10b77ad34c1103fc8dd390db9937496e41716fdf
1#!/usr/bin/env perl
2# Copyright (C) 2006, Eric Wong <normalperson@yhbt.net>
3# License: GPL v2 or later
4use warnings;
5use strict;
6use vars qw/ $AUTHOR $VERSION
7 $sha1 $sha1_short $_revision $_repository
8 $_q $_authors $_authors_prog %users/;
9$AUTHOR = 'Eric Wong <normalperson@yhbt.net>';
10$VERSION = '@@GIT_VERSION@@';
11# From which subdir have we been invoked?
13my $cmd_dir_prefix = eval {
14 command_oneline([qw/rev-parse --show-prefix/], STDERR => 0)
15} || '';
16my $git_dir_user_set = 1 if defined $ENV{GIT_DIR};
18$ENV{GIT_DIR} ||= '.git';
19$Git::SVN::default_repo_id = 'svn';
20$Git::SVN::default_ref_id = $ENV{GIT_SVN_ID} || 'git-svn';
21$Git::SVN::Ra::_log_window_size = 100;
22$Git::SVN::_minimize_url = 'unset';
23$Git::SVN::Log::TZ = $ENV{TZ};
25$ENV{TZ} = 'UTC';
26$| = 1; # unbuffer STDOUT
27sub fatal (@) { print STDERR "@_\n"; exit 1 }
29require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
30require SVN::Ra;
31require SVN::Delta;
32if ($SVN::Core::VERSION lt '1.1.0') {
33 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
34}
35my $can_compress = eval { require Compress::Zlib; 1};
36push @Git::SVN::Ra::ISA, 'SVN::Ra';
37push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
38push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
39use Carp qw/croak/;
40use Digest::MD5;
41use IO::File qw//;
42use File::Basename qw/dirname basename/;
43use File::Path qw/mkpath/;
44use File::Spec;
45use File::Find;
46use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
47use IPC::Open3;
48use Git;
49BEGIN {
51 # import functions from Git into our packages, en masse
52 no strict 'refs';
53 foreach (qw/command command_oneline command_noisy command_output_pipe
54 command_input_pipe command_close_pipe
55 command_bidi_pipe command_close_bidi_pipe/) {
56 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
57 Git::SVN::Migration Git::SVN::Log Git::SVN),
58 __PACKAGE__) {
59 *{"${package}::$_"} = \&{"Git::$_"};
60 }
61 }
62}
63my ($SVN);
65$sha1 = qr/[a-f\d]{40}/;
67$sha1_short = qr/[a-f\d]{4,40}/;
68my ($_stdin, $_help, $_edit,
69 $_message, $_file, $_branch_dest,
70 $_template, $_shared,
71 $_version, $_fetch_all, $_no_rebase, $_fetch_parent,
72 $_merge, $_strategy, $_dry_run, $_local,
73 $_prefix, $_no_checkout, $_url, $_verbose,
74 $_git_format, $_commit_url, $_tag);
75$Git::SVN::_follow_parent = 1;
76$_q ||= 0;
77my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
78 'config-dir=s' => \$Git::SVN::Ra::config_dir,
79 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache,
80 'ignore-paths=s' => \$SVN::Git::Fetcher::_ignore_regex );
81my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
82 'authors-file|A=s' => \$_authors,
83 'authors-prog=s' => \$_authors_prog,
84 'repack:i' => \$Git::SVN::_repack,
85 'noMetadata' => \$Git::SVN::_no_metadata,
86 'useSvmProps' => \$Git::SVN::_use_svm_props,
87 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
88 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
89 'no-checkout' => \$_no_checkout,
90 'quiet|q+' => \$_q,
91 'repack-flags|repack-args|repack-opts=s' =>
92 \$Git::SVN::_repack_flags,
93 'use-log-author' => \$Git::SVN::_use_log_author,
94 'add-author-from' => \$Git::SVN::_add_author_from,
95 'localtime' => \$Git::SVN::_localtime,
96 %remote_opts );
97my ($_trunk, @_tags, @_branches, $_stdlayout);
99my %icv;
100my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
101 'trunk|T=s' => \$_trunk, 'tags|t=s@' => \@_tags,
102 'branches|b=s@' => \@_branches, 'prefix=s' => \$_prefix,
103 'stdlayout|s' => \$_stdlayout,
104 'minimize-url|m!' => \$Git::SVN::_minimize_url,
105 'no-metadata' => sub { $icv{noMetadata} = 1 },
106 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
107 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
108 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
109 %remote_opts );
110my %cmt_opts = ( 'edit|e' => \$_edit,
111 'rmdir' => \$SVN::Git::Editor::_rmdir,
112 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
113 'l=i' => \$SVN::Git::Editor::_rename_limit,
114 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
115);
116my %cmd = (
118 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
119 { 'revision|r=s' => \$_revision,
120 'fetch-all|all' => \$_fetch_all,
121 'parent|p' => \$_fetch_parent,
122 %fc_opts } ],
123 clone => [ \&cmd_clone, "Initialize and fetch revisions",
124 { 'revision|r=s' => \$_revision,
125 %fc_opts, %init_opts } ],
126 init => [ \&cmd_init, "Initialize a repo for tracking" .
127 " (requires URL argument)",
128 \%init_opts ],
129 'multi-init' => [ \&cmd_multi_init,
130 "Deprecated alias for ".
131 "'$0 init -T<trunk> -b<branches> -t<tags>'",
132 \%init_opts ],
133 dcommit => [ \&cmd_dcommit,
134 'Commit several diffs to merge with upstream',
135 { 'merge|m|M' => \$_merge,
136 'strategy|s=s' => \$_strategy,
137 'verbose|v' => \$_verbose,
138 'dry-run|n' => \$_dry_run,
139 'fetch-all|all' => \$_fetch_all,
140 'commit-url=s' => \$_commit_url,
141 'revision|r=i' => \$_revision,
142 'no-rebase' => \$_no_rebase,
143 %cmt_opts, %fc_opts } ],
144 branch => [ \&cmd_branch,
145 'Create a branch in the SVN repository',
146 { 'message|m=s' => \$_message,
147 'destination|d=s' => \$_branch_dest,
148 'dry-run|n' => \$_dry_run,
149 'tag|t' => \$_tag } ],
150 tag => [ sub { $_tag = 1; cmd_branch(@_) },
151 'Create a tag in the SVN repository',
152 { 'message|m=s' => \$_message,
153 'destination|d=s' => \$_branch_dest,
154 'dry-run|n' => \$_dry_run } ],
155 'set-tree' => [ \&cmd_set_tree,
156 "Set an SVN repository to a git tree-ish",
157 { 'stdin' => \$_stdin, %cmt_opts, %fc_opts, } ],
158 'create-ignore' => [ \&cmd_create_ignore,
159 'Create a .gitignore per svn:ignore',
160 { 'revision|r=i' => \$_revision
161 } ],
162 'propget' => [ \&cmd_propget,
163 'Print the value of a property on a file or directory',
164 { 'revision|r=i' => \$_revision } ],
165 'proplist' => [ \&cmd_proplist,
166 'List all properties of a file or directory',
167 { 'revision|r=i' => \$_revision } ],
168 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
169 { 'revision|r=i' => \$_revision
170 } ],
171 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
172 { 'revision|r=i' => \$_revision
173 } ],
174 'multi-fetch' => [ \&cmd_multi_fetch,
175 "Deprecated alias for $0 fetch --all",
176 { 'revision|r=s' => \$_revision, %fc_opts } ],
177 'migrate' => [ sub { },
178 # no-op, we automatically run this anyways,
179 'Migrate configuration/metadata/layout from
180 previous versions of git-svn',
181 { 'minimize' => \$Git::SVN::Migration::_minimize,
182 %remote_opts } ],
183 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
184 { 'limit=i' => \$Git::SVN::Log::limit,
185 'revision|r=s' => \$_revision,
186 'verbose|v' => \$Git::SVN::Log::verbose,
187 'incremental' => \$Git::SVN::Log::incremental,
188 'oneline' => \$Git::SVN::Log::oneline,
189 'show-commit' => \$Git::SVN::Log::show_commit,
190 'non-recursive' => \$Git::SVN::Log::non_recursive,
191 'authors-file|A=s' => \$_authors,
192 'color' => \$Git::SVN::Log::color,
193 'pager=s' => \$Git::SVN::Log::pager
194 } ],
195 'find-rev' => [ \&cmd_find_rev,
196 "Translate between SVN revision numbers and tree-ish",
197 {} ],
198 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
199 { 'merge|m|M' => \$_merge,
200 'verbose|v' => \$_verbose,
201 'strategy|s=s' => \$_strategy,
202 'local|l' => \$_local,
203 'fetch-all|all' => \$_fetch_all,
204 'dry-run|n' => \$_dry_run,
205 %fc_opts } ],
206 'commit-diff' => [ \&cmd_commit_diff,
207 'Commit a diff between two trees',
208 { 'message|m=s' => \$_message,
209 'file|F=s' => \$_file,
210 'revision|r=s' => \$_revision,
211 %cmt_opts } ],
212 'info' => [ \&cmd_info,
213 "Show info about the latest SVN revision
214 on the current branch",
215 { 'url' => \$_url, } ],
216 'blame' => [ \&Git::SVN::Log::cmd_blame,
217 "Show what revision and author last modified each line of a file",
218 { 'git-format' => \$_git_format } ],
219 'reset' => [ \&cmd_reset,
220 "Undo fetches back to the specified SVN revision",
221 { 'revision|r=s' => \$_revision,
222 'parent|p' => \$_fetch_parent } ],
223 'gc' => [ \&cmd_gc,
224 "Compress unhandled.log files in .git/svn and remove " .
225 "index files in .git/svn",
226 {} ],
227);
228my $cmd;
230for (my $i = 0; $i < @ARGV; $i++) {
231 if (defined $cmd{$ARGV[$i]}) {
232 $cmd = $ARGV[$i];
233 splice @ARGV, $i, 1;
234 last;
235 } elsif ($ARGV[$i] eq 'help') {
236 $cmd = $ARGV[$i+1];
237 usage(0);
238 }
239};
240# make sure we're always running at the top-level working directory
242unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
243 unless (-d $ENV{GIT_DIR}) {
244 if ($git_dir_user_set) {
245 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
246 "but it is not a directory\n";
247 }
248 my $git_dir = delete $ENV{GIT_DIR};
249 my $cdup = undef;
250 git_cmd_try {
251 $cdup = command_oneline(qw/rev-parse --show-cdup/);
252 $git_dir = '.' unless ($cdup);
253 chomp $cdup if ($cdup);
254 $cdup = "." unless ($cdup && length $cdup);
255 } "Already at toplevel, but $git_dir not found\n";
256 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
257 unless (-d $git_dir) {
258 die "$git_dir still not found after going to ",
259 "'$cdup'\n";
260 }
261 $ENV{GIT_DIR} = $git_dir;
262 }
263 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
264}
265my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
267read_repo_config(\%opts);
269if ($cmd && ($cmd eq 'log' || $cmd eq 'blame')) {
270 Getopt::Long::Configure('pass_through');
271}
272my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
273 'minimize-connections' => \$Git::SVN::Migration::_minimize,
274 'id|i=s' => \$Git::SVN::default_ref_id,
275 'svn-remote|remote|R=s' => sub {
276 $Git::SVN::no_reuse_existing = 1;
277 $Git::SVN::default_repo_id = $_[1] });
278exit 1 if (!$rv && $cmd && $cmd ne 'log');
279usage(0) if $_help;
281version() if $_version;
282usage(1) unless defined $cmd;
283load_authors() if $_authors;
284if (defined $_authors_prog) {
285 $_authors_prog = "'" . File::Spec->rel2abs($_authors_prog) . "'";
286}
287unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
289 Git::SVN::Migration::migration_check();
290}
291Git::SVN::init_vars();
292eval {
293 Git::SVN::verify_remotes_sanity();
294 $cmd{$cmd}->[0]->(@ARGV);
295};
296fatal $@ if $@;
297post_fetch_checkout();
298exit 0;
299####################### primary functions ######################
301sub usage {
302 my $exit = shift || 0;
303 my $fd = $exit ? \*STDERR : \*STDOUT;
304 print $fd <<"";
305git-svn - bidirectional operations between a single Subversion tree and git
306Usage: git svn <command> [options] [arguments]\n
307 print $fd "Available commands:\n" unless $cmd;
309 foreach (sort keys %cmd) {
311 next if $cmd && $cmd ne $_;
312 next if /^multi-/; # don't show deprecated commands
313 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
314 foreach (sort keys %{$cmd{$_}->[2]}) {
315 # mixed-case options are for .git/config only
316 next if /[A-Z]/ && /^[a-z]+$/i;
317 # prints out arguments as they should be passed:
318 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
319 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
320 "--$_" : "-$_" }
321 split /\|/,$_)," $x\n";
322 }
323 }
324 print $fd <<"";
325\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
326arbitrary identifier if you're tracking multiple SVN branches/repositories in
327one git repository and want to keep them separate. See git-svn(1) for more
328information.
329 exit $exit;
331}
332sub version {
334 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
335 exit 0;
336}
337sub do_git_init_db {
339 unless (-d $ENV{GIT_DIR}) {
340 my @init_db = ('init');
341 push @init_db, "--template=$_template" if defined $_template;
342 if (defined $_shared) {
343 if ($_shared =~ /[a-z]/) {
344 push @init_db, "--shared=$_shared";
345 } else {
346 push @init_db, "--shared";
347 }
348 }
349 command_noisy(@init_db);
350 $_repository = Git->repository(Repository => ".git");
351 }
352 command_noisy('config', 'core.autocrlf', 'false');
353 my $set;
354 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
355 foreach my $i (keys %icv) {
356 die "'$set' and '$i' cannot both be set\n" if $set;
357 next unless defined $icv{$i};
358 command_noisy('config', "$pfx.$i", $icv{$i});
359 $set = $i;
360 }
361 my $ignore_regex = \$SVN::Git::Fetcher::_ignore_regex;
362 command_noisy('config', "$pfx.ignore-paths", $$ignore_regex)
363 if defined $$ignore_regex;
364}
365sub init_subdir {
367 my $repo_path = shift or return;
368 mkpath([$repo_path]) unless -d $repo_path;
369 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
370 $ENV{GIT_DIR} = '.git';
371 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
372}
373sub cmd_clone {
375 my ($url, $path) = @_;
376 if (!defined $path &&
377 (defined $_trunk || @_branches || @_tags ||
378 defined $_stdlayout) &&
379 $url !~ m#^[a-z\+]+://#) {
380 $path = $url;
381 }
382 $path = basename($url) if !defined $path || !length $path;
383 cmd_init($url, $path);
384 Git::SVN::fetch_all($Git::SVN::default_repo_id);
385 command_oneline('config', 'svn.authorsfile', $_authors) if $_authors;
386}
387sub cmd_init {
389 if (defined $_stdlayout) {
390 $_trunk = 'trunk' if (!defined $_trunk);
391 @_tags = 'tags' if (! @_tags);
392 @_branches = 'branches' if (! @_branches);
393 }
394 if (defined $_trunk || @_branches || @_tags) {
395 return cmd_multi_init(@_);
396 }
397 my $url = shift or die "SVN repository location required ",
398 "as a command-line argument\n";
399 $url = canonicalize_url($url);
400 init_subdir(@_);
401 do_git_init_db();
402 if ($Git::SVN::_minimize_url eq 'unset') {
404 $Git::SVN::_minimize_url = 0;
405 }
406 Git::SVN->init($url);
408}
409sub cmd_fetch {
411 if (grep /^\d+=./, @_) {
412 die "'<rev>=<commit>' fetch arguments are ",
413 "no longer supported.\n";
414 }
415 my ($remote) = @_;
416 if (@_ > 1) {
417 die "Usage: $0 fetch [--all] [--parent] [svn-remote]\n";
418 }
419 if ($_fetch_parent) {
420 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
421 unless ($gs) {
422 die "Unable to determine upstream SVN information from ",
423 "working tree history\n";
424 }
425 # just fetch, don't checkout.
426 $_no_checkout = 'true';
427 $_fetch_all ? $gs->fetch_all : $gs->fetch;
428 } elsif ($_fetch_all) {
429 cmd_multi_fetch();
430 } else {
431 $remote ||= $Git::SVN::default_repo_id;
432 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
433 }
434}
435sub cmd_set_tree {
437 my (@commits) = @_;
438 if ($_stdin || !@commits) {
439 print "Reading from stdin...\n";
440 @commits = ();
441 while (<STDIN>) {
442 if (/\b($sha1_short)\b/o) {
443 unshift @commits, $1;
444 }
445 }
446 }
447 my @revs;
448 foreach my $c (@commits) {
449 my @tmp = command('rev-parse',$c);
450 if (scalar @tmp == 1) {
451 push @revs, $tmp[0];
452 } elsif (scalar @tmp > 1) {
453 push @revs, reverse(command('rev-list',@tmp));
454 } else {
455 fatal "Failed to rev-parse $c";
456 }
457 }
458 my $gs = Git::SVN->new;
459 my ($r_last, $cmt_last) = $gs->last_rev_commit;
460 $gs->fetch;
461 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
462 fatal "There are new revisions that were fetched ",
463 "and need to be merged (or acknowledged) ",
464 "before committing.\nlast rev: $r_last\n",
465 " current: $gs->{last_rev}";
466 }
467 $gs->set_tree($_) foreach @revs;
468 print "Done committing ",scalar @revs," revisions to SVN\n";
469 unlink $gs->{index};
470}
471sub cmd_dcommit {
473 my $head = shift;
474 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
475 'Cannot dcommit with a dirty index. Commit your changes first, '
476 . "or stash them with `git stash'.\n";
477 $head ||= 'HEAD';
478 my $old_head;
480 if ($head ne 'HEAD') {
481 $old_head = eval {
482 command_oneline([qw/symbolic-ref -q HEAD/])
483 };
484 if ($old_head) {
485 $old_head =~ s{^refs/heads/}{};
486 } else {
487 $old_head = eval { command_oneline(qw/rev-parse HEAD/) };
488 }
489 command(['checkout', $head], STDERR => 0);
490 }
491 my @refs;
493 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD', \@refs);
494 unless ($gs) {
495 die "Unable to determine upstream SVN information from ",
496 "$head history.\nPerhaps the repository is empty.";
497 }
498 if (defined $_commit_url) {
500 $url = $_commit_url;
501 } else {
502 $url = eval { command_oneline('config', '--get',
503 "svn-remote.$gs->{repo_id}.commiturl") };
504 if (!$url) {
505 $url = $gs->full_url
506 }
507 }
508 my $last_rev = $_revision if defined $_revision;
510 if ($url) {
511 print "Committing to $url ...\n";
512 }
513 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
514 if ($_no_rebase && scalar(@$linear_refs) > 1) {
515 warn "Attempting to commit more than one change while ",
516 "--no-rebase is enabled.\n",
517 "If these changes depend on each other, re-running ",
518 "without --no-rebase may be required."
519 }
520 my $expect_url = $url;
521 Git::SVN::remove_username($expect_url);
522 while (1) {
523 my $d = shift @$linear_refs or last;
524 unless (defined $last_rev) {
525 (undef, $last_rev, undef) = cmt_metadata("$d~1");
526 unless (defined $last_rev) {
527 fatal "Unable to extract revision information ",
528 "from commit $d~1";
529 }
530 }
531 if ($_dry_run) {
532 print "diff-tree $d~1 $d\n";
533 } else {
534 my $cmt_rev;
535 my %ed_opts = ( r => $last_rev,
536 log => get_commit_entry($d)->{log},
537 ra => Git::SVN::Ra->new($url),
538 config => SVN::Core::config_get_config(
539 $Git::SVN::Ra::config_dir
540 ),
541 tree_a => "$d~1",
542 tree_b => $d,
543 editor_cb => sub {
544 print "Committed r$_[0]\n";
545 $cmt_rev = $_[0];
546 },
547 svn_path => '');
548 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
549 print "No changes\n$d~1 == $d\n";
550 } elsif ($parents->{$d} && @{$parents->{$d}}) {
551 $gs->{inject_parents_dcommit}->{$cmt_rev} =
552 $parents->{$d};
553 }
554 $_fetch_all ? $gs->fetch_all : $gs->fetch;
555 $last_rev = $cmt_rev;
556 next if $_no_rebase;
557 # we always want to rebase against the current HEAD,
559 # not any head that was passed to us
560 my @diff = command('diff-tree', $d,
561 $gs->refname, '--');
562 my @finish;
563 if (@diff) {
564 @finish = rebase_cmd();
565 print STDERR "W: $d and ", $gs->refname,
566 " differ, using @finish:\n",
567 join("\n", @diff), "\n";
568 } else {
569 print "No changes between current HEAD and ",
570 $gs->refname,
571 "\nResetting to the latest ",
572 $gs->refname, "\n";
573 @finish = qw/reset --mixed/;
574 }
575 command_noisy(@finish, $gs->refname);
576 if (@diff) {
577 @refs = ();
578 my ($url_, $rev_, $uuid_, $gs_) =
579 working_head_info('HEAD', \@refs);
580 my ($linear_refs_, $parents_) =
581 linearize_history($gs_, \@refs);
582 if (scalar(@$linear_refs) !=
583 scalar(@$linear_refs_)) {
584 fatal "# of revisions changed ",
585 "\nbefore:\n",
586 join("\n", @$linear_refs),
587 "\n\nafter:\n",
588 join("\n", @$linear_refs_), "\n",
589 'If you are attempting to commit ',
590 "merges, try running:\n\t",
591 'git rebase --interactive',
592 '--preserve-merges ',
593 $gs->refname,
594 "\nBefore dcommitting";
595 }
596 if ($url_ ne $expect_url) {
597 fatal "URL mismatch after rebase: ",
598 "$url_ != $expect_url";
599 }
600 if ($uuid_ ne $uuid) {
601 fatal "uuid mismatch after rebase: ",
602 "$uuid_ != $uuid";
603 }
604 # remap parents
605 my (%p, @l, $i);
606 for ($i = 0; $i < scalar @$linear_refs; $i++) {
607 my $new = $linear_refs_->[$i] or next;
608 $p{$new} =
609 $parents->{$linear_refs->[$i]};
610 push @l, $new;
611 }
612 $parents = \%p;
613 $linear_refs = \@l;
614 }
615 }
616 }
617 if ($old_head) {
619 my $new_head = command_oneline(qw/rev-parse HEAD/);
620 my $new_is_symbolic = eval {
621 command_oneline(qw/symbolic-ref -q HEAD/);
622 };
623 if ($new_is_symbolic) {
624 print "dcommitted the branch ", $head, "\n";
625 } else {
626 print "dcommitted on a detached HEAD because you gave ",
627 "a revision argument.\n",
628 "The rewritten commit is: ", $new_head, "\n";
629 }
630 command(['checkout', $old_head], STDERR => 0);
631 }
632 unlink $gs->{index};
634}
635sub cmd_branch {
637 my ($branch_name, $head) = @_;
638 unless (defined $branch_name && length $branch_name) {
640 die(($_tag ? "tag" : "branch") . " name required\n");
641 }
642 $head ||= 'HEAD';
643 my ($src, $rev, undef, $gs) = working_head_info($head);
645 my $remote = Git::SVN::read_all_remotes()->{$gs->{repo_id}};
647 my $allglobs = $remote->{ $_tag ? 'tags' : 'branches' };
648 my $glob;
649 if ($#{$allglobs} == 0) {
650 $glob = $allglobs->[0];
651 } else {
652 unless(defined $_branch_dest) {
653 die "Multiple ",
654 $_tag ? "tag" : "branch",
655 " paths defined for Subversion repository.\n",
656 "You must specify where you want to create the ",
657 $_tag ? "tag" : "branch",
658 " with the --destination argument.\n";
659 }
660 foreach my $g (@{$allglobs}) {
661 # SVN::Git::Editor could probably be moved to Git.pm..
