cc35f5046fd07edfffca07b3e01cadf846d63db2
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 $_authors %users/;
9$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::Log::TZ = $ENV{TZ};
24$ENV{TZ} = 'UTC';
25$| = 1; # unbuffer STDOUT
26sub fatal (@) { print STDERR "@_\n"; exit 1 }
28require SVN::Core; # use()-ing this causes segfaults for me... *shrug*
29require SVN::Ra;
30require SVN::Delta;
31if ($SVN::Core::VERSION lt '1.1.0') {
32 fatal "Need SVN::Core 1.1.0 or better (got $SVN::Core::VERSION)";
33}
34push @Git::SVN::Ra::ISA, 'SVN::Ra';
35push @SVN::Git::Editor::ISA, 'SVN::Delta::Editor';
36push @SVN::Git::Fetcher::ISA, 'SVN::Delta::Editor';
37use Carp qw/croak/;
38use Digest::MD5;
39use IO::File qw//;
40use File::Basename qw/dirname basename/;
41use File::Path qw/mkpath/;
42use Getopt::Long qw/:config gnu_getopt no_ignore_case auto_abbrev/;
43use IPC::Open3;
44use Git;
45BEGIN {
47 # import functions from Git into our packages, en masse
48 no strict 'refs';
49 foreach (qw/command command_oneline command_noisy command_output_pipe
50 command_input_pipe command_close_pipe/) {
51 for my $package ( qw(SVN::Git::Editor SVN::Git::Fetcher
52 Git::SVN::Migration Git::SVN::Log Git::SVN),
53 __PACKAGE__) {
54 *{"${package}::$_"} = \&{"Git::$_"};
55 }
56 }
57}
58my ($SVN);
60$sha1 = qr/[a-f\d]{40}/;
62$sha1_short = qr/[a-f\d]{4,40}/;
63my ($_stdin, $_help, $_edit,
64 $_message, $_file,
65 $_template, $_shared,
66 $_version, $_fetch_all, $_no_rebase,
67 $_merge, $_strategy, $_dry_run, $_local,
68 $_prefix, $_no_checkout, $_url, $_verbose,
69 $_git_format);
70$Git::SVN::_follow_parent = 1;
71my %remote_opts = ( 'username=s' => \$Git::SVN::Prompt::_username,
72 'config-dir=s' => \$Git::SVN::Ra::config_dir,
73 'no-auth-cache' => \$Git::SVN::Prompt::_no_auth_cache );
74my %fc_opts = ( 'follow-parent|follow!' => \$Git::SVN::_follow_parent,
75 'authors-file|A=s' => \$_authors,
76 'repack:i' => \$Git::SVN::_repack,
77 'noMetadata' => \$Git::SVN::_no_metadata,
78 'useSvmProps' => \$Git::SVN::_use_svm_props,
79 'useSvnsyncProps' => \$Git::SVN::_use_svnsync_props,
80 'log-window-size=i' => \$Git::SVN::Ra::_log_window_size,
81 'no-checkout' => \$_no_checkout,
82 'quiet|q' => \$_q,
83 'repack-flags|repack-args|repack-opts=s' =>
84 \$Git::SVN::_repack_flags,
85 'use-log-author' => \$Git::SVN::_use_log_author,
86 'add-author-from' => \$Git::SVN::_add_author_from,
87 %remote_opts );
88my ($_trunk, $_tags, $_branches, $_stdlayout);
90my %icv;
91my %init_opts = ( 'template=s' => \$_template, 'shared:s' => \$_shared,
92 'trunk|T=s' => \$_trunk, 'tags|t=s' => \$_tags,
93 'branches|b=s' => \$_branches, 'prefix=s' => \$_prefix,
94 'stdlayout|s' => \$_stdlayout,
95 'minimize-url|m' => \$Git::SVN::_minimize_url,
96 'no-metadata' => sub { $icv{noMetadata} = 1 },
97 'use-svm-props' => sub { $icv{useSvmProps} = 1 },
98 'use-svnsync-props' => sub { $icv{useSvnsyncProps} = 1 },
99 'rewrite-root=s' => sub { $icv{rewriteRoot} = $_[1] },
100 %remote_opts );
101my %cmt_opts = ( 'edit|e' => \$_edit,
102 'rmdir' => \$SVN::Git::Editor::_rmdir,
103 'find-copies-harder' => \$SVN::Git::Editor::_find_copies_harder,
104 'l=i' => \$SVN::Git::Editor::_rename_limit,
105 'copy-similarity|C=i'=> \$SVN::Git::Editor::_cp_similarity
106);
107my %cmd = (
109 fetch => [ \&cmd_fetch, "Download new revisions from SVN",
110 { 'revision|r=s' => \$_revision,
111 'fetch-all|all' => \$_fetch_all,
112 %fc_opts } ],
113 clone => [ \&cmd_clone, "Initialize and fetch revisions",
114 { 'revision|r=s' => \$_revision,
115 %fc_opts, %init_opts } ],
116 init => [ \&cmd_init, "Initialize a repo for tracking" .
117 " (requires URL argument)",
118 \%init_opts ],
119 'multi-init' => [ \&cmd_multi_init,
120 "Deprecated alias for ".
121 "'$0 init -T<trunk> -b<branches> -t<tags>'",
122 \%init_opts ],
123 dcommit => [ \&cmd_dcommit,
124 'Commit several diffs to merge with upstream',
125 { 'merge|m|M' => \$_merge,
126 'strategy|s=s' => \$_strategy,
127 'verbose|v' => \$_verbose,
128 'dry-run|n' => \$_dry_run,
129 'fetch-all|all' => \$_fetch_all,
130 'no-rebase' => \$_no_rebase,
131 %cmt_opts, %fc_opts } ],
132 'set-tree' => [ \&cmd_set_tree,
133 "Set an SVN repository to a git tree-ish",
134 { 'stdin|' => \$_stdin, %cmt_opts, %fc_opts, } ],
135 'create-ignore' => [ \&cmd_create_ignore,
136 'Create a .gitignore per svn:ignore',
137 { 'revision|r=i' => \$_revision
138 } ],
139 'propget' => [ \&cmd_propget,
140 'Print the value of a property on a file or directory',
141 { 'revision|r=i' => \$_revision } ],
142 'proplist' => [ \&cmd_proplist,
143 'List all properties of a file or directory',
144 { 'revision|r=i' => \$_revision } ],
145 'show-ignore' => [ \&cmd_show_ignore, "Show svn:ignore listings",
146 { 'revision|r=i' => \$_revision
147 } ],
148 'show-externals' => [ \&cmd_show_externals, "Show svn:externals listings",
149 { 'revision|r=i' => \$_revision
150 } ],
151 'multi-fetch' => [ \&cmd_multi_fetch,
152 "Deprecated alias for $0 fetch --all",
153 { 'revision|r=s' => \$_revision, %fc_opts } ],
154 'migrate' => [ sub { },
155 # no-op, we automatically run this anyways,
156 'Migrate configuration/metadata/layout from
157 previous versions of git-svn',
158 { 'minimize' => \$Git::SVN::Migration::_minimize,
159 %remote_opts } ],
160 'log' => [ \&Git::SVN::Log::cmd_show_log, 'Show commit logs',
161 { 'limit=i' => \$Git::SVN::Log::limit,
162 'revision|r=s' => \$_revision,
163 'verbose|v' => \$Git::SVN::Log::verbose,
164 'incremental' => \$Git::SVN::Log::incremental,
165 'oneline' => \$Git::SVN::Log::oneline,
166 'show-commit' => \$Git::SVN::Log::show_commit,
167 'non-recursive' => \$Git::SVN::Log::non_recursive,
168 'authors-file|A=s' => \$_authors,
169 'color' => \$Git::SVN::Log::color,
170 'pager=s' => \$Git::SVN::Log::pager
171 } ],
172 'find-rev' => [ \&cmd_find_rev, "Translate between SVN revision numbers and tree-ish",
173 {} ],
174 'rebase' => [ \&cmd_rebase, "Fetch and rebase your working directory",
175 { 'merge|m|M' => \$_merge,
176 'verbose|v' => \$_verbose,
177 'strategy|s=s' => \$_strategy,
178 'local|l' => \$_local,
179 'fetch-all|all' => \$_fetch_all,
180 'dry-run|n' => \$_dry_run,
181 %fc_opts } ],
182 'commit-diff' => [ \&cmd_commit_diff,
183 'Commit a diff between two trees',
184 { 'message|m=s' => \$_message,
185 'file|F=s' => \$_file,
186 'revision|r=s' => \$_revision,
187 %cmt_opts } ],
188 'info' => [ \&cmd_info,
189 "Show info about the latest SVN revision
190 on the current branch",
191 { 'url' => \$_url, } ],
192 'blame' => [ \&Git::SVN::Log::cmd_blame,
193 "Show what revision and author last modified each line of a file",
194 { 'git-format' => \$_git_format } ],
195);
196my $cmd;
198for (my $i = 0; $i < @ARGV; $i++) {
199 if (defined $cmd{$ARGV[$i]}) {
200 $cmd = $ARGV[$i];
201 splice @ARGV, $i, 1;
202 last;
203 }
204};
205# make sure we're always running at the top-level working directory
207unless ($cmd && $cmd =~ /(?:clone|init|multi-init)$/) {
208 unless (-d $ENV{GIT_DIR}) {
209 if ($git_dir_user_set) {
210 die "GIT_DIR=$ENV{GIT_DIR} explicitly set, ",
211 "but it is not a directory\n";
212 }
213 my $git_dir = delete $ENV{GIT_DIR};
214 chomp(my $cdup = command_oneline(qw/rev-parse --show-cdup/));
215 unless (length $cdup) {
216 die "Already at toplevel, but $git_dir ",
217 "not found '$cdup'\n";
218 }
219 chdir $cdup or die "Unable to chdir up to '$cdup'\n";
220 unless (-d $git_dir) {
221 die "$git_dir still not found after going to ",
222 "'$cdup'\n";
223 }
224 $ENV{GIT_DIR} = $git_dir;
225 }
226 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
227}
228my %opts = %{$cmd{$cmd}->[2]} if (defined $cmd);
230read_repo_config(\%opts);
232Getopt::Long::Configure('pass_through') if ($cmd && ($cmd eq 'log' || $cmd eq 'blame'));
233my $rv = GetOptions(%opts, 'help|H|h' => \$_help, 'version|V' => \$_version,
234 'minimize-connections' => \$Git::SVN::Migration::_minimize,
235 'id|i=s' => \$Git::SVN::default_ref_id,
236 'svn-remote|remote|R=s' => sub {
237 $Git::SVN::no_reuse_existing = 1;
238 $Git::SVN::default_repo_id = $_[1] });
239exit 1 if (!$rv && $cmd && $cmd ne 'log');
240usage(0) if $_help;
242version() if $_version;
243usage(1) unless defined $cmd;
244load_authors() if $_authors;
245unless ($cmd =~ /^(?:clone|init|multi-init|commit-diff)$/) {
247 Git::SVN::Migration::migration_check();
248}
249Git::SVN::init_vars();
250eval {
251 Git::SVN::verify_remotes_sanity();
252 $cmd{$cmd}->[0]->(@ARGV);
253};
254fatal $@ if $@;
255post_fetch_checkout();
256exit 0;
257####################### primary functions ######################
259sub usage {
260 my $exit = shift || 0;
261 my $fd = $exit ? \*STDERR : \*STDOUT;
262 print $fd <<"";
263git-svn - bidirectional operations between a single Subversion tree and git
264Usage: git svn <command> [options] [arguments]\n
265 print $fd "Available commands:\n" unless $cmd;
267 foreach (sort keys %cmd) {
269 next if $cmd && $cmd ne $_;
270 next if /^multi-/; # don't show deprecated commands
271 print $fd ' ',pack('A17',$_),$cmd{$_}->[1],"\n";
272 foreach (sort keys %{$cmd{$_}->[2]}) {
273 # mixed-case options are for .git/config only
274 next if /[A-Z]/ && /^[a-z]+$/i;
275 # prints out arguments as they should be passed:
276 my $x = s#[:=]s$## ? '<arg>' : s#[:=]i$## ? '<num>' : '';
277 print $fd ' ' x 21, join(', ', map { length $_ > 1 ?
278 "--$_" : "-$_" }
279 split /\|/,$_)," $x\n";
280 }
281 }
282 print $fd <<"";
283\nGIT_SVN_ID may be set in the environment or via the --id/-i switch to an
284arbitrary identifier if you're tracking multiple SVN branches/repositories in
285one git repository and want to keep them separate. See git-svn(1) for more
286information.
287 exit $exit;
289}
290sub version {
292 print "git-svn version $VERSION (svn $SVN::Core::VERSION)\n";
293 exit 0;
294}
295sub do_git_init_db {
297 unless (-d $ENV{GIT_DIR}) {
298 my @init_db = ('init');
299 push @init_db, "--template=$_template" if defined $_template;
300 if (defined $_shared) {
301 if ($_shared =~ /[a-z]/) {
302 push @init_db, "--shared=$_shared";
303 } else {
304 push @init_db, "--shared";
305 }
306 }
307 command_noisy(@init_db);
308 $_repository = Git->repository(Repository => ".git");
309 }
310 my $set;
311 my $pfx = "svn-remote.$Git::SVN::default_repo_id";
312 foreach my $i (keys %icv) {
313 die "'$set' and '$i' cannot both be set\n" if $set;
314 next unless defined $icv{$i};
315 command_noisy('config', "$pfx.$i", $icv{$i});
316 $set = $i;
317 }
318}
319sub init_subdir {
321 my $repo_path = shift or return;
322 mkpath([$repo_path]) unless -d $repo_path;
323 chdir $repo_path or die "Couldn't chdir to $repo_path: $!\n";
324 $ENV{GIT_DIR} = '.git';
325 $_repository = Git->repository(Repository => $ENV{GIT_DIR});
326}
327sub cmd_clone {
329 my ($url, $path) = @_;
330 if (!defined $path &&
331 (defined $_trunk || defined $_branches || defined $_tags ||
332 defined $_stdlayout) &&
333 $url !~ m#^[a-z\+]+://#) {
334 $path = $url;
335 }
336 $path = basename($url) if !defined $path || !length $path;
337 cmd_init($url, $path);
338 Git::SVN::fetch_all($Git::SVN::default_repo_id);
339}
340sub cmd_init {
342 if (defined $_stdlayout) {
343 $_trunk = 'trunk' if (!defined $_trunk);
344 $_tags = 'tags' if (!defined $_tags);
345 $_branches = 'branches' if (!defined $_branches);
346 }
347 if (defined $_trunk || defined $_branches || defined $_tags) {
348 return cmd_multi_init(@_);
349 }
350 my $url = shift or die "SVN repository location required ",
351 "as a command-line argument\n";
352 init_subdir(@_);
353 do_git_init_db();
354 Git::SVN->init($url);
356}
357sub cmd_fetch {
359 if (grep /^\d+=./, @_) {
360 die "'<rev>=<commit>' fetch arguments are ",
361 "no longer supported.\n";
362 }
363 my ($remote) = @_;
364 if (@_ > 1) {
365 die "Usage: $0 fetch [--all] [svn-remote]\n";
366 }
367 $remote ||= $Git::SVN::default_repo_id;
368 if ($_fetch_all) {
369 cmd_multi_fetch();
370 } else {
371 Git::SVN::fetch_all($remote, Git::SVN::read_all_remotes());
372 }
373}
374sub cmd_set_tree {
376 my (@commits) = @_;
377 if ($_stdin || !@commits) {
378 print "Reading from stdin...\n";
379 @commits = ();
380 while (<STDIN>) {
381 if (/\b($sha1_short)\b/o) {
382 unshift @commits, $1;
383 }
384 }
385 }
386 my @revs;
387 foreach my $c (@commits) {
388 my @tmp = command('rev-parse',$c);
389 if (scalar @tmp == 1) {
390 push @revs, $tmp[0];
391 } elsif (scalar @tmp > 1) {
392 push @revs, reverse(command('rev-list',@tmp));
393 } else {
394 fatal "Failed to rev-parse $c";
395 }
396 }
397 my $gs = Git::SVN->new;
398 my ($r_last, $cmt_last) = $gs->last_rev_commit;
399 $gs->fetch;
400 if (defined $gs->{last_rev} && $r_last != $gs->{last_rev}) {
401 fatal "There are new revisions that were fetched ",
402 "and need to be merged (or acknowledged) ",
403 "before committing.\nlast rev: $r_last\n",
404 " current: $gs->{last_rev}";
405 }
406 $gs->set_tree($_) foreach @revs;
407 print "Done committing ",scalar @revs," revisions to SVN\n";
408 unlink $gs->{index};
409}
410sub cmd_dcommit {
412 my $head = shift;
413 git_cmd_try { command_oneline(qw/diff-index --quiet HEAD/) }
414 'Cannot dcommit with a dirty index. Commit your changes first, '
415 . "or stash them with `git stash'.\n";
416 $head ||= 'HEAD';
417 my @refs;
418 my ($url, $rev, $uuid, $gs) = working_head_info($head, \@refs);
419 if ($url) {
420 print "Committing to $url ...\n";
421 }
422 unless ($gs) {
423 die "Unable to determine upstream SVN information from ",
424 "$head history.\nPerhaps the repository is empty.";
425 }
426 my $last_rev;
427 my ($linear_refs, $parents) = linearize_history($gs, \@refs);
428 if ($_no_rebase && scalar(@$linear_refs) > 1) {
429 warn "Attempting to commit more than one change while ",
430 "--no-rebase is enabled.\n",
431 "If these changes depend on each other, re-running ",
432 "without --no-rebase may be required."
433 }
434 while (1) {
435 my $d = shift @$linear_refs or last;
436 unless (defined $last_rev) {
437 (undef, $last_rev, undef) = cmt_metadata("$d~1");
438 unless (defined $last_rev) {
439 fatal "Unable to extract revision information ",
440 "from commit $d~1";
441 }
442 }
443 if ($_dry_run) {
444 print "diff-tree $d~1 $d\n";
445 } else {
446 my $cmt_rev;
447 my %ed_opts = ( r => $last_rev,
448 log => get_commit_entry($d)->{log},
449 ra => Git::SVN::Ra->new($gs->full_url),
450 config => SVN::Core::config_get_config(
451 $Git::SVN::Ra::config_dir
452 ),
453 tree_a => "$d~1",
454 tree_b => $d,
455 editor_cb => sub {
456 print "Committed r$_[0]\n";
457 $cmt_rev = $_[0];
458 },
459 svn_path => '');
460 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
461 print "No changes\n$d~1 == $d\n";
462 } elsif ($parents->{$d} && @{$parents->{$d}}) {
463 $gs->{inject_parents_dcommit}->{$cmt_rev} =
464 $parents->{$d};
465 }
466 $_fetch_all ? $gs->fetch_all : $gs->fetch;
467 $last_rev = $cmt_rev;
468 next if $_no_rebase;
469 # we always want to rebase against the current HEAD,
471 # not any head that was passed to us
472 my @diff = command('diff-tree', $d,
473 $gs->refname, '--');
474 my @finish;
475 if (@diff) {
476 @finish = rebase_cmd();
477 print STDERR "W: $d and ", $gs->refname,
478 " differ, using @finish:\n",
479 join("\n", @diff), "\n";
480 } else {
481 print "No changes between current HEAD and ",
482 $gs->refname,
483 "\nResetting to the latest ",
484 $gs->refname, "\n";
485 @finish = qw/reset --mixed/;
486 }
487 command_noisy(@finish, $gs->refname);
488 if (@diff) {
489 @refs = ();
490 my ($url_, $rev_, $uuid_, $gs_) =
491 working_head_info($head, \@refs);
492 my ($linear_refs_, $parents_) =
493 linearize_history($gs_, \@refs);
494 if (scalar(@$linear_refs) !=
495 scalar(@$linear_refs_)) {
496 fatal "# of revisions changed ",
497 "\nbefore:\n",
498 join("\n", @$linear_refs),
499 "\n\nafter:\n",
500 join("\n", @$linear_refs_), "\n",
501 'If you are attempting to commit ',
502 "merges, try running:\n\t",
503 'git rebase --interactive',
504 '--preserve-merges ',
505 $gs->refname,
506 "\nBefore dcommitting";
507 }
508 if ($url_ ne $url) {
509 fatal "URL mismatch after rebase: ",
510 "$url_ != $url";
511 }
512 if ($uuid_ ne $uuid) {
513 fatal "uuid mismatch after rebase: ",
514 "$uuid_ != $uuid";
515 }
516 # remap parents
517 my (%p, @l, $i);
518 for ($i = 0; $i < scalar @$linear_refs; $i++) {
519 my $new = $linear_refs_->[$i] or next;
520 $p{$new} =
521 $parents->{$linear_refs->[$i]};
522 push @l, $new;
523 }
524 $parents = \%p;
525 $linear_refs = \@l;
526 }
527 }
528 }
529 unlink $gs->{index};
530}
531sub cmd_find_rev {
533 my $revision_or_hash = shift or die "SVN or git revision required ",
534 "as a command-line argument\n";
535 my $result;
536 if ($revision_or_hash =~ /^r\d+$/) {
537 my $head = shift;
538 $head ||= 'HEAD';
539 my @refs;
540 my (undef, undef, $uuid, $gs) = working_head_info($head, \@refs);
541 unless ($gs) {
542 die "Unable to determine upstream SVN information from ",
543 "$head history\n";
544 }
545 my $desired_revision = substr($revision_or_hash, 1);
546 $result = $gs->rev_map_get($desired_revision, $uuid);
547 } else {
548 my (undef, $rev, undef) = cmt_metadata($revision_or_hash);
549 $result = $rev;
550 }
551 print "$result\n" if $result;
552}
553sub cmd_rebase {
555 command_noisy(qw/update-index --refresh/);
556 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
557 unless ($gs) {
558 die "Unable to determine upstream SVN information from ",
559 "working tree history\n";
560 }
561 if ($_dry_run) {
562 print "Remote Branch: " . $gs->refname . "\n";
563 print "SVN URL: " . $url . "\n";
564 return;
565 }
566 if (command(qw/diff-index HEAD --/)) {
567 print STDERR "Cannot rebase with uncommited changes:\n";
568 command_noisy('status');
569 exit 1;
570 }
571 unless ($_local) {
572 # rebase will checkout for us, so no need to do it explicitly
573 $_no_checkout = 'true';
574 $_fetch_all ? $gs->fetch_all : $gs->fetch;
575 }
576 command_noisy(rebase_cmd(), $gs->refname);
577}
578sub cmd_show_ignore {
580 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
581 $gs ||= Git::SVN->new;
582 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
583 $gs->prop_walk($gs->{path}, $r, sub {
584 my ($gs, $path, $props) = @_;
585 print STDOUT "\n# $path\n";
586 my $s = $props->{'svn:ignore'} or return;
587 $s =~ s/[\r\n]+/\n/g;
588 chomp $s;
589 $s =~ s#^#$path#gm;
590 print STDOUT "$s\n";
591 });
592}
593sub cmd_show_externals {
595 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
596 $gs ||= Git::SVN->new;
597 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
598 $gs->prop_walk($gs->{path}, $r, sub {
599 my ($gs, $path, $props) = @_;
600 print STDOUT "\n# $path\n";
601 my $s = $props->{'svn:externals'} or return;
602 $s =~ s/[\r\n]+/\n/g;
603 chomp $s;
604 $s =~ s#^#$path#gm;
605 print STDOUT "$s\n";
606 });
607}
608sub cmd_create_ignore {
610 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
611 $gs ||= Git::SVN->new;
612 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
613 $gs->prop_walk($gs->{path}, $r, sub {
614 my ($gs, $path, $props) = @_;
615 # $path is of the form /path/to/dir/
616 my $ignore = '.' . $path . '.gitignore';
617 my $s = $props->{'svn:ignore'} or return;
618 open(GITIGNORE, '>', $ignore)
619 or fatal("Failed to open `$ignore' for writing: $!");
620 $s =~ s/[\r\n]+/\n/g;
621 chomp $s;
622 # Prefix all patterns so that the ignore doesn't apply
623 # to sub-directories.
624 $s =~ s#^#/#gm;
625 print GITIGNORE "$s\n";
626 close(GITIGNORE)
627 or fatal("Failed to close `$ignore': $!");
628 command_noisy('add', '-f', $ignore);
629 });
630}
631sub canonicalize_path {
633 my ($path) = @_;
634 my $dot_slash_added = 0;
635 if (substr($path, 0, 1) ne "/") {
636 $path = "./" . $path;
637 $dot_slash_added = 1;
638 }
639 # File::Spec->canonpath doesn't collapse x/../y into y (for a
640 # good reason), so let's do this manually.
