e5a553f0078fb87d6aab01a5ece4ff2ab94a1c89
1#!/usr/bin/env perl
2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3# License: GPL v2 or later
4use 5.008;
5use warnings;
6use strict;
7use vars qw/ $AUTHOR $VERSION
8 $sha1 $sha1_short $_revision $_repository
9 $_q $_authors $_authors_prog %users/;
10$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
11$VERSION = '@@GIT_VERSION@@';
12# From which subdir have we been invoked?
14my $cmd_dir_prefix = eval {
15 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
16} || '';
17my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
19$ENV{GIT_DIR} ||= '.git';
20$Git::SVN::default_repo_id = 'svn';
21$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
22$Git::SVN::Ra::_log_window_size = 100;
23$Git::SVN::_minimize_url = 'unset';
24if (! exists $ENV{SVN_SSH}) {
26 if (exists $ENV{GIT_SSH}) {
27 $ENV{SVN_SSH} = $ENV{GIT_SSH};
28 if ($^O eq 'msys') {
29 $ENV{SVN_SSH} =~ s/\\/\\\\/g;
30 $ENV{SVN_SSH} =~ s/(.*)/"$1"/;
31 }
32 }
33}
34$Git::SVN::Log::TZ = $ENV{TZ};
36$ENV{TZ} = 'UTC';
37$| = 1; # unbuffer STDOUT
38sub fatal (@) { print STDERR "@_\n"; exit 1 }
40sub _req_svn {
41 require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
42 require SVN::Ra;
43 require SVN::Delta;
44 if ($SVN::Core::VERSION lt '1.1.0') {
45 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
46 }
47}
48my $can_compress = eval { require Compress::Zlib; 1};
49push @Git::SVN::Ra::ISA, 'SVN::Ra';
50push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
51push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
52use Carp qw/croak/;
53use Digest::MD5;
54use IO::File qw//;
55use File::Basename qw/dirname basename/;
56use File::Path qw/mkpath/;
57use File::Spec;
58use File::Find;
59use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
60use IPC::Open3;
61use Git;
62BEGIN {
64 # import functions from Git into our packages, en masse
65 no strict 'refs';
66 foreach (qw/command command_oneline command_noisy command_output_pipe
67 command_input_pipe command_close_pipe
68 command_bidi_pipe command_close_bidi_pipe/) {
69 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
70 Git::SVN::Migration Git::SVN::Log Git::SVN),
71 __PACKAGE__) {
72 *{"${package}::$_"} = \&{"Git::$_"};
73 }
74 }
75}
76my ($SVN);
78$sha1 = qr/[a-f\d]{40}/;
80$sha1_short = qr/[a-f\d]{4,40}/;
81my ($_stdin, $_help, $_edit,
82 $_message, $_file, $_branch_dest,
83 $_template, $_shared,
84 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
85 $_merge, $_strategy, $_dry_run, $_local,
86 $_prefix, $_no_checkout, $_url, $_verbose,
87 $_git_format, $_commit_url, $_tag, $_merge_info);
88$Git::SVN::_follow_parent = 1;
89$_q ||= 0;
90my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
91 'config-dir=s' => \$Git::SVN::Ra::config_dir,
92 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
93 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
94my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
95 'authors-file|A=s' => \$_authors,
96 'authors-prog=s' => \$_authors_prog,
97 'repack:i' => \$Git::SVN::_repack,
98 'noMetadata' => \$Git::SVN::_no_metadata,
99 'useSvmProps' => \$Git::SVN::_use_svm_props,
100 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
101 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
102 'no-checkout' => \$_no_checkout,
103 'quiet|q+' => \$_q,
104 'repack-flags|repack-args|repack-opts=s' =>
105 \$Git::SVN::_repack_flags,
106 'use-log-author' => \$Git::SVN::_use_log_author,
107 'add-author-from' => \$Git::SVN::_add_author_from,
108 'localtime' => \$Git::SVN::_localtime,
109 %remote_opts );
110my ($_trunk, @_tags, @_branches, $_stdlayout);
112my %icv;
113my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
114 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
115 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
116 'stdlayout|s' => \$_stdlayout,
117 'minimize-url|m!' => \$Git::SVN::_minimize_url,
118 'no-metadata' => sub { $icv{noMetadata} = 1 },
119 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
120 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
121 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
122 'rewrite-uuid=s' => sub { $icv{rewriteUUID} = $_[1] },
123 %remote_opts );
124my %cmt_opts = ( 'edit|e' => \$_edit,
125 'rmdir' => \$SVN::Git::Editor::_rmdir,
126 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
127 'l=i' => \$SVN::Git::Editor::_rename_limit,
128 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
129);
130my %cmd = (
132 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
133 { 'revision|r=s' => \$_revision,
134 'fetch-all|all' => \$_fetch_all,
135 'parent|p' => \$_fetch_parent,
136 %fc_opts } ],
137 clone => [ \&cmd_clone, "Initialize and fetch revisions",
138 { 'revision|r=s' => \$_revision,
139 %fc_opts, %init_opts } ],
140 init => [ \&cmd_init, "Initialize a repo for tracking" .
141 " (requires URL argument)",
142 \%init_opts ],
143 'multi-init' => [ \&cmd_multi_init,
144 "Deprecated alias for ".
145 "'$0 init -T<trunk> -b<branches> -t<tags>'",
146 \%init_opts ],
147 dcommit => [ \&cmd_dcommit,
148 'Commit several diffs to merge with upstream',
149 { 'merge|m|M' => \$_merge,
150 'strategy|s=s' => \$_strategy,
151 'verbose|v' => \$_verbose,
152 'dry-run|n' => \$_dry_run,
153 'fetch-all|all' => \$_fetch_all,
154 'commit-url=s' => \$_commit_url,
155 'revision|r=i' => \$_revision,
156 'no-rebase' => \$_no_rebase,
157 'mergeinfo=s' => \$_merge_info,
158 %cmt_opts, %fc_opts } ],
159 branch => [ \&cmd_branch,
160 'Create a branch in the SVN repository',
161 { 'message|m=s' => \$_message,
162 'destination|d=s' => \$_branch_dest,
163 'dry-run|n' => \$_dry_run,
164 'tag|t' => \$_tag,
165 'username=s' => \$Git::SVN::Prompt::_username,
166 'commit-url=s' => \$_commit_url } ],
167 tag => [ sub { $_tag = 1; cmd_branch(@_) },
168 'Create a tag in the SVN repository',
169 { 'message|m=s' => \$_message,
170 'destination|d=s' => \$_branch_dest,
171 'dry-run|n' => \$_dry_run,
172 'username=s' => \$Git::SVN::Prompt::_username,
173 'commit-url=s' => \$_commit_url } ],
174 'set-tree' => [ \&cmd_set_tree,
175 "Set an SVN repository to a git tree-ish",
176 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
177 'create-ignore' => [ \&cmd_create_ignore,
178 'Create a .gitignore per svn:ignore',
179 { 'revision|r=i' => \$_revision
180 } ],
181 'mkdirs' => [ \&cmd_mkdirs ,
182 "recreate empty directories after a checkout",
183 { 'revision|r=i' => \$_revision } ],
184 'propget' => [ \&cmd_propget,
185 'Print the value of a property on a file or directory',
186 { 'revision|r=i' => \$_revision } ],
187 'proplist' => [ \&cmd_proplist,
188 'List all properties of a file or directory',
189 { 'revision|r=i' => \$_revision } ],
190 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
191 { 'revision|r=i' => \$_revision
192 } ],
193 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
194 { 'revision|r=i' => \$_revision
195 } ],
196 'multi-fetch' => [ \&cmd_multi_fetch,
197 "Deprecated alias for $0 fetch --all",
198 { 'revision|r=s' => \$_revision, %fc_opts } ],
199 'migrate' => [ sub { },
200 # no-op, we automatically run this anyways,
201 'Migrate configuration/metadata/layout from
202 previous versions of git-svn',
203 { 'minimize' => \$Git::SVN::Migration::_minimize,
204 %remote_opts } ],
205 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
206 { 'limit=i' => \$Git::SVN::Log::limit,
207 'revision|r=s' => \$_revision,
208 'verbose|v' => \$Git::SVN::Log::verbose,
209 'incremental' => \$Git::SVN::Log::incremental,
210 'oneline' => \$Git::SVN::Log::oneline,
211 'show-commit' => \$Git::SVN::Log::show_commit,
212 'non-recursive' => \$Git::SVN::Log::non_recursive,
213 'authors-file|A=s' => \$_authors,
214 'color' => \$Git::SVN::Log::color,
215 'pager=s' => \$Git::SVN::Log::pager
216 } ],
217 'find-rev' => [ \&cmd_find_rev,
218 "Translate between SVN revision numbers and tree-ish",
219 {} ],
220 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
221 { 'merge|m|M' => \$_merge,
222 'verbose|v' => \$_verbose,
223 'strategy|s=s' => \$_strategy,
224 'local|l' => \$_local,
225 'fetch-all|all' => \$_fetch_all,
226 'dry-run|n' => \$_dry_run,
227 %fc_opts } ],
228 'commit-diff' => [ \&cmd_commit_diff,
229 'Commit a diff between two trees',
230 { 'message|m=s' => \$_message,
231 'file|F=s' => \$_file,
232 'revision|r=s' => \$_revision,
233 %cmt_opts } ],
234 'info' => [ \&cmd_info,
235 "Show info about the latest SVN revision
236 on the current branch",
237 { 'url' => \$_url, } ],
238 'blame' => [ \&Git::SVN::Log::cmd_blame,
239 "Show what revision and author last modified each line of a file",
240 { 'git-format' => \$_git_format } ],
241 'reset' => [ \&cmd_reset,
242 "Undo fetches back to the specified SVN revision",
243 { 'revision|r=s' => \$_revision,
244 'parent|p' => \$_fetch_parent } ],
245 'gc' => [ \&cmd_gc,
246 "Compress unhandled.log files in .git/svn and remove " .
247 "index files in .git/svn",
248 {} ],
249);
250my $cmd;
252for (my $i = 0; $i < @ARGV; $i++) {
253 if (defined $cmd{$ARGV[$i]}) {
254 $cmd = $ARGV[$i];
255 splice @ARGV, $i, 1;
256 last;
257 } elsif ($ARGV[$i] eq 'help') {
258 $cmd = $ARGV[$i+1];
259 usage(0);
260 }
261};
262# make sure we're always running at the top-level working directory
264unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
265 unless (-d $ENV{GIT_DIR}) {
266 if ($git_dir_user_set) {
267 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
268 "but it is not a directory\n";
269 }
270 my $git_dir = delete $ENV{GIT_DIR};
271 my $cdup = undef;
272 git_cmd_try {
273 $cdup = command_oneline(qw/rev-parse --show-cdup/);
274 $git_dir = '.' unless ($cdup);
275 chomp $cdup if ($cdup);
276 $cdup = "." unless ($cdup && length $cdup);
277 } "Already at toplevel, but $git_dir not found\n";
278 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
279 unless (-d $git_dir) {
280 die "$git_dir still not found after going to ",
281 "'$cdup'\n";
282 }
283 $ENV{GIT_DIR} = $git_dir;
284 }
285 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
286}
287my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
289read_git_config(\%opts);
291if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
292 Getopt::Long::Configure('pass_through');
293}
294my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
295 'minimize-connections' => \$Git::SVN::Migration::_minimize,
296 'id|i=s' => \$Git::SVN::default_ref_id,
297 'svn-remote|remote|R=s' => sub {
298 $Git::SVN::no_reuse_existing = 1;
299 $Git::SVN::default_repo_id = $_[1] });
300exit 1 if (!$rv && $cmd && $cmd ne 'log');
301usage(0) if $_help;
303version() if $_version;
304usage(1) unless defined $cmd;
305load_authors() if $_authors;
306if (defined $_authors_prog) {
307 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
308}
309unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
311 Git::SVN::Migration::migration_check();
312}
313Git::SVN::init_vars();
314eval {
315 Git::SVN::verify_remotes_sanity();
316 $cmd{$cmd}->[0]->(@ARGV);
317};
318fatal $@ if $@;
319post_fetch_checkout();
320exit 0;
321####################### primary functions ######################
323sub usage {
324 my $exit = shift || 0;
325 my $fd = $exit ? \*STDERR : \*STDOUT;
326 print $fd <<"";
327git-svn - bidirectional operations between a single Subversion tree and git
328Usage: git svn <command> [options] [arguments]\n
329 print $fd "Available commands:\n" unless $cmd;
331 foreach (sort keys %cmd) {
333 next if $cmd && $cmd ne $_;
334 next if /^multi-/; # don't show deprecated commands
335 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
336 foreach (sort keys %{$cmd{$_}->[2]}) {
337 # mixed-case options are for .git/config only
338 next if /[A-Z]/ && /^[a-z]+$/i;
339 # prints out arguments as they should be passed:
340 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
341 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
342 "--$_" : "-$_" }
343 split /\|/,$_)," $x\n";
344 }
345 }
346 print $fd <<"";
347\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
348arbitrary identifier if you're tracking multiple SVN branches/repositories in
349one git repository and want to keep them separate. See git-svn(1) for more
350information.
351 exit $exit;
353}
354sub version {
356 ::_req_svn();
357 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
358 exit 0;
359}
360sub do_git_init_db {
362 unless (-d $ENV{GIT_DIR}) {
363 my @init_db = ('init');
364 push @init_db, "--template=$_template" if defined $_template;
365 if (defined $_shared) {
366 if ($_shared =~ /[a-z]/) {
367 push @init_db, "--shared=$_shared";
368 } else {
369 push @init_db, "--shared";
370 }
371 }
372 command_noisy(@init_db);
373 $_repository = Git->repository(Repository => ".git");
374 }
375 my $set;
376 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
377 foreach my $i (keys %icv) {
378 die "'$set' and '$i' cannot both be set\n" if $set;
379 next unless defined $icv{$i};
380 command_noisy('config', "$pfx.$i", $icv{$i});
381 $set = $i;
382 }
383 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
384 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
385 if defined $$ignore_regex;
386}
387sub init_subdir {
389 my $repo_path = shift or return;
390 mkpath([$repo_path]) unless -d $repo_path;
391 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
392 $ENV{GIT_DIR} = '.git';
393 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
394}
395sub cmd_clone {
397 my ($url, $path) = @_;
398 if (!defined $path &&
399 (defined $_trunk || @_branches || @_tags ||
400 defined $_stdlayout) &&
401 $url !~ m#^[a-z\+]+://#) {
402 $path = $url;
403 }
404 $path = basename($url) if !defined $path || !length $path;
405 my $authors_absolute = $_authors ? File::Spec->rel2abs($_authors) : "";
406 cmd_init($url, $path);
407 command_oneline('config', 'svn.authorsfile', $authors_absolute)
408 if $_authors;
409 Git::SVN::fetch_all($Git::SVN::default_repo_id);
410}
411sub cmd_init {
413 if (defined $_stdlayout) {
414 $_trunk = 'trunk' if (!defined $_trunk);
415 @_tags = 'tags' if (! @_tags);
416 @_branches = 'branches' if (! @_branches);
417 }
418 if (defined $_trunk || @_branches || @_tags) {
419 return cmd_multi_init(@_);
420 }
421 my $url = shift or die "SVN repository location required ",
422 "as a command-line argument\n";
423 $url = canonicalize_url($url);
424 init_subdir(@_);
425 do_git_init_db();
426 if ($Git::SVN::_minimize_url eq 'unset') {
428 $Git::SVN::_minimize_url = 0;
429 }
430 Git::SVN->init($url);
432}
433sub cmd_fetch {
435 if (grep /^\d+=./, @_) {
436 die "'<rev>=<commit>' fetch arguments are ",
437 "no longer supported.\n";
438 }
439 my ($remote) = @_;
440 if (@_ > 1) {
441 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
442 }
443 $Git::SVN::no_reuse_existing = undef;
444 if ($_fetch_parent) {
445 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
446 unless ($gs) {
447 die "Unable to determine upstream SVN information from ",
448 "working tree history\n";
449 }
450 # just fetch, don't checkout.
451 $_no_checkout = 'true';
452 $_fetch_all ? $gs->fetch_all : $gs->fetch;
453 } elsif ($_fetch_all) {
454 cmd_multi_fetch();
455 } else {
456 $remote ||= $Git::SVN::default_repo_id;
457 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
458 }
459}
460sub cmd_set_tree {
462 my (@commits) = @_;
463 if ($_stdin || !@commits) {
464 print "Reading from stdin...\n";
465 @commits = ();
466 while (<STDIN>) {
467 if (/\b($sha1_short)\b/o) {
468 unshift @commits, $1;
469 }
470 }
471 }
472 my @revs;
473 foreach my $c (@commits) {
474 my @tmp = command('rev-parse',$c);
475 if (scalar @tmp == 1) {
476 push @revs, $tmp[0];
477 } elsif (scalar @tmp > 1) {
478 push @revs, reverse(command('rev-list',@tmp));
479 } else {
480 fatal "Failed to rev-parse $c";
481 }
482 }
483 my $gs = Git::SVN->new;
484 my ($r_last, $cmt_last) = $gs->last_rev_commit;
485 $gs->fetch;
486 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
487 fatal "There are new revisions that were fetched ",
488 "and need to be merged (or acknowledged) ",
489 "before committing.\nlast rev: $r_last\n",
490 " current: $gs->{last_rev}";
491 }
492 $gs->set_tree($_) foreach @revs;
493 print "Done committing ",scalar @revs," revisions to SVN\n";
494 unlink $gs->{index};
495}
496sub cmd_dcommit {
498 my $head = shift;
499 command_noisy(qw/update-index --refresh/);
500 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
501 'Cannot dcommit with a dirty index. Commit your changes first, '
502 . "or stash them with `git stash'.\n";
503 $head ||= 'HEAD';
504 my $old_head;
506 if ($head ne 'HEAD') {
507 $old_head = eval {
508 command_oneline([qw/symbolic-ref -q HEAD/])
509 };
510 if ($old_head) {
511 $old_head =~ s{^refs/heads/}{};
512 } else {
513 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
514 }
515 command(['checkout', $head], STDERR => 0);
516 }
517 my @refs;
519 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
520 unless ($gs) {
521 die "Unable to determine upstream SVN information from ",
522 "$head history.\nPerhaps the repository is empty.";
523 }
524 if (defined $_commit_url) {
526 $url = $_commit_url;
527 } else {
528 $url = eval { command_oneline('config', '--get',
529 "svn-remote.$gs->{repo_id}.commiturl") };
530 if (!$url) {
531 $url = $gs->full_pushurl
532 }
533 }
534 my $last_rev = $_revision if defined $_revision;
536 if ($url) {
537 print "Committing to $url ...\n";
538 }
539 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
540 if ($_no_rebase && scalar(@$linear_refs) > 1) {
541 warn "Attempting to commit more than one change while ",
542 "--no-rebase is enabled.\n",
543 "If these changes depend on each other, re-running ",
544 "without --no-rebase may be required."
545 }
546 my $expect_url = $url;
547 Git::SVN::remove_username($expect_url);
548 while (1) {
549 my $d = shift @$linear_refs or last;
550 unless (defined $last_rev) {
551 (undef, $last_rev, undef) = cmt_metadata("$d~1");
552 unless (defined $last_rev) {
553 fatal "Unable to extract revision information ",
554 "from commit $d~1";
555 }
556 }
557 if ($_dry_run) {
558 print "diff-tree $d~1 $d\n";
559 } else {
560 my $cmt_rev;
561 my %ed_opts = ( r => $last_rev,
562 log => get_commit_entry($d)->{log},
563 ra => Git::SVN::Ra->new($url),
564 config => SVN::Core::config_get_config(
565 $Git::SVN::Ra::config_dir
566 ),
567 tree_a => "$d~1",
568 tree_b => $d,
569 editor_cb => sub {
570 print "Committed r$_[0]\n";
571 $cmt_rev = $_[0];
572 },
573 mergeinfo => $_merge_info,
574 svn_path => '');
575 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
576 print "No changes\n$d~1 == $d\n";
577 } elsif ($parents->{$d} && @{$parents->{$d}}) {
578 $gs->{inject_parents_dcommit}->{$cmt_rev} =
579 $parents->{$d};
580 }
581 $_fetch_all ? $gs->fetch_all : $gs->fetch;
582 $last_rev = $cmt_rev;
583 next if $_no_rebase;
584 # we always want to rebase against the current HEAD,
586 # not any head that was passed to us
587 my @diff = command('diff-tree', $d,
588 $gs->refname, '--');
589 my @finish;
590 if (@diff) {
591 @finish = rebase_cmd();
592 print STDERR "W: $d and ", $gs->refname,
593 " differ, using @finish:\n",
594 join("\n", @diff), "\n";
595 } else {
596 print "No changes between current HEAD and ",
597 $gs->refname,
598 "\nResetting to the latest ",
599 $gs->refname, "\n";
600 @finish = qw/reset --mixed/;
601 }
602 command_noisy(@finish, $gs->refname);
603 if (@diff) {
604 @refs = ();
605 my ($url_, $rev_, $uuid_, $gs_) =
606 working_head_info('HEAD', \@refs);
607 my ($linear_refs_, $parents_) =
608 linearize_history($gs_, \@refs);
609 if (scalar(@$linear_refs) !=
610 scalar(@$linear_refs_)) {
611 fatal "# of revisions changed ",
612 "\nbefore:\n",
613 join("\n", @$linear_refs),
614 "\n\nafter:\n",
615 join("\n", @$linear_refs_), "\n",
616 'If you are attempting to commit ',
617 "merges, try running:\n\t",
618 'git rebase --interactive',
619 '--preserve-merges ',
620 $gs->refname,
621 "\nBefore dcommitting";
622 }
623 if ($url_ ne $expect_url) {
624 if ($url_ eq $gs->metadata_url) {
625 print
626 "Accepting rewritten URL:",
627 " $url_\n";
628 } else {
629 fatal
630 "URL mismatch after rebase:",
631 " $url_ != $expect_url";
632 }
633 }
634 if ($uuid_ ne $uuid) {
635 fatal "uuid mismatch after rebase: ",
636 "$uuid_ != $uuid";
637 }
638 # remap parents
639 my (%p, @l, $i);
640 for ($i = 0; $i < scalar @$linear_refs; $i++) {
641 my $new = $linear_refs_->[$i] or next;
642 $p{$new} =
643 $parents->{$linear_refs->[$i]};
644 push @l, $new;
645 }
646 $parents = \%p;
647 $linear_refs = \@l;
648 }
649 }
650 }
651 if ($old_head) {
653 my $new_head = command_oneline(qw/rev-parse HEAD/);
654 my $new_is_symbolic = eval {
655 command_oneline(qw/symbolic-ref -q HEAD/);
656 };
657 if ($new_is_symbolic) {
658 print "dcommitted the branch ", $head, "\n";
659 } else {
660 print "dcommitted on a detached HEAD because you gave ",
661 "a revision argument.\n",
662 "The rewritten commit is: ", $new_head, "\n";
663 }
664 command(['checkout', $old_head], STDERR => 0);
665 }
666 unlink $gs->{index};
668}
669sub cmd_branch {
671 my ($branch_name, $head) = @_;
672 unless (defined $branch_name && length $branch_name) {
674 die(($_tag ? "tag" : "branch") . " name required\n");
675 }
676 $head ||= 'HEAD';
677 my (undef, $rev, undef, $gs) = working_head_info($head);
679 my $src = $gs->full_pushurl;
680 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
682 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
683 my $glob;
684 if ($#{$allglobs} == 0) {
685 $glob = $allglobs->[0];
686 } else {
687 unless(defined $_branch_dest) {
688 die "Multiple ",
689 $_tag ? "tag" : "branch",
690 " paths defined for Subversion repository.\n",
691 "You must specify where you want to create the ",
692 $_tag ? "tag" : "branch",
693 " with the --destination argument.\n";
694 }
695 foreach my $g (@{$allglobs}) {
696 # SVN::Git::Editor could probably be moved to Git.pm..