662 my $re = SVN::Git::Editor::glob2pat($g->{path}->{left});
663 if ($_branch_dest =~ /$re/) {
664 $glob = $g;
665 last;
666 }
667 }
668 unless (defined $glob) {
669 die "Unknown ",
670 $_tag ? "tag" : "branch",
671 " destination $_branch_dest\n";
672 }
673 }
674 my ($lft, $rgt) = @{ $glob->{path} }{qw/left right/};
675 my $dst = join '/', $remote->{url}, $lft, $branch_name, ($rgt || ());
676 my $ctx = SVN::Client->new(
678 auth => Git::SVN::Ra::_auth_providers(),
679 log_msg => sub {
680 ${ $_[0] } = defined $_message
681 ? $_message
682 : 'Create ' . ($_tag ? 'tag ' : 'branch ' )
683 . $branch_name;
684 },
685 );
686 eval {
688 $ctx->ls($dst, 'HEAD', 0);
689 } and die "branch ${branch_name} already exists\n";
690 print "Copying ${src} at r${rev} to ${dst}...\n";
692 $ctx->copy($src, $rev, $dst)
693 unless $_dry_run;
694 $gs->fetch_all;
696}
697sub cmd_find_rev {
699 my $revision_or_hash = shift or die "SVN or git revision required ",
700 "as a command-line argument\n";
701 my $result;
702 if ($revision_or_hash =~ /^r\d+$/) {
703 my $head = shift;
704 $head ||= 'HEAD';
705 my @refs;
706 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
707 unless ($gs) {
708 die "Unable to determine upstream SVN information from ",
709 "$head history\n";
710 }
711 my $desired_revision = substr($revision_or_hash, 1);
712 $result = $gs->rev_map_get($desired_revision, $uuid);
713 } else {
714 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
715 $result = $rev;
716 }
717 print "$result\n" if $result;
718}
719sub cmd_rebase {
721 command_noisy(qw/update-index --refresh/);
722 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
723 unless ($gs) {
724 die "Unable to determine upstream SVN information from ",
725 "working tree history\n";
726 }
727 if ($_dry_run) {
728 print "Remote Branch: " . $gs->refname . "\n";
729 print "SVN URL: " . $url . "\n";
730 return;
731 }
732 if (command(qw/diff-index HEAD --/)) {
733 print STDERR "Cannot rebase with uncommited changes:\n";
734 command_noisy('status');
735 exit 1;
736 }
737 unless ($_local) {
738 # rebase will checkout for us, so no need to do it explicitly
739 $_no_checkout = 'true';
740 $_fetch_all ? $gs->fetch_all : $gs->fetch;
741 }
742 command_noisy(rebase_cmd(), $gs->refname);
743}
744sub cmd_show_ignore {
746 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
747 $gs ||= Git::SVN->new;
748 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
749 $gs->prop_walk($gs->{path}, $r, sub {
750 my ($gs, $path, $props) = @_;
751 print STDOUT "\n# $path\n";
752 my $s = $props->{'svn:ignore'} or return;
753 $s =~ s/[\r\n]+/\n/g;
754 chomp $s;
755 $s =~ s#^#$path#gm;
756 print STDOUT "$s\n";
757 });
758}
759sub cmd_show_externals {
761 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
762 $gs ||= Git::SVN->new;
763 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
764 $gs->prop_walk($gs->{path}, $r, sub {
765 my ($gs, $path, $props) = @_;
766 print STDOUT "\n# $path\n";
767 my $s = $props->{'svn:externals'} or return;
768 $s =~ s/[\r\n]+/\n/g;
769 chomp $s;
770 $s =~ s#^#$path#gm;
771 print STDOUT "$s\n";
772 });
773}
774sub cmd_create_ignore {
776 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
777 $gs ||= Git::SVN->new;
778 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
779 $gs->prop_walk($gs->{path}, $r, sub {
780 my ($gs, $path, $props) = @_;
781 # $path is of the form /path/to/dir/
782 $path = '.' . $path;
783 # SVN can have attributes on empty directories,
784 # which git won't track
785 mkpath([$path]) unless -d $path;
786 my $ignore = $path . '.gitignore';
787 my $s = $props->{'svn:ignore'} or return;
788 open(GITIGNORE, '>', $ignore)
789 or fatal("Failed to open `$ignore' for writing: $!");
790 $s =~ s/[\r\n]+/\n/g;
791 chomp $s;
792 # Prefix all patterns so that the ignore doesn't apply
793 # to sub-directories.
794 $s =~ s#^#/#gm;
795 print GITIGNORE "$s\n";
796 close(GITIGNORE)
797 or fatal("Failed to close `$ignore': $!");
798 command_noisy('add', '-f', $ignore);
799 });
800}
801sub canonicalize_path {
803 my ($path) = @_;
804 my $dot_slash_added = 0;
805 if (substr($path, 0, 1) ne "/") {
806 $path = "./" . $path;
807 $dot_slash_added = 1;
808 }
809 # File::Spec->canonpath doesn't collapse x/../y into y (for a
810 # good reason), so let's do this manually.
811 $path =~ s#/+#/#g;
812 $path =~ s#/\.(?:/|$)#/#g;
813 $path =~ s#/[^/]+/\.\.##g;
814 $path =~ s#/$##g;
815 $path =~ s#^\./## if $dot_slash_added;
816 $path =~ s#^/##;
817 $path =~ s#^\.$##;
818 return $path;
819}
820sub canonicalize_url {
822 my ($url) = @_;
823 $url =~ s#^([^:]+://[^/]*/)(.*)$#$1 . canonicalize_path($2)#e;
824 return $url;
825}
826# get_svnprops(PATH)
828# ------------------
829# Helper for cmd_propget and cmd_proplist below.
830sub get_svnprops {
831 my $path = shift;
832 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
833 $gs ||= Git::SVN->new;
834 # prefix THE PATH by the sub-directory from which the user
836 # invoked us.
837 $path = $cmd_dir_prefix . $path;
838 fatal("No such file or directory: $path") unless -e $path;
839 my $is_dir = -d $path ? 1 : 0;
840 $path = $gs->{path} . '/' . $path;
841 # canonicalize the path (otherwise libsvn will abort or fail to
843 # find the file)
844 $path = canonicalize_path($path);
845 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
847 my $props;
848 if ($is_dir) {
849 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
850 }
851 else {
852 (undef, $props) = $gs->ra->get_file($path, $r, undef);
853 }
854 return $props;
855}
856# cmd_propget (PROP, PATH)
858# ------------------------
859# Print the SVN property PROP for PATH.
860sub cmd_propget {
861 my ($prop, $path) = @_;
862 $path = '.' if not defined $path;
863 usage(1) if not defined $prop;
864 my $props = get_svnprops($path);
865 if (not defined $props->{$prop}) {
866 fatal("`$path' does not have a `$prop' SVN property.");
867 }
868 print $props->{$prop} . "\n";
869}
870# cmd_proplist (PATH)
872# -------------------
873# Print the list of SVN properties for PATH.
874sub cmd_proplist {
875 my $path = shift;
876 $path = '.' if not defined $path;
877 my $props = get_svnprops($path);
878 print "Properties on '$path':\n";
879 foreach (sort keys %{$props}) {
880 print " $_\n";
881 }
882}
883sub cmd_multi_init {
885 my $url = shift;
886 unless (defined $_trunk || @_branches || @_tags) {
887 usage(1);
888 }
889 $_prefix = '' unless defined $_prefix;
891 if (defined $url) {
892 $url = canonicalize_url($url);
893 init_subdir(@_);
894 }
895 do_git_init_db();
896 if (defined $_trunk) {
897 my $trunk_ref = $_prefix . 'trunk';
898 # try both old-style and new-style lookups:
899 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
900 unless ($gs_trunk) {
901 my ($trunk_url, $trunk_path) =
902 complete_svn_url($url, $_trunk);
903 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
904 undef, $trunk_ref);
905 }
906 }
907 return unless @_branches || @_tags;
908 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
909 foreach my $path (@_branches) {
910 complete_url_ls_init($ra, $path, '--branches/-b', $_prefix);
911 }
912 foreach my $path (@_tags) {
913 complete_url_ls_init($ra, $path, '--tags/-t', $_prefix.'tags/');
914 }
915}
916sub cmd_multi_fetch {
918 my $remotes = Git::SVN::read_all_remotes();
919 foreach my $repo_id (sort keys %$remotes) {
920 if ($remotes->{$repo_id}->{url}) {
921 Git::SVN::fetch_all($repo_id, $remotes);
922 }
923 }
924}
925# this command is special because it requires no metadata
927sub cmd_commit_diff {
928 my ($ta, $tb, $url) = @_;
929 my $usage = "Usage: $0 commit-diff -r<revision> ".
930 "<tree-ish> <tree-ish> [<URL>]";
931 fatal($usage) if (!defined $ta || !defined $tb);
932 my $svn_path = '';
933 if (!defined $url) {
934 my $gs = eval { Git::SVN->new };
935 if (!$gs) {
936 fatal("Needed URL or usable git-svn --id in ",
937 "the command-line\n", $usage);
938 }
939 $url = $gs->{url};
940 $svn_path = $gs->{path};
941 }
942 unless (defined $_revision) {
943 fatal("-r|--revision is a required argument\n", $usage);
944 }
945 if (defined $_message && defined $_file) {
946 fatal("Both --message/-m and --file/-F specified ",
947 "for the commit message.\n",
948 "I have no idea what you mean");
949 }
950 if (defined $_file) {
951 $_message = file_to_s($_file);
952 } else {
953 $_message ||= get_commit_entry($tb)->{log};
954 }
955 my $ra ||= Git::SVN::Ra->new($url);
956 my $r = $_revision;
957 if ($r eq 'HEAD') {
958 $r = $ra->get_latest_revnum;
959 } elsif ($r !~ /^\d+$/) {
960 die "revision argument: $r not understood by git-svn\n";
961 }
962 my %ed_opts = ( r => $r,
963 log => $_message,
964 ra => $ra,
965 tree_a => $ta,
966 tree_b => $tb,
967 editor_cb => sub { print "Committed r$_[0]\n" },
968 svn_path => $svn_path );
969 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
970 print "No changes\n$ta == $tb\n";
971 }
972}
973sub escape_uri_only {
975 my ($uri) = @_;
976 my @tmp;
977 foreach (split m{/}, $uri) {
978 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
979 push @tmp, $_;
980 }
981 join('/', @tmp);
982}
983sub escape_url {
985 my ($url) = @_;
986 if ($url =~ m#^([^:]+)://([^/]*)(.*)$#) {
987 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
988 $url = "$scheme://$domain$uri";
989 }
990 $url;
991}
992sub cmd_info {
994 my $path = canonicalize_path(defined($_[0]) ? $_[0] : ".");
995 my $fullpath = canonicalize_path($cmd_dir_prefix . $path);
996 if (exists $_[1]) {
997 die "Too many arguments specified\n";
998 }
999 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
1001 if (!$file_type && !$diff_status) {
1003 print STDERR "svn: '$path' is not under version control\n";
1004 exit 1;
1005 }
1006 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1008 unless ($gs) {
1009 die "Unable to determine upstream SVN information from ",
1010 "working tree history\n";
1011 }
1012 # canonicalize_path() will return "" to make libsvn 1.5.x happy,
1014 $path = "." if $path eq "";
1015 my $full_url = $url . ($fullpath eq "" ? "" : "/$fullpath");
1017 if ($_url) {
1019 print escape_url($full_url), "\n";
1020 return;
1021 }
1022 my $result = "Path: $path\n";
1024 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
1025 $result .= "URL: " . escape_url($full_url) . "\n";
1026 eval {
1028 my $repos_root = $gs->repos_root;
1029 Git::SVN::remove_username($repos_root);
1030 $result .= "Repository Root: " . escape_url($repos_root) . "\n";
1031 };
1032 if ($@) {
1033 $result .= "Repository Root: (offline)\n";
1034 }
1035 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A" &&
1036 ($SVN::Core::VERSION le '1.5.4' || $file_type ne "dir");
1037 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
1038 $result .= "Node Kind: " .
1040 ($file_type eq "dir" ? "directory" : "file") . "\n";
1041 my $schedule = $diff_status eq "A"
1043 ? "add"
1044 : ($diff_status eq "D" ? "delete" : "normal");
1045 $result .= "Schedule: $schedule\n";
1046 if ($diff_status eq "A") {
1048 print $result, "\n";
1049 return;
1050 }
1051 my ($lc_author, $lc_rev, $lc_date_utc);
1053 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $fullpath);
1054 my $log = command_output_pipe(@args);
1055 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
1056 while (<$log>) {
1057 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
1058 $lc_author = $1;
1059 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
1060 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
1061 (undef, $lc_rev, undef) = ::extract_metadata($1);
1062 }
1063 }
1064 close $log;
1065 Git::SVN::Log::set_local_timezone();
1067 $result .= "Last Changed Author: $lc_author\n";
1069 $result .= "Last Changed Rev: $lc_rev\n";
1070 $result .= "Last Changed Date: " .
1071 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
1072 if ($file_type ne "dir") {
1074 my $text_last_updated_date =
1075 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
1076 $result .=
1077 "Text Last Updated: " .
1078 Git::SVN::Log::format_svn_date($text_last_updated_date) .
1079 "\n";
1080 my $checksum;
1081 if ($diff_status eq "D") {
1082 my ($fh, $ctx) =
1083 command_output_pipe(qw(cat-file blob), "HEAD:$path");
1084 if ($file_type eq "link") {
1085 my $file_name = <$fh>;
1086 $checksum = md5sum("link $file_name");
1087 } else {
1088 $checksum = md5sum($fh);
1089 }
1090 command_close_pipe($fh, $ctx);
1091 } elsif ($file_type eq "link") {
1092 my $file_name =
1093 command(qw(cat-file blob), "HEAD:$path");
1094 $checksum =
1095 md5sum("link " . $file_name);
1096 } else {
1097 open FILE, "<", $path or die $!;
1098 $checksum = md5sum(\*FILE);
1099 close FILE or die $!;
1100 }
1101 $result .= "Checksum: " . $checksum . "\n";
1102 }
1103 print $result, "\n";
1105}
1106sub cmd_reset {
1108 my $target = shift || $_revision or die "SVN revision required\n";
1109 $target = $1 if $target =~ /^r(\d+)$/;
1110 $target =~ /^\d+$/ or die "Numeric SVN revision expected\n";
1111 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
1112 unless ($gs) {
1113 die "Unable to determine upstream SVN information from ".
1114 "history\n";
1115 }
1116 my ($r, $c) = $gs->find_rev_before($target, not $_fetch_parent);
1117 $gs->rev_map_set($r, $c, 'reset', $uuid);
1118 print "r$r = $c ($gs->{ref_id})\n";
1119}
1120sub cmd_gc {
1122 if (!$can_compress) {
1123 warn "Compress::Zlib could not be found; unhandled.log " .
1124 "files will not be compressed.\n";
1125 }
1126 find({ wanted => \&gc_directory, no_chdir => 1}, "$ENV{GIT_DIR}/svn");
1127}
1128########################### utility functions #########################
1130sub rebase_cmd {
1132 my @cmd = qw/rebase/;
1133 push @cmd, '-v' if $_verbose;
1134 push @cmd, qw/--merge/ if $_merge;
1135 push @cmd, "--strategy=$_strategy" if $_strategy;
1136 @cmd;
1137}
1138sub post_fetch_checkout {
1140 return if $_no_checkout;
1141 my $gs = $Git::SVN::_head or return;
1142 return if verify_ref('refs/heads/master^0');
1143 my $valid_head = verify_ref('HEAD^0');
1145 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
1146 return if ($valid_head || !verify_ref('HEAD^0'));
1147 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
1149 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
1150 return if -f $index;
1151 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
1153 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
1154 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
1155 print STDERR "Checked out HEAD:\n ",
1156 $gs->full_url, " r", $gs->last_rev, "\n";
1157}
1158sub complete_svn_url {
1160 my ($url, $path) = @_;
1161 $path =~ s#/+$##;
1162 if ($path !~ m#^[a-z\+]+://#) {
1163 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
1164 fatal("E: '$path' is not a complete URL ",
1165 "and a separate URL is not specified");
1166 }
1167 return ($url, $path);
1168 }
1169 return ($path, '');
1170}
1171sub complete_url_ls_init {
1173 my ($ra, $repo_path, $switch, $pfx) = @_;
1174 unless ($repo_path) {
1175 print STDERR "W: $switch not specified\n";
1176 return;
1177 }
1178 $repo_path =~ s#/+$##;
1179 if ($repo_path =~ m#^[a-z\+]+://#) {
1180 $ra = Git::SVN::Ra->new($repo_path);
1181 $repo_path = '';
1182 } else {
1183 $repo_path =~ s#^/+##;
1184 unless ($ra) {
1185 fatal("E: '$repo_path' is not a complete URL ",
1186 "and a separate URL is not specified");
1187 }
1188 }
1189 my $url = $ra->{url};
1190 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
1191 my $k = "svn-remote.$gs->{repo_id}.url";
1192 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
1193 if ($orig_url && ($orig_url ne $gs->{url})) {
1194 die "$k already set: $orig_url\n",
1195 "wanted to set to: $gs->{url}\n";
1196 }
1197 command_oneline('config', $k, $gs->{url}) unless $orig_url;
1198 my $remote_path = "$gs->{path}/$repo_path";
1199 $remote_path =~ s#/+#/#g;
1200 $remote_path =~ s#^/##g;
1201 $remote_path .= "/*" if $remote_path !~ /\*/;
1202 my ($n) = ($switch =~ /^--(\w+)/);
1203 if (length $pfx && $pfx !~ m#/$#) {
1204 die "--prefix='$pfx' must have a trailing slash '/'\n";
1205 }
1206 command_noisy('config',
1207 '--add',
1208 "svn-remote.$gs->{repo_id}.$n",
1209 "$remote_path:refs/remotes/$pfx*" .
1210 ('/*' x (($remote_path =~ tr/*/*/) - 1)) );
1211}
1212sub verify_ref {
1214 my ($ref) = @_;
1215 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
1216 { STDERR => 0 }); };
1217}
1218sub get_tree_from_treeish {
1220 my ($treeish) = @_;
1221 # $treeish can be a symbolic ref, too:
1222 my $type = command_oneline(qw/cat-file -t/, $treeish);
1223 my $expected;
1224 while ($type eq 'tag') {
1225 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1226 }
1227 if ($type eq 'commit') {
1228 $expected = (grep /^tree /, command(qw/cat-file commit/,
1229 $treeish))[0];
1230 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1231 die "Unable to get tree from $treeish\n" unless $expected;
1232 } elsif ($type eq 'tree') {
1233 $expected = $treeish;
1234 } else {
1235 die "$treeish is a $type, expected tree, tag or commit\n";
1236 }
1237 return $expected;
1238}
1239sub get_commit_entry {
1241 my ($treeish) = shift;
1242 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1243 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1244 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1245 open my $log_fh, '>', $commit_editmsg or croak $!;
1246 my $type = command_oneline(qw/cat-file -t/, $treeish);
1248 if ($type eq 'commit' || $type eq 'tag') {
1249 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1250 $type, $treeish);
1251 my $in_msg = 0;
1252 my $author;
1253 my $saw_from = 0;
1254 my $msgbuf = "";
1255 while (<$msg_fh>) {
1256 if (!$in_msg) {
1257 $in_msg = 1 if (/^\s*$/);
1258 $author = $1 if (/^author (.*>)/);
1259 } elsif (/^git-svn-id: /) {
1260 # skip this for now, we regenerate the
1261 # correct one on re-fetch anyways
1262 # TODO: set *:merge properties or like...
1263 } else {
1264 if (/^From:/ || /^Signed-off-by:/) {
1265 $saw_from = 1;
1266 }
1267 $msgbuf .= $_;
1268 }
1269 }
1270 $msgbuf =~ s/\s+$//s;
1271 if ($Git::SVN::_add_author_from && defined($author)
1272 && !$saw_from) {
1273 $msgbuf .= "\n\nFrom: $author";
1274 }
1275 print $log_fh $msgbuf or croak $!;
1276 command_close_pipe($msg_fh, $ctx);
1277 }
1278 close $log_fh or croak $!;
1279 if ($_edit || ($type eq 'tree')) {
1281 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1282 # TODO: strip out spaces, comments, like git-commit.sh
1283 system($editor, $commit_editmsg);
1284 }
1285 rename $commit_editmsg, $commit_msg or croak $!;
1286 {
1287 require Encode;
1288 # SVN requires messages to be UTF-8 when entering the repo
1289 local $/;
1290 open $log_fh, '<', $commit_msg or croak $!;
1291 binmode $log_fh;
1292 chomp($log_entry{log} = <$log_fh>);
1293 my $enc = Git::config('i18n.commitencoding') || 'UTF-8';
1295 my $msg = $log_entry{log};
1296 eval { $msg = Encode::decode($enc, $msg, 1) };
1298 if ($@) {
1299 die "Could not decode as $enc:\n", $msg,
1300 "\nPerhaps you need to set i18n.commitencoding\n";
1301 }
1302 eval { $msg = Encode::encode('UTF-8', $msg, 1) };
1304 die "Could not encode as UTF-8:\n$msg\n" if $@;
1305 $log_entry{log} = $msg;
1307 close $log_fh or croak $!;
1309 }
1310 unlink $commit_msg;
1311 \%log_entry;
1312}
1313sub s_to_file {
1315 my ($str, $file, $mode) = @_;
1316 open my $fd,'>',$file or croak $!;
1317 print $fd $str,"\n" or croak $!;
1318 close $fd or croak $!;
1319 chmod ($mode &~ umask, $file) if (defined $mode);
1320}
1321sub file_to_s {
1323 my $file = shift;
1324 open my $fd,'<',$file or croak "$!: file: $file\n";
1325 local $/;
1326 my $ret = <$fd>;
1327 close $fd or croak $!;
1328 $ret =~ s/\s*$//s;
1329 return $ret;
1330}
1331# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1333sub load_authors {
1334 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1335 my $log = $cmd eq 'log';
1336 while (<$authors>) {
1337 chomp;
1338 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1339 my ($user, $name, $email) = ($1, $2, $3);
1340 if ($log) {
1341 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1342 } else {
1343 $users{$user} = [$name, $email];
1344 }
1345 }
1346 close $authors or croak $!;
1347}
1348# convert GetOpt::Long specs for use by git-config
1350sub read_repo_config {
1351 return unless -d $ENV{GIT_DIR};
1352 my $opts = shift;
1353 my @config_only;
1354 foreach my $o (keys %$opts) {
1355 # if we have mixedCase and a long option-only, then
1356 # it's a config-only variable that we don't need for
1357 # the command-line.