641 $path =~ s#/+#/#g;
642 $path =~ s#/\.(?:/|$)#/#g;
643 $path =~ s#/[^/]+/\.\.##g;
644 $path =~ s#/$##g;
645 $path =~ s#^\./## if $dot_slash_added;
646 $path =~ s#^/##;
647 $path =~ s#^\.$##;
648 return $path;
649}
650# get_svnprops(PATH)
652# ------------------
653# Helper for cmd_propget and cmd_proplist below.
654sub get_svnprops {
655 my $path = shift;
656 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
657 $gs ||= Git::SVN->new;
658 # prefix THE PATH by the sub-directory from which the user
660 # invoked us.
661 $path = $cmd_dir_prefix . $path;
662 fatal("No such file or directory: $path") unless -e $path;
663 my $is_dir = -d $path ? 1 : 0;
664 $path = $gs->{path} . '/' . $path;
665 # canonicalize the path (otherwise libsvn will abort or fail to
667 # find the file)
668 $path = canonicalize_path($path);
669 my $r = (defined $_revision ? $_revision : $gs->ra->get_latest_revnum);
671 my $props;
672 if ($is_dir) {
673 (undef, undef, $props) = $gs->ra->get_dir($path, $r);
674 }
675 else {
676 (undef, $props) = $gs->ra->get_file($path, $r, undef);
677 }
678 return $props;
679}
680# cmd_propget (PROP, PATH)
682# ------------------------
683# Print the SVN property PROP for PATH.
684sub cmd_propget {
685 my ($prop, $path) = @_;
686 $path = '.' if not defined $path;
687 usage(1) if not defined $prop;
688 my $props = get_svnprops($path);
689 if (not defined $props->{$prop}) {
690 fatal("`$path' does not have a `$prop' SVN property.");
691 }
692 print $props->{$prop} . "\n";
693}
694# cmd_proplist (PATH)
696# -------------------
697# Print the list of SVN properties for PATH.
698sub cmd_proplist {
699 my $path = shift;
700 $path = '.' if not defined $path;
701 my $props = get_svnprops($path);
702 print "Properties on '$path':\n";
703 foreach (sort keys %{$props}) {
704 print " $_\n";
705 }
706}
707sub cmd_multi_init {
709 my $url = shift;
710 unless (defined $_trunk || defined $_branches || defined $_tags) {
711 usage(1);
712 }
713 # there are currently some bugs that prevent multi-init/multi-fetch
715 # setups from working well without this.
716 $Git::SVN::_minimize_url = 1;
717 $_prefix = '' unless defined $_prefix;
719 if (defined $url) {
720 $url =~ s#/+$##;
721 init_subdir(@_);
722 }
723 do_git_init_db();
724 if (defined $_trunk) {
725 my $trunk_ref = $_prefix . 'trunk';
726 # try both old-style and new-style lookups:
727 my $gs_trunk = eval { Git::SVN->new($trunk_ref) };
728 unless ($gs_trunk) {
729 my ($trunk_url, $trunk_path) =
730 complete_svn_url($url, $_trunk);
731 $gs_trunk = Git::SVN->init($trunk_url, $trunk_path,
732 undef, $trunk_ref);
733 }
734 }
735 return unless defined $_branches || defined $_tags;
736 my $ra = $url ? Git::SVN::Ra->new($url) : undef;
737 complete_url_ls_init($ra, $_branches, '--branches/-b', $_prefix);
738 complete_url_ls_init($ra, $_tags, '--tags/-t', $_prefix . 'tags/');
739}
740sub cmd_multi_fetch {
742 my $remotes = Git::SVN::read_all_remotes();
743 foreach my $repo_id (sort keys %$remotes) {
744 if ($remotes->{$repo_id}->{url}) {
745 Git::SVN::fetch_all($repo_id, $remotes);
746 }
747 }
748}
749# this command is special because it requires no metadata
751sub cmd_commit_diff {
752 my ($ta, $tb, $url) = @_;
753 my $usage = "Usage: $0 commit-diff -r<revision> ".
754 "<tree-ish> <tree-ish> [<URL>]";
755 fatal($usage) if (!defined $ta || !defined $tb);
756 my $svn_path = '';
757 if (!defined $url) {
758 my $gs = eval { Git::SVN->new };
759 if (!$gs) {
760 fatal("Needed URL or usable git-svn --id in ",
761 "the command-line\n", $usage);
762 }
763 $url = $gs->{url};
764 $svn_path = $gs->{path};
765 }
766 unless (defined $_revision) {
767 fatal("-r|--revision is a required argument\n", $usage);
768 }
769 if (defined $_message && defined $_file) {
770 fatal("Both --message/-m and --file/-F specified ",
771 "for the commit message.\n",
772 "I have no idea what you mean");
773 }
774 if (defined $_file) {
775 $_message = file_to_s($_file);
776 } else {
777 $_message ||= get_commit_entry($tb)->{log};
778 }
779 my $ra ||= Git::SVN::Ra->new($url);
780 my $r = $_revision;
781 if ($r eq 'HEAD') {
782 $r = $ra->get_latest_revnum;
783 } elsif ($r !~ /^\d+$/) {
784 die "revision argument: $r not understood by git-svn\n";
785 }
786 my %ed_opts = ( r => $r,
787 log => $_message,
788 ra => $ra,
789 tree_a => $ta,
790 tree_b => $tb,
791 editor_cb => sub { print "Committed r$_[0]\n" },
792 svn_path => $svn_path );
793 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
794 print "No changes\n$ta == $tb\n";
795 }
796}
797sub cmd_info {
799 my $path = canonicalize_path(shift or ".");
800 unless (scalar(@_) == 0) {
801 die "Too many arguments specified\n";
802 }
803 my ($file_type, $diff_status) = find_file_type_and_diff_status($path);
805 if (!$file_type && !$diff_status) {
807 print STDERR "$path: (Not a versioned resource)\n\n";
808 return;
809 }
810 my ($url, $rev, $uuid, $gs) = working_head_info('HEAD');
812 unless ($gs) {
813 die "Unable to determine upstream SVN information from ",
814 "working tree history\n";
815 }
816 my $full_url = $url . ($path eq "." ? "" : "/$path");
817 if ($_url) {
819 print $full_url, "\n";
820 return;
821 }
822 my $result = "Path: $path\n";
824 $result .= "Name: " . basename($path) . "\n" if $file_type ne "dir";
825 $result .= "URL: " . $full_url . "\n";
826 eval {
828 my $repos_root = $gs->repos_root;
829 Git::SVN::remove_username($repos_root);
830 $result .= "Repository Root: $repos_root\n";
831 };
832 if ($@) {
833 $result .= "Repository Root: (offline)\n";
834 }
835 $result .= "Repository UUID: $uuid\n" unless $diff_status eq "A";
836 $result .= "Revision: " . ($diff_status eq "A" ? 0 : $rev) . "\n";
837 $result .= "Node Kind: " .
839 ($file_type eq "dir" ? "directory" : "file") . "\n";
840 my $schedule = $diff_status eq "A"
842 ? "add"
843 : ($diff_status eq "D" ? "delete" : "normal");
844 $result .= "Schedule: $schedule\n";
845 if ($diff_status eq "A") {
847 print $result, "\n";
848 return;
849 }
850 my ($lc_author, $lc_rev, $lc_date_utc);
852 my @args = Git::SVN::Log::git_svn_log_cmd($rev, $rev, "--", $path);
853 my $log = command_output_pipe(@args);
854 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
855 while (<$log>) {
856 if (/^${esc_color}author (.+) <[^>]+> (\d+) ([\-\+]?\d+)$/o) {
857 $lc_author = $1;
858 $lc_date_utc = Git::SVN::Log::parse_git_date($2, $3);
859 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
860 (undef, $lc_rev, undef) = ::extract_metadata($1);
861 }
862 }
863 close $log;
864 Git::SVN::Log::set_local_timezone();
866 $result .= "Last Changed Author: $lc_author\n";
868 $result .= "Last Changed Rev: $lc_rev\n";
869 $result .= "Last Changed Date: " .
870 Git::SVN::Log::format_svn_date($lc_date_utc) . "\n";
871 if ($file_type ne "dir") {
873 my $text_last_updated_date =
874 ($diff_status eq "D" ? $lc_date_utc : (stat $path)[9]);
875 $result .=
876 "Text Last Updated: " .
877 Git::SVN::Log::format_svn_date($text_last_updated_date) .
878 "\n";
879 my $checksum;
880 if ($diff_status eq "D") {
881 my ($fh, $ctx) =
882 command_output_pipe(qw(cat-file blob), "HEAD:$path");
883 if ($file_type eq "link") {
884 my $file_name = <$fh>;
885 $checksum = md5sum("link $file_name");
886 } else {
887 $checksum = md5sum($fh);
888 }
889 command_close_pipe($fh, $ctx);
890 } elsif ($file_type eq "link") {
891 my $file_name =
892 command(qw(cat-file blob), "HEAD:$path");
893 $checksum =
894 md5sum("link " . $file_name);
895 } else {
896 open FILE, "<", $path or die $!;
897 $checksum = md5sum(\*FILE);
898 close FILE or die $!;
899 }
900 $result .= "Checksum: " . $checksum . "\n";
901 }
902 print $result, "\n";
904}
905########################### utility functions #########################
907sub rebase_cmd {
909 my @cmd = qw/rebase/;
910 push @cmd, '-v' if $_verbose;
911 push @cmd, qw/--merge/ if $_merge;
912 push @cmd, "--strategy=$_strategy" if $_strategy;
913 @cmd;
914}
915sub post_fetch_checkout {
917 return if $_no_checkout;
918 my $gs = $Git::SVN::_head or return;
919 return if verify_ref('refs/heads/master^0');
920 my $valid_head = verify_ref('HEAD^0');
922 command_noisy(qw(update-ref refs/heads/master), $gs->refname);
923 return if ($valid_head || !verify_ref('HEAD^0'));
924 return if $ENV{GIT_DIR} !~ m#^(?:.*/)?\.git$#;
926 my $index = $ENV{GIT_INDEX_FILE} || "$ENV{GIT_DIR}/index";
927 return if -f $index;
928 return if command_oneline(qw/rev-parse --is-inside-work-tree/) eq 'false';
930 return if command_oneline(qw/rev-parse --is-inside-git-dir/) eq 'true';
931 command_noisy(qw/read-tree -m -u -v HEAD HEAD/);
932 print STDERR "Checked out HEAD:\n ",
933 $gs->full_url, " r", $gs->last_rev, "\n";
934}
935sub complete_svn_url {
937 my ($url, $path) = @_;
938 $path =~ s#/+$##;
939 if ($path !~ m#^[a-z\+]+://#) {
940 if (!defined $url || $url !~ m#^[a-z\+]+://#) {
941 fatal("E: '$path' is not a complete URL ",
942 "and a separate URL is not specified");
943 }
944 return ($url, $path);
945 }
946 return ($path, '');
947}
948sub complete_url_ls_init {
950 my ($ra, $repo_path, $switch, $pfx) = @_;
951 unless ($repo_path) {
952 print STDERR "W: $switch not specified\n";
953 return;
954 }
955 $repo_path =~ s#/+$##;
956 if ($repo_path =~ m#^[a-z\+]+://#) {
957 $ra = Git::SVN::Ra->new($repo_path);
958 $repo_path = '';
959 } else {
960 $repo_path =~ s#^/+##;
961 unless ($ra) {
962 fatal("E: '$repo_path' is not a complete URL ",
963 "and a separate URL is not specified");
964 }
965 }
966 my $url = $ra->{url};
967 my $gs = Git::SVN->init($url, undef, undef, undef, 1);
968 my $k = "svn-remote.$gs->{repo_id}.url";
969 my $orig_url = eval { command_oneline(qw/config --get/, $k) };
970 if ($orig_url && ($orig_url ne $gs->{url})) {
971 die "$k already set: $orig_url\n",
972 "wanted to set to: $gs->{url}\n";
973 }
974 command_oneline('config', $k, $gs->{url}) unless $orig_url;
975 my $remote_path = "$ra->{svn_path}/$repo_path";
976 $remote_path =~ s#/+#/#g;
977 $remote_path =~ s#^/##g;
978 $remote_path .= "/*" if $remote_path !~ /\*/;
979 my ($n) = ($switch =~ /^--(\w+)/);
980 if (length $pfx && $pfx !~ m#/$#) {
981 die "--prefix='$pfx' must have a trailing slash '/'\n";
982 }
983 command_noisy('config', "svn-remote.$gs->{repo_id}.$n",
984 "$remote_path:refs/remotes/$pfx*");
985}
986sub verify_ref {
988 my ($ref) = @_;
989 eval { command_oneline([ 'rev-parse', '--verify', $ref ],
990 { STDERR => 0 }); };
991}
992sub get_tree_from_treeish {
994 my ($treeish) = @_;
995 # $treeish can be a symbolic ref, too:
996 my $type = command_oneline(qw/cat-file -t/, $treeish);
997 my $expected;
998 while ($type eq 'tag') {
999 ($treeish, $type) = command(qw/cat-file tag/, $treeish);
1000 }
1001 if ($type eq 'commit') {
1002 $expected = (grep /^tree /, command(qw/cat-file commit/,
1003 $treeish))[0];
1004 ($expected) = ($expected =~ /^tree ($sha1)$/o);
1005 die "Unable to get tree from $treeish\n" unless $expected;
1006 } elsif ($type eq 'tree') {
1007 $expected = $treeish;
1008 } else {
1009 die "$treeish is a $type, expected tree, tag or commit\n";
1010 }
1011 return $expected;
1012}
1013sub get_commit_entry {
1015 my ($treeish) = shift;
1016 my %log_entry = ( log => '', tree => get_tree_from_treeish($treeish) );
1017 my $commit_editmsg = "$ENV{GIT_DIR}/COMMIT_EDITMSG";
1018 my $commit_msg = "$ENV{GIT_DIR}/COMMIT_MSG";
1019 open my $log_fh, '>', $commit_editmsg or croak $!;
1020 my $type = command_oneline(qw/cat-file -t/, $treeish);
1022 if ($type eq 'commit' || $type eq 'tag') {
1023 my ($msg_fh, $ctx) = command_output_pipe('cat-file',
1024 $type, $treeish);
1025 my $in_msg = 0;
1026 my $author;
1027 my $saw_from = 0;
1028 my $msgbuf = "";
1029 while (<$msg_fh>) {
1030 if (!$in_msg) {
1031 $in_msg = 1 if (/^\s*$/);
1032 $author = $1 if (/^author (.*>)/);
1033 } elsif (/^git-svn-id: /) {
1034 # skip this for now, we regenerate the
1035 # correct one on re-fetch anyways
1036 # TODO: set *:merge properties or like...
1037 } else {
1038 if (/^From:/ || /^Signed-off-by:/) {
1039 $saw_from = 1;
1040 }
1041 $msgbuf .= $_;
1042 }
1043 }
1044 $msgbuf =~ s/\s+$//s;
1045 if ($Git::SVN::_add_author_from && defined($author)
1046 && !$saw_from) {
1047 $msgbuf .= "\n\nFrom: $author";
1048 }
1049 print $log_fh $msgbuf or croak $!;
1050 command_close_pipe($msg_fh, $ctx);
1051 }
1052 close $log_fh or croak $!;
1053 if ($_edit || ($type eq 'tree')) {
1055 my $editor = $ENV{VISUAL} || $ENV{EDITOR} || 'vi';
1056 # TODO: strip out spaces, comments, like git-commit.sh
1057 system($editor, $commit_editmsg);
1058 }
1059 rename $commit_editmsg, $commit_msg or croak $!;
1060 open $log_fh, '<', $commit_msg or croak $!;
1061 { local $/; chomp($log_entry{log} = <$log_fh>); }
1062 close $log_fh or croak $!;
1063 unlink $commit_msg;
1064 \%log_entry;
1065}
1066sub s_to_file {
1068 my ($str, $file, $mode) = @_;
1069 open my $fd,'>',$file or croak $!;
1070 print $fd $str,"\n" or croak $!;
1071 close $fd or croak $!;
1072 chmod ($mode &~ umask, $file) if (defined $mode);
1073}
1074sub file_to_s {
1076 my $file = shift;
1077 open my $fd,'<',$file or croak "$!: file: $file\n";
1078 local $/;
1079 my $ret = <$fd>;
1080 close $fd or croak $!;
1081 $ret =~ s/\s*$//s;
1082 return $ret;
1083}
1084# '<svn username> = real-name <email address>' mapping based on git-svnimport:
1086sub load_authors {
1087 open my $authors, '<', $_authors or die "Can't open $_authors $!\n";
1088 my $log = $cmd eq 'log';
1089 while (<$authors>) {
1090 chomp;
1091 next unless /^(.+?|\(no author\))\s*=\s*(.+?)\s*<(.+)>\s*$/;
1092 my ($user, $name, $email) = ($1, $2, $3);
1093 if ($log) {
1094 $Git::SVN::Log::rusers{"$name <$email>"} = $user;
1095 } else {
1096 $users{$user} = [$name, $email];
1097 }
1098 }
1099 close $authors or croak $!;
1100}
1101# convert GetOpt::Long specs for use by git-config
1103sub read_repo_config {
1104 return unless -d $ENV{GIT_DIR};
1105 my $opts = shift;
1106 my @config_only;
1107 foreach my $o (keys %$opts) {
1108 # if we have mixedCase and a long option-only, then
1109 # it's a config-only variable that we don't need for
1110 # the command-line.
