) = @_;
5328 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5330 $g->{path}->{depth},
5331 $r);
5332 foreach my $de (@dirs) {
5334 my $p = $g->{path}->full_path($de);
5335 next if $exists->{$p};
5336 next if (length $g->{path}->{right} &&
5337 ($self->check_path($p, $r) !=
5338 $SVN::Node::dir));
5339 next unless $p =~ /$g->{path}->{regex}/;
5340 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5341 $g->{ref}->full_path($de), 1);
5342 }
5343 }
5344 foreach my $g (@$globs) {
5345 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5346 if ($path->{action} =~ /^[AR]$/) {
5347 get_dir_check($self, $exists, $g, $r);
5348 }
5349 }
5350 foreach (keys %$paths) {
5351 if (/$g->{path}->{left_regex}/ &&
5352 !/$g->{path}->{regex}/) {
5353 next if $paths->{$_}->{action} !~ /^[AR]$/;
5354 get_dir_check($self, $exists, $g, $r);
5355 }
5356 next unless /$g->{path}->{regex}/;
5357 my $p = $1;
5358 my $pathname = $g->{path}->full_path($p);
5359 next if $exists->{$pathname};
5360 next if ($self->check_path($pathname, $r) !=
5361 $SVN::Node::dir);
5362 $exists->{$pathname} = Git::SVN->init(
5363 $self->{url}, $pathname, undef,
5364 $g->{ref}->full_path($p), 1);
5365 }
5366 my $c = '';
5367 foreach (split m#/#, $g->{path}->{left}) {
5368 $c .= "/$_";
5369 next unless ($paths->{$c} &&
5370 ($paths->{$c}->{action} =~ /^[AR]$/));
5371 get_dir_check($self, $exists, $g, $r);
5372 }
5373 }
5374 values %$exists;
5375}
5376sub minimize_url {
5378 my ($self) = @_;
5379 return $self->{url} if ($self->{url} eq $self->{repos_root});
5380 my $url = $self->{repos_root};
5381 my @components = split(m!/!, $self->{svn_path});
5382 my $c = '';
5383 do {
5384 $url .= "/$c" if length $c;
5385 eval {
5386 my $ra = (ref $self)->new($url);
5387 my $latest = $ra->get_latest_revnum;
5388 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5389 };
5390 } while ($@ && ($c = shift @components));
5391 $url;
5392}
5393sub can_do_switch {
5395 my $self = shift;
5396 unless (defined $can_do_switch) {
5397 my $pool = SVN::Pool->new;
5398 my $rep = eval {
5399 $self->do_switch(1, '', 0, $self->{url},
5400 SVN::Delta::Editor->new, $pool);
5401 };
5402 if ($@) {
5403 $can_do_switch = 0;
5404 } else {
5405 $rep->abort_report($pool);
5406 $can_do_switch = 1;
5407 }
5408 $pool->clear;
5409 }
5410 $can_do_switch;
5411}
5412sub skip_unknown_revs {
5414 my ($err) = @_;
5415 my $errno = $err->apr_err();
5416 # Maybe the branch we're tracking didn't
5417 # exist when the repo started, so it's
5418 # not an error if it doesn't, just continue
5419 #
5420 # Wonderfully consistent library, eh?
5421 # 160013 - svn:// and file://
5422 # 175002 - http(s)://
5423 # 175007 - http(s):// (this repo required authorization, too...)
5424 # More codes may be discovered later...