1358 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1359 my $v = $opts->{$o};
1360 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1361 $key =~ s/-//g;
1362 my $arg = 'git config';
1363 $arg .= ' --int' if ($o =~ /[:=]i$/);
1364 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1365 if (ref $v eq 'ARRAY') {
1366 chomp(my @tmp = `$arg --get-all svn.$key`);
1367 @$v = @tmp if @tmp;
1368 } else {
1369 chomp(my $tmp = `$arg --get svn.$key`);
1370 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1371 $$v = $tmp;
1372 }
1373 }
1374 }
1375 delete @$opts{@config_only} if @config_only;
1376}
1377sub extract_metadata {
1379 my $id = shift or return (undef, undef, undef);
1380 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1381 \s([a-f\d\-]+)$/ix);
1382 if (!defined $rev || !$uuid || !$url) {
1383 # some of the original repositories I made had
1384 # identifiers like this:
1385 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/i);
1386 }
1387 return ($url, $rev, $uuid);
1388}
1389sub cmt_metadata {
1391 return extract_metadata((grep(/^git-svn-id: /,
1392 command(qw/cat-file commit/, shift)))[-1]);
1393}
1394sub cmt_sha2rev_batch {
1396 my %s2r;
1397 my ($pid, $in, $out, $ctx) = command_bidi_pipe(qw/cat-file --batch/);
1398 my $list = shift;
1399 foreach my $sha (@{$list}) {
1401 my $first = 1;
1402 my $size = 0;
1403 print $out $sha, "\n";
1404 while (my $line = <$in>) {
1406 if ($first && $line =~ /^[[:xdigit:]]{40}\smissing$/) {
1407 last;
1408 } elsif ($first &&
1409 $line =~ /^[[:xdigit:]]{40}\scommit\s(\d+)$/) {
1410 $first = 0;
1411 $size = $1;
1412 next;
1413 } elsif ($line =~ /^(git-svn-id: )/) {
1414 my (undef, $rev, undef) =
1415 extract_metadata($line);
1416 $s2r{$sha} = $rev;
1417 }
1418 $size -= length($line);
1420 last if ($size == 0);
1421 }
1422 }
1423 command_close_bidi_pipe($pid, $in, $out, $ctx);
1425 return \%s2r;
1427}
1428sub working_head_info {
1430 my ($head, $refs) = @_;
1431 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1432 my ($fh, $ctx) = command_output_pipe(@args, $head);
1433 my $hash;
1434 my %max;
1435 while (<$fh>) {
1436 if ( m{^commit ($::sha1)$} ) {
1437 unshift @$refs, $hash if $hash and $refs;
1438 $hash = $1;
1439 next;
1440 }
1441 next unless s{^\s*(git-svn-id:)}{$1};
1442 my ($url, $rev, $uuid) = extract_metadata($_);
1443 if (defined $url && defined $rev) {
1444 next if $max{$url} and $max{$url} < $rev;
1445 if (my $gs = Git::SVN->find_by_url($url)) {
1446 my $c = $gs->rev_map_get($rev, $uuid);
1447 if ($c && $c eq $hash) {
1448 close $fh; # break the pipe
1449 return ($url, $rev, $uuid, $gs);
1450 } else {
1451 $max{$url} ||= $gs->rev_map_max;
1452 }
1453 }
1454 }
1455 }
1456 command_close_pipe($fh, $ctx);
1457 (undef, undef, undef, undef);
1458}
1459sub read_commit_parents {
1461 my ($parents, $c) = @_;
1462 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1463 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1464 @{$parents->{$c}} = split(/ /, $p);
1465}
1466sub linearize_history {
1468 my ($gs, $refs) = @_;
1469 my %parents;
1470 foreach my $c (@$refs) {
1471 read_commit_parents(\%parents, $c);
1472 }
1473 my @linear_refs;
1475 my %skip = ();
1476 my $last_svn_commit = $gs->last_commit;
1477 foreach my $c (reverse @$refs) {
1478 next if $c eq $last_svn_commit;
1479 last if $skip{$c};
1480 unshift @linear_refs, $c;
1482 $skip{$c} = 1;
1483 # we only want the first parent to diff against for linear
1485 # history, we save the rest to inject when we finalize the
1486 # svn commit
1487 my $fp_a = verify_ref("$c~1");
1488 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1489 if (!$fp_a || !$fp_b) {
1490 die "Commit $c\n",
1491 "has no parent commit, and therefore ",
1492 "nothing to diff against.\n",
1493 "You should be working from a repository ",
1494 "originally created by git-svn\n";
1495 }
1496 if ($fp_a ne $fp_b) {
1497 die "$c~1 = $fp_a, however parsing commit $c ",
1498 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1499 }
1500 foreach my $p (@{$parents{$c}}) {
1502 $skip{$p} = 1;
1503 }
1504 }
1505 (\@linear_refs, \%parents);
1506}
1507sub find_file_type_and_diff_status {
1509 my ($path) = @_;
1510 return ('dir', '') if $path eq '';
1511 my $diff_output =
1513 command_oneline(qw(diff --cached --name-status --), $path) || "";
1514 my $diff_status = (split(' ', $diff_output))[0] || "";
1515 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1517 return (undef, undef) if !$diff_status && !$ls_tree;
1519 if ($diff_status eq "A") {
1521 return ("link", $diff_status) if -l $path;
1522 return ("dir", $diff_status) if -d $path;
1523 return ("file", $diff_status);
1524 }
1525 my $mode = (split(' ', $ls_tree))[0] || "";
1527 return ("link", $diff_status) if $mode eq "120000";
1529 return ("dir", $diff_status) if $mode eq "040000";
1530 return ("file", $diff_status);
1531}
1532sub md5sum {
1534 my $arg = shift;
1535 my $ref = ref $arg;
1536 my $md5 = Digest::MD5->new();
1537 if ($ref eq 'GLOB' || $ref eq 'IO::File' || $ref eq 'File::Temp') {
1538 $md5->addfile($arg) or croak $!;
1539 } elsif ($ref eq 'SCALAR') {
1540 $md5->add($$arg) or croak $!;
1541 } elsif (!$ref) {
1542 $md5->add($arg) or croak $!;
1543 } else {
1544 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1545 }
1546 return $md5->hexdigest();
1547}
1548sub gc_directory {
1550 if ($can_compress && -f $_ && basename($_) eq "unhandled.log") {
1551 my $out_filename = $_ . ".gz";
1552 open my $in_fh, "<", $_ or die "Unable to open $_: $!\n";
1553 binmode $in_fh;
1554 my $gz = Compress::Zlib::gzopen($out_filename, "ab") or
1555 die "Unable to open $out_filename: $!\n";
1556 my $res;
1558 while ($res = sysread($in_fh, my $str, 1024)) {
1559 $gz->gzwrite($str) or
1560 die "Unable to write: ".$gz->gzerror()."!\n";
1561 }
1562 unlink $_ or die "unlink $File::Find::name: $!\n";
1563 } elsif (-f $_ && basename($_) eq "index") {
1564 unlink $_ or die "unlink $_: $!\n";
1565 }
1566}
1567package Git::SVN;
1569use strict;
1570use warnings;
1571use Fcntl qw/:DEFAULT :seek/;
1572use constant rev_map_fmt => 'NH40';
1573use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1574 $_repack $_repack_flags $_use_svm_props $_head
1575 $_use_svnsync_props $no_reuse_existing $_minimize_url
1576 $_use_log_author $_add_author_from $_localtime/;
1577use Carp qw/croak/;
1578use File::Path qw/mkpath/;
1579use File::Copy qw/copy/;
1580use IPC::Open3;
1581my ($_gc_nr, $_gc_period);
1583# properties that we do not log:
1585my %SKIP_PROP;
1586BEGIN {
1587 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1588 svn:special svn:executable
1589 svn:entry:committed-rev
1590 svn:entry:last-author
1591 svn:entry:uuid
1592 svn:entry:committed-date/;
1593 # some options are read globally, but can be overridden locally
1595 # per [svn-remote "..."] section. Command-line options will *NOT*
1596 # override options set in an [svn-remote "..."] section
1597 no strict 'refs';
1598 for my $option (qw/follow_parent no_metadata use_svm_props
1599 use_svnsync_props/) {
1600 my $key = $option;
1601 $key =~ tr/_//d;
1602 my $prop = "-$option";
1603 *$option = sub {
1604 my ($self) = @_;
1605 return $self->{$prop} if exists $self->{$prop};
1606 my $k = "svn-remote.$self->{repo_id}.$key";
1607 eval { command_oneline(qw/config --get/, $k) };
1608 if ($@) {
1609 $self->{$prop} = ${"Git::SVN::_$option"};
1610 } else {
1611 my $v = command_oneline(qw/config --bool/,$k);
1612 $self->{$prop} = $v eq 'false' ? 0 : 1;
1613 }
1614 return $self->{$prop};
1615 }
1616 }
1617}
1618my (%LOCKFILES, %INDEX_FILES);
1621END {
1622 unlink keys %LOCKFILES if %LOCKFILES;
1623 unlink keys %INDEX_FILES if %INDEX_FILES;
1624}
1625sub resolve_local_globs {
1627 my ($url, $fetch, $glob_spec) = @_;
1628 return unless defined $glob_spec;
1629 my $ref = $glob_spec->{ref};
1630 my $path = $glob_spec->{path};
1631 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1632 next unless m#^refs/remotes/$ref->{regex}$#;
1633 my $p = $1;
1634 my $pathname = desanitize_refname($path->full_path($p));
1635 my $refname = desanitize_refname($ref->full_path($p));
1636 if (my $existing = $fetch->{$pathname}) {
1637 if ($existing ne $refname) {
1638 die "Refspec conflict:\n",
1639 "existing: refs/remotes/$existing\n",
1640 " globbed: refs/remotes/$refname\n";
1641 }
1642 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1643 $u =~ s!^\Q$url\E(/|$)!! or die
1644 "refs/remotes/$refname: '$url' not found in '$u'\n";
1645 if ($pathname ne $u) {
1646 warn "W: Refspec glob conflict ",
1647 "(ref: refs/remotes/$refname):\n",
1648 "expected path: $pathname\n",
1649 " real path: $u\n",
1650 "Continuing ahead with $u\n";
1651 next;
1652 }
1653 } else {
1654 $fetch->{$pathname} = $refname;
1655 }
1656 }
1657}
1658sub parse_revision_argument {
1660 my ($base, $head) = @_;
1661 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1662 return ($base, $head);
1663 }
1664 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1665 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1666 return ($head, $head) if ($::_revision eq 'HEAD');
1667 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1668 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1669 die "revision argument: $::_revision not understood by git-svn\n";
1670}
1671sub fetch_all {
1673 my ($repo_id, $remotes) = @_;
1674 if (ref $repo_id) {
1675 my $gs = $repo_id;
1676 $repo_id = undef;
1677 $repo_id = $gs->{repo_id};
1678 }
1679 $remotes ||= read_all_remotes();
1680 my $remote = $remotes->{$repo_id} or
1681 die "[svn-remote \"$repo_id\"] unknown\n";
1682 my $fetch = $remote->{fetch};
1683 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1684 my (@gs, @globs);
1685 my $ra = Git::SVN::Ra->new($url);
1686 my $uuid = $ra->get_uuid;
1687 my $head = $ra->get_latest_revnum;
1688 $ra->get_log("", $head, 0, 1, 0, 1, sub { $head = $_[1] });
1689 my $base = defined $fetch ? $head : 0;
1690 # read the max revs for wildcard expansion (branches/*, tags/*)
1692 foreach my $t (qw/branches tags/) {
1693 defined $remote->{$t} or next;
1694 push @globs, @{$remote->{$t}};
1695 my $max_rev = eval { tmp_config(qw/--int --get/,
1697 "svn-remote.$repo_id.${t}-maxRev") };
1698 if (defined $max_rev && ($max_rev < $base)) {
1699 $base = $max_rev;
1700 } elsif (!defined $max_rev) {
1701 $base = 0;
1702 }
1703 }
1704 if ($fetch) {
1706 foreach my $p (sort keys %$fetch) {
1707 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1708 my $lr = $gs->rev_map_max;
1709 if (defined $lr) {
1710 $base = $lr if ($lr < $base);
1711 }
1712 push @gs, $gs;
1713 }
1714 }
1715 ($base, $head) = parse_revision_argument($base, $head);
1717 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1718}
1719sub read_all_remotes {
1721 my $r = {};
1722 my $use_svm_props = eval { command_oneline(qw/config --bool
1723 svn.useSvmProps/) };
1724 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1725 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1726 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1727 my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1728 die("svn-remote.$remote: remote ref '$_remote_ref' "
1729 . "must start with 'refs/remotes/'\n")
1730 unless $_remote_ref =~ m{^refs/remotes/(.+)};
1731 my $remote_ref = $1;
1732 $local_ref =~ s{^/}{};
1733 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1734 $r->{$remote}->{svm} = {} if $use_svm_props;
1735 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1736 $r->{$1}->{svm} = {};
1737 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1738 $r->{$1}->{url} = $2;
1739 } elsif (m!^(.+)\.(branches|tags)=
1740 (.*):refs/remotes/(.+)\s*$/!x) {
1741 my ($p, $g) = ($3, $4);
1742 my $rs = {
1743 t => $2,
1744 remote => $1,
1745 path => Git::SVN::GlobSpec->new($p),
1746 ref => Git::SVN::GlobSpec->new($g) };
1747 if (length($rs->{ref}->{right}) != 0) {
1748 die "The '*' glob character must be the last ",
1749 "character of '$g'\n";
1750 }
1751 push @{ $r->{$1}->{$2} }, $rs;
1752 }
1753 }
1754 map {
1756 if (defined $r->{$_}->{svm}) {
1757 my $svm;
1758 eval {
1759 my $section = "svn-remote.$_";
1760 $svm = {
1761 source => tmp_config('--get',
1762 "$section.svm-source"),
1763 replace => tmp_config('--get',
1764 "$section.svm-replace"),
1765 }
1766 };
1767 $r->{$_}->{svm} = $svm;
1768 }
1769 } keys %$r;
1770 $r;
1772}
1773sub init_vars {
1775 $_gc_nr = $_gc_period = 1000;
1776 if (defined $_repack || defined $_repack_flags) {
1777 warn "Repack options are obsolete; they have no effect.\n";
1778 }
1779}
1780sub verify_remotes_sanity {
1782 return unless -d $ENV{GIT_DIR};
1783 my %seen;
1784 foreach (command(qw/config -l/)) {
1785 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1786 if ($seen{$1}) {
1787 die "Remote ref refs/remote/$1 is tracked by",
1788 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1789 "Please resolve this ambiguity in ",
1790 "your git configuration file before ",
1791 "continuing\n";
1792 }
1793 $seen{$1} = $_;
1794 }
1795 }
1796}
1797sub find_existing_remote {
1799 my ($url, $remotes) = @_;
1800 return undef if $no_reuse_existing;
1801 my $existing;
1802 foreach my $repo_id (keys %$remotes) {
1803 my $u = $remotes->{$repo_id}->{url} or next;
1804 next if $u ne $url;
1805 $existing = $repo_id;
1806 last;
1807 }
1808 $existing;
1809}
1810sub init_remote_config {
1812 my ($self, $url, $no_write) = @_;
1813 $url =~ s!/+$!!; # strip trailing slash
1814 my $r = read_all_remotes();
1815 my $existing = find_existing_remote($url, $r);
1816 if ($existing) {
1817 unless ($no_write) {
1818 print STDERR "Using existing ",
1819 "[svn-remote \"$existing\"]\n";
1820 }
1821 $self->{repo_id} = $existing;
1822 } elsif ($_minimize_url) {
1823 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1824 $existing = find_existing_remote($min_url, $r);
1825 if ($existing) {
1826 unless ($no_write) {
1827 print STDERR "Using existing ",
1828 "[svn-remote \"$existing\"]\n";
1829 }
1830 $self->{repo_id} = $existing;
1831 }
1832 if ($min_url ne $url) {
1833 unless ($no_write) {
1834 print STDERR "Using higher level of URL: ",
1835 "$url => $min_url\n";
1836 }
1837 my $old_path = $self->{path};
1838 $self->{path} = $url;
1839 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1840 if (length $old_path) {
1841 $self->{path} .= "/$old_path";
1842 }
1843 $url = $min_url;
1844 }
1845 }
1846 my $orig_url;
1847 if (!$existing) {
1848 # verify that we aren't overwriting anything:
1849 $orig_url = eval {
1850 command_oneline('config', '--get',
1851 "svn-remote.$self->{repo_id}.url")
1852 };
1853 if ($orig_url && ($orig_url ne $url)) {
1854 die "svn-remote.$self->{repo_id}.url already set: ",
1855 "$orig_url\nwanted to set to: $url\n";
1856 }
1857 }
1858 my ($xrepo_id, $xpath) = find_ref($self->refname);
1859 if (defined $xpath) {
1860 die "svn-remote.$xrepo_id.fetch already set to track ",
1861 "$xpath:refs/remotes/", $self->refname, "\n";
1862 }
1863 unless ($no_write) {
1864 command_noisy('config',
1865 "svn-remote.$self->{repo_id}.url", $url);
1866 $self->{path} =~ s{^/}{};
1867 command_noisy('config', '--add',
1868 "svn-remote.$self->{repo_id}.fetch",
1869 "$self->{path}:".$self->refname);
1870 }
1871 $self->{url} = $url;
1872}
1873sub find_by_url { # repos_root and, path are optional
1875 my ($class, $full_url, $repos_root, $path) = @_;
1876 return undef unless defined $full_url;
1878 remove_username($full_url);
1879 remove_username($repos_root) if defined $repos_root;
1880 my $remotes = read_all_remotes();
1881 if (defined $full_url && defined $repos_root && !defined $path) {
1882 $path = $full_url;
1883 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1884 }
1885 foreach my $repo_id (keys %$remotes) {
1886 my $u = $remotes->{$repo_id}->{url} or next;
1887 remove_username($u);
1888 next if defined $repos_root && $repos_root ne $u;
1889 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1891 foreach my $t (qw/branches tags/) {
1892 foreach my $globspec (@{$remotes->{$repo_id}->{$t}}) {
1893 resolve_local_globs($u, $fetch, $globspec);
1894 }
1895 }
1896 my $p = $path;
1897 my $rwr = rewrite_root({repo_id => $repo_id});
1898 my $svm = $remotes->{$repo_id}->{svm}
1899 if defined $remotes->{$repo_id}->{svm};
1900 unless (defined $p) {
1901 $p = $full_url;
1902 my $z = $u;
1903 my $prefix = '';
1904 if ($rwr) {
1905 $z = $rwr;
1906 remove_username($z);
1907 } elsif (defined $svm) {
1908 $z = $svm->{source};
1909 $prefix = $svm->{replace};
1910 $prefix =~ s#^\Q$u\E(?:/|$)##;
1911 $prefix =~ s#/$##;
1912 }
1913 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1914 }
1915 foreach my $f (keys %$fetch) {
1916 next if $f ne $p;
1917 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1918 }
1919 }
1920 undef;
1921}
1922sub init {
1924 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1925 my $self = _new($class, $repo_id, $ref_id, $path);
1926 if (defined $url) {
1927 $self->init_remote_config($url, $no_write);
1928 }
1929 $self;
1930}
1931sub find_ref {
1933 my ($ref_id) = @_;
1934 foreach (command(qw/config -l/)) {
1935 next unless m!^svn-remote\.(.+)\.fetch=
1936 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1937 my ($repo_id, $path, $ref) = ($1, $2, $3);
1938 if ($ref eq $ref_id) {
1939 $path = '' if ($path =~ m#^\./?#);
1940 return ($repo_id, $path);
1941 }
1942 }
1943 (undef, undef, undef);
1944}
1945sub new {
1947 my ($class, $ref_id, $repo_id, $path) = @_;
1948 if (defined $ref_id && !defined $repo_id && !defined $path) {
1949 ($repo_id, $path) = find_ref($ref_id);
1950 if (!defined $repo_id) {
1951 die "Could not find a \"svn-remote.*.fetch\" key ",
1952 "in the repository configuration matching: ",
1953 "refs/remotes/$ref_id\n";
1954 }
1955 }
1956 my $self = _new($class, $repo_id, $ref_id, $path);
1957 if (!defined $self->{path} || !length $self->{path}) {
1958 my $fetch = command_oneline('config', '--get',
1959 "svn-remote.$repo_id.fetch",
1960 ":refs/remotes/$ref_id\$") or
1961 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1962 "\":refs/remotes/$ref_id\$\" in config\n";
1963 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1964 }
1965 $self->{url} = command_oneline('config', '--get',
1966 "svn-remote.$repo_id.url") or
1967 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1968 $self->rebuild;
1969 $self;
1970}
1971sub refname {
1973 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1974 # It cannot end with a slash /, we'll throw up on this because
1976 # SVN can't have directories with a slash in their name, either:
1977 if ($refname =~ m{/$}) {
1978 die "ref: '$refname' ends with a trailing slash, this is ",
1979 "not permitted by git nor Subversion\n";
1980 }
1981 # It cannot have ASCII control character space, tilde ~, caret ^,
1983 # colon :, question-mark ?, asterisk *, space, or open bracket [
1984 # anywhere.
1985 #
1986 # Additionally, % must be escaped because it is used for escaping
1987 # and we want our escaped refname to be reversible
1988 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1989 # no slash-separated component can begin with a dot .
1991 # /.* becomes /%2E*
1992 $refname =~ s{/\.}{/%2E}g;
1993 # It cannot have two consecutive dots .. anywhere
1995 # .. becomes %2E%2E
1996 $refname =~ s{\.\.}{%2E%2E}g;
1997 return $refname;
1999}
2000sub desanitize_refname {
2002 my ($refname) = @_;
2003 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
2004 return $refname;
2005}
2006sub svm_uuid {
2008 my ($self) = @_;
2009 return $self->{svm}->{uuid} if $self->svm;
2010 $self->ra;
2011 unless ($self->{svm}) {
2012 die "SVM UUID not cached, and reading remotely failed\n";
2013 }
2014 $self->{svm}->{uuid};
2015}
2016sub svm {
2018 my ($self) = @_;
2019 return $self->{svm} if $self->{svm};
2020 my $svm;
2021 # see if we have it in our config, first:
2022 eval {
2023 my $section = "svn-remote.$self->{repo_id}";
2024 $svm = {
2025 source => tmp_config('--get', "$section.svm-source"),
2026 uuid => tmp_config('--get', "$section.svm-uuid"),
2027 replace => tmp_config('--get', "$section.svm-replace"),
2028 }
2029 };
2030 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
2031 $self->{svm} = $svm;
2032 }
2033 $self->{svm};
2034}
2035sub _set_svm_vars {
2037 my ($self, $ra) = @_;
2038 return $ra if $self->svm;
2039 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
2041 "(svm:source, svm:uuid) ",
2042 "from the following URLs:\n" );
2043 sub read_svm_props {
2044 my ($self, $ra, $path, $r) = @_;
2045 my $props = ($ra->get_dir($path, $r))[2];
2046 my $src = $props->{'svm:source'};
2047 my $uuid = $props->{'svm:uuid'};
2048 return undef if (!$src || !$uuid);
2049 chomp($src, $uuid);
2051 $uuid =~ m{^[0-9a-f\-]{30,}$}i
2053 or die "doesn't look right - svm:uuid is '$uuid'\n";
2054 # the '!' is used to mark the repos_root!/relative/path
2056 $src =~ s{/?!/?}{/};
2057 $src =~ s{/+$}{}; # no trailing slashes please
2058 # username is of no interest
2059 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
2060 my $replace = $ra->{url};
2062 $replace .= "/$path" if length $path;
2063 my $section = "svn-remote.$self->{repo_id}";
2065 tmp_config("$section.svm-source", $src);
2066 tmp_config("$section.svm-replace", $replace);
2067 tmp_config("$section.svm-uuid", $uuid);
2068 $self->{svm} = {
2069 source => $src,
2070 uuid => $uuid,
2071 replace => $replace
2072 };
2073 }
2074 my $r = $ra->get_latest_revnum;
2076 my $path = $self->{path};
2077 my %tried;
2078 while (length $path) {
2079 unless ($tried{"$self->{url}/$path"}) {
2080 return $ra if $self->read_svm_props($ra, $path, $r);
2081 $tried{"$self->{url}/$path"} = 1;
2082 }
2083 $path =~ s#/?[^/]+$##;
2084 }
2085 die "Path: '$path' should be ''\n" if $path ne '';
2086 return $ra if $self->read_svm_props($ra, $path, $r);
2087 $tried{"$self->{url}/$path"} = 1;
2088 if ($ra->{repos_root} eq $self->{url}) {
2090 die @err, (map { " $_\n" } keys %tried), "\n";
2091 }
2092 # nope, make sure we're connected to the repository root:
2094 my $ok;
2095 my @tried_b;
2096 $path = $ra->{svn_path};
2097 $ra = Git::SVN::Ra->new($ra->{repos_root});
2098 while (length $path) {
2099 unless ($tried{"$ra->{url}/$path"}) {
2100 $ok = $self->read_svm_props($ra, $path, $r);
2101 last if $ok;
2102 $tried{"$ra->{url}/$path"} = 1;
2103 }
2104 $path =~ s#/?[^/]+$##;
2105 }
2106 die "Path: '$path' should be ''\n" if $path ne '';
2107 $ok ||= $self->read_svm_props($ra, $path, $r);
2108 $tried{"$ra->{url}/$path"} = 1;
2109 if (!$ok) {
2110 die @err, (map { " $_\n" } keys %tried), "\n";
2111 }
2112 Git::SVN::Ra->new($self->{url});
2113}
2114sub svnsync {
2116 my ($self) = @_;
2117 return $self->{svnsync} if $self->{svnsync};
2118 if ($self->no_metadata) {
2120 die "Can't have both 'noMetadata' and ",
2121 "'useSvnsyncProps' options set!\n";
2122 }
2123 if ($self->rewrite_root) {
2124 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
2125 "options set!\n";
2126 }
2127 my $svnsync;
2129 # see if we have it in our config, first:
2130 eval {
2131 my $section = "svn-remote.$self->{repo_id}";
2132 my $url = tmp_config('--get', "$section.svnsync-url");
2134 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2135 die "doesn't look right - svn:sync-from-url is '$url'\n";
2136 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
2138 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2139 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2140 $svnsync = { url => $url, uuid => $uuid }
2142 };
2143 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
2144 return $self->{svnsync} = $svnsync;
2145 }
2146 my $err = "useSvnsyncProps set, but failed to read " .