697 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
698 if ($_branch_dest =~ /$re/) {
699 $glob = $g;
700 last;
701 }
702 }
703 unless (defined $glob) {
704 my $dest_re = qr/\b\Q$_branch_dest\E\b/;
705 foreach my $g (@{$allglobs}) {
706 $g->{path}->{left} =~ /$dest_re/ or next;
707 if (defined $glob) {
708 die "Ambiguous destination: ",
709 $_branch_dest, "\nmatches both '",
710 $glob->{path}->{left}, "' and '",
711 $g->{path}->{left}, "'\n";
712 }
713 $glob = $g;
714 }
715 unless (defined $glob) {
716 die "Unknown ",
717 $_tag ? "tag" : "branch",
718 " destination $_branch_dest\n";
719 }
720 }
721 }
722 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
723 my $url;
724 if (defined $_commit_url) {
725 $url = $_commit_url;
726 } else {
727 $url = eval { command_oneline('config', '--get',
728 "svn-remote.$gs->{repo_id}.commiturl") };
729 if (!$url) {
730 $url = $remote->{pushurl} || $remote->{url};
731 }
732 }
733 my $dst = join '/', $url, $lft, $branch_name, ($rgt || ());
734 if ($dst =~ /^https:/ && $src =~ /^http:/) {
736 $src=~s/^http:/https:/;
737 }
738 ::_req_svn();
740 my $ctx = SVN::Client->new(
742 auth => Git::SVN::Ra::_auth_providers(),
743 log_msg => sub {
744 ${ $_[0] } = defined $_message
745 ? $_message
746 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
747 . $branch_name;
748 },
749 );
750 eval {
752 $ctx->ls($dst, 'HEAD', 0);
753 } and die "branch ${branch_name} already exists\n";
754 print "Copying ${src} at r${rev} to ${dst}...\n";
756 $ctx->copy($src, $rev, $dst)
757 unless $_dry_run;
758 $gs->fetch_all;
760}
761sub cmd_find_rev {
763 my $revision_or_hash = shift or die "SVN or git revision required ",
764 "as a command-line argument\n";
765 my $result;
766 if ($revision_or_hash =~ /^r\d+$/) {
767 my $head = shift;
768 $head ||= 'HEAD';
769 my @refs;
770 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
771 unless ($gs) {
772 die "Unable to determine upstream SVN information from ",
773 "$head history\n";
774 }
775 my $desired_revision = substr($revision_or_hash, 1);
776 $result = $gs->rev_map_get($desired_revision, $uuid);
777 } else {
778 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
779 $result = $rev;
780 }
781 print "$result\n" if $result;
782}
783sub cmd_rebase {
785 command_noisy(qw/update-index --refresh/);
786 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
787 unless ($gs) {
788 die "Unable to determine upstream SVN information from ",
789 "working tree history\n";
790 }
791 if ($_dry_run) {
792 print "Remote Branch: " . $gs->refname . "\n";
793 print "SVN URL: " . $url . "\n";
794 return;
795 }
796 if (command(qw/diff-index HEAD --/)) {
797 print STDERR "Cannot rebase with uncommited changes:\n";
798 command_noisy('status');
799 exit 1;
800 }
801 unless ($_local) {
802 # rebase will checkout for us, so no need to do it explicitly
803 $_no_checkout = 'true';
804 $_fetch_all ? $gs->fetch_all : $gs->fetch;
805 }
806 command_noisy(rebase_cmd(), $gs->refname);
807 $gs->mkemptydirs;
808}
809sub cmd_show_ignore {
811 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
812 $gs ||= Git::SVN->new;
813 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
814 $gs->prop_walk($gs->{path}, $r, sub {
815 my ($gs, $path, $props) = @_;
816 print STDOUT "\n# $path\n";
817 my $s = $props->{'svn:ignore'} or return;
818 $s =~ s/[\r\n]+/\n/g;
819 $s =~ s/^\n+//;
820 chomp $s;
821 $s =~ s#^#$path#gm;
822 print STDOUT "$s\n";
823 });
824}
825sub cmd_show_externals {
827 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
828 $gs ||= Git::SVN->new;
829 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
830 $gs->prop_walk($gs->{path}, $r, sub {
831 my ($gs, $path, $props) = @_;
832 print STDOUT "\n# $path\n";
833 my $s = $props->{'svn:externals'} or return;
834 $s =~ s/[\r\n]+/\n/g;
835 chomp $s;
836 $s =~ s#^#$path#gm;
837 print STDOUT "$s\n";
838 });
839}
840sub cmd_create_ignore {
842 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
843 $gs ||= Git::SVN->new;
844 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
845 $gs->prop_walk($gs->{path}, $r, sub {
846 my ($gs, $path, $props) = @_;
847 # $path is of the form /path/to/dir/
848 $path = '.' . $path;
849 # SVN can have attributes on empty directories,
850 # which git won't track
851 mkpath([$path]) unless -d $path;
852 my $ignore = $path . '.gitignore';
853 my $s = $props->{'svn:ignore'} or return;
854 open(GITIGNORE, '>', $ignore)
855 or fatal("Failed to open `$ignore' for writing: $!");
856 $s =~ s/[\r\n]+/\n/g;
857 $s =~ s/^\n+//;
858 chomp $s;
859 # Prefix all patterns so that the ignore doesn't apply
860 # to sub-directories.
861 $s =~ s#^#/#gm;
862 print GITIGNORE "$s\n";
863 close(GITIGNORE)
864 or fatal("Failed to close `$ignore': $!");
865 command_noisy('add', '-f', $ignore);
866 });
867}
868sub cmd_mkdirs {
870 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
871 $gs ||= Git::SVN->new;
872 $gs->mkemptydirs($_revision);
873}
874sub canonicalize_path {
876 my ($path) = @_;
877 my $dot_slash_added = 0;
878 if (substr($path, 0, 1) ne "/") {
879 $path = "./" . $path;
880 $dot_slash_added = 1;
881 }
882 # File::Spec->canonpath doesn't collapse x/../y into y (for a
883 # good reason), so let's do this manually.
884 $path =~ s#/+#/#g;
885 $path =~ s#/\.(?:/|$)#/#g;
886 $path =~ s#/[^/]+/\.\.##g;
887 $path =~ s#/$##g;
888 $path =~ s#^\./## if $dot_slash_added;
889 $path =~ s#^/##;
890 $path =~ s#^\.$##;
891 return $path;
892}
893sub canonicalize_url {
895 my ($url) = @_;
896 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
897 return $url;
898}
899# get_svnprops(PATH)
901# ------------------
902# Helper for cmd_propget and cmd_proplist below.
903sub get_svnprops {
904 my $path = shift;
905 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
906 $gs ||= Git::SVN->new;
907 # prefix THE PATH by the sub-directory from which the user
909 # invoked us.
910 $path = $cmd_dir_prefix . $path;
911 fatal("No such file or directory: $path") unless -e $path;
912 my $is_dir = -d $path ? 1 : 0;
913 $path = $gs->{path} . '/' . $path;
914 # canonicalize the path (otherwise libsvn will abort or fail to
916 # find the file)
917 $path = canonicalize_path($path);
918 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
920 my $props;
921 if ($is_dir) {
922 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
923 }
924 else {
925 (undef, $props) = $gs->ra->get_file($path, $r, undef);
926 }
927 return $props;
928}
929# cmd_propget (PROP, PATH)
931# ------------------------
932# Print the SVN property PROP for PATH.
933sub cmd_propget {
934 my ($prop, $path) = @_;
935 $path = '.' if not defined $path;
936 usage(1) if not defined $prop;
937 my $props = get_svnprops($path);
938 if (not defined $props->{$prop}) {
939 fatal("`$path' does not have a `$prop' SVN property.");
940 }
941 print $props->{$prop} . "\n";
942}
943# cmd_proplist (PATH)
945# -------------------
946# Print the list of SVN properties for PATH.
947sub cmd_proplist {
948 my $path = shift;
949 $path = '.' if not defined $path;
950 my $props = get_svnprops($path);
951 print "Properties on '$path':\n";
952 foreach (sort keys %{$props}) {
953 print " $_\n";
954 }
955}
956sub cmd_multi_init {
958 my $url = shift;
959 unless (defined $_trunk || @_branches || @_tags) {
960 usage(1);
961 }
962 $_prefix = '' unless defined $_prefix;
964 if (defined $url) {
965 $url = canonicalize_url($url);
966 init_subdir(@_);
967 }
968 do_git_init_db();
969 if (defined $_trunk) {
970 $_trunk =~ s#^/+##;
971 my $trunk_ref = 'refs/remotes/' . $_prefix . 'trunk';
972 # try both old-style and new-style lookups:
973 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
974 unless ($gs_trunk) {
975 my ($trunk_url, $trunk_path) =
976 complete_svn_url($url, $_trunk);
977 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
978 undef, $trunk_ref);
979 }
980 }
981 return unless @_branches || @_tags;
982 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
983 foreach my $path (@_branches) {
984 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
985 }
986 foreach my $path (@_tags) {
987 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
988 }
989}
990sub cmd_multi_fetch {
992 $Git::SVN::no_reuse_existing = undef;
993 my $remotes = Git::SVN::read_all_remotes();
994 foreach my $repo_id (sort keys %$remotes) {
995 if ($remotes->{$repo_id}->{url}) {
996 Git::SVN::fetch_all($repo_id, $remotes);
997 }
998 }
999}
1000# this command is special because it requires no metadata
1002sub cmd_commit_diff {
1003 my ($ta, $tb, $url) = @_;
1004 my $usage = "Usage: $0 commit-diff -r<revision> ".
1005 "<tree-ish> <tree-ish> [<URL>]";
1006 fatal($usage) if (!defined $ta || !defined $tb);
1007 my $svn_path = '';
1008 if (!defined $url) {
1009 my $gs = eval { Git::SVN->new };
1010 if (!$gs) {
1011 fatal("Needed URL or usable git-svn --id in ",
1012 "the command-line\n", $usage);
1013 }
1014 $url = $gs->{url};
1015 $svn_path = $gs->{path};
1016 }
1017 unless (defined $_revision) {
1018 fatal("-r|--revision is a required argument\n", $usage);
1019 }
1020 if (defined $_message && defined $_file) {
1021 fatal("Both --message/-m and --file/-F specified ",
1022 "for the commit message.\n",
1023 "I have no idea what you mean");
1024 }
1025 if (defined $_file) {
1026 $_message = file_to_s($_file);
1027 } else {
1028 $_message ||= get_commit_entry($tb)->{log};
1029 }
1030 my $ra ||= Git::SVN::Ra->new($url);
1031 my $r = $_revision;
1032 if ($r eq 'HEAD') {
1033 $r = $ra->get_latest_revnum;
1034 } elsif ($r !~ /^\d+$/) {
1035 die "revision argument: $r not understood by git-svn\n";
1036 }
1037 my %ed_opts = ( r => $r,
1038 log => $_message,
1039 ra => $ra,
1040 tree_a => $ta,
1041 tree_b => $tb,
1042 editor_cb => sub { print "Committed r$_[0]\n" },
1043 svn_path => $svn_path );
1044 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
1045 print "No changes\n$ta == $tb\n";
1046 }
1047}
1048sub escape_uri_only {
1050 my ($uri) = @_;
1051 my @tmp;
1052 foreach (split m{/}, $uri) {
1053 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
1054 push @tmp, $_;
1055 }
1056 join('/', @tmp);
1057}
1058sub escape_url {
1060 my ($url) = @_;
1061 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
1062 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
1063 $url = "$scheme://$domain$uri";
1064 }
1065 $url;
1066}
1067sub cmd_info {
1069 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
1070 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
1071 if (exists $_[1]) {
1072 die "Too many arguments specified\n";
1073 }
1074 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1076 if (!$file_type && !$diff_status) {
1078 print STDERR "svn: '$path' is not under version control\n";
1079 exit 1;
1080 }
1081 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1083 unless ($gs) {
1084 die "Unable to determine upstream SVN information from ",
1085 "working tree history\n";
1086 }
1087 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1089 $path = "." if $path eq "";
1090 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1092 if ($_url) {
1094 print escape_url($full_url), "\n";
1095 return;
1096 }
1097 my $result = "Path: $path\n";
1099 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1100 $result .= "URL: " . escape_url($full_url) . "\n";
1101 eval {
1103 my $repos_root = $gs->repos_root;
1104 Git::SVN::remove_username($repos_root);
1105 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1106 };
1107 if ($@) {
1108 $result .= "Repository Root: (offline)\n";
1109 }
1110 ::_req_svn();
1111 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1112 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1113 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1114 $result .= "Node Kind: " .
1116 ($file_type eq "dir" ? "directory" : "file") . "\n";
1117 my $schedule = $diff_status eq "A"
1119 ? "add"
1120 : ($diff_status eq "D" ? "delete" : "normal");
1121 $result .= "Schedule: $schedule\n";
1122 if ($diff_status eq "A") {
1124 print $result, "\n";
1125 return;
1126 }
1127 my ($lc_author, $lc_rev, $lc_date_utc);
1129 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1130 my $log = command_output_pipe(@args);
1131 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1132 while (<$log>) {
1133 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1134 $lc_author = $1;
1135 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1136 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1137 (undef, $lc_rev, undef) = ::extract_metadata($1);
1138 }
1139 }
1140 close $log;
1141 Git::SVN::Log::set_local_timezone();
1143 $result .= "Last Changed Author: $lc_author\n";
1145 $result .= "Last Changed Rev: $lc_rev\n";
1146 $result .= "Last Changed Date: " .
1147 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1148 if ($file_type ne "dir") {
1150 my $text_last_updated_date =
1151 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1152 $result .=
1153 "Text Last Updated: " .
1154 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1155 "\n";
1156 my $checksum;
1157 if ($diff_status eq "D") {
1158 my ($fh, $ctx) =
1159 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1160 if ($file_type eq "link") {
1161 my $file_name = <$fh>;
1162 $checksum = md5sum("link $file_name");
1163 } else {
1164 $checksum = md5sum($fh);
1165 }
1166 command_close_pipe($fh, $ctx);
1167 } elsif ($file_type eq "link") {
1168 my $file_name =
1169 command(qw(cat-file blob), "HEAD:$path");
1170 $checksum =
1171 md5sum("link " . $file_name);
1172 } else {
1173 open FILE, "<", $path or die $!;
1174 $checksum = md5sum(\*FILE);
1175 close FILE or die $!;
1176 }
1177 $result .= "Checksum: " . $checksum . "\n";
1178 }
1179 print $result, "\n";
1181}
1182sub cmd_reset {
1184 my $target = shift || $_revision or die "SVN revision required\n";
1185 $target = $1 if $target =~ /^r(\d+)$/;
1186 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1187 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1188 unless ($gs) {
1189 die "Unable to determine upstream SVN information from ".
1190 "history\n";
1191 }
1192 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1193 die "Cannot find SVN revision $target\n" unless defined($c);
1194 $gs->rev_map_set($r, $c, 'reset', $uuid);
1195 print "r$r = $c ($gs->{ref_id})\n";
1196}
1197sub cmd_gc {
1199 if (!$can_compress) {
1200 warn "Compress::Zlib could not be found; unhandled.log " .
1201 "files will not be compressed.\n";
1202 }
1203 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1204}
1205########################### utility functions #########################
1207sub rebase_cmd {
1209 my @cmd = qw/rebase/;
1210 push @cmd, '-v' if $_verbose;
1211 push @cmd, qw/--merge/ if $_merge;
1212 push @cmd, "--strategy=$_strategy" if $_strategy;
1213 @cmd;
1214}
1215sub post_fetch_checkout {
1217 return if $_no_checkout;
1218 my $gs = $Git::SVN::_head or return;
1219 return if verify_ref('refs/heads/master^0');
1220 # look for "trunk" ref if it exists
1222 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
1223 my $fetch = $remote->{fetch};
1224 if ($fetch) {
1225 foreach my $p (keys %$fetch) {
1226 basename($fetch->{$p}) eq 'trunk' or next;
1227 $gs = Git::SVN->new($fetch->{$p}, $gs->{repo_id}, $p);
1228 last;
1229 }
1230 }
1231 my $valid_head = verify_ref('HEAD^0');
1233 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1234 return if ($valid_head || !verify_ref('HEAD^0'));
1235 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1237 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1238 return if -f $index;
1239 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1241 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1242 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1243 print STDERR "Checked out HEAD:\n ",
1244 $gs->full_url, " r", $gs->last_rev, "\n";
1245 $gs->mkemptydirs($gs->last_rev);
1246}
1247sub complete_svn_url {
1249 my ($url, $path) = @_;
1250 $path =~ s#/+$##;
1251 if ($path !~ m#^[a-z\+]+://#) {
1252 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1253 fatal("E: '$path' is not a complete URL ",
1254 "and a separate URL is not specified");
1255 }
1256 return ($url, $path);
1257 }
1258 return ($path, '');
1259}
1260sub complete_url_ls_init {
1262 my ($ra, $repo_path, $switch, $pfx) = @_;
1263 unless ($repo_path) {
1264 print STDERR "W: $switch not specified\n";
1265 return;
1266 }
1267 $repo_path =~ s#/+$##;
1268 if ($repo_path =~ m#^[a-z\+]+://#) {
1269 $ra = Git::SVN::Ra->new($repo_path);
1270 $repo_path = '';
1271 } else {
1272 $repo_path =~ s#^/+##;
1273 unless ($ra) {
1274 fatal("E: '$repo_path' is not a complete URL ",
1275 "and a separate URL is not specified");
1276 }
1277 }
1278 my $url = $ra->{url};
1279 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1280 my $k = "svn-remote.$gs->{repo_id}.url";
1281 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1282 if ($orig_url && ($orig_url ne $gs->{url})) {
1283 die "$k already set: $orig_url\n",
1284 "wanted to set to: $gs->{url}\n";
1285 }
1286 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1287 my $remote_path = "$gs->{path}/$repo_path";
1288 $remote_path =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1289 $remote_path =~ s#/+#/#g;
1290 $remote_path =~ s#^/##g;
1291 $remote_path .= "/*" if $remote_path !~ /\*/;
1292 my ($n) = ($switch =~ /^--(\w+)/);
1293 if (length $pfx && $pfx !~ m#/$#) {
1294 die "--prefix='$pfx' must have a trailing slash '/'\n";
1295 }
1296 command_noisy('config',
1297 '--add',
1298 "svn-remote.$gs->{repo_id}.$n",
1299 "$remote_path:refs/remotes/$pfx*" .
1300 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1301}
1302sub verify_ref {
1304 my ($ref) = @_;
1305 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1306 { STDERR => 0 }); };
1307}
1308sub get_tree_from_treeish {
1310 my ($treeish) = @_;
1311 # $treeish can be a symbolic ref, too:
1312 my $type = command_oneline(qw/cat-file -t/, $treeish);
1313 my $expected;
1314 while ($type eq 'tag') {
1315 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1316 }
1317 if ($type eq 'commit') {
1318 $expected = (grep /^tree /, command(qw/cat-file commit/,
1319 $treeish))[0];
1320 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1321 die "Unable to get tree from $treeish\n" unless $expected;
1322 } elsif ($type eq 'tree') {
1323 $expected = $treeish;
1324 } else {
1325 die "$treeish is a $type, expected tree, tag or commit\n";
1326 }
1327 return $expected;
1328}
1329sub get_commit_entry {
1331 my ($treeish) = shift;
1332 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1333 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1334 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1335 open my $log_fh, '>', $commit_editmsg or croak $!;
1336 my $type = command_oneline(qw/cat-file -t/, $treeish);
1338 if ($type eq 'commit' || $type eq 'tag') {
1339 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1340 $type, $treeish);
1341 my $in_msg = 0;
1342 my $author;
1343 my $saw_from = 0;
1344 my $msgbuf = "";
1345 while (<$msg_fh>) {
1346 if (!$in_msg) {
1347 $in_msg = 1 if (/^\s*$/);
1348 $author = $1 if (/^author (.*>)/);
1349 } elsif (/^git-svn-id: /) {
1350 # skip this for now, we regenerate the
1351 # correct one on re-fetch anyways
1352 # TODO: set *:merge properties or like...
1353 } else {
1354 if (/^From:/ || /^Signed-off-by:/) {
1355 $saw_from = 1;
1356 }
1357 $msgbuf .= $_;
1358 }
1359 }
1360 $msgbuf =~ s/\s+$//s;
1361 if ($Git::SVN::_add_author_from && defined($author)
1362 && !$saw_from) {
1363 $msgbuf .= "\n\nFrom: $author";
1364 }
1365 print $log_fh $msgbuf or croak $!;
1366 command_close_pipe($msg_fh, $ctx);
1367 }
1368 close $log_fh or croak $!;
1369 if ($_edit || ($type eq 'tree')) {
1371 chomp(my $editor = command_oneline(qw(var GIT_EDITOR)));
1372 system('sh', '-c', $editor.' "$@"', $editor, $commit_editmsg);
1373 }
1374 rename $commit_editmsg, $commit_msg or croak $!;
1375 {
1376 require Encode;
1377 # SVN requires messages to be UTF-8 when entering the repo
1378 local $/;
1379 open $log_fh, '<', $commit_msg or croak $!;
1380 binmode $log_fh;
1381 chomp($log_entry{log} = <$log_fh>);
1382 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1384 my $msg = $log_entry{log};
1385 eval { $msg = Encode::decode($enc, $msg, 1) };
1387 if ($@) {
1388 die "Could not decode as $enc:\n", $msg,
1389 "\nPerhaps you need to set i18n.commitencoding\n";
1390 }
1391 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1393 die "Could not encode as UTF-8:\n$msg\n" if $@;
1394 $log_entry{log} = $msg;
1396 close $log_fh or croak $!;
1398 }
1399 unlink $commit_msg;
1400 \%log_entry;
1401}
1402sub s_to_file {
1404 my ($str, $file, $mode) = @_;
1405 open my $fd,'>',$file or croak $!;
1406 print $fd $str,"\n" or croak $!;
1407 close $fd or croak $!;
1408 chmod ($mode &~ umask, $file) if (defined $mode);
1409}
1410sub file_to_s {
1412 my $file = shift;
1413 open my $fd,'<',$file or croak "$!: file: $file\n";
1414 local $/;
1415 my $ret = <$fd>;
1416 close $fd or croak $!;
1417 $ret =~ s/\s*$//s;
1418 return $ret;
1419}
1420# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1422sub load_authors {
1423 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1424 my $log = $cmd eq 'log';
1425 while (<$authors>) {
1426 chomp;
1427 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1428 my ($user, $name, $email) = ($1, $2, $3);
1429 if ($log) {
1430 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1431 } else {
1432 $users{$user} = [$name, $email];
1433 }
1434 }
1435 close $authors or croak $!;
1436}
1437# convert GetOpt::Long specs for use by git-config
1439sub read_git_config {
1440 my $opts = shift;
1441 my @config_only;
1442 foreach my $o (keys %$opts) {
1443 # if we have mixedCase and a long option-only, then
1444 # it's a config-only variable that we don't need for
1445 # the command-line.