5425 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5426 my $err_key = $err->expanded_message;
5427 # revision numbers change every time, filter them out
5428 $err_key =~ s/\d+/\0/g;
5429 $err_key = "$errno\0$err_key";
5430 unless ($ignored_err{$err_key}) {
5431 warn "W: Ignoring error from SVN, path probably ",
5432 "does not exist: ($errno): ",
5433 $err->expanded_message,"\n";
5434 warn "W: Do not be alarmed at the above message ",
5435 "git-svn is just searching aggressively for ",
5436 "old history.\n",
5437 "This may take a while on large repositories\n";
5438 $ignored_err{$err_key} = 1;
5439 }
5440 return;
5441 }
5442 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5443}
5444package Git::SVN::Log;
5446use strict;
5447use warnings;
5448use POSIX qw/strftime/;
5449use Time::Local;
5450use constant commit_log_separator => ('-' x 72) . "\n";
5451use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5452 %rusers $show_commit $incremental/;
5453my $l_fmt;
5454sub cmt_showable {
5456 my ($c) = @_;
5457 return 1 if defined $c->{r};
5458 # big commit message got truncated by the 16k pretty buffer in rev-list
5460 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5461 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5462 @{$c->{l}} = ();
5463 my @log = command(qw/cat-file commit/, $c->{c});
5464 # shift off the headers
5466 shift @log while ($log[0] ne '');
5467 shift @log;
5468 # TODO: make $c->{l} not have a trailing newline in the future
5470 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5471 (undef, $c->{r}, undef) = ::extract_metadata(
5473 (grep(/^git-svn-id: /, @log))[-1]);
5474 }
5475 return defined $c->{r};
5476}
5477sub log_use_color {
5479 return $color || Git->repository->get_colorbool('color.diff');
5480}
5481sub git_svn_log_cmd {
5483 my ($r_min, $r_max, @args) = @_;
5484 my $head = 'HEAD';
5485 my (@files, @log_opts);
5486 foreach my $x (@args) {
5487 if ($x eq '--' || @files) {
5488 push @files, $x;
5489 } else {
5490 if (::verify_ref("$x^0")) {
5491 $head = $x;
5492 } else {
5493 push @log_opts, $x;
5494 }
5495 }
5496 }
5497 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5499 $gs ||= Git::SVN->_new;
5500 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5501 $gs->refname);
5502 push @cmd, '-r' unless $non_recursive;
5503 push @cmd, qw/--raw --name-status/ if $verbose;
5504 push @cmd, '--color' if log_use_color();
5505 push @cmd, @log_opts;
5506 if (defined $r_max && $r_max == $r_min) {
5507 push @cmd, '--max-count=1';
5508 if (my $c = $gs->rev_map_get($r_max)) {
5509 push @cmd, $c;
5510 }
5511 } elsif (defined $r_max) {
5512 if ($r_max < $r_min) {
5513 ($r_min, $r_max) = ($r_max, $r_min);
5514 }
5515 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5516 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5517 # If there are no commits in the range, both $c_max and $c_min
5518 # will be undefined. If there is at least 1 commit in the
5519 # range, both will be defined.
5520 return () if !defined $c_min || !defined $c_max;
5521 if ($c_min eq $c_max) {
5522 push @cmd, '--max-count=1', $c_min;
5523 } else {
5524 push @cmd, '--boundary', "$c_min..$c_max";
5525 }
5526 }
5527 return (@cmd, @files);
5528}
5529# adapted from pager.c
5531sub config_pager {
5532 if (! -t *STDOUT) {
5533 $ENV{GIT_PAGER_IN_USE} = 'false';
5534 $pager = undef;
5535 return;
5536 }
5537 chomp($pager = command_oneline(qw(var GIT_PAGER)));
5538 if ($pager eq 'cat') {
5539 $pager = undef;
5540 }
5541 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5542}
5543sub run_pager {
5545 return unless defined $pager;
5546 pipe my ($rfd, $wfd) or return;
5547 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5548 if (!$pid) {
5549 open STDOUT, '>&', $wfd or
5550 ::fatal "Can't redirect to stdout: $!";
5551 return;
5552 }
5553 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5554 $ENV{LESS} ||= 'FRSX';
5555 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5556}
5557sub format_svn_date {
5559 # some systmes don't handle or mishandle %z, so be creative.