2148 "svnsync property: svn:sync-from-";
2149 my $rp = $self->ra->rev_proplist(0);
2150 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
2152 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
2153 die "doesn't look right - svn:sync-from-url is '$url'\n";
2154 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
2156 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}i) or
2157 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
2158 my $section = "svn-remote.$self->{repo_id}";
2160 tmp_config('--add', "$section.svnsync-uuid", $uuid);
2161 tmp_config('--add', "$section.svnsync-url", $url);
2162 return $self->{svnsync} = { url => $url, uuid => $uuid };
2163}
2164# this allows us to memoize our SVN::Ra UUID locally and avoid a
2166# remote lookup (useful for 'git svn log').
2167sub ra_uuid {
2168 my ($self) = @_;
2169 unless ($self->{ra_uuid}) {
2170 my $key = "svn-remote.$self->{repo_id}.uuid";
2171 my $uuid = eval { tmp_config('--get', $key) };
2172 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/i) {
2173 $self->{ra_uuid} = $uuid;
2174 } else {
2175 die "ra_uuid called without URL\n" unless $self->{url};
2176 $self->{ra_uuid} = $self->ra->get_uuid;
2177 tmp_config('--add', $key, $self->{ra_uuid});
2178 }
2179 }
2180 $self->{ra_uuid};
2181}
2182sub _set_repos_root {
2184 my ($self, $repos_root) = @_;
2185 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2186 $repos_root ||= $self->ra->{repos_root};
2187 tmp_config($k, $repos_root);
2188 $repos_root;
2189}
2190sub repos_root {
2192 my ($self) = @_;
2193 my $k = "svn-remote.$self->{repo_id}.reposRoot";
2194 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
2195}
2196sub ra {
2198 my ($self) = shift;
2199 my $ra = Git::SVN::Ra->new($self->{url});
2200 $self->_set_repos_root($ra->{repos_root});
2201 if ($self->use_svm_props && !$self->{svm}) {
2202 if ($self->no_metadata) {
2203 die "Can't have both 'noMetadata' and ",
2204 "'useSvmProps' options set!\n";
2205 } elsif ($self->use_svnsync_props) {
2206 die "Can't have both 'useSvnsyncProps' and ",
2207 "'useSvmProps' options set!\n";
2208 }
2209 $ra = $self->_set_svm_vars($ra);
2210 $self->{-want_revprops} = 1;
2211 }
2212 $ra;
2213}
2214# prop_walk(PATH, REV, SUB)
2216# -------------------------
2217# Recursively traverse PATH at revision REV and invoke SUB for each
2218# directory that contains a SVN property. SUB will be invoked as
2219# follows: &SUB(gs, path, props); where `gs' is this instance of
2220# Git::SVN, `path' the path to the directory where the properties
2221# `props' were found. The `path' will be relative to point of checkout,
2222# that is, if url://repo/trunk is the current Git branch, and that
2223# directory contains a sub-directory `d', SUB will be invoked with `/d/'
2224# as `path' (note the trailing `/').
2225sub prop_walk {
2226 my ($self, $path, $rev, $sub) = @_;
2227 $path =~ s#^/##;
2229 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
2230 $path =~ s#^/*#/#g;
2231 my $p = $path;
2232 # Strip the irrelevant part of the path.
2233 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
2234 # Ensure the path is terminated by a `/'.
2235 $p =~ s#/*$#/#;
2236 # The properties contain all the internal SVN stuff nobody
2238 # (usually) cares about.
2239 my $interesting_props = 0;
2240 foreach (keys %{$props}) {
2241 # If it doesn't start with `svn:', it must be a
2242 # user-defined property.
2243 ++$interesting_props and next if $_ !~ /^svn:/;
2244 # FIXME: Fragile, if SVN adds new public properties,
2245 # this needs to be updated.
2246 ++$interesting_props if /^svn:(?:ignore|keywords|executable
2247 |eol-style|mime-type
2248 |externals|needs-lock)$/x;
2249 }
2250 &$sub($self, $p, $props) if $interesting_props;
2251 foreach (sort keys %$dirent) {
2253 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
2254 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
2255 }
2256}
2257sub last_rev { ($_[0]->last_rev_commit)[0] }
2259sub last_commit { ($_[0]->last_rev_commit)[1] }
2260# returns the newest SVN revision number and newest commit SHA1
2262sub last_rev_commit {
2263 my ($self) = @_;
2264 if (defined $self->{last_rev} && defined $self->{last_commit}) {
2265 return ($self->{last_rev}, $self->{last_commit});
2266 }
2267 my $c = ::verify_ref($self->refname.'^0');
2268 if ($c && !$self->use_svm_props && !$self->no_metadata) {
2269 my $rev = (::cmt_metadata($c))[1];
2270 if (defined $rev) {
2271 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
2272 return ($rev, $c);
2273 }
2274 }
2275 my $map_path = $self->map_path;
2276 unless (-e $map_path) {
2277 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
2278 return (undef, undef);
2279 }
2280 my ($rev, $commit) = $self->rev_map_max(1);
2281 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
2282 return ($rev, $commit);
2283}
2284sub get_fetch_range {
2286 my ($self, $min, $max) = @_;
2287 $max ||= $self->ra->get_latest_revnum;
2288 $min ||= $self->rev_map_max;
2289 (++$min, $max);
2290}
2291sub tmp_config {
2293 my (@args) = @_;
2294 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
2295 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2296 if (! -f $config && -f $old_def_config) {
2297 rename $old_def_config, $config or
2298 die "Failed rename $old_def_config => $config: $!\n";
2299 }
2300 my $old_config = $ENV{GIT_CONFIG};
2301 $ENV{GIT_CONFIG} = $config;
2302 $@ = undef;
2303 my @ret = eval {
2304 unless (-f $config) {
2305 mkfile($config);
2306 open my $fh, '>', $config or
2307 die "Can't open $config: $!\n";
2308 print $fh "; This file is used internally by ",
2309 "git-svn\n" or die
2310 "Couldn't write to $config: $!\n";
2311 print $fh "; You should not have to edit it\n" or
2312 die "Couldn't write to $config: $!\n";
2313 close $fh or die "Couldn't close $config: $!\n";
2314 }
2315 command('config', @args);
2316 };
2317 my $err = $@;
2318 if (defined $old_config) {
2319 $ENV{GIT_CONFIG} = $old_config;
2320 } else {
2321 delete $ENV{GIT_CONFIG};
2322 }
2323 die $err if $err;
2324 wantarray ? @ret : $ret[0];
2325}
2326sub tmp_index_do {
2328 my ($self, $sub) = @_;
2329 my $old_index = $ENV{GIT_INDEX_FILE};
2330 $ENV{GIT_INDEX_FILE} = $self->{index};
2331 $@ = undef;
2332 my @ret = eval {
2333 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2334 mkpath([$dir]) unless -d $dir;
2335 &$sub;
2336 };
2337 my $err = $@;
2338 if (defined $old_index) {
2339 $ENV{GIT_INDEX_FILE} = $old_index;
2340 } else {
2341 delete $ENV{GIT_INDEX_FILE};
2342 }
2343 die $err if $err;
2344 wantarray ? @ret : $ret[0];
2345}
2346sub assert_index_clean {
2348 my ($self, $treeish) = @_;
2349 $self->tmp_index_do(sub {
2351 command_noisy('read-tree', $treeish) unless -e $self->{index};
2352 my $x = command_oneline('write-tree');
2353 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2354 /^tree ($::sha1)/mo);
2355 return if $y eq $x;
2356 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2358 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2359 command_noisy('read-tree', $treeish);
2360 $x = command_oneline('write-tree');
2361 if ($y ne $x) {
2362 ::fatal "trees ($treeish) $y != $x\n",
2363 "Something is seriously wrong...";
2364 }
2365 });
2366}
2367sub get_commit_parents {
2369 my ($self, $log_entry) = @_;
2370 my (%seen, @ret, @tmp);
2371 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2372 if (my $ip = $self->{inject_parents}) {
2373 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2374 push @tmp, $commit;
2375 }
2376 }
2377 if (my $cur = ::verify_ref($self->refname.'^0')) {
2378 push @tmp, $cur;
2379 }
2380 if (my $ipd = $self->{inject_parents_dcommit}) {
2381 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2382 push @tmp, @$commit;
2383 }
2384 }
2385 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2386 while (my $p = shift @tmp) {
2387 next if $seen{$p};
2388 $seen{$p} = 1;
2389 push @ret, $p;
2390 # MAXPARENT is defined to 16 in commit-tree.c:
2391 last if @ret >= 16;
2392 }
2393 if (@tmp) {
2394 die "r$log_entry->{revision}: No room for parents:\n\t",
2395 join("\n\t", @tmp), "\n";
2396 }
2397 @ret;
2398}
2399sub rewrite_root {
2401 my ($self) = @_;
2402 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2403 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2404 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2405 if ($rwr) {
2406 $rwr =~ s#/+$##;
2407 if ($rwr !~ m#^[a-z\+]+://#) {
2408 die "$rwr is not a valid URL (key: $k)\n";
2409 }
2410 }
2411 $self->{-rewrite_root} = $rwr;
2412}
2413sub metadata_url {
2415 my ($self) = @_;
2416 ($self->rewrite_root || $self->{url}) .
2417 (length $self->{path} ? '/' . $self->{path} : '');
2418}
2419sub full_url {
2421 my ($self) = @_;
2422 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2423}
2424sub set_commit_header_env {
2427 my ($log_entry) = @_;
2428 my %env;
2429 foreach my $ned (qw/NAME EMAIL DATE/) {
2430 foreach my $ac (qw/AUTHOR COMMITTER/) {
2431 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2432 }
2433 }
2434 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2436 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2437 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2438 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2440 ? $log_entry->{commit_name}
2441 : $log_entry->{name};
2442 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2443 ? $log_entry->{commit_email}
2444 : $log_entry->{email};
2445 \%env;
2446}
2447sub restore_commit_header_env {
2449 my ($env) = @_;
2450 foreach my $ned (qw/NAME EMAIL DATE/) {
2451 foreach my $ac (qw/AUTHOR COMMITTER/) {
2452 my $k = "GIT_${ac}_${ned}";
2453 if (defined $env->{$k}) {
2454 $ENV{$k} = $env->{$k};
2455 } else {
2456 delete $ENV{$k};
2457 }
2458 }
2459 }
2460}
2461sub gc {
2463 command_noisy('gc', '--auto');
2464};
2465sub do_git_commit {
2467 my ($self, $log_entry) = @_;
2468 my $lr = $self->last_rev;
2469 if (defined $lr && $lr >= $log_entry->{revision}) {
2470 die "Last fetched revision of ", $self->refname,
2471 " was r$lr, but we are about to fetch: ",
2472 "r$log_entry->{revision}!\n";
2473 }
2474 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2475 croak "$log_entry->{revision} = $c already exists! ",
2476 "Why are we refetching it?\n";
2477 }
2478 my $old_env = set_commit_header_env($log_entry);
2479 my $tree = $log_entry->{tree};
2480 if (!defined $tree) {
2481 $tree = $self->tmp_index_do(sub {
2482 command_oneline('write-tree') });
2483 }
2484 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2485 my @exec = ('git', 'commit-tree', $tree);
2487 foreach ($self->get_commit_parents($log_entry)) {
2488 push @exec, '-p', $_;
2489 }
2490 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2491 or croak $!;
2492 binmode $msg_fh;
2493 # we always get UTF-8 from SVN, but we may want our commits in
2495 # a different encoding.
2496 if (my $enc = Git::config('i18n.commitencoding')) {
2497 require Encode;
2498 Encode::from_to($log_entry->{log}, 'UTF-8', $enc);
2499 }
2500 print $msg_fh $log_entry->{log} or croak $!;
2501 restore_commit_header_env($old_env);
2502 unless ($self->no_metadata) {
2503 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2504 or croak $!;
2505 }
2506 $msg_fh->flush == 0 or croak $!;
2507 close $msg_fh or croak $!;
2508 chomp(my $commit = do { local $/; <$out_fh> });
2509 close $out_fh or croak $!;
2510 waitpid $pid, 0;
2511 croak $? if $?;
2512 if ($commit !~ /^$::sha1$/o) {
2513 die "Failed to commit, invalid sha1: $commit\n";
2514 }
2515 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2517 $self->{last_rev} = $log_entry->{revision};
2519 $self->{last_commit} = $commit;
2520 print "r$log_entry->{revision}" unless $::_q > 1;
2521 if (defined $log_entry->{svm_revision}) {
2522 print " (\@$log_entry->{svm_revision})" unless $::_q > 1;
2523 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2524 0, $self->svm_uuid);
2525 }
2526 print " = $commit ($self->{ref_id})\n" unless $::_q > 1;
2527 if (--$_gc_nr == 0) {
2528 $_gc_nr = $_gc_period;
2529 gc();
2530 }
2531 return $commit;
2532}
2533sub match_paths {
2535 my ($self, $paths, $r) = @_;
2536 return 1 if $self->{path} eq '';
2537 if (my $path = $paths->{"/$self->{path}"}) {
2538 return ($path->{action} eq 'D') ? 0 : 1;
2539 }
2540 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2541 if (grep /$self->{path_regex}/, keys %$paths) {
2542 return 1;
2543 }
2544 my $c = '';
2545 foreach (split m#/#, $self->{path}) {
2546 $c .= "/$_";
2547 next unless ($paths->{$c} &&
2548 ($paths->{$c}->{action} =~ /^[AR]$/));
2549 if ($self->ra->check_path($self->{path}, $r) ==
2550 $SVN::Node::dir) {
2551 return 1;
2552 }
2553 }
2554 return 0;
2555}
2556sub find_parent_branch {
2558 my ($self, $paths, $rev) = @_;
2559 return undef unless $self->follow_parent;
2560 unless (defined $paths) {
2561 my $err_handler = $SVN::Error::handler;
2562 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2563 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1,
2564 sub { $paths = $_[0] });
2565 $SVN::Error::handler = $err_handler;
2566 }
2567 return undef unless defined $paths;
2568 # look for a parent from another branch:
2570 my @b_path_components = split m#/#, $self->{path};
2571 my @a_path_components;
2572 my $i;
2573 while (@b_path_components) {
2574 $i = $paths->{'/'.join('/', @b_path_components)};
2575 last if $i && defined $i->{copyfrom_path};
2576 unshift(@a_path_components, pop(@b_path_components));
2577 }
2578 return undef unless defined $i && defined $i->{copyfrom_path};
2579 my $branch_from = $i->{copyfrom_path};
2580 if (@a_path_components) {
2581 print STDERR "branch_from: $branch_from => ";
2582 $branch_from .= '/'.join('/', @a_path_components);
2583 print STDERR $branch_from, "\n";
2584 }
2585 my $r = $i->{copyfrom_rev};
2586 my $repos_root = $self->ra->{repos_root};
2587 my $url = $self->ra->{url};
2588 my $new_url = $url . $branch_from;
2589 print STDERR "Found possible branch point: ",
2590 "$new_url => ", $self->full_url, ", $r\n";
2591 $branch_from =~ s#^/##;
2592 my $gs = $self->other_gs($new_url, $url,
2593 $branch_from, $r, $self->{ref_id});
2594 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2595 {
2596 my ($base, $head);
2597 if (!defined $r0 || !defined $parent) {
2598 ($base, $head) = parse_revision_argument(0, $r);
2599 } else {
2600 if ($r0 < $r) {
2601 $gs->ra->get_log([$gs->{path}], $r0 + 1, $r, 1,
2602 0, 1, sub { $base = $_[1] - 1 });
2603 }
2604 }
2605 if (defined $base && $base <= $r) {
2606 $gs->fetch($base, $r);
2607 }
2608 ($r0, $parent) = $gs->find_rev_before($r, 1);
2609 }
2610 if (defined $r0 && defined $parent) {
2611 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2612 my $ed;
2613 if ($self->ra->can_do_switch) {
2614 $self->assert_index_clean($parent);
2615 print STDERR "Following parent with do_switch\n";
2616 # do_switch works with svn/trunk >= r22312, but that
2617 # is not included with SVN 1.4.3 (the latest version
2618 # at the moment), so we can't rely on it
2619 $self->{last_rev} = $r0;
2620 $self->{last_commit} = $parent;
2621 $ed = SVN::Git::Fetcher->new($self, $gs->{path});
2622 $gs->ra->gs_do_switch($r0, $rev, $gs,
2623 $self->full_url, $ed)
2624 or die "SVN connection failed somewhere...\n";
2625 } elsif ($self->ra->trees_match($new_url, $r0,
2626 $self->full_url, $rev)) {
2627 print STDERR "Trees match:\n",
2628 " $new_url\@$r0\n",
2629 " ${\$self->full_url}\@$rev\n",
2630 "Following parent with no changes\n";
2631 $self->tmp_index_do(sub {
2632 command_noisy('read-tree', $parent);
2633 });
2634 $self->{last_commit} = $parent;
2635 } else {
2636 print STDERR "Following parent with do_update\n";
2637 $ed = SVN::Git::Fetcher->new($self);
2638 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2639 or die "SVN connection failed somewhere...\n";
2640 }
2641 print STDERR "Successfully followed parent\n";
2642 return $self->make_log_entry($rev, [$parent], $ed);
2643 }
2644 return undef;
2645}
2646sub do_fetch {
2648 my ($self, $paths, $rev) = @_;
2649 my $ed;
2650 my ($last_rev, @parents);
2651 if (my $lc = $self->last_commit) {
2652 # we can have a branch that was deleted, then re-added
2653 # under the same name but copied from another path, in
2654 # which case we'll have multiple parents (we don't
2655 # want to break the original ref, nor lose copypath info):
2656 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2657 push @{$log_entry->{parents}}, $lc;
2658 return $log_entry;
2659 }
2660 $ed = SVN::Git::Fetcher->new($self);
2661 $last_rev = $self->{last_rev};
2662 $ed->{c} = $lc;
2663 @parents = ($lc);
2664 } else {
2665 $last_rev = $rev;
2666 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2667 return $log_entry;
2668 }
2669 $ed = SVN::Git::Fetcher->new($self);
2670 }
2671 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2672 die "SVN connection failed somewhere...\n";
2673 }
2674 $self->make_log_entry($rev, \@parents, $ed);
2675}
2676sub get_untracked {
2678 my ($self, $ed) = @_;
2679 my @out;
2680 my $h = $ed->{empty};
2681 foreach (sort keys %$h) {
2682 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2683 push @out, " $act: " . uri_encode($_);
2684 warn "W: $act: $_\n";
2685 }
2686 foreach my $t (qw/dir_prop file_prop/) {
2687 $h = $ed->{$t} or next;
2688 foreach my $path (sort keys %$h) {
2689 my $ppath = $path eq '' ? '.' : $path;
2690 foreach my $prop (sort keys %{$h->{$path}}) {
2691 next if $SKIP_PROP{$prop};
2692 my $v = $h->{$path}->{$prop};
2693 my $t_ppath_prop = "$t: " .
2694 uri_encode($ppath) . ' ' .
2695 uri_encode($prop);
2696 if (defined $v) {
2697 push @out, " +$t_ppath_prop " .
2698 uri_encode($v);
2699 } else {
2700 push @out, " -$t_ppath_prop";
2701 }
2702 }
2703 }
2704 }
2705 foreach my $t (qw/absent_file absent_directory/) {
2706 $h = $ed->{$t} or next;
2707 foreach my $parent (sort keys %$h) {
2708 foreach my $path (sort @{$h->{$parent}}) {
2709 push @out, " $t: " .
2710 uri_encode("$parent/$path");
2711 warn "W: $t: $parent/$path ",
2712 "Insufficient permissions?\n";
2713 }
2714 }
2715 }
2716 \@out;
2717}
2718# parse_svn_date(DATE)
2720# --------------------
2721# Given a date (in UTC) from Subversion, return a string in the format
2722# "<TZ Offset> <local date/time>" that Git will use.
2723#
2724# By default the parsed date will be in UTC; if $Git::SVN::_localtime
2725# is true we'll convert it to the local timezone instead.
2726sub parse_svn_date {
2727 my $date = shift || return '+0000 1970-01-01 00:00:00';
2728 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2729 (\d\d)\:(\d\d)\:(\d\d)\.\d*Z$/x) or
2730 croak "Unable to parse date: $date\n";
2731 my $parsed_date; # Set next.
2732 if ($Git::SVN::_localtime) {
2734 # Translate the Subversion datetime to an epoch time.
2735 # Begin by switching ourselves to $date's timezone, UTC.
2736 my $old_env_TZ = $ENV{TZ};
2737 $ENV{TZ} = 'UTC';
2738 my $epoch_in_UTC =
2740 POSIX::strftime('%s', $S, $M, $H, $d, $m - 1, $Y - 1900);
2741 # Determine our local timezone (including DST) at the
2743 # time of $epoch_in_UTC. $Git::SVN::Log::TZ stored the
2744 # value of TZ, if any, at the time we were run.
2745 if (defined $Git::SVN::Log::TZ) {
2746 $ENV{TZ} = $Git::SVN::Log::TZ;
2747 } else {
2748 delete $ENV{TZ};
2749 }
2750 my $our_TZ =
2752 POSIX::strftime('%Z', $S, $M, $H, $d, $m - 1, $Y - 1900);
2753 # This converts $epoch_in_UTC into our local timezone.
2755 my ($sec, $min, $hour, $mday, $mon, $year,
2756 $wday, $yday, $isdst) = localtime($epoch_in_UTC);
2757 $parsed_date = sprintf('%s %04d-%02d-%02d %02d:%02d:%02d',
2759 $our_TZ, $year + 1900, $mon + 1,
2760 $mday, $hour, $min, $sec);
2761 # Reset us to the timezone in effect when we entered
2763 # this routine.