1446 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1447 my $v = $opts->{$o};
1448 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1449 $key =~ s/-//g;
1450 my $arg = 'git config';
1451 $arg .= ' --int' if ($o =~ /[:=]i$/);
1452 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1453 if (ref $v eq 'ARRAY') {
1454 chomp(my @tmp = `$arg --get-all svn.$key`);
1455 @$v = @tmp if @tmp;
1456 } else {
1457 chomp(my $tmp = `$arg --get svn.$key`);
1458 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1459 $$v = $tmp;
1460 }
1461 }
1462 }
1463 delete @$opts{@config_only} if @config_only;
1464}
1465sub extract_metadata {
1467 my $id = shift or return (undef, undef, undef);
1468 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1469 \s([a-f\d\-]+)$/ix);
1470 if (!defined $rev || !$uuid || !$url) {
1471 # some of the original repositories I made had
1472 # identifiers like this:
1473 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1474 }
1475 return ($url, $rev, $uuid);
1476}
1477sub cmt_metadata {
1479 return extract_metadata((grep(/^git-svn-id: /,
1480 command(qw/cat-file commit/, shift)))[-1]);
1481}
1482sub cmt_sha2rev_batch {
1484 my %s2r;
1485 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1486 my $list = shift;
1487 foreach my $sha (@{$list}) {
1489 my $first = 1;
1490 my $size = 0;
1491 print $out $sha, "\n";
1492 while (my $line = <$in>) {
1494 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1495 last;
1496 } elsif ($first &&
1497 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1498 $first = 0;
1499 $size = $1;
1500 next;
1501 } elsif ($line =~ /^(git-svn-id: )/) {
1502 my (undef, $rev, undef) =
1503 extract_metadata($line);
1504 $s2r{$sha} = $rev;
1505 }
1506 $size -= length($line);
1508 last if ($size == 0);
1509 }
1510 }
1511 command_close_bidi_pipe($pid, $in, $out, $ctx);
1513 return \%s2r;
1515}
1516sub working_head_info {
1518 my ($head, $refs) = @_;
1519 my @args = qw/log --no-color --no-decorate --first-parent
1520 --pretty=medium/;
1521 my ($fh, $ctx) = command_output_pipe(@args, $head);
1522 my $hash;
1523 my %max;
1524 while (<$fh>) {
1525 if ( m{^commit ($::sha1)$} ) {
1526 unshift @$refs, $hash if $hash and $refs;
1527 $hash = $1;
1528 next;
1529 }
1530 next unless s{^\s*(git-svn-id:)}{$1};
1531 my ($url, $rev, $uuid) = extract_metadata($_);
1532 if (defined $url && defined $rev) {
1533 next if $max{$url} and $max{$url} < $rev;
1534 if (my $gs = Git::SVN->find_by_url($url)) {
1535 my $c = $gs->rev_map_get($rev, $uuid);
1536 if ($c && $c eq $hash) {
1537 close $fh; # break the pipe
1538 return ($url, $rev, $uuid, $gs);
1539 } else {
1540 $max{$url} ||= $gs->rev_map_max;
1541 }
1542 }
1543 }
1544 }
1545 command_close_pipe($fh, $ctx);
1546 (undef, undef, undef, undef);
1547}
1548sub read_commit_parents {
1550 my ($parents, $c) = @_;
1551 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1552 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1553 @{$parents->{$c}} = split(/ /, $p);
1554}
1555sub linearize_history {
1557 my ($gs, $refs) = @_;
1558 my %parents;
1559 foreach my $c (@$refs) {
1560 read_commit_parents(\%parents, $c);
1561 }
1562 my @linear_refs;
1564 my %skip = ();
1565 my $last_svn_commit = $gs->last_commit;
1566 foreach my $c (reverse @$refs) {
1567 next if $c eq $last_svn_commit;
1568 last if $skip{$c};
1569 unshift @linear_refs, $c;
1571 $skip{$c} = 1;
1572 # we only want the first parent to diff against for linear
1574 # history, we save the rest to inject when we finalize the
1575 # svn commit
1576 my $fp_a = verify_ref("$c~1");
1577 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1578 if (!$fp_a || !$fp_b) {
1579 die "Commit $c\n",
1580 "has no parent commit, and therefore ",
1581 "nothing to diff against.\n",
1582 "You should be working from a repository ",
1583 "originally created by git-svn\n";
1584 }
1585 if ($fp_a ne $fp_b) {
1586 die "$c~1 = $fp_a, however parsing commit $c ",
1587 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1588 }
1589 foreach my $p (@{$parents{$c}}) {
1591 $skip{$p} = 1;
1592 }
1593 }
1594 (\@linear_refs, \%parents);
1595}
1596sub find_file_type_and_diff_status {
1598 my ($path) = @_;
1599 return ('dir', '') if $path eq '';
1600 my $diff_output =
1602 command_oneline(qw(diff --cached --name-status --), $path) || "";
1603 my $diff_status = (split(' ', $diff_output))[0] || "";
1604 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1606 return (undef, undef) if !$diff_status && !$ls_tree;
1608 if ($diff_status eq "A") {
1610 return ("link", $diff_status) if -l $path;
1611 return ("dir", $diff_status) if -d $path;
1612 return ("file", $diff_status);
1613 }
1614 my $mode = (split(' ', $ls_tree))[0] || "";
1616 return ("link", $diff_status) if $mode eq "120000";
1618 return ("dir", $diff_status) if $mode eq "040000";
1619 return ("file", $diff_status);
1620}
1621sub md5sum {
1623 my $arg = shift;
1624 my $ref = ref $arg;
1625 my $md5 = Digest::MD5->new();
1626 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1627 $md5->addfile($arg) or croak $!;
1628 } elsif ($ref eq 'SCALAR') {
1629 $md5->add($$arg) or croak $!;
1630 } elsif (!$ref) {
1631 $md5->add($arg) or croak $!;
1632 } else {
1633 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1634 }
1635 return $md5->hexdigest();
1636}
1637sub gc_directory {
1639 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1640 my $out_filename = $_ . ".gz";
1641 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1642 binmode $in_fh;
1643 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1644 die "Unable to open $out_filename: $!\n";
1645 my $res;
1647 while ($res = sysread($in_fh, my $str, 1024)) {
1648 $gz->gzwrite($str) or
1649 die "Unable to write: ".$gz->gzerror()."!\n";
1650 }
1651 unlink $_ or die "unlink $File::Find::name: $!\n";
1652 } elsif (-f $_ && basename($_) eq "index") {
1653 unlink $_ or die "unlink $_: $!\n";
1654 }
1655}
1656package Git::SVN;
1658use strict;
1659use warnings;
1660use Fcntl qw/:DEFAULT :seek/;
1661use constant rev_map_fmt => 'NH40';
1662use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1663 $_repack $_repack_flags $_use_svm_props $_head
1664 $_use_svnsync_props $no_reuse_existing $_minimize_url
1665 $_use_log_author $_add_author_from $_localtime/;
1666use Carp qw/croak/;
1667use File::Path qw/mkpath/;
1668use File::Copy qw/copy/;
1669use IPC::Open3;
1670use Memoize; # core since 5.8.0, Jul 2002
1671use Memoize::Storable;
1672my ($_gc_nr, $_gc_period);
1674# properties that we do not log:
1676my %SKIP_PROP;
1677BEGIN {
1678 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1679 svn:special svn:executable
1680 svn:entry:committed-rev
1681 svn:entry:last-author
1682 svn:entry:uuid
1683 svn:entry:committed-date/;
1684 # some options are read globally, but can be overridden locally
1686 # per [svn-remote "..."] section. Command-line options will *NOT*
1687 # override options set in an [svn-remote "..."] section
1688 no strict 'refs';
1689 for my $option (qw/follow_parent no_metadata use_svm_props
1690 use_svnsync_props/) {
1691 my $key = $option;
1692 $key =~ tr/_//d;
1693 my $prop = "-$option";
1694 *$option = sub {
1695 my ($self) = @_;
1696 return $self->{$prop} if exists $self->{$prop};
1697 my $k = "svn-remote.$self->{repo_id}.$key";
1698 eval { command_oneline(qw/config --get/, $k) };
1699 if ($@) {
1700 $self->{$prop} = ${"Git::SVN::_$option"};
1701 } else {
1702 my $v = command_oneline(qw/config --bool/,$k);
1703 $self->{$prop} = $v eq 'false' ? 0 : 1;
1704 }
1705 return $self->{$prop};
1706 }
1707 }
1708}
1709my (%LOCKFILES, %INDEX_FILES);
1712END {
1713 unlink keys %LOCKFILES if %LOCKFILES;
1714 unlink keys %INDEX_FILES if %INDEX_FILES;
1715}
1716sub resolve_local_globs {
1718 my ($url, $fetch, $glob_spec) = @_;
1719 return unless defined $glob_spec;
1720 my $ref = $glob_spec->{ref};
1721 my $path = $glob_spec->{path};
1722 foreach (command(qw#for-each-ref --format=%(refname) refs/#)) {
1723 next unless m#^$ref->{regex}$#;
1724 my $p = $1;
1725 my $pathname = desanitize_refname($path->full_path($p));
1726 my $refname = desanitize_refname($ref->full_path($p));
1727 if (my $existing = $fetch->{$pathname}) {
1728 if ($existing ne $refname) {
1729 die "Refspec conflict:\n",
1730 "existing: $existing\n",
1731 " globbed: $refname\n";
1732 }
1733 my $u = (::cmt_metadata("$refname"))[0];
1734 $u =~ s!^\Q$url\E(/|$)!! or die
1735 "$refname: '$url' not found in '$u'\n";
1736 if ($pathname ne $u) {
1737 warn "W: Refspec glob conflict ",
1738 "(ref: $refname):\n",
1739 "expected path: $pathname\n",
1740 " real path: $u\n",
1741 "Continuing ahead with $u\n";
1742 next;
1743 }
1744 } else {
1745 $fetch->{$pathname} = $refname;
1746 }
1747 }
1748}
1749sub parse_revision_argument {
1751 my ($base, $head) = @_;
1752 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1753 return ($base, $head);
1754 }
1755 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1756 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1757 return ($head, $head) if ($::_revision eq 'HEAD');
1758 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1759 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1760 die "revision argument: $::_revision not understood by git-svn\n";
1761}
1762sub fetch_all {
1764 my ($repo_id, $remotes) = @_;
1765 if (ref $repo_id) {
1766 my $gs = $repo_id;
1767 $repo_id = undef;
1768 $repo_id = $gs->{repo_id};
1769 }
1770 $remotes ||= read_all_remotes();
1771 my $remote = $remotes->{$repo_id} or
1772 die "[svn-remote \"$repo_id\"] unknown\n";
1773 my $fetch = $remote->{fetch};
1774 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1775 my (@gs, @globs);
1776 my $ra = Git::SVN::Ra->new($url);
1777 my $uuid = $ra->get_uuid;
1778 my $head = $ra->get_latest_revnum;
1779 # ignore errors, $head revision may not even exist anymore
1781 eval { $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] }) };
1782 warn "W: $@\n" if $@;
1783 my $base = defined $fetch ? $head : 0;
1785 # read the max revs for wildcard expansion (branches/*, tags/*)
1787 foreach my $t (qw/branches tags/) {
1788 defined $remote->{$t} or next;
1789 push @globs, @{$remote->{$t}};
1790 my $max_rev = eval { tmp_config(qw/--int --get/,
1792 "svn-remote.$repo_id.${t}-maxRev") };
1793 if (defined $max_rev && ($max_rev < $base)) {
1794 $base = $max_rev;
1795 } elsif (!defined $max_rev) {
1796 $base = 0;
1797 }
1798 }
1799 if ($fetch) {
1801 foreach my $p (sort keys %$fetch) {
1802 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1803 my $lr = $gs->rev_map_max;
1804 if (defined $lr) {
1805 $base = $lr if ($lr < $base);
1806 }
1807 push @gs, $gs;
1808 }
1809 }
1810 ($base, $head) = parse_revision_argument($base, $head);
1812 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1813}
1814sub read_all_remotes {
1816 my $r = {};
1817 my $use_svm_props = eval { command_oneline(qw/config --bool
1818 svn.useSvmProps/) };
1819 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1820 my $svn_refspec = qr{\s*(.*?)\s*:\s*(.+?)\s*};
1821 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1822 if (m!^(.+)\.fetch=$svn_refspec$!) {
1823 my ($remote, $local_ref, $remote_ref) = ($1, $2, $3);
1824 die("svn-remote.$remote: remote ref '$remote_ref' "
1825 . "must start with 'refs/'\n")
1826 unless $remote_ref =~ m{^refs/};
1827 $local_ref = uri_decode($local_ref);
1828 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1829 $r->{$remote}->{svm} = {} if $use_svm_props;
1830 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1831 $r->{$1}->{svm} = {};
1832 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1833 $r->{$1}->{url} = $2;
1834 } elsif (m!^(.+)\.pushurl=\s*(.*)\s*$!) {
1835 $r->{$1}->{pushurl} = $2;
1836 } elsif (m!^(.+)\.(branches|tags)=$svn_refspec$!) {
1837 my ($remote, $t, $local_ref, $remote_ref) =
1838 ($1, $2, $3, $4);
1839 die("svn-remote.$remote: remote ref '$remote_ref' ($t) "
1840 . "must start with 'refs/'\n")
1841 unless $remote_ref =~ m{^refs/};
1842 $local_ref = uri_decode($local_ref);
1843 my $rs = {
1844 t => $t,
1845 remote => $remote,
1846 path => Git::SVN::GlobSpec->new($local_ref, 1),
1847 ref => Git::SVN::GlobSpec->new($remote_ref, 0) };
1848 if (length($rs->{ref}->{right}) != 0) {
1849 die "The '*' glob character must be the last ",
1850 "character of '$remote_ref'\n";
1851 }
1852 push @{ $r->{$remote}->{$t} }, $rs;
1853 }
1854 }
1855 map {
1857 if (defined $r->{$_}->{svm}) {
1858 my $svm;
1859 eval {
1860 my $section = "svn-remote.$_";
1861 $svm = {
1862 source => tmp_config('--get',
1863 "$section.svm-source"),
1864 replace => tmp_config('--get',
1865 "$section.svm-replace"),
1866 }
1867 };
1868 $r->{$_}->{svm} = $svm;
1869 }
1870 } keys %$r;
1871 $r;
1873}
1874sub init_vars {
1876 $_gc_nr = $_gc_period = 1000;
1877 if (defined $_repack || defined $_repack_flags) {
1878 warn "Repack options are obsolete; they have no effect.\n";
1879 }
1880}
1881sub verify_remotes_sanity {
1883 return unless -d $ENV{GIT_DIR};
1884 my %seen;
1885 foreach (command(qw/config -l/)) {
1886 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1887 if ($seen{$1}) {
1888 die "Remote ref refs/remote/$1 is tracked by",
1889 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1890 "Please resolve this ambiguity in ",
1891 "your git configuration file before ",
1892 "continuing\n";
1893 }
1894 $seen{$1} = $_;
1895 }
1896 }
1897}
1898sub find_existing_remote {
1900 my ($url, $remotes) = @_;
1901 return undef if $no_reuse_existing;
1902 my $existing;
1903 foreach my $repo_id (keys %$remotes) {
1904 my $u = $remotes->{$repo_id}->{url} or next;
1905 next if $u ne $url;
1906 $existing = $repo_id;
1907 last;
1908 }
1909 $existing;
1910}
1911sub init_remote_config {
1913 my ($self, $url, $no_write) = @_;
1914 $url =~ s!/+$!!; # strip trailing slash
1915 my $r = read_all_remotes();
1916 my $existing = find_existing_remote($url, $r);
1917 if ($existing) {
1918 unless ($no_write) {
1919 print STDERR "Using existing ",
1920 "[svn-remote \"$existing\"]\n";
1921 }
1922 $self->{repo_id} = $existing;
1923 } elsif ($_minimize_url) {
1924 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1925 $existing = find_existing_remote($min_url, $r);
1926 if ($existing) {
1927 unless ($no_write) {
1928 print STDERR "Using existing ",
1929 "[svn-remote \"$existing\"]\n";
1930 }
1931 $self->{repo_id} = $existing;
1932 }
1933 if ($min_url ne $url) {
1934 unless ($no_write) {
1935 print STDERR "Using higher level of URL: ",
1936 "$url => $min_url\n";
1937 }
1938 my $old_path = $self->{path};
1939 $self->{path} = $url;
1940 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1941 if (length $old_path) {
1942 $self->{path} .= "/$old_path";
1943 }
1944 $url = $min_url;
1945 }
1946 }
1947 my $orig_url;
1948 if (!$existing) {
1949 # verify that we aren't overwriting anything:
1950 $orig_url = eval {
1951 command_oneline('config', '--get',
1952 "svn-remote.$self->{repo_id}.url")
1953 };
1954 if ($orig_url && ($orig_url ne $url)) {
1955 die "svn-remote.$self->{repo_id}.url already set: ",
1956 "$orig_url\nwanted to set to: $url\n";
1957 }
1958 }
1959 my ($xrepo_id, $xpath) = find_ref($self->refname);
1960 if (!$no_write && defined $xpath) {
1961 die "svn-remote.$xrepo_id.fetch already set to track ",
1962 "$xpath:", $self->refname, "\n";
1963 }
1964 unless ($no_write) {
1965 command_noisy('config',
1966 "svn-remote.$self->{repo_id}.url", $url);
1967 $self->{path} =~ s{^/}{};
1968 $self->{path} =~ s{%([0-9A-F]{2})}{chr hex($1)}ieg;
1969 command_noisy('config', '--add',
1970 "svn-remote.$self->{repo_id}.fetch",
1971 "$self->{path}:".$self->refname);
1972 }
1973 $self->{url} = $url;
1974}
1975sub find_by_url { # repos_root and, path are optional
1977 my ($class, $full_url, $repos_root, $path) = @_;
1978 return undef unless defined $full_url;
1980 remove_username($full_url);
1981 remove_username($repos_root) if defined $repos_root;
1982 my $remotes = read_all_remotes();
1983 if (defined $full_url && defined $repos_root && !defined $path) {
1984 $path = $full_url;
1985 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1986 }
1987 foreach my $repo_id (keys %$remotes) {
1988 my $u = $remotes->{$repo_id}->{url} or next;
1989 remove_username($u);
1990 next if defined $repos_root && $repos_root ne $u;
1991 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1993 foreach my $t (qw/branches tags/) {
1994 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1995 resolve_local_globs($u, $fetch, $globspec);
1996 }
1997 }
1998 my $p = $path;
1999 my $rwr = rewrite_root({repo_id => $repo_id});
2000 my $svm = $remotes->{$repo_id}->{svm}
2001 if defined $remotes->{$repo_id}->{svm};
2002 unless (defined $p) {
2003 $p = $full_url;
2004 my $z = $u;
2005 my $prefix = '';
2006 if ($rwr) {
2007 $z = $rwr;
2008 remove_username($z);
2009 } elsif (defined $svm) {
2010 $z = $svm->{source};
2011 $prefix = $svm->{replace};
2012 $prefix =~ s#^\Q$u\E(?:/|$)##;
2013 $prefix =~ s#/$##;
2014 }
2015 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
2016 }
2017 foreach my $f (keys %$fetch) {
2018 next if $f ne $p;
2019 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
2020 }
2021 }
2022 undef;
2023}
2024sub init {
2026 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
2027 my $self = _new($class, $repo_id, $ref_id, $path);
2028 if (defined $url) {
2029 $self->init_remote_config($url, $no_write);
2030 }
2031 $self;
2032}
2033sub find_ref {
2035 my ($ref_id) = @_;
2036 foreach (command(qw/config -l/)) {
2037 next unless m!^svn-remote\.(.+)\.fetch=
2038 \s*(.*?)\s*:\s*(.+?)\s*$!x;
2039 my ($repo_id, $path, $ref) = ($1, $2, $3);
2040 if ($ref eq $ref_id) {
2041 $path = '' if ($path =~ m#^\./?#);
2042 return ($repo_id, $path);
2043 }
2044 }
2045 (undef, undef, undef);
2046}
2047sub new {
2049 my ($class, $ref_id, $repo_id, $path) = @_;
2050 if (defined $ref_id && !defined $repo_id && !defined $path) {
2051 ($repo_id, $path) = find_ref($ref_id);
2052 if (!defined $repo_id) {
2053 die "Could not find a \"svn-remote.*.fetch\" key ",
2054 "in the repository configuration matching: ",
2055 "$ref_id\n";
2056 }
2057 }
2058 my $self = _new($class, $repo_id, $ref_id, $path);
2059 if (!defined $self->{path} || !length $self->{path}) {
2060 my $fetch = command_oneline('config', '--get',
2061 "svn-remote.$repo_id.fetch",
2062 ":$ref_id\$") or
2063 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
2064 "\":$ref_id\$\" in config\n";
2065 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
2066 }
2067 $self->{path} =~ s{/+}{/}g;
2068 $self->{path} =~ s{\A/}{};
2069 $self->{path} =~ s{/\z}{};
2070 $self->{url} = command_oneline('config', '--get',
2071 "svn-remote.$repo_id.url") or
2072 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
2073 $self->{pushurl} = eval { command_oneline('config', '--get',
2074 "svn-remote.$repo_id.pushurl") };
2075 $self->rebuild;
2076 $self;
2077}
2078sub refname {
2080 my ($refname) = $_[0]->{ref_id} ;
2081 # It cannot end with a slash /, we'll throw up on this because
2083 # SVN can't have directories with a slash in their name, either:
2084 if ($refname =~ m{/$}) {
2085 die "ref: '$refname' ends with a trailing slash, this is ",
2086 "not permitted by git nor Subversion\n";
2087 }
2088 # It cannot have ASCII control character space, tilde ~, caret ^,
2090 # colon :, question-mark ?, asterisk *, space, or open bracket [
2091 # anywhere.
2092 #
2093 # Additionally, % must be escaped because it is used for escaping
2094 # and we want our escaped refname to be reversible
2095 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
2096 # no slash-separated component can begin with a dot .
2098 # /.* becomes /%2E*
2099 $refname =~ s{/\.}{/%2E}g;
2100 # It cannot have two consecutive dots .. anywhere
2102 # .. becomes %2E%2E
2103 $refname =~ s{\.\.}{%2E%2E}g;
2104 # trailing dots and .lock are not allowed
2106 # .$ becomes %2E and .lock becomes %2Elock
2107 $refname =~ s{\.(?=$|lock$)}{%2E};
2108 # the sequence @{ is used to access the reflog
2110 # @{ becomes %40{
2111 $refname =~ s{\@\{}{%40\{}g;
2112 return $refname;
2114}
2115sub desanitize_refname {
2117 my ($refname) = @_;
2118 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2119 return $refname;
2120}
2121sub svm_uuid {
2123 my ($self) = @_;
2124 return $self->{svm}->{uuid} if $self->svm;
2125 $self->ra;
2126 unless ($self->{svm}) {
2127 die "SVM UUID not cached, and reading remotely failed\n";
2128 }
2129 $self->{svm}->{uuid};
2130}
2131sub svm {
2133 my ($self) = @_;
2134 return $self->{svm} if $self->{svm};
2135 my $svm;
2136 # see if we have it in our config, first:
2137 eval {
2138 my $section = "svn-remote.$self->{repo_id}";
2139 $svm = {
2140 source => tmp_config('--get', "$section.svm-source"),
2141 uuid => tmp_config('--get', "$section.svm-uuid"),
2142 replace => tmp_config('--get', "$section.svm-replace"),
2143 }
2144 };
2145 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2146 $self->{svm} = $svm;
2147 }
2148 $self->{svm};
2149}
2150sub _set_svm_vars {
2152 my ($self, $ra) = @_;
2153 return $ra if $self->svm;
2154 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2156 "(svm:source, svm:uuid) ",
2157 "from the following URLs:\n" );
2158 sub read_svm_props {
2159 my ($self, $ra, $path, $r) = @_;
2160 my $props = ($ra->get_dir($path, $r))[2];
2161 my $src = $props->{'svm:source'};
2162 my $uuid = $props->{'svm:uuid'};
2163 return undef if (!$src || !$uuid);
2164 chomp($src, $uuid);
2166 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2168 or die "doesn't look right - svm:uuid is '$uuid'\n";
2169 # the '!' is used to mark the repos_root!/relative/path
2171 $src =~ s{/?!/?}{/};
2172 $src =~ s{/+$}{}; # no trailing slashes please
2173 # username is of no interest
2174 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2175 my $replace = $ra->{url};
2177 $replace .= "/$path" if length $path;
2178 my $section = "svn-remote.$self->{repo_id}";
2180 tmp_config("$section.svm-source", $src);
2181 tmp_config("$section.svm-replace", $replace);
2182 tmp_config("$section.svm-uuid", $uuid);
2183 $self->{svm} = {
2184 source => $src,
2185 uuid => $uuid,
2186 replace => $replace
2187 };
2188 }
2189 my $r = $ra->get_latest_revnum;
2191 my $path = $self->{path};
2192 my %tried;
2193 while (length $path) {
2194 unless ($tried{"$self->{url}/$path"}) {
2195 return $ra if $self->read_svm_props($ra, $path, $r);
2196 $tried{"$self->{url}/$path"} = 1;
2197 }
2198 $path =~ s#/?[^/]+$##;
2199 }
2200 die "Path: '$path' should be ''\n" if $path ne '';
2201 return $ra if $self->read_svm_props($ra, $path, $r);
2202 $tried{"$self->{url}/$path"} = 1;
2203 if ($ra->{repos_root} eq $self->{url}) {
2205 die @err, (map { " $_\n" } keys %tried), "\n";
2206 }
2207 # nope, make sure we're connected to the repository root:
2209 my $ok;
2210 my @tried_b;
2211 $path = $ra->{svn_path};
2212 $ra = Git::SVN::Ra->new($ra->{repos_root});
2213 while (length $path) {
2214 unless ($tried{"$ra->{url}/$path"}) {
2215 $ok = $self->read_svm_props($ra, $path, $r);
2216 last if $ok;
2217 $tried{"$ra->{url}/$path"} = 1;
2218 }
2219 $path =~ s#/?[^/]+$##;
2220 }
2221 die "Path: '$path' should be ''\n" if $path ne '';
2222 $ok ||= $self->read_svm_props($ra, $path, $r);
2223 $tried{"$ra->{url}/$path"} = 1;
2224 if (!$ok) {
2225 die @err, (map { " $_\n" } keys %tried), "\n";
2226 }
2227 Git::SVN::Ra->new($self->{url});
2228}
2229sub svnsync {
2231 my ($self) = @_;
2232 return $self->{svnsync} if $self->{svnsync};
2233 if ($self->no_metadata) {
2235 die "Can't have both 'noMetadata' and ",
2236 "'useSvnsyncProps' options set!\n";
2237 }
2238 if ($self->rewrite_root) {
2239 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2240 "options set!\n";
2241 }
2242 if ($self->rewrite_uuid) {
2243 die "Can't have both 'useSvnsyncProps' and 'rewriteUUID' ",
2244 "options set!\n";
2245 }
2246 my $svnsync;
2248 # see if we have it in our config, first:
2249 eval {
2250 my $section = "svn-remote.$self->{repo_id}";
2251 my $url = tmp_config('--get', "$section.svnsync-url");
2253 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2254 die "doesn't look right - svn:sync-from-url is '$url'\n";
2255 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2257 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2258 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2259 $svnsync = { url => $url, uuid => $uuid }
2261 };
2262 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2263 return $self->{svnsync} = $svnsync;
2264 }
2265 my $err = "useSvnsyncProps set, but failed to read " .
2267 "svnsync property: svn:sync-from-";
2268 my $rp = $self->ra->rev_proplist(0);
2269 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2271 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2272 die "doesn't look right - svn:sync-from-url is '$url'\n";
2273 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2275 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2276 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2277 my $section = "svn-remote.$self->{repo_id}";
2279 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2280 tmp_config('--add', "$section.svnsync-url", $url);
2281 return $self->{svnsync} = { url => $url, uuid => $uuid };
2282}
2283# this allows us to memoize our SVN::Ra UUID locally and avoid a
2285# remote lookup (useful for 'git svn log').
2286sub ra_uuid {
2287 my ($self) = @_;
2288 unless ($self->{ra_uuid}) {
2289 my $key = "svn-remote.$self->{repo_id}.uuid";
2290 my $uuid = eval { tmp_config('--get', $key) };
2291 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2292 $self->{ra_uuid} = $uuid;
2293 } else {
2294 die "ra_uuid called without URL\n" unless $self->{url};
2295 $self->{ra_uuid} = $self->ra->get_uuid;
2296 tmp_config('--add', $key, $self->{ra_uuid});
2297 }
2298 }
2299 $self->{ra_uuid};
2300}
2301sub _set_repos_root {
2303 my ($self, $repos_root) = @_;
2304 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2305 $repos_root ||= $self->ra->{repos_root};
2306 tmp_config($k, $repos_root);
2307 $repos_root;
2308}
2309sub repos_root {
2311 my ($self) = @_;
2312 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2313 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2314}
2315sub ra {
2317 my ($self) = shift;
2318 my $ra = Git::SVN::Ra->new($self->{url});
2319 $self->_set_repos_root($ra->{repos_root});
2320 if ($self->use_svm_props && !$self->{svm}) {
2321 if ($self->no_metadata) {
2322 die "Can't have both 'noMetadata' and ",
2323 "'useSvmProps' options set!\n";
2324 } elsif ($self->use_svnsync_props) {
2325 die "Can't have both 'useSvnsyncProps' and ",
2326 "'useSvmProps' options set!\n";
2327 }
2328 $ra = $self->_set_svm_vars($ra);
2329 $self->{-want_revprops} = 1;
2330 }
2331 $ra;
2332}
2333# prop_walk(PATH, REV, SUB)
2335# -------------------------
2336# Recursively traverse PATH at revision REV and invoke SUB for each
2337# directory that contains a SVN property. SUB will be invoked as
2338# follows: &SUB(gs, path, props); where `gs' is this instance of
2339# Git::SVN, `path' the path to the directory where the properties
2340# `props' were found. The `path' will be relative to point of checkout,
2341# that is, if url://repo/trunk is the current Git branch, and that
2342# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2343# as `path' (note the trailing `/').