5560 my $t = shift || time;
5561 my $gm = timelocal(gmtime($t));
5562 my $sign = qw( + + - )[ $t <=> $gm ];
5563 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5564 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5565}
5566sub parse_git_date {
5568 my ($t, $tz) = @_;
5569 # Date::Parse isn't in the standard Perl distro :(
5570 if ($tz =~ s/^\+//) {
5571 $t += tz_to_s_offset($tz);
5572 } elsif ($tz =~ s/^\-//) {
5573 $t -= tz_to_s_offset($tz);
5574 }
5575 return $t;
5576}
5577sub set_local_timezone {
5579 if (defined $TZ) {
5580 $ENV{TZ} = $TZ;
5581 } else {
5582 delete $ENV{TZ};
5583 }
5584}
5585sub tz_to_s_offset {
5587 my ($tz) = @_;
5588 $tz =~ s/(\d\d)$//;
5589 return ($1 * 60) + ($tz * 3600);
5590}
5591sub get_author_info {
5593 my ($dest, $author, $t, $tz) = @_;
5594 $author =~ s/(?:^\s*|\s*$)//g;
5595 $dest->{a_raw} = $author;
5596 my $au;
5597 if ($::_authors) {
5598 $au = $rusers{$author} || undef;
5599 }
5600 if (!$au) {
5601 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5602 }
5603 $dest->{t} = $t;
5604 $dest->{tz} = $tz;
5605 $dest->{a} = $au;
5606 $dest->{t_utc} = parse_git_date($t, $tz);
5607}
5608sub process_commit {
5610 my ($c, $r_min, $r_max, $defer) = @_;
5611 if (defined $r_min && defined $r_max) {
5612 if ($r_min == $c->{r} && $r_min == $r_max) {
5613 show_commit($c);
5614 return 0;
5615 }
5616 return 1 if $r_min == $r_max;
5617 if ($r_min < $r_max) {
5618 # we need to reverse the print order
5619 return 0 if (defined $limit && --$limit < 0);
5620 push @$defer, $c;
5621 return 1;
5622 }
5623 if ($r_min != $r_max) {
5624 return 1 if ($r_min < $c->{r});
5625 return 1 if ($r_max > $c->{r});
5626 }
5627 }
5628 return 0 if (defined $limit && --$limit < 0);
5629 show_commit($c);
5630 return 1;
5631}
5632sub show_commit {
5634 my $c = shift;
5635 if ($oneline) {
5636 my $x = "\n";
5637 if (my $l = $c->{l}) {
5638 while ($l->[0] =~ /^\s*$/) { shift @$l }
5639 $x = $l->[0];
5640 }
5641 $l_fmt ||= 'A' . length($c->{r});
5642 print 'r',pack($l_fmt, $c->{r}),' | ';
5643 print "$c->{c} | " if $show_commit;
5644 print $x;
5645 } else {
5646 show_commit_normal($c);
5647 }
5648}
5649sub show_commit_changed_paths {
5651 my ($c) = @_;
5652 return unless $c->{changed};
5653 print "Changed paths:\n", @{$c->{changed}};
5654}
5655sub show_commit_normal {
5657 my ($c) = @_;
5658 print commit_log_separator, "r$c->{r} | ";
5659 print "$c->{c} | " if $show_commit;
5660 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5661 my $nr_line = 0;
5662 if (my $l = $c->{l}) {
5664 while ($l->[$#$l] eq "\n" && $#$l > 0
5665 && $l->[($#$l - 1)] eq "\n") {
5666 pop @$l;
5667 }
5668 $nr_line = scalar @$l;
5669 if (!$nr_line) {
5670 print "1 line\n\n\n";
5671 } else {
5672 if ($nr_line == 1) {
5673 $nr_line = '1 line';
5674 } else {
5675 $nr_line .= ' lines';
5676 }
5677 print $nr_line, "\n";
5678 show_commit_changed_paths($c);
5679 print "\n";
5680 print $_ foreach @$l;
5681 }
5682 } else {
5683 print "1 line\n";
5684 show_commit_changed_paths($c);
5685 print "\n";
5686 }
5688 foreach my $x (qw/raw stat diff/) {
5689 if ($c->{$x}) {
5690 print "\n";
5691 print $_ foreach @{$c->{$x}}
5692 }
5693 }
5694}
5695sub cmd_show_log {
5697 my (@args) = @_;
5698 my ($r_min, $r_max);
5699 my $r_last = -1; # prevent dupes
5700 set_local_timezone();
5701 if (defined $::_revision) {
5702 if ($::_revision =~ /^(\d+):(\d+)$/) {
5703 ($r_min, $r_max) = ($1, $2);
5704 } elsif ($::_revision =~ /^\d+$/) {
5705 $r_min = $r_max = $::_revision;
5706 } else {
5707 ::fatal "-r$::_revision is not supported, use ",
5708 "standard 'git log' arguments instead";
5709 }
5710 }
5711 config_pager();
5713 @args = git_svn_log_cmd($r_min, $r_max, @args);
5714 if (!@args) {
5715 print commit_log_separator unless $incremental || $oneline;
5716 return;
5717 }
5718 my $log = command_output_pipe(@args);
5719 run_pager();
5720 my (@k, $c, $d, $stat);
5721 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5722 while (<$log>) {
5723 if (/^${esc_color}commit -?($::sha1_short)/o) {
5724 my $cmt = $1;
5725 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5726 $r_last = $c->{r};
5727 process_commit($c, $r_min, $r_max, \@k) or
5728 goto out;
5729 }
5730 $d = undef;
5731 $c = { c => $cmt };
5732 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5733 get_author_info($c, $1, $2, $3);
5734 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5735 # ignore
5736 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5737 push @{$c->{raw}}, $_;
5738 } elsif (/^${esc_color}[ACRMDT]\t/) {
5739 # we could add $SVN->{svn_path} here, but that requires
5740 # remote access at the moment (repo_path_split)...