2764 if (defined $old_env_TZ) {
2765 $ENV{TZ} = $old_env_TZ;
2766 } else {
2767 delete $ENV{TZ};
2768 }
2769 } else {
2770 $parsed_date = "+0000 $Y-$m-$d $H:$M:$S";
2771 }
2772 return $parsed_date;
2774}
2775sub other_gs {
2777 my ($self, $new_url, $url,
2778 $branch_from, $r, $old_ref_id) = @_;
2779 my $gs = Git::SVN->find_by_url($new_url, $url, $branch_from);
2780 unless ($gs) {
2781 my $ref_id = $old_ref_id;
2782 $ref_id =~ s/\@\d+$//;
2783 $ref_id .= "\@$r";
2784 # just grow a tail if we're not unique enough :x
2785 $ref_id .= '-' while find_ref($ref_id);
2786 print STDERR "Initializing parent: $ref_id\n";
2787 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2788 if ($u =~ s#^\Q$url\E(/|$)##) {
2789 $p = $u;
2790 $u = $url;
2791 $repo_id = $self->{repo_id};
2792 }
2793 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2794 }
2795 $gs
2796}
2797sub call_authors_prog {
2799 my ($orig_author) = @_;
2800 my $author = `$::_authors_prog $orig_author`;
2801 if ($? != 0) {
2802 die "$::_authors_prog failed with exit code $?\n"
2803 }
2804 if ($author =~ /^\s*(.+?)\s*<(.*)>\s*$/) {
2805 my ($name, $email) = ($1, $2);
2806 $email = undef if length $2 == 0;
2807 return [$name, $email];
2808 } else {
2809 die "Author: $orig_author: $::_authors_prog returned "
2810 . "invalid author format: $author\n";
2811 }
2812}
2813sub check_author {
2815 my ($author) = @_;
2816 if (!defined $author || length $author == 0) {
2817 $author = '(no author)';
2818 }
2819 if (!defined $::users{$author}) {
2820 if (defined $::_authors_prog) {
2821 $::users{$author} = call_authors_prog($author);
2822 } elsif (defined $::_authors) {
2823 die "Author: $author not defined in $::_authors file\n";
2824 }
2825 }
2826 $author;
2827}
2828sub make_log_entry {
2830 my ($self, $rev, $parents, $ed) = @_;
2831 my $untracked = $self->get_untracked($ed);
2832 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2834 print $un "r$rev\n" or croak $!;
2835 print $un $_, "\n" foreach @$untracked;
2836 my %log_entry = ( parents => $parents || [], revision => $rev,
2837 log => '');
2838 my $headrev;
2840 my $logged = delete $self->{logged_rev_props};
2841 if (!$logged || $self->{-want_revprops}) {
2842 my $rp = $self->ra->rev_proplist($rev);
2843 foreach (sort keys %$rp) {
2844 my $v = $rp->{$_};
2845 if (/^svn:(author|date|log)$/) {
2846 $log_entry{$1} = $v;
2847 } elsif ($_ eq 'svm:headrev') {
2848 $headrev = $v;
2849 } else {
2850 print $un " rev_prop: ", uri_encode($_), ' ',
2851 uri_encode($v), "\n";
2852 }
2853 }
2854 } else {
2855 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2856 }
2857 close $un or croak $!;
2858 $log_entry{date} = parse_svn_date($log_entry{date});
2860 $log_entry{log} .= "\n";
2861 my $author = $log_entry{author} = check_author($log_entry{author});
2862 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2863 : ($author, undef);
2864 my ($commit_name, $commit_email) = ($name, $email);
2866 if ($_use_log_author) {
2867 my $name_field;
2868 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2869 $name_field = $1;
2870 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2871 $name_field = $1;
2872 }
2873 if (!defined $name_field) {
2874 if (!defined $email) {
2875 $email = $name;
2876 }
2877 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2878 ($name, $email) = ($1, $2);
2879 } elsif ($name_field =~ /(.*)@/) {
2880 ($name, $email) = ($1, $name_field);
2881 } else {
2882 ($name, $email) = ($name_field, $name_field);
2883 }
2884 }
2885 if (defined $headrev && $self->use_svm_props) {
2886 if ($self->rewrite_root) {
2887 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2888 "options set!\n";
2889 }
2890 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$}i;
2891 # we don't want "SVM: initializing mirror for junk" ...
2892 return undef if $r == 0;
2893 my $svm = $self->svm;
2894 if ($uuid ne $svm->{uuid}) {
2895 die "UUID mismatch on SVM path:\n",
2896 "expected: $svm->{uuid}\n",
2897 " got: $uuid\n";
2898 }
2899 my $full_url = $self->full_url;
2900 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2901 die "Failed to replace '$svm->{replace}' with ",
2902 "'$svm->{source}' in $full_url\n";
2903 # throw away username for storing in records
2904 remove_username($full_url);
2905 $log_entry{metadata} = "$full_url\@$r $uuid";
2906 $log_entry{svm_revision} = $r;
2907 $email ||= "$author\@$uuid";
2908 $commit_email ||= "$author\@$uuid";
2909 } elsif ($self->use_svnsync_props) {
2910 my $full_url = $self->svnsync->{url};
2911 $full_url .= "/$self->{path}" if length $self->{path};
2912 remove_username($full_url);
2913 my $uuid = $self->svnsync->{uuid};
2914 $log_entry{metadata} = "$full_url\@$rev $uuid";
2915 $email ||= "$author\@$uuid";
2916 $commit_email ||= "$author\@$uuid";
2917 } else {
2918 my $url = $self->metadata_url;
2919 remove_username($url);
2920 $log_entry{metadata} = "$url\@$rev " .
2921 $self->ra->get_uuid;
2922 $email ||= "$author\@" . $self->ra->get_uuid;
2923 $commit_email ||= "$author\@" . $self->ra->get_uuid;
2924 }
2925 $log_entry{name} = $name;
2926 $log_entry{email} = $email;
2927 $log_entry{commit_name} = $commit_name;
2928 $log_entry{commit_email} = $commit_email;
2929 \%log_entry;
2930}
2931sub fetch {
2933 my ($self, $min_rev, $max_rev, @parents) = @_;
2934 my ($last_rev, $last_commit) = $self->last_rev_commit;
2935 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2936 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2937}
2938sub set_tree_cb {
2940 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2941 $self->{inject_parents} = { $rev => $tree };
2942 $self->fetch(undef, undef);
2943}
2944sub set_tree {
2946 my ($self, $tree) = (shift, shift);
2947 my $log_entry = ::get_commit_entry($tree);
2948 unless ($self->{last_rev}) {
2949 ::fatal("Must have an existing revision to commit");
2950 }
2951 my %ed_opts = ( r => $self->{last_rev},
2952 log => $log_entry->{log},
2953 ra => $self->ra,
2954 tree_a => $self->{last_commit},
2955 tree_b => $tree,
2956 editor_cb => sub {
2957 $self->set_tree_cb($log_entry, $tree, @_) },
2958 svn_path => $self->{path} );
2959 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2960 print "No changes\nr$self->{last_rev} = $tree\n";
2961 }
2962}
2963sub rebuild_from_rev_db {
2965 my ($self, $path) = @_;
2966 my $r = -1;
2967 open my $fh, '<', $path or croak "open: $!";
2968 binmode $fh or croak "binmode: $!";
2969 while (<$fh>) {
2970 length($_) == 41 or croak "inconsistent size in ($_) != 41";
2971 chomp($_);
2972 ++$r;
2973 next if $_ eq ('0' x 40);
2974 $self->rev_map_set($r, $_);
2975 print "r$r = $_\n";
2976 }
2977 close $fh or croak "close: $!";
2978 unlink $path or croak "unlink: $!";
2979}
2980sub rebuild {
2982 my ($self) = @_;
2983 my $map_path = $self->map_path;
2984 my $partial = (-e $map_path && ! -z $map_path);
2985 return unless ::verify_ref($self->refname.'^0');
2986 if (!$partial && ($self->use_svm_props || $self->no_metadata)) {
2987 my $rev_db = $self->rev_db_path;
2988 $self->rebuild_from_rev_db($rev_db);
2989 if ($self->use_svm_props) {
2990 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2991 $self->rebuild_from_rev_db($svm_rev_db);
2992 }
2993 $self->unlink_rev_db_symlink;
2994 return;
2995 }
2996 print "Rebuilding $map_path ...\n" if (!$partial);
2997 my ($base_rev, $head) = ($partial ? $self->rev_map_max_norebuild(1) :
2998 (undef, undef));
2999 my ($log, $ctx) =
3000 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
3001 ($head ? "$head.." : "") . $self->refname,
3002 '--');
3003 my $metadata_url = $self->metadata_url;
3004 remove_username($metadata_url);
3005 my $svn_uuid = $self->ra_uuid;
3006 my $c;
3007 while (<$log>) {
3008 if ( m{^commit ($::sha1)$} ) {
3009 $c = $1;
3010 next;
3011 }
3012 next unless s{^\s*(git-svn-id:)}{$1};
3013 my ($url, $rev, $uuid) = ::extract_metadata($_);
3014 remove_username($url);
3015 # ignore merges (from set-tree)
3017 next if (!defined $rev || !$uuid);
3018 # if we merged or otherwise started elsewhere, this is
3020 # how we break out of it
3021 if (($uuid ne $svn_uuid) ||
3022 ($metadata_url && $url && ($url ne $metadata_url))) {
3023 next;
3024 }
3025 if ($partial && $head) {
3026 print "Partial-rebuilding $map_path ...\n";
3027 print "Currently at $base_rev = $head\n";
3028 $head = undef;
3029 }
3030 $self->rev_map_set($rev, $c);
3032 print "r$rev = $c\n";
3033 }
3034 command_close_pipe($log, $ctx);
3035 print "Done rebuilding $map_path\n" if (!$partial || !$head);
3036 my $rev_db_path = $self->rev_db_path;
3037 if (-f $self->rev_db_path) {
3038 unlink $self->rev_db_path or croak "unlink: $!";
3039 }
3040 $self->unlink_rev_db_symlink;
3041}
3042# rev_map:
3044# Tie::File seems to be prone to offset errors if revisions get sparse,
3045# it's not that fast, either. Tie::File is also not in Perl 5.6. So
3046# one of my favorite modules is out :< Next up would be one of the DBM
3047# modules, but I'm not sure which is most portable...
3048#
3049# This is the replacement for the rev_db format, which was too big
3050# and inefficient for large repositories with a lot of sparse history
3051# (mainly tags)
3052#
3053# The format is this:
3054# - 24 bytes for every record,
3055# * 4 bytes for the integer representing an SVN revision number
3056# * 20 bytes representing the sha1 of a git commit
3057# - No empty padding records like the old format
3058# (except the last record, which can be overwritten)
3059# - new records are written append-only since SVN revision numbers
3060# increase monotonically
3061# - lookups on SVN revision number are done via a binary search
3062# - Piping the file to xxd -c24 is a good way of dumping it for
3063# viewing or editing (piped back through xxd -r), should the need
3064# ever arise.
3065# - The last record can be padding revision with an all-zero sha1
3066# This is used to optimize fetch performance when using multiple
3067# "fetch" directives in .git/config
3068#
3069# These files are disposable unless noMetadata or useSvmProps is set
3070sub _rev_map_set {
3072 my ($fh, $rev, $commit) = @_;
3073 binmode $fh or croak "binmode: $!";
3075 my $size = (stat($fh))[7];
3076 ($size % 24) == 0 or croak "inconsistent size: $size";
3077 my $wr_offset = 0;
3079 if ($size > 0) {
3080 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3081 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
3082 $read == 24 or croak "read only $read bytes (!= 24)";
3083 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
3084 if ($last_commit eq ('0' x40)) {
3085 if ($size >= 48) {
3086 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3087 $read = sysread($fh, $buf, 24) or
3088 croak "read: $!";
3089 $read == 24 or
3090 croak "read only $read bytes (!= 24)";
3091 ($last_rev, $last_commit) =
3092 unpack(rev_map_fmt, $buf);
3093 if ($last_commit eq ('0' x40)) {
3094 croak "inconsistent .rev_map\n";
3095 }
3096 }
3097 if ($last_rev >= $rev) {
3098 croak "last_rev is higher!: $last_rev >= $rev";
3099 }
3100 $wr_offset = -24;
3101 }
3102 }
3103 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
3104 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
3105 croak "write: $!";
3106}
3107sub _rev_map_reset {
3109 my ($fh, $rev, $commit) = @_;
3110 my $c = _rev_map_get($fh, $rev);
3111 $c eq $commit or die "_rev_map_reset(@_) commit $c does not match!\n";
3112 my $offset = sysseek($fh, 0, SEEK_CUR) or croak "seek: $!";
3113 truncate $fh, $offset or croak "truncate: $!";
3114}
3115sub mkfile {
3117 my ($path) = @_;
3118 unless (-e $path) {
3119 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
3120 mkpath([$dir]) unless -d $dir;
3121 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
3122 close $fh or die "Couldn't close (create) $path: $!\n";
3123 }
3124}
3125sub rev_map_set {
3127 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
3128 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
3129 my $db = $self->map_path($uuid);
3130 my $db_lock = "$db.lock";
3131 my $sig;
3132 $update_ref ||= 0;
3133 if ($update_ref) {
3134 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3135 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
3136 }
3137 mkfile($db);
3138 $LOCKFILES{$db_lock} = 1;
3140 my $sync;
3141 # both of these options make our .rev_db file very, very important
3142 # and we can't afford to lose it because rebuild() won't work
3143 if ($self->use_svm_props || $self->no_metadata) {
3144 $sync = 1;
3145 copy($db, $db_lock) or die "rev_map_set(@_): ",
3146 "Failed to copy: ",
3147 "$db => $db_lock ($!)\n";
3148 } else {
3149 rename $db, $db_lock or die "rev_map_set(@_): ",
3150 "Failed to rename: ",
3151 "$db => $db_lock ($!)\n";
3152 }
3153 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
3155 or croak "Couldn't open $db_lock: $!\n";
3156 $update_ref eq 'reset' ? _rev_map_reset($fh, $rev, $commit) :
3157 _rev_map_set($fh, $rev, $commit);
3158 if ($sync) {
3159 $fh->flush or die "Couldn't flush $db_lock: $!\n";
3160 $fh->sync or die "Couldn't sync $db_lock: $!\n";
3161 }
3162 close $fh or croak $!;
3163 if ($update_ref) {
3164 $_head = $self;
3165 my $note = "";
3166 $note = " ($update_ref)" if ($update_ref !~ /^\d*$/);
3167 command_noisy('update-ref', '-m', "r$rev$note",
3168 $self->refname, $commit);
3169 }
3170 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
3171 "$db_lock => $db ($!)\n";
3172 delete $LOCKFILES{$db_lock};
3173 if ($update_ref) {
3174 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
3175 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
3176 kill $sig, $$ if defined $sig;
3177 }
3178}
3179# If want_commit, this will return an array of (rev, commit) where
3181# commit _must_ be a valid commit in the archive.
3182# Otherwise, it'll return the max revision (whether or not the
3183# commit is valid or just a 0x40 placeholder).
3184sub rev_map_max {
3185 my ($self, $want_commit) = @_;
3186 $self->rebuild;
3187 my ($r, $c) = $self->rev_map_max_norebuild($want_commit);
3188 $want_commit ? ($r, $c) : $r;
3189}
3190sub rev_map_max_norebuild {
3192 my ($self, $want_commit) = @_;
3193 my $map_path = $self->map_path;
3194 stat $map_path or return $want_commit ? (0, undef) : 0;
3195 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3196 binmode $fh or croak "binmode: $!";
3197 my $size = (stat($fh))[7];
3198 ($size % 24) == 0 or croak "inconsistent size: $size";
3199 if ($size == 0) {
3201 close $fh or croak "close: $!";
3202 return $want_commit ? (0, undef) : 0;
3203 }
3204 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
3206 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3207 my ($r, $c) = unpack(rev_map_fmt, $buf);
3208 if ($want_commit && $c eq ('0' x40)) {
3209 if ($size < 48) {
3210 return $want_commit ? (0, undef) : 0;
3211 }
3212 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
3213 sysread($fh, $buf, 24) == 24 or croak "read: $!";
3214 ($r, $c) = unpack(rev_map_fmt, $buf);
3215 if ($c eq ('0'x40)) {
3216 croak "Penultimate record is all-zeroes in $map_path";
3217 }
3218 }
3219 close $fh or croak "close: $!";
3220 $want_commit ? ($r, $c) : $r;
3221}
3222sub rev_map_get {
3224 my ($self, $rev, $uuid) = @_;
3225 my $map_path = $self->map_path($uuid);
3226 return undef unless -e $map_path;
3227 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
3229 my $c = _rev_map_get($fh, $rev);
3230 close($fh) or croak "close: $!";
3231 $c
3232}
3233sub _rev_map_get {
3235 my ($fh, $rev) = @_;
3236 binmode $fh or croak "binmode: $!";
3238 my $size = (stat($fh))[7];
3239 ($size % 24) == 0 or croak "inconsistent size: $size";
3240 if ($size == 0) {
3242 return undef;
3243 }
3244 my ($l, $u) = (0, $size - 24);
3246 my ($r, $c, $buf);
3247 while ($l <= $u) {
3249 my $i = int(($l/24 + $u/24) / 2) * 24;
3250 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
3251 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
3252 my ($r, $c) = unpack('NH40', $buf);
3253 if ($r < $rev) {
3255 $l = $i + 24;
3256 } elsif ($r > $rev) {
3257 $u = $i - 24;
3258 } else { # $r == $rev
3259 return $c eq ('0' x 40) ? undef : $c;
3260 }
3261 }
3262 undef;
3263}
3264# Finds the first svn revision that exists on (if $eq_ok is true) or
3266# before $rev for the current branch. It will not search any lower
3267# than $min_rev. Returns the git commit hash and svn revision number
3268# if found, else (undef, undef).
3269sub find_rev_before {
3270 my ($self, $rev, $eq_ok, $min_rev) = @_;
3271 --$rev unless $eq_ok;
3272 $min_rev ||= 1;
3273 my $max_rev = $self->rev_map_max;
3274 $rev = $max_rev if ($rev > $max_rev);
3275 while ($rev >= $min_rev) {
3276 if (my $c = $self->rev_map_get($rev)) {
3277 return ($rev, $c);
3278 }
3279 --$rev;
3280 }
3281 return (undef, undef);
3282}
3283# Finds the first svn revision that exists on (if $eq_ok is true) or
3285# after $rev for the current branch. It will not search any higher
3286# than $max_rev. Returns the git commit hash and svn revision number
3287# if found, else (undef, undef).
3288sub find_rev_after {
3289 my ($self, $rev, $eq_ok, $max_rev) = @_;
3290 ++$rev unless $eq_ok;
3291 $max_rev ||= $self->rev_map_max;
3292 while ($rev <= $max_rev) {
3293 if (my $c = $self->rev_map_get($rev)) {
3294 return ($rev, $c);
3295 }
3296 ++$rev;
3297 }
3298 return (undef, undef);
3299}
3300sub _new {
3302 my ($class, $repo_id, $ref_id, $path) = @_;
3303 unless (defined $repo_id && length $repo_id) {
3304 $repo_id = $Git::SVN::default_repo_id;
3305 }
3306 unless (defined $ref_id && length $ref_id) {
3307 $_[2] = $ref_id = $Git::SVN::default_ref_id;
3308 }
3309 $_[1] = $repo_id;
3310 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
3311 $_[3] = $path = '' unless (defined $path);
3312 mkpath(["$ENV{GIT_DIR}/svn"]);
3313 bless {
3314 ref_id => $ref_id, dir => $dir, index => "$dir/index",
3315 path => $path, config => "$ENV{GIT_DIR}/svn/config",
3316 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
3317}
3318# for read-only access of old .rev_db formats
3320sub unlink_rev_db_symlink {
3321 my ($self) = @_;
3322 my $link = $self->rev_db_path;
3323 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
3324 if (-l $link) {
3325 unlink $link or croak "unlink: $link failed!";
3326 }
3327}
3328sub rev_db_path {
3330 my ($self, $uuid) = @_;
3331 my $db_path = $self->map_path($uuid);
3332 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
3333 or croak "map_path: $db_path does not contain '/.rev_map.' !";
3334 $db_path;
3335}
3336# the new replacement for .rev_db
3338sub map_path {
3339 my ($self, $uuid) = @_;
3340 $uuid ||= $self->ra_uuid;
3341 "$self->{map_root}.$uuid";
3342}
3343sub uri_encode {
3345 my ($f) = @_;
3346 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
3347 $f
3348}
3349sub remove_username {
3351 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
3352}
3353package Git::SVN::Prompt;
3355use strict;
3356use warnings;
3357require SVN::Core;
3358use vars qw/$_no_auth_cache $_username/;
3359sub simple {
3361 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
3362 $may_save = undef if $_no_auth_cache;
3363 $default_username = $_username if defined $_username;
3364 if (defined $default_username && length $default_username) {
3365 if (defined $realm && length $realm) {
3366 print STDERR "Authentication realm: $realm\n";
3367 STDERR->flush;
3368 }
3369 $cred->username($default_username);
3370 } else {
3371 username($cred, $realm, $may_save, $pool);
3372 }
3373 $cred->password(_read_password("Password for '" .
3374 $cred->username . "': ", $realm));
3375 $cred->may_save($may_save);
3376 $SVN::_Core::SVN_NO_ERROR;
3377}
3378sub ssl_server_trust {
3380 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
3381 $may_save = undef if $_no_auth_cache;
3382 print STDERR "Error validating server certificate for '$realm':\n";
3383 {
3384 no warnings 'once';
3385 # All variables SVN::Auth::SSL::* are used only once,
3386 # so we're shutting up Perl warnings about this.
3387 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
3388 print STDERR " - The certificate is not issued ",
3389 "by a trusted authority. Use the\n",
3390 " fingerprint to validate ",
3391 "the certificate manually!\n";
3392 }
3393 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
3394 print STDERR " - The certificate hostname ",
3395 "does not match.\n";
3396 }
3397 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
3398 print STDERR " - The certificate is not yet valid.\n";
3399 }
3400 if ($failures & $SVN::Auth::SSL::EXPIRED) {
3401 print STDERR " - The certificate has expired.\n";
3402 }
3403 if ($failures & $SVN::Auth::SSL::OTHER) {
3404 print STDERR " - The certificate has ",
3405 "an unknown error.\n";
3406 }
3407 } # no warnings 'once'
3408 printf STDERR
3409 "Certificate information:\n".
3410 " - Hostname: %s\n".
3411 " - Valid: from %s until %s\n".
3412 " - Issuer: %s\n".
3413 " - Fingerprint: %s\n",
3414 map $cert_info->$_, qw(hostname valid_from valid_until
3415 issuer_dname fingerprint);
3416 my $choice;
3417prompt:
3418 print STDERR $may_save ?
3419 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
3420 "(R)eject or accept (t)emporarily? ";
3421 STDERR->flush;
3422 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3423 if ($choice =~ /^t$/i) {
3424 $cred->may_save(undef);
3425 } elsif ($choice =~ /^r$/i) {
3426 return -1;
3427 } elsif ($may_save && $choice =~ /^p$/i) {
3428 $cred->may_save($may_save);
3429 } else {
3430 goto prompt;
3431 }
3432 $cred->accepted_failures($failures);
3433 $SVN::_Core::SVN_NO_ERROR;
3434}
3435sub ssl_client_cert {
3437 my ($cred, $realm, $may_save, $pool) = @_;
3438 $may_save = undef if $_no_auth_cache;
3439 print STDERR "Client certificate filename: ";
3440 STDERR->flush;
3441 chomp(my $filename = <STDIN>);
3442 $cred->cert_file($filename);
3443 $cred->may_save($may_save);
3444 $SVN::_Core::SVN_NO_ERROR;
3445}
3446sub ssl_client_cert_pw {
3448 my ($cred, $realm, $may_save, $pool) = @_;
3449 $may_save = undef if $_no_auth_cache;
3450 $cred->password(_read_password("Password: ", $realm));
3451 $cred->may_save($may_save);
3452 $SVN::_Core::SVN_NO_ERROR;
3453}
3454sub username {
3456 my ($cred, $realm, $may_save, $pool) = @_;
3457 $may_save = undef if $_no_auth_cache;
3458 if (defined $realm && length $realm) {
3459 print STDERR "Authentication realm: $realm\n";
3460 }
3461 my $username;
3462 if (defined $_username) {
3463 $username = $_username;
3464 } else {
3465 print STDERR "Username: ";
3466 STDERR->flush;
3467 chomp($username = <STDIN>);
3468 }
3469 $cred->username($username);
3470 $cred->may_save($may_save);
3471 $SVN::_Core::SVN_NO_ERROR;
3472}
3473sub _read_password {
3475 my ($prompt, $realm) = @_;
3476 print STDERR $prompt;
3477 STDERR->flush;
3478 require Term::ReadKey;
3479 Term::ReadKey::ReadMode('noecho');
3480 my $password = '';
3481 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3482 last if $key =~ /[\012\015]/; # \n\r
3483 $password .= $key;
3484 }
3485 Term::ReadKey::ReadMode('restore');
3486 print STDERR "\n";
3487 STDERR->flush;
3488 $password;
3489}
3490package SVN::Git::Fetcher;
3492use vars qw/@ISA/;
3493use strict;
3494use warnings;
3495use Carp qw/croak/;
3496use File::Temp qw/tempfile/;
3497use IO::File qw//;
3498use vars qw/$_ignore_regex/;
3499# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3501sub new {
3502 my ($class, $git_svn, $switch_path) = @_;
3503 my $self = SVN::Delta::Editor->new;
3504 bless $self, $class;
3505 if (exists $git_svn->{last_commit}) {
3506 $self->{c} = $git_svn->{last_commit};
3507 $self->{empty_symlinks} =
3508 _mark_empty_symlinks($git_svn, $switch_path);
3509 }
3510 $self->{ignore_regex} = eval { command_oneline('config', '--get',
3511 "svn-remote.$git_svn->{repo_id}.ignore-paths") };
3512 $self->{empty} = {};
3513 $self->{dir_prop} = {};
3514 $self->{file_prop} = {};
3515 $self->{absent_dir} = {};
3516 $self->{absent_file} = {};
3517 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3518 $self;
3519}
3520# this uses the Ra object, so it must be called before do_{switch,update},
3522# not inside them (when the Git::SVN::Fetcher object is passed) to
3523# do_{switch,update}
3524sub _mark_empty_symlinks {
3525 my ($git_svn, $switch_path) = @_;
3526 my $bool = Git::config_bool('svn.brokenSymlinkWorkaround');
3527 return {} if (!defined($bool)) || (defined($bool) && ! $bool);
3528 my %ret;
3530 my ($rev, $cmt) = $git_svn->last_rev_commit;
3531 return {} unless ($rev && $cmt);
3532 # allow the warning to be printed for each revision we fetch to
3534 # ensure the user sees it. The user can also disable the workaround
3535 # on the repository even while git svn is running and the next
3536 # revision fetched will skip this expensive function.