2344sub prop_walk {
2345 my ($self, $path, $rev, $sub) = @_;
2346 $path =~ s#^/##;
2348 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2349 $path =~ s#^/*#/#g;
2350 my $p = $path;
2351 # Strip the irrelevant part of the path.
2352 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2353 # Ensure the path is terminated by a `/'.
2354 $p =~ s#/*$#/#;
2355 # The properties contain all the internal SVN stuff nobody
2357 # (usually) cares about.
2358 my $interesting_props = 0;
2359 foreach (keys %{$props}) {
2360 # If it doesn't start with `svn:', it must be a
2361 # user-defined property.
2362 ++$interesting_props and next if $_ !~ /^svn:/;
2363 # FIXME: Fragile, if SVN adds new public properties,
2364 # this needs to be updated.
2365 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2366 |eol-style|mime-type
2367 |externals|needs-lock)$/x;
2368 }
2369 &$sub($self, $p, $props) if $interesting_props;
2370 foreach (sort keys %$dirent) {
2372 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2373 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2374 }
2375}
2376sub last_rev { ($_[0]->last_rev_commit)[0] }
2378sub last_commit { ($_[0]->last_rev_commit)[1] }
2379# returns the newest SVN revision number and newest commit SHA1
2381sub last_rev_commit {
2382 my ($self) = @_;
2383 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2384 return ($self->{last_rev}, $self->{last_commit});
2385 }
2386 my $c = ::verify_ref($self->refname.'^0');
2387 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2388 my $rev = (::cmt_metadata($c))[1];
2389 if (defined $rev) {
2390 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2391 return ($rev, $c);
2392 }
2393 }
2394 my $map_path = $self->map_path;
2395 unless (-e $map_path) {
2396 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2397 return (undef, undef);
2398 }
2399 my ($rev, $commit) = $self->rev_map_max(1);
2400 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2401 return ($rev, $commit);
2402}
2403sub get_fetch_range {
2405 my ($self, $min, $max) = @_;
2406 $max ||= $self->ra->get_latest_revnum;
2407 $min ||= $self->rev_map_max;
2408 (++$min, $max);
2409}
2410sub tmp_config {
2412 my (@args) = @_;
2413 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2414 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2415 if (! -f $config && -f $old_def_config) {
2416 rename $old_def_config, $config or
2417 die "Failed rename $old_def_config => $config: $!\n";
2418 }
2419 my $old_config = $ENV{GIT_CONFIG};
2420 $ENV{GIT_CONFIG} = $config;
2421 $@ = undef;
2422 my @ret = eval {
2423 unless (-f $config) {
2424 mkfile($config);
2425 open my $fh, '>', $config or
2426 die "Can't open $config: $!\n";
2427 print $fh "; This file is used internally by ",
2428 "git-svn\n" or die
2429 "Couldn't write to $config: $!\n";
2430 print $fh "; You should not have to edit it\n" or
2431 die "Couldn't write to $config: $!\n";
2432 close $fh or die "Couldn't close $config: $!\n";
2433 }
2434 command('config', @args);
2435 };
2436 my $err = $@;
2437 if (defined $old_config) {
2438 $ENV{GIT_CONFIG} = $old_config;
2439 } else {
2440 delete $ENV{GIT_CONFIG};
2441 }
2442 die $err if $err;
2443 wantarray ? @ret : $ret[0];
2444}
2445sub tmp_index_do {
2447 my ($self, $sub) = @_;
2448 my $old_index = $ENV{GIT_INDEX_FILE};
2449 $ENV{GIT_INDEX_FILE} = $self->{index};
2450 $@ = undef;
2451 my @ret = eval {
2452 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2453 mkpath([$dir]) unless -d $dir;
2454 &$sub;
2455 };
2456 my $err = $@;
2457 if (defined $old_index) {
2458 $ENV{GIT_INDEX_FILE} = $old_index;
2459 } else {
2460 delete $ENV{GIT_INDEX_FILE};
2461 }
2462 die $err if $err;
2463 wantarray ? @ret : $ret[0];
2464}
2465sub assert_index_clean {
2467 my ($self, $treeish) = @_;
2468 $self->tmp_index_do(sub {
2470 command_noisy('read-tree', $treeish) unless -e $self->{index};
2471 my $x = command_oneline('write-tree');
2472 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2473 /^tree ($::sha1)/mo);
2474 return if $y eq $x;
2475 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2477 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2478 command_noisy('read-tree', $treeish);
2479 $x = command_oneline('write-tree');
2480 if ($y ne $x) {
2481 ::fatal "trees ($treeish) $y != $x\n",
2482 "Something is seriously wrong...";
2483 }
2484 });
2485}
2486sub get_commit_parents {
2488 my ($self, $log_entry) = @_;
2489 my (%seen, @ret, @tmp);
2490 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2491 if (my $ip = $self->{inject_parents}) {
2492 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2493 push @tmp, $commit;
2494 }
2495 }
2496 if (my $cur = ::verify_ref($self->refname.'^0')) {
2497 push @tmp, $cur;
2498 }
2499 if (my $ipd = $self->{inject_parents_dcommit}) {
2500 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2501 push @tmp, @$commit;
2502 }
2503 }
2504 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2505 while (my $p = shift @tmp) {
2506 next if $seen{$p};
2507 $seen{$p} = 1;
2508 push @ret, $p;
2509 }
2510 @ret;
2511}
2512sub rewrite_root {
2514 my ($self) = @_;
2515 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2516 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2517 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2518 if ($rwr) {
2519 $rwr =~ s#/+$##;
2520 if ($rwr !~ m#^[a-z\+]+://#) {
2521 die "$rwr is not a valid URL (key: $k)\n";
2522 }
2523 }
2524 $self->{-rewrite_root} = $rwr;
2525}
2526sub rewrite_uuid {
2528 my ($self) = @_;
2529 return $self->{-rewrite_uuid} if exists $self->{-rewrite_uuid};
2530 my $k = "svn-remote.$self->{repo_id}.rewriteUUID";
2531 my $rwid = eval { command_oneline(qw/config --get/, $k) };
2532 if ($rwid) {
2533 $rwid =~ s#/+$##;
2534 if ($rwid !~ m#^[a-f0-9]{8}-(?:[a-f0-9]{4}-){3}[a-f0-9]{12}$#) {
2535 die "$rwid is not a valid UUID (key: $k)\n";
2536 }
2537 }
2538 $self->{-rewrite_uuid} = $rwid;
2539}
2540sub metadata_url {
2542 my ($self) = @_;
2543 ($self->rewrite_root || $self->{url}) .
2544 (length $self->{path} ? '/' . $self->{path} : '');
2545}
2546sub full_url {
2548 my ($self) = @_;
2549 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2550}
2551sub full_pushurl {
2553 my ($self) = @_;
2554 if ($self->{pushurl}) {
2555 return $self->{pushurl} . (length $self->{path} ? '/' .
2556 $self->{path} : '');
2557 } else {
2558 return $self->full_url;
2559 }
2560}
2561sub set_commit_header_env {
2563 my ($log_entry) = @_;
2564 my %env;
2565 foreach my $ned (qw/NAME EMAIL DATE/) {
2566 foreach my $ac (qw/AUTHOR COMMITTER/) {
2567 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2568 }
2569 }
2570 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2572 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2573 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2574 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2576 ? $log_entry->{commit_name}
2577 : $log_entry->{name};
2578 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2579 ? $log_entry->{commit_email}
2580 : $log_entry->{email};
2581 \%env;
2582}
2583sub restore_commit_header_env {
2585 my ($env) = @_;
2586 foreach my $ned (qw/NAME EMAIL DATE/) {
2587 foreach my $ac (qw/AUTHOR COMMITTER/) {
2588 my $k = "GIT_${ac}_${ned}";
2589 if (defined $env->{$k}) {
2590 $ENV{$k} = $env->{$k};
2591 } else {
2592 delete $ENV{$k};
2593 }
2594 }
2595 }
2596}
2597sub gc {
2599 command_noisy('gc', '--auto');
2600};
2601sub do_git_commit {
2603 my ($self, $log_entry) = @_;
2604 my $lr = $self->last_rev;
2605 if (defined $lr && $lr >= $log_entry->{revision}) {
2606 die "Last fetched revision of ", $self->refname,
2607 " was r$lr, but we are about to fetch: ",
2608 "r$log_entry->{revision}!\n";
2609 }
2610 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2611 croak "$log_entry->{revision} = $c already exists! ",
2612 "Why are we refetching it?\n";
2613 }
2614 my $old_env = set_commit_header_env($log_entry);
2615 my $tree = $log_entry->{tree};
2616 if (!defined $tree) {
2617 $tree = $self->tmp_index_do(sub {
2618 command_oneline('write-tree') });
2619 }
2620 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2621 my @exec = ('git', 'commit-tree', $tree);
2623 foreach ($self->get_commit_parents($log_entry)) {
2624 push @exec, '-p', $_;
2625 }
2626 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2627 or croak $!;
2628 binmode $msg_fh;
2629 # we always get UTF-8 from SVN, but we may want our commits in
2631 # a different encoding.
2632 if (my $enc = Git::config('i18n.commitencoding')) {
2633 require Encode;
2634 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2635 }
2636 print $msg_fh $log_entry->{log} or croak $!;
2637 restore_commit_header_env($old_env);
2638 unless ($self->no_metadata) {
2639 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2640 or croak $!;
2641 }
2642 $msg_fh->flush == 0 or croak $!;
2643 close $msg_fh or croak $!;
2644 chomp(my $commit = do { local $/; <$out_fh> });
2645 close $out_fh or croak $!;
2646 waitpid $pid, 0;
2647 croak $? if $?;
2648 if ($commit !~ /^$::sha1$/o) {
2649 die "Failed to commit, invalid sha1: $commit\n";
2650 }
2651 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2653 $self->{last_rev} = $log_entry->{revision};
2655 $self->{last_commit} = $commit;
2656 print "r$log_entry->{revision}" unless $::_q > 1;
2657 if (defined $log_entry->{svm_revision}) {
2658 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2659 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2660 0, $self->svm_uuid);
2661 }
2662 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2663 if (--$_gc_nr == 0) {
2664 $_gc_nr = $_gc_period;
2665 gc();
2666 }
2667 return $commit;
2668}
2669sub match_paths {
2671 my ($self, $paths, $r) = @_;
2672 return 1 if $self->{path} eq '';
2673 if (my $path = $paths->{"/$self->{path}"}) {
2674 return ($path->{action} eq 'D') ? 0 : 1;
2675 }
2676 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2677 if (grep /$self->{path_regex}/, keys %$paths) {
2678 return 1;
2679 }
2680 my $c = '';
2681 foreach (split m#/#, $self->{path}) {
2682 $c .= "/$_";
2683 next unless ($paths->{$c} &&
2684 ($paths->{$c}->{action} =~ /^[AR]$/));
2685 if ($self->ra->check_path($self->{path}, $r) ==
2686 $SVN::Node::dir) {
2687 return 1;
2688 }
2689 }
2690 return 0;
2691}
2692sub find_parent_branch {
2694 my ($self, $paths, $rev) = @_;
2695 return undef unless $self->follow_parent;
2696 unless (defined $paths) {
2697 my $err_handler = $SVN::Error::handler;
2698 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2699 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2700 sub { $paths = $_[0] });
2701 $SVN::Error::handler = $err_handler;
2702 }
2703 return undef unless defined $paths;
2704 # look for a parent from another branch:
2706 my @b_path_components = split m#/#, $self->{path};
2707 my @a_path_components;
2708 my $i;
2709 while (@b_path_components) {
2710 $i = $paths->{'/'.join('/', @b_path_components)};
2711 last if $i && defined $i->{copyfrom_path};
2712 unshift(@a_path_components, pop(@b_path_components));
2713 }
2714 return undef unless defined $i && defined $i->{copyfrom_path};
2715 my $branch_from = $i->{copyfrom_path};
2716 if (@a_path_components) {
2717 print STDERR "branch_from: $branch_from => ";
2718 $branch_from .= '/'.join('/', @a_path_components);
2719 print STDERR $branch_from, "\n";
2720 }
2721 my $r = $i->{copyfrom_rev};
2722 my $repos_root = $self->ra->{repos_root};
2723 my $url = $self->ra->{url};
2724 my $new_url = $url . $branch_from;
2725 print STDERR "Found possible branch point: ",
2726 "$new_url => ", $self->full_url, ", $r\n"
2727 unless $::_q > 1;
2728 $branch_from =~ s#^/##;
2729 my $gs = $self->other_gs($new_url, $url,
2730 $branch_from, $r, $self->{ref_id});
2731 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2732 {
2733 my ($base, $head);
2734 if (!defined $r0 || !defined $parent) {
2735 ($base, $head) = parse_revision_argument(0, $r);
2736 } else {
2737 if ($r0 < $r) {
2738 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2739 0, 1, sub { $base = $_[1] - 1 });
2740 }
2741 }
2742 if (defined $base && $base <= $r) {
2743 $gs->fetch($base, $r);
2744 }
2745 ($r0, $parent) = $gs->find_rev_before($r, 1);
2746 }
2747 if (defined $r0 && defined $parent) {
2748 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n"
2749 unless $::_q > 1;
2750 my $ed;
2751 if ($self->ra->can_do_switch) {
2752 $self->assert_index_clean($parent);
2753 print STDERR "Following parent with do_switch\n"
2754 unless $::_q > 1;
2755 # do_switch works with svn/trunk >= r22312, but that
2756 # is not included with SVN 1.4.3 (the latest version
2757 # at the moment), so we can't rely on it
2758 $self->{last_rev} = $r0;
2759 $self->{last_commit} = $parent;
2760 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2761 $gs->ra->gs_do_switch($r0, $rev, $gs,
2762 $self->full_url, $ed)
2763 or die "SVN connection failed somewhere...\n";
2764 } elsif ($self->ra->trees_match($new_url, $r0,
2765 $self->full_url, $rev)) {
2766 print STDERR "Trees match:\n",
2767 " $new_url\@$r0\n",
2768 " ${\$self->full_url}\@$rev\n",
2769 "Following parent with no changes\n"
2770 unless $::_q > 1;
2771 $self->tmp_index_do(sub {
2772 command_noisy('read-tree', $parent);
2773 });
2774 $self->{last_commit} = $parent;
2775 } else {
2776 print STDERR "Following parent with do_update\n"
2777 unless $::_q > 1;
2778 $ed = SVN::Git::Fetcher->new($self);
2779 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2780 or die "SVN connection failed somewhere...\n";
2781 }
2782 print STDERR "Successfully followed parent\n" unless $::_q > 1;
2783 return $self->make_log_entry($rev, [$parent], $ed);
2784 }
2785 return undef;
2786}
2787sub do_fetch {
2789 my ($self, $paths, $rev) = @_;
2790 my $ed;
2791 my ($last_rev, @parents);
2792 if (my $lc = $self->last_commit) {
2793 # we can have a branch that was deleted, then re-added
2794 # under the same name but copied from another path, in
2795 # which case we'll have multiple parents (we don't
2796 # want to break the original ref, nor lose copypath info):
2797 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2798 push @{$log_entry->{parents}}, $lc;
2799 return $log_entry;
2800 }
2801 $ed = SVN::Git::Fetcher->new($self);
2802 $last_rev = $self->{last_rev};
2803 $ed->{c} = $lc;
2804 @parents = ($lc);
2805 } else {
2806 $last_rev = $rev;
2807 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2808 return $log_entry;
2809 }
2810 $ed = SVN::Git::Fetcher->new($self);
2811 }
2812 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2813 die "SVN connection failed somewhere...\n";
2814 }
2815 $self->make_log_entry($rev, \@parents, $ed);
2816}
2817sub mkemptydirs {
2819 my ($self, $r) = @_;
2820 sub scan {
2822 my ($r, $empty_dirs, $line) = @_;
2823 if (defined $r && $line =~ /^r(\d+)$/) {
2824 return 0 if $1 > $r;
2825 } elsif ($line =~ /^ \+empty_dir: (.+)$/) {
2826 $empty_dirs->{$1} = 1;
2827 } elsif ($line =~ /^ \-empty_dir: (.+)$/) {
2828 my @d = grep {m[^\Q$1\E(/|$)]} (keys %$empty_dirs);
2829 delete @$empty_dirs{@d};
2830 }
2831 1; # continue
2832 };
2833 my %empty_dirs = ();
2835 my $gz_file = "$self->{dir}/unhandled.log.gz";
2836 if (-f $gz_file) {
2837 if (!$can_compress) {
2838 warn "Compress::Zlib could not be found; ",
2839 "empty directories in $gz_file will not be read\n";
2840 } else {
2841 my $gz = Compress::Zlib::gzopen($gz_file, "rb") or
2842 die "Unable to open $gz_file: $!\n";
2843 my $line;
2844 while ($gz->gzreadline($line) > 0) {
2845 scan($r, \%empty_dirs, $line) or last;
2846 }
2847 $gz->gzclose;
2848 }
2849 }
2850 if (open my $fh, '<', "$self->{dir}/unhandled.log") {
2852 binmode $fh or croak "binmode: $!";
2853 while (<$fh>) {
2854 scan($r, \%empty_dirs, $_) or last;
2855 }
2856 close $fh;
2857 }
2858 my $strip = qr/\A\Q$self->{path}\E(?:\/|$)/;
2860 foreach my $d (sort keys %empty_dirs) {
2861 $d = uri_decode($d);
2862 $d =~ s/$strip//;
2863 next unless length($d);
2864 next if -d $d;
2865 if (-e $d) {
2866 warn "$d exists but is not a directory\n";
2867 } else {
2868 print "creating empty directory: $d\n";
2869 mkpath([$d]);
2870 }
2871 }
2872}
2873sub get_untracked {
2875 my ($self, $ed) = @_;
2876 my @out;
2877 my $h = $ed->{empty};
2878 foreach (sort keys %$h) {
2879 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2880 push @out, " $act: " . uri_encode($_);
2881 warn "W: $act: $_\n";
2882 }
2883 foreach my $t (qw/dir_prop file_prop/) {
2884 $h = $ed->{$t} or next;
2885 foreach my $path (sort keys %$h) {
2886 my $ppath = $path eq '' ? '.' : $path;
2887 foreach my $prop (sort keys %{$h->{$path}}) {
2888 next if $SKIP_PROP{$prop};
2889 my $v = $h->{$path}->{$prop};
2890 my $t_ppath_prop = "$t: " .
2891 uri_encode($ppath) . ' ' .
2892 uri_encode($prop);
2893 if (defined $v) {
2894 push @out, " +$t_ppath_prop " .
2895 uri_encode($v);
2896 } else {
2897 push @out, " -$t_ppath_prop";
2898 }
2899 }
2900 }
2901 }
2902 foreach my $t (qw/absent_file absent_directory/) {
2903 $h = $ed->{$t} or next;
2904 foreach my $parent (sort keys %$h) {
2905 foreach my $path (sort @{$h->{$parent}}) {
2906 push @out, " $t: " .
2907 uri_encode("$parent/$path");
2908 warn "W: $t: $parent/$path ",
2909 "Insufficient permissions?\n";
2910 }
2911 }
2912 }
2913 \@out;
2914}
2915# parse_svn_date(DATE)
2917# --------------------
2918# Given a date (in UTC) from Subversion, return a string in the format
2919# "<TZ Offset> <local date/time>" that Git will use.
2920#
2921# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2922# is true we'll convert it to the local timezone instead.
2923sub parse_svn_date {
2924 my $date = shift || return '+0000 1970-01-01 00:00:00';
2925 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2926 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2927 croak "Unable to parse date: $date\n";
2928 my $parsed_date; # Set next.
2929 if ($Git::SVN::_localtime) {
2931 # Translate the Subversion datetime to an epoch time.
2932 # Begin by switching ourselves to $date's timezone, UTC.
2933 my $old_env_TZ = $ENV{TZ};
2934 $ENV{TZ} = 'UTC';
2935 my $epoch_in_UTC =
2937 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2938 # Determine our local timezone (including DST) at the
2940 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2941 # value of TZ, if any, at the time we were run.
2942 if (defined $Git::SVN::Log::TZ) {
2943 $ENV{TZ} = $Git::SVN::Log::TZ;
2944 } else {
2945 delete $ENV{TZ};
2946 }
2947 my $our_TZ =
2949 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2950 # This converts $epoch_in_UTC into our local timezone.
2952 my ($sec, $min, $hour, $mday, $mon, $year,
2953 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2954 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2956 $our_TZ, $year + 1900, $mon + 1,
2957 $mday, $hour, $min, $sec);
2958 # Reset us to the timezone in effect when we entered
2960 # this routine.
2961 if (defined $old_env_TZ) {
2962 $ENV{TZ} = $old_env_TZ;
2963 } else {
2964 delete $ENV{TZ};
2965 }
2966 } else {
2967 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2968 }
2969 return $parsed_date;
2971}
2972sub other_gs {
2974 my ($self, $new_url, $url,
2975 $branch_from, $r, $old_ref_id) = @_;
2976 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2977 unless ($gs) {
2978 my $ref_id = $old_ref_id;
2979 $ref_id =~ s/\@\d+-*$//;
2980 $ref_id .= "\@$r";
2981 # just grow a tail if we're not unique enough :x
2982 $ref_id .= '-' while find_ref($ref_id);
2983 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2984 if ($u =~ s#^\Q$url\E(/|$)##) {
2985 $p = $u;
2986 $u = $url;
2987 $repo_id = $self->{repo_id};
2988 }
2989 while (1) {
2990 # It is possible to tag two different subdirectories at
2991 # the same revision. If the url for an existing ref
2992 # does not match, we must either find a ref with a
2993 # matching url or create a new ref by growing a tail.
2994 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2995 my (undef, $max_commit) = $gs->rev_map_max(1);
2996 last if (!$max_commit);
2997 my ($url) = ::cmt_metadata($max_commit);
2998 last if ($url eq $gs->full_url);
2999 $ref_id .= '-';
3000 }
3001 print STDERR "Initializing parent: $ref_id\n" unless $::_q > 1;
3002 }
3003 $gs
3004}
3005sub call_authors_prog {
3007 my ($orig_author) = @_;
3008 $orig_author = command_oneline('rev-parse', '--sq-quote', $orig_author);
3009 my $author = `$::_authors_prog $orig_author`;
3010 if ($? != 0) {
3011 die "$::_authors_prog failed with exit code $?\n"
3012 }
3013 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
3014 my ($name, $email) = ($1, $2);
3015 $email = undef if length $2 == 0;
3016 return [$name, $email];
3017 } else {
3018 die "Author: $orig_author: $::_authors_prog returned "
3019 . "invalid author format: $author\n";
3020 }
3021}
3022sub check_author {
3024 my ($author) = @_;
3025 if (!defined $author || length $author == 0) {
3026 $author = '(no author)';
3027 }
3028 if (!defined $::users{$author}) {
3029 if (defined $::_authors_prog) {
3030 $::users{$author} = call_authors_prog($author);
3031 } elsif (defined $::_authors) {
3032 die "Author: $author not defined in $::_authors file\n";
3033 }
3034 }
3035 $author;
3036}
3037sub find_extra_svk_parents {
3039 my ($self, $ed, $tickets, $parents) = @_;
3040 # aha! svk:merge property changed...
3041 my @tickets = split "\n", $tickets;
3042 my @known_parents;
3043 for my $ticket ( @tickets ) {
3044 my ($uuid, $path, $rev) = split /:/, $ticket;
3045 if ( $uuid eq $self->ra_uuid ) {
3046 my $url = $self->{url};
3047 my $repos_root = $url;
3048 my $branch_from = $path;
3049 $branch_from =~ s{^/}{};
3050 my $gs = $self->other_gs($repos_root."/".$branch_from,
3051 $url,
3052 $branch_from,
3053 $rev,
3054 $self->{ref_id});
3055 if ( my $commit = $gs->rev_map_get($rev, $uuid) ) {
3056 # wahey! we found it, but it might be
3057 # an old one (!)
3058 push @known_parents, [ $rev, $commit ];
3059 }
3060 }
3061 }
3062 # Ordering matters; highest-numbered commit merge tickets
3063 # first, as they may account for later merge ticket additions
3064 # or changes.
3065 @known_parents = map {$_->[1]} sort {$b->[0] <=> $a->[0]} @known_parents;
3066 for my $parent ( @known_parents ) {
3067 my @cmd = ('rev-list', $parent, map { "^$_" } @$parents );
3068 my ($msg_fh, $ctx) = command_output_pipe(@cmd);
3069 my $new;
3070 while ( <$msg_fh> ) {
3071 $new=1;last;
3072 }
3073 command_close_pipe($msg_fh, $ctx);
3074 if ( $new ) {
3075 print STDERR
3076 "Found merge parent (svk:merge ticket): $parent\n";
3077 push @$parents, $parent;
3078 }
3079 }
3080}
3081sub lookup_svn_merge {
3083 my $uuid = shift;
3084 my $url = shift;
3085 my $merge = shift;
3086 my ($source, $revs) = split ":", $merge;
3088 my $path = $source;
3089 $path =~ s{^/}{};
3090 my $gs = Git::SVN->find_by_url($url.$source, $url, $path);
3091 if ( !$gs ) {
3092 warn "Couldn't find revmap for $url$source\n";
3093 return;
3094 }
3095 my @ranges = split ",", $revs;
3096 my ($tip, $tip_commit);
3097 my @merged_commit_ranges;
3098 # find the tip
3099 for my $range ( @ranges ) {
3100 my ($bottom, $top) = split "-", $range;
3101 $top ||= $bottom;
3102 my $bottom_commit = $gs->find_rev_after( $bottom, 1, $top );
3103 my $top_commit = $gs->find_rev_before( $top, 1, $bottom );
3104 unless ($top_commit and $bottom_commit) {
3106 warn "W:unknown path/rev in svn:mergeinfo "
3107 ."dirprop: $source:$range\n";
3108 next;
3109 }
3110 push @merged_commit_ranges,
3112 "$bottom_commit^..$top_commit";
3113 if ( !defined $tip or $top > $tip ) {
3115 $tip = $top;
3116 $tip_commit = $top_commit;
3117 }
3118 }
3119 return ($tip_commit, @merged_commit_ranges);
3120}
3121sub _rev_list {
3123 my ($msg_fh, $ctx) = command_output_pipe(
3124 "rev-list", @_,
3125 );
3126 my @rv;
3127 while ( <$msg_fh> ) {
3128 chomp;
3129 push @rv, $_;
3130 }
3131 command_close_pipe($msg_fh, $ctx);
3132 @rv;
3133}
3134sub check_cherry_pick {
3136 my $base = shift;
3137 my $tip = shift;
3138 my $parents = shift;
3139 my @ranges = @_;
3140 my %commits = map { $_ => 1 }
3141 _rev_list("--no-merges", $tip, "--not", $base, @$parents);
3142 for my $range ( @ranges ) {
3143 delete @commits{_rev_list($range)};
3144 }
3145 for my $commit (keys %commits) {
3146 if (has_no_changes($commit)) {
3147 delete $commits{$commit};
3148 }
3149 }
3150 return (keys %commits);
3151}
3152sub has_no_changes {
3154 my $commit = shift;
3155 my @revs = split / /, command_oneline(
3157 qw(rev-list --parents -1 -m), $commit);
3158 # Commits with no parents, e.g. the start of a partial branch,
3160 # have changes by definition.