5741 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5742 push @{$c->{changed}}, $_;
5743 } elsif (/^${esc_color}diff /o) {
5744 $d = 1;
5745 push @{$c->{diff}}, $_;
5746 } elsif ($d) {
5747 push @{$c->{diff}}, $_;
5748 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5749 $esc_color*[\+\-]*$esc_color$/x) {
5750 $stat = 1;
5751 push @{$c->{stat}}, $_;
5752 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5753 push @{$c->{stat}}, $_;
5754 $stat = undef;
5755 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5756 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5757 } elsif (s/^${esc_color} //o) {
5758 push @{$c->{l}}, $_;
5759 }
5760 }
5761 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5762 $r_last = $c->{r};
5763 process_commit($c, $r_min, $r_max, \@k);
5764 }
5765 if (@k) {
5766 ($r_min, $r_max) = ($r_max, $r_min);
5767 process_commit($_, $r_min, $r_max) foreach reverse @k;
5768 }
5769out:
5770 close $log;
5771 print commit_log_separator unless $incremental || $oneline;
5772}
5773sub cmd_blame {
5775 my $path = pop;
5776 config_pager();
5778 run_pager();
5779 my ($fh, $ctx, $rev);
5781 if ($_git_format) {
5783 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5784 while (my $line = <$fh>) {
5785 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5786 # Uncommitted edits show up as a rev ID of
5787 # all zeros, which we can't look up with
5788 # cmt_metadata
5789 if ($1 !~ /^0+$/) {
5790 (undef, $rev, undef) =
5791 ::cmt_metadata($1);
5792 $rev = '0' if (!$rev);
5793 } else {
5794 $rev = '0';
5795 }
5796 $rev = sprintf('%-10s', $rev);
5797 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5798 }
5799 print $line;
5800 }
5801 } else {
5802 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5803 '--', $path);
5804 my ($sha1);
5805 my %authors;
5806 my @buffer;
5807 my %dsha; #distinct sha keys
5808 while (my $line = <$fh>) {
5810 push @buffer, $line;
5811 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5812 $dsha{$1} = 1;
5813 }
5814 }
5815 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5817 foreach my $line (@buffer) {
5819 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5820 $rev = $s2r->{$1};
5821 $rev = '0' if (!$rev)
5822 }
5823 elsif ($line =~ /^author (.*)/) {
5824 $authors{$rev} = $1;
5825 $authors{$rev} =~ s/\s/_/g;
5826 }
5827 elsif ($line =~ /^\t(.*)$/) {
5828 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5829 }
5830 }
5831 }
5832 command_close_pipe($fh, $ctx);
5833}
5834package Git::SVN::Migration;
5836# these version numbers do NOT correspond to actual version numbers
5837# of git nor git-svn. They are just relative.
5838#
5839# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5840#
5841# v1 layout: .git/$id/info/url, refs/remotes/$id
5842#
5843# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5844#
5845# v3 layout: .git/svn/$id, refs/remotes/$id
5846# - info/url may remain for backwards compatibility
5847# - this is what we migrate up to this layout automatically,
5848# - this will be used by git svn init on single branches
5849# v3.1 layout (auto migrated):
5850# - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5851# for backwards compatibility
5852#
5853# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5854# - this is only created for newly multi-init-ed
5855# repositories. Similar in spirit to the
5856# --use-separate-remotes option in git-clone (now default)
5857# - we do not automatically migrate to this (following
5858# the example set by core git)
5859#
5860# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5861# - newer, more-efficient format that uses 24-bytes per record
5862# with no filler space.