3537 my $printed_warning;
3538 chomp(my $empty_blob = `git hash-object -t blob --stdin < /dev/null`);
3539 my ($ls, $ctx) = command_output_pipe(qw/ls-tree -r -z/, $cmt);
3540 local $/ = "\0";
3541 my $pfx = defined($switch_path) ? $switch_path : $git_svn->{path};
3542 $pfx .= '/' if length($pfx);
3543 while (<$ls>) {
3544 chomp;
3545 s/\A100644 blob $empty_blob\t//o or next;
3546 unless ($printed_warning) {
3547 print STDERR "Scanning for empty symlinks, ",
3548 "this may take a while if you have ",
3549 "many empty files\n",
3550 "You may disable this with `",
3551 "git config svn.brokenSymlinkWorkaround ",
3552 "false'.\n",
3553 "This may be done in a different ",
3554 "terminal without restarting ",
3555 "git svn\n";
3556 $printed_warning = 1;
3557 }
3558 my $path = $_;
3559 my (undef, $props) =
3560 $git_svn->ra->get_file($pfx.$path, $rev, undef);
3561 if ($props->{'svn:special'}) {
3562 $ret{$path} = 1;
3563 }
3564 }
3565 command_close_pipe($ls, $ctx);
3566 \%ret;
3567}
3568# returns true if a given path is inside a ".git" directory
3570sub in_dot_git {
3571 $_[0] =~ m{(?:^|/)\.git(?:/|$)};
3572}
3573# return value: 0 -- don't ignore, 1 -- ignore
3575sub is_path_ignored {
3576 my ($self, $path) = @_;
3577 return 1 if in_dot_git($path);
3578 return 1 if defined($self->{ignore_regex}) &&
3579 $path =~ m!$self->{ignore_regex}!;
3580 return 0 unless defined($_ignore_regex);
3581 return 1 if $path =~ m!$_ignore_regex!o;
3582 return 0;
3583}
3584sub set_path_strip {
3586 my ($self, $path) = @_;
3587 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3588}
3589sub open_root {
3591 { path => '' };
3592}
3593sub open_directory {
3595 my ($self, $path, $pb, $rev) = @_;
3596 { path => $path };
3597}
3598sub git_path {
3600 my ($self, $path) = @_;
3601 if ($self->{path_strip}) {
3602 $path =~ s!$self->{path_strip}!! or
3603 die "Failed to strip path '$path' ($self->{path_strip})\n";
3604 }
3605 $path;
3606}
3607sub delete_entry {
3609 my ($self, $path, $rev, $pb) = @_;
3610 return undef if $self->is_path_ignored($path);
3611 my $gpath = $self->git_path($path);
3613 return undef if ($gpath eq '');
3614 # remove entire directories.
3616 my ($tree) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3617 =~ /\A040000 tree ([a-f\d]{40})\t\Q$gpath\E\0/);
3618 if ($tree) {
3619 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3620 -r --name-only -z/,
3621 $tree);
3622 local $/ = "\0";
3623 while (<$ls>) {
3624 chomp;
3625 my $rmpath = "$gpath/$_";
3626 $self->{gii}->remove($rmpath);
3627 print "\tD\t$rmpath\n" unless $::_q;
3628 }
3629 print "\tD\t$gpath/\n" unless $::_q;
3630 command_close_pipe($ls, $ctx);
3631 $self->{empty}->{$path} = 0
3632 } else {
3633 $self->{gii}->remove($gpath);
3634 print "\tD\t$gpath\n" unless $::_q;
3635 }
3636 undef;
3637}
3638sub open_file {
3640 my ($self, $path, $pb, $rev) = @_;
3641 my ($mode, $blob);
3642 goto out if $self->is_path_ignored($path);
3644 my $gpath = $self->git_path($path);
3646 ($mode, $blob) = (command('ls-tree', '-z', $self->{c}, "./$gpath")
3647 =~ /\A(\d{6}) blob ([a-f\d]{40})\t\Q$gpath\E\0/);
3648 unless (defined $mode && defined $blob) {
3649 die "$path was not found in commit $self->{c} (r$rev)\n";
3650 }
3651 if ($mode eq '100644' && $self->{empty_symlinks}->{$path}) {
3652 $mode = '120000';
3653 }
3654out:
3655 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3656 pool => SVN::Pool->new, action => 'M' };
3657}
3658sub add_file {
3660 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3661 my $mode;
3662 if (!$self->is_path_ignored($path)) {
3664 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3665 delete $self->{empty}->{$dir};
3666 $mode = '100644';
3667 }
3668 { path => $path, mode_a => $mode, mode_b => $mode,
3669 pool => SVN::Pool->new, action => 'A' };
3670}
3671sub add_directory {
3673 my ($self, $path, $cp_path, $cp_rev) = @_;
3674 goto out if $self->is_path_ignored($path);
3675 my $gpath = $self->git_path($path);
3676 if ($gpath eq '') {
3677 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3678 -r --name-only -z/,
3679 $self->{c});
3680 local $/ = "\0";
3681 while (<$ls>) {
3682 chomp;
3683 $self->{gii}->remove($_);
3684 print "\tD\t$_\n" unless $::_q;
3685 }
3686 command_close_pipe($ls, $ctx);
3687 $self->{empty}->{$path} = 0;
3688 }
3689 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3690 delete $self->{empty}->{$dir};
3691 $self->{empty}->{$path} = 1;
3692out:
3693 { path => $path };
3694}
3695sub change_dir_prop {
3697 my ($self, $db, $prop, $value) = @_;
3698 return undef if $self->is_path_ignored($db->{path});
3699 $self->{dir_prop}->{$db->{path}} ||= {};
3700 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3701 undef;
3702}
3703sub absent_directory {
3705 my ($self, $path, $pb) = @_;
3706 return undef if $self->is_path_ignored($path);
3707 $self->{absent_dir}->{$pb->{path}} ||= [];
3708 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3709 undef;
3710}
3711sub absent_file {
3713 my ($self, $path, $pb) = @_;
3714 return undef if $self->is_path_ignored($path);
3715 $self->{absent_file}->{$pb->{path}} ||= [];
3716 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3717 undef;
3718}
3719sub change_file_prop {
3721 my ($self, $fb, $prop, $value) = @_;
3722 return undef if $self->is_path_ignored($fb->{path});
3723 if ($prop eq 'svn:executable') {
3724 if ($fb->{mode_b} != 120000) {
3725 $fb->{mode_b} = defined $value ? 100755 : 100644;
3726 }
3727 } elsif ($prop eq 'svn:special') {
3728 $fb->{mode_b} = defined $value ? 120000 : 100644;
3729 } else {
3730 $self->{file_prop}->{$fb->{path}} ||= {};
3731 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3732 }
3733 undef;
3734}
3735sub apply_textdelta {
3737 my ($self, $fb, $exp) = @_;
3738 return undef if $self->is_path_ignored($fb->{path});
3739 my $fh = $::_repository->temp_acquire('svn_delta');
3740 # $fh gets auto-closed() by SVN::TxDelta::apply(),
3741 # (but $base does not,) so dup() it for reading in close_file
3742 open my $dup, '<&', $fh or croak $!;
3743 my $base = $::_repository->temp_acquire('git_blob');
3744 if ($fb->{blob}) {
3746 my ($base_is_link, $size);
3747 if ($fb->{mode_a} eq '120000' &&
3749 ! $self->{empty_symlinks}->{$fb->{path}}) {
3750 print $base 'link ' or die "print $!\n";
3751 $base_is_link = 1;
3752 }
3753 retry:
3754 $size = $::_repository->cat_blob($fb->{blob}, $base);
3755 die "Failed to read object $fb->{blob}" if ($size < 0);
3756 if (defined $exp) {
3758 seek $base, 0, 0 or croak $!;
3759 my $got = ::md5sum($base);
3760 if ($got ne $exp) {
3761 my $err = "Checksum mismatch: ".
3762 "$fb->{path} $fb->{blob}\n" .
3763 "expected: $exp\n" .
3764 " got: $got\n";
3765 if ($base_is_link) {
3766 warn $err,
3767 "Retrying... (possibly ",
3768 "a bad symlink from SVN)\n";
3769 $::_repository->temp_reset($base);
3770 $base_is_link = 0;
3771 goto retry;
3772 }
3773 die $err;
3774 }
3775 }
3776 }
3777 seek $base, 0, 0 or croak $!;
3778 $fb->{fh} = $fh;
3779 $fb->{base} = $base;
3780 [ SVN::TxDelta::apply($base, $dup, undef, $fb->{path}, $fb->{pool}) ];
3781}
3782sub close_file {
3784 my ($self, $fb, $exp) = @_;
3785 return undef if $self->is_path_ignored($fb->{path});
3786 my $hash;
3788 my $path = $self->git_path($fb->{path});
3789 if (my $fh = $fb->{fh}) {
3790 if (defined $exp) {
3791 seek($fh, 0, 0) or croak $!;
3792 my $got = ::md5sum($fh);
3793 if ($got ne $exp) {
3794 die "Checksum mismatch: $path\n",
3795 "expected: $exp\n got: $got\n";
3796 }
3797 }
3798 if ($fb->{mode_b} == 120000) {
3799 sysseek($fh, 0, 0) or croak $!;
3800 my $rd = sysread($fh, my $buf, 5);
3801 if (!defined $rd) {
3803 croak "sysread: $!\n";
3804 } elsif ($rd == 0) {
3805 warn "$path has mode 120000",
3806 " but it points to nothing\n",
3807 "converting to an empty file with mode",
3808 " 100644\n";
3809 $fb->{mode_b} = '100644';
3810 } elsif ($buf ne 'link ') {
3811 warn "$path has mode 120000",
3812 " but is not a link\n";
3813 } else {
3814 my $tmp_fh = $::_repository->temp_acquire(
3815 'svn_hash');
3816 my $res;
3817 while ($res = sysread($fh, my $str, 1024)) {
3818 my $out = syswrite($tmp_fh, $str, $res);
3819 defined($out) && $out == $res
3820 or croak("write ",
3821 Git::temp_path($tmp_fh),
3822 ": $!\n");
3823 }
3824 defined $res or croak $!;
3825 ($fh, $tmp_fh) = ($tmp_fh, $fh);
3827 Git::temp_release($tmp_fh, 1);
3828 }
3829 }
3830 $hash = $::_repository->hash_and_insert_object(
3832 Git::temp_path($fh));
3833 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3834 Git::temp_release($fb->{base}, 1);
3836 Git::temp_release($fh, 1);
3837 } else {
3838 $hash = $fb->{blob} or die "no blob information\n";
3839 }
3840 $fb->{pool}->clear;
3841 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3842 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3843 undef;
3844}
3845sub abort_edit {
3847 my $self = shift;
3848 $self->{nr} = $self->{gii}->{nr};
3849 delete $self->{gii};
3850 $self->SUPER::abort_edit(@_);
3851}
3852sub close_edit {
3854 my $self = shift;
3855 $self->{git_commit_ok} = 1;
3856 $self->{nr} = $self->{gii}->{nr};
3857 delete $self->{gii};
3858 $self->SUPER::close_edit(@_);
3859}
3860package SVN::Git::Editor;
3862use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3863use strict;
3864use warnings;
3865use Carp qw/croak/;
3866use IO::File;
3867sub new {
3869 my ($class, $opts) = @_;
3870 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3871 die "$_ required!\n" unless (defined $opts->{$_});
3872 }
3873 my $pool = SVN::Pool->new;
3875 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3876 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3877 $opts->{r}, $mods);
3878 # $opts->{ra} functions should not be used after this:
3880 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3881 $opts->{editor_cb}, $pool);
3882 my $self = SVN::Delta::Editor->new(@ce, $pool);
3883 bless $self, $class;
3884 foreach (qw/svn_path r tree_a tree_b/) {
3885 $self->{$_} = $opts->{$_};
3886 }
3887 $self->{url} = $opts->{ra}->{url};
3888 $self->{mods} = $mods;
3889 $self->{types} = $types;
3890 $self->{pool} = $pool;
3891 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3892 $self->{rm} = { };
3893 $self->{path_prefix} = length $self->{svn_path} ?
3894 "$self->{svn_path}/" : '';
3895 $self->{config} = $opts->{config};
3896 return $self;
3897}
3898sub generate_diff {
3900 my ($tree_a, $tree_b) = @_;
3901 my @diff_tree = qw(diff-tree -z -r);
3902 if ($_cp_similarity) {
3903 push @diff_tree, "-C$_cp_similarity";
3904 } else {
3905 push @diff_tree, '-C';
3906 }
3907 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3908 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3909 push @diff_tree, $tree_a, $tree_b;
3910 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3911 local $/ = "\0";
3912 my $state = 'meta';
3913 my @mods;
3914 while (<$diff_fh>) {
3915 chomp $_; # this gets rid of the trailing "\0"
3916 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3917 ($::sha1)\s($::sha1)\s
3918 ([MTCRAD])\d*$/xo) {
3919 push @mods, { mode_a => $1, mode_b => $2,
3920 sha1_a => $3, sha1_b => $4,
3921 chg => $5 };
3922 if ($5 =~ /^(?:C|R)$/) {
3923 $state = 'file_a';
3924 } else {
3925 $state = 'file_b';
3926 }
3927 } elsif ($state eq 'file_a') {
3928 my $x = $mods[$#mods] or croak "Empty array\n";
3929 if ($x->{chg} !~ /^(?:C|R)$/) {
3930 croak "Error parsing $_, $x->{chg}\n";
3931 }
3932 $x->{file_a} = $_;
3933 $state = 'file_b';
3934 } elsif ($state eq 'file_b') {
3935 my $x = $mods[$#mods] or croak "Empty array\n";
3936 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3937 croak "Error parsing $_, $x->{chg}\n";
3938 }
3939 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3940 croak "Error parsing $_, $x->{chg}\n";
3941 }
3942 $x->{file_b} = $_;
3943 $state = 'meta';
3944 } else {
3945 croak "Error parsing $_\n";
3946 }
3947 }
3948 command_close_pipe($diff_fh, $ctx);
3949 \@mods;
3950}
3951sub check_diff_paths {
3953 my ($ra, $pfx, $rev, $mods) = @_;
3954 my %types;
3955 $pfx .= '/' if length $pfx;
3956 sub type_diff_paths {
3958 my ($ra, $types, $path, $rev) = @_;
3959 my @p = split m#/+#, $path;
3960 my $c = shift @p;
3961 unless (defined $types->{$c}) {
3962 $types->{$c} = $ra->check_path($c, $rev);
3963 }
3964 while (@p) {
3965 $c .= '/' . shift @p;
3966 next if defined $types->{$c};
3967 $types->{$c} = $ra->check_path($c, $rev);
3968 }
3969 }
3970 foreach my $m (@$mods) {
3972 foreach my $f (qw/file_a file_b/) {
3973 next unless defined $m->{$f};
3974 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3975 if (length $pfx.$dir && ! defined $types{$dir}) {
3976 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3977 }
3978 }
3979 }
3980 \%types;
3981}
3982sub split_path {
3984 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3985}
3986sub repo_path {
3988 my ($self, $path) = @_;
3989 $self->{path_prefix}.(defined $path ? $path : '');
3990}
3991sub url_path {
3993 my ($self, $path) = @_;
3994 if ($self->{url} =~ m#^https?://#) {
3995 $path =~ s/([^~a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3996 }
3997 $self->{url} . '/' . $self->repo_path($path);
3998}
3999sub rmdirs {
4001 my ($self) = @_;
4002 my $rm = $self->{rm};
4003 delete $rm->{''}; # we never delete the url we're tracking
4004 return unless %$rm;
4005 foreach (keys %$rm) {
4007 my @d = split m#/#, $_;
4008 my $c = shift @d;
4009 $rm->{$c} = 1;
4010 while (@d) {
4011 $c .= '/' . shift @d;
4012 $rm->{$c} = 1;
4013 }
4014 }
4015 delete $rm->{$self->{svn_path}};
4016 delete $rm->{''}; # we never delete the url we're tracking
4017 return unless %$rm;
4018 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
4020 $self->{tree_b});
4021 local $/ = "\0";
4022 while (<$fh>) {
4023 chomp;
4024 my @dn = split m#/#, $_;
4025 while (pop @dn) {
4026 delete $rm->{join '/', @dn};
4027 }
4028 unless (%$rm) {
4029 close $fh;
4030 return;
4031 }
4032 }
4033 command_close_pipe($fh, $ctx);
4034 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
4036 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
4037 $self->close_directory($bat->{$d}, $p);
4038 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
4039 print "\tD+\t$d/\n" unless $::_q;
4040 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
4041 delete $bat->{$d};
4042 }
4043}
4044sub open_or_add_dir {
4046 my ($self, $full_path, $baton) = @_;
4047 my $t = $self->{types}->{$full_path};
4048 if (!defined $t) {
4049 die "$full_path not known in r$self->{r} or we have a bug!\n";
4050 }
4051 {
4052 no warnings 'once';
4053 # SVN::Node::none and SVN::Node::file are used only once,
4054 # so we're shutting up Perl's warnings about them.
4055 if ($t == $SVN::Node::none) {
4056 return $self->add_directory($full_path, $baton,
4057 undef, -1, $self->{pool});
4058 } elsif ($t == $SVN::Node::dir) {
4059 return $self->open_directory($full_path, $baton,
4060 $self->{r}, $self->{pool});
4061 } # no warnings 'once'
4062 print STDERR "$full_path already exists in repository at ",
4063 "r$self->{r} and it is not a directory (",
4064 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
4065 } # no warnings 'once'
4066 exit 1;
4067}
4068sub ensure_path {
4070 my ($self, $path) = @_;
4071 my $bat = $self->{bat};
4072 my $repo_path = $self->repo_path($path);
4073 return $bat->{''} unless (length $repo_path);
4074 my @p = split m#/+#, $repo_path;
4075 my $c = shift @p;
4076 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
4077 while (@p) {
4078 my $c0 = $c;
4079 $c .= '/' . shift @p;
4080 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
4081 }
4082 return $bat->{$c};
4083}
4084# Subroutine to convert a globbing pattern to a regular expression.
4086# From perl cookbook.
4087sub glob2pat {
4088 my $globstr = shift;
4089 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
4090 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
4091 return '^' . $globstr . '$';
4092}
4093sub check_autoprop {
4095 my ($self, $pattern, $properties, $file, $fbat) = @_;
4096 # Convert the globbing pattern to a regular expression.
4097 my $regex = glob2pat($pattern);
4098 # Check if the pattern matches the file name.
4099 if($file =~ m/($regex)/) {
4100 # Parse the list of properties to set.
4101 my @props = split(/;/, $properties);
4102 foreach my $prop (@props) {
4103 # Parse 'name=value' syntax and set the property.
4104 if ($prop =~ /([^=]+)=(.*)/) {
4105 my ($n,$v) = ($1,$2);
4106 for ($n, $v) {
4107 s/^\s+//; s/\s+$//;
4108 }
4109 $self->change_file_prop($fbat, $n, $v);
4110 }
4111 }
4112 }
4113}
4114sub apply_autoprops {
4116 my ($self, $file, $fbat) = @_;
4117 my $conf_t = ${$self->{config}}{'config'};
4118 no warnings 'once';
4119 # Check [miscellany]/enable-auto-props in svn configuration.
4120 if (SVN::_Core::svn_config_get_bool(
4121 $conf_t,
4122 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
4123 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
4124 0)) {
4125 # Auto-props are enabled. Enumerate them to look for matches.