3161 return 1 if (@revs < 2);
3162 # Commits with multiple parents, e.g a merge, have no changes
3164 # by definition.
3165 return 0 if (@revs > 2);
3166 return (command_oneline("rev-parse", "$commit^{tree}") eq
3168 command_oneline("rev-parse", "$commit~1^{tree}"));
3169}
3170# The GIT_DIR environment variable is not always set until after the command
3172# line arguments are processed, so we can't memoize in a BEGIN block.
3173{
3174 my $memoized = 0;
3175 sub memoize_svn_mergeinfo_functions {
3177 return if $memoized;
3178 $memoized = 1;
3179 my $cache_path = "$ENV{GIT_DIR}/svn/.caches/";
3181 mkpath([$cache_path]) unless -d $cache_path;
3182 tie my %lookup_svn_merge_cache => 'Memoize::Storable',
3184 "$cache_path/lookup_svn_merge.db", 'nstore';
3185 memoize 'lookup_svn_merge',
3186 SCALAR_CACHE => 'FAULT',
3187 LIST_CACHE => ['HASH' => \%lookup_svn_merge_cache],
3188 ;
3189 tie my %check_cherry_pick_cache => 'Memoize::Storable',
3191 "$cache_path/check_cherry_pick.db", 'nstore';
3192 memoize 'check_cherry_pick',
3193 SCALAR_CACHE => 'FAULT',
3194 LIST_CACHE => ['HASH' => \%check_cherry_pick_cache],
3195 ;
3196 tie my %has_no_changes_cache => 'Memoize::Storable',
3198 "$cache_path/has_no_changes.db", 'nstore';
3199 memoize 'has_no_changes',
3200 SCALAR_CACHE => ['HASH' => \%has_no_changes_cache],
3201 LIST_CACHE => 'FAULT',
3202 ;
3203 }
3204 sub unmemoize_svn_mergeinfo_functions {
3206 return if not $memoized;
3207 $memoized = 0;
3208 Memoize::unmemoize 'lookup_svn_merge';
3210 Memoize::unmemoize 'check_cherry_pick';
3211 Memoize::unmemoize 'has_no_changes';
3212 }
3213}
3214END {
3216 # Force cache writeout explicitly instead of waiting for
3217 # global destruction to avoid segfault in Storable:
3218 # http://rt.cpan.org/Public/Bug/Display.html?id=36087
3219 unmemoize_svn_mergeinfo_functions();
3220}
3221sub parents_exclude {
3223 my $parents = shift;
3224 my @commits = @_;
3225 return unless @commits;
3226 my @excluded;
3228 my $excluded;
3229 do {
3230 my @cmd = ('rev-list', "-1", @commits, "--not", @$parents );
3231 $excluded = command_oneline(@cmd);
3232 if ( $excluded ) {
3233 my @new;
3234 my $found;
3235 for my $commit ( @commits ) {
3236 if ( $commit eq $excluded ) {
3237 push @excluded, $commit;
3238 $found++;
3239 last;
3240 }
3241 else {
3242 push @new, $commit;
3243 }
3244 }
3245 die "saw commit '$excluded' in rev-list output, "
3246 ."but we didn't ask for that commit (wanted: @commits --not @$parents)"
3247 unless $found;
3248 @commits = @new;
3249 }
3250 }
3251 while ($excluded and @commits);
3252 return @excluded;
3254}
3255# note: this function should only be called if the various dirprops
3258# have actually changed
3259sub find_extra_svn_parents {
3260 my ($self, $ed, $mergeinfo, $parents) = @_;
3261 # aha! svk:merge property changed...
3262 memoize_svn_mergeinfo_functions();
3264 # We first search for merged tips which are not in our
3266 # history. Then, we figure out which git revisions are in
3267 # that tip, but not this revision. If all of those revisions
3268 # are now marked as merge, we can add the tip as a parent.
3269 my @merges = split "\n", $mergeinfo;
3270 my @merge_tips;
3271 my $url = $self->{url};
3272 my $uuid = $self->ra_uuid;
3273 my %ranges;
3274 for my $merge ( @merges ) {
3275 my ($tip_commit, @ranges) =
3276 lookup_svn_merge( $uuid, $url, $merge );
3277 unless (!$tip_commit or
3278 grep { $_ eq $tip_commit } @$parents ) {
3279 push @merge_tips, $tip_commit;
3280 $ranges{$tip_commit} = \@ranges;
3281 } else {
3282 push @merge_tips, undef;
3283 }
3284 }
3285 my %excluded = map { $_ => 1 }
3287 parents_exclude($parents, grep { defined } @merge_tips);
3288 # check merge tips for new parents
3290 my @new_parents;
3291 for my $merge_tip ( @merge_tips ) {
3292 my $spec = shift @merges;
3293 next unless $merge_tip and $excluded{$merge_tip};
3294 my $ranges = $ranges{$merge_tip};
3296 # check out 'new' tips
3298 my $merge_base;
3299 eval {
3300 $merge_base = command_oneline(
3301 "merge-base",
3302 @$parents, $merge_tip,
3303 );
3304 };
3305 if ($@) {
3306 die "An error occurred during merge-base"
3307 unless $@->isa("Git::Error::Command");
3308 warn "W: Cannot find common ancestor between ".
3310 "@$parents and $merge_tip. Ignoring merge info.\n";
3311 next;
3312 }
3313 # double check that there are no missing non-merge commits
3315 my (@incomplete) = check_cherry_pick(
3316 $merge_base, $merge_tip,
3317 $parents,
3318 @$ranges,
3319 );
3320 if ( @incomplete ) {
3322 warn "W:svn cherry-pick ignored ($spec) - missing "
3323 .@incomplete." commit(s) (eg $incomplete[0])\n";
3324 } else {
3325 warn
3326 "Found merge parent (svn:mergeinfo prop): ",
3327 $merge_tip, "\n";
3328 push @new_parents, $merge_tip;
3329 }
3330 }
3331 # cater for merges which merge commits from multiple branches
3333 if ( @new_parents > 1 ) {
3334 for ( my $i = 0; $i <= $#new_parents; $i++ ) {
3335 for ( my $j = 0; $j <= $#new_parents; $j++ ) {
3336 next if $i == $j;
3337 next unless $new_parents[$i];
3338 next unless $new_parents[$j];
3339 my $revs = command_oneline(
3340 "rev-list", "-1",
3341 "$new_parents[$i]..$new_parents[$j]",
3342 );
3343 if ( !$revs ) {
3344 undef($new_parents[$j]);
3345 }
3346 }
3347 }
3348 }
3349 push @$parents, grep { defined } @new_parents;
3350}
3351sub make_log_entry {
3353 my ($self, $rev, $parents, $ed) = @_;
3354 my $untracked = $self->get_untracked($ed);
3355 my @parents = @$parents;
3357 my $ps = $ed->{path_strip} || "";
3358 for my $path ( grep { m/$ps/ } %{$ed->{dir_prop}} ) {
3359 my $props = $ed->{dir_prop}{$path};
3360 if ( $props->{"svk:merge"} ) {
3361 $self->find_extra_svk_parents
3362 ($ed, $props->{"svk:merge"}, \@parents);
3363 }
3364 if ( $props->{"svn:mergeinfo"} ) {
3365 $self->find_extra_svn_parents
3366 ($ed,
3367 $props->{"svn:mergeinfo"},
3368 \@parents);
3369 }
3370 }
3371 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
3373 print $un "r$rev\n" or croak $!;
3374 print $un $_, "\n" foreach @$untracked;
3375 my %log_entry = ( parents => \@parents, revision => $rev,
3376 log => '');
3377 my $headrev;
3379 my $logged = delete $self->{logged_rev_props};
3380 if (!$logged || $self->{-want_revprops}) {
3381 my $rp = $self->ra->rev_proplist($rev);
3382 foreach (sort keys %$rp) {
3383 my $v = $rp->{$_};
3384 if (/^svn:(author|date|log)$/) {
3385 $log_entry{$1} = $v;
3386 } elsif ($_ eq 'svm:headrev') {
3387 $headrev = $v;
3388 } else {
3389 print $un " rev_prop: ", uri_encode($_), ' ',
3390 uri_encode($v), "\n";
3391 }
3392 }
3393 } else {
3394 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
3395 }
3396 close $un or croak $!;
3397 $log_entry{date} = parse_svn_date($log_entry{date});
3399 $log_entry{log} .= "\n";
3400 my $author = $log_entry{author} = check_author($log_entry{author});
3401 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
3402 : ($author, undef);
3403 my ($commit_name, $commit_email) = ($name, $email);
3405 if ($_use_log_author) {
3406 my $name_field;
3407 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
3408 $name_field = $1;
3409 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
3410 $name_field = $1;
3411 }
3412 if (!defined $name_field) {
3413 if (!defined $email) {
3414 $email = $name;
3415 }
3416 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
3417 ($name, $email) = ($1, $2);
3418 } elsif ($name_field =~ /(.*)@/) {
3419 ($name, $email) = ($1, $name_field);
3420 } else {
3421 ($name, $email) = ($name_field, $name_field);
3422 }
3423 }
3424 if (defined $headrev && $self->use_svm_props) {
3425 if ($self->rewrite_root) {
3426 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
3427 "options set!\n";
3428 }
3429 if ($self->rewrite_uuid) {
3430 die "Can't have both 'useSvmProps' and 'rewriteUUID' ",
3431 "options set!\n";
3432 }
3433 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
3434 # we don't want "SVM: initializing mirror for junk" ...
3435 return undef if $r == 0;
3436 my $svm = $self->svm;
3437 if ($uuid ne $svm->{uuid}) {
3438 die "UUID mismatch on SVM path:\n",
3439 "expected: $svm->{uuid}\n",
3440 " got: $uuid\n";
3441 }
3442 my $full_url = $self->full_url;
3443 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
3444 die "Failed to replace '$svm->{replace}' with ",
3445 "'$svm->{source}' in $full_url\n";
3446 # throw away username for storing in records
3447 remove_username($full_url);
3448 $log_entry{metadata} = "$full_url\@$r $uuid";
3449 $log_entry{svm_revision} = $r;
3450 $email ||= "$author\@$uuid";
3451 $commit_email ||= "$author\@$uuid";
3452 } elsif ($self->use_svnsync_props) {
3453 my $full_url = $self->svnsync->{url};
3454 $full_url .= "/$self->{path}" if length $self->{path};
3455 remove_username($full_url);
3456 my $uuid = $self->svnsync->{uuid};
3457 $log_entry{metadata} = "$full_url\@$rev $uuid";
3458 $email ||= "$author\@$uuid";
3459 $commit_email ||= "$author\@$uuid";
3460 } else {
3461 my $url = $self->metadata_url;
3462 remove_username($url);
3463 my $uuid = $self->rewrite_uuid || $self->ra->get_uuid;
3464 $log_entry{metadata} = "$url\@$rev " . $uuid;
3465 $email ||= "$author\@" . $uuid;
3466 $commit_email ||= "$author\@" . $uuid;
3467 }
3468 $log_entry{name} = $name;
3469 $log_entry{email} = $email;
3470 $log_entry{commit_name} = $commit_name;
3471 $log_entry{commit_email} = $commit_email;
3472 \%log_entry;
3473}
3474sub fetch {
3476 my ($self, $min_rev, $max_rev, @parents) = @_;
3477 my ($last_rev, $last_commit) = $self->last_rev_commit;
3478 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
3479 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
3480}
3481sub set_tree_cb {
3483 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
3484 $self->{inject_parents} = { $rev => $tree };
3485 $self->fetch(undef, undef);
3486}
3487sub set_tree {
3489 my ($self, $tree) = (shift, shift);
3490 my $log_entry = ::get_commit_entry($tree);
3491 unless ($self->{last_rev}) {
3492 ::fatal("Must have an existing revision to commit");
3493 }
3494 my %ed_opts = ( r => $self->{last_rev},
3495 log => $log_entry->{log},
3496 ra => $self->ra,
3497 tree_a => $self->{last_commit},
3498 tree_b => $tree,
3499 editor_cb => sub {
3500 $self->set_tree_cb($log_entry, $tree, @_) },
3501 svn_path => $self->{path} );
3502 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
3503 print "No changes\nr$self->{last_rev} = $tree\n";
3504 }
3505}
3506sub rebuild_from_rev_db {
3508 my ($self, $path) = @_;
3509 my $r = -1;
3510 open my $fh, '<', $path or croak "open: $!";
3511 binmode $fh or croak "binmode: $!";
3512 while (<$fh>) {
3513 length($_) == 41 or croak "inconsistent size in ($_) != 41";
3514 chomp($_);
3515 ++$r;
3516 next if $_ eq ('0' x 40);
3517 $self->rev_map_set($r, $_);
3518 print "r$r = $_\n";
3519 }
3520 close $fh or croak "close: $!";
3521 unlink $path or croak "unlink: $!";
3522}
3523sub rebuild {
3525 my ($self) = @_;
3526 my $map_path = $self->map_path;
3527 my $partial = (-e $map_path && ! -z $map_path);
3528 return unless ::verify_ref($self->refname.'^0');
3529 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
3530 my $rev_db = $self->rev_db_path;
3531 $self->rebuild_from_rev_db($rev_db);
3532 if ($self->use_svm_props) {
3533 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
3534 $self->rebuild_from_rev_db($svm_rev_db);
3535 }
3536 $self->unlink_rev_db_symlink;
3537 return;
3538 }
3539 print "Rebuilding $map_path ...\n" if (!$partial);
3540 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
3541 (undef, undef));
3542 my ($log, $ctx) =
3543 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3544 ($head ? "$head.." : "") . $self->refname,
3545 '--');
3546 my $metadata_url = $self->metadata_url;
3547 remove_username($metadata_url);
3548 my $svn_uuid = $self->rewrite_uuid || $self->ra_uuid;
3549 my $c;
3550 while (<$log>) {
3551 if ( m{^commit ($::sha1)$} ) {
3552 $c = $1;
3553 next;
3554 }
3555 next unless s{^\s*(git-svn-id:)}{$1};
3556 my ($url, $rev, $uuid) = ::extract_metadata($_);
3557 remove_username($url);
3558 # ignore merges (from set-tree)
3560 next if (!defined $rev || !$uuid);
3561 # if we merged or otherwise started elsewhere, this is
3563 # how we break out of it
3564 if (($uuid ne $svn_uuid) ||
3565 ($metadata_url && $url && ($url ne $metadata_url))) {
3566 next;
3567 }
3568 if ($partial && $head) {
3569 print "Partial-rebuilding $map_path ...\n";
3570 print "Currently at $base_rev = $head\n";
3571 $head = undef;
3572 }
3573 $self->rev_map_set($rev, $c);
3575 print "r$rev = $c\n";
3576 }
3577 command_close_pipe($log, $ctx);
3578 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3579 my $rev_db_path = $self->rev_db_path;
3580 if (-f $self->rev_db_path) {
3581 unlink $self->rev_db_path or croak "unlink: $!";
3582 }
3583 $self->unlink_rev_db_symlink;
3584}
3585# rev_map:
3587# Tie::File seems to be prone to offset errors if revisions get sparse,
3588# it's not that fast, either. Tie::File is also not in Perl 5.6. So
3589# one of my favorite modules is out :< Next up would be one of the DBM
3590# modules, but I'm not sure which is most portable...
3591#
3592# This is the replacement for the rev_db format, which was too big
3593# and inefficient for large repositories with a lot of sparse history
3594# (mainly tags)
3595#
3596# The format is this:
3597# - 24 bytes for every record,
3598# * 4 bytes for the integer representing an SVN revision number
3599# * 20 bytes representing the sha1 of a git commit
3600# - No empty padding records like the old format
3601# (except the last record, which can be overwritten)
3602# - new records are written append-only since SVN revision numbers
3603# increase monotonically
3604# - lookups on SVN revision number are done via a binary search
3605# - Piping the file to xxd -c24 is a good way of dumping it for
3606# viewing or editing (piped back through xxd -r), should the need
3607# ever arise.
3608# - The last record can be padding revision with an all-zero sha1
3609# This is used to optimize fetch performance when using multiple
3610# "fetch" directives in .git/config
3611#
3612# These files are disposable unless noMetadata or useSvmProps is set
3613sub _rev_map_set {
3615 my ($fh, $rev, $commit) = @_;
3616 binmode $fh or croak "binmode: $!";
3618 my $size = (stat($fh))[7];
3619 ($size % 24) == 0 or croak "inconsistent size: $size";
3620 my $wr_offset = 0;
3622 if ($size > 0) {
3623 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3624 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3625 $read == 24 or croak "read only $read bytes (!= 24)";
3626 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3627 if ($last_commit eq ('0' x40)) {
3628 if ($size >= 48) {
3629 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3630 $read = sysread($fh, $buf, 24) or
3631 croak "read: $!";
3632 $read == 24 or
3633 croak "read only $read bytes (!= 24)";
3634 ($last_rev, $last_commit) =
3635 unpack(rev_map_fmt, $buf);
3636 if ($last_commit eq ('0' x40)) {
3637 croak "inconsistent .rev_map\n";
3638 }
3639 }
3640 if ($last_rev >= $rev) {
3641 croak "last_rev is higher!: $last_rev >= $rev";
3642 }
3643 $wr_offset = -24;
3644 }
3645 }
3646 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3647 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3648 croak "write: $!";
3649}
3650sub _rev_map_reset {
3652 my ($fh, $rev, $commit) = @_;
3653 my $c = _rev_map_get($fh, $rev);
3654 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3655 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3656 truncate $fh, $offset or croak "truncate: $!";
3657}
3658sub mkfile {
3660 my ($path) = @_;
3661 unless (-e $path) {
3662 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3663 mkpath([$dir]) unless -d $dir;
3664 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3665 close $fh or die "Couldn't close (create) $path: $!\n";
3666 }
3667}
3668sub rev_map_set {
3670 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3671 defined $commit or die "missing arg3\n";
3672 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3673 my $db = $self->map_path($uuid);
3674 my $db_lock = "$db.lock";
3675 my $sig;
3676 $update_ref ||= 0;
3677 if ($update_ref) {
3678 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3679 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3680 }
3681 mkfile($db);
3682 $LOCKFILES{$db_lock} = 1;
3684 my $sync;
3685 # both of these options make our .rev_db file very, very important
3686 # and we can't afford to lose it because rebuild() won't work
3687 if ($self->use_svm_props || $self->no_metadata) {
3688 $sync = 1;
3689 copy($db, $db_lock) or die "rev_map_set(@_): ",
3690 "Failed to copy: ",
3691 "$db => $db_lock ($!)\n";
3692 } else {
3693 rename $db, $db_lock or die "rev_map_set(@_): ",
3694 "Failed to rename: ",
3695 "$db => $db_lock ($!)\n";
3696 }
3697 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3699 or croak "Couldn't open $db_lock: $!\n";
3700 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3701 _rev_map_set($fh, $rev, $commit);
3702 if ($sync) {
3703 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3704 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3705 }
3706 close $fh or croak $!;
3707 if ($update_ref) {
3708 $_head = $self;
3709 my $note = "";
3710 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3711 command_noisy('update-ref', '-m', "r$rev$note",
3712 $self->refname, $commit);
3713 }
3714 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3715 "$db_lock => $db ($!)\n";
3716 delete $LOCKFILES{$db_lock};
3717 if ($update_ref) {
3718 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3719 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3720 kill $sig, $$ if defined $sig;
3721 }
3722}
3723# If want_commit, this will return an array of (rev, commit) where
3725# commit _must_ be a valid commit in the archive.
3726# Otherwise, it'll return the max revision (whether or not the
3727# commit is valid or just a 0x40 placeholder).
3728sub rev_map_max {
3729 my ($self, $want_commit) = @_;
3730 $self->rebuild;
3731 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3732 $want_commit ? ($r, $c) : $r;
3733}
3734sub rev_map_max_norebuild {
3736 my ($self, $want_commit) = @_;
3737 my $map_path = $self->map_path;
3738 stat $map_path or return $want_commit ? (0, undef) : 0;
3739 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3740 binmode $fh or croak "binmode: $!";
3741 my $size = (stat($fh))[7];
3742 ($size % 24) == 0 or croak "inconsistent size: $size";
3743 if ($size == 0) {
3745 close $fh or croak "close: $!";
3746 return $want_commit ? (0, undef) : 0;
3747 }
3748 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3750 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3751 my ($r, $c) = unpack(rev_map_fmt, $buf);
3752 if ($want_commit && $c eq ('0' x40)) {
3753 if ($size < 48) {
3754 return $want_commit ? (0, undef) : 0;
3755 }
3756 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3757 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3758 ($r, $c) = unpack(rev_map_fmt, $buf);
3759 if ($c eq ('0'x40)) {
3760 croak "Penultimate record is all-zeroes in $map_path";
3761 }
3762 }
3763 close $fh or croak "close: $!";
3764 $want_commit ? ($r, $c) : $r;
3765}
3766sub rev_map_get {
3768 my ($self, $rev, $uuid) = @_;
3769 my $map_path = $self->map_path($uuid);
3770 return undef unless -e $map_path;
3771 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3773 my $c = _rev_map_get($fh, $rev);
3774 close($fh) or croak "close: $!";
3775 $c
3776}
3777sub _rev_map_get {
3779 my ($fh, $rev) = @_;
3780 binmode $fh or croak "binmode: $!";
3782 my $size = (stat($fh))[7];
3783 ($size % 24) == 0 or croak "inconsistent size: $size";
3784 if ($size == 0) {
3786 return undef;
3787 }
3788 my ($l, $u) = (0, $size - 24);
3790 my ($r, $c, $buf);
3791 while ($l <= $u) {
3793 my $i = int(($l/24 + $u/24) / 2) * 24;
3794 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3795 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3796 my ($r, $c) = unpack(rev_map_fmt, $buf);
3797 if ($r < $rev) {
3799 $l = $i + 24;
3800 } elsif ($r > $rev) {
3801 $u = $i - 24;
3802 } else { # $r == $rev
3803 return $c eq ('0' x 40) ? undef : $c;
3804 }
3805 }
3806 undef;
3807}
3808# Finds the first svn revision that exists on (if $eq_ok is true) or
3810# before $rev for the current branch. It will not search any lower
3811# than $min_rev. Returns the git commit hash and svn revision number
3812# if found, else (undef, undef).
3813sub find_rev_before {
3814 my ($self, $rev, $eq_ok, $min_rev) = @_;
3815 --$rev unless $eq_ok;
3816 $min_rev ||= 1;
3817 my $max_rev = $self->rev_map_max;
3818 $rev = $max_rev if ($rev > $max_rev);
3819 while ($rev >= $min_rev) {
3820 if (my $c = $self->rev_map_get($rev)) {
3821 return ($rev, $c);
3822 }
3823 --$rev;
3824 }
3825 return (undef, undef);
3826}
3827# Finds the first svn revision that exists on (if $eq_ok is true) or
3829# after $rev for the current branch. It will not search any higher
3830# than $max_rev. Returns the git commit hash and svn revision number
3831# if found, else (undef, undef).