5863# - use xxd -c24 < .rev_map.$UUID to view and debug
5864# - This is a one-way migration, repositories updated to the
5865# new format will not be able to use old git-svn without
5866# rebuilding the .rev_db. Rebuilding the rev_db is not
5867# possible if noMetadata or useSvmProps are set; but should
5868# be no problem for users that use the (sensible) defaults.
5869use strict;
5870use warnings;
5871use Carp qw/croak/;
5872use File::Path qw/mkpath/;
5873use File::Basename qw/dirname basename/;
5874use vars qw/$_minimize/;
5875sub migrate_from_v0 {
5877 my $git_dir = $ENV{GIT_DIR};
5878 return undef unless -d $git_dir;
5879 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5880 my $migrated = 0;
5881 while (<$fh>) {
5882 chomp;
5883 my ($id, $orig_ref) = ($_, $_);
5884 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5885 next unless -f "$git_dir/$id/info/url";
5886 my $new_ref = "refs/remotes/$id";
5887 if (::verify_ref("$new_ref^0")) {
5888 print STDERR "W: $orig_ref is probably an old ",
5889 "branch used by an ancient version of ",
5890 "git-svn.\n",
5891 "However, $new_ref also exists.\n",
5892 "We will not be able ",
5893 "to use this branch until this ",
5894 "ambiguity is resolved.\n";
5895 next;
5896 }
5897 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5898 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5899 command_noisy('update-ref', $new_ref, $orig_ref);
5900 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5901 $migrated++;
5902 }
5903 command_close_pipe($fh, $ctx);
5904 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5905 $migrated;
5906}
5907sub migrate_from_v1 {
5909 my $git_dir = $ENV{GIT_DIR};
5910 my $migrated = 0;
5911 return $migrated unless -d $git_dir;
5912 my $svn_dir = "$git_dir/svn";
5913 # just in case somebody used 'svn' as their $id at some point...
5915 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5916 print STDERR "Migrating from a git-svn v1 layout...\n";
5918 mkpath([$svn_dir]);
5919 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5920 "$svn_dir\n\t(required for this version ",
5921 "($::VERSION) of git-svn) does not exist.\n";
5922 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5923 while (<$fh>) {
5924 my $x = $_;
5925 next unless $x =~ s#^refs/remotes/##;
5926 chomp $x;
5927 next unless -f "$git_dir/$x/info/url";
5928 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5929 next unless $u;
5930 my $dn = dirname("$git_dir/svn/$x");
5931 mkpath([$dn]) unless -d $dn;
5932 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5933 mkpath(["$git_dir/svn/svn"]);
5934 print STDERR " - $git_dir/$x/info => ",
5935 "$git_dir/svn/$x/info\n";
5936 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5937 croak "$!: $x";
5938 # don't worry too much about these, they probably
5939 # don't exist with repos this old (save for index,
5940 # and we can easily regenerate that)
5941 foreach my $f (qw/unhandled.log index .rev_db/) {
5942 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5943 }
5944 } else {
5945 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5946 rename "$git_dir/$x", "$git_dir/svn/$x" or
5947 croak "$!: $x";
5948 }
5949 $migrated++;
5950 }
5951 command_close_pipe($fh, $ctx);
5952 print STDERR "Done migrating from a git-svn v1 layout\n";
5953 $migrated;
5954}
5955sub read_old_urls {
5957 my ($l_map, $pfx, $path) = @_;
5958 my @dir;
5959 foreach (<$path/*>) {
5960 if (-r "$_/info/url") {
5961 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5962 my $ref_id = $pfx . basename $_;
5963 my $url = ::file_to_s("$_/info/url");
5964 $l_map->{$ref_id} = $url;
5965 } elsif (-d $_) {
5966 push @dir, $_;
5967 }
5968 }
5969 foreach (@dir) {
5970 my $x = $_;
5971 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5972 read_old_urls($l_map, $x, $_);
5973 }
5974}
5975sub migrate_from_v2 {
5977 my @cfg = command(qw/config -l/);
5978 return if grep /^svn-remote\..+\.url=/, @cfg;
5979 my %l_map;