1111 push @config_only, $o if ($o =~ /[A-Z]/ && $o =~ /^[a-z]+$/i);
1112 my $v = $opts->{$o};
1113 my ($key) = ($o =~ /^([a-zA-Z\-]+)/);
1114 $key =~ s/-//g;
1115 my $arg = 'git-config';
1116 $arg .= ' --int' if ($o =~ /[:=]i$/);
1117 $arg .= ' --bool' if ($o !~ /[:=][sfi]$/);
1118 if (ref $v eq 'ARRAY') {
1119 chomp(my @tmp = `$arg --get-all svn.$key`);
1120 @$v = @tmp if @tmp;
1121 } else {
1122 chomp(my $tmp = `$arg --get svn.$key`);
1123 if ($tmp && !($arg =~ / --bool/ && $tmp eq 'false')) {
1124 $$v = $tmp;
1125 }
1126 }
1127 }
1128 delete @$opts{@config_only} if @config_only;
1129}
1130sub extract_metadata {
1132 my $id = shift or return (undef, undef, undef);
1133 my ($url, $rev, $uuid) = ($id =~ /^\s*git-svn-id:\s+(.*)\@(\d+)
1134 \s([a-f\d\-]+)$/x);
1135 if (!defined $rev || !$uuid || !$url) {
1136 # some of the original repositories I made had
1137 # identifiers like this:
1138 ($rev, $uuid) = ($id =~/^\s*git-svn-id:\s(\d+)\@([a-f\d\-]+)/);
1139 }
1140 return ($url, $rev, $uuid);
1141}
1142sub cmt_metadata {
1144 return extract_metadata((grep(/^git-svn-id: /,
1145 command(qw/cat-file commit/, shift)))[-1]);
1146}
1147sub working_head_info {
1149 my ($head, $refs) = @_;
1150 my @args = ('log', '--no-color', '--first-parent', '--pretty=medium');
1151 my ($fh, $ctx) = command_output_pipe(@args, $head);
1152 my $hash;
1153 my %max;
1154 while (<$fh>) {
1155 if ( m{^commit ($::sha1)$} ) {
1156 unshift @$refs, $hash if $hash and $refs;
1157 $hash = $1;
1158 next;
1159 }
1160 next unless s{^\s*(git-svn-id:)}{$1};
1161 my ($url, $rev, $uuid) = extract_metadata($_);
1162 if (defined $url && defined $rev) {
1163 next if $max{$url} and $max{$url} < $rev;
1164 if (my $gs = Git::SVN->find_by_url($url)) {
1165 my $c = $gs->rev_map_get($rev, $uuid);
1166 if ($c && $c eq $hash) {
1167 close $fh; # break the pipe
1168 return ($url, $rev, $uuid, $gs);
1169 } else {
1170 $max{$url} ||= $gs->rev_map_max;
1171 }
1172 }
1173 }
1174 }
1175 command_close_pipe($fh, $ctx);
1176 (undef, undef, undef, undef);
1177}
1178sub read_commit_parents {
1180 my ($parents, $c) = @_;
1181 chomp(my $p = command_oneline(qw/rev-list --parents -1/, $c));
1182 $p =~ s/^($c)\s*// or die "rev-list --parents -1 $c failed!\n";
1183 @{$parents->{$c}} = split(/ /, $p);
1184}
1185sub linearize_history {
1187 my ($gs, $refs) = @_;
1188 my %parents;
1189 foreach my $c (@$refs) {
1190 read_commit_parents(\%parents, $c);
1191 }
1192 my @linear_refs;
1194 my %skip = ();
1195 my $last_svn_commit = $gs->last_commit;
1196 foreach my $c (reverse @$refs) {
1197 next if $c eq $last_svn_commit;
1198 last if $skip{$c};
1199 unshift @linear_refs, $c;
1201 $skip{$c} = 1;
1202 # we only want the first parent to diff against for linear
1204 # history, we save the rest to inject when we finalize the
1205 # svn commit
1206 my $fp_a = verify_ref("$c~1");
1207 my $fp_b = shift @{$parents{$c}} if $parents{$c};
1208 if (!$fp_a || !$fp_b) {
1209 die "Commit $c\n",
1210 "has no parent commit, and therefore ",
1211 "nothing to diff against.\n",
1212 "You should be working from a repository ",
1213 "originally created by git-svn\n";
1214 }
1215 if ($fp_a ne $fp_b) {
1216 die "$c~1 = $fp_a, however parsing commit $c ",
1217 "revealed that:\n$c~1 = $fp_b\nBUG!\n";
1218 }
1219 foreach my $p (@{$parents{$c}}) {
1221 $skip{$p} = 1;
1222 }
1223 }
1224 (\@linear_refs, \%parents);
1225}
1226sub find_file_type_and_diff_status {
1228 my ($path) = @_;
1229 return ('dir', '') if $path eq '';
1230 my $diff_output =
1232 command_oneline(qw(diff --cached --name-status --), $path) || "";
1233 my $diff_status = (split(' ', $diff_output))[0] || "";
1234 my $ls_tree = command_oneline(qw(ls-tree HEAD), $path) || "";
1236 return (undef, undef) if !$diff_status && !$ls_tree;
1238 if ($diff_status eq "A") {
1240 return ("link", $diff_status) if -l $path;
1241 return ("dir", $diff_status) if -d $path;
1242 return ("file", $diff_status);
1243 }
1244 my $mode = (split(' ', $ls_tree))[0] || "";
1246 return ("link", $diff_status) if $mode eq "120000";
1248 return ("dir", $diff_status) if $mode eq "040000";
1249 return ("file", $diff_status);
1250}
1251sub md5sum {
1253 my $arg = shift;
1254 my $ref = ref $arg;
1255 my $md5 = Digest::MD5->new();
1256 if ($ref eq 'GLOB' || $ref eq 'IO::File') {
1257 $md5->addfile($arg) or croak $!;
1258 } elsif ($ref eq 'SCALAR') {
1259 $md5->add($$arg) or croak $!;
1260 } elsif (!$ref) {
1261 $md5->add($arg) or croak $!;
1262 } else {
1263 ::fatal "Can't provide MD5 hash for unknown ref type: '", $ref, "'";
1264 }
1265 return $md5->hexdigest();
1266}
1267package Git::SVN;
1269use strict;
1270use warnings;
1271use Fcntl qw/:DEFAULT :seek/;
1272use constant rev_map_fmt => 'NH40';
1273use vars qw/$default_repo_id $default_ref_id $_no_metadata $_follow_parent
1274 $_repack $_repack_flags $_use_svm_props $_head
1275 $_use_svnsync_props $no_reuse_existing $_minimize_url
1276 $_use_log_author $_add_author_from/;
1277use Carp qw/croak/;
1278use File::Path qw/mkpath/;
1279use File::Copy qw/copy/;
1280use IPC::Open3;
1281my ($_gc_nr, $_gc_period);
1283# properties that we do not log:
1285my %SKIP_PROP;
1286BEGIN {
1287 %SKIP_PROP = map { $_ => 1 } qw/svn:wc:ra_dav:version-url
1288 svn:special svn:executable
1289 svn:entry:committed-rev
1290 svn:entry:last-author
1291 svn:entry:uuid
1292 svn:entry:committed-date/;
1293 # some options are read globally, but can be overridden locally
1295 # per [svn-remote "..."] section. Command-line options will *NOT*
1296 # override options set in an [svn-remote "..."] section
1297 no strict 'refs';
1298 for my $option (qw/follow_parent no_metadata use_svm_props
1299 use_svnsync_props/) {
1300 my $key = $option;
1301 $key =~ tr/_//d;
1302 my $prop = "-$option";
1303 *$option = sub {
1304 my ($self) = @_;
1305 return $self->{$prop} if exists $self->{$prop};
1306 my $k = "svn-remote.$self->{repo_id}.$key";
1307 eval { command_oneline(qw/config --get/, $k) };
1308 if ($@) {
1309 $self->{$prop} = ${"Git::SVN::_$option"};
1310 } else {
1311 my $v = command_oneline(qw/config --bool/,$k);
1312 $self->{$prop} = $v eq 'false' ? 0 : 1;
1313 }
1314 return $self->{$prop};
1315 }
1316 }
1317}
1318my (%LOCKFILES, %INDEX_FILES);
1320END {
1321 unlink keys %LOCKFILES if %LOCKFILES;
1322 unlink keys %INDEX_FILES if %INDEX_FILES;
1323}
1324sub resolve_local_globs {
1326 my ($url, $fetch, $glob_spec) = @_;
1327 return unless defined $glob_spec;
1328 my $ref = $glob_spec->{ref};
1329 my $path = $glob_spec->{path};
1330 foreach (command(qw#for-each-ref --format=%(refname) refs/remotes#)) {
1331 next unless m#^refs/remotes/$ref->{regex}$#;
1332 my $p = $1;
1333 my $pathname = desanitize_refname($path->full_path($p));
1334 my $refname = desanitize_refname($ref->full_path($p));
1335 if (my $existing = $fetch->{$pathname}) {
1336 if ($existing ne $refname) {
1337 die "Refspec conflict:\n",
1338 "existing: refs/remotes/$existing\n",
1339 " globbed: refs/remotes/$refname\n";
1340 }
1341 my $u = (::cmt_metadata("refs/remotes/$refname"))[0];
1342 $u =~ s!^\Q$url\E(/|$)!! or die
1343 "refs/remotes/$refname: '$url' not found in '$u'\n";
1344 if ($pathname ne $u) {
1345 warn "W: Refspec glob conflict ",
1346 "(ref: refs/remotes/$refname):\n",
1347 "expected path: $pathname\n",
1348 " real path: $u\n",
1349 "Continuing ahead with $u\n";
1350 next;
1351 }
1352 } else {
1353 $fetch->{$pathname} = $refname;
1354 }
1355 }
1356}
1357sub parse_revision_argument {
1359 my ($base, $head) = @_;
1360 if (!defined $::_revision || $::_revision eq 'BASE:HEAD') {
1361 return ($base, $head);
1362 }
1363 return ($1, $2) if ($::_revision =~ /^(\d+):(\d+)$/);
1364 return ($::_revision, $::_revision) if ($::_revision =~ /^\d+$/);
1365 return ($head, $head) if ($::_revision eq 'HEAD');
1366 return ($base, $1) if ($::_revision =~ /^BASE:(\d+)$/);
1367 return ($1, $head) if ($::_revision =~ /^(\d+):HEAD$/);
1368 die "revision argument: $::_revision not understood by git-svn\n";
1369}
1370sub fetch_all {
1372 my ($repo_id, $remotes) = @_;
1373 if (ref $repo_id) {
1374 my $gs = $repo_id;
1375 $repo_id = undef;
1376 $repo_id = $gs->{repo_id};
1377 }
1378 $remotes ||= read_all_remotes();
1379 my $remote = $remotes->{$repo_id} or
1380 die "[svn-remote \"$repo_id\"] unknown\n";
1381 my $fetch = $remote->{fetch};
1382 my $url = $remote->{url} or die "svn-remote.$repo_id.url not defined\n";
1383 my (@gs, @globs);
1384 my $ra = Git::SVN::Ra->new($url);
1385 my $uuid = $ra->get_uuid;
1386 my $head = $ra->get_latest_revnum;
1387 my $base = defined $fetch ? $head : 0;
1388 # read the max revs for wildcard expansion (branches/*, tags/*)
1390 foreach my $t (qw/branches tags/) {
1391 defined $remote->{$t} or next;
1392 push @globs, $remote->{$t};
1393 my $max_rev = eval { tmp_config(qw/--int --get/,
1394 "svn-remote.$repo_id.${t}-maxRev") };
1395 if (defined $max_rev && ($max_rev < $base)) {
1396 $base = $max_rev;
1397 } elsif (!defined $max_rev) {
1398 $base = 0;
1399 }
1400 }
1401 if ($fetch) {
1403 foreach my $p (sort keys %$fetch) {
1404 my $gs = Git::SVN->new($fetch->{$p}, $repo_id, $p);
1405 my $lr = $gs->rev_map_max;
1406 if (defined $lr) {
1407 $base = $lr if ($lr < $base);
1408 }
1409 push @gs, $gs;
1410 }
1411 }
1412 ($base, $head) = parse_revision_argument($base, $head);
1414 $ra->gs_fetch_loop_common($base, $head, \@gs, \@globs);
1415}
1416sub read_all_remotes {
1418 my $r = {};
1419 my $use_svm_props = eval { command_oneline(qw/config --bool
1420 svn.useSvmProps/) };
1421 $use_svm_props = $use_svm_props eq 'true' if $use_svm_props;
1422 foreach (grep { s/^svn-remote\.// } command(qw/config -l/)) {
1423 if (m!^(.+)\.fetch=\s*(.*)\s*:\s*(.+)\s*$!) {
1424 my ($remote, $local_ref, $_remote_ref) = ($1, $2, $3);
1425 die("svn-remote.$remote: remote ref '$_remote_ref' "
1426 . "must start with 'refs/remotes/'\n")
1427 unless $_remote_ref =~ m{^refs/remotes/(.+)};
1428 my $remote_ref = $1;
1429 $local_ref =~ s{^/}{};
1430 $r->{$remote}->{fetch}->{$local_ref} = $remote_ref;
1431 $r->{$remote}->{svm} = {} if $use_svm_props;
1432 } elsif (m!^(.+)\.usesvmprops=\s*(.*)\s*$!) {
1433 $r->{$1}->{svm} = {};
1434 } elsif (m!^(.+)\.url=\s*(.*)\s*$!) {
1435 $r->{$1}->{url} = $2;
1436 } elsif (m!^(.+)\.(branches|tags)=
1437 (.*):refs/remotes/(.+)\s*$/!x) {
1438 my ($p, $g) = ($3, $4);
1439 my $rs = $r->{$1}->{$2} = {
1440 t => $2,
1441 remote => $1,
1442 path => Git::SVN::GlobSpec->new($p),
1443 ref => Git::SVN::GlobSpec->new($g) };
1444 if (length($rs->{ref}->{right}) != 0) {
1445 die "The '*' glob character must be the last ",
1446 "character of '$g'\n";
1447 }
1448 }
1449 }
1450 map {
1452 if (defined $r->{$_}->{svm}) {
1453 my $svm;
1454 eval {
1455 my $section = "svn-remote.$_";
1456 $svm = {
1457 source => tmp_config('--get',
1458 "$section.svm-source"),
1459 replace => tmp_config('--get',
1460 "$section.svm-replace"),
1461 }
1462 };
1463 $r->{$_}->{svm} = $svm;
1464 }
1465 } keys %$r;
1466 $r;
1468}
1469sub init_vars {
1471 $_gc_nr = $_gc_period = 1000;
1472 if (defined $_repack || defined $_repack_flags) {
1473 warn "Repack options are obsolete; they have no effect.\n";
1474 }
1475}
1476sub verify_remotes_sanity {
1478 return unless -d $ENV{GIT_DIR};
1479 my %seen;
1480 foreach (command(qw/config -l/)) {
1481 if (m!^svn-remote\.(?:.+)\.fetch=.*:refs/remotes/(\S+)\s*$!) {
1482 if ($seen{$1}) {
1483 die "Remote ref refs/remote/$1 is tracked by",
1484 "\n \"$_\"\nand\n \"$seen{$1}\"\n",
1485 "Please resolve this ambiguity in ",
1486 "your git configuration file before ",
1487 "continuing\n";
1488 }
1489 $seen{$1} = $_;
1490 }
1491 }
1492}
1493sub find_existing_remote {
1495 my ($url, $remotes) = @_;
1496 return undef if $no_reuse_existing;
1497 my $existing;
1498 foreach my $repo_id (keys %$remotes) {
1499 my $u = $remotes->{$repo_id}->{url} or next;
1500 next if $u ne $url;
1501 $existing = $repo_id;
1502 last;
1503 }
1504 $existing;
1505}
1506sub init_remote_config {
1508 my ($self, $url, $no_write) = @_;
1509 $url =~ s!/+$!!; # strip trailing slash
1510 my $r = read_all_remotes();
1511 my $existing = find_existing_remote($url, $r);
1512 if ($existing) {
1513 unless ($no_write) {
1514 print STDERR "Using existing ",
1515 "[svn-remote \"$existing\"]\n";
1516 }
1517 $self->{repo_id} = $existing;
1518 } elsif ($_minimize_url) {
1519 my $min_url = Git::SVN::Ra->new($url)->minimize_url;
1520 $existing = find_existing_remote($min_url, $r);
1521 if ($existing) {
1522 unless ($no_write) {
1523 print STDERR "Using existing ",
1524 "[svn-remote \"$existing\"]\n";
1525 }
1526 $self->{repo_id} = $existing;
1527 }
1528 if ($min_url ne $url) {
1529 unless ($no_write) {
1530 print STDERR "Using higher level of URL: ",
1531 "$url => $min_url\n";
1532 }
1533 my $old_path = $self->{path};
1534 $self->{path} = $url;
1535 $self->{path} =~ s!^\Q$min_url\E(/|$)!!;
1536 if (length $old_path) {
1537 $self->{path} .= "/$old_path";
1538 }
1539 $url = $min_url;
1540 }
1541 }
1542 my $orig_url;
1543 if (!$existing) {
1544 # verify that we aren't overwriting anything:
1545 $orig_url = eval {
1546 command_oneline('config', '--get',
1547 "svn-remote.$self->{repo_id}.url")
1548 };
1549 if ($orig_url && ($orig_url ne $url)) {
1550 die "svn-remote.$self->{repo_id}.url already set: ",
1551 "$orig_url\nwanted to set to: $url\n";
1552 }
1553 }
1554 my ($xrepo_id, $xpath) = find_ref($self->refname);
1555 if (defined $xpath) {
1556 die "svn-remote.$xrepo_id.fetch already set to track ",
1557 "$xpath:refs/remotes/", $self->refname, "\n";
1558 }
1559 unless ($no_write) {
1560 command_noisy('config',
1561 "svn-remote.$self->{repo_id}.url", $url);
1562 $self->{path} =~ s{^/}{};
1563 command_noisy('config', '--add',
1564 "svn-remote.$self->{repo_id}.fetch",
1565 "$self->{path}:".$self->refname);
1566 }
1567 $self->{url} = $url;
1568}
1569sub find_by_url { # repos_root and, path are optional
1571 my ($class, $full_url, $repos_root, $path) = @_;
1572 return undef unless defined $full_url;
1574 remove_username($full_url);
1575 remove_username($repos_root) if defined $repos_root;
1576 my $remotes = read_all_remotes();
1577 if (defined $full_url && defined $repos_root && !defined $path) {
1578 $path = $full_url;
1579 $path =~ s#^\Q$repos_root\E(?:/|$)##;
1580 }
1581 foreach my $repo_id (keys %$remotes) {
1582 my $u = $remotes->{$repo_id}->{url} or next;
1583 remove_username($u);
1584 next if defined $repos_root && $repos_root ne $u;
1585 my $fetch = $remotes->{$repo_id}->{fetch} || {};
1587 foreach (qw/branches tags/) {
1588 resolve_local_globs($u, $fetch,
1589 $remotes->{$repo_id}->{$_});
1590 }
1591 my $p = $path;
1592 my $rwr = rewrite_root({repo_id => $repo_id});
1593 my $svm = $remotes->{$repo_id}->{svm}
1594 if defined $remotes->{$repo_id}->{svm};
1595 unless (defined $p) {
1596 $p = $full_url;
1597 my $z = $u;
1598 my $prefix = '';
1599 if ($rwr) {
1600 $z = $rwr;
1601 } elsif (defined $svm) {
1602 $z = $svm->{source};
1603 $prefix = $svm->{replace};
1604 $prefix =~ s#^\Q$u\E(?:/|$)##;
1605 $prefix =~ s#/$##;
1606 }
1607 $p =~ s#^\Q$z\E(?:/|$)#$prefix# or next;
1608 }
1609 foreach my $f (keys %$fetch) {
1610 next if $f ne $p;
1611 return Git::SVN->new($fetch->{$f}, $repo_id, $f);
1612 }
1613 }
1614 undef;
1615}
1616sub init {
1618 my ($class, $url, $path, $repo_id, $ref_id, $no_write) = @_;
1619 my $self = _new($class, $repo_id, $ref_id, $path);
1620 if (defined $url) {
1621 $self->init_remote_config($url, $no_write);
1622 }
1623 $self;
1624}
1625sub find_ref {
1627 my ($ref_id) = @_;
1628 foreach (command(qw/config -l/)) {
1629 next unless m!^svn-remote\.(.+)\.fetch=
1630 \s*(.*)\s*:\s*refs/remotes/(.+)\s*$!x;
1631 my ($repo_id, $path, $ref) = ($1, $2, $3);
1632 if ($ref eq $ref_id) {
1633 $path = '' if ($path =~ m#^\./?#);
1634 return ($repo_id, $path);
1635 }
1636 }
1637 (undef, undef, undef);
1638}
1639sub new {
1641 my ($class, $ref_id, $repo_id, $path) = @_;
1642 if (defined $ref_id && !defined $repo_id && !defined $path) {
1643 ($repo_id, $path) = find_ref($ref_id);
1644 if (!defined $repo_id) {
1645 die "Could not find a \"svn-remote.*.fetch\" key ",
1646 "in the repository configuration matching: ",
1647 "refs/remotes/$ref_id\n";
1648 }
1649 }
1650 my $self = _new($class, $repo_id, $ref_id, $path);
1651 if (!defined $self->{path} || !length $self->{path}) {
1652 my $fetch = command_oneline('config', '--get',
1653 "svn-remote.$repo_id.fetch",
1654 ":refs/remotes/$ref_id\$") or
1655 die "Failed to read \"svn-remote.$repo_id.fetch\" ",
1656 "\":refs/remotes/$ref_id\$\" in config\n";
1657 ($self->{path}, undef) = split(/\s*:\s*/, $fetch);
1658 }
1659 $self->{url} = command_oneline('config', '--get',
1660 "svn-remote.$repo_id.url") or
1661 die "Failed to read \"svn-remote.$repo_id.url\" in config\n";
1662 $self->rebuild;
1663 $self;
1664}
1665sub refname {
1667 my ($refname) = "refs/remotes/$_[0]->{ref_id}" ;
1668 # It cannot end with a slash /, we'll throw up on this because
1670 # SVN can't have directories with a slash in their name, either:
1671 if ($refname =~ m{/$}) {
1672 die "ref: '$refname' ends with a trailing slash, this is ",
1673 "not permitted by git nor Subversion\n";
1674 }
1675 # It cannot have ASCII control character space, tilde ~, caret ^,
1677 # colon :, question-mark ?, asterisk *, space, or open bracket [
1678 # anywhere.
1679 #
1680 # Additionally, % must be escaped because it is used for escaping
1681 # and we want our escaped refname to be reversible
1682 $refname =~ s{([ \%~\^:\?\*\[\t])}{uc sprintf('%%%02x',ord($1))}eg;
1683 # no slash-separated component can begin with a dot .
1685 # /.* becomes /%2E*
1686 $refname =~ s{/\.}{/%2E}g;
1687 # It cannot have two consecutive dots .. anywhere
1689 # .. becomes %2E%2E
1690 $refname =~ s{\.\.}{%2E%2E}g;
1691 return $refname;
1693}
1694sub desanitize_refname {
1696 my ($refname) = @_;
1697 $refname =~ s{%(?:([0-9A-F]{2}))}{chr hex($1)}eg;
1698 return $refname;
1699}
1700sub svm_uuid {
1702 my ($self) = @_;
1703 return $self->{svm}->{uuid} if $self->svm;
1704 $self->ra;
1705 unless ($self->{svm}) {
1706 die "SVM UUID not cached, and reading remotely failed\n";
1707 }
1708 $self->{svm}->{uuid};
1709}
1710sub svm {
1712 my ($self) = @_;
1713 return $self->{svm} if $self->{svm};
1714 my $svm;
1715 # see if we have it in our config, first:
1716 eval {
1717 my $section = "svn-remote.$self->{repo_id}";
1718 $svm = {
1719 source => tmp_config('--get', "$section.svm-source"),
1720 uuid => tmp_config('--get', "$section.svm-uuid"),
1721 replace => tmp_config('--get', "$section.svm-replace"),
1722 }
1723 };
1724 if ($svm && $svm->{source} && $svm->{uuid} && $svm->{replace}) {
1725 $self->{svm} = $svm;
1726 }
1727 $self->{svm};
1728}
1729sub _set_svm_vars {
1731 my ($self, $ra) = @_;
1732 return $ra if $self->svm;
1733 my @err = ( "useSvmProps set, but failed to read SVM properties\n",
1735 "(svm:source, svm:uuid) ",
1736 "from the following URLs:\n" );
1737 sub read_svm_props {
1738 my ($self, $ra, $path, $r) = @_;
1739 my $props = ($ra->get_dir($path, $r))[2];
1740 my $src = $props->{'svm:source'};
1741 my $uuid = $props->{'svm:uuid'};
1742 return undef if (!$src || !$uuid);
1743 chomp($src, $uuid);
1745 $uuid =~ m{^[0-9a-f\-]{30,}$}
1747 or die "doesn't look right - svm:uuid is '$uuid'\n";
1748 # the '!' is used to mark the repos_root!/relative/path
1750 $src =~ s{/?!/?}{/};
1751 $src =~ s{/+$}{}; # no trailing slashes please
1752 # username is of no interest
1753 $src =~ s{(^[a-z\+]*://)[^/@]*@}{$1};
1754 my $replace = $ra->{url};
1756 $replace .= "/$path" if length $path;
1757 my $section = "svn-remote.$self->{repo_id}";
1759 tmp_config("$section.svm-source", $src);
1760 tmp_config("$section.svm-replace", $replace);
1761 tmp_config("$section.svm-uuid", $uuid);
1762 $self->{svm} = {
1763 source => $src,
1764 uuid => $uuid,
1765 replace => $replace
1766 };
1767 }
1768 my $r = $ra->get_latest_revnum;
1770 my $path = $self->{path};
1771 my %tried;
1772 while (length $path) {
1773 unless ($tried{"$self->{url}/$path"}) {
1774 return $ra if $self->read_svm_props($ra, $path, $r);
1775 $tried{"$self->{url}/$path"} = 1;
1776 }
1777 $path =~ s#/?[^/]+$##;
1778 }
1779 die "Path: '$path' should be ''\n" if $path ne '';
1780 return $ra if $self->read_svm_props($ra, $path, $r);
1781 $tried{"$self->{url}/$path"} = 1;
1782 if ($ra->{repos_root} eq $self->{url}) {
1784 die @err, (map { " $_\n" } keys %tried), "\n";
1785 }
1786 # nope, make sure we're connected to the repository root:
1788 my $ok;
1789 my @tried_b;
1790 $path = $ra->{svn_path};
1791 $ra = Git::SVN::Ra->new($ra->{repos_root});
1792 while (length $path) {
1793 unless ($tried{"$ra->{url}/$path"}) {
1794 $ok = $self->read_svm_props($ra, $path, $r);
1795 last if $ok;
1796 $tried{"$ra->{url}/$path"} = 1;
1797 }
1798 $path =~ s#/?[^/]+$##;
1799 }
1800 die "Path: '$path' should be ''\n" if $path ne '';
1801 $ok ||= $self->read_svm_props($ra, $path, $r);
1802 $tried{"$ra->{url}/$path"} = 1;
1803 if (!$ok) {
1804 die @err, (map { " $_\n" } keys %tried), "\n";
1805 }
1806 Git::SVN::Ra->new($self->{url});
1807}
1808sub svnsync {
1810 my ($self) = @_;
1811 return $self->{svnsync} if $self->{svnsync};
1812 if ($self->no_metadata) {
1814 die "Can't have both 'noMetadata' and ",
1815 "'useSvnsyncProps' options set!\n";
1816 }
1817 if ($self->rewrite_root) {
1818 die "Can't have both 'useSvnsyncProps' and 'rewriteRoot' ",
1819 "options set!\n";
1820 }
1821 my $svnsync;
1823 # see if we have it in our config, first:
1824 eval {
1825 my $section = "svn-remote.$self->{repo_id}";
1826 my $url = tmp_config('--get', "$section.svnsync-url");
1828 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1829 die "doesn't look right - svn:sync-from-url is '$url'\n";
1830 my $uuid = tmp_config('--get', "$section.svnsync-uuid");
1832 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1833 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1834 $svnsync = { url => $url, uuid => $uuid }
1836 };
1837 if ($svnsync && $svnsync->{url} && $svnsync->{uuid}) {
1838 return $self->{svnsync} = $svnsync;
1839 }
1840 my $err = "useSvnsyncProps set, but failed to read " .
1842 "svnsync property: svn:sync-from-";
1843 my $rp = $self->ra->rev_proplist(0);
1844 my $url = $rp->{'svn:sync-from-url'} or die $err . "url\n";
1846 ($url) = ($url =~ m{^([a-z\+]+://\S+)$}) or
1847 die "doesn't look right - svn:sync-from-url is '$url'\n";
1848 my $uuid = $rp->{'svn:sync-from-uuid'} or die $err . "uuid\n";
1850 ($uuid) = ($uuid =~ m{^([0-9a-f\-]{30,})$}) or
1851 die "doesn't look right - svn:sync-from-uuid is '$uuid'\n";
1852 my $section = "svn-remote.$self->{repo_id}";
1854 tmp_config('--add', "$section.svnsync-uuid", $uuid);
1855 tmp_config('--add', "$section.svnsync-url", $url);
1856 return $self->{svnsync} = { url => $url, uuid => $uuid };
1857}
1858# this allows us to memoize our SVN::Ra UUID locally and avoid a
1860# remote lookup (useful for 'git svn log').
1861sub ra_uuid {
1862 my ($self) = @_;
1863 unless ($self->{ra_uuid}) {
1864 my $key = "svn-remote.$self->{repo_id}.uuid";
1865 my $uuid = eval { tmp_config('--get', $key) };
1866 if (!$@ && $uuid && $uuid =~ /^([a-f\d\-]{30,})$/) {
1867 $self->{ra_uuid} = $uuid;
1868 } else {
1869 die "ra_uuid called without URL\n" unless $self->{url};
1870 $self->{ra_uuid} = $self->ra->get_uuid;
1871 tmp_config('--add', $key, $self->{ra_uuid});
1872 }
1873 }
1874 $self->{ra_uuid};
1875}
1876sub _set_repos_root {
1878 my ($self, $repos_root) = @_;
1879 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1880 $repos_root ||= $self->ra->{repos_root};
1881 tmp_config($k, $repos_root);
1882 $repos_root;
1883}
1884sub repos_root {
1886 my ($self) = @_;
1887 my $k = "svn-remote.$self->{repo_id}.reposRoot";
1888 eval { tmp_config('--get', $k) } || $self->_set_repos_root;
1889}
1890sub ra {
1892 my ($self) = shift;
1893 my $ra = Git::SVN::Ra->new($self->{url});
1894 $self->_set_repos_root($ra->{repos_root});
1895 if ($self->use_svm_props && !$self->{svm}) {
1896 if ($self->no_metadata) {
1897 die "Can't have both 'noMetadata' and ",
1898 "'useSvmProps' options set!\n";
1899 } elsif ($self->use_svnsync_props) {
1900 die "Can't have both 'useSvnsyncProps' and ",
1901 "'useSvmProps' options set!\n";
1902 }
1903 $ra = $self->_set_svm_vars($ra);
1904 $self->{-want_revprops} = 1;
1905 }
1906 $ra;
1907}
1908sub rel_path {
1910 my ($self) = @_;
1911 my $repos_root = $self->ra->{repos_root};
1912 return $self->{path} if ($self->{url} eq $repos_root);
1913 my $url = $self->{url} .