3832sub find_rev_after {
3833 my ($self, $rev, $eq_ok, $max_rev) = @_;
3834 ++$rev unless $eq_ok;
3835 $max_rev ||= $self->rev_map_max;
3836 while ($rev <= $max_rev) {
3837 if (my $c = $self->rev_map_get($rev)) {
3838 return ($rev, $c);
3839 }
3840 ++$rev;
3841 }
3842 return (undef, undef);
3843}
3844sub _new {
3846 my ($class, $repo_id, $ref_id, $path) = @_;
3847 unless (defined $repo_id && length $repo_id) {
3848 $repo_id = $Git::SVN::default_repo_id;
3849 }
3850 unless (defined $ref_id && length $ref_id) {
3851 $_prefix = '' unless defined($_prefix);
3852 $_[2] = $ref_id =
3853 "refs/remotes/$_prefix$Git::SVN::default_ref_id";
3854 }
3855 $_[1] = $repo_id;
3856 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3857 # Older repos imported by us used $GIT_DIR/svn/foo instead of
3859 # $GIT_DIR/svn/refs/remotes/foo when tracking refs/remotes/foo
3860 if ($ref_id =~ m{^refs/remotes/(.*)}) {
3861 my $old_dir = "$ENV{GIT_DIR}/svn/$1";
3862 if (-d $old_dir && ! -d $dir) {
3863 $dir = $old_dir;
3864 }
3865 }
3866 $_[3] = $path = '' unless (defined $path);
3868 mkpath([$dir]);
3869 bless {
3870 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3871 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3872 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3873}
3874# for read-only access of old .rev_db formats
3876sub unlink_rev_db_symlink {
3877 my ($self) = @_;
3878 my $link = $self->rev_db_path;
3879 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3880 if (-l $link) {
3881 unlink $link or croak "unlink: $link failed!";
3882 }
3883}
3884sub rev_db_path {
3886 my ($self, $uuid) = @_;
3887 my $db_path = $self->map_path($uuid);
3888 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3889 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3890 $db_path;
3891}
3892# the new replacement for .rev_db
3894sub map_path {
3895 my ($self, $uuid) = @_;
3896 $uuid ||= $self->ra_uuid;
3897 "$self->{map_root}.$uuid";
3898}
3899sub uri_encode {
3901 my ($f) = @_;
3902 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3903 $f
3904}
3905sub uri_decode {
3907 my ($f) = @_;
3908 $f =~ s#%([0-9a-fA-F]{2})#chr(hex($1))#eg;
3909 $f
3910}
3911sub remove_username {
3913 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3914}
3915package Git::SVN::Prompt;
3917use strict;
3918use warnings;
3919require SVN::Core;
3920use vars qw/$_no_auth_cache $_username/;
3921sub simple {
3923 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3924 $may_save = undef if $_no_auth_cache;
3925 $default_username = $_username if defined $_username;
3926 if (defined $default_username && length $default_username) {
3927 if (defined $realm && length $realm) {
3928 print STDERR "Authentication realm: $realm\n";
3929 STDERR->flush;
3930 }
3931 $cred->username($default_username);
3932 } else {
3933 username($cred, $realm, $may_save, $pool);
3934 }
3935 $cred->password(_read_password("Password for '" .
3936 $cred->username . "': ", $realm));
3937 $cred->may_save($may_save);
3938 $SVN::_Core::SVN_NO_ERROR;
3939}
3940sub ssl_server_trust {
3942 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3943 $may_save = undef if $_no_auth_cache;
3944 print STDERR "Error validating server certificate for '$realm':\n";
3945 {
3946 no warnings 'once';
3947 # All variables SVN::Auth::SSL::* are used only once,
3948 # so we're shutting up Perl warnings about this.
3949 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3950 print STDERR " - The certificate is not issued ",
3951 "by a trusted authority. Use the\n",
3952 " fingerprint to validate ",
3953 "the certificate manually!\n";
3954 }
3955 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3956 print STDERR " - The certificate hostname ",
3957 "does not match.\n";
3958 }
3959 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3960 print STDERR " - The certificate is not yet valid.\n";
3961 }
3962 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3963 print STDERR " - The certificate has expired.\n";
3964 }
3965 if ($failures & $SVN::Auth::SSL::OTHER) {
3966 print STDERR " - The certificate has ",
3967 "an unknown error.\n";
3968 }
3969 } # no warnings 'once'
3970 printf STDERR
3971 "Certificate information:\n".
3972 " - Hostname: %s\n".
3973 " - Valid: from %s until %s\n".
3974 " - Issuer: %s\n".
3975 " - Fingerprint: %s\n",
3976 map $cert_info->$_, qw(hostname valid_from valid_until
3977 issuer_dname fingerprint);
3978 my $choice;
3979prompt:
3980 print STDERR $may_save ?
3981 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3982 "(R)eject or accept (t)emporarily? ";
3983 STDERR->flush;
3984 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3985 if ($choice =~ /^t$/i) {
3986 $cred->may_save(undef);
3987 } elsif ($choice =~ /^r$/i) {
3988 return -1;
3989 } elsif ($may_save && $choice =~ /^p$/i) {
3990 $cred->may_save($may_save);
3991 } else {
3992 goto prompt;
3993 }
3994 $cred->accepted_failures($failures);
3995 $SVN::_Core::SVN_NO_ERROR;
3996}
3997sub ssl_client_cert {
3999 my ($cred, $realm, $may_save, $pool) = @_;
4000 $may_save = undef if $_no_auth_cache;
4001 print STDERR "Client certificate filename: ";
4002 STDERR->flush;
4003 chomp(my $filename = <STDIN>);
4004 $cred->cert_file($filename);
4005 $cred->may_save($may_save);
4006 $SVN::_Core::SVN_NO_ERROR;
4007}
4008sub ssl_client_cert_pw {
4010 my ($cred, $realm, $may_save, $pool) = @_;
4011 $may_save = undef if $_no_auth_cache;
4012 $cred->password(_read_password("Password: ", $realm));
4013 $cred->may_save($may_save);
4014 $SVN::_Core::SVN_NO_ERROR;
4015}
4016sub username {
4018 my ($cred, $realm, $may_save, $pool) = @_;
4019 $may_save = undef if $_no_auth_cache;
4020 if (defined $realm && length $realm) {
4021 print STDERR "Authentication realm: $realm\n";
4022 }
4023 my $username;
4024 if (defined $_username) {
4025 $username = $_username;
4026 } else {
4027 print STDERR "Username: ";
4028 STDERR->flush;
4029 chomp($username = <STDIN>);
4030 }
4031 $cred->username($username);
4032 $cred->may_save($may_save);
4033 $SVN::_Core::SVN_NO_ERROR;
4034}
4035sub _read_password {
4037 my ($prompt, $realm) = @_;
4038 my $password = '';
4039 if (exists $ENV{GIT_ASKPASS}) {
4040 open(PH, "-|", $ENV{GIT_ASKPASS}, $prompt);
4041 $password = <PH>;
4042 $password =~ s/[\012\015]//; # \n\r
4043 close(PH);
4044 } else {
4045 print STDERR $prompt;
4046 STDERR->flush;
4047 require Term::ReadKey;
4048 Term::ReadKey::ReadMode('noecho');
4049 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
4050 last if $key =~ /[\012\015]/; # \n\r
4051 $password .= $key;
4052 }
4053 Term::ReadKey::ReadMode('restore');
4054 print STDERR "\n";
4055 STDERR->flush;
4056 }
4057 $password;
4058}
4059package SVN::Git::Fetcher;
4061use vars qw/@ISA/;
4062use strict;
4063use warnings;
4064use Carp qw/croak/;
4065use IO::File qw//;
4066use vars qw/$_ignore_regex/;
4067# file baton members: path, mode_a, mode_b, pool, fh, blob, base
4069sub new {
4070 my ($class, $git_svn, $switch_path) = @_;
4071 my $self = SVN::Delta::Editor->new;
4072 bless $self, $class;
4073 if (exists $git_svn->{last_commit}) {
4074 $self->{c} = $git_svn->{last_commit};
4075 $self->{empty_symlinks} =
4076 _mark_empty_symlinks($git_svn, $switch_path);
4077 }
4078 $self->{ignore_regex} = eval { command_oneline('config', '--get',
4079 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
4080 $self->{empty} = {};
4081 $self->{dir_prop} = {};
4082 $self->{file_prop} = {};
4083 $self->{absent_dir} = {};
4084 $self->{absent_file} = {};
4085 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
4086 $self->{pathnameencoding} = Git::config('svn.pathnameencoding');
4087 $self;
4088}
4089# this uses the Ra object, so it must be called before do_{switch,update},
4091# not inside them (when the Git::SVN::Fetcher object is passed) to
4092# do_{switch,update}
4093sub _mark_empty_symlinks {
4094 my ($git_svn, $switch_path) = @_;
4095 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
4096 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
4097 my %ret;
4099 my ($rev, $cmt) = $git_svn->last_rev_commit;
4100 return {} unless ($rev && $cmt);
4101 # allow the warning to be printed for each revision we fetch to
4103 # ensure the user sees it. The user can also disable the workaround
4104 # on the repository even while git svn is running and the next
4105 # revision fetched will skip this expensive function.
4106 my $printed_warning;
4107 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
4108 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
4109 local $/ = "\0";
4110 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
4111 $pfx .= '/' if length($pfx);
4112 while (<$ls>) {
4113 chomp;
4114 s/\A100644 blob $empty_blob\t//o or next;
4115 unless ($printed_warning) {
4116 print STDERR "Scanning for empty symlinks, ",
4117 "this may take a while if you have ",
4118 "many empty files\n",
4119 "You may disable this with `",
4120 "git config svn.brokenSymlinkWorkaround ",
4121 "false'.\n",
4122 "This may be done in a different ",
4123 "terminal without restarting ",
4124 "git svn\n";
4125 $printed_warning = 1;
4126 }
4127 my $path = $_;
4128 my (undef, $props) =
4129 $git_svn->ra->get_file($pfx.$path, $rev, undef);
4130 if ($props->{'svn:special'}) {
4131 $ret{$path} = 1;
4132 }
4133 }
4134 command_close_pipe($ls, $ctx);
4135 \%ret;
4136}
4137# returns true if a given path is inside a ".git" directory
4139sub in_dot_git {
4140 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
4141}
4142# return value: 0 -- don't ignore, 1 -- ignore
4144sub is_path_ignored {
4145 my ($self, $path) = @_;
4146 return 1 if in_dot_git($path);
4147 return 1 if defined($self->{ignore_regex}) &&
4148 $path =~ m!$self->{ignore_regex}!;
4149 return 0 unless defined($_ignore_regex);
4150 return 1 if $path =~ m!$_ignore_regex!o;
4151 return 0;
4152}
4153sub set_path_strip {
4155 my ($self, $path) = @_;
4156 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
4157}
4158sub open_root {
4160 { path => '' };
4161}
4162sub open_directory {
4164 my ($self, $path, $pb, $rev) = @_;
4165 { path => $path };
4166}
4167sub git_path {
4169 my ($self, $path) = @_;
4170 if (my $enc = $self->{pathnameencoding}) {
4171 require Encode;
4172 Encode::from_to($path, 'UTF-8', $enc);
4173 }
4174 if ($self->{path_strip}) {
4175 $path =~ s!$self->{path_strip}!! or
4176 die "Failed to strip path '$path' ($self->{path_strip})\n";
4177 }
4178 $path;
4179}
4180sub delete_entry {
4182 my ($self, $path, $rev, $pb) = @_;
4183 return undef if $self->is_path_ignored($path);
4184 my $gpath = $self->git_path($path);
4186 return undef if ($gpath eq '');
4187 # remove entire directories.
4189 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4190 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
4191 if ($tree) {
4192 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4193 -r --name-only -z/,
4194 $tree);
4195 local $/ = "\0";
4196 while (<$ls>) {
4197 chomp;
4198 my $rmpath = "$gpath/$_";
4199 $self->{gii}->remove($rmpath);
4200 print "\tD\t$rmpath\n" unless $::_q;
4201 }
4202 print "\tD\t$gpath/\n" unless $::_q;
4203 command_close_pipe($ls, $ctx);
4204 } else {
4205 $self->{gii}->remove($gpath);
4206 print "\tD\t$gpath\n" unless $::_q;
4207 }
4208 $self->{empty}->{$path} = 0;
4209 undef;
4210}
4211sub open_file {
4213 my ($self, $path, $pb, $rev) = @_;
4214 my ($mode, $blob);
4215 goto out if $self->is_path_ignored($path);
4217 my $gpath = $self->git_path($path);
4219 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
4220 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
4221 unless (defined $mode && defined $blob) {
4222 die "$path was not found in commit $self->{c} (r$rev)\n";
4223 }
4224 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
4225 $mode = '120000';
4226 }
4227out:
4228 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
4229 pool => SVN::Pool->new, action => 'M' };
4230}
4231sub add_file {
4233 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
4234 my $mode;
4235 if (!$self->is_path_ignored($path)) {
4237 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4238 delete $self->{empty}->{$dir};
4239 $mode = '100644';
4240 }
4241 { path => $path, mode_a => $mode, mode_b => $mode,
4242 pool => SVN::Pool->new, action => 'A' };
4243}
4244sub add_directory {
4246 my ($self, $path, $cp_path, $cp_rev) = @_;
4247 goto out if $self->is_path_ignored($path);
4248 my $gpath = $self->git_path($path);
4249 if ($gpath eq '') {
4250 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
4251 -r --name-only -z/,
4252 $self->{c});
4253 local $/ = "\0";
4254 while (<$ls>) {
4255 chomp;
4256 $self->{gii}->remove($_);
4257 print "\tD\t$_\n" unless $::_q;
4258 }
4259 command_close_pipe($ls, $ctx);
4260 $self->{empty}->{$path} = 0;
4261 }
4262 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
4263 delete $self->{empty}->{$dir};
4264 $self->{empty}->{$path} = 1;
4265out:
4266 { path => $path };
4267}
4268sub change_dir_prop {
4270 my ($self, $db, $prop, $value) = @_;
4271 return undef if $self->is_path_ignored($db->{path});
4272 $self->{dir_prop}->{$db->{path}} ||= {};
4273 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
4274 undef;
4275}
4276sub absent_directory {
4278 my ($self, $path, $pb) = @_;
4279 return undef if $self->is_path_ignored($path);
4280 $self->{absent_dir}->{$pb->{path}} ||= [];
4281 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
4282 undef;
4283}
4284sub absent_file {
4286 my ($self, $path, $pb) = @_;
4287 return undef if $self->is_path_ignored($path);
4288 $self->{absent_file}->{$pb->{path}} ||= [];
4289 push @{$self->{absent_file}->{$pb->{path}}}, $path;
4290 undef;
4291}
4292sub change_file_prop {
4294 my ($self, $fb, $prop, $value) = @_;
4295 return undef if $self->is_path_ignored($fb->{path});
4296 if ($prop eq 'svn:executable') {
4297 if ($fb->{mode_b} != 120000) {
4298 $fb->{mode_b} = defined $value ? 100755 : 100644;
4299 }
4300 } elsif ($prop eq 'svn:special') {
4301 $fb->{mode_b} = defined $value ? 120000 : 100644;
4302 } else {
4303 $self->{file_prop}->{$fb->{path}} ||= {};
4304 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
4305 }
4306 undef;
4307}
4308sub apply_textdelta {
4310 my ($self, $fb, $exp) = @_;
4311 return undef if $self->is_path_ignored($fb->{path});
4312 my $fh = $::_repository->temp_acquire('svn_delta');
4313 # $fh gets auto-closed() by SVN::TxDelta::apply(),
4314 # (but $base does not,) so dup() it for reading in close_file
4315 open my $dup, '<&', $fh or croak $!;
4316 my $base = $::_repository->temp_acquire('git_blob');
4317 if ($fb->{blob}) {
4319 my ($base_is_link, $size);
4320 if ($fb->{mode_a} eq '120000' &&
4322 ! $self->{empty_symlinks}->{$fb->{path}}) {
4323 print $base 'link ' or die "print $!\n";
4324 $base_is_link = 1;
4325 }
4326 retry:
4327 $size = $::_repository->cat_blob($fb->{blob}, $base);
4328 die "Failed to read object $fb->{blob}" if ($size < 0);
4329 if (defined $exp) {
4331 seek $base, 0, 0 or croak $!;
4332 my $got = ::md5sum($base);
4333 if ($got ne $exp) {
4334 my $err = "Checksum mismatch: ".
4335 "$fb->{path} $fb->{blob}\n" .
4336 "expected: $exp\n" .
4337 " got: $got\n";
4338 if ($base_is_link) {
4339 warn $err,
4340 "Retrying... (possibly ",
4341 "a bad symlink from SVN)\n";
4342 $::_repository->temp_reset($base);
4343 $base_is_link = 0;
4344 goto retry;
4345 }
4346 die $err;
4347 }
4348 }
4349 }
4350 seek $base, 0, 0 or croak $!;
4351 $fb->{fh} = $fh;
4352 $fb->{base} = $base;
4353 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
4354}
4355sub close_file {
4357 my ($self, $fb, $exp) = @_;
4358 return undef if $self->is_path_ignored($fb->{path});
4359 my $hash;
4361 my $path = $self->git_path($fb->{path});
4362 if (my $fh = $fb->{fh}) {
4363 if (defined $exp) {
4364 seek($fh, 0, 0) or croak $!;
4365 my $got = ::md5sum($fh);
4366 if ($got ne $exp) {
4367 die "Checksum mismatch: $path\n",
4368 "expected: $exp\n got: $got\n";
4369 }
4370 }
4371 if ($fb->{mode_b} == 120000) {
4372 sysseek($fh, 0, 0) or croak $!;
4373 my $rd = sysread($fh, my $buf, 5);
4374 if (!defined $rd) {
4376 croak "sysread: $!\n";
4377 } elsif ($rd == 0) {
4378 warn "$path has mode 120000",
4379 " but it points to nothing\n",
4380 "converting to an empty file with mode",
4381 " 100644\n";
4382 $fb->{mode_b} = '100644';
4383 } elsif ($buf ne 'link ') {
4384 warn "$path has mode 120000",
4385 " but is not a link\n";
4386 } else {
4387 my $tmp_fh = $::_repository->temp_acquire(
4388 'svn_hash');
4389 my $res;
4390 while ($res = sysread($fh, my $str, 1024)) {
4391 my $out = syswrite($tmp_fh, $str, $res);
4392 defined($out) && $out == $res
4393 or croak("write ",
4394 Git::temp_path($tmp_fh),
4395 ": $!\n");
4396 }
4397 defined $res or croak $!;
4398 ($fh, $tmp_fh) = ($tmp_fh, $fh);
4400 Git::temp_release($tmp_fh, 1);
4401 }
4402 }
4403 $hash = $::_repository->hash_and_insert_object(
4405 Git::temp_path($fh));
4406 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
4407 Git::temp_release($fb->{base}, 1);
4409 Git::temp_release($fh, 1);
4410 } else {
4411 $hash = $fb->{blob} or die "no blob information\n";
4412 }
4413 $fb->{pool}->clear;
4414 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
4415 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
4416 undef;
4417}
4418sub abort_edit {
4420 my $self = shift;
4421 $self->{nr} = $self->{gii}->{nr};
4422 delete $self->{gii};
4423 $self->SUPER::abort_edit(@_);
4424}
4425sub close_edit {
4427 my $self = shift;
4428 $self->{git_commit_ok} = 1;
4429 $self->{nr} = $self->{gii}->{nr};
4430 delete $self->{gii};
4431 $self->SUPER::close_edit(@_);
4432}
4433package SVN::Git::Editor;
4435use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
4436use strict;
4437use warnings;
4438use Carp qw/croak/;
4439use IO::File;
4440sub new {
4442 my ($class, $opts) = @_;
4443 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
4444 die "$_ required!\n" unless (defined $opts->{$_});
4445 }
4446 my $pool = SVN::Pool->new;
4448 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
4449 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
4450 $opts->{r}, $mods);
4451 # $opts->{ra} functions should not be used after this:
4453 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
4454 $opts->{editor_cb}, $pool);
4455 my $self = SVN::Delta::Editor->new(@ce, $pool);
4456 bless $self, $class;
4457 foreach (qw/svn_path r tree_a tree_b/) {
4458 $self->{$_} = $opts->{$_};
4459 }
4460 $self->{url} = $opts->{ra}->{url};
4461 $self->{mods} = $mods;
4462 $self->{types} = $types;
4463 $self->{pool} = $pool;
4464 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
4465 $self->{rm} = { };
4466 $self->{path_prefix} = length $self->{svn_path} ?
4467 "$self->{svn_path}/" : '';
4468 $self->{config} = $opts->{config};
4469 $self->{mergeinfo} = $opts->{mergeinfo};
4470 return $self;
4471}
4472sub generate_diff {
4474 my ($tree_a, $tree_b) = @_;
4475 my @diff_tree = qw(diff-tree -z -r);
4476 if ($_cp_similarity) {
4477 push @diff_tree, "-C$_cp_similarity";
4478 } else {
4479 push @diff_tree, '-C';
4480 }
4481 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
4482 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
4483 push @diff_tree, $tree_a, $tree_b;
4484 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
4485 local $/ = "\0";
4486 my $state = 'meta';
4487 my @mods;
4488 while (<$diff_fh>) {
4489 chomp $_; # this gets rid of the trailing "\0"
4490 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
4491 ($::sha1)\s($::sha1)\s
4492 ([MTCRAD])\d*$/xo) {
4493 push @mods, { mode_a => $1, mode_b => $2,
4494 sha1_a => $3, sha1_b => $4,
4495 chg => $5 };
4496 if ($5 =~ /^(?:C|R)$/) {
4497 $state = 'file_a';
4498 } else {
4499 $state = 'file_b';
4500 }
4501 } elsif ($state eq 'file_a') {
4502 my $x = $mods[$#mods] or croak "Empty array\n";
4503 if ($x->{chg} !~ /^(?:C|R)$/) {
4504 croak "Error parsing $_, $x->{chg}\n";
4505 }
4506 $x->{file_a} = $_;
4507 $state = 'file_b';
4508 } elsif ($state eq 'file_b') {
4509 my $x = $mods[$#mods] or croak "Empty array\n";
4510 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
4511 croak "Error parsing $_, $x->{chg}\n";
4512 }
4513 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
4514 croak "Error parsing $_, $x->{chg}\n";
4515 }
4516 $x->{file_b} = $_;
4517 $state = 'meta';
4518 } else {
4519 croak "Error parsing $_\n";
4520 }
4521 }
4522 command_close_pipe($diff_fh, $ctx);
4523 \@mods;
4524}
4525sub check_diff_paths {
4527 my ($ra, $pfx, $rev, $mods) = @_;
4528 my %types;
4529 $pfx .= '/' if length $pfx;
4530 sub type_diff_paths {
4532 my ($ra, $types, $path, $rev) = @_;
4533 my @p = split m#/+#, $path;
4534 my $c = shift @p;
4535 unless (defined $types->{$c}) {
4536 $types->{$c} = $ra->check_path($c, $rev);
4537 }
4538 while (@p) {
4539 $c .= '/' . shift @p;
4540 next if defined $types->{$c};
4541 $types->{$c} = $ra->check_path($c, $rev);
4542 }
4543 }
4544 foreach my $m (@$mods) {
4546 foreach my $f (qw/file_a file_b/) {
4547 next unless defined $m->{$f};
4548 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
4549 if (length $pfx.$dir && ! defined $types{$dir}) {
4550 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
4551 }
4552 }
4553 }
4554 \%types;
4555}
4556sub split_path {
4558 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
4559}
4560sub repo_path {
4562 my ($self, $path) = @_;
4563 if (my $enc = $self->{pathnameencoding}) {
4564 require Encode;
4565 Encode::from_to($path, $enc, 'UTF-8');
4566 }
4567 $self->{path_prefix}.(defined $path ? $path : '');
4568}
4569sub url_path {
4571 my ($self, $path) = @_;
4572 if ($self->{url} =~ m#^https?://#) {
4573 $path =~ s!([^~a-zA-Z0-9_./-])!uc sprintf("%%%02x",ord($1))!eg;
4574 }
4575 $self->{url} . '/' . $self->repo_path($path);
4576}
4577sub rmdirs {
4579 my ($self) = @_;
4580 my $rm = $self->{rm};
4581 delete $rm->{''}; # we never delete the url we're tracking
4582 return unless %$rm;
4583 foreach (keys %$rm) {
4585 my @d = split m#/#, $_;
4586 my $c = shift @d;
4587 $rm->{$c} = 1;
4588 while (@d) {
4589 $c .= '/' . shift @d;
4590 $rm->{$c} = 1;
4591 }
4592 }
4593 delete $rm->{$self->{svn_path}};
4594 delete $rm->{''}; # we never delete the url we're tracking
4595 return unless %$rm;
4596 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4598 $self->{tree_b});
4599 local $/ = "\0";
4600 while (<$fh>) {
4601 chomp;
4602 my @dn = split m#/#, $_;
4603 while (pop @dn) {
4604 delete $rm->{join '/', @dn};
4605 }
4606 unless (%$rm) {
4607 close $fh;
4608 return;
4609 }
4610 }
4611 command_close_pipe($fh, $ctx);
4612 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4614 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4615 $self->close_directory($bat->{$d}, $p);
4616 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4617 print "\tD+\t$d/\n" unless $::_q;
4618 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4619 delete $bat->{$d};
4620 }
4621}
4622sub open_or_add_dir {
4624 my ($self, $full_path, $baton) = @_;
4625 my $t = $self->{types}->{$full_path};
4626 if (!defined $t) {
4627 die "$full_path not known in r$self->{r} or we have a bug!\n";
4628 }
4629 {
4630 no warnings 'once';
4631 # SVN::Node::none and SVN::Node::file are used only once,
4632 # so we're shutting up Perl's warnings about them.
4633 if ($t == $SVN::Node::none) {
4634 return $self->add_directory($full_path, $baton,
4635 undef, -1, $self->{pool});
4636 } elsif ($t == $SVN::Node::dir) {
4637 return $self->open_directory($full_path, $baton,
4638 $self->{r}, $self->{pool});
4639 } # no warnings 'once'
4640 print STDERR "$full_path already exists in repository at ",
4641 "r$self->{r} and it is not a directory (",
4642 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4643 } # no warnings 'once'
4644 exit 1;
4645}
4646sub ensure_path {
4648 my ($self, $path) = @_;
4649 my $bat = $self->{bat};
4650 my $repo_path = $self->repo_path($path);
4651 return $bat->{''} unless (length $repo_path);
4652 my @p = split m#/+#, $repo_path;
4653 my $c = shift @p;
4654 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4655 while (@p) {
4656 my $c0 = $c;
4657 $c .= '/' . shift @p;
4658 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4659 }
4660 return $bat->{$c};
4661}
4662# Subroutine to convert a globbing pattern to a regular expression.
4664# From perl cookbook.