5980 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5981 my $migrated = 0;
5982 foreach my $ref_id (sort keys %l_map) {
5984 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5985 if ($@) {
5986 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5987 }
5988 $migrated++;
5989 }
5990 $migrated;
5991}
5992sub minimize_connections {
5994 my $r = Git::SVN::read_all_remotes();
5995 my $new_urls = {};
5996 my $root_repos = {};
5997 foreach my $repo_id (keys %$r) {
5998 my $url = $r->{$repo_id}->{url} or next;
5999 my $fetch = $r->{$repo_id}->{fetch} or next;
6000 my $ra = Git::SVN::Ra->new($url);
6001 # skip existing cases where we already connect to the root
6003 if (($ra->{url} eq $ra->{repos_root}) ||
6004 ($ra->{repos_root} eq $repo_id)) {
6005 $root_repos->{$ra->{url}} = $repo_id;
6006 next;
6007 }
6008 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
6010 my $root_path = $ra->{url};
6011 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
6012 foreach my $path (keys %$fetch) {
6013 my $ref_id = $fetch->{$path};
6014 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
6015 # make sure we can read when connecting to
6017 # a higher level of a repository
6018 my ($last_rev, undef) = $gs->last_rev_commit;
6019 if (!defined $last_rev) {
6020 $last_rev = eval {
6021 $root_ra->get_latest_revnum;
6022 };
6023 next if $@;
6024 }
6025 my $new = $root_path;
6026 $new .= length $path ? "/$path" : '';
6027 eval {
6028 $root_ra->get_log([$new], $last_rev, $last_rev,
6029 0, 0, 1, sub { });
6030 };
6031 next if $@;
6032 $new_urls->{$ra->{repos_root}}->{$new} =
6033 { ref_id => $ref_id,
6034 old_repo_id => $repo_id,
6035 old_path => $path };
6036 }
6037 }
6038 my @emptied;
6040 foreach my $url (keys %$new_urls) {
6041 # see if we can re-use an existing [svn-remote "repo_id"]
6042 # instead of creating a(n ugly) new section:
6043 my $repo_id = $root_repos->{$url} || $url;
6044 my $fetch = $new_urls->{$url};
6046 foreach my $path (keys %$fetch) {
6047 my $x = $fetch->{$path};
6048 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
6049 my $pfx = "svn-remote.$x->{old_repo_id}";
6050 my $old_fetch = quotemeta("$x->{old_path}:".
6052 "$x->{ref_id}");
6053 command_noisy(qw/config --unset/,
6054 "$pfx.fetch", '^'. $old_fetch . '$');
6055 delete $r->{$x->{old_repo_id}}->
6056 {fetch}->{$x->{old_path}};
6057 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6058 command_noisy(qw/config --unset/,
6059 "$pfx.url");
6060 push @emptied, $x->{old_repo_id}
6061 }
6062 }
6063 }
6064 if (@emptied) {
6065 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6066 print STDERR <<EOF;
6067The following [svn-remote] sections in your config file ($file) are empty
6068and can be safely removed:
6069EOF
6070 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6071 }
6072}
6073sub migration_check {
6075 migrate_from_v0();
6076 migrate_from_v1();
6077 migrate_from_v2();
6078 minimize_connections() if $_minimize;
6079}
6080package Git::IndexInfo;
6082use strict;
6083use warnings;
6084use Git qw/command_input_pipe command_close_pipe/;
6085sub new {
6087 my ($class) = @_;
6088 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6089 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6090}
6091sub remove {
6093 my ($self, $path) = @_;
6094 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6095 return ++$self->{nr};
6096 }
6097 undef;
6098}
6099sub update {
6101 my ($self, $mode, $hash, $path) = @_;
6102 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6103 return ++$self->{nr};
6104 }
6105 undef;
6106}
6107sub DESTROY {
6109 my ($self) = @_;
6110 command_close_pipe($self->{gui}, $self->{ctx});
6111}
6112package Git::SVN::GlobSpec;
6114use strict;
6115use warnings;
6116sub new {
6118 my ($class, $glob, $pattern_ok) = @_;
6119 my $re = $glob;
6120 $re =~ s!/+$!!g; # no need for trailing slashes
6121 my (@left, @right, @patterns);
6122 my $state = "left";
6123 my $die_msg = "Only one set of wildcard directories " .