1914 (length $self->{path} ? "/$self->{path}" : $self->{path});
1915 $url =~ s!^\Q$repos_root\E(?:/+|$)!!g;
1916 $url;
1917}
1918# prop_walk(PATH, REV, SUB)
1920# -------------------------
1921# Recursively traverse PATH at revision REV and invoke SUB for each
1922# directory that contains a SVN property. SUB will be invoked as
1923# follows: &SUB(gs, path, props); where `gs' is this instance of
1924# Git::SVN, `path' the path to the directory where the properties
1925# `props' were found. The `path' will be relative to point of checkout,
1926# that is, if url://repo/trunk is the current Git branch, and that
1927# directory contains a sub-directory `d', SUB will be invoked with `/d/'
1928# as `path' (note the trailing `/').
1929sub prop_walk {
1930 my ($self, $path, $rev, $sub) = @_;
1931 $path =~ s#^/##;
1933 my ($dirent, undef, $props) = $self->ra->get_dir($path, $rev);
1934 $path =~ s#^/*#/#g;
1935 my $p = $path;
1936 # Strip the irrelevant part of the path.
1937 $p =~ s#^/+\Q$self->{path}\E(/|$)#/#;
1938 # Ensure the path is terminated by a `/'.
1939 $p =~ s#/*$#/#;
1940 # The properties contain all the internal SVN stuff nobody
1942 # (usually) cares about.
1943 my $interesting_props = 0;
1944 foreach (keys %{$props}) {
1945 # If it doesn't start with `svn:', it must be a
1946 # user-defined property.
1947 ++$interesting_props and next if $_ !~ /^svn:/;
1948 # FIXME: Fragile, if SVN adds new public properties,
1949 # this needs to be updated.
1950 ++$interesting_props if /^svn:(?:ignore|keywords|executable
1951 |eol-style|mime-type
1952 |externals|needs-lock)$/x;
1953 }
1954 &$sub($self, $p, $props) if $interesting_props;
1955 foreach (sort keys %$dirent) {
1957 next if $dirent->{$_}->{kind} != $SVN::Node::dir;
1958 $self->prop_walk($self->{path} . $p . $_, $rev, $sub);
1959 }
1960}
1961sub last_rev { ($_[0]->last_rev_commit)[0] }
1963sub last_commit { ($_[0]->last_rev_commit)[1] }
1964# returns the newest SVN revision number and newest commit SHA1
1966sub last_rev_commit {
1967 my ($self) = @_;
1968 if (defined $self->{last_rev} && defined $self->{last_commit}) {
1969 return ($self->{last_rev}, $self->{last_commit});
1970 }
1971 my $c = ::verify_ref($self->refname.'^0');
1972 if ($c && !$self->use_svm_props && !$self->no_metadata) {
1973 my $rev = (::cmt_metadata($c))[1];
1974 if (defined $rev) {
1975 ($self->{last_rev}, $self->{last_commit}) = ($rev, $c);
1976 return ($rev, $c);
1977 }
1978 }
1979 my $map_path = $self->map_path;
1980 unless (-e $map_path) {
1981 ($self->{last_rev}, $self->{last_commit}) = (undef, undef);
1982 return (undef, undef);
1983 }
1984 my ($rev, $commit) = $self->rev_map_max(1);
1985 ($self->{last_rev}, $self->{last_commit}) = ($rev, $commit);
1986 return ($rev, $commit);
1987}
1988sub get_fetch_range {
1990 my ($self, $min, $max) = @_;
1991 $max ||= $self->ra->get_latest_revnum;
1992 $min ||= $self->rev_map_max;
1993 (++$min, $max);
1994}
1995sub tmp_config {
1997 my (@args) = @_;
1998 my $old_def_config = "$ENV{GIT_DIR}/svn/config";
1999 my $config = "$ENV{GIT_DIR}/svn/.metadata";
2000 if (! -f $config && -f $old_def_config) {
2001 rename $old_def_config, $config or
2002 die "Failed rename $old_def_config => $config: $!\n";
2003 }
2004 my $old_config = $ENV{GIT_CONFIG};
2005 $ENV{GIT_CONFIG} = $config;
2006 $@ = undef;
2007 my @ret = eval {
2008 unless (-f $config) {
2009 mkfile($config);
2010 open my $fh, '>', $config or
2011 die "Can't open $config: $!\n";
2012 print $fh "; This file is used internally by ",
2013 "git-svn\n" or die
2014 "Couldn't write to $config: $!\n";
2015 print $fh "; You should not have to edit it\n" or
2016 die "Couldn't write to $config: $!\n";
2017 close $fh or die "Couldn't close $config: $!\n";
2018 }
2019 command('config', @args);
2020 };
2021 my $err = $@;
2022 if (defined $old_config) {
2023 $ENV{GIT_CONFIG} = $old_config;
2024 } else {
2025 delete $ENV{GIT_CONFIG};
2026 }
2027 die $err if $err;
2028 wantarray ? @ret : $ret[0];
2029}
2030sub tmp_index_do {
2032 my ($self, $sub) = @_;
2033 my $old_index = $ENV{GIT_INDEX_FILE};
2034 $ENV{GIT_INDEX_FILE} = $self->{index};
2035 $@ = undef;
2036 my @ret = eval {
2037 my ($dir, $base) = ($self->{index} =~ m#^(.*?)/?([^/]+)$#);
2038 mkpath([$dir]) unless -d $dir;
2039 &$sub;
2040 };
2041 my $err = $@;
2042 if (defined $old_index) {
2043 $ENV{GIT_INDEX_FILE} = $old_index;
2044 } else {
2045 delete $ENV{GIT_INDEX_FILE};
2046 }
2047 die $err if $err;
2048 wantarray ? @ret : $ret[0];
2049}
2050sub assert_index_clean {
2052 my ($self, $treeish) = @_;
2053 $self->tmp_index_do(sub {
2055 command_noisy('read-tree', $treeish) unless -e $self->{index};
2056 my $x = command_oneline('write-tree');
2057 my ($y) = (command(qw/cat-file commit/, $treeish) =~
2058 /^tree ($::sha1)/mo);
2059 return if $y eq $x;
2060 warn "Index mismatch: $y != $x\nrereading $treeish\n";
2062 unlink $self->{index} or die "unlink $self->{index}: $!\n";
2063 command_noisy('read-tree', $treeish);
2064 $x = command_oneline('write-tree');
2065 if ($y ne $x) {
2066 ::fatal "trees ($treeish) $y != $x\n",
2067 "Something is seriously wrong...";
2068 }
2069 });
2070}
2071sub get_commit_parents {
2073 my ($self, $log_entry) = @_;
2074 my (%seen, @ret, @tmp);
2075 # legacy support for 'set-tree'; this is only used by set_tree_cb:
2076 if (my $ip = $self->{inject_parents}) {
2077 if (my $commit = delete $ip->{$log_entry->{revision}}) {
2078 push @tmp, $commit;
2079 }
2080 }
2081 if (my $cur = ::verify_ref($self->refname.'^0')) {
2082 push @tmp, $cur;
2083 }
2084 if (my $ipd = $self->{inject_parents_dcommit}) {
2085 if (my $commit = delete $ipd->{$log_entry->{revision}}) {
2086 push @tmp, @$commit;
2087 }
2088 }
2089 push @tmp, $_ foreach (@{$log_entry->{parents}}, @tmp);
2090 while (my $p = shift @tmp) {
2091 next if $seen{$p};
2092 $seen{$p} = 1;
2093 push @ret, $p;
2094 # MAXPARENT is defined to 16 in commit-tree.c:
2095 last if @ret >= 16;
2096 }
2097 if (@tmp) {
2098 die "r$log_entry->{revision}: No room for parents:\n\t",
2099 join("\n\t", @tmp), "\n";
2100 }
2101 @ret;
2102}
2103sub rewrite_root {
2105 my ($self) = @_;
2106 return $self->{-rewrite_root} if exists $self->{-rewrite_root};
2107 my $k = "svn-remote.$self->{repo_id}.rewriteRoot";
2108 my $rwr = eval { command_oneline(qw/config --get/, $k) };
2109 if ($rwr) {
2110 $rwr =~ s#/+$##;
2111 if ($rwr !~ m#^[a-z\+]+://#) {
2112 die "$rwr is not a valid URL (key: $k)\n";
2113 }
2114 }
2115 $self->{-rewrite_root} = $rwr;
2116}
2117sub metadata_url {
2119 my ($self) = @_;
2120 ($self->rewrite_root || $self->{url}) .
2121 (length $self->{path} ? '/' . $self->{path} : '');
2122}
2123sub full_url {
2125 my ($self) = @_;
2126 $self->{url} . (length $self->{path} ? '/' . $self->{path} : '');
2127}
2128sub set_commit_header_env {
2131 my ($log_entry) = @_;
2132 my %env;
2133 foreach my $ned (qw/NAME EMAIL DATE/) {
2134 foreach my $ac (qw/AUTHOR COMMITTER/) {
2135 $env{"GIT_${ac}_${ned}"} = $ENV{"GIT_${ac}_${ned}"};
2136 }
2137 }
2138 $ENV{GIT_AUTHOR_NAME} = $log_entry->{name};
2140 $ENV{GIT_AUTHOR_EMAIL} = $log_entry->{email};
2141 $ENV{GIT_AUTHOR_DATE} = $ENV{GIT_COMMITTER_DATE} = $log_entry->{date};
2142 $ENV{GIT_COMMITTER_NAME} = (defined $log_entry->{commit_name})
2144 ? $log_entry->{commit_name}
2145 : $log_entry->{name};
2146 $ENV{GIT_COMMITTER_EMAIL} = (defined $log_entry->{commit_email})
2147 ? $log_entry->{commit_email}
2148 : $log_entry->{email};
2149 \%env;
2150}
2151sub restore_commit_header_env {
2153 my ($env) = @_;
2154 foreach my $ned (qw/NAME EMAIL DATE/) {
2155 foreach my $ac (qw/AUTHOR COMMITTER/) {
2156 my $k = "GIT_${ac}_${ned}";
2157 if (defined $env->{$k}) {
2158 $ENV{$k} = $env->{$k};
2159 } else {
2160 delete $ENV{$k};
2161 }
2162 }
2163 }
2164}
2165sub gc {
2167 command_noisy('gc', '--auto');
2168};
2169sub do_git_commit {
2171 my ($self, $log_entry) = @_;
2172 my $lr = $self->last_rev;
2173 if (defined $lr && $lr >= $log_entry->{revision}) {
2174 die "Last fetched revision of ", $self->refname,
2175 " was r$lr, but we are about to fetch: ",
2176 "r$log_entry->{revision}!\n";
2177 }
2178 if (my $c = $self->rev_map_get($log_entry->{revision})) {
2179 croak "$log_entry->{revision} = $c already exists! ",
2180 "Why are we refetching it?\n";
2181 }
2182 my $old_env = set_commit_header_env($log_entry);
2183 my $tree = $log_entry->{tree};
2184 if (!defined $tree) {
2185 $tree = $self->tmp_index_do(sub {
2186 command_oneline('write-tree') });
2187 }
2188 die "Tree is not a valid sha1: $tree\n" if $tree !~ /^$::sha1$/o;
2189 my @exec = ('git-commit-tree', $tree);
2191 foreach ($self->get_commit_parents($log_entry)) {
2192 push @exec, '-p', $_;
2193 }
2194 defined(my $pid = open3(my $msg_fh, my $out_fh, '>&STDERR', @exec))
2195 or croak $!;
2196 print $msg_fh $log_entry->{log} or croak $!;
2197 restore_commit_header_env($old_env);
2198 unless ($self->no_metadata) {
2199 print $msg_fh "\ngit-svn-id: $log_entry->{metadata}\n"
2200 or croak $!;
2201 }
2202 $msg_fh->flush == 0 or croak $!;
2203 close $msg_fh or croak $!;
2204 chomp(my $commit = do { local $/; <$out_fh> });
2205 close $out_fh or croak $!;
2206 waitpid $pid, 0;
2207 croak $? if $?;
2208 if ($commit !~ /^$::sha1$/o) {
2209 die "Failed to commit, invalid sha1: $commit\n";
2210 }
2211 $self->rev_map_set($log_entry->{revision}, $commit, 1);
2213 $self->{last_rev} = $log_entry->{revision};
2215 $self->{last_commit} = $commit;
2216 print "r$log_entry->{revision}";
2217 if (defined $log_entry->{svm_revision}) {
2218 print " (\@$log_entry->{svm_revision})";
2219 $self->rev_map_set($log_entry->{svm_revision}, $commit,
2220 0, $self->svm_uuid);
2221 }
2222 print " = $commit ($self->{ref_id})\n";
2223 if (--$_gc_nr == 0) {
2224 $_gc_nr = $_gc_period;
2225 gc();
2226 }
2227 return $commit;
2228}
2229sub match_paths {
2231 my ($self, $paths, $r) = @_;
2232 return 1 if $self->{path} eq '';
2233 if (my $path = $paths->{"/$self->{path}"}) {
2234 return ($path->{action} eq 'D') ? 0 : 1;
2235 }
2236 $self->{path_regex} ||= qr/^\/\Q$self->{path}\E\//;
2237 if (grep /$self->{path_regex}/, keys %$paths) {
2238 return 1;
2239 }
2240 my $c = '';
2241 foreach (split m#/#, $self->{path}) {
2242 $c .= "/$_";
2243 next unless ($paths->{$c} &&
2244 ($paths->{$c}->{action} =~ /^[AR]$/));
2245 if ($self->ra->check_path($self->{path}, $r) ==
2246 $SVN::Node::dir) {
2247 return 1;
2248 }
2249 }
2250 return 0;
2251}
2252sub find_parent_branch {
2254 my ($self, $paths, $rev) = @_;
2255 return undef unless $self->follow_parent;
2256 unless (defined $paths) {
2257 my $err_handler = $SVN::Error::handler;
2258 $SVN::Error::handler = \&Git::SVN::Ra::skip_unknown_revs;
2259 $self->ra->get_log([$self->{path}], $rev, $rev, 0, 1, 1, sub {
2260 $paths =
2261 Git::SVN::Ra::dup_changed_paths($_[0]) });
2262 $SVN::Error::handler = $err_handler;
2263 }
2264 return undef unless defined $paths;
2265 # look for a parent from another branch:
2267 my @b_path_components = split m#/#, $self->rel_path;
2268 my @a_path_components;
2269 my $i;
2270 while (@b_path_components) {
2271 $i = $paths->{'/'.join('/', @b_path_components)};
2272 last if $i && defined $i->{copyfrom_path};
2273 unshift(@a_path_components, pop(@b_path_components));
2274 }
2275 return undef unless defined $i && defined $i->{copyfrom_path};
2276 my $branch_from = $i->{copyfrom_path};
2277 if (@a_path_components) {
2278 print STDERR "branch_from: $branch_from => ";
2279 $branch_from .= '/'.join('/', @a_path_components);
2280 print STDERR $branch_from, "\n";
2281 }
2282 my $r = $i->{copyfrom_rev};
2283 my $repos_root = $self->ra->{repos_root};
2284 my $url = $self->ra->{url};
2285 my $new_url = $repos_root . $branch_from;
2286 print STDERR "Found possible branch point: ",
2287 "$new_url => ", $self->full_url, ", $r\n";
2288 $branch_from =~ s#^/##;
2289 my $gs = Git::SVN->find_by_url($new_url, $repos_root, $branch_from);
2290 unless ($gs) {
2291 my $ref_id = $self->{ref_id};
2292 $ref_id =~ s/\@\d+$//;
2293 $ref_id .= "\@$r";
2294 # just grow a tail if we're not unique enough :x
2295 $ref_id .= '-' while find_ref($ref_id);
2296 print STDERR "Initializing parent: $ref_id\n";
2297 my ($u, $p, $repo_id) = ($new_url, '', $ref_id);
2298 if ($u =~ s#^\Q$url\E(/|$)##) {
2299 $p = $u;
2300 $u = $url;
2301 $repo_id = $self->{repo_id};
2302 }
2303 $gs = Git::SVN->init($u, $p, $repo_id, $ref_id, 1);
2304 }
2305 my ($r0, $parent) = $gs->find_rev_before($r, 1);
2306 if (!defined $r0 || !defined $parent) {
2307 my ($base, $head) = parse_revision_argument(0, $r);
2308 if ($base <= $r) {
2309 $gs->fetch($base, $r);
2310 }
2311 ($r0, $parent) = $gs->last_rev_commit;
2312 }
2313 if (defined $r0 && defined $parent) {
2314 print STDERR "Found branch parent: ($self->{ref_id}) $parent\n";
2315 my $ed;
2316 if ($self->ra->can_do_switch) {
2317 $self->assert_index_clean($parent);
2318 print STDERR "Following parent with do_switch\n";
2319 # do_switch works with svn/trunk >= r22312, but that
2320 # is not included with SVN 1.4.3 (the latest version
2321 # at the moment), so we can't rely on it
2322 $self->{last_commit} = $parent;
2323 $ed = SVN::Git::Fetcher->new($self);
2324 $gs->ra->gs_do_switch($r0, $rev, $gs,
2325 $self->full_url, $ed)
2326 or die "SVN connection failed somewhere...\n";
2327 } elsif ($self->ra->trees_match($new_url, $r0,
2328 $self->full_url, $rev)) {
2329 print STDERR "Trees match:\n",
2330 " $new_url\@$r0\n",
2331 " ${\$self->full_url}\@$rev\n",
2332 "Following parent with no changes\n";
2333 $self->tmp_index_do(sub {
2334 command_noisy('read-tree', $parent);
2335 });
2336 $self->{last_commit} = $parent;
2337 } else {
2338 print STDERR "Following parent with do_update\n";
2339 $ed = SVN::Git::Fetcher->new($self);
2340 $self->ra->gs_do_update($rev, $rev, $self, $ed)
2341 or die "SVN connection failed somewhere...\n";
2342 }
2343 print STDERR "Successfully followed parent\n";
2344 return $self->make_log_entry($rev, [$parent], $ed);
2345 }
2346 return undef;
2347}
2348sub do_fetch {
2350 my ($self, $paths, $rev) = @_;
2351 my $ed;
2352 my ($last_rev, @parents);
2353 if (my $lc = $self->last_commit) {
2354 # we can have a branch that was deleted, then re-added
2355 # under the same name but copied from another path, in
2356 # which case we'll have multiple parents (we don't
2357 # want to break the original ref, nor lose copypath info):
2358 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2359 push @{$log_entry->{parents}}, $lc;
2360 return $log_entry;
2361 }
2362 $ed = SVN::Git::Fetcher->new($self);
2363 $last_rev = $self->{last_rev};
2364 $ed->{c} = $lc;
2365 @parents = ($lc);
2366 } else {
2367 $last_rev = $rev;
2368 if (my $log_entry = $self->find_parent_branch($paths, $rev)) {
2369 return $log_entry;
2370 }
2371 $ed = SVN::Git::Fetcher->new($self);
2372 }
2373 unless ($self->ra->gs_do_update($last_rev, $rev, $self, $ed)) {
2374 die "SVN connection failed somewhere...\n";
2375 }
2376 $self->make_log_entry($rev, \@parents, $ed);
2377}
2378sub get_untracked {
2380 my ($self, $ed) = @_;
2381 my @out;
2382 my $h = $ed->{empty};
2383 foreach (sort keys %$h) {
2384 my $act = $h->{$_} ? '+empty_dir' : '-empty_dir';
2385 push @out, " $act: " . uri_encode($_);
2386 warn "W: $act: $_\n";
2387 }
2388 foreach my $t (qw/dir_prop file_prop/) {
2389 $h = $ed->{$t} or next;
2390 foreach my $path (sort keys %$h) {
2391 my $ppath = $path eq '' ? '.' : $path;
2392 foreach my $prop (sort keys %{$h->{$path}}) {
2393 next if $SKIP_PROP{$prop};
2394 my $v = $h->{$path}->{$prop};
2395 my $t_ppath_prop = "$t: " .
2396 uri_encode($ppath) . ' ' .
2397 uri_encode($prop);
2398 if (defined $v) {
2399 push @out, " +$t_ppath_prop " .
2400 uri_encode($v);
2401 } else {
2402 push @out, " -$t_ppath_prop";
2403 }
2404 }
2405 }
2406 }
2407 foreach my $t (qw/absent_file absent_directory/) {
2408 $h = $ed->{$t} or next;
2409 foreach my $parent (sort keys %$h) {
2410 foreach my $path (sort @{$h->{$parent}}) {
2411 push @out, " $t: " .
2412 uri_encode("$parent/$path");
2413 warn "W: $t: $parent/$path ",
2414 "Insufficient permissions?\n";
2415 }
2416 }
2417 }
2418 \@out;
2419}
2420sub parse_svn_date {
2422 my $date = shift || return '+0000 1970-01-01 00:00:00';
2423 my ($Y,$m,$d,$H,$M,$S) = ($date =~ /^(\d{4})\-(\d\d)\-(\d\d)T
2424 (\d\d)\:(\d\d)\:(\d\d).\d+Z$/x) or
2425 croak "Unable to parse date: $date\n";
2426 "+0000 $Y-$m-$d $H:$M:$S";
2427}
2428sub check_author {
2430 my ($author) = @_;
2431 if (!defined $author || length $author == 0) {
2432 $author = '(no author)';
2433 } elsif (defined $::_authors && ! defined $::users{$author}) {
2434 die "Author: $author not defined in $::_authors file\n";
2435 }
2436 $author;
2437}
2438sub make_log_entry {
2440 my ($self, $rev, $parents, $ed) = @_;
2441 my $untracked = $self->get_untracked($ed);
2442 open my $un, '>>', "$self->{dir}/unhandled.log" or croak $!;
2444 print $un "r$rev\n" or croak $!;
2445 print $un $_, "\n" foreach @$untracked;
2446 my %log_entry = ( parents => $parents || [], revision => $rev,
2447 log => '');
2448 my $headrev;
2450 my $logged = delete $self->{logged_rev_props};
2451 if (!$logged || $self->{-want_revprops}) {
2452 my $rp = $self->ra->rev_proplist($rev);
2453 foreach (sort keys %$rp) {
2454 my $v = $rp->{$_};
2455 if (/^svn:(author|date|log)$/) {
2456 $log_entry{$1} = $v;
2457 } elsif ($_ eq 'svm:headrev') {
2458 $headrev = $v;
2459 } else {
2460 print $un " rev_prop: ", uri_encode($_), ' ',
2461 uri_encode($v), "\n";
2462 }
2463 }
2464 } else {
2465 map { $log_entry{$_} = $logged->{$_} } keys %$logged;
2466 }
2467 close $un or croak $!;
2468 $log_entry{date} = parse_svn_date($log_entry{date});
2470 $log_entry{log} .= "\n";
2471 my $author = $log_entry{author} = check_author($log_entry{author});
2472 my ($name, $email) = defined $::users{$author} ? @{$::users{$author}}
2473 : ($author, undef);
2474 my ($commit_name, $commit_email) = ($name, $email);
2476 if ($_use_log_author) {
2477 my $name_field;
2478 if ($log_entry{log} =~ /From:\s+(.*\S)\s*\n/i) {
2479 $name_field = $1;
2480 } elsif ($log_entry{log} =~ /Signed-off-by:\s+(.*\S)\s*\n/i) {
2481 $name_field = $1;
2482 }
2483 if (!defined $name_field) {
2484 if (!defined $email) {
2485 $email = $name;
2486 }
2487 } elsif ($name_field =~ /(.*?)\s+<(.*)>/) {
2488 ($name, $email) = ($1, $2);
2489 } elsif ($name_field =~ /(.*)@/) {
2490 ($name, $email) = ($1, $name_field);
2491 } else {
2492 ($name, $email) = ($name_field, $name_field);
2493 }
2494 }
2495 if (defined $headrev && $self->use_svm_props) {
2496 if ($self->rewrite_root) {
2497 die "Can't have both 'useSvmProps' and 'rewriteRoot' ",
2498 "options set!\n";
2499 }
2500 my ($uuid, $r) = $headrev =~ m{^([a-f\d\-]{30,}):(\d+)$};
2501 # we don't want "SVM: initializing mirror for junk" ...