4665sub glob2pat {
4666 my $globstr = shift;
4667 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4668 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4669 return '^' . $globstr . '$';
4670}
4671sub check_autoprop {
4673 my ($self, $pattern, $properties, $file, $fbat) = @_;
4674 # Convert the globbing pattern to a regular expression.
4675 my $regex = glob2pat($pattern);
4676 # Check if the pattern matches the file name.
4677 if($file =~ m/($regex)/) {
4678 # Parse the list of properties to set.
4679 my @props = split(/;/, $properties);
4680 foreach my $prop (@props) {
4681 # Parse 'name=value' syntax and set the property.
4682 if ($prop =~ /([^=]+)=(.*)/) {
4683 my ($n,$v) = ($1,$2);
4684 for ($n, $v) {
4685 s/^\s+//; s/\s+$//;
4686 }
4687 $self->change_file_prop($fbat, $n, $v);
4688 }
4689 }
4690 }
4691}
4692sub apply_autoprops {
4694 my ($self, $file, $fbat) = @_;
4695 my $conf_t = ${$self->{config}}{'config'};
4696 no warnings 'once';
4697 # Check [miscellany]/enable-auto-props in svn configuration.
4698 if (SVN::_Core::svn_config_get_bool(
4699 $conf_t,
4700 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4701 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4702 0)) {
4703 # Auto-props are enabled. Enumerate them to look for matches.
4704 my $callback = sub {
4705 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4706 };
4707 SVN::_Core::svn_config_enumerate(
4708 $conf_t,
4709 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4710 $callback);
4711 }
4712}
4713sub A {
4715 my ($self, $m) = @_;
4716 my ($dir, $file) = split_path($m->{file_b});
4717 my $pbat = $self->ensure_path($dir);
4718 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4719 undef, -1);
4720 print "\tA\t$m->{file_b}\n" unless $::_q;
4721 $self->apply_autoprops($file, $fbat);
4722 $self->chg_file($fbat, $m);
4723 $self->close_file($fbat,undef,$self->{pool});
4724}
4725sub C {
4727 my ($self, $m) = @_;
4728 my ($dir, $file) = split_path($m->{file_b});
4729 my $pbat = $self->ensure_path($dir);
4730 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4731 $self->url_path($m->{file_a}), $self->{r});
4732 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4733 $self->chg_file($fbat, $m);
4734 $self->close_file($fbat,undef,$self->{pool});
4735}
4736sub delete_entry {
4738 my ($self, $path, $pbat) = @_;
4739 my $rpath = $self->repo_path($path);
4740 my ($dir, $file) = split_path($rpath);
4741 $self->{rm}->{$dir} = 1;
4742 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4743}
4744sub R {
4746 my ($self, $m) = @_;
4747 my ($dir, $file) = split_path($m->{file_b});
4748 my $pbat = $self->ensure_path($dir);
4749 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4750 $self->url_path($m->{file_a}), $self->{r});
4751 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4752 $self->apply_autoprops($file, $fbat);
4753 $self->chg_file($fbat, $m);
4754 $self->close_file($fbat,undef,$self->{pool});
4755 ($dir, $file) = split_path($m->{file_a});
4757 $pbat = $self->ensure_path($dir);
4758 $self->delete_entry($m->{file_a}, $pbat);
4759}
4760sub M {
4762 my ($self, $m) = @_;
4763 my ($dir, $file) = split_path($m->{file_b});
4764 my $pbat = $self->ensure_path($dir);
4765 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4766 $pbat,$self->{r},$self->{pool});
4767 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4768 $self->chg_file($fbat, $m);
4769 $self->close_file($fbat,undef,$self->{pool});
4770}
4771sub T { shift->M(@_) }
4773sub change_file_prop {
4775 my ($self, $fbat, $pname, $pval) = @_;
4776 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4777}
4778sub change_dir_prop {
4780 my ($self, $pbat, $pname, $pval) = @_;
4781 $self->SUPER::change_dir_prop($pbat, $pname, $pval, $self->{pool});
4782}
4783sub _chg_file_get_blob ($$$$) {
4785 my ($self, $fbat, $m, $which) = @_;
4786 my $fh = $::_repository->temp_acquire("git_blob_$which");
4787 if ($m->{"mode_$which"} =~ /^120/) {
4788 print $fh 'link ' or croak $!;
4789 $self->change_file_prop($fbat,'svn:special','*');
4790 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4791 $self->change_file_prop($fbat,'svn:special',undef);
4792 }
4793 my $blob = $m->{"sha1_$which"};
4794 return ($fh,) if ($blob =~ /^0{40}$/);
4795 my $size = $::_repository->cat_blob($blob, $fh);
4796 croak "Failed to read object $blob" if ($size < 0);
4797 $fh->flush == 0 or croak $!;
4798 seek $fh, 0, 0 or croak $!;
4799 my $exp = ::md5sum($fh);
4801 seek $fh, 0, 0 or croak $!;
4802 return ($fh, $exp);
4803}
4804sub chg_file {
4806 my ($self, $fbat, $m) = @_;
4807 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4808 $self->change_file_prop($fbat,'svn:executable','*');
4809 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4810 $self->change_file_prop($fbat,'svn:executable',undef);
4811 }
4812 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4813 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4814 my $pool = SVN::Pool->new;
4815 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4816 if (-s $fh_a) {
4817 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4818 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4819 if (defined $res) {
4820 die "Unexpected result from send_txstream: $res\n",
4821 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4822 }
4823 } else {
4824 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4825 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4826 if ($got ne $exp_b);
4827 }
4828 Git::temp_release($fh_b, 1);
4829 Git::temp_release($fh_a, 1);
4830 $pool->clear;
4831}
4832sub D {
4834 my ($self, $m) = @_;
4835 my ($dir, $file) = split_path($m->{file_b});
4836 my $pbat = $self->ensure_path($dir);
4837 print "\tD\t$m->{file_b}\n" unless $::_q;
4838 $self->delete_entry($m->{file_b}, $pbat);
4839}
4840sub close_edit {
4842 my ($self) = @_;
4843 my ($p,$bat) = ($self->{pool}, $self->{bat});
4844 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4845 next if $_ eq '';
4846 $self->close_directory($bat->{$_}, $p);
4847 }
4848 $self->close_directory($bat->{''}, $p);
4849 $self->SUPER::close_edit($p);
4850 $p->clear;
4851}
4852sub abort_edit {
4854 my ($self) = @_;
4855 $self->SUPER::abort_edit($self->{pool});
4856}
4857sub DESTROY {
4859 my $self = shift;
4860 $self->SUPER::DESTROY(@_);
4861 $self->{pool}->clear;
4862}
4863# this drives the editor
4865sub apply_diff {
4866 my ($self) = @_;
4867 my $mods = $self->{mods};
4868 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4869 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4870 my $f = $m->{chg};
4871 if (defined $o{$f}) {
4872 $self->$f($m);
4873 } else {
4874 fatal("Invalid change type: $f");
4875 }
4876 }
4877 if (defined($self->{mergeinfo})) {
4879 $self->change_dir_prop($self->{bat}{''}, "svn:mergeinfo",
4880 $self->{mergeinfo});
4881 }
4882 $self->rmdirs if $_rmdir;
4883 if (@$mods == 0) {
4884 $self->abort_edit;
4885 } else {
4886 $self->close_edit;
4887 }
4888 return scalar @$mods;
4889}
4890package Git::SVN::Ra;
4892use vars qw/@ISA $config_dir $_log_window_size/;
4893use strict;
4894use warnings;
4895my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4896BEGIN {
4898 # enforce temporary pool usage for some simple functions
4899 no strict 'refs';
4900 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4901 get_file/) {
4902 my $SUPER = "SUPER::$f";
4903 *$f = sub {
4904 my $self = shift;
4905 my $pool = SVN::Pool->new;
4906 my @ret = $self->$SUPER(@_,$pool);
4907 $pool->clear;
4908 wantarray ? @ret : $ret[0];
4909 };
4910 }
4911}
4912sub _auth_providers () {
4914 [
4915 SVN::Client::get_simple_provider(),
4916 SVN::Client::get_ssl_server_trust_file_provider(),
4917 SVN::Client::get_simple_prompt_provider(
4918 \&Git::SVN::Prompt::simple, 2),
4919 SVN::Client::get_ssl_client_cert_file_provider(),
4920 SVN::Client::get_ssl_client_cert_prompt_provider(
4921 \&Git::SVN::Prompt::ssl_client_cert, 2),
4922 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4923 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4924 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4925 SVN::Client::get_username_provider(),
4926 SVN::Client::get_ssl_server_trust_prompt_provider(
4927 \&Git::SVN::Prompt::ssl_server_trust),
4928 SVN::Client::get_username_prompt_provider(
4929 \&Git::SVN::Prompt::username, 2)
4930 ]
4931}
4932sub escape_uri_only {
4934 my ($uri) = @_;
4935 my @tmp;
4936 foreach (split m{/}, $uri) {
4937 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4938 push @tmp, $_;
4939 }
4940 join('/', @tmp);
4941}
4942sub escape_url {
4944 my ($url) = @_;
4945 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4946 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4947 $url = "$scheme://$domain$uri";
4948 }
4949 $url;
4950}
4951sub new {
4953 my ($class, $url) = @_;
4954 $url =~ s!/+$!!;
4955 return $RA if ($RA && $RA->{url} eq $url);
4956 ::_req_svn();
4958 SVN::_Core::svn_config_ensure($config_dir, undef);
4960 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4961 my $config = SVN::Core::config_get_config($config_dir);
4962 $RA = undef;
4963 my $dont_store_passwords = 1;
4964 my $conf_t = ${$config}{'config'};
4965 {
4966 no warnings 'once';
4967 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4968 # produces warnings that variables are used only once.
4969 # I had not found the better way to shut them up, so
4970 # the warnings of type 'once' are disabled in this block.
4971 if (SVN::_Core::svn_config_get_bool($conf_t,
4972 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4973 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4974 1) == 0) {
4975 SVN::_Core::svn_auth_set_parameter($baton,
4976 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4977 bless (\$dont_store_passwords, "_p_void"));
4978 }
4979 if (SVN::_Core::svn_config_get_bool($conf_t,
4980 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4981 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4982 1) == 0) {
4983 $Git::SVN::Prompt::_no_auth_cache = 1;
4984 }
4985 } # no warnings 'once'
4986 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4987 config => $config,
4988 pool => SVN::Pool->new,
4989 auth_provider_callbacks => $callbacks);
4990 $self->{url} = $url;
4991 $self->{svn_path} = $url;
4992 $self->{repos_root} = $self->get_repos_root;
4993 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4994 $self->{cache} = { check_path => { r => 0, data => {} },
4995 get_dir => { r => 0, data => {} } };
4996 $RA = bless $self, $class;
4997}
4998sub check_path {
5000 my ($self, $path, $r) = @_;
5001 my $cache = $self->{cache}->{check_path};
5002 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
5003 return $cache->{data}->{$path};
5004 }
5005 my $pool = SVN::Pool->new;
5006 my $t = $self->SUPER::check_path($path, $r, $pool);
5007 $pool->clear;
5008 if ($r != $cache->{r}) {
5009 %{$cache->{data}} = ();
5010 $cache->{r} = $r;
5011 }
5012 $cache->{data}->{$path} = $t;
5013}
5014sub get_dir {
5016 my ($self, $dir, $r) = @_;
5017 my $cache = $self->{cache}->{get_dir};
5018 if ($r == $cache->{r}) {
5019 if (my $x = $cache->{data}->{$dir}) {
5020 return wantarray ? @$x : $x->[0];
5021 }
5022 }
5023 my $pool = SVN::Pool->new;
5024 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
5025 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
5026 $pool->clear;
5027 if ($r != $cache->{r}) {
5028 %{$cache->{data}} = ();
5029 $cache->{r} = $r;
5030 }
5031 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
5032 wantarray ? (\%dirents, $r, $props) : \%dirents;
5033}
5034sub DESTROY {
5036 # do not call the real DESTROY since we store ourselves in $RA
5037}
5038# get_log(paths, start, end, limit,
5040# discover_changed_paths, strict_node_history, receiver)
5041sub get_log {
5042 my ($self, @args) = @_;
5043 my $pool = SVN::Pool->new;
5044 # svn_log_changed_path_t objects passed to get_log are likely to be
5046 # overwritten even if only the refs are copied to an external variable,
5047 # so we should dup the structures in their entirety. Using an
5048 # externally passed pool (instead of our temporary and quickly cleared
5049 # pool in Git::SVN::Ra) does not help matters at all...
5050 my $receiver = pop @args;
5051 my $prefix = "/".$self->{svn_path};
5052 $prefix =~ s#/+($)##;
5053 my $prefix_regex = qr#^\Q$prefix\E#;
5054 push(@args, sub {
5055 my ($paths) = $_[0];
5056 return &$receiver(@_) unless $paths;
5057 $_[0] = ();
5058 foreach my $p (keys %$paths) {
5059 my $i = $paths->{$p};
5060 # Make path relative to our url, not repos_root
5061 $p =~ s/$prefix_regex//;
5062 my %s = map { $_ => $i->$_; }
5063 qw/copyfrom_path copyfrom_rev action/;
5064 if ($s{'copyfrom_path'}) {
5065 $s{'copyfrom_path'} =~ s/$prefix_regex//;
5066 }
5067 $_[0]{$p} = \%s;
5068 }
5069 &$receiver(@_);
5070 });
5071 # the limit parameter was not supported in SVN 1.1.x, so we
5074 # drop it. Therefore, the receiver callback passed to it
5075 # is made aware of this limitation by being wrapped if
5076 # the limit passed to is being wrapped.
5077 if ($SVN::Core::VERSION le '1.2.0') {
5078 my $limit = splice(@args, 3, 1);
5079 if ($limit > 0) {
5080 my $receiver = pop @args;
5081 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
5082 }
5083 }
5084 my $ret = $self->SUPER::get_log(@args, $pool);
5085 $pool->clear;
5086 $ret;
5087}
5088sub trees_match {
5090 my ($self, $url1, $rev1, $url2, $rev2) = @_;
5091 my $ctx = SVN::Client->new(auth => _auth_providers);
5092 my $out = IO::File->new_tmpfile;
5093 # older SVN (1.1.x) doesn't take $pool as the last parameter for
5095 # $ctx->diff(), so we'll create a default one
5096 my $pool = SVN::Pool->new_default_sub;
5097 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
5099 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
5100 $out->flush;
5101 my $ret = (($out->stat)[7] == 0);
5102 close $out or croak $!;
5103 $ret;
5105}
5106sub get_commit_editor {
5108 my ($self, $log, $cb, $pool) = @_;
5109 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
5110 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
5111}
5112sub gs_do_update {
5114 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
5115 my $new = ($rev_a == $rev_b);
5116 my $path = $gs->{path};
5117 if ($new && -e $gs->{index}) {
5119 unlink $gs->{index} or die
5120 "Couldn't unlink index: $gs->{index}: $!\n";
5121 }
5122 my $pool = SVN::Pool->new;
5123 $editor->set_path_strip($path);
5124 my (@pc) = split m#/#, $path;
5125 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
5126 1, $editor, $pool);
5127 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5128 # Since we can't rely on svn_ra_reparent being available, we'll
5130 # just have to do some magic with set_path to make it so
5131 # we only want a partial path.
5132 my $sp = '';
5133 my $final = join('/', @pc);
5134 while (@pc) {
5135 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
5136 $sp .= '/' if length $sp;
5137 $sp .= shift @pc;
5138 }
5139 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
5140 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
5142 $reporter->finish_report($pool);
5144 $pool->clear;
5145 $editor->{git_commit_ok};
5146}
5147# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
5149# svn_ra_reparent didn't work before 1.4)
5150sub gs_do_switch {
5151 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
5152 my $path = $gs->{path};
5153 my $pool = SVN::Pool->new;
5154 my $full_url = $self->{url};
5156 my $old_url = $full_url;
5157 $full_url .= '/' . $path if length $path;
5158 my ($ra, $reparented);
5159 if ($old_url =~ m#^svn(\+ssh)?://# ||
5161 ($full_url =~ m#^https?://# &&
5162 escape_url($full_url) ne $full_url)) {
5163 $_[0] = undef;
5164 $self = undef;
5165 $RA = undef;
5166 $ra = Git::SVN::Ra->new($full_url);
5167 $ra_invalid = 1;
5168 } elsif ($old_url ne $full_url) {
5169 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
5170 $self->{url} = $full_url;
5171 $reparented = 1;
5172 }
5173 $ra ||= $self;
5175 $url_b = escape_url($url_b);
5176 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
5177 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
5178 $reporter->set_path('', $rev_a, 0, @lock, $pool);
5179 $reporter->finish_report($pool);
5180 if ($reparented) {
5182 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
5183 $self->{url} = $old_url;
5184 }
5185 $pool->clear;
5187 $editor->{git_commit_ok};
5188}
5189sub longest_common_path {
5191 my ($gsv, $globs) = @_;
5192 my %common;
5193 my $common_max = scalar @$gsv;
5194 foreach my $gs (@$gsv) {
5196 my @tmp = split m#/#, $gs->{path};
5197 my $p = '';
5198 foreach (@tmp) {
5199 $p .= length($p) ? "/$_" : $_;
5200 $common{$p} ||= 0;
5201 $common{$p}++;
5202 }
5203 }
5204 $globs ||= [];
5205 $common_max += scalar @$globs;
5206 foreach my $glob (@$globs) {
5207 my @tmp = split m#/#, $glob->{path}->{left};
5208 my $p = '';
5209 foreach (@tmp) {
5210 $p .= length($p) ? "/$_" : $_;
5211 $common{$p} ||= 0;
5212 $common{$p}++;
5213 }
5214 }
5215 my $longest_path = '';
5217 foreach (sort {length $b <=> length $a} keys %common) {
5218 if ($common{$_} == $common_max) {
5219 $longest_path = $_;
5220 last;
5221 }
5222 }
5223 $longest_path;
5224}
5225sub gs_fetch_loop_common {
5227 my ($self, $base, $head, $gsv, $globs) = @_;
5228 return if ($base > $head);
5229 my $inc = $_log_window_size;
5230 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
5231 my $longest_path = longest_common_path($gsv, $globs);
5232 my $ra_url = $self->{url};
5233 my $find_trailing_edge;
5234 while (1) {
5235 my %revs;
5236 my $err;
5237 my $err_handler = $SVN::Error::handler;
5238 $SVN::Error::handler = sub {
5239 ($err) = @_;
5240 skip_unknown_revs($err);
5241 };
5242 sub _cb {
5243 my ($paths, $r, $author, $date, $log) = @_;
5244 [ $paths,
5245 { author => $author, date => $date, log => $log } ];
5246 }
5247 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
5248 sub { $revs{$_[1]} = _cb(@_) });
5249 if ($err) {
5250 print "Checked through r$max\r";
5251 } else {
5252 $find_trailing_edge = 1;
5253 }
5254 if ($err and $find_trailing_edge) {
5255 print STDERR "Path '$longest_path' ",
5256 "was probably deleted:\n",
5257 $err->expanded_message,
5258 "\nWill attempt to follow ",
5259 "revisions r$min .. r$max ",
5260 "committed before the deletion\n";
5261 my $hi = $max;
5262 while (--$hi >= $min) {
5263 my $ok;
5264 $self->get_log([$longest_path], $min, $hi,
5265 0, 1, 1, sub {
5266 $ok = $_[1];
5267 $revs{$_[1]} = _cb(@_) });
5268 if ($ok) {
5269 print STDERR "r$min .. r$ok OK\n";
5270 last;
5271 }
5272 }
5273 $find_trailing_edge = 0;
5274 }
5275 $SVN::Error::handler = $err_handler;
5276 my %exists = map { $_->{path} => $_ } @$gsv;
5278 foreach my $r (sort {$a <=> $b} keys %revs) {
5279 my ($paths, $logged) = @{$revs{$r}};
5280 foreach my $gs ($self->match_globs(\%exists, $paths,
5282 $globs, $r)) {
5283 if ($gs->rev_map_max >= $r) {
5284 next;
5285 }
5286 next unless $gs->match_paths($paths, $r);
5287 $gs->{logged_rev_props} = $logged;
5288 if (my $last_commit = $gs->last_commit) {
5289 $gs->assert_index_clean($last_commit);
5290 }
5291 my $log_entry = $gs->do_fetch($paths, $r);
5292 if ($log_entry) {
5293 $gs->do_git_commit($log_entry);
5294 }
5295 $INDEX_FILES{$gs->{index}} = 1;
5296 }
5297 foreach my $g (@$globs) {
5298 my $k = "svn-remote.$g->{remote}." .
5299 "$g->{t}-maxRev";
5300 Git::SVN::tmp_config($k, $r);
5301 }
5302 if ($ra_invalid) {
5303 $_[0] = undef;
5304 $self = undef;
5305 $RA = undef;
5306 $self = Git::SVN::Ra->new($ra_url);
5307 $ra_invalid = undef;
5308 }
5309 }
5310 # pre-fill the .rev_db since it'll eventually get filled in
5311 # with '0' x40 if something new gets committed
5312 foreach my $gs (@$gsv) {
5313 next if $gs->rev_map_max >= $max;
5314 next if defined $gs->rev_map_get($max);
5315 $gs->rev_map_set($max, 0 x40);
5316 }
5317 foreach my $g (@$globs) {
5318 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
5319 Git::SVN::tmp_config($k, $max);
5320 }
5321 last if $max >= $head;
5322 $min = $max + 1;
5323 $max += $inc;
5324 $max = $head if ($max > $head);
5325 }
5326 Git::SVN::gc();
5327}
5328sub get_dir_globbed {
5330 my ($self, $left, $depth, $r) = @_;
5331 my @x = eval { $self->get_dir($left, $r) };
5333 return unless scalar @x == 3;
5334 my $dirents = $x[0];
5335 my @finalents;
5336 foreach my $de (keys %$dirents) {
5337 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
5338 if ($depth > 1) {
5339 my @args = ("$left/$de", $depth - 1, $r);
5340 foreach my $dir ($self->get_dir_globbed(@args)) {
5341 push @finalents, "$de/$dir";
5342 }
5343 } else {
5344 push @finalents, $de;
5345 }
5346 }
5347 @finalents;
5348}
5349sub match_globs {
5351 my ($self, $exists, $paths, $globs, $r) = @_;
5352 sub get_dir_check {
5354 my ($self, $exists, $g, $r) = @_;
5355 my @dirs = $self->get_dir_globbed($g->{path}->{left},
5357 $g->{path}->{depth},
5358 $r);
5359 foreach my $de (@dirs) {
5361 my $p = $g->{path}->full_path($de);
5362 next if $exists->{$p};
5363 next if (length $g->{path}->{right} &&
5364 ($self->check_path($p, $r) !=
5365 $SVN::Node::dir));
5366 next unless $p =~ /$g->{path}->{regex}/;
5367 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
5368 $g->{ref}->full_path($de), 1);
5369 }
5370 }
5371 foreach my $g (@$globs) {
5372 if (my $path = $paths->{"/$g->{path}->{left}"}) {
5373 if ($path->{action} =~ /^[AR]$/) {
5374 get_dir_check($self, $exists, $g, $r);
5375 }
5376 }
5377 foreach (keys %$paths) {
5378 if (/$g->{path}->{left_regex}/ &&
5379 !/$g->{path}->{regex}/) {
5380 next if $paths->{$_}->{action} !~ /^[AR]$/;
5381 get_dir_check($self, $exists, $g, $r);
5382 }
5383 next unless /$g->{path}->{regex}/;
5384 my $p = $1;
5385 my $pathname = $g->{path}->full_path($p);
5386 next if $exists->{$pathname};
5387 next if ($self->check_path($pathname, $r) !=
5388 $SVN::Node::dir);
5389 $exists->{$pathname} = Git::SVN->init(
5390 $self->{url}, $pathname, undef,
5391 $g->{ref}->full_path($p), 1);
5392 }
5393 my $c = '';
5394 foreach (split m#/#, $g->{path}->{left}) {
5395 $c .= "/$_";
5396 next unless ($paths->{$c} &&
5397 ($paths->{$c}->{action} =~ /^[AR]$/));
5398 get_dir_check($self, $exists, $g, $r);
5399 }
5400 }
5401 values %$exists;
5402}
5403sub minimize_url {
5405 my ($self) = @_;
5406 return $self->{url} if ($self->{url} eq $self->{repos_root});
5407 my $url = $self->{repos_root};
5408 my @components = split(m!/!, $self->{svn_path});
5409 my $c = '';
5410 do {
5411 $url .= "/$c" if length $c;
5412 eval {
5413 my $ra = (ref $self)->new($url);
5414 my $latest = $ra->get_latest_revnum;
5415 $ra->get_log("", $latest, 0, 1, 0, 1, sub {});
5416 };
5417 } while ($@ && ($c = shift @components));
5418 $url;
5419}
5420sub can_do_switch {
5422 my $self = shift;
5423 unless (defined $can_do_switch) {
5424 my $pool = SVN::Pool->new;
5425 my $rep = eval {
5426 $self->do_switch(1, '', 0, $self->{url},
5427 SVN::Delta::Editor->new, $pool);
5428 };
5429 if ($@) {
5430 $can_do_switch = 0;
5431 } else {
5432 $rep->abort_report($pool);
5433 $can_do_switch = 1;
5434 }
5435 $pool->clear;
5436 }
5437 $can_do_switch;
5438}
5439sub skip_unknown_revs {
5441 my ($err) = @_;
5442 my $errno = $err->apr_err();
5443 # Maybe the branch we're tracking didn't
5444 # exist when the repo started, so it's
5445 # not an error if it doesn't, just continue
5446 #
5447 # Wonderfully consistent library, eh?