6124 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6125 for my $part (split(m|/|, $glob)) {
6126 if ($part =~ /\*/ && $part ne "*") {
6127 die "Invalid pattern in '$glob': $part\n";
6128 } elsif ($pattern_ok && $part =~ /[{}]/ &&
6129 $part !~ /^\{[^{}]+\}/) {
6130 die "Invalid pattern in '$glob': $part\n";
6131 }
6132 if ($part eq "*") {
6133 die $die_msg if $state eq "right";
6134 $state = "pattern";
6135 push(@patterns, "[^/]*");
6136 } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6137 die $die_msg if $state eq "right";
6138 $state = "pattern";
6139 my $p = quotemeta($1);
6140 $p =~ s/\\,/|/g;
6141 push(@patterns, "(?:$p)");
6142 } else {
6143 if ($state eq "left") {
6144 push(@left, $part);
6145 } else {
6146 push(@right, $part);
6147 $state = "right";
6148 }
6149 }
6150 }
6151 my $depth = @patterns;
6152 if ($depth == 0) {
6153 die "One '*' is needed in glob: '$glob'\n";
6154 }
6155 my $left = join('/', @left);
6156 my $right = join('/', @right);
6157 $re = join('/', @patterns);
6158 $re = join('\/',
6159 grep(length, quotemeta($left), "($re)", quotemeta($right)));
6160 my $left_re = qr/^\/\Q$left\E(\/|$)/;
6161 bless { left => $left, right => $right, left_regex => $left_re,
6162 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6163}
6164sub full_path {
6166 my ($self, $path) = @_;
6167 return (length $self->{left} ? "$self->{left}/" : '') .
6168 $path . (length $self->{right} ? "/$self->{right}" : '');
6169}
6170__END__
6172Data structures:
6174$remotes = { # returned by read_all_remotes()
6177 'svn' => {
6178 # svn-remote.svn.url=https://svn.musicpd.org
6179 url => 'https://svn.musicpd.org',
6180 # svn-remote.svn.fetch=mpd/trunk:trunk
6181 fetch => {
6182 'mpd/trunk' => 'trunk',
6183 },
6184 # svn-remote.svn.tags=mpd/tags/*:tags/*
6185 tags => {
6186 path => {
6187 left => 'mpd/tags',
6188 right => '',
6189 regex => qr!mpd/tags/([^/]+)$!,
6190 glob => 'tags/*',
6191 },
6192 ref => {
6193 left => 'tags',
6194 right => '',
6195 regex => qr!tags/([^/]+)$!,
6196 glob => 'tags/*',
6197 },
6198 }
6199 }
6200};
6201$log_entry hashref as returned by libsvn_log_entry()
6203{
6204 log => 'whitespace-formatted log entry
6205', # trailing newline is preserved
6206 revision => '8', # integer
6207 date => '2004-02-24T17:01:44.108345Z', # commit date
6208 author => 'committer name'
6209};
6210# this is generated by generate_diff();
6213@mods = array of diff-index line hashes, each element represents one line
6214 of diff-index output
6215diff-index line ($m hash)
6217{
6218 mode_a => first column of diff-index output, no leading ':',
6219 mode_b => second column of diff-index output,
6220 sha1_b => sha1sum of the final blob,
6221 chg => change type [MCRADT],
6222 file_a => original file name of a file (iff chg is 'C' or 'R')
6223 file_b => new/current file name of a file (any chg)
6224}
6225;
6226# retval of read_url_paths{,_all}();
6228$l_map = {
6229 # repository root url
6230 'https://svn.musicpd.org' => {
6231 # repository path # GIT_SVN_ID
6232 'mpd/trunk' => 'trunk',
6233 'mpd/tags/0.11.5' => 'tags/0.11.5',
6234 },
6235}
6236Notes:
6238 I don't trust the each() function on unless I created %hash myself
6239 because the internal iterator may not have started at base.