2502 return undef if $r == 0;
2503 my $svm = $self->svm;
2504 if ($uuid ne $svm->{uuid}) {
2505 die "UUID mismatch on SVM path:\n",
2506 "expected: $svm->{uuid}\n",
2507 " got: $uuid\n";
2508 }
2509 my $full_url = $self->full_url;
2510 $full_url =~ s#^\Q$svm->{replace}\E(/|$)#$svm->{source}$1# or
2511 die "Failed to replace '$svm->{replace}' with ",
2512 "'$svm->{source}' in $full_url\n";
2513 # throw away username for storing in records
2514 remove_username($full_url);
2515 $log_entry{metadata} = "$full_url\@$r $uuid";
2516 $log_entry{svm_revision} = $r;
2517 $email ||= "$author\@$uuid";
2518 $commit_email ||= "$author\@$uuid";
2519 } elsif ($self->use_svnsync_props) {
2520 my $full_url = $self->svnsync->{url};
2521 $full_url .= "/$self->{path}" if length $self->{path};
2522 remove_username($full_url);
2523 my $uuid = $self->svnsync->{uuid};
2524 $log_entry{metadata} = "$full_url\@$rev $uuid";
2525 $email ||= "$author\@$uuid";
2526 $commit_email ||= "$author\@$uuid";
2527 } else {
2528 my $url = $self->metadata_url;
2529 remove_username($url);
2530 $log_entry{metadata} = "$url\@$rev " .
2531 $self->ra->get_uuid;
2532 $email ||= "$author\@" . $self->ra->get_uuid;
2533 $commit_email ||= "$author\@" . $self->ra->get_uuid;
2534 }
2535 $log_entry{name} = $name;
2536 $log_entry{email} = $email;
2537 $log_entry{commit_name} = $commit_name;
2538 $log_entry{commit_email} = $commit_email;
2539 \%log_entry;
2540}
2541sub fetch {
2543 my ($self, $min_rev, $max_rev, @parents) = @_;
2544 my ($last_rev, $last_commit) = $self->last_rev_commit;
2545 my ($base, $head) = $self->get_fetch_range($min_rev, $max_rev);
2546 $self->ra->gs_fetch_loop_common($base, $head, [$self]);
2547}
2548sub set_tree_cb {
2550 my ($self, $log_entry, $tree, $rev, $date, $author) = @_;
2551 $self->{inject_parents} = { $rev => $tree };
2552 $self->fetch(undef, undef);
2553}
2554sub set_tree {
2556 my ($self, $tree) = (shift, shift);
2557 my $log_entry = ::get_commit_entry($tree);
2558 unless ($self->{last_rev}) {
2559 fatal("Must have an existing revision to commit");
2560 }
2561 my %ed_opts = ( r => $self->{last_rev},
2562 log => $log_entry->{log},
2563 ra => $self->ra,
2564 tree_a => $self->{last_commit},
2565 tree_b => $tree,
2566 editor_cb => sub {
2567 $self->set_tree_cb($log_entry, $tree, @_) },
2568 svn_path => $self->{path} );
2569 if (!SVN::Git::Editor->new(\%ed_opts)->apply_diff) {
2570 print "No changes\nr$self->{last_rev} = $tree\n";
2571 }
2572}
2573sub rebuild_from_rev_db {
2575 my ($self, $path) = @_;
2576 my $r = -1;
2577 open my $fh, '<', $path or croak "open: $!";
2578 binmode $fh or croak "binmode: $!";
2579 while (<$fh>) {
2580 length($_) == 41 or croak "inconsistent size in ($_) != 41";
2581 chomp($_);
2582 ++$r;
2583 next if $_ eq ('0' x 40);
2584 $self->rev_map_set($r, $_);
2585 print "r$r = $_\n";
2586 }
2587 close $fh or croak "close: $!";
2588 unlink $path or croak "unlink: $!";
2589}
2590sub rebuild {
2592 my ($self) = @_;
2593 my $map_path = $self->map_path;
2594 return if (-e $map_path && ! -z $map_path);
2595 return unless ::verify_ref($self->refname.'^0');
2596 if ($self->use_svm_props || $self->no_metadata) {
2597 my $rev_db = $self->rev_db_path;
2598 $self->rebuild_from_rev_db($rev_db);
2599 if ($self->use_svm_props) {
2600 my $svm_rev_db = $self->rev_db_path($self->svm_uuid);
2601 $self->rebuild_from_rev_db($svm_rev_db);
2602 }
2603 $self->unlink_rev_db_symlink;
2604 return;
2605 }
2606 print "Rebuilding $map_path ...\n";
2607 my ($log, $ctx) =
2608 command_output_pipe(qw/rev-list --pretty=raw --no-color --reverse/,
2609 $self->refname, '--');
2610 my $metadata_url = $self->metadata_url;
2611 remove_username($metadata_url);
2612 my $svn_uuid = $self->ra_uuid;
2613 my $c;
2614 while (<$log>) {
2615 if ( m{^commit ($::sha1)$} ) {
2616 $c = $1;
2617 next;
2618 }
2619 next unless s{^\s*(git-svn-id:)}{$1};
2620 my ($url, $rev, $uuid) = ::extract_metadata($_);
2621 remove_username($url);
2622 # ignore merges (from set-tree)
2624 next if (!defined $rev || !$uuid);
2625 # if we merged or otherwise started elsewhere, this is
2627 # how we break out of it
2628 if (($uuid ne $svn_uuid) ||
2629 ($metadata_url && $url && ($url ne $metadata_url))) {
2630 next;
2631 }
2632 $self->rev_map_set($rev, $c);
2634 print "r$rev = $c\n";
2635 }
2636 command_close_pipe($log, $ctx);
2637 print "Done rebuilding $map_path\n";
2638 my $rev_db_path = $self->rev_db_path;
2639 if (-f $self->rev_db_path) {
2640 unlink $self->rev_db_path or croak "unlink: $!";
2641 }
2642 $self->unlink_rev_db_symlink;
2643}
2644# rev_map:
2646# Tie::File seems to be prone to offset errors if revisions get sparse,
2647# it's not that fast, either. Tie::File is also not in Perl 5.6. So
2648# one of my favorite modules is out :< Next up would be one of the DBM
2649# modules, but I'm not sure which is most portable...
2650#
2651# This is the replacement for the rev_db format, which was too big
2652# and inefficient for large repositories with a lot of sparse history
2653# (mainly tags)
2654#
2655# The format is this:
2656# - 24 bytes for every record,
2657# * 4 bytes for the integer representing an SVN revision number
2658# * 20 bytes representing the sha1 of a git commit
2659# - No empty padding records like the old format
2660# (except the last record, which can be overwritten)
2661# - new records are written append-only since SVN revision numbers
2662# increase monotonically
2663# - lookups on SVN revision number are done via a binary search
2664# - Piping the file to xxd -c24 is a good way of dumping it for
2665# viewing or editing (piped back through xxd -r), should the need
2666# ever arise.
2667# - The last record can be padding revision with an all-zero sha1
2668# This is used to optimize fetch performance when using multiple
2669# "fetch" directives in .git/config
2670#
2671# These files are disposable unless noMetadata or useSvmProps is set
2672sub _rev_map_set {
2674 my ($fh, $rev, $commit) = @_;
2675 binmode $fh or croak "binmode: $!";
2677 my $size = (stat($fh))[7];
2678 ($size % 24) == 0 or croak "inconsistent size: $size";
2679 my $wr_offset = 0;
2681 if ($size > 0) {
2682 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2683 my $read = sysread($fh, my $buf, 24) or croak "read: $!";
2684 $read == 24 or croak "read only $read bytes (!= 24)";
2685 my ($last_rev, $last_commit) = unpack(rev_map_fmt, $buf);
2686 if ($last_commit eq ('0' x40)) {
2687 if ($size >= 48) {
2688 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2689 $read = sysread($fh, $buf, 24) or
2690 croak "read: $!";
2691 $read == 24 or
2692 croak "read only $read bytes (!= 24)";
2693 ($last_rev, $last_commit) =
2694 unpack(rev_map_fmt, $buf);
2695 if ($last_commit eq ('0' x40)) {
2696 croak "inconsistent .rev_map\n";
2697 }
2698 }
2699 if ($last_rev >= $rev) {
2700 croak "last_rev is higher!: $last_rev >= $rev";
2701 }
2702 $wr_offset = -24;
2703 }
2704 }
2705 sysseek($fh, $wr_offset, SEEK_END) or croak "seek: $!";
2706 syswrite($fh, pack(rev_map_fmt, $rev, $commit), 24) == 24 or
2707 croak "write: $!";
2708}
2709sub mkfile {
2711 my ($path) = @_;
2712 unless (-e $path) {
2713 my ($dir, $base) = ($path =~ m#^(.*?)/?([^/]+)$#);
2714 mkpath([$dir]) unless -d $dir;
2715 open my $fh, '>>', $path or die "Couldn't create $path: $!\n";
2716 close $fh or die "Couldn't close (create) $path: $!\n";
2717 }
2718}
2719sub rev_map_set {
2721 my ($self, $rev, $commit, $update_ref, $uuid) = @_;
2722 length $commit == 40 or die "arg3 must be a full SHA1 hexsum\n";
2723 my $db = $self->map_path($uuid);
2724 my $db_lock = "$db.lock";
2725 my $sig;
2726 if ($update_ref) {
2727 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2728 $SIG{USR1} = $SIG{USR2} = sub { $sig = $_[0] };
2729 }
2730 mkfile($db);
2731 $LOCKFILES{$db_lock} = 1;
2733 my $sync;
2734 # both of these options make our .rev_db file very, very important
2735 # and we can't afford to lose it because rebuild() won't work
2736 if ($self->use_svm_props || $self->no_metadata) {
2737 $sync = 1;
2738 copy($db, $db_lock) or die "rev_map_set(@_): ",
2739 "Failed to copy: ",
2740 "$db => $db_lock ($!)\n";
2741 } else {
2742 rename $db, $db_lock or die "rev_map_set(@_): ",
2743 "Failed to rename: ",
2744 "$db => $db_lock ($!)\n";
2745 }
2746 sysopen(my $fh, $db_lock, O_RDWR | O_CREAT)
2748 or croak "Couldn't open $db_lock: $!\n";
2749 _rev_map_set($fh, $rev, $commit);
2750 if ($sync) {
2751 $fh->flush or die "Couldn't flush $db_lock: $!\n";
2752 $fh->sync or die "Couldn't sync $db_lock: $!\n";
2753 }
2754 close $fh or croak $!;
2755 if ($update_ref) {
2756 $_head = $self;
2757 command_noisy('update-ref', '-m', "r$rev",
2758 $self->refname, $commit);
2759 }
2760 rename $db_lock, $db or die "rev_map_set(@_): ", "Failed to rename: ",
2761 "$db_lock => $db ($!)\n";
2762 delete $LOCKFILES{$db_lock};
2763 if ($update_ref) {
2764 $SIG{INT} = $SIG{HUP} = $SIG{TERM} = $SIG{ALRM} = $SIG{PIPE} =
2765 $SIG{USR1} = $SIG{USR2} = 'DEFAULT';
2766 kill $sig, $$ if defined $sig;
2767 }
2768}
2769# If want_commit, this will return an array of (rev, commit) where
2771# commit _must_ be a valid commit in the archive.
2772# Otherwise, it'll return the max revision (whether or not the
2773# commit is valid or just a 0x40 placeholder).
2774sub rev_map_max {
2775 my ($self, $want_commit) = @_;
2776 $self->rebuild;
2777 my $map_path = $self->map_path;
2778 stat $map_path or return $want_commit ? (0, undef) : 0;
2779 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2780 binmode $fh or croak "binmode: $!";
2781 my $size = (stat($fh))[7];
2782 ($size % 24) == 0 or croak "inconsistent size: $size";
2783 if ($size == 0) {
2785 close $fh or croak "close: $!";
2786 return $want_commit ? (0, undef) : 0;
2787 }
2788 sysseek($fh, -24, SEEK_END) or croak "seek: $!";
2790 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2791 my ($r, $c) = unpack(rev_map_fmt, $buf);
2792 if ($want_commit && $c eq ('0' x40)) {
2793 if ($size < 48) {
2794 return $want_commit ? (0, undef) : 0;
2795 }
2796 sysseek($fh, -48, SEEK_END) or croak "seek: $!";
2797 sysread($fh, $buf, 24) == 24 or croak "read: $!";
2798 ($r, $c) = unpack(rev_map_fmt, $buf);
2799 if ($c eq ('0'x40)) {
2800 croak "Penultimate record is all-zeroes in $map_path";
2801 }
2802 }
2803 close $fh or croak "close: $!";
2804 $want_commit ? ($r, $c) : $r;
2805}
2806sub rev_map_get {
2808 my ($self, $rev, $uuid) = @_;
2809 my $map_path = $self->map_path($uuid);
2810 return undef unless -e $map_path;
2811 sysopen(my $fh, $map_path, O_RDONLY) or croak "open: $!";
2813 binmode $fh or croak "binmode: $!";
2814 my $size = (stat($fh))[7];
2815 ($size % 24) == 0 or croak "inconsistent size: $size";
2816 if ($size == 0) {
2818 close $fh or croak "close: $fh";
2819 return undef;
2820 }
2821 my ($l, $u) = (0, $size - 24);
2823 my ($r, $c, $buf);
2824 while ($l <= $u) {
2826 my $i = int(($l/24 + $u/24) / 2) * 24;
2827 sysseek($fh, $i, SEEK_SET) or croak "seek: $!";
2828 sysread($fh, my $buf, 24) == 24 or croak "read: $!";
2829 my ($r, $c) = unpack('NH40', $buf);
2830 if ($r < $rev) {
2832 $l = $i + 24;
2833 } elsif ($r > $rev) {
2834 $u = $i - 24;
2835 } else { # $r == $rev
2836 close($fh) or croak "close: $!";
2837 return $c eq ('0' x 40) ? undef : $c;
2838 }
2839 }
2840 close($fh) or croak "close: $!";
2841 undef;
2842}
2843# Finds the first svn revision that exists on (if $eq_ok is true) or
2845# before $rev for the current branch. It will not search any lower
2846# than $min_rev. Returns the git commit hash and svn revision number
2847# if found, else (undef, undef).
2848sub find_rev_before {
2849 my ($self, $rev, $eq_ok, $min_rev) = @_;
2850 --$rev unless $eq_ok;
2851 $min_rev ||= 1;
2852 while ($rev >= $min_rev) {
2853 if (my $c = $self->rev_map_get($rev)) {
2854 return ($rev, $c);
2855 }
2856 --$rev;
2857 }
2858 return (undef, undef);
2859}
2860# Finds the first svn revision that exists on (if $eq_ok is true) or
2862# after $rev for the current branch. It will not search any higher
2863# than $max_rev. Returns the git commit hash and svn revision number
2864# if found, else (undef, undef).
2865sub find_rev_after {
2866 my ($self, $rev, $eq_ok, $max_rev) = @_;
2867 ++$rev unless $eq_ok;
2868 $max_rev ||= $self->rev_map_max;
2869 while ($rev <= $max_rev) {
2870 if (my $c = $self->rev_map_get($rev)) {
2871 return ($rev, $c);
2872 }
2873 ++$rev;
2874 }
2875 return (undef, undef);
2876}
2877sub _new {
2879 my ($class, $repo_id, $ref_id, $path) = @_;
2880 unless (defined $repo_id && length $repo_id) {
2881 $repo_id = $Git::SVN::default_repo_id;
2882 }
2883 unless (defined $ref_id && length $ref_id) {
2884 $_[2] = $ref_id = $Git::SVN::default_ref_id;
2885 }
2886 $_[1] = $repo_id;
2887 my $dir = "$ENV{GIT_DIR}/svn/$ref_id";
2888 $_[3] = $path = '' unless (defined $path);
2889 mkpath(["$ENV{GIT_DIR}/svn"]);
2890 bless {
2891 ref_id => $ref_id, dir => $dir, index => "$dir/index",
2892 path => $path, config => "$ENV{GIT_DIR}/svn/config",
2893 map_root => "$dir/.rev_map", repo_id => $repo_id }, $class;
2894}
2895# for read-only access of old .rev_db formats
2897sub unlink_rev_db_symlink {
2898 my ($self) = @_;
2899 my $link = $self->rev_db_path;
2900 $link =~ s/\.[\w-]+$// or croak "missing UUID at the end of $link";
2901 if (-l $link) {
2902 unlink $link or croak "unlink: $link failed!";
2903 }
2904}
2905sub rev_db_path {
2907 my ($self, $uuid) = @_;
2908 my $db_path = $self->map_path($uuid);
2909 $db_path =~ s{/\.rev_map\.}{/\.rev_db\.}
2910 or croak "map_path: $db_path does not contain '/.rev_map.' !";
2911 $db_path;
2912}
2913# the new replacement for .rev_db
2915sub map_path {
2916 my ($self, $uuid) = @_;
2917 $uuid ||= $self->ra_uuid;
2918 "$self->{map_root}.$uuid";
2919}
2920sub uri_encode {
2922 my ($f) = @_;
2923 $f =~ s#([^a-zA-Z0-9\*!\:_\./\-])#uc sprintf("%%%02x",ord($1))#eg;
2924 $f
2925}
2926sub remove_username {
2928 $_[0] =~ s{^([^:]*://)[^@]+@}{$1};
2929}
2930package Git::SVN::Prompt;
2932use strict;
2933use warnings;
2934require SVN::Core;
2935use vars qw/$_no_auth_cache $_username/;
2936sub simple {
2938 my ($cred, $realm, $default_username, $may_save, $pool) = @_;
2939 $may_save = undef if $_no_auth_cache;
2940 $default_username = $_username if defined $_username;
2941 if (defined $default_username && length $default_username) {
2942 if (defined $realm && length $realm) {
2943 print STDERR "Authentication realm: $realm\n";
2944 STDERR->flush;
2945 }
2946 $cred->username($default_username);
2947 } else {
2948 username($cred, $realm, $may_save, $pool);
2949 }
2950 $cred->password(_read_password("Password for '" .
2951 $cred->username . "': ", $realm));
2952 $cred->may_save($may_save);
2953 $SVN::_Core::SVN_NO_ERROR;
2954}
2955sub ssl_server_trust {
2957 my ($cred, $realm, $failures, $cert_info, $may_save, $pool) = @_;
2958 $may_save = undef if $_no_auth_cache;
2959 print STDERR "Error validating server certificate for '$realm':\n";
2960 {
2961 no warnings 'once';
2962 # All variables SVN::Auth::SSL::* are used only once,
2963 # so we're shutting up Perl warnings about this.
2964 if ($failures & $SVN::Auth::SSL::UNKNOWNCA) {
2965 print STDERR " - The certificate is not issued ",
2966 "by a trusted authority. Use the\n",
2967 " fingerprint to validate ",
2968 "the certificate manually!\n";
2969 }
2970 if ($failures & $SVN::Auth::SSL::CNMISMATCH) {
2971 print STDERR " - The certificate hostname ",
2972 "does not match.\n";
2973 }
2974 if ($failures & $SVN::Auth::SSL::NOTYETVALID) {
2975 print STDERR " - The certificate is not yet valid.\n";
2976 }
2977 if ($failures & $SVN::Auth::SSL::EXPIRED) {
2978 print STDERR " - The certificate has expired.\n";
2979 }
2980 if ($failures & $SVN::Auth::SSL::OTHER) {
2981 print STDERR " - The certificate has ",
2982 "an unknown error.\n";
2983 }
2984 } # no warnings 'once'
2985 printf STDERR
2986 "Certificate information:\n".
2987 " - Hostname: %s\n".
2988 " - Valid: from %s until %s\n".
2989 " - Issuer: %s\n".
2990 " - Fingerprint: %s\n",
2991 map $cert_info->$_, qw(hostname valid_from valid_until
2992 issuer_dname fingerprint);
2993 my $choice;
2994prompt:
2995 print STDERR $may_save ?
2996 "(R)eject, accept (t)emporarily or accept (p)ermanently? " :
2997 "(R)eject or accept (t)emporarily? ";
2998 STDERR->flush;
2999 $choice = lc(substr(<STDIN> || 'R', 0, 1));
3000 if ($choice =~ /^t$/i) {
3001 $cred->may_save(undef);
3002 } elsif ($choice =~ /^r$/i) {
3003 return -1;
3004 } elsif ($may_save && $choice =~ /^p$/i) {
3005 $cred->may_save($may_save);
3006 } else {
3007 goto prompt;
3008 }
3009 $cred->accepted_failures($failures);
3010 $SVN::_Core::SVN_NO_ERROR;
3011}
3012sub ssl_client_cert {
3014 my ($cred, $realm, $may_save, $pool) = @_;
3015 $may_save = undef if $_no_auth_cache;
3016 print STDERR "Client certificate filename: ";
3017 STDERR->flush;
3018 chomp(my $filename = <STDIN>);
3019 $cred->cert_file($filename);
3020 $cred->may_save($may_save);
3021 $SVN::_Core::SVN_NO_ERROR;
3022}
3023sub ssl_client_cert_pw {
3025 my ($cred, $realm, $may_save, $pool) = @_;
3026 $may_save = undef if $_no_auth_cache;
3027 $cred->password(_read_password("Password: ", $realm));
3028 $cred->may_save($may_save);
3029 $SVN::_Core::SVN_NO_ERROR;
3030}
3031sub username {
3033 my ($cred, $realm, $may_save, $pool) = @_;
3034 $may_save = undef if $_no_auth_cache;
3035 if (defined $realm && length $realm) {
3036 print STDERR "Authentication realm: $realm\n";
3037 }
3038 my $username;
3039 if (defined $_username) {
3040 $username = $_username;
3041 } else {
3042 print STDERR "Username: ";
3043 STDERR->flush;
3044 chomp($username = <STDIN>);
3045 }
3046 $cred->username($username);
3047 $cred->may_save($may_save);
3048 $SVN::_Core::SVN_NO_ERROR;
3049}
3050sub _read_password {
3052 my ($prompt, $realm) = @_;
3053 print STDERR $prompt;
3054 STDERR->flush;
3055 require Term::ReadKey;
3056 Term::ReadKey::ReadMode('noecho');
3057 my $password = '';
3058 while (defined(my $key = Term::ReadKey::ReadKey(0))) {
3059 last if $key =~ /[\012\015]/; # \n\r
3060 $password .= $key;
3061 }
3062 Term::ReadKey::ReadMode('restore');
3063 print STDERR "\n";
3064 STDERR->flush;
3065 $password;
3066}
3067package SVN::Git::Fetcher;
3069use vars qw/@ISA/;
3070use strict;
3071use warnings;
3072use Carp qw/croak/;
3073use File::Temp qw/tempfile/;
3074use IO::File qw//;
3075# file baton members: path, mode_a, mode_b, pool, fh, blob, base
3077sub new {
3078 my ($class, $git_svn) = @_;
3079 my $self = SVN::Delta::Editor->new;
3080 bless $self, $class;
3081 $self->{c} = $git_svn->{last_commit} if exists $git_svn->{last_commit};
3082 $self->{empty} = {};
3083 $self->{dir_prop} = {};
3084 $self->{file_prop} = {};
3085 $self->{absent_dir} = {};
3086 $self->{absent_file} = {};
3087 $self->{gii} = $git_svn->tmp_index_do(sub { Git::IndexInfo->new });
3088 $self;
3089}
3090sub set_path_strip {
3092 my ($self, $path) = @_;
3093 $self->{path_strip} = qr/^\Q$path\E(\/|$)/ if length $path;
3094}
3095sub open_root {
3097 { path => '' };
3098}
3099sub open_directory {
3101 my ($self, $path, $pb, $rev) = @_;
3102 { path => $path };
3103}
3104sub git_path {
3106 my ($self, $path) = @_;
3107 if ($self->{path_strip}) {
3108 $path =~ s!$self->{path_strip}!! or
3109 die "Failed to strip path '$path' ($self->{path_strip})\n";
3110 }
3111 $path;
3112}
3113sub delete_entry {
3115 my ($self, $path, $rev, $pb) = @_;
3116 my $gpath = $self->git_path($path);
3118 return undef if ($gpath eq '');
3119 # remove entire directories.