5448 # 160013 - svn:// and file://
5449 # 175002 - http(s)://
5450 # 175007 - http(s):// (this repo required authorization, too...)
5451 # More codes may be discovered later...
5452 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
5453 my $err_key = $err->expanded_message;
5454 # revision numbers change every time, filter them out
5455 $err_key =~ s/\d+/\0/g;
5456 $err_key = "$errno\0$err_key";
5457 unless ($ignored_err{$err_key}) {
5458 warn "W: Ignoring error from SVN, path probably ",
5459 "does not exist: ($errno): ",
5460 $err->expanded_message,"\n";
5461 warn "W: Do not be alarmed at the above message ",
5462 "git-svn is just searching aggressively for ",
5463 "old history.\n",
5464 "This may take a while on large repositories\n";
5465 $ignored_err{$err_key} = 1;
5466 }
5467 return;
5468 }
5469 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
5470}
5471package Git::SVN::Log;
5473use strict;
5474use warnings;
5475use POSIX qw/strftime/;
5476use Time::Local;
5477use constant commit_log_separator => ('-' x 72) . "\n";
5478use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
5479 %rusers $show_commit $incremental/;
5480my $l_fmt;
5481sub cmt_showable {
5483 my ($c) = @_;
5484 return 1 if defined $c->{r};
5485 # big commit message got truncated by the 16k pretty buffer in rev-list
5487 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
5488 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
5489 @{$c->{l}} = ();
5490 my @log = command(qw/cat-file commit/, $c->{c});
5491 # shift off the headers
5493 shift @log while ($log[0] ne '');
5494 shift @log;
5495 # TODO: make $c->{l} not have a trailing newline in the future
5497 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
5498 (undef, $c->{r}, undef) = ::extract_metadata(
5500 (grep(/^git-svn-id: /, @log))[-1]);
5501 }
5502 return defined $c->{r};
5503}
5504sub log_use_color {
5506 return $color || Git->repository->get_colorbool('color.diff');
5507}
5508sub git_svn_log_cmd {
5510 my ($r_min, $r_max, @args) = @_;
5511 my $head = 'HEAD';
5512 my (@files, @log_opts);
5513 foreach my $x (@args) {
5514 if ($x eq '--' || @files) {
5515 push @files, $x;
5516 } else {
5517 if (::verify_ref("$x^0")) {
5518 $head = $x;
5519 } else {
5520 push @log_opts, $x;
5521 }
5522 }
5523 }
5524 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
5526 $gs ||= Git::SVN->_new;
5527 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
5528 $gs->refname);
5529 push @cmd, '-r' unless $non_recursive;
5530 push @cmd, qw/--raw --name-status/ if $verbose;
5531 push @cmd, '--color' if log_use_color();
5532 push @cmd, @log_opts;
5533 if (defined $r_max && $r_max == $r_min) {
5534 push @cmd, '--max-count=1';
5535 if (my $c = $gs->rev_map_get($r_max)) {
5536 push @cmd, $c;
5537 }
5538 } elsif (defined $r_max) {
5539 if ($r_max < $r_min) {
5540 ($r_min, $r_max) = ($r_max, $r_min);
5541 }
5542 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
5543 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
5544 # If there are no commits in the range, both $c_max and $c_min
5545 # will be undefined. If there is at least 1 commit in the
5546 # range, both will be defined.
5547 return () if !defined $c_min || !defined $c_max;
5548 if ($c_min eq $c_max) {
5549 push @cmd, '--max-count=1', $c_min;
5550 } else {
5551 push @cmd, '--boundary', "$c_min..$c_max";
5552 }
5553 }
5554 return (@cmd, @files);
5555}
5556# adapted from pager.c
5558sub config_pager {
5559 if (! -t *STDOUT) {
5560 $ENV{GIT_PAGER_IN_USE} = 'false';
5561 $pager = undef;
5562 return;
5563 }
5564 chomp($pager = command_oneline(qw(var GIT_PAGER)));
5565 if ($pager eq 'cat') {
5566 $pager = undef;
5567 }
5568 $ENV{GIT_PAGER_IN_USE} = defined($pager);
5569}
5570sub run_pager {
5572 return unless defined $pager;
5573 pipe my ($rfd, $wfd) or return;
5574 defined(my $pid = fork) or ::fatal "Can't fork: $!";
5575 if (!$pid) {
5576 open STDOUT, '>&', $wfd or
5577 ::fatal "Can't redirect to stdout: $!";
5578 return;
5579 }
5580 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
5581 $ENV{LESS} ||= 'FRSX';
5582 exec $pager or ::fatal "Can't run pager: $! ($pager)";
5583}
5584sub format_svn_date {
5586 # some systmes don't handle or mishandle %z, so be creative.
5587 my $t = shift || time;
5588 my $gm = timelocal(gmtime($t));
5589 my $sign = qw( + + - )[ $t <=> $gm ];
5590 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
5591 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
5592}
5593sub parse_git_date {
5595 my ($t, $tz) = @_;
5596 # Date::Parse isn't in the standard Perl distro :(
5597 if ($tz =~ s/^\+//) {
5598 $t += tz_to_s_offset($tz);
5599 } elsif ($tz =~ s/^\-//) {
5600 $t -= tz_to_s_offset($tz);
5601 }
5602 return $t;
5603}
5604sub set_local_timezone {
5606 if (defined $TZ) {
5607 $ENV{TZ} = $TZ;
5608 } else {
5609 delete $ENV{TZ};
5610 }
5611}
5612sub tz_to_s_offset {
5614 my ($tz) = @_;
5615 $tz =~ s/(\d\d)$//;
5616 return ($1 * 60) + ($tz * 3600);
5617}
5618sub get_author_info {
5620 my ($dest, $author, $t, $tz) = @_;
5621 $author =~ s/(?:^\s*|\s*$)//g;
5622 $dest->{a_raw} = $author;
5623 my $au;
5624 if ($::_authors) {
5625 $au = $rusers{$author} || undef;
5626 }
5627 if (!$au) {
5628 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5629 }
5630 $dest->{t} = $t;
5631 $dest->{tz} = $tz;
5632 $dest->{a} = $au;
5633 $dest->{t_utc} = parse_git_date($t, $tz);
5634}
5635sub process_commit {
5637 my ($c, $r_min, $r_max, $defer) = @_;
5638 if (defined $r_min && defined $r_max) {
5639 if ($r_min == $c->{r} && $r_min == $r_max) {
5640 show_commit($c);
5641 return 0;
5642 }
5643 return 1 if $r_min == $r_max;
5644 if ($r_min < $r_max) {
5645 # we need to reverse the print order
5646 return 0 if (defined $limit && --$limit < 0);
5647 push @$defer, $c;
5648 return 1;
5649 }
5650 if ($r_min != $r_max) {
5651 return 1 if ($r_min < $c->{r});
5652 return 1 if ($r_max > $c->{r});
5653 }
5654 }
5655 return 0 if (defined $limit && --$limit < 0);
5656 show_commit($c);
5657 return 1;
5658}
5659sub show_commit {
5661 my $c = shift;
5662 if ($oneline) {
5663 my $x = "\n";
5664 if (my $l = $c->{l}) {
5665 while ($l->[0] =~ /^\s*$/) { shift @$l }
5666 $x = $l->[0];
5667 }
5668 $l_fmt ||= 'A' . length($c->{r});
5669 print 'r',pack($l_fmt, $c->{r}),' | ';
5670 print "$c->{c} | " if $show_commit;
5671 print $x;
5672 } else {
5673 show_commit_normal($c);
5674 }
5675}
5676sub show_commit_changed_paths {
5678 my ($c) = @_;
5679 return unless $c->{changed};
5680 print "Changed paths:\n", @{$c->{changed}};
5681}
5682sub show_commit_normal {
5684 my ($c) = @_;
5685 print commit_log_separator, "r$c->{r} | ";
5686 print "$c->{c} | " if $show_commit;
5687 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5688 my $nr_line = 0;
5689 if (my $l = $c->{l}) {
5691 while ($l->[$#$l] eq "\n" && $#$l > 0
5692 && $l->[($#$l - 1)] eq "\n") {
5693 pop @$l;
5694 }
5695 $nr_line = scalar @$l;
5696 if (!$nr_line) {
5697 print "1 line\n\n\n";
5698 } else {
5699 if ($nr_line == 1) {
5700 $nr_line = '1 line';
5701 } else {
5702 $nr_line .= ' lines';
5703 }
5704 print $nr_line, "\n";
5705 show_commit_changed_paths($c);
5706 print "\n";
5707 print $_ foreach @$l;
5708 }
5709 } else {
5710 print "1 line\n";
5711 show_commit_changed_paths($c);
5712 print "\n";
5713 }
5715 foreach my $x (qw/raw stat diff/) {
5716 if ($c->{$x}) {
5717 print "\n";
5718 print $_ foreach @{$c->{$x}}
5719 }
5720 }
5721}
5722sub cmd_show_log {
5724 my (@args) = @_;
5725 my ($r_min, $r_max);
5726 my $r_last = -1; # prevent dupes
5727 set_local_timezone();
5728 if (defined $::_revision) {
5729 if ($::_revision =~ /^(\d+):(\d+)$/) {
5730 ($r_min, $r_max) = ($1, $2);
5731 } elsif ($::_revision =~ /^\d+$/) {
5732 $r_min = $r_max = $::_revision;
5733 } else {
5734 ::fatal "-r$::_revision is not supported, use ",
5735 "standard 'git log' arguments instead";
5736 }
5737 }
5738 config_pager();
5740 @args = git_svn_log_cmd($r_min, $r_max, @args);
5741 if (!@args) {
5742 print commit_log_separator unless $incremental || $oneline;
5743 return;
5744 }
5745 my $log = command_output_pipe(@args);
5746 run_pager();
5747 my (@k, $c, $d, $stat);
5748 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5749 while (<$log>) {
5750 if (/^${esc_color}commit (- )?($::sha1_short)/o) {
5751 my $cmt = $1;
5752 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5753 $r_last = $c->{r};
5754 process_commit($c, $r_min, $r_max, \@k) or
5755 goto out;
5756 }
5757 $d = undef;
5758 $c = { c => $cmt };
5759 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5760 get_author_info($c, $1, $2, $3);
5761 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5762 # ignore
5763 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5764 push @{$c->{raw}}, $_;
5765 } elsif (/^${esc_color}[ACRMDT]\t/) {
5766 # we could add $SVN->{svn_path} here, but that requires
5767 # remote access at the moment (repo_path_split)...
5768 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5769 push @{$c->{changed}}, $_;
5770 } elsif (/^${esc_color}diff /o) {
5771 $d = 1;
5772 push @{$c->{diff}}, $_;
5773 } elsif ($d) {
5774 push @{$c->{diff}}, $_;
5775 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5776 $esc_color*[\+\-]*$esc_color$/x) {
5777 $stat = 1;
5778 push @{$c->{stat}}, $_;
5779 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5780 push @{$c->{stat}}, $_;
5781 $stat = undef;
5782 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5783 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5784 } elsif (s/^${esc_color} //o) {
5785 push @{$c->{l}}, $_;
5786 }
5787 }
5788 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5789 $r_last = $c->{r};
5790 process_commit($c, $r_min, $r_max, \@k);
5791 }
5792 if (@k) {
5793 ($r_min, $r_max) = ($r_max, $r_min);
5794 process_commit($_, $r_min, $r_max) foreach reverse @k;
5795 }
5796out:
5797 close $log;
5798 print commit_log_separator unless $incremental || $oneline;
5799}
5800sub cmd_blame {
5802 my $path = pop;
5803 config_pager();
5805 run_pager();
5806 my ($fh, $ctx, $rev);
5808 if ($_git_format) {
5810 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5811 while (my $line = <$fh>) {
5812 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5813 # Uncommitted edits show up as a rev ID of
5814 # all zeros, which we can't look up with
5815 # cmt_metadata
5816 if ($1 !~ /^0+$/) {
5817 (undef, $rev, undef) =
5818 ::cmt_metadata($1);
5819 $rev = '0' if (!$rev);
5820 } else {
5821 $rev = '0';
5822 }
5823 $rev = sprintf('%-10s', $rev);
5824 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5825 }
5826 print $line;
5827 }
5828 } else {
5829 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5830 '--', $path);
5831 my ($sha1);
5832 my %authors;
5833 my @buffer;
5834 my %dsha; #distinct sha keys
5835 while (my $line = <$fh>) {
5837 push @buffer, $line;
5838 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5839 $dsha{$1} = 1;
5840 }
5841 }
5842 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5844 foreach my $line (@buffer) {
5846 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5847 $rev = $s2r->{$1};
5848 $rev = '0' if (!$rev)
5849 }
5850 elsif ($line =~ /^author (.*)/) {
5851 $authors{$rev} = $1;
5852 $authors{$rev} =~ s/\s/_/g;
5853 }
5854 elsif ($line =~ /^\t(.*)$/) {
5855 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5856 }
5857 }
5858 }
5859 command_close_pipe($fh, $ctx);
5860}
5861package Git::SVN::Migration;
5863# these version numbers do NOT correspond to actual version numbers
5864# of git nor git-svn. They are just relative.
5865#
5866# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5867#
5868# v1 layout: .git/$id/info/url, refs/remotes/$id
5869#
5870# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5871#
5872# v3 layout: .git/svn/$id, refs/remotes/$id
5873# - info/url may remain for backwards compatibility
5874# - this is what we migrate up to this layout automatically,
5875# - this will be used by git svn init on single branches
5876# v3.1 layout (auto migrated):
5877# - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5878# for backwards compatibility
5879#
5880# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5881# - this is only created for newly multi-init-ed
5882# repositories. Similar in spirit to the
5883# --use-separate-remotes option in git-clone (now default)
5884# - we do not automatically migrate to this (following
5885# the example set by core git)
5886#
5887# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5888# - newer, more-efficient format that uses 24-bytes per record
5889# with no filler space.
5890# - use xxd -c24 < .rev_map.$UUID to view and debug
5891# - This is a one-way migration, repositories updated to the
5892# new format will not be able to use old git-svn without
5893# rebuilding the .rev_db. Rebuilding the rev_db is not
5894# possible if noMetadata or useSvmProps are set; but should
5895# be no problem for users that use the (sensible) defaults.
5896use strict;
5897use warnings;
5898use Carp qw/croak/;
5899use File::Path qw/mkpath/;
5900use File::Basename qw/dirname basename/;
5901use vars qw/$_minimize/;
5902sub migrate_from_v0 {
5904 my $git_dir = $ENV{GIT_DIR};
5905 return undef unless -d $git_dir;
5906 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5907 my $migrated = 0;
5908 while (<$fh>) {
5909 chomp;
5910 my ($id, $orig_ref) = ($_, $_);
5911 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5912 next unless -f "$git_dir/$id/info/url";
5913 my $new_ref = "refs/remotes/$id";
5914 if (::verify_ref("$new_ref^0")) {
5915 print STDERR "W: $orig_ref is probably an old ",
5916 "branch used by an ancient version of ",
5917 "git-svn.\n",
5918 "However, $new_ref also exists.\n",
5919 "We will not be able ",
5920 "to use this branch until this ",
5921 "ambiguity is resolved.\n";
5922 next;
5923 }
5924 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5925 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5926 command_noisy('update-ref', $new_ref, $orig_ref);
5927 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5928 $migrated++;
5929 }
5930 command_close_pipe($fh, $ctx);
5931 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5932 $migrated;
5933}
5934sub migrate_from_v1 {
5936 my $git_dir = $ENV{GIT_DIR};
5937 my $migrated = 0;
5938 return $migrated unless -d $git_dir;
5939 my $svn_dir = "$git_dir/svn";
5940 # just in case somebody used 'svn' as their $id at some point...
5942 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5943 print STDERR "Migrating from a git-svn v1 layout...\n";
5945 mkpath([$svn_dir]);
5946 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5947 "$svn_dir\n\t(required for this version ",
5948 "($::VERSION) of git-svn) does not exist.\n";
5949 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5950 while (<$fh>) {
5951 my $x = $_;
5952 next unless $x =~ s#^refs/remotes/##;
5953 chomp $x;
5954 next unless -f "$git_dir/$x/info/url";
5955 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5956 next unless $u;
5957 my $dn = dirname("$git_dir/svn/$x");
5958 mkpath([$dn]) unless -d $dn;
5959 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5960 mkpath(["$git_dir/svn/svn"]);
5961 print STDERR " - $git_dir/$x/info => ",
5962 "$git_dir/svn/$x/info\n";
5963 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5964 croak "$!: $x";
5965 # don't worry too much about these, they probably
5966 # don't exist with repos this old (save for index,
5967 # and we can easily regenerate that)
5968 foreach my $f (qw/unhandled.log index .rev_db/) {
5969 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5970 }
5971 } else {
5972 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5973 rename "$git_dir/$x", "$git_dir/svn/$x" or
5974 croak "$!: $x";
5975 }
5976 $migrated++;
5977 }
5978 command_close_pipe($fh, $ctx);
5979 print STDERR "Done migrating from a git-svn v1 layout\n";
5980 $migrated;
5981}
5982sub read_old_urls {
5984 my ($l_map, $pfx, $path) = @_;
5985 my @dir;
5986 foreach (<$path/*>) {
5987 if (-r "$_/info/url") {
5988 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5989 my $ref_id = $pfx . basename $_;
5990 my $url = ::file_to_s("$_/info/url");
5991 $l_map->{$ref_id} = $url;
5992 } elsif (-d $_) {
5993 push @dir, $_;
5994 }
5995 }
5996 foreach (@dir) {
5997 my $x = $_;
5998 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5999 read_old_urls($l_map, $x, $_);
6000 }
6001}
6002sub migrate_from_v2 {
6004 my @cfg = command(qw/config -l/);
6005 return if grep /^svn-remote\..+\.url=/, @cfg;
6006 my %l_map;
6007 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
6008 my $migrated = 0;
6009 foreach my $ref_id (sort keys %l_map) {
6011 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
6012 if ($@) {
6013 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
6014 }
6015 $migrated++;
6016 }
6017 $migrated;
6018}
6019sub minimize_connections {
6021 my $r = Git::SVN::read_all_remotes();
6022 my $new_urls = {};
6023 my $root_repos = {};
6024 foreach my $repo_id (keys %$r) {
6025 my $url = $r->{$repo_id}->{url} or next;
6026 my $fetch = $r->{$repo_id}->{fetch} or next;
6027 my $ra = Git::SVN::Ra->new($url);
6028 # skip existing cases where we already connect to the root
6030 if (($ra->{url} eq $ra->{repos_root}) ||
6031 ($ra->{repos_root} eq $repo_id)) {
6032 $root_repos->{$ra->{url}} = $repo_id;
6033 next;
6034 }
6035 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
6037 my $root_path = $ra->{url};
6038 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
6039 foreach my $path (keys %$fetch) {
6040 my $ref_id = $fetch->{$path};
6041 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
6042 # make sure we can read when connecting to
6044 # a higher level of a repository
6045 my ($last_rev, undef) = $gs->last_rev_commit;
6046 if (!defined $last_rev) {
6047 $last_rev = eval {
6048 $root_ra->get_latest_revnum;
6049 };
6050 next if $@;
6051 }
6052 my $new = $root_path;
6053 $new .= length $path ? "/$path" : '';
6054 eval {
6055 $root_ra->get_log([$new], $last_rev, $last_rev,
6056 0, 0, 1, sub { });
6057 };
6058 next if $@;
6059 $new_urls->{$ra->{repos_root}}->{$new} =
6060 { ref_id => $ref_id,
6061 old_repo_id => $repo_id,
6062 old_path => $path };
6063 }
6064 }
6065 my @emptied;
6067 foreach my $url (keys %$new_urls) {
6068 # see if we can re-use an existing [svn-remote "repo_id"]
6069 # instead of creating a(n ugly) new section:
6070 my $repo_id = $root_repos->{$url} || $url;
6071 my $fetch = $new_urls->{$url};
6073 foreach my $path (keys %$fetch) {
6074 my $x = $fetch->{$path};
6075 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
6076 my $pfx = "svn-remote.$x->{old_repo_id}";
6077 my $old_fetch = quotemeta("$x->{old_path}:".
6079 "$x->{ref_id}");
6080 command_noisy(qw/config --unset/,
6081 "$pfx.fetch", '^'. $old_fetch . '$');
6082 delete $r->{$x->{old_repo_id}}->
6083 {fetch}->{$x->{old_path}};
6084 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
6085 command_noisy(qw/config --unset/,
6086 "$pfx.url");
6087 push @emptied, $x->{old_repo_id}
6088 }
6089 }
6090 }
6091 if (@emptied) {
6092 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
6093 print STDERR <<EOF;
6094The following [svn-remote] sections in your config file ($file) are empty
6095and can be safely removed:
6096EOF
6097 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
6098 }
6099}
6100sub migration_check {
6102 migrate_from_v0();
6103 migrate_from_v1();
6104 migrate_from_v2();
6105 minimize_connections() if $_minimize;
6106}
6107package Git::IndexInfo;
6109use strict;
6110use warnings;
6111use Git qw/command_input_pipe command_close_pipe/;
6112sub new {
6114 my ($class) = @_;
6115 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
6116 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
6117}
6118sub remove {
6120 my ($self, $path) = @_;
6121 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
6122 return ++$self->{nr};
6123 }
6124 undef;
6125}
6126sub update {
6128 my ($self, $mode, $hash, $path) = @_;
6129 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
6130 return ++$self->{nr};
6131 }
6132 undef;
6133}
6134sub DESTROY {
6136 my ($self) = @_;
6137 command_close_pipe($self->{gui}, $self->{ctx});
6138}
6139package Git::SVN::GlobSpec;
6141use strict;
6142use warnings;
6143sub new {
6145 my ($class, $glob, $pattern_ok) = @_;
6146 my $re = $glob;
6147 $re =~ s!/+$!!g; # no need for trailing slashes
6148 my (@left, @right, @patterns);
6149 my $state = "left";
6150 my $die_msg = "Only one set of wildcard directories " .
6151 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
6152 for my $part (split(m|/|, $glob)) {
6153 if ($part =~ /\*/ && $part ne "*") {
6154 die "Invalid pattern in '$glob': $part\n";
6155 } elsif ($pattern_ok && $part =~ /[{}]/ &&
6156 $part !~ /^\{[^{}]+\}/) {
6157 die "Invalid pattern in '$glob': $part\n";
6158 }
6159 if ($part eq "*") {
6160 die $die_msg if $state eq "right";
6161 $state = "pattern";
6162 push(@patterns, "[^/]*");
6163 } elsif ($pattern_ok && $part =~ /^\{(.*)\}$/) {
6164 die $die_msg if $state eq "right";
6165 $state = "pattern";
6166 my $p = quotemeta($1);
6167 $p =~ s/\\,/|/g;
6168 push(@patterns, "(?:$p)");
6169 } else {
6170 if ($state eq "left") {
6171 push(@left, $part);
6172 } else {
6173 push(@right, $part);
6174 $state = "right";
6175 }
6176 }
6177 }
6178 my $depth = @patterns;
6179 if ($depth == 0) {
6180 die "One '*' is needed in glob: '$glob'\n";
6181 }
6182 my $left = join('/', @left);
6183 my $right = join('/', @right);
6184 $re = join('/', @patterns);
6185 $re = join('\/',
6186 grep(length, quotemeta($left), "($re)", quotemeta($right)));
6187 my $left_re = qr/^\/\Q$left\E(\/|$)/;
6188 bless { left => $left, right => $right, left_regex => $left_re,
6189 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
6190}
6191sub full_path {
6193 my ($self, $path) = @_;
6194 return (length $self->{left} ? "$self->{left}/" : '') .
6195 $path . (length $self->{right} ? "/$self->{right}" : '');
6196}
6197__END__
6199Data structures:
6201$remotes = { # returned by read_all_remotes()
6204 'svn' => {
6205 # svn-remote.svn.url=https://svn.musicpd.org
6206 url => 'https://svn.musicpd.org',
6207 # svn-remote.svn.fetch=mpd/trunk:trunk
6208 fetch => {
6209 'mpd/trunk' => 'trunk',
6210 },
6211 # svn-remote.svn.tags=mpd/tags/*:tags/*
6212 tags => {
6213 path => {
6214 left => 'mpd/tags',
6215 right => '',
6216 regex => qr!mpd/tags/([^/]+)$!,
6217 glob => 'tags/*',
6218 },
6219 ref => {
6220 left => 'tags',
6221 right => '',
6222 regex => qr!tags/([^/]+)$!,
6223 glob => 'tags/*',
6224 },
6225 }
6226 }
6227};
6228$log_entry hashref as returned by libsvn_log_entry()
6230{
6231 log => 'whitespace-formatted log entry
6232', # trailing newline is preserved
6233 revision => '8', # integer
6234 date => '2004-02-24T17:01:44.108345Z', # commit date
6235 author => 'committer name'
6236};
6237# this is generated by generate_diff();
6240@mods = array of diff-index line hashes, each element represents one line
6241 of diff-index output
6242diff-index line ($m hash)
6244{
6245 mode_a => first column of diff-index output, no leading ':',
6246 mode_b => second column of diff-index output,
6247 sha1_b => sha1sum of the final blob,
6248 chg => change type [MCRADT],
6249 file_a => original file name of a file (iff chg is 'C' or 'R')
6250 file_b => new/current file name of a file (any chg)
6251}
6252;
6253# retval of read_url_paths{,_all}();
6255$l_map = {
6256 # repository root url
6257 'https://svn.musicpd.org' => {
6258 # repository path # GIT_SVN_ID
6259 'mpd/trunk' => 'trunk',
6260 'mpd/tags/0.11.5' => 'tags/0.11.5',
6261 },
6262}
6263Notes:
6265 I don't trust the each() function on unless I created %hash myself
6266 because the internal iterator may not have started at base.