4126 my $callback = sub {
4127 $self->check_autoprop($_[0], $_[1], $file, $fbat);
4128 };
4129 SVN::_Core::svn_config_enumerate(
4130 $conf_t,
4131 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
4132 $callback);
4133 }
4134}
4135sub A {
4137 my ($self, $m) = @_;
4138 my ($dir, $file) = split_path($m->{file_b});
4139 my $pbat = $self->ensure_path($dir);
4140 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4141 undef, -1);
4142 print "\tA\t$m->{file_b}\n" unless $::_q;
4143 $self->apply_autoprops($file, $fbat);
4144 $self->chg_file($fbat, $m);
4145 $self->close_file($fbat,undef,$self->{pool});
4146}
4147sub C {
4149 my ($self, $m) = @_;
4150 my ($dir, $file) = split_path($m->{file_b});
4151 my $pbat = $self->ensure_path($dir);
4152 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4153 $self->url_path($m->{file_a}), $self->{r});
4154 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4155 $self->chg_file($fbat, $m);
4156 $self->close_file($fbat,undef,$self->{pool});
4157}
4158sub delete_entry {
4160 my ($self, $path, $pbat) = @_;
4161 my $rpath = $self->repo_path($path);
4162 my ($dir, $file) = split_path($rpath);
4163 $self->{rm}->{$dir} = 1;
4164 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
4165}
4166sub R {
4168 my ($self, $m) = @_;
4169 my ($dir, $file) = split_path($m->{file_b});
4170 my $pbat = $self->ensure_path($dir);
4171 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
4172 $self->url_path($m->{file_a}), $self->{r});
4173 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
4174 $self->apply_autoprops($file, $fbat);
4175 $self->chg_file($fbat, $m);
4176 $self->close_file($fbat,undef,$self->{pool});
4177 ($dir, $file) = split_path($m->{file_a});
4179 $pbat = $self->ensure_path($dir);
4180 $self->delete_entry($m->{file_a}, $pbat);
4181}
4182sub M {
4184 my ($self, $m) = @_;
4185 my ($dir, $file) = split_path($m->{file_b});
4186 my $pbat = $self->ensure_path($dir);
4187 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
4188 $pbat,$self->{r},$self->{pool});
4189 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
4190 $self->chg_file($fbat, $m);
4191 $self->close_file($fbat,undef,$self->{pool});
4192}
4193sub T { shift->M(@_) }
4195sub change_file_prop {
4197 my ($self, $fbat, $pname, $pval) = @_;
4198 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
4199}
4200sub _chg_file_get_blob ($$$$) {
4202 my ($self, $fbat, $m, $which) = @_;
4203 my $fh = $::_repository->temp_acquire("git_blob_$which");
4204 if ($m->{"mode_$which"} =~ /^120/) {
4205 print $fh 'link ' or croak $!;
4206 $self->change_file_prop($fbat,'svn:special','*');
4207 } elsif ($m->{mode_a} =~ /^120/ && $m->{"mode_$which"} !~ /^120/) {
4208 $self->change_file_prop($fbat,'svn:special',undef);
4209 }
4210 my $blob = $m->{"sha1_$which"};
4211 return ($fh,) if ($blob =~ /^0{40}$/);
4212 my $size = $::_repository->cat_blob($blob, $fh);
4213 croak "Failed to read object $blob" if ($size < 0);
4214 $fh->flush == 0 or croak $!;
4215 seek $fh, 0, 0 or croak $!;
4216 my $exp = ::md5sum($fh);
4218 seek $fh, 0, 0 or croak $!;
4219 return ($fh, $exp);
4220}
4221sub chg_file {
4223 my ($self, $fbat, $m) = @_;
4224 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
4225 $self->change_file_prop($fbat,'svn:executable','*');
4226 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
4227 $self->change_file_prop($fbat,'svn:executable',undef);
4228 }
4229 my ($fh_a, $exp_a) = _chg_file_get_blob $self, $fbat, $m, 'a';
4230 my ($fh_b, $exp_b) = _chg_file_get_blob $self, $fbat, $m, 'b';
4231 my $pool = SVN::Pool->new;
4232 my $atd = $self->apply_textdelta($fbat, $exp_a, $pool);
4233 if (-s $fh_a) {
4234 my $txstream = SVN::TxDelta::new ($fh_a, $fh_b, $pool);
4235 my $res = SVN::TxDelta::send_txstream($txstream, @$atd, $pool);
4236 if (defined $res) {
4237 die "Unexpected result from send_txstream: $res\n",
4238 "(SVN::Core::VERSION: $SVN::Core::VERSION)\n";
4239 }
4240 } else {
4241 my $got = SVN::TxDelta::send_stream($fh_b, @$atd, $pool);
4242 die "Checksum mismatch\nexpected: $exp_b\ngot: $got\n"
4243 if ($got ne $exp_b);
4244 }
4245 Git::temp_release($fh_b, 1);
4246 Git::temp_release($fh_a, 1);
4247 $pool->clear;
4248}
4249sub D {
4251 my ($self, $m) = @_;
4252 my ($dir, $file) = split_path($m->{file_b});
4253 my $pbat = $self->ensure_path($dir);
4254 print "\tD\t$m->{file_b}\n" unless $::_q;
4255 $self->delete_entry($m->{file_b}, $pbat);
4256}
4257sub close_edit {
4259 my ($self) = @_;
4260 my ($p,$bat) = ($self->{pool}, $self->{bat});
4261 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
4262 next if $_ eq '';
4263 $self->close_directory($bat->{$_}, $p);
4264 }
4265 $self->close_directory($bat->{''}, $p);
4266 $self->SUPER::close_edit($p);
4267 $p->clear;
4268}
4269sub abort_edit {
4271 my ($self) = @_;
4272 $self->SUPER::abort_edit($self->{pool});
4273}
4274sub DESTROY {
4276 my $self = shift;
4277 $self->SUPER::DESTROY(@_);
4278 $self->{pool}->clear;
4279}
4280# this drives the editor
4282sub apply_diff {
4283 my ($self) = @_;
4284 my $mods = $self->{mods};
4285 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
4286 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
4287 my $f = $m->{chg};
4288 if (defined $o{$f}) {
4289 $self->$f($m);
4290 } else {
4291 fatal("Invalid change type: $f");
4292 }
4293 }
4294 $self->rmdirs if $_rmdir;
4295 if (@$mods == 0) {
4296 $self->abort_edit;
4297 } else {
4298 $self->close_edit;
4299 }
4300 return scalar @$mods;
4301}
4302package Git::SVN::Ra;
4304use vars qw/@ISA $config_dir $_log_window_size/;
4305use strict;
4306use warnings;
4307my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
4308BEGIN {
4310 # enforce temporary pool usage for some simple functions
4311 no strict 'refs';
4312 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root
4313 get_file/) {
4314 my $SUPER = "SUPER::$f";
4315 *$f = sub {
4316 my $self = shift;
4317 my $pool = SVN::Pool->new;
4318 my @ret = $self->$SUPER(@_,$pool);
4319 $pool->clear;
4320 wantarray ? @ret : $ret[0];
4321 };
4322 }
4323}
4324sub _auth_providers () {
4326 [
4327 SVN::Client::get_simple_provider(),
4328 SVN::Client::get_ssl_server_trust_file_provider(),
4329 SVN::Client::get_simple_prompt_provider(
4330 \&Git::SVN::Prompt::simple, 2),
4331 SVN::Client::get_ssl_client_cert_file_provider(),
4332 SVN::Client::get_ssl_client_cert_prompt_provider(
4333 \&Git::SVN::Prompt::ssl_client_cert, 2),
4334 SVN::Client::get_ssl_client_cert_pw_file_provider(),
4335 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
4336 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
4337 SVN::Client::get_username_provider(),
4338 SVN::Client::get_ssl_server_trust_prompt_provider(
4339 \&Git::SVN::Prompt::ssl_server_trust),
4340 SVN::Client::get_username_prompt_provider(
4341 \&Git::SVN::Prompt::username, 2)
4342 ]
4343}
4344sub escape_uri_only {
4346 my ($uri) = @_;
4347 my @tmp;
4348 foreach (split m{/}, $uri) {
4349 s/([^~\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
4350 push @tmp, $_;
4351 }
4352 join('/', @tmp);
4353}
4354sub escape_url {
4356 my ($url) = @_;
4357 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
4358 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
4359 $url = "$scheme://$domain$uri";
4360 }
4361 $url;
4362}
4363sub new {
4365 my ($class, $url) = @_;
4366 $url =~ s!/+$!!;
4367 return $RA if ($RA && $RA->{url} eq $url);
4368 SVN::_Core::svn_config_ensure($config_dir, undef);
4370 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
4371 my $config = SVN::Core::config_get_config($config_dir);
4372 $RA = undef;
4373 my $dont_store_passwords = 1;
4374 my $conf_t = ${$config}{'config'};
4375 {
4376 no warnings 'once';
4377 # The usage of $SVN::_Core::SVN_CONFIG_* variables
4378 # produces warnings that variables are used only once.
4379 # I had not found the better way to shut them up, so
4380 # the warnings of type 'once' are disabled in this block.
4381 if (SVN::_Core::svn_config_get_bool($conf_t,
4382 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4383 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
4384 1) == 0) {
4385 SVN::_Core::svn_auth_set_parameter($baton,
4386 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
4387 bless (\$dont_store_passwords, "_p_void"));
4388 }
4389 if (SVN::_Core::svn_config_get_bool($conf_t,
4390 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
4391 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
4392 1) == 0) {
4393 $Git::SVN::Prompt::_no_auth_cache = 1;
4394 }
4395 } # no warnings 'once'
4396 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
4397 config => $config,
4398 pool => SVN::Pool->new,
4399 auth_provider_callbacks => $callbacks);
4400 $self->{url} = $url;
4401 $self->{svn_path} = $url;
4402 $self->{repos_root} = $self->get_repos_root;
4403 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
4404 $self->{cache} = { check_path => { r => 0, data => {} },
4405 get_dir => { r => 0, data => {} } };
4406 $RA = bless $self, $class;
4407}
4408sub check_path {
4410 my ($self, $path, $r) = @_;
4411 my $cache = $self->{cache}->{check_path};
4412 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
4413 return $cache->{data}->{$path};
4414 }
4415 my $pool = SVN::Pool->new;
4416 my $t = $self->SUPER::check_path($path, $r, $pool);
4417 $pool->clear;
4418 if ($r != $cache->{r}) {
4419 %{$cache->{data}} = ();
4420 $cache->{r} = $r;
4421 }
4422 $cache->{data}->{$path} = $t;
4423}
4424sub get_dir {
4426 my ($self, $dir, $r) = @_;
4427 my $cache = $self->{cache}->{get_dir};
4428 if ($r == $cache->{r}) {
4429 if (my $x = $cache->{data}->{$dir}) {
4430 return wantarray ? @$x : $x->[0];
4431 }
4432 }
4433 my $pool = SVN::Pool->new;
4434 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
4435 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
4436 $pool->clear;
4437 if ($r != $cache->{r}) {
4438 %{$cache->{data}} = ();
4439 $cache->{r} = $r;
4440 }
4441 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
4442 wantarray ? (\%dirents, $r, $props) : \%dirents;
4443}
4444sub DESTROY {
4446 # do not call the real DESTROY since we store ourselves in $RA
4447}
4448# get_log(paths, start, end, limit,
4450# discover_changed_paths, strict_node_history, receiver)
4451sub get_log {
4452 my ($self, @args) = @_;
4453 my $pool = SVN::Pool->new;
4454 # svn_log_changed_path_t objects passed to get_log are likely to be
4456 # overwritten even if only the refs are copied to an external variable,
4457 # so we should dup the structures in their entirety. Using an
4458 # externally passed pool (instead of our temporary and quickly cleared
4459 # pool in Git::SVN::Ra) does not help matters at all...
4460 my $receiver = pop @args;
4461 my $prefix = "/".$self->{svn_path};
4462 $prefix =~ s#/+($)##;
4463 my $prefix_regex = qr#^\Q$prefix\E#;
4464 push(@args, sub {
4465 my ($paths) = $_[0];
4466 return &$receiver(@_) unless $paths;
4467 $_[0] = ();
4468 foreach my $p (keys %$paths) {
4469 my $i = $paths->{$p};
4470 # Make path relative to our url, not repos_root
4471 $p =~ s/$prefix_regex//;
4472 my %s = map { $_ => $i->$_; }
4473 qw/copyfrom_path copyfrom_rev action/;
4474 if ($s{'copyfrom_path'}) {
4475 $s{'copyfrom_path'} =~ s/$prefix_regex//;
4476 }
4477 $_[0]{$p} = \%s;
4478 }
4479 &$receiver(@_);
4480 });
4481 # the limit parameter was not supported in SVN 1.1.x, so we
4484 # drop it. Therefore, the receiver callback passed to it
4485 # is made aware of this limitation by being wrapped if
4486 # the limit passed to is being wrapped.
4487 if ($SVN::Core::VERSION le '1.2.0') {
4488 my $limit = splice(@args, 3, 1);
4489 if ($limit > 0) {
4490 my $receiver = pop @args;
4491 push(@args, sub { &$receiver(@_) if (--$limit >= 0) });
4492 }
4493 }
4494 my $ret = $self->SUPER::get_log(@args, $pool);
4495 $pool->clear;
4496 $ret;
4497}
4498sub trees_match {
4500 my ($self, $url1, $rev1, $url2, $rev2) = @_;
4501 my $ctx = SVN::Client->new(auth => _auth_providers);
4502 my $out = IO::File->new_tmpfile;
4503 # older SVN (1.1.x) doesn't take $pool as the last parameter for
4505 # $ctx->diff(), so we'll create a default one
4506 my $pool = SVN::Pool->new_default_sub;
4507 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
4509 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
4510 $out->flush;
4511 my $ret = (($out->stat)[7] == 0);
4512 close $out or croak $!;
4513 $ret;
4515}
4516sub get_commit_editor {
4518 my ($self, $log, $cb, $pool) = @_;
4519 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
4520 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
4521}
4522sub gs_do_update {
4524 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
4525 my $new = ($rev_a == $rev_b);
4526 my $path = $gs->{path};
4527 if ($new && -e $gs->{index}) {
4529 unlink $gs->{index} or die
4530 "Couldn't unlink index: $gs->{index}: $!\n";
4531 }
4532 my $pool = SVN::Pool->new;
4533 $editor->set_path_strip($path);
4534 my (@pc) = split m#/#, $path;
4535 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
4536 1, $editor, $pool);
4537 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4538 # Since we can't rely on svn_ra_reparent being available, we'll
4540 # just have to do some magic with set_path to make it so
4541 # we only want a partial path.
4542 my $sp = '';
4543 my $final = join('/', @pc);
4544 while (@pc) {
4545 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
4546 $sp .= '/' if length $sp;
4547 $sp .= shift @pc;
4548 }
4549 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
4550 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
4552 $reporter->finish_report($pool);
4554 $pool->clear;
4555 $editor->{git_commit_ok};
4556}
4557# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
4559# svn_ra_reparent didn't work before 1.4)
4560sub gs_do_switch {
4561 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
4562 my $path = $gs->{path};
4563 my $pool = SVN::Pool->new;
4564 my $full_url = $self->{url};
4566 my $old_url = $full_url;
4567 $full_url .= '/' . $path if length $path;
4568 my ($ra, $reparented);
4569 if ($old_url =~ m#^svn(\+ssh)?://# ||
4571 ($full_url =~ m#^https?://# &&
4572 escape_url($full_url) ne $full_url)) {
4573 $_[0] = undef;
4574 $self = undef;
4575 $RA = undef;
4576 $ra = Git::SVN::Ra->new($full_url);
4577 $ra_invalid = 1;
4578 } elsif ($old_url ne $full_url) {
4579 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url, $pool);
4580 $self->{url} = $full_url;
4581 $reparented = 1;
4582 }
4583 $ra ||= $self;
4585 $url_b = escape_url($url_b);
4586 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
4587 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
4588 $reporter->set_path('', $rev_a, 0, @lock, $pool);
4589 $reporter->finish_report($pool);
4590 if ($reparented) {
4592 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
4593 $self->{url} = $old_url;
4594 }
4595 $pool->clear;
4597 $editor->{git_commit_ok};
4598}
4599sub longest_common_path {
4601 my ($gsv, $globs) = @_;
4602 my %common;
4603 my $common_max = scalar @$gsv;
4604 foreach my $gs (@$gsv) {
4606 my @tmp = split m#/#, $gs->{path};
4607 my $p = '';
4608 foreach (@tmp) {
4609 $p .= length($p) ? "/$_" : $_;
4610 $common{$p} ||= 0;
4611 $common{$p}++;
4612 }
4613 }
4614 $globs ||= [];
4615 $common_max += scalar @$globs;
4616 foreach my $glob (@$globs) {
4617 my @tmp = split m#/#, $glob->{path}->{left};
4618 my $p = '';
4619 foreach (@tmp) {
4620 $p .= length($p) ? "/$_" : $_;
4621 $common{$p} ||= 0;
4622 $common{$p}++;
4623 }
4624 }
4625 my $longest_path = '';
4627 foreach (sort {length $b <=> length $a} keys %common) {
4628 if ($common{$_} == $common_max) {
4629 $longest_path = $_;
4630 last;
4631 }
4632 }
4633 $longest_path;
4634}
4635sub gs_fetch_loop_common {
4637 my ($self, $base, $head, $gsv, $globs) = @_;
4638 return if ($base > $head);
4639 my $inc = $_log_window_size;
4640 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4641 my $longest_path = longest_common_path($gsv, $globs);
4642 my $ra_url = $self->{url};
4643 my $find_trailing_edge;
4644 while (1) {
4645 my %revs;
4646 my $err;
4647 my $err_handler = $SVN::Error::handler;
4648 $SVN::Error::handler = sub {
4649 ($err) = @_;
4650 skip_unknown_revs($err);
4651 };
4652 sub _cb {
4653 my ($paths, $r, $author, $date, $log) = @_;
4654 [ $paths,
4655 { author => $author, date => $date, log => $log } ];
4656 }
4657 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4658 sub { $revs{$_[1]} = _cb(@_) });
4659 if ($err) {
4660 print "Checked through r$max\r";
4661 } else {
4662 $find_trailing_edge = 1;
4663 }
4664 if ($err and $find_trailing_edge) {
4665 print STDERR "Path '$longest_path' ",
4666 "was probably deleted:\n",
4667 $err->expanded_message,
4668 "\nWill attempt to follow ",
4669 "revisions r$min .. r$max ",
4670 "committed before the deletion\n";
4671 my $hi = $max;
4672 while (--$hi >= $min) {
4673 my $ok;
4674 $self->get_log([$longest_path], $min, $hi,
4675 0, 1, 1, sub {
4676 $ok = $_[1];
4677 $revs{$_[1]} = _cb(@_) });
4678 if ($ok) {
4679 print STDERR "r$min .. r$ok OK\n";
4680 last;
4681 }
4682 }
4683 $find_trailing_edge = 0;
4684 }
4685 $SVN::Error::handler = $err_handler;
4686 my %exists = map { $_->{path} => $_ } @$gsv;
4688 foreach my $r (sort {$a <=> $b} keys %revs) {
4689 my ($paths, $logged) = @{$revs{$r}};
4690 foreach my $gs ($self->match_globs(\%exists, $paths,
4692 $globs, $r)) {
4693 if ($gs->rev_map_max >= $r) {
4694 next;
4695 }
4696 next unless $gs->match_paths($paths, $r);
4697 $gs->{logged_rev_props} = $logged;
4698 if (my $last_commit = $gs->last_commit) {
4699 $gs->assert_index_clean($last_commit);
4700 }
4701 my $log_entry = $gs->do_fetch($paths, $r);
4702 if ($log_entry) {
4703 $gs->do_git_commit($log_entry);
4704 }
4705 $INDEX_FILES{$gs->{index}} = 1;
4706 }
4707 foreach my $g (@$globs) {
4708 my $k = "svn-remote.$g->{remote}." .
4709 "$g->{t}-maxRev";
4710 Git::SVN::tmp_config($k, $r);
4711 }
4712 if ($ra_invalid) {
4713 $_[0] = undef;
4714 $self = undef;
4715 $RA = undef;
4716 $self = Git::SVN::Ra->new($ra_url);
4717 $ra_invalid = undef;
4718 }
4719 }
4720 # pre-fill the .rev_db since it'll eventually get filled in
4721 # with '0' x40 if something new gets committed
4722 foreach my $gs (@$gsv) {
4723 next if $gs->rev_map_max >= $max;
4724 next if defined $gs->rev_map_get($max);
4725 $gs->rev_map_set($max, 0 x40);
4726 }
4727 foreach my $g (@$globs) {
4728 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4729 Git::SVN::tmp_config($k, $max);
4730 }
4731 last if $max >= $head;
4732 $min = $max + 1;
4733 $max += $inc;
4734 $max = $head if ($max > $head);
4735 }
4736 Git::SVN::gc();
4737}
4738sub get_dir_globbed {
4740 my ($self, $left, $depth, $r) = @_;
4741 my @x = eval { $self->get_dir($left, $r) };
4743 return unless scalar @x == 3;
4744 my $dirents = $x[0];
4745 my @finalents;
4746 foreach my $de (keys %$dirents) {
4747 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4748 if ($depth > 1) {
4749 my @args = ("$left/$de", $depth - 1, $r);
4750 foreach my $dir ($self->get_dir_globbed(@args)) {
4751 push @finalents, "$de/$dir";
4752 }
4753 } else {
4754 push @finalents, $de;
4755 }
4756 }
4757 @finalents;
4758}
4759sub match_globs {
4761 my ($self, $exists, $paths, $globs, $r) = @_;
4762 sub get_dir_check {
4764 my ($self, $exists, $g, $r) = @_;
4765 my @dirs = $self->get_dir_globbed($g->{path}->{left},
4767 $g->{path}->{depth},
4768 $r);
4769 foreach my $de (@dirs) {
4771 my $p = $g->{path}->full_path($de);
4772 next if $exists->{$p};
4773 next if (length $g->{path}->{right} &&
4774 ($self->check_path($p, $r) !=
4775 $SVN::Node::dir));
4776 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4777 $g->{ref}->full_path($de), 1);
4778 }
4779 }
4780 foreach my $g (@$globs) {
4781 if (my $path = $paths->{"/$g->{path}->{left}"}) {
4782 if ($path->{action} =~ /^[AR]$/) {
4783 get_dir_check($self, $exists, $g, $r);
4784 }
4785 }
4786 foreach (keys %$paths) {
4787 if (/$g->{path}->{left_regex}/ &&
4788 !/$g->{path}->{regex}/) {
4789 next if $paths->{$_}->{action} !~ /^[AR]$/;
4790 get_dir_check($self, $exists, $g, $r);
4791 }
4792 next unless /$g->{path}->{regex}/;
4793 my $p = $1;
4794 my $pathname = $g->{path}->full_path($p);
4795 next if $exists->{$pathname};
4796 next if ($self->check_path($pathname, $r) !=
4797 $SVN::Node::dir);
4798 $exists->{$pathname} = Git::SVN->init(
4799 $self->{url}, $pathname, undef,
4800 $g->{ref}->full_path($p), 1);
4801 }
4802 my $c = '';
4803 foreach (split m#/#, $g->{path}->{left}) {
4804 $c .= "/$_";
4805 next unless ($paths->{$c} &&
4806 ($paths->{$c}->{action} =~ /^[AR]$/));
4807 get_dir_check($self, $exists, $g, $r);
4808 }
4809 }
4810 values %$exists;
4811}
4812sub minimize_url {
4814 my ($self) = @_;
4815 return $self->{url} if ($self->{url} eq $self->{repos_root});
4816 my $url = $self->{repos_root};
4817 my @components = split(m!/!, $self->{svn_path});
4818 my $c = '';
4819 do {
4820 $url .= "/$c" if length $c;
4821 eval { (ref $self)->new($url)->get_latest_revnum };
4822 } while ($@ && ($c = shift @components));
4823 $url;
4824}
4825sub can_do_switch {
4827 my $self = shift;
4828 unless (defined $can_do_switch) {
4829 my $pool = SVN::Pool->new;
4830 my $rep = eval {
4831 $self->do_switch(1, '', 0, $self->{url},
4832 SVN::Delta::Editor->new, $pool);
4833 };
4834 if ($@) {
4835 $can_do_switch = 0;
4836 } else {
4837 $rep->abort_report($pool);
4838 $can_do_switch = 1;
4839 }
4840 $pool->clear;
4841 }
4842 $can_do_switch;
4843}
4844sub skip_unknown_revs {
4846 my ($err) = @_;
4847 my $errno = $err->apr_err();
4848 # Maybe the branch we're tracking didn't
4849 # exist when the repo started, so it's
4850 # not an error if it doesn't, just continue
4851 #
4852 # Wonderfully consistent library, eh?
4853 # 160013 - svn:// and file://
4854 # 175002 - http(s)://
4855 # 175007 - http(s):// (this repo required authorization, too...)
4856 # More codes may be discovered later...
4857 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4858 my $err_key = $err->expanded_message;
4859 # revision numbers change every time, filter them out
4860 $err_key =~ s/\d+/\0/g;
4861 $err_key = "$errno\0$err_key";
4862 unless ($ignored_err{$err_key}) {
4863 warn "W: Ignoring error from SVN, path probably ",
4864 "does not exist: ($errno): ",
4865 $err->expanded_message,"\n";
4866 warn "W: Do not be alarmed at the above message ",
4867 "git-svn is just searching aggressively for ",
4868 "old history.\n",
4869 "This may take a while on large repositories\n";
4870 $ignored_err{$err_key} = 1;
4871 }
4872 return;
4873 }
4874 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4875}
4876package Git::SVN::Log;
4878use strict;
4879use warnings;
4880use POSIX qw/strftime/;
4881use Time::Local;
4882use constant commit_log_separator => ('-' x 72) . "\n";
4883use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4884 %rusers $show_commit $incremental/;
4885my $l_fmt;
4886sub cmt_showable {
4888 my ($c) = @_;
4889 return 1 if defined $c->{r};
4890 # big commit message got truncated by the 16k pretty buffer in rev-list
4892 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4893 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4894 @{$c->{l}} = ();
4895 my @log = command(qw/cat-file commit/, $c->{c});
4896 # shift off the headers
4898 shift @log while ($log[0] ne '');
4899 shift @log;
4900 # TODO: make $c->{l} not have a trailing newline in the future
4902 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4903 (undef, $c->{r}, undef) = ::extract_metadata(
4905 (grep(/^git-svn-id: /, @log))[-1]);
4906 }
4907 return defined $c->{r};
4908}
4909sub log_use_color {
4911 return $color || Git->repository->get_colorbool('color.diff');
4912}
4913sub git_svn_log_cmd {
4915 my ($r_min, $r_max, @args) = @_;
4916 my $head = 'HEAD';
4917 my (@files, @log_opts);
4918 foreach my $x (@args) {
4919 if ($x eq '--' || @files) {
4920 push @files, $x;
4921 } else {
4922 if (::verify_ref("$x^0")) {
4923 $head = $x;
4924 } else {
4925 push @log_opts, $x;
4926 }
4927 }
4928 }
4929 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4931 $gs ||= Git::SVN->_new;
4932 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4933 $gs->refname);
4934 push @cmd, '-r' unless $non_recursive;
4935 push @cmd, qw/--raw --name-status/ if $verbose;
4936 push @cmd, '--color' if log_use_color();
4937 push @cmd, @log_opts;
4938 if (defined $r_max && $r_max == $r_min) {
4939 push @cmd, '--max-count=1';
4940 if (my $c = $gs->rev_map_get($r_max)) {
4941 push @cmd, $c;
4942 }
4943 } elsif (defined $r_max) {
4944 if ($r_max < $r_min) {
4945 ($r_min, $r_max) = ($r_max, $r_min);
4946 }
4947 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4948 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4949 # If there are no commits in the range, both $c_max and $c_min
4950 # will be undefined. If there is at least 1 commit in the
4951 # range, both will be defined.