3121 if (command('ls-tree', $self->{c}, '--', $gpath) =~ /^040000 tree/) {
3122 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3123 -r --name-only -z/,
3124 $self->{c}, '--', $gpath);
3125 local $/ = "\0";
3126 while (<$ls>) {
3127 chomp;
3128 $self->{gii}->remove($_);
3129 print "\tD\t$_\n" unless $::_q;
3130 }
3131 print "\tD\t$gpath/\n" unless $::_q;
3132 command_close_pipe($ls, $ctx);
3133 $self->{empty}->{$path} = 0
3134 } else {
3135 $self->{gii}->remove($gpath);
3136 print "\tD\t$gpath\n" unless $::_q;
3137 }
3138 undef;
3139}
3140sub open_file {
3142 my ($self, $path, $pb, $rev) = @_;
3143 my $gpath = $self->git_path($path);
3144 my ($mode, $blob) = (command('ls-tree', $self->{c}, '--', $gpath)
3145 =~ /^(\d{6}) blob ([a-f\d]{40})\t/);
3146 unless (defined $mode && defined $blob) {
3147 die "$path was not found in commit $self->{c} (r$rev)\n";
3148 }
3149 { path => $path, mode_a => $mode, mode_b => $mode, blob => $blob,
3150 pool => SVN::Pool->new, action => 'M' };
3151}
3152sub add_file {
3154 my ($self, $path, $pb, $cp_path, $cp_rev) = @_;
3155 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3156 delete $self->{empty}->{$dir};
3157 { path => $path, mode_a => 100644, mode_b => 100644,
3158 pool => SVN::Pool->new, action => 'A' };
3159}
3160sub add_directory {
3162 my ($self, $path, $cp_path, $cp_rev) = @_;
3163 my $gpath = $self->git_path($path);
3164 if ($gpath eq '') {
3165 my ($ls, $ctx) = command_output_pipe(qw/ls-tree
3166 -r --name-only -z/,
3167 $self->{c});
3168 local $/ = "\0";
3169 while (<$ls>) {
3170 chomp;
3171 $self->{gii}->remove($_);
3172 print "\tD\t$_\n" unless $::_q;
3173 }
3174 command_close_pipe($ls, $ctx);
3175 $self->{empty}->{$path} = 0;
3176 }
3177 my ($dir, $file) = ($path =~ m#^(.*?)/?([^/]+)$#);
3178 delete $self->{empty}->{$dir};
3179 $self->{empty}->{$path} = 1;
3180 { path => $path };
3181}
3182sub change_dir_prop {
3184 my ($self, $db, $prop, $value) = @_;
3185 $self->{dir_prop}->{$db->{path}} ||= {};
3186 $self->{dir_prop}->{$db->{path}}->{$prop} = $value;
3187 undef;
3188}
3189sub absent_directory {
3191 my ($self, $path, $pb) = @_;
3192 $self->{absent_dir}->{$pb->{path}} ||= [];
3193 push @{$self->{absent_dir}->{$pb->{path}}}, $path;
3194 undef;
3195}
3196sub absent_file {
3198 my ($self, $path, $pb) = @_;
3199 $self->{absent_file}->{$pb->{path}} ||= [];
3200 push @{$self->{absent_file}->{$pb->{path}}}, $path;
3201 undef;
3202}
3203sub change_file_prop {
3205 my ($self, $fb, $prop, $value) = @_;
3206 if ($prop eq 'svn:executable') {
3207 if ($fb->{mode_b} != 120000) {
3208 $fb->{mode_b} = defined $value ? 100755 : 100644;
3209 }
3210 } elsif ($prop eq 'svn:special') {
3211 $fb->{mode_b} = defined $value ? 120000 : 100644;
3212 } else {
3213 $self->{file_prop}->{$fb->{path}} ||= {};
3214 $self->{file_prop}->{$fb->{path}}->{$prop} = $value;
3215 }
3216 undef;
3217}
3218sub apply_textdelta {
3220 my ($self, $fb, $exp) = @_;
3221 my $fh = IO::File->new_tmpfile;
3222 $fh->autoflush(1);
3223 # $fh gets auto-closed() by SVN::TxDelta::apply(),
3224 # (but $base does not,) so dup() it for reading in close_file
3225 open my $dup, '<&', $fh or croak $!;
3226 my $base = IO::File->new_tmpfile;
3227 $base->autoflush(1);
3228 if ($fb->{blob}) {
3229 print $base 'link ' if ($fb->{mode_a} == 120000);
3230 my $size = $::_repository->cat_blob($fb->{blob}, $base);
3231 die "Failed to read object $fb->{blob}" if ($size < 0);
3232 if (defined $exp) {
3234 seek $base, 0, 0 or croak $!;
3235 my $got = ::md5sum($base);
3236 die "Checksum mismatch: $fb->{path} $fb->{blob}\n",
3237 "expected: $exp\n",
3238 " got: $got\n" if ($got ne $exp);
3239 }
3240 }
3241 seek $base, 0, 0 or croak $!;
3242 $fb->{fh} = $dup;
3243 $fb->{base} = $base;
3244 [ SVN::TxDelta::apply($base, $fh, undef, $fb->{path}, $fb->{pool}) ];
3245}
3246sub close_file {
3248 my ($self, $fb, $exp) = @_;
3249 my $hash;
3250 my $path = $self->git_path($fb->{path});
3251 if (my $fh = $fb->{fh}) {
3252 if (defined $exp) {
3253 seek($fh, 0, 0) or croak $!;
3254 my $got = ::md5sum($fh);
3255 if ($got ne $exp) {
3256 die "Checksum mismatch: $path\n",
3257 "expected: $exp\n got: $got\n";
3258 }
3259 }
3260 sysseek($fh, 0, 0) or croak $!;
3261 if ($fb->{mode_b} == 120000) {
3262 eval {
3263 sysread($fh, my $buf, 5) == 5 or croak $!;
3264 $buf eq 'link ' or die "$path has mode 120000",
3265 " but is not a link";
3266 };
3267 if ($@) {
3268 warn "$@\n";
3269 sysseek($fh, 0, 0) or croak $!;
3270 }
3271 }
3272 my ($tmp_fh, $tmp_filename) = File::Temp::tempfile(UNLINK => 1);
3274 my $result;
3275 while ($result = sysread($fh, my $string, 1024)) {
3276 my $wrote = syswrite($tmp_fh, $string, $result);
3277 defined($wrote) && $wrote == $result
3278 or croak("write $tmp_filename: $!\n");
3279 }
3280 defined $result or croak $!;
3281 close $tmp_fh or croak $!;
3282 close $fh or croak $!;
3284 $hash = $::_repository->hash_and_insert_object($tmp_filename);
3286 unlink($tmp_filename);
3287 $hash =~ /^[a-f\d]{40}$/ or die "not a sha1: $hash\n";
3288 close $fb->{base} or croak $!;
3289 } else {
3290 $hash = $fb->{blob} or die "no blob information\n";
3291 }
3292 $fb->{pool}->clear;
3293 $self->{gii}->update($fb->{mode_b}, $hash, $path) or croak $!;
3294 print "\t$fb->{action}\t$path\n" if $fb->{action} && ! $::_q;
3295 undef;
3296}
3297sub abort_edit {
3299 my $self = shift;
3300 $self->{nr} = $self->{gii}->{nr};
3301 delete $self->{gii};
3302 $self->SUPER::abort_edit(@_);
3303}
3304sub close_edit {
3306 my $self = shift;
3307 $self->{git_commit_ok} = 1;
3308 $self->{nr} = $self->{gii}->{nr};
3309 delete $self->{gii};
3310 $self->SUPER::close_edit(@_);
3311}
3312package SVN::Git::Editor;
3314use vars qw/@ISA $_rmdir $_cp_similarity $_find_copies_harder $_rename_limit/;
3315use strict;
3316use warnings;
3317use Carp qw/croak/;
3318use IO::File;
3319sub new {
3321 my ($class, $opts) = @_;
3322 foreach (qw/svn_path r ra tree_a tree_b log editor_cb/) {
3323 die "$_ required!\n" unless (defined $opts->{$_});
3324 }
3325 my $pool = SVN::Pool->new;
3327 my $mods = generate_diff($opts->{tree_a}, $opts->{tree_b});
3328 my $types = check_diff_paths($opts->{ra}, $opts->{svn_path},
3329 $opts->{r}, $mods);
3330 # $opts->{ra} functions should not be used after this:
3332 my @ce = $opts->{ra}->get_commit_editor($opts->{log},
3333 $opts->{editor_cb}, $pool);
3334 my $self = SVN::Delta::Editor->new(@ce, $pool);
3335 bless $self, $class;
3336 foreach (qw/svn_path r tree_a tree_b/) {
3337 $self->{$_} = $opts->{$_};
3338 }
3339 $self->{url} = $opts->{ra}->{url};
3340 $self->{mods} = $mods;
3341 $self->{types} = $types;
3342 $self->{pool} = $pool;
3343 $self->{bat} = { '' => $self->open_root($self->{r}, $self->{pool}) };
3344 $self->{rm} = { };
3345 $self->{path_prefix} = length $self->{svn_path} ?
3346 "$self->{svn_path}/" : '';
3347 $self->{config} = $opts->{config};
3348 return $self;
3349}
3350sub generate_diff {
3352 my ($tree_a, $tree_b) = @_;
3353 my @diff_tree = qw(diff-tree -z -r);
3354 if ($_cp_similarity) {
3355 push @diff_tree, "-C$_cp_similarity";
3356 } else {
3357 push @diff_tree, '-C';
3358 }
3359 push @diff_tree, '--find-copies-harder' if $_find_copies_harder;
3360 push @diff_tree, "-l$_rename_limit" if defined $_rename_limit;
3361 push @diff_tree, $tree_a, $tree_b;
3362 my ($diff_fh, $ctx) = command_output_pipe(@diff_tree);
3363 local $/ = "\0";
3364 my $state = 'meta';
3365 my @mods;
3366 while (<$diff_fh>) {
3367 chomp $_; # this gets rid of the trailing "\0"
3368 if ($state eq 'meta' && /^:(\d{6})\s(\d{6})\s
3369 $::sha1\s($::sha1)\s
3370 ([MTCRAD])\d*$/xo) {
3371 push @mods, { mode_a => $1, mode_b => $2,
3372 sha1_b => $3, chg => $4 };
3373 if ($4 =~ /^(?:C|R)$/) {
3374 $state = 'file_a';
3375 } else {
3376 $state = 'file_b';
3377 }
3378 } elsif ($state eq 'file_a') {
3379 my $x = $mods[$#mods] or croak "Empty array\n";
3380 if ($x->{chg} !~ /^(?:C|R)$/) {
3381 croak "Error parsing $_, $x->{chg}\n";
3382 }
3383 $x->{file_a} = $_;
3384 $state = 'file_b';
3385 } elsif ($state eq 'file_b') {
3386 my $x = $mods[$#mods] or croak "Empty array\n";
3387 if (exists $x->{file_a} && $x->{chg} !~ /^(?:C|R)$/) {
3388 croak "Error parsing $_, $x->{chg}\n";
3389 }
3390 if (!exists $x->{file_a} && $x->{chg} =~ /^(?:C|R)$/) {
3391 croak "Error parsing $_, $x->{chg}\n";
3392 }
3393 $x->{file_b} = $_;
3394 $state = 'meta';
3395 } else {
3396 croak "Error parsing $_\n";
3397 }
3398 }
3399 command_close_pipe($diff_fh, $ctx);
3400 \@mods;
3401}
3402sub check_diff_paths {
3404 my ($ra, $pfx, $rev, $mods) = @_;
3405 my %types;
3406 $pfx .= '/' if length $pfx;
3407 sub type_diff_paths {
3409 my ($ra, $types, $path, $rev) = @_;
3410 my @p = split m#/+#, $path;
3411 my $c = shift @p;
3412 unless (defined $types->{$c}) {
3413 $types->{$c} = $ra->check_path($c, $rev);
3414 }
3415 while (@p) {
3416 $c .= '/' . shift @p;
3417 next if defined $types->{$c};
3418 $types->{$c} = $ra->check_path($c, $rev);
3419 }
3420 }
3421 foreach my $m (@$mods) {
3423 foreach my $f (qw/file_a file_b/) {
3424 next unless defined $m->{$f};
3425 my ($dir) = ($m->{$f} =~ m#^(.*?)/?(?:[^/]+)$#);
3426 if (length $pfx.$dir && ! defined $types{$dir}) {
3427 type_diff_paths($ra, \%types, $pfx.$dir, $rev);
3428 }
3429 }
3430 }
3431 \%types;
3432}
3433sub split_path {
3435 return ($_[0] =~ m#^(.*?)/?([^/]+)$#);
3436}
3437sub repo_path {
3439 my ($self, $path) = @_;
3440 $self->{path_prefix}.(defined $path ? $path : '');
3441}
3442sub url_path {
3444 my ($self, $path) = @_;
3445 if ($self->{url} =~ m#^https?://#) {
3446 $path =~ s/([^a-zA-Z0-9_.-])/uc sprintf("%%%02x",ord($1))/eg;
3447 }
3448 $self->{url} . '/' . $self->repo_path($path);
3449}
3450sub rmdirs {
3452 my ($self) = @_;
3453 my $rm = $self->{rm};
3454 delete $rm->{''}; # we never delete the url we're tracking
3455 return unless %$rm;
3456 foreach (keys %$rm) {
3458 my @d = split m#/#, $_;
3459 my $c = shift @d;
3460 $rm->{$c} = 1;
3461 while (@d) {
3462 $c .= '/' . shift @d;
3463 $rm->{$c} = 1;
3464 }
3465 }
3466 delete $rm->{$self->{svn_path}};
3467 delete $rm->{''}; # we never delete the url we're tracking
3468 return unless %$rm;
3469 my ($fh, $ctx) = command_output_pipe(qw/ls-tree --name-only -r -z/,
3471 $self->{tree_b});
3472 local $/ = "\0";
3473 while (<$fh>) {
3474 chomp;
3475 my @dn = split m#/#, $_;
3476 while (pop @dn) {
3477 delete $rm->{join '/', @dn};
3478 }
3479 unless (%$rm) {
3480 close $fh;
3481 return;
3482 }
3483 }
3484 command_close_pipe($fh, $ctx);
3485 my ($r, $p, $bat) = ($self->{r}, $self->{pool}, $self->{bat});
3487 foreach my $d (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$rm) {
3488 $self->close_directory($bat->{$d}, $p);
3489 my ($dn) = ($d =~ m#^(.*?)/?(?:[^/]+)$#);
3490 print "\tD+\t$d/\n" unless $::_q;
3491 $self->SUPER::delete_entry($d, $r, $bat->{$dn}, $p);
3492 delete $bat->{$d};
3493 }
3494}
3495sub open_or_add_dir {
3497 my ($self, $full_path, $baton) = @_;
3498 my $t = $self->{types}->{$full_path};
3499 if (!defined $t) {
3500 die "$full_path not known in r$self->{r} or we have a bug!\n";
3501 }
3502 {
3503 no warnings 'once';
3504 # SVN::Node::none and SVN::Node::file are used only once,
3505 # so we're shutting up Perl's warnings about them.
3506 if ($t == $SVN::Node::none) {
3507 return $self->add_directory($full_path, $baton,
3508 undef, -1, $self->{pool});
3509 } elsif ($t == $SVN::Node::dir) {
3510 return $self->open_directory($full_path, $baton,
3511 $self->{r}, $self->{pool});
3512 } # no warnings 'once'
3513 print STDERR "$full_path already exists in repository at ",
3514 "r$self->{r} and it is not a directory (",
3515 ($t == $SVN::Node::file ? 'file' : 'unknown'),"/$t)\n";
3516 } # no warnings 'once'
3517 exit 1;
3518}
3519sub ensure_path {
3521 my ($self, $path) = @_;
3522 my $bat = $self->{bat};
3523 my $repo_path = $self->repo_path($path);
3524 return $bat->{''} unless (length $repo_path);
3525 my @p = split m#/+#, $repo_path;
3526 my $c = shift @p;
3527 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{''});
3528 while (@p) {
3529 my $c0 = $c;
3530 $c .= '/' . shift @p;
3531 $bat->{$c} ||= $self->open_or_add_dir($c, $bat->{$c0});
3532 }
3533 return $bat->{$c};
3534}
3535# Subroutine to convert a globbing pattern to a regular expression.
3537# From perl cookbook.
3538sub glob2pat {
3539 my $globstr = shift;
3540 my %patmap = ('*' => '.*', '?' => '.', '[' => '[', ']' => ']');
3541 $globstr =~ s{(.)} { $patmap{$1} || "\Q$1" }ge;
3542 return '^' . $globstr . '$';
3543}
3544sub check_autoprop {
3546 my ($self, $pattern, $properties, $file, $fbat) = @_;
3547 # Convert the globbing pattern to a regular expression.
3548 my $regex = glob2pat($pattern);
3549 # Check if the pattern matches the file name.
3550 if($file =~ m/($regex)/) {
3551 # Parse the list of properties to set.
3552 my @props = split(/;/, $properties);
3553 foreach my $prop (@props) {
3554 # Parse 'name=value' syntax and set the property.
3555 if ($prop =~ /([^=]+)=(.*)/) {
3556 my ($n,$v) = ($1,$2);
3557 for ($n, $v) {
3558 s/^\s+//; s/\s+$//;
3559 }
3560 $self->change_file_prop($fbat, $n, $v);
3561 }
3562 }
3563 }
3564}
3565sub apply_autoprops {
3567 my ($self, $file, $fbat) = @_;
3568 my $conf_t = ${$self->{config}}{'config'};
3569 no warnings 'once';
3570 # Check [miscellany]/enable-auto-props in svn configuration.
3571 if (SVN::_Core::svn_config_get_bool(
3572 $conf_t,
3573 $SVN::_Core::SVN_CONFIG_SECTION_MISCELLANY,
3574 $SVN::_Core::SVN_CONFIG_OPTION_ENABLE_AUTO_PROPS,
3575 0)) {
3576 # Auto-props are enabled. Enumerate them to look for matches.
3577 my $callback = sub {
3578 $self->check_autoprop($_[0], $_[1], $file, $fbat);
3579 };
3580 SVN::_Core::svn_config_enumerate(
3581 $conf_t,
3582 $SVN::_Core::SVN_CONFIG_SECTION_AUTO_PROPS,
3583 $callback);
3584 }
3585}
3586sub A {
3588 my ($self, $m) = @_;
3589 my ($dir, $file) = split_path($m->{file_b});
3590 my $pbat = $self->ensure_path($dir);
3591 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3592 undef, -1);
3593 print "\tA\t$m->{file_b}\n" unless $::_q;
3594 $self->apply_autoprops($file, $fbat);
3595 $self->chg_file($fbat, $m);
3596 $self->close_file($fbat,undef,$self->{pool});
3597}
3598sub C {
3600 my ($self, $m) = @_;
3601 my ($dir, $file) = split_path($m->{file_b});
3602 my $pbat = $self->ensure_path($dir);
3603 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3604 $self->url_path($m->{file_a}), $self->{r});
3605 print "\tC\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3606 $self->chg_file($fbat, $m);
3607 $self->close_file($fbat,undef,$self->{pool});
3608}
3609sub delete_entry {
3611 my ($self, $path, $pbat) = @_;
3612 my $rpath = $self->repo_path($path);
3613 my ($dir, $file) = split_path($rpath);
3614 $self->{rm}->{$dir} = 1;
3615 $self->SUPER::delete_entry($rpath, $self->{r}, $pbat, $self->{pool});
3616}
3617sub R {
3619 my ($self, $m) = @_;
3620 my ($dir, $file) = split_path($m->{file_b});
3621 my $pbat = $self->ensure_path($dir);
3622 my $fbat = $self->add_file($self->repo_path($m->{file_b}), $pbat,
3623 $self->url_path($m->{file_a}), $self->{r});
3624 print "\tR\t$m->{file_a} => $m->{file_b}\n" unless $::_q;
3625 $self->chg_file($fbat, $m);
3626 $self->close_file($fbat,undef,$self->{pool});
3627 ($dir, $file) = split_path($m->{file_a});
3629 $pbat = $self->ensure_path($dir);
3630 $self->delete_entry($m->{file_a}, $pbat);
3631}
3632sub M {
3634 my ($self, $m) = @_;
3635 my ($dir, $file) = split_path($m->{file_b});
3636 my $pbat = $self->ensure_path($dir);
3637 my $fbat = $self->open_file($self->repo_path($m->{file_b}),
3638 $pbat,$self->{r},$self->{pool});
3639 print "\t$m->{chg}\t$m->{file_b}\n" unless $::_q;
3640 $self->chg_file($fbat, $m);
3641 $self->close_file($fbat,undef,$self->{pool});
3642}
3643sub T { shift->M(@_) }
3645sub change_file_prop {
3647 my ($self, $fbat, $pname, $pval) = @_;
3648 $self->SUPER::change_file_prop($fbat, $pname, $pval, $self->{pool});
3649}
3650sub chg_file {
3652 my ($self, $fbat, $m) = @_;
3653 if ($m->{mode_b} =~ /755$/ && $m->{mode_a} !~ /755$/) {
3654 $self->change_file_prop($fbat,'svn:executable','*');
3655 } elsif ($m->{mode_b} !~ /755$/ && $m->{mode_a} =~ /755$/) {
3656 $self->change_file_prop($fbat,'svn:executable',undef);
3657 }
3658 my $fh = IO::File->new_tmpfile or croak $!;
3659 if ($m->{mode_b} =~ /^120/) {
3660 print $fh 'link ' or croak $!;
3661 $self->change_file_prop($fbat,'svn:special','*');
3662 } elsif ($m->{mode_a} =~ /^120/ && $m->{mode_b} !~ /^120/) {
3663 $self->change_file_prop($fbat,'svn:special',undef);
3664 }
3665 my $size = $::_repository->cat_blob($m->{sha1_b}, $fh);
3666 croak "Failed to read object $m->{sha1_b}" if ($size < 0);
3667 $fh->flush == 0 or croak $!;
3668 seek $fh, 0, 0 or croak $!;
3669 my $exp = ::md5sum($fh);
3671 seek $fh, 0, 0 or croak $!;
3672 my $pool = SVN::Pool->new;
3674 my $atd = $self->apply_textdelta($fbat, undef, $pool);
3675 my $got = SVN::TxDelta::send_stream($fh, @$atd, $pool);
3676 die "Checksum mismatch\nexpected: $exp\ngot: $got\n" if ($got ne $exp);
3677 $pool->clear;
3678 close $fh or croak $!;
3680}
3681sub D {
3683 my ($self, $m) = @_;
3684 my ($dir, $file) = split_path($m->{file_b});
3685 my $pbat = $self->ensure_path($dir);
3686 print "\tD\t$m->{file_b}\n" unless $::_q;
3687 $self->delete_entry($m->{file_b}, $pbat);
3688}
3689sub close_edit {
3691 my ($self) = @_;
3692 my ($p,$bat) = ($self->{pool}, $self->{bat});
3693 foreach (sort { $b =~ tr#/#/# <=> $a =~ tr#/#/# } keys %$bat) {
3694 next if $_ eq '';
3695 $self->close_directory($bat->{$_}, $p);
3696 }
3697 $self->close_directory($bat->{''}, $p);
3698 $self->SUPER::close_edit($p);
3699 $p->clear;
3700}
3701sub abort_edit {
3703 my ($self) = @_;
3704 $self->SUPER::abort_edit($self->{pool});
3705}
3706sub DESTROY {
3708 my $self = shift;
3709 $self->SUPER::DESTROY(@_);
3710 $self->{pool}->clear;
3711}
3712# this drives the editor
3714sub apply_diff {
3715 my ($self) = @_;
3716 my $mods = $self->{mods};
3717 my %o = ( D => 1, R => 0, C => -1, A => 3, M => 3, T => 3 );
3718 foreach my $m (sort { $o{$a->{chg}} <=> $o{$b->{chg}} } @$mods) {
3719 my $f = $m->{chg};
3720 if (defined $o{$f}) {
3721 $self->$f($m);
3722 } else {
3723 fatal("Invalid change type: $f");
3724 }
3725 }
3726 $self->rmdirs if $_rmdir;
3727 if (@$mods == 0) {
3728 $self->abort_edit;
3729 } else {
3730 $self->close_edit;
3731 }
3732 return scalar @$mods;
3733}
3734package Git::SVN::Ra;
3736use vars qw/@ISA $config_dir $_log_window_size/;
3737use strict;
3738use warnings;
3739my ($ra_invalid, $can_do_switch, %ignored_err, $RA);
3740BEGIN {
3742 # enforce temporary pool usage for some simple functions
3743 no strict 'refs';
3744 for my $f (qw/rev_proplist get_latest_revnum get_uuid get_repos_root/) {
3745 my $SUPER = "SUPER::$f";
3746 *$f = sub {
3747 my $self = shift;
3748 my $pool = SVN::Pool->new;
3749 my @ret = $self->$SUPER(@_,$pool);
3750 $pool->clear;
3751 wantarray ? @ret : $ret[0];
3752 };
3753 }
3754}
3755sub _auth_providers () {
3757 [
3758 SVN::Client::get_simple_provider(),
3759 SVN::Client::get_ssl_server_trust_file_provider(),
3760 SVN::Client::get_simple_prompt_provider(
3761 \&Git::SVN::Prompt::simple, 2),
3762 SVN::Client::get_ssl_client_cert_file_provider(),
3763 SVN::Client::get_ssl_client_cert_prompt_provider(
3764 \&Git::SVN::Prompt::ssl_client_cert, 2),
3765 SVN::Client::get_ssl_client_cert_pw_file_provider(),
3766 SVN::Client::get_ssl_client_cert_pw_prompt_provider(
3767 \&Git::SVN::Prompt::ssl_client_cert_pw, 2),
3768 SVN::Client::get_username_provider(),
3769 SVN::Client::get_ssl_server_trust_prompt_provider(
3770 \&Git::SVN::Prompt::ssl_server_trust),
3771 SVN::Client::get_username_prompt_provider(
3772 \&Git::SVN::Prompt::username, 2)
3773 ]
3774}
3775sub escape_uri_only {
3777 my ($uri) = @_;
3778 my @tmp;
3779 foreach (split m{/}, $uri) {
3780 s/([^\w.%+-]|%(?![a-fA-F0-9]{2}))/sprintf("%%%02X",ord($1))/eg;
3781 push @tmp, $_;
3782 }
3783 join('/', @tmp);
3784}
3785sub escape_url {
3787 my ($url) = @_;
3788 if ($url =~ m#^(https?)://([^/]+)(.*)$#) {
3789 my ($scheme, $domain, $uri) = ($1, $2, escape_uri_only($3));
3790 $url = "$scheme://$domain$uri";
3791 }
3792 $url;
3793}
3794sub new {
3796 my ($class, $url) = @_;
3797 $url =~ s!/+$!!;
3798 return $RA if ($RA && $RA->{url} eq $url);
3799 SVN::_Core::svn_config_ensure($config_dir, undef);
3801 my ($baton, $callbacks) = SVN::Core::auth_open_helper(_auth_providers);
3802 my $config = SVN::Core::config_get_config($config_dir);
3803 $RA = undef;
3804 my $dont_store_passwords = 1;
3805 my $conf_t = ${$config}{'config'};
3806 {
3807 no warnings 'once';
3808 # The usage of $SVN::_Core::SVN_CONFIG_* variables
3809 # produces warnings that variables are used only once.