4952 return () if !defined $c_min || !defined $c_max;
4953 if ($c_min eq $c_max) {
4954 push @cmd, '--max-count=1', $c_min;
4955 } else {
4956 push @cmd, '--boundary', "$c_min..$c_max";
4957 }
4958 }
4959 return (@cmd, @files);
4960}
4961# adapted from pager.c
4963sub config_pager {
4964 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4965 if (!defined $pager) {
4966 $pager = 'less';
4967 } elsif (length $pager == 0 || $pager eq 'cat') {
4968 $pager = undef;
4969 }
4970 $ENV{GIT_PAGER_IN_USE} = defined($pager);
4971}
4972sub run_pager {
4974 return unless -t *STDOUT && defined $pager;
4975 pipe my ($rfd, $wfd) or return;
4976 defined(my $pid = fork) or ::fatal "Can't fork: $!";
4977 if (!$pid) {
4978 open STDOUT, '>&', $wfd or
4979 ::fatal "Can't redirect to stdout: $!";
4980 return;
4981 }
4982 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4983 $ENV{LESS} ||= 'FRSX';
4984 exec $pager or ::fatal "Can't run pager: $! ($pager)";
4985}
4986sub format_svn_date {
4988 # some systmes don't handle or mishandle %z, so be creative.
4989 my $t = shift || time;
4990 my $gm = timelocal(gmtime($t));
4991 my $sign = qw( + + - )[ $t <=> $gm ];
4992 my $gmoff = sprintf("%s%02d%02d", $sign, (gmtime(abs($t - $gm)))[2,1]);
4993 return strftime("%Y-%m-%d %H:%M:%S $gmoff (%a, %d %b %Y)", localtime($t));
4994}
4995sub parse_git_date {
4997 my ($t, $tz) = @_;
4998 # Date::Parse isn't in the standard Perl distro :(
4999 if ($tz =~ s/^\+//) {
5000 $t += tz_to_s_offset($tz);
5001 } elsif ($tz =~ s/^\-//) {
5002 $t -= tz_to_s_offset($tz);
5003 }
5004 return $t;
5005}
5006sub set_local_timezone {
5008 if (defined $TZ) {
5009 $ENV{TZ} = $TZ;
5010 } else {
5011 delete $ENV{TZ};
5012 }
5013}
5014sub tz_to_s_offset {
5016 my ($tz) = @_;
5017 $tz =~ s/(\d\d)$//;
5018 return ($1 * 60) + ($tz * 3600);
5019}
5020sub get_author_info {
5022 my ($dest, $author, $t, $tz) = @_;
5023 $author =~ s/(?:^\s*|\s*$)//g;
5024 $dest->{a_raw} = $author;
5025 my $au;
5026 if ($::_authors) {
5027 $au = $rusers{$author} || undef;
5028 }
5029 if (!$au) {
5030 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
5031 }
5032 $dest->{t} = $t;
5033 $dest->{tz} = $tz;
5034 $dest->{a} = $au;
5035 $dest->{t_utc} = parse_git_date($t, $tz);
5036}
5037sub process_commit {
5039 my ($c, $r_min, $r_max, $defer) = @_;
5040 if (defined $r_min && defined $r_max) {
5041 if ($r_min == $c->{r} && $r_min == $r_max) {
5042 show_commit($c);
5043 return 0;
5044 }
5045 return 1 if $r_min == $r_max;
5046 if ($r_min < $r_max) {
5047 # we need to reverse the print order
5048 return 0 if (defined $limit && --$limit < 0);
5049 push @$defer, $c;
5050 return 1;
5051 }
5052 if ($r_min != $r_max) {
5053 return 1 if ($r_min < $c->{r});
5054 return 1 if ($r_max > $c->{r});
5055 }
5056 }
5057 return 0 if (defined $limit && --$limit < 0);
5058 show_commit($c);
5059 return 1;
5060}
5061sub show_commit {
5063 my $c = shift;
5064 if ($oneline) {
5065 my $x = "\n";
5066 if (my $l = $c->{l}) {
5067 while ($l->[0] =~ /^\s*$/) { shift @$l }
5068 $x = $l->[0];
5069 }
5070 $l_fmt ||= 'A' . length($c->{r});
5071 print 'r',pack($l_fmt, $c->{r}),' | ';
5072 print "$c->{c} | " if $show_commit;
5073 print $x;
5074 } else {
5075 show_commit_normal($c);
5076 }
5077}
5078sub show_commit_changed_paths {
5080 my ($c) = @_;
5081 return unless $c->{changed};
5082 print "Changed paths:\n", @{$c->{changed}};
5083}
5084sub show_commit_normal {
5086 my ($c) = @_;
5087 print commit_log_separator, "r$c->{r} | ";
5088 print "$c->{c} | " if $show_commit;
5089 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
5090 my $nr_line = 0;
5091 if (my $l = $c->{l}) {
5093 while ($l->[$#$l] eq "\n" && $#$l > 0
5094 && $l->[($#$l - 1)] eq "\n") {
5095 pop @$l;
5096 }
5097 $nr_line = scalar @$l;
5098 if (!$nr_line) {
5099 print "1 line\n\n\n";
5100 } else {
5101 if ($nr_line == 1) {
5102 $nr_line = '1 line';
5103 } else {
5104 $nr_line .= ' lines';
5105 }
5106 print $nr_line, "\n";
5107 show_commit_changed_paths($c);
5108 print "\n";
5109 print $_ foreach @$l;
5110 }
5111 } else {
5112 print "1 line\n";
5113 show_commit_changed_paths($c);
5114 print "\n";
5115 }
5117 foreach my $x (qw/raw stat diff/) {
5118 if ($c->{$x}) {
5119 print "\n";
5120 print $_ foreach @{$c->{$x}}
5121 }
5122 }
5123}
5124sub cmd_show_log {
5126 my (@args) = @_;
5127 my ($r_min, $r_max);
5128 my $r_last = -1; # prevent dupes
5129 set_local_timezone();
5130 if (defined $::_revision) {
5131 if ($::_revision =~ /^(\d+):(\d+)$/) {
5132 ($r_min, $r_max) = ($1, $2);
5133 } elsif ($::_revision =~ /^\d+$/) {
5134 $r_min = $r_max = $::_revision;
5135 } else {
5136 ::fatal "-r$::_revision is not supported, use ",
5137 "standard 'git log' arguments instead";
5138 }
5139 }
5140 config_pager();
5142 @args = git_svn_log_cmd($r_min, $r_max, @args);
5143 if (!@args) {
5144 print commit_log_separator unless $incremental || $oneline;
5145 return;
5146 }
5147 my $log = command_output_pipe(@args);
5148 run_pager();
5149 my (@k, $c, $d, $stat);
5150 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
5151 while (<$log>) {
5152 if (/^${esc_color}commit -?($::sha1_short)/o) {
5153 my $cmt = $1;
5154 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
5155 $r_last = $c->{r};
5156 process_commit($c, $r_min, $r_max, \@k) or
5157 goto out;
5158 }
5159 $d = undef;
5160 $c = { c => $cmt };
5161 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
5162 get_author_info($c, $1, $2, $3);
5163 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
5164 # ignore
5165 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
5166 push @{$c->{raw}}, $_;
5167 } elsif (/^${esc_color}[ACRMDT]\t/) {
5168 # we could add $SVN->{svn_path} here, but that requires
5169 # remote access at the moment (repo_path_split)...
5170 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
5171 push @{$c->{changed}}, $_;
5172 } elsif (/^${esc_color}diff /o) {
5173 $d = 1;
5174 push @{$c->{diff}}, $_;
5175 } elsif ($d) {
5176 push @{$c->{diff}}, $_;
5177 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
5178 $esc_color*[\+\-]*$esc_color$/x) {
5179 $stat = 1;
5180 push @{$c->{stat}}, $_;
5181 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
5182 push @{$c->{stat}}, $_;
5183 $stat = undef;
5184 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
5185 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
5186 } elsif (s/^${esc_color} //o) {
5187 push @{$c->{l}}, $_;
5188 }
5189 }
5190 if ($c && defined $c->{r} && $c->{r} != $r_last) {
5191 $r_last = $c->{r};
5192 process_commit($c, $r_min, $r_max, \@k);
5193 }
5194 if (@k) {
5195 ($r_min, $r_max) = ($r_max, $r_min);
5196 process_commit($_, $r_min, $r_max) foreach reverse @k;
5197 }
5198out:
5199 close $log;
5200 print commit_log_separator unless $incremental || $oneline;
5201}
5202sub cmd_blame {
5204 my $path = pop;
5205 config_pager();
5207 run_pager();
5208 my ($fh, $ctx, $rev);
5210 if ($_git_format) {
5212 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
5213 while (my $line = <$fh>) {
5214 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
5215 # Uncommitted edits show up as a rev ID of
5216 # all zeros, which we can't look up with
5217 # cmt_metadata
5218 if ($1 !~ /^0+$/) {
5219 (undef, $rev, undef) =
5220 ::cmt_metadata($1);
5221 $rev = '0' if (!$rev);
5222 } else {
5223 $rev = '0';
5224 }
5225 $rev = sprintf('%-10s', $rev);
5226 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
5227 }
5228 print $line;
5229 }
5230 } else {
5231 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
5232 '--', $path);
5233 my ($sha1);
5234 my %authors;
5235 my @buffer;
5236 my %dsha; #distinct sha keys
5237 while (my $line = <$fh>) {
5239 push @buffer, $line;
5240 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5241 $dsha{$1} = 1;
5242 }
5243 }
5244 my $s2r = ::cmt_sha2rev_batch([keys %dsha]);
5246 foreach my $line (@buffer) {
5248 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
5249 $rev = $s2r->{$1};
5250 $rev = '0' if (!$rev)
5251 }
5252 elsif ($line =~ /^author (.*)/) {
5253 $authors{$rev} = $1;
5254 $authors{$rev} =~ s/\s/_/g;
5255 }
5256 elsif ($line =~ /^\t(.*)$/) {
5257 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
5258 }
5259 }
5260 }
5261 command_close_pipe($fh, $ctx);
5262}
5263package Git::SVN::Migration;
5265# these version numbers do NOT correspond to actual version numbers
5266# of git nor git-svn. They are just relative.
5267#
5268# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
5269#
5270# v1 layout: .git/$id/info/url, refs/remotes/$id
5271#
5272# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
5273#
5274# v3 layout: .git/svn/$id, refs/remotes/$id
5275# - info/url may remain for backwards compatibility
5276# - this is what we migrate up to this layout automatically,
5277# - this will be used by git svn init on single branches
5278# v3.1 layout (auto migrated):
5279# - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
5280# for backwards compatibility
5281#
5282# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
5283# - this is only created for newly multi-init-ed
5284# repositories. Similar in spirit to the
5285# --use-separate-remotes option in git-clone (now default)
5286# - we do not automatically migrate to this (following
5287# the example set by core git)
5288#
5289# v5 layout: .rev_db.$UUID => .rev_map.$UUID
5290# - newer, more-efficient format that uses 24-bytes per record
5291# with no filler space.
5292# - use xxd -c24 < .rev_map.$UUID to view and debug
5293# - This is a one-way migration, repositories updated to the
5294# new format will not be able to use old git-svn without
5295# rebuilding the .rev_db. Rebuilding the rev_db is not
5296# possible if noMetadata or useSvmProps are set; but should
5297# be no problem for users that use the (sensible) defaults.
5298use strict;
5299use warnings;
5300use Carp qw/croak/;
5301use File::Path qw/mkpath/;
5302use File::Basename qw/dirname basename/;
5303use vars qw/$_minimize/;
5304sub migrate_from_v0 {
5306 my $git_dir = $ENV{GIT_DIR};
5307 return undef unless -d $git_dir;
5308 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5309 my $migrated = 0;
5310 while (<$fh>) {
5311 chomp;
5312 my ($id, $orig_ref) = ($_, $_);
5313 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
5314 next unless -f "$git_dir/$id/info/url";
5315 my $new_ref = "refs/remotes/$id";
5316 if (::verify_ref("$new_ref^0")) {
5317 print STDERR "W: $orig_ref is probably an old ",
5318 "branch used by an ancient version of ",
5319 "git-svn.\n",
5320 "However, $new_ref also exists.\n",
5321 "We will not be able ",
5322 "to use this branch until this ",
5323 "ambiguity is resolved.\n";
5324 next;
5325 }
5326 print STDERR "Migrating from v0 layout...\n" if !$migrated;
5327 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
5328 command_noisy('update-ref', $new_ref, $orig_ref);
5329 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
5330 $migrated++;
5331 }
5332 command_close_pipe($fh, $ctx);
5333 print STDERR "Done migrating from v0 layout...\n" if $migrated;
5334 $migrated;
5335}
5336sub migrate_from_v1 {
5338 my $git_dir = $ENV{GIT_DIR};
5339 my $migrated = 0;
5340 return $migrated unless -d $git_dir;
5341 my $svn_dir = "$git_dir/svn";
5342 # just in case somebody used 'svn' as their $id at some point...
5344 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
5345 print STDERR "Migrating from a git-svn v1 layout...\n";
5347 mkpath([$svn_dir]);
5348 print STDERR "Data from a previous version of git-svn exists, but\n\t",
5349 "$svn_dir\n\t(required for this version ",
5350 "($::VERSION) of git-svn) does not exist.\n";
5351 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
5352 while (<$fh>) {
5353 my $x = $_;
5354 next unless $x =~ s#^refs/remotes/##;
5355 chomp $x;
5356 next unless -f "$git_dir/$x/info/url";
5357 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
5358 next unless $u;
5359 my $dn = dirname("$git_dir/svn/$x");
5360 mkpath([$dn]) unless -d $dn;
5361 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
5362 mkpath(["$git_dir/svn/svn"]);
5363 print STDERR " - $git_dir/$x/info => ",
5364 "$git_dir/svn/$x/info\n";
5365 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
5366 croak "$!: $x";
5367 # don't worry too much about these, they probably
5368 # don't exist with repos this old (save for index,
5369 # and we can easily regenerate that)
5370 foreach my $f (qw/unhandled.log index .rev_db/) {
5371 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
5372 }
5373 } else {
5374 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
5375 rename "$git_dir/$x", "$git_dir/svn/$x" or
5376 croak "$!: $x";
5377 }
5378 $migrated++;
5379 }
5380 command_close_pipe($fh, $ctx);
5381 print STDERR "Done migrating from a git-svn v1 layout\n";
5382 $migrated;
5383}
5384sub read_old_urls {
5386 my ($l_map, $pfx, $path) = @_;
5387 my @dir;
5388 foreach (<$path/*>) {
5389 if (-r "$_/info/url") {
5390 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
5391 my $ref_id = $pfx . basename $_;
5392 my $url = ::file_to_s("$_/info/url");
5393 $l_map->{$ref_id} = $url;
5394 } elsif (-d $_) {
5395 push @dir, $_;
5396 }
5397 }
5398 foreach (@dir) {
5399 my $x = $_;
5400 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
5401 read_old_urls($l_map, $x, $_);
5402 }
5403}
5404sub migrate_from_v2 {
5406 my @cfg = command(qw/config -l/);
5407 return if grep /^svn-remote\..+\.url=/, @cfg;
5408 my %l_map;
5409 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
5410 my $migrated = 0;
5411 foreach my $ref_id (sort keys %l_map) {
5413 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
5414 if ($@) {
5415 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
5416 }
5417 $migrated++;
5418 }
5419 $migrated;
5420}
5421sub minimize_connections {
5423 my $r = Git::SVN::read_all_remotes();
5424 my $new_urls = {};
5425 my $root_repos = {};
5426 foreach my $repo_id (keys %$r) {
5427 my $url = $r->{$repo_id}->{url} or next;
5428 my $fetch = $r->{$repo_id}->{fetch} or next;
5429 my $ra = Git::SVN::Ra->new($url);
5430 # skip existing cases where we already connect to the root
5432 if (($ra->{url} eq $ra->{repos_root}) ||
5433 ($ra->{repos_root} eq $repo_id)) {
5434 $root_repos->{$ra->{url}} = $repo_id;
5435 next;
5436 }
5437 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
5439 my $root_path = $ra->{url};
5440 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
5441 foreach my $path (keys %$fetch) {
5442 my $ref_id = $fetch->{$path};
5443 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
5444 # make sure we can read when connecting to
5446 # a higher level of a repository
5447 my ($last_rev, undef) = $gs->last_rev_commit;
5448 if (!defined $last_rev) {
5449 $last_rev = eval {
5450 $root_ra->get_latest_revnum;
5451 };
5452 next if $@;
5453 }
5454 my $new = $root_path;
5455 $new .= length $path ? "/$path" : '';
5456 eval {
5457 $root_ra->get_log([$new], $last_rev, $last_rev,
5458 0, 0, 1, sub { });
5459 };
5460 next if $@;
5461 $new_urls->{$ra->{repos_root}}->{$new} =
5462 { ref_id => $ref_id,
5463 old_repo_id => $repo_id,
5464 old_path => $path };
5465 }
5466 }
5467 my @emptied;
5469 foreach my $url (keys %$new_urls) {
5470 # see if we can re-use an existing [svn-remote "repo_id"]
5471 # instead of creating a(n ugly) new section:
5472 my $repo_id = $root_repos->{$url} || $url;
5473 my $fetch = $new_urls->{$url};
5475 foreach my $path (keys %$fetch) {
5476 my $x = $fetch->{$path};
5477 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
5478 my $pfx = "svn-remote.$x->{old_repo_id}";
5479 my $old_fetch = quotemeta("$x->{old_path}:".
5481 "refs/remotes/$x->{ref_id}");
5482 command_noisy(qw/config --unset/,
5483 "$pfx.fetch", '^'. $old_fetch . '$');
5484 delete $r->{$x->{old_repo_id}}->
5485 {fetch}->{$x->{old_path}};
5486 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
5487 command_noisy(qw/config --unset/,
5488 "$pfx.url");
5489 push @emptied, $x->{old_repo_id}
5490 }
5491 }
5492 }
5493 if (@emptied) {
5494 my $file = $ENV{GIT_CONFIG} || "$ENV{GIT_DIR}/config";
5495 print STDERR <<EOF;
5496The following [svn-remote] sections in your config file ($file) are empty
5497and can be safely removed:
5498EOF
5499 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
5500 }
5501}
5502sub migration_check {
5504 migrate_from_v0();
5505 migrate_from_v1();
5506 migrate_from_v2();
5507 minimize_connections() if $_minimize;
5508}
5509package Git::IndexInfo;
5511use strict;
5512use warnings;
5513use Git qw/command_input_pipe command_close_pipe/;
5514sub new {
5516 my ($class) = @_;
5517 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
5518 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
5519}
5520sub remove {
5522 my ($self, $path) = @_;
5523 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
5524 return ++$self->{nr};
5525 }
5526 undef;
5527}
5528sub update {
5530 my ($self, $mode, $hash, $path) = @_;
5531 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
5532 return ++$self->{nr};
5533 }
5534 undef;
5535}
5536sub DESTROY {
5538 my ($self) = @_;
5539 command_close_pipe($self->{gui}, $self->{ctx});
5540}
5541package Git::SVN::GlobSpec;
5543use strict;
5544use warnings;
5545sub new {
5547 my ($class, $glob) = @_;
5548 my $re = $glob;
5549 $re =~ s!/+$!!g; # no need for trailing slashes
5550 $re =~ m!^([^*]*)(\*(?:/\*)*)([^*]*)$!;
5551 my $temp = $re;
5552 my ($left, $right) = ($1, $3);
5553 $re = $2;
5554 my $depth = $re =~ tr/*/*/;
5555 if ($depth != $temp =~ tr/*/*/) {
5556 die "Only one set of wildcard directories " .
5557 "(e.g. '*' or '*/*/*') is supported: '$glob'\n";
5558 }
5559 if ($depth == 0) {
5560 die "One '*' is needed for glob: '$glob'\n";
5561 }
5562 $re =~ s!\*!\[^/\]*!g;
5563 $re = quotemeta($left) . "($re)" . quotemeta($right);
5564 if (length $left && !($left =~ s!/+$!!g)) {
5565 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
5566 }
5567 if (length $right && !($right =~ s!^/+!!g)) {
5568 die "Missing leading '/' on right side of: '$glob' ($right)\n";
5569 }
5570 my $left_re = qr/^\/\Q$left\E(\/|$)/;
5571 bless { left => $left, right => $right, left_regex => $left_re,
5572 regex => qr/$re/, glob => $glob, depth => $depth }, $class;
5573}
5574sub full_path {
5576 my ($self, $path) = @_;
5577 return (length $self->{left} ? "$self->{left}/" : '') .
5578 $path . (length $self->{right} ? "/$self->{right}" : '');
5579}
5580__END__
5582Data structures:
5584$remotes = { # returned by read_all_remotes()
5587 'svn' => {
5588 # svn-remote.svn.url=https://svn.musicpd.org
5589 url => 'https://svn.musicpd.org',
5590 # svn-remote.svn.fetch=mpd/trunk:trunk
5591 fetch => {
5592 'mpd/trunk' => 'trunk',
5593 },
5594 # svn-remote.svn.tags=mpd/tags/*:tags/*
5595 tags => {
5596 path => {
5597 left => 'mpd/tags',
5598 right => '',
5599 regex => qr!mpd/tags/([^/]+)$!,
5600 glob => 'tags/*',
5601 },
5602 ref => {
5603 left => 'tags',
5604 right => '',
5605 regex => qr!tags/([^/]+)$!,
5606 glob => 'tags/*',
5607 },
5608 }
5609 }
5610};
5611$log_entry hashref as returned by libsvn_log_entry()
5613{
5614 log => 'whitespace-formatted log entry
5615', # trailing newline is preserved
5616 revision => '8', # integer
5617 date => '2004-02-24T17:01:44.108345Z', # commit date
5618 author => 'committer name'
5619};
5620# this is generated by generate_diff();
5623@mods = array of diff-index line hashes, each element represents one line
5624 of diff-index output
5625diff-index line ($m hash)
5627{
5628 mode_a => first column of diff-index output, no leading ':',
5629 mode_b => second column of diff-index output,
5630 sha1_b => sha1sum of the final blob,
5631 chg => change type [MCRADT],
5632 file_a => original file name of a file (iff chg is 'C' or 'R')
5633 file_b => new/current file name of a file (any chg)
5634}
5635;
5636# retval of read_url_paths{,_all}();
5638$l_map = {
5639 # repository root url
5640 'https://svn.musicpd.org' => {
5641 # repository path # GIT_SVN_ID
5642 'mpd/trunk' => 'trunk',
5643 'mpd/tags/0.11.5' => 'tags/0.11.5',
5644 },
5645}
5646Notes:
5648 I don't trust the each() function on unless I created %hash myself
5649 because the internal iterator may not have started at base.