3810 # I had not found the better way to shut them up, so
3811 # the warnings of type 'once' are disabled in this block.
3812 if (SVN::_Core::svn_config_get_bool($conf_t,
3813 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3814 $SVN::_Core::SVN_CONFIG_OPTION_STORE_PASSWORDS,
3815 1) == 0) {
3816 SVN::_Core::svn_auth_set_parameter($baton,
3817 $SVN::_Core::SVN_AUTH_PARAM_DONT_STORE_PASSWORDS,
3818 bless (\$dont_store_passwords, "_p_void"));
3819 }
3820 if (SVN::_Core::svn_config_get_bool($conf_t,
3821 $SVN::_Core::SVN_CONFIG_SECTION_AUTH,
3822 $SVN::_Core::SVN_CONFIG_OPTION_STORE_AUTH_CREDS,
3823 1) == 0) {
3824 $Git::SVN::Prompt::_no_auth_cache = 1;
3825 }
3826 } # no warnings 'once'
3827 my $self = SVN::Ra->new(url => escape_url($url), auth => $baton,
3828 config => $config,
3829 pool => SVN::Pool->new,
3830 auth_provider_callbacks => $callbacks);
3831 $self->{url} = $url;
3832 $self->{svn_path} = $url;
3833 $self->{repos_root} = $self->get_repos_root;
3834 $self->{svn_path} =~ s#^\Q$self->{repos_root}\E(/|$)##;
3835 $self->{cache} = { check_path => { r => 0, data => {} },
3836 get_dir => { r => 0, data => {} } };
3837 $RA = bless $self, $class;
3838}
3839sub check_path {
3841 my ($self, $path, $r) = @_;
3842 my $cache = $self->{cache}->{check_path};
3843 if ($r == $cache->{r} && exists $cache->{data}->{$path}) {
3844 return $cache->{data}->{$path};
3845 }
3846 my $pool = SVN::Pool->new;
3847 my $t = $self->SUPER::check_path($path, $r, $pool);
3848 $pool->clear;
3849 if ($r != $cache->{r}) {
3850 %{$cache->{data}} = ();
3851 $cache->{r} = $r;
3852 }
3853 $cache->{data}->{$path} = $t;
3854}
3855sub get_dir {
3857 my ($self, $dir, $r) = @_;
3858 my $cache = $self->{cache}->{get_dir};
3859 if ($r == $cache->{r}) {
3860 if (my $x = $cache->{data}->{$dir}) {
3861 return wantarray ? @$x : $x->[0];
3862 }
3863 }
3864 my $pool = SVN::Pool->new;
3865 my ($d, undef, $props) = $self->SUPER::get_dir($dir, $r, $pool);
3866 my %dirents = map { $_ => { kind => $d->{$_}->kind } } keys %$d;
3867 $pool->clear;
3868 if ($r != $cache->{r}) {
3869 %{$cache->{data}} = ();
3870 $cache->{r} = $r;
3871 }
3872 $cache->{data}->{$dir} = [ \%dirents, $r, $props ];
3873 wantarray ? (\%dirents, $r, $props) : \%dirents;
3874}
3875sub DESTROY {
3877 # do not call the real DESTROY since we store ourselves in $RA
3878}
3879sub get_log {
3881 my ($self, @args) = @_;
3882 my $pool = SVN::Pool->new;
3883 splice(@args, 3, 1) if ($SVN::Core::VERSION le '1.2.0');
3884 my $ret = $self->SUPER::get_log(@args, $pool);
3885 $pool->clear;
3886 $ret;
3887}
3888sub trees_match {
3890 my ($self, $url1, $rev1, $url2, $rev2) = @_;
3891 my $ctx = SVN::Client->new(auth => _auth_providers);
3892 my $out = IO::File->new_tmpfile;
3893 # older SVN (1.1.x) doesn't take $pool as the last parameter for
3895 # $ctx->diff(), so we'll create a default one
3896 my $pool = SVN::Pool->new_default_sub;
3897 $ra_invalid = 1; # this will open a new SVN::Ra connection to $url1
3899 $ctx->diff([], $url1, $rev1, $url2, $rev2, 1, 1, 0, $out, $out);
3900 $out->flush;
3901 my $ret = (($out->stat)[7] == 0);
3902 close $out or croak $!;
3903 $ret;
3905}
3906sub get_commit_editor {
3908 my ($self, $log, $cb, $pool) = @_;
3909 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef, 0) : ();
3910 $self->SUPER::get_commit_editor($log, $cb, @lock, $pool);
3911}
3912sub gs_do_update {
3914 my ($self, $rev_a, $rev_b, $gs, $editor) = @_;
3915 my $new = ($rev_a == $rev_b);
3916 my $path = $gs->{path};
3917 if ($new && -e $gs->{index}) {
3919 unlink $gs->{index} or die
3920 "Couldn't unlink index: $gs->{index}: $!\n";
3921 }
3922 my $pool = SVN::Pool->new;
3923 $editor->set_path_strip($path);
3924 my (@pc) = split m#/#, $path;
3925 my $reporter = $self->do_update($rev_b, (@pc ? shift @pc : ''),
3926 1, $editor, $pool);
3927 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3928 # Since we can't rely on svn_ra_reparent being available, we'll
3930 # just have to do some magic with set_path to make it so
3931 # we only want a partial path.
3932 my $sp = '';
3933 my $final = join('/', @pc);
3934 while (@pc) {
3935 $reporter->set_path($sp, $rev_b, 0, @lock, $pool);
3936 $sp .= '/' if length $sp;
3937 $sp .= shift @pc;
3938 }
3939 die "BUG: '$sp' != '$final'\n" if ($sp ne $final);
3940 $reporter->set_path($sp, $rev_a, $new, @lock, $pool);
3942 $reporter->finish_report($pool);
3944 $pool->clear;
3945 $editor->{git_commit_ok};
3946}
3947# this requires SVN 1.4.3 or later (do_switch didn't work before 1.4.3, and
3949# svn_ra_reparent didn't work before 1.4)
3950sub gs_do_switch {
3951 my ($self, $rev_a, $rev_b, $gs, $url_b, $editor) = @_;
3952 my $path = $gs->{path};
3953 my $pool = SVN::Pool->new;
3954 my $full_url = $self->{url};
3956 my $old_url = $full_url;
3957 $full_url .= '/' . escape_uri_only($path) if length $path;
3958 my ($ra, $reparented);
3959 if ($old_url ne $full_url) {
3960 if ($old_url !~ m#^svn(\+ssh)?://#) {
3961 SVN::_Ra::svn_ra_reparent($self->{session}, $full_url,
3962 $pool);
3963 $self->{url} = $full_url;
3964 $reparented = 1;
3965 } else {
3966 $_[0] = undef;
3967 $self = undef;
3968 $RA = undef;
3969 $ra = Git::SVN::Ra->new($full_url);
3970 $ra_invalid = 1;
3971 }
3972 }
3973 $ra ||= $self;
3974 my $reporter = $ra->do_switch($rev_b, '', 1, $url_b, $editor, $pool);
3975 my @lock = $SVN::Core::VERSION ge '1.2.0' ? (undef) : ();
3976 $reporter->set_path('', $rev_a, 0, @lock, $pool);
3977 $reporter->finish_report($pool);
3978 if ($reparented) {
3980 SVN::_Ra::svn_ra_reparent($self->{session}, $old_url, $pool);
3981 $self->{url} = $old_url;
3982 }
3983 $pool->clear;
3985 $editor->{git_commit_ok};
3986}
3987sub longest_common_path {
3989 my ($gsv, $globs) = @_;
3990 my %common;
3991 my $common_max = scalar @$gsv;
3992 foreach my $gs (@$gsv) {
3994 my @tmp = split m#/#, $gs->{path};
3995 my $p = '';
3996 foreach (@tmp) {
3997 $p .= length($p) ? "/$_" : $_;
3998 $common{$p} ||= 0;
3999 $common{$p}++;
4000 }
4001 }
4002 $globs ||= [];
4003 $common_max += scalar @$globs;
4004 foreach my $glob (@$globs) {
4005 my @tmp = split m#/#, $glob->{path}->{left};
4006 my $p = '';
4007 foreach (@tmp) {
4008 $p .= length($p) ? "/$_" : $_;
4009 $common{$p} ||= 0;
4010 $common{$p}++;
4011 }
4012 }
4013 my $longest_path = '';
4015 foreach (sort {length $b <=> length $a} keys %common) {
4016 if ($common{$_} == $common_max) {
4017 $longest_path = $_;
4018 last;
4019 }
4020 }
4021 $longest_path;
4022}
4023sub gs_fetch_loop_common {
4025 my ($self, $base, $head, $gsv, $globs) = @_;
4026 return if ($base > $head);
4027 my $inc = $_log_window_size;
4028 my ($min, $max) = ($base, $head < $base + $inc ? $head : $base + $inc);
4029 my $longest_path = longest_common_path($gsv, $globs);
4030 my $ra_url = $self->{url};
4031 while (1) {
4032 my %revs;
4033 my $err;
4034 my $err_handler = $SVN::Error::handler;
4035 $SVN::Error::handler = sub {
4036 ($err) = @_;
4037 skip_unknown_revs($err);
4038 };
4039 sub _cb {
4040 my ($paths, $r, $author, $date, $log) = @_;
4041 [ dup_changed_paths($paths),
4042 { author => $author, date => $date, log => $log } ];
4043 }
4044 $self->get_log([$longest_path], $min, $max, 0, 1, 1,
4045 sub { $revs{$_[1]} = _cb(@_) });
4046 if ($err && $max >= $head) {
4047 print STDERR "Path '$longest_path' ",
4048 "was probably deleted:\n",
4049 $err->expanded_message,
4050 "\nWill attempt to follow ",
4051 "revisions r$min .. r$max ",
4052 "committed before the deletion\n";
4053 my $hi = $max;
4054 while (--$hi >= $min) {
4055 my $ok;
4056 $self->get_log([$longest_path], $min, $hi,
4057 0, 1, 1, sub {
4058 $ok ||= $_[1];
4059 $revs{$_[1]} = _cb(@_) });
4060 if ($ok) {
4061 print STDERR "r$min .. r$ok OK\n";
4062 last;
4063 }
4064 }
4065 }
4066 $SVN::Error::handler = $err_handler;
4067 my %exists = map { $_->{path} => $_ } @$gsv;
4069 foreach my $r (sort {$a <=> $b} keys %revs) {
4070 my ($paths, $logged) = @{$revs{$r}};
4071 foreach my $gs ($self->match_globs(\%exists, $paths,
4073 $globs, $r)) {
4074 if ($gs->rev_map_max >= $r) {
4075 next;
4076 }
4077 next unless $gs->match_paths($paths, $r);
4078 $gs->{logged_rev_props} = $logged;
4079 if (my $last_commit = $gs->last_commit) {
4080 $gs->assert_index_clean($last_commit);
4081 }
4082 my $log_entry = $gs->do_fetch($paths, $r);
4083 if ($log_entry) {
4084 $gs->do_git_commit($log_entry);
4085 }
4086 $INDEX_FILES{$gs->{index}} = 1;
4087 }
4088 foreach my $g (@$globs) {
4089 my $k = "svn-remote.$g->{remote}." .
4090 "$g->{t}-maxRev";
4091 Git::SVN::tmp_config($k, $r);
4092 }
4093 if ($ra_invalid) {
4094 $_[0] = undef;
4095 $self = undef;
4096 $RA = undef;
4097 $self = Git::SVN::Ra->new($ra_url);
4098 $ra_invalid = undef;
4099 }
4100 }
4101 # pre-fill the .rev_db since it'll eventually get filled in
4102 # with '0' x40 if something new gets committed
4103 foreach my $gs (@$gsv) {
4104 next if $gs->rev_map_max >= $max;
4105 next if defined $gs->rev_map_get($max);
4106 $gs->rev_map_set($max, 0 x40);
4107 }
4108 foreach my $g (@$globs) {
4109 my $k = "svn-remote.$g->{remote}.$g->{t}-maxRev";
4110 Git::SVN::tmp_config($k, $max);
4111 }
4112 last if $max >= $head;
4113 $min = $max + 1;
4114 $max += $inc;
4115 $max = $head if ($max > $head);
4116 }
4117 Git::SVN::gc();
4118}
4119sub match_globs {
4121 my ($self, $exists, $paths, $globs, $r) = @_;
4122 sub get_dir_check {
4124 my ($self, $exists, $g, $r) = @_;
4125 my @x = eval { $self->get_dir($g->{path}->{left}, $r) };
4126 return unless scalar @x == 3;
4127 my $dirents = $x[0];
4128 foreach my $de (keys %$dirents) {
4129 next if $dirents->{$de}->{kind} != $SVN::Node::dir;
4130 my $p = $g->{path}->full_path($de);
4131 next if $exists->{$p};
4132 next if (length $g->{path}->{right} &&
4133 ($self->check_path($p, $r) !=
4134 $SVN::Node::dir));
4135 $exists->{$p} = Git::SVN->init($self->{url}, $p, undef,
4136 $g->{ref}->full_path($de), 1);
4137 }
4138 }
4139 foreach my $g (@$globs) {
4140 if (my $path = $paths->{"/$g->{path}->{left}"}) {
4141 if ($path->{action} =~ /^[AR]$/) {
4142 get_dir_check($self, $exists, $g, $r);
4143 }
4144 }
4145 foreach (keys %$paths) {
4146 if (/$g->{path}->{left_regex}/ &&
4147 !/$g->{path}->{regex}/) {
4148 next if $paths->{$_}->{action} !~ /^[AR]$/;
4149 get_dir_check($self, $exists, $g, $r);
4150 }
4151 next unless /$g->{path}->{regex}/;
4152 my $p = $1;
4153 my $pathname = $g->{path}->full_path($p);
4154 next if $exists->{$pathname};
4155 next if ($self->check_path($pathname, $r) !=
4156 $SVN::Node::dir);
4157 $exists->{$pathname} = Git::SVN->init(
4158 $self->{url}, $pathname, undef,
4159 $g->{ref}->full_path($p), 1);
4160 }
4161 my $c = '';
4162 foreach (split m#/#, $g->{path}->{left}) {
4163 $c .= "/$_";
4164 next unless ($paths->{$c} &&
4165 ($paths->{$c}->{action} =~ /^[AR]$/));
4166 get_dir_check($self, $exists, $g, $r);
4167 }
4168 }
4169 values %$exists;
4170}
4171sub minimize_url {
4173 my ($self) = @_;
4174 return $self->{url} if ($self->{url} eq $self->{repos_root});
4175 my $url = $self->{repos_root};
4176 my @components = split(m!/!, $self->{svn_path});
4177 my $c = '';
4178 do {
4179 $url .= "/$c" if length $c;
4180 eval { (ref $self)->new($url)->get_latest_revnum };
4181 } while ($@ && ($c = shift @components));
4182 $url;
4183}
4184sub can_do_switch {
4186 my $self = shift;
4187 unless (defined $can_do_switch) {
4188 my $pool = SVN::Pool->new;
4189 my $rep = eval {
4190 $self->do_switch(1, '', 0, $self->{url},
4191 SVN::Delta::Editor->new, $pool);
4192 };
4193 if ($@) {
4194 $can_do_switch = 0;
4195 } else {
4196 $rep->abort_report($pool);
4197 $can_do_switch = 1;
4198 }
4199 $pool->clear;
4200 }
4201 $can_do_switch;
4202}
4203sub skip_unknown_revs {
4205 my ($err) = @_;
4206 my $errno = $err->apr_err();
4207 # Maybe the branch we're tracking didn't
4208 # exist when the repo started, so it's
4209 # not an error if it doesn't, just continue
4210 #
4211 # Wonderfully consistent library, eh?
4212 # 160013 - svn:// and file://
4213 # 175002 - http(s)://
4214 # 175007 - http(s):// (this repo required authorization, too...)
4215 # More codes may be discovered later...
4216 if ($errno == 175007 || $errno == 175002 || $errno == 160013) {
4217 my $err_key = $err->expanded_message;
4218 # revision numbers change every time, filter them out
4219 $err_key =~ s/\d+/\0/g;
4220 $err_key = "$errno\0$err_key";
4221 unless ($ignored_err{$err_key}) {
4222 warn "W: Ignoring error from SVN, path probably ",
4223 "does not exist: ($errno): ",
4224 $err->expanded_message,"\n";
4225 warn "W: Do not be alarmed at the above message ",
4226 "git-svn is just searching aggressively for ",
4227 "old history.\n",
4228 "This may take a while on large repositories\n";
4229 $ignored_err{$err_key} = 1;
4230 }
4231 return;
4232 }
4233 die "Error from SVN, ($errno): ", $err->expanded_message,"\n";
4234}
4235# svn_log_changed_path_t objects passed to get_log are likely to be
4237# overwritten even if only the refs are copied to an external variable,
4238# so we should dup the structures in their entirety. Using an externally
4239# passed pool (instead of our temporary and quickly cleared pool in
4240# Git::SVN::Ra) does not help matters at all...
4241sub dup_changed_paths {
4242 my ($paths) = @_;
4243 return undef unless $paths;
4244 my %ret;
4245 foreach my $p (keys %$paths) {
4246 my $i = $paths->{$p};
4247 my %s = map { $_ => $i->$_ }
4248 qw/copyfrom_path copyfrom_rev action/;
4249 $ret{$p} = \%s;
4250 }
4251 \%ret;
4252}
4253package Git::SVN::Log;
4255use strict;
4256use warnings;
4257use POSIX qw/strftime/;
4258use constant commit_log_separator => ('-' x 72) . "\n";
4259use vars qw/$TZ $limit $color $pager $non_recursive $verbose $oneline
4260 %rusers $show_commit $incremental/;
4261my $l_fmt;
4262sub cmt_showable {
4264 my ($c) = @_;
4265 return 1 if defined $c->{r};
4266 # big commit message got truncated by the 16k pretty buffer in rev-list
4268 if ($c->{l} && $c->{l}->[-1] eq "...\n" &&
4269 $c->{a_raw} =~ /\@([a-f\d\-]+)>$/) {
4270 @{$c->{l}} = ();
4271 my @log = command(qw/cat-file commit/, $c->{c});
4272 # shift off the headers
4274 shift @log while ($log[0] ne '');
4275 shift @log;
4276 # TODO: make $c->{l} not have a trailing newline in the future
4278 @{$c->{l}} = map { "$_\n" } grep !/^git-svn-id: /, @log;
4279 (undef, $c->{r}, undef) = ::extract_metadata(
4281 (grep(/^git-svn-id: /, @log))[-1]);
4282 }
4283 return defined $c->{r};
4284}
4285sub log_use_color {
4287 return $color || Git->repository->get_colorbool('color.diff');
4288}
4289sub git_svn_log_cmd {
4291 my ($r_min, $r_max, @args) = @_;
4292 my $head = 'HEAD';
4293 my (@files, @log_opts);
4294 foreach my $x (@args) {
4295 if ($x eq '--' || @files) {
4296 push @files, $x;
4297 } else {
4298 if (::verify_ref("$x^0")) {
4299 $head = $x;
4300 } else {
4301 push @log_opts, $x;
4302 }
4303 }
4304 }
4305 my ($url, $rev, $uuid, $gs) = ::working_head_info($head);
4307 $gs ||= Git::SVN->_new;
4308 my @cmd = (qw/log --abbrev-commit --pretty=raw --default/,
4309 $gs->refname);
4310 push @cmd, '-r' unless $non_recursive;
4311 push @cmd, qw/--raw --name-status/ if $verbose;
4312 push @cmd, '--color' if log_use_color();
4313 push @cmd, @log_opts;
4314 if (defined $r_max && $r_max == $r_min) {
4315 push @cmd, '--max-count=1';
4316 if (my $c = $gs->rev_map_get($r_max)) {
4317 push @cmd, $c;
4318 }
4319 } elsif (defined $r_max) {
4320 if ($r_max < $r_min) {
4321 ($r_min, $r_max) = ($r_max, $r_min);
4322 }
4323 my (undef, $c_max) = $gs->find_rev_before($r_max, 1, $r_min);
4324 my (undef, $c_min) = $gs->find_rev_after($r_min, 1, $r_max);
4325 # If there are no commits in the range, both $c_max and $c_min
4326 # will be undefined. If there is at least 1 commit in the
4327 # range, both will be defined.
4328 return () if !defined $c_min || !defined $c_max;
4329 if ($c_min eq $c_max) {
4330 push @cmd, '--max-count=1', $c_min;
4331 } else {
4332 push @cmd, '--boundary', "$c_min..$c_max";
4333 }
4334 }
4335 return (@cmd, @files);
4336}
4337# adapted from pager.c
4339sub config_pager {
4340 $pager ||= $ENV{GIT_PAGER} || $ENV{PAGER};
4341 if (!defined $pager) {
4342 $pager = 'less';
4343 } elsif (length $pager == 0 || $pager eq 'cat') {
4344 $pager = undef;
4345 }
4346 $ENV{GIT_PAGER_IN_USE} = defined($pager);
4347}
4348sub run_pager {
4350 return unless -t *STDOUT && defined $pager;
4351 pipe my $rfd, my $wfd or return;
4352 defined(my $pid = fork) or ::fatal "Can't fork: $!";
4353 if (!$pid) {
4354 open STDOUT, '>&', $wfd or
4355 ::fatal "Can't redirect to stdout: $!";
4356 return;
4357 }
4358 open STDIN, '<&', $rfd or ::fatal "Can't redirect stdin: $!";
4359 $ENV{LESS} ||= 'FRSX';
4360 exec $pager or ::fatal "Can't run pager: $! ($pager)";
4361}
4362sub format_svn_date {
4364 return strftime("%Y-%m-%d %H:%M:%S %z (%a, %d %b %Y)", localtime(shift));
4365}
4366sub parse_git_date {
4368 my ($t, $tz) = @_;
4369 # Date::Parse isn't in the standard Perl distro :(
4370 if ($tz =~ s/^\+//) {
4371 $t += tz_to_s_offset($tz);
4372 } elsif ($tz =~ s/^\-//) {
4373 $t -= tz_to_s_offset($tz);
4374 }
4375 return $t;
4376}
4377sub set_local_timezone {
4379 if (defined $TZ) {
4380 $ENV{TZ} = $TZ;
4381 } else {
4382 delete $ENV{TZ};
4383 }
4384}
4385sub tz_to_s_offset {
4387 my ($tz) = @_;
4388 $tz =~ s/(\d\d)$//;
4389 return ($1 * 60) + ($tz * 3600);
4390}
4391sub get_author_info {
4393 my ($dest, $author, $t, $tz) = @_;
4394 $author =~ s/(?:^\s*|\s*$)//g;
4395 $dest->{a_raw} = $author;
4396 my $au;
4397 if ($::_authors) {
4398 $au = $rusers{$author} || undef;
4399 }
4400 if (!$au) {
4401 ($au) = ($author =~ /<([^>]+)\@[^>]+>$/);
4402 }
4403 $dest->{t} = $t;
4404 $dest->{tz} = $tz;
4405 $dest->{a} = $au;
4406 $dest->{t_utc} = parse_git_date($t, $tz);
4407}
4408sub process_commit {
4410 my ($c, $r_min, $r_max, $defer) = @_;
4411 if (defined $r_min && defined $r_max) {
4412 if ($r_min == $c->{r} && $r_min == $r_max) {
4413 show_commit($c);
4414 return 0;
4415 }
4416 return 1 if $r_min == $r_max;
4417 if ($r_min < $r_max) {
4418 # we need to reverse the print order
4419 return 0 if (defined $limit && --$limit < 0);
4420 push @$defer, $c;
4421 return 1;
4422 }
4423 if ($r_min != $r_max) {
4424 return 1 if ($r_min < $c->{r});
4425 return 1 if ($r_max > $c->{r});
4426 }
4427 }
4428 return 0 if (defined $limit && --$limit < 0);
4429 show_commit($c);
4430 return 1;
4431}
4432sub show_commit {
4434 my $c = shift;
4435 if ($oneline) {
4436 my $x = "\n";
4437 if (my $l = $c->{l}) {
4438 while ($l->[0] =~ /^\s*$/) { shift @$l }
4439 $x = $l->[0];
4440 }
4441 $l_fmt ||= 'A' . length($c->{r});
4442 print 'r',pack($l_fmt, $c->{r}),' | ';
4443 print "$c->{c} | " if $show_commit;
4444 print $x;
4445 } else {
4446 show_commit_normal($c);
4447 }
4448}
4449sub show_commit_changed_paths {
4451 my ($c) = @_;
4452 return unless $c->{changed};
4453 print "Changed paths:\n", @{$c->{changed}};
4454}
4455sub show_commit_normal {
4457 my ($c) = @_;
4458 print commit_log_separator, "r$c->{r} | ";
4459 print "$c->{c} | " if $show_commit;
4460 print "$c->{a} | ", format_svn_date($c->{t_utc}), ' | ';
4461 my $nr_line = 0;
4462 if (my $l = $c->{l}) {
4464 while ($l->[$#$l] eq "\n" && $#$l > 0
4465 && $l->[($#$l - 1)] eq "\n") {
4466 pop @$l;
4467 }
4468 $nr_line = scalar @$l;
4469 if (!$nr_line) {
4470 print "1 line\n\n\n";
4471 } else {
4472 if ($nr_line == 1) {
4473 $nr_line = '1 line';
4474 } else {
4475 $nr_line .= ' lines';
4476 }
4477 print $nr_line, "\n";
4478 show_commit_changed_paths($c);
4479 print "\n";
4480 print $_ foreach @$l;
4481 }
4482 } else {
4483 print "1 line\n";
4484 show_commit_changed_paths($c);
4485 print "\n";
4486 }
4488 foreach my $x (qw/raw stat diff/) {
4489 if ($c->{$x}) {
4490 print "\n";
4491 print $_ foreach @{$c->{$x}}
4492 }
4493 }
4494}
4495sub cmd_show_log {
4497 my (@args) = @_;
4498 my ($r_min, $r_max);
4499 my $r_last = -1; # prevent dupes
4500 set_local_timezone();
4501 if (defined $::_revision) {
4502 if ($::_revision =~ /^(\d+):(\d+)$/) {
4503 ($r_min, $r_max) = ($1, $2);
4504 } elsif ($::_revision =~ /^\d+$/) {
4505 $r_min = $r_max = $::_revision;
4506 } else {
4507 ::fatal "-r$::_revision is not supported, use ",
4508 "standard 'git log' arguments instead";
4509 }
4510 }
4511 config_pager();
4513 @args = git_svn_log_cmd($r_min, $r_max, @args);
4514 if (!@args) {
4515 print commit_log_separator unless $incremental || $oneline;
4516 return;
4517 }
4518 my $log = command_output_pipe(@args);
4519 run_pager();
4520 my (@k, $c, $d, $stat);
4521 my $esc_color = qr/(?:\033\[(?:(?:\d+;)*\d*)?m)*/;
4522 while (<$log>) {
4523 if (/^${esc_color}commit -?($::sha1_short)/o) {
4524 my $cmt = $1;
4525 if ($c && cmt_showable($c) && $c->{r} != $r_last) {
4526 $r_last = $c->{r};
4527 process_commit($c, $r_min, $r_max, \@k) or
4528 goto out;
4529 }
4530 $d = undef;
4531 $c = { c => $cmt };
4532 } elsif (/^${esc_color}author (.+) (\d+) ([\-\+]?\d+)$/o) {
4533 get_author_info($c, $1, $2, $3);
4534 } elsif (/^${esc_color}(?:tree|parent|committer) /o) {
4535 # ignore
4536 } elsif (/^${esc_color}:\d{6} \d{6} $::sha1_short/o) {
4537 push @{$c->{raw}}, $_;
4538 } elsif (/^${esc_color}[ACRMDT]\t/) {
4539 # we could add $SVN->{svn_path} here, but that requires
4540 # remote access at the moment (repo_path_split)...
4541 s#^(${esc_color})([ACRMDT])\t#$1 $2 #o;
4542 push @{$c->{changed}}, $_;
4543 } elsif (/^${esc_color}diff /o) {
4544 $d = 1;
4545 push @{$c->{diff}}, $_;
4546 } elsif ($d) {
4547 push @{$c->{diff}}, $_;
4548 } elsif (/^\ .+\ \|\s*\d+\ $esc_color[\+\-]*
4549 $esc_color*[\+\-]*$esc_color$/x) {
4550 $stat = 1;
4551 push @{$c->{stat}}, $_;
4552 } elsif ($stat && /^ \d+ files changed, \d+ insertions/) {
4553 push @{$c->{stat}}, $_;
4554 $stat = undef;
4555 } elsif (/^${esc_color} (git-svn-id:.+)$/o) {
4556 ($c->{url}, $c->{r}, undef) = ::extract_metadata($1);
4557 } elsif (s/^${esc_color} //o) {
4558 push @{$c->{l}}, $_;
4559 }
4560 }
4561 if ($c && defined $c->{r} && $c->{r} != $r_last) {
4562 $r_last = $c->{r};
4563 process_commit($c, $r_min, $r_max, \@k);
4564 }
4565 if (@k) {
4566 ($r_min, $r_max) = ($r_max, $r_min);
4567 process_commit($_, $r_min, $r_max) foreach reverse @k;
4568 }
4569out:
4570 close $log;
4571 print commit_log_separator unless $incremental || $oneline;
4572}
4573sub cmd_blame {
4575 my $path = pop;
4576 config_pager();
4578 run_pager();
4579 my ($fh, $ctx, $rev);
4581 if ($_git_format) {
4583 ($fh, $ctx) = command_output_pipe('blame', @_, $path);
4584 while (my $line = <$fh>) {
4585 if ($line =~ /^\^?([[:xdigit:]]+)\s/) {
4586 # Uncommitted edits show up as a rev ID of
4587 # all zeros, which we can't look up with
4588 # cmt_metadata
4589 if ($1 !~ /^0+$/) {
4590 (undef, $rev, undef) =
4591 ::cmt_metadata($1);
4592 $rev = '0' if (!$rev);
4593 } else {
4594 $rev = '0';
4595 }
4596 $rev = sprintf('%-10s', $rev);
4597 $line =~ s/^\^?[[:xdigit:]]+(\s)/$rev$1/;
4598 }
4599 print $line;
4600 }
4601 } else {
4602 ($fh, $ctx) = command_output_pipe('blame', '-p', @_, 'HEAD',
4603 '--', $path);
4604 my ($sha1);
4605 my %authors;
4606 while (my $line = <$fh>) {
4607 if ($line =~ /^([[:xdigit:]]{40})\s\d+\s\d+/) {
4608 $sha1 = $1;
4609 (undef, $rev, undef) = ::cmt_metadata($1);
4610 $rev = '0' if (!$rev);
4611 }
4612 elsif ($line =~ /^author (.*)/) {
4613 $authors{$rev} = $1;
4614 $authors{$rev} =~ s/\s/_/g;
4615 }
4616 elsif ($line =~ /^\t(.*)$/) {
4617 printf("%6s %10s %s\n", $rev, $authors{$rev}, $1);
4618 }
4619 }
4620 }
4621 command_close_pipe($fh, $ctx);
4622}
4623package Git::SVN::Migration;
4625# these version numbers do NOT correspond to actual version numbers
4626# of git nor git-svn. They are just relative.
4627#
4628# v0 layout: .git/$id/info/url, refs/heads/$id-HEAD
4629#
4630# v1 layout: .git/$id/info/url, refs/remotes/$id
4631#
4632# v2 layout: .git/svn/$id/info/url, refs/remotes/$id
4633#
4634# v3 layout: .git/svn/$id, refs/remotes/$id
4635# - info/url may remain for backwards compatibility
4636# - this is what we migrate up to this layout automatically,
4637# - this will be used by git svn init on single branches
4638# v3.1 layout (auto migrated):
4639# - .rev_db => .rev_db.$UUID, .rev_db will remain as a symlink
4640# for backwards compatibility
4641#
4642# v4 layout: .git/svn/$repo_id/$id, refs/remotes/$repo_id/$id
4643# - this is only created for newly multi-init-ed
4644# repositories. Similar in spirit to the
4645# --use-separate-remotes option in git-clone (now default)
4646# - we do not automatically migrate to this (following
4647# the example set by core git)
4648#
4649# v5 layout: .rev_db.$UUID => .rev_map.$UUID
4650# - newer, more-efficient format that uses 24-bytes per record
4651# with no filler space.
4652# - use xxd -c24 < .rev_map.$UUID to view and debug
4653# - This is a one-way migration, repositories updated to the
4654# new format will not be able to use old git-svn without
4655# rebuilding the .rev_db. Rebuilding the rev_db is not
4656# possible if noMetadata or useSvmProps are set; but should
4657# be no problem for users that use the (sensible) defaults.
4658use strict;
4659use warnings;
4660use Carp qw/croak/;
4661use File::Path qw/mkpath/;
4662use File::Basename qw/dirname basename/;
4663use vars qw/$_minimize/;
4664sub migrate_from_v0 {
4666 my $git_dir = $ENV{GIT_DIR};
4667 return undef unless -d $git_dir;
4668 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4669 my $migrated = 0;
4670 while (<$fh>) {
4671 chomp;
4672 my ($id, $orig_ref) = ($_, $_);
4673 next unless $id =~ s#^refs/heads/(.+)-HEAD$#$1#;
4674 next unless -f "$git_dir/$id/info/url";
4675 my $new_ref = "refs/remotes/$id";
4676 if (::verify_ref("$new_ref^0")) {
4677 print STDERR "W: $orig_ref is probably an old ",
4678 "branch used by an ancient version of ",
4679 "git-svn.\n",
4680 "However, $new_ref also exists.\n",
4681 "We will not be able ",
4682 "to use this branch until this ",
4683 "ambiguity is resolved.\n";
4684 next;
4685 }
4686 print STDERR "Migrating from v0 layout...\n" if !$migrated;
4687 print STDERR "Renaming ref: $orig_ref => $new_ref\n";
4688 command_noisy('update-ref', $new_ref, $orig_ref);
4689 command_noisy('update-ref', '-d', $orig_ref, $orig_ref);
4690 $migrated++;
4691 }
4692 command_close_pipe($fh, $ctx);
4693 print STDERR "Done migrating from v0 layout...\n" if $migrated;
4694 $migrated;
4695}
4696sub migrate_from_v1 {
4698 my $git_dir = $ENV{GIT_DIR};
4699 my $migrated = 0;
4700 return $migrated unless -d $git_dir;
4701 my $svn_dir = "$git_dir/svn";
4702 # just in case somebody used 'svn' as their $id at some point...
4704 return $migrated if -d $svn_dir && ! -f "$svn_dir/info/url";
4705 print STDERR "Migrating from a git-svn v1 layout...\n";
4707 mkpath([$svn_dir]);
4708 print STDERR "Data from a previous version of git-svn exists, but\n\t",
4709 "$svn_dir\n\t(required for this version ",
4710 "($::VERSION) of git-svn) does not exist.\n";
4711 my ($fh, $ctx) = command_output_pipe(qw/rev-parse --symbolic --all/);
4712 while (<$fh>) {
4713 my $x = $_;
4714 next unless $x =~ s#^refs/remotes/##;
4715 chomp $x;
4716 next unless -f "$git_dir/$x/info/url";
4717 my $u = eval { ::file_to_s("$git_dir/$x/info/url") };
4718 next unless $u;
4719 my $dn = dirname("$git_dir/svn/$x");
4720 mkpath([$dn]) unless -d $dn;
4721 if ($x eq 'svn') { # they used 'svn' as GIT_SVN_ID:
4722 mkpath(["$git_dir/svn/svn"]);
4723 print STDERR " - $git_dir/$x/info => ",
4724 "$git_dir/svn/$x/info\n";
4725 rename "$git_dir/$x/info", "$git_dir/svn/$x/info" or
4726 croak "$!: $x";
4727 # don't worry too much about these, they probably
4728 # don't exist with repos this old (save for index,
4729 # and we can easily regenerate that)
4730 foreach my $f (qw/unhandled.log index .rev_db/) {
4731 rename "$git_dir/$x/$f", "$git_dir/svn/$x/$f";
4732 }
4733 } else {
4734 print STDERR " - $git_dir/$x => $git_dir/svn/$x\n";
4735 rename "$git_dir/$x", "$git_dir/svn/$x" or
4736 croak "$!: $x";
4737 }
4738 $migrated++;
4739 }
4740 command_close_pipe($fh, $ctx);
4741 print STDERR "Done migrating from a git-svn v1 layout\n";
4742 $migrated;
4743}
4744sub read_old_urls {
4746 my ($l_map, $pfx, $path) = @_;
4747 my @dir;
4748 foreach (<$path/*>) {
4749 if (-r "$_/info/url") {
4750 $pfx .= '/' if $pfx && $pfx !~ m!/$!;
4751 my $ref_id = $pfx . basename $_;
4752 my $url = ::file_to_s("$_/info/url");
4753 $l_map->{$ref_id} = $url;
4754 } elsif (-d $_) {
4755 push @dir, $_;
4756 }
4757 }
4758 foreach (@dir) {
4759 my $x = $_;
4760 $x =~ s!^\Q$ENV{GIT_DIR}\E/svn/!!o;
4761 read_old_urls($l_map, $x, $_);
4762 }
4763}
4764sub migrate_from_v2 {
4766 my @cfg = command(qw/config -l/);
4767 return if grep /^svn-remote\..+\.url=/, @cfg;
4768 my %l_map;
4769 read_old_urls(\%l_map, '', "$ENV{GIT_DIR}/svn");
4770 my $migrated = 0;
4771 foreach my $ref_id (sort keys %l_map) {
4773 eval { Git::SVN->init($l_map{$ref_id}, '', undef, $ref_id) };
4774 if ($@) {
4775 Git::SVN->init($l_map{$ref_id}, '', $ref_id, $ref_id);
4776 }
4777 $migrated++;
4778 }
4779 $migrated;
4780}
4781sub minimize_connections {
4783 my $r = Git::SVN::read_all_remotes();
4784 my $new_urls = {};
4785 my $root_repos = {};
4786 foreach my $repo_id (keys %$r) {
4787 my $url = $r->{$repo_id}->{url} or next;
4788 my $fetch = $r->{$repo_id}->{fetch} or next;
4789 my $ra = Git::SVN::Ra->new($url);
4790 # skip existing cases where we already connect to the root
4792 if (($ra->{url} eq $ra->{repos_root}) ||
4793 ($ra->{repos_root} eq $repo_id)) {
4794 $root_repos->{$ra->{url}} = $repo_id;
4795 next;
4796 }
4797 my $root_ra = Git::SVN::Ra->new($ra->{repos_root});
4799 my $root_path = $ra->{url};
4800 $root_path =~ s#^\Q$ra->{repos_root}\E(/|$)##;
4801 foreach my $path (keys %$fetch) {
4802 my $ref_id = $fetch->{$path};
4803 my $gs = Git::SVN->new($ref_id, $repo_id, $path);
4804 # make sure we can read when connecting to
4806 # a higher level of a repository
4807 my ($last_rev, undef) = $gs->last_rev_commit;
4808 if (!defined $last_rev) {
4809 $last_rev = eval {
4810 $root_ra->get_latest_revnum;
4811 };
4812 next if $@;
4813 }
4814 my $new = $root_path;
4815 $new .= length $path ? "/$path" : '';
4816 eval {
4817 $root_ra->get_log([$new], $last_rev, $last_rev,
4818 0, 0, 1, sub { });
4819 };
4820 next if $@;
4821 $new_urls->{$ra->{repos_root}}->{$new} =
4822 { ref_id => $ref_id,
4823 old_repo_id => $repo_id,
4824 old_path => $path };
4825 }
4826 }
4827 my @emptied;
4829 foreach my $url (keys %$new_urls) {
4830 # see if we can re-use an existing [svn-remote "repo_id"]
4831 # instead of creating a(n ugly) new section:
4832 my $repo_id = $root_repos->{$url} || $url;
4833 my $fetch = $new_urls->{$url};
4835 foreach my $path (keys %$fetch) {
4836 my $x = $fetch->{$path};
4837 Git::SVN->init($url, $path, $repo_id, $x->{ref_id});
4838 my $pfx = "svn-remote.$x->{old_repo_id}";
4839 my $old_fetch = quotemeta("$x->{old_path}:".
4841 "refs/remotes/$x->{ref_id}");
4842 command_noisy(qw/config --unset/,
4843 "$pfx.fetch", '^'. $old_fetch . '$');
4844 delete $r->{$x->{old_repo_id}}->
4845 {fetch}->{$x->{old_path}};
4846 if (!keys %{$r->{$x->{old_repo_id}}->{fetch}}) {
4847 command_noisy(qw/config --unset/,
4848 "$pfx.url");
4849 push @emptied, $x->{old_repo_id}
4850 }
4851 }
4852 }
4853 if (@emptied) {
4854 my $file = $ENV{GIT_CONFIG} || $ENV{GIT_CONFIG_LOCAL} ||
4855 "$ENV{GIT_DIR}/config";
4856 print STDERR <<EOF;
4857The following [svn-remote] sections in your config file ($file) are empty
4858and can be safely removed:
4859EOF
4860 print STDERR "[svn-remote \"$_\"]\n" foreach @emptied;
4861 }
4862}
4863sub migration_check {
4865 migrate_from_v0();
4866 migrate_from_v1();
4867 migrate_from_v2();
4868 minimize_connections() if $_minimize;
4869}
4870package Git::IndexInfo;
4872use strict;
4873use warnings;
4874use Git qw/command_input_pipe command_close_pipe/;
4875sub new {
4877 my ($class) = @_;
4878 my ($gui, $ctx) = command_input_pipe(qw/update-index -z --index-info/);
4879 bless { gui => $gui, ctx => $ctx, nr => 0}, $class;
4880}
4881sub remove {
4883 my ($self, $path) = @_;
4884 if (print { $self->{gui} } '0 ', 0 x 40, "\t", $path, "\0") {
4885 return ++$self->{nr};
4886 }
4887 undef;
4888}
4889sub update {
4891 my ($self, $mode, $hash, $path) = @_;
4892 if (print { $self->{gui} } $mode, ' ', $hash, "\t", $path, "\0") {
4893 return ++$self->{nr};
4894 }
4895 undef;
4896}
4897sub DESTROY {
4899 my ($self) = @_;
4900 command_close_pipe($self->{gui}, $self->{ctx});
4901}
4902package Git::SVN::GlobSpec;
4904use strict;
4905use warnings;
4906sub new {
4908 my ($class, $glob) = @_;
4909 my $re = $glob;
4910 $re =~ s!/+$!!g; # no need for trailing slashes
4911 my $nr = ($re =~ s!^(.*)\*(.*)$!\(\[^/\]+\)!g);
4912 my ($left, $right) = ($1, $2);
4913 if ($nr > 1) {
4914 die "Only one '*' wildcard expansion ",
4915 "is supported (got $nr): '$glob'\n";
4916 } elsif ($nr == 0) {
4917 die "One '*' is needed for glob: '$glob'\n";
4918 }
4919 $re = quotemeta($left) . $re . quotemeta($right);
4920 if (length $left && !($left =~ s!/+$!!g)) {
4921 die "Missing trailing '/' on left side of: '$glob' ($left)\n";
4922 }
4923 if (length $right && !($right =~ s!^/+!!g)) {
4924 die "Missing leading '/' on right side of: '$glob' ($right)\n";
4925 }
4926 my $left_re = qr/^\/\Q$left\E(\/|$)/;
4927 bless { left => $left, right => $right, left_regex => $left_re,
4928 regex => qr/$re/, glob => $glob }, $class;
4929}
4930sub full_path {
4932 my ($self, $path) = @_;
4933 return (length $self->{left} ? "$self->{left}/" : '') .
4934 $path . (length $self->{right} ? "/$self->{right}" : '');
4935}
4936__END__
4938Data structures:
4940$remotes = { # returned by read_all_remotes()
4943 'svn' => {
4944 # svn-remote.svn.url=https://svn.musicpd.org
4945 url => 'https://svn.musicpd.org',
4946 # svn-remote.svn.fetch=mpd/trunk:trunk
4947 fetch => {
4948 'mpd/trunk' => 'trunk',
4949 },
4950 # svn-remote.svn.tags=mpd/tags/*:tags/*
4951 tags => {
4952 path => {
4953 left => 'mpd/tags',
4954 right => '',
4955 regex => qr!mpd/tags/([^/]+)$!,
4956 glob => 'tags/*',
4957 },
4958 ref => {
4959 left => 'tags',
4960 right => '',
4961 regex => qr!tags/([^/]+)$!,
4962 glob => 'tags/*',
4963 },
4964 }
4965 }
4966};
4967$log_entry hashref as returned by libsvn_log_entry()
4969{
4970 log => 'whitespace-formatted log entry
4971', # trailing newline is preserved
4972 revision => '8', # integer
4973 date => '2004-02-24T17:01:44.108345Z', # commit date
4974 author => 'committer name'
4975};
4976# this is generated by generate_diff();
4979@mods = array of diff-index line hashes, each element represents one line
4980 of diff-index output
4981diff-index line ($m hash)
4983{
4984 mode_a => first column of diff-index output, no leading ':',
4985 mode_b => second column of diff-index output,
4986 sha1_b => sha1sum of the final blob,
4987 chg => change type [MCRADT],
4988 file_a => original file name of a file (iff chg is 'C' or 'R')
4989 file_b => new/current file name of a file (any chg)
4990}
4991;
4992# retval of read_url_paths{,_all}();
4994$l_map = {
4995 # repository root url
4996 'https://svn.musicpd.org' => {
4997 # repository path # GIT_SVN_ID
4998 'mpd/trunk' => 'trunk',
4999 'mpd/tags/0.11.5' => 'tags/0.11.5',
5000 },
5001}
5002Notes:
5004 I don't trust the each() function on unless I created %hash myself
5005 because the internal iterator may not have started